diff --git a/Makefile b/Makefile index 9d765ebcccf1e2..c0ab2a78a3e561 100644 --- a/Makefile +++ b/Makefile @@ -1606,7 +1606,7 @@ MRPROPER_FILES += include/config include/generated \ certs/x509.genkey \ vmlinux-gdb.py \ *.spec rpmbuild \ - rust/libmacros.so + rust/libmacros.so rust/libserde_derive.so # clean - Delete most, but leave enough to build external modules # @@ -1848,8 +1848,18 @@ PHONY += rustfmt rustfmtcheck rustfmt: $(Q)find $(abs_srctree) -type f -name '*.rs' \ -o -path $(abs_srctree)/rust/alloc -prune \ + -o -path $(abs_srctree)/rust/proc-macro2 -prune \ + -o -path $(abs_srctree)/rust/quote -prune \ + -o -path $(abs_srctree)/rust/syn -prune \ + -o -path $(abs_srctree)/rust/serde -prune \ + -o -path $(abs_srctree)/rust/serde_derive -prune \ -o -path $(abs_objtree)/rust/test -prune \ | grep -Fv $(abs_srctree)/rust/alloc \ + | grep -Fv $(abs_srctree)/rust/proc-macro2 \ + | grep -Fv $(abs_srctree)/rust/quote \ + | grep -Fv $(abs_srctree)/rust/syn \ + | grep -Fv $(abs_srctree)/rust/serde \ + | grep -Fv $(abs_srctree)/rust/serde_derive \ | grep -Fv $(abs_objtree)/rust/test \ | grep -Fv generated \ | xargs $(RUSTFMT) $(rustfmt_flags) diff --git a/rust/.gitignore b/rust/.gitignore index 21552992b401fa..f6c52116012326 100644 --- a/rust/.gitignore +++ b/rust/.gitignore @@ -1,5 +1,6 @@ # SPDX-License-Identifier: GPL-2.0 +*.rlib bindings_generated.rs bindings_helpers_generated.rs uapi_generated.rs diff --git a/rust/Makefile b/rust/Makefile index 7c9d9f11aec505..37952e93ff068c 100644 --- a/rust/Makefile +++ b/rust/Makefile @@ -8,11 +8,12 @@ always-$(CONFIG_RUST) += exports_core_generated.h obj-$(CONFIG_RUST) += helpers.o CFLAGS_REMOVE_helpers.o = -Wmissing-prototypes -Wmissing-declarations -always-$(CONFIG_RUST) += libmacros.so -no-clean-files += libmacros.so +always-$(CONFIG_RUST) += libproc_macro2.rlib libquote.rlib libsyn.rlib +always-$(CONFIG_RUST) += libserde_derive.so libmacros.so +no-clean-files += libserde_derive.so libmacros.so always-$(CONFIG_RUST) += bindings/bindings_generated.rs bindings/bindings_helpers_generated.rs -obj-$(CONFIG_RUST) += alloc.o bindings.o kernel.o +obj-$(CONFIG_RUST) += alloc.o bindings.o serde.o kernel.o always-$(CONFIG_RUST) += exports_alloc_generated.h exports_bindings_generated.h \ exports_kernel_generated.h @@ -60,11 +61,73 @@ alloc-cfgs = \ --cfg no_sync \ --cfg no_thin +proc_macro2-skip_flags := \ + --edition=2021 \ + -Drust_2018_idioms \ + -Dunreachable_pub \ + -Dunsafe_op_in_unsafe_fn + +proc_macro2-flags := \ + --edition=2018 \ + -Amissing_docs \ + --cfg 'feature="proc-macro"' \ + --cfg use_proc_macro \ + --cfg wrap_proc_macro + +quote-skip_flags := \ + --edition=2021 \ + -Drust_2018_idioms + +quote-flags := \ + --edition=2018 \ + -Amissing_docs \ + --extern proc_macro2 \ + --cfg 'feature="proc-macro"' + +syn-skip_flags := \ + --edition=2021 \ + -Drust_2018_idioms \ + -Dunreachable_pub \ + -Dunsafe_op_in_unsafe_fn + +syn-flags := \ + --edition=2018 \ + -Amissing_docs \ + --cfg 'feature="clone-impls"' \ + --cfg 'feature="derive"' \ + --cfg 'feature="extra-traits"' \ + --cfg 'feature="fold"' \ + --cfg 'feature="full"' \ + --cfg 'feature="parsing"' \ + --cfg 'feature="printing"' \ + --cfg 'feature="proc-macro"' \ + --cfg 'feature="quote"' \ + --cfg 'feature="visit"' \ + --cfg 'feature="visit-mut"' + +serde_derive-skip_flags := \ + --edition=2021 \ + -Drust_2018_idioms \ + -Dunreachable_pub + +serde_derive-flags := \ + -Amissing_docs + +serde-skip_flags := \ + --edition=2021 \ + -Drust_2018_idioms \ + -Dunreachable_pub + +serde-flags := \ + -Amissing_docs \ + --cfg no_fp_fmt_parse + quiet_cmd_rustdoc = RUSTDOC $(if $(rustdoc_host),H, ) $< cmd_rustdoc = \ OBJTREE=$(abspath $(objtree)) \ - $(RUSTDOC) $(if $(rustdoc_host),$(rust_common_flags),$(rust_flags)) \ + $(RUSTDOC) $(filter-out $(skip_flags),$(if $(rustdoc_host),$(rust_common_flags),$(rust_flags))) \ $(rustc_target_flags) -L$(objtree)/$(obj) \ + --extern quote --extern syn \ --output $(objtree)/$(obj)/doc \ --crate-name $(subst rustdoc-,,$@) \ @$(objtree)/include/generated/rustc_cfg $< @@ -81,7 +144,7 @@ quiet_cmd_rustdoc = RUSTDOC $(if $(rustdoc_host),H, ) $< # command-like flags to solve the issue. Meanwhile, we use the non-custom case # and then retouch the generated files. rustdoc: rustdoc-core rustdoc-macros rustdoc-compiler_builtins \ - rustdoc-alloc rustdoc-kernel + rustdoc-alloc rustdoc-kernel rustdoc-serde rustdoc-serde_derive $(Q)cp $(srctree)/Documentation/images/logo.svg $(objtree)/$(obj)/doc $(Q)cp $(srctree)/Documentation/images/COPYING-logo $(objtree)/$(obj)/doc $(Q)find $(objtree)/$(obj)/doc -name '*.html' -type f -print0 | xargs -0 sed -Ei \ @@ -116,17 +179,31 @@ rustdoc-alloc: $(src)/alloc/lib.rs rustdoc-core rustdoc-compiler_builtins FORCE rustdoc-kernel: private rustc_target_flags = --extern alloc \ --extern build_error --extern macros=$(objtree)/$(obj)/libmacros.so \ - --extern bindings --extern uapi + --extern bindings --extern uapi --extern serde --extern serde_derive rustdoc-kernel: $(src)/kernel/lib.rs rustdoc-core rustdoc-macros \ - rustdoc-compiler_builtins rustdoc-alloc $(obj)/libmacros.so \ - $(obj)/bindings.o FORCE + rustdoc-compiler_builtins rustdoc-alloc rustdoc-serde $(obj)/libmacros.so \ + $(obj)/bindings.o $(obj)/libserde_derive.so FORCE + $(call if_changed,rustdoc) + +rustdoc-serde_derive: private rustdoc_host = yes +rustdoc-serde_derive: private skip_flags = $(serde_derive-skip_flags) +rustdoc-serde_derive: private rustc_target_flags = --crate-type proc-macro \ + --extern proc_macro -Amissing_docs +rustdoc-serde_derive: $(src)/serde_derive/lib.rs FORCE + $(call if_changed,rustdoc) + +rustdoc-serde: private skip_flags = $(serde-skip_flags) +rustdoc-serde: private rustc_target_flags = --extern alloc --extern serde \ + -Arustdoc::broken_intra_doc_links +rustdoc-serde: $(src)/serde/lib.rs rustdoc-core FORCE $(call if_changed,rustdoc) quiet_cmd_rustc_test_library = RUSTC TL $< cmd_rustc_test_library = \ OBJTREE=$(abspath $(objtree)) \ - $(RUSTC) $(rust_common_flags) \ - @$(objtree)/include/generated/rustc_cfg $(rustc_target_flags) \ + $(RUSTC) \ + $(filter-out $(skip_flags),$(rust_common_flags) $(rustc_target_flags)) \ + @$(objtree)/include/generated/rustc_cfg \ --crate-type $(if $(rustc_test_library_proc),proc-macro,rlib) \ --out-dir $(objtree)/$(obj)/test --cfg testlib \ --sysroot $(objtree)/$(obj)/test/sysroot \ @@ -136,9 +213,25 @@ quiet_cmd_rustc_test_library = RUSTC TL $< rusttestlib-build_error: $(src)/build_error.rs rusttest-prepare FORCE $(call if_changed,rustc_test_library) -rusttestlib-macros: private rustc_target_flags = --extern proc_macro +rusttestlib-proc_macro2: private skip_flags = $(proc_macro2-skip_flags) +rusttestlib-proc_macro2: private rustc_target_flags = $(proc_macro2-flags) +rusttestlib-proc_macro2: $(src)/proc-macro2/lib.rs rusttest-prepare FORCE + $(call if_changed,rustc_test_library) + +rusttestlib-quote: private skip_flags = $(quote-skip_flags) +rusttestlib-quote: private rustc_target_flags = $(quote-flags) +rusttestlib-quote: $(src)/quote/lib.rs rusttestlib-proc_macro2 FORCE + $(call if_changed,rustc_test_library) + +rusttestlib-syn: private skip_flags = $(syn-skip_flags) +rusttestlib-syn: private rustc_target_flags = $(syn-flags) +rusttestlib-syn: $(src)/syn/lib.rs rusttestlib-quote FORCE + $(call if_changed,rustc_test_library) + +rusttestlib-macros: private rustc_target_flags = --extern proc_macro \ + --extern quote --extern syn rusttestlib-macros: private rustc_test_library_proc = yes -rusttestlib-macros: $(src)/macros/lib.rs rusttest-prepare FORCE +rusttestlib-macros: $(src)/macros/lib.rs rusttestlib-syn FORCE $(call if_changed,rustc_test_library) rusttestlib-bindings: $(src)/bindings/lib.rs rusttest-prepare FORCE @@ -147,6 +240,18 @@ rusttestlib-bindings: $(src)/bindings/lib.rs rusttest-prepare FORCE rusttestlib-uapi: $(src)/uapi/lib.rs rusttest-prepare FORCE $(call if_changed,rustc_test_library) +rusttestlib-serde_derive: private skip_flags = $(serde_derive-skip_flags) +rusttestlib-serde_derive: private rustc_target_flags = --extern proc_macro \ + --extern quote --extern syn $(serde_derive-flags) +rusttestlib-serde_derive: private rustc_test_library_proc = yes +rusttestlib-serde_derive: $(src)/serde_derive/lib.rs rusttestlib-syn FORCE + $(call if_changed,rustc_test_library) + +rusttestlib-serde: private skip_flags = $(serde-skip_flags) +rusttestlib-serde: private rustc_target_flags = $(serde-flags) +rusttestlib-serde: $(src)/serde/lib.rs rusttest-prepare FORCE + $(call if_changed,rustc_test_library) + quiet_cmd_rustdoc_test = RUSTDOC T $< cmd_rustdoc_test = \ OBJTREE=$(abspath $(objtree)) \ @@ -222,17 +327,19 @@ quiet_cmd_rustsysroot = RUSTSYSROOT rusttest-prepare: FORCE $(call if_changed,rustsysroot) -rusttest-macros: private rustc_target_flags = --extern proc_macro +rusttest-macros: private rustc_target_flags = --extern proc_macro \ + --extern quote --extern syn rusttest-macros: private rustdoc_test_target_flags = --crate-type proc-macro -rusttest-macros: $(src)/macros/lib.rs rusttest-prepare FORCE +rusttest-macros: $(src)/macros/lib.rs rusttestlib-syn FORCE $(call if_changed,rustc_test) $(call if_changed,rustdoc_test) rusttest-kernel: private rustc_target_flags = --extern alloc \ - --extern build_error --extern macros --extern bindings --extern uapi + --extern build_error --extern macros --extern bindings --extern uapi \ + --extern serde --extern serde_derive rusttest-kernel: $(src)/kernel/lib.rs rusttest-prepare \ rusttestlib-build_error rusttestlib-macros rusttestlib-bindings \ - rusttestlib-uapi FORCE + rusttestlib-uapi rusttestlib-serde rusttestlib-serde_derive FORCE $(call if_changed,rustc_test) $(call if_changed,rustc_test_library) @@ -348,17 +455,52 @@ $(obj)/exports_bindings_generated.h: $(obj)/bindings.o FORCE $(obj)/exports_kernel_generated.h: $(obj)/kernel.o FORCE $(call if_changed,exports) +quiet_cmd_rustc_hostlibrary = $(RUSTC_OR_CLIPPY_QUIET) H $@ + cmd_rustc_hostlibrary = \ + $(if $(skip_clippy),$(RUSTC),$(RUSTC_OR_CLIPPY)) \ + $(filter-out $(skip_flags),$(rust_common_flags) $(rustc_target_flags)) \ + --emit=dep-info,link --crate-type rlib -O \ + --out-dir $(objtree)/$(obj) -L$(objtree)/$(obj) \ + --crate-name $(patsubst lib%.rlib,%,$(notdir $@)) $<; \ + mv $(objtree)/$(obj)/$(patsubst lib%.rlib,%,$(notdir $@)).d $(depfile); \ + sed -i '/^\#/d' $(depfile) + +$(obj)/libproc_macro2.rlib: private skip_clippy = 1 +$(obj)/libproc_macro2.rlib: private skip_flags = $(proc_macro2-skip_flags) +$(obj)/libproc_macro2.rlib: private rustc_target_flags = $(proc_macro2-flags) +$(obj)/libproc_macro2.rlib: $(src)/proc-macro2/lib.rs FORCE + $(call if_changed_dep,rustc_hostlibrary) + +$(obj)/libquote.rlib: private skip_clippy = 1 +$(obj)/libquote.rlib: private skip_flags = $(quote-skip_flags) +$(obj)/libquote.rlib: private rustc_target_flags = $(quote-flags) +$(obj)/libquote.rlib: $(src)/quote/lib.rs $(obj)/libproc_macro2.rlib FORCE + $(call if_changed_dep,rustc_hostlibrary) + +$(obj)/libsyn.rlib: private skip_clippy = 1 +$(obj)/libsyn.rlib: private skip_flags = $(syn-skip_flags) +$(obj)/libsyn.rlib: private rustc_target_flags = $(syn-flags) +$(obj)/libsyn.rlib: $(src)/syn/lib.rs $(obj)/libquote.rlib FORCE + $(call if_changed_dep,rustc_hostlibrary) + quiet_cmd_rustc_procmacro = $(RUSTC_OR_CLIPPY_QUIET) P $@ cmd_rustc_procmacro = \ - $(RUSTC_OR_CLIPPY) $(rust_common_flags) \ + $(if $(skip_clippy),$(RUSTC),$(RUSTC_OR_CLIPPY)) \ + $(filter-out $(skip_flags),$(rust_common_flags) $(rustc_target_flags)) \ --emit=dep-info=$(depfile) --emit=link=$@ --extern proc_macro \ - --crate-type proc-macro \ - --crate-name $(patsubst lib%.so,%,$(notdir $@)) $< + --extern quote --extern syn --crate-type proc-macro \ + -L$(objtree)/$(obj) --crate-name $(patsubst lib%.so,%,$(notdir $@)) $< # Procedural macros can only be used with the `rustc` that compiled it. # Therefore, to get `libmacros.so` automatically recompiled when the compiler # version changes, we add `core.o` as a dependency (even if it is not needed). -$(obj)/libmacros.so: $(src)/macros/lib.rs $(obj)/core.o FORCE +$(obj)/libserde_derive.so: private skip_clippy = 1 +$(obj)/libserde_derive.so: private skip_flags = $(serde_derive-skip_flags) +$(obj)/libserde_derive.so: private rustc_target_flags = $(serde_derive-flags) +$(obj)/libserde_derive.so: $(src)/serde_derive/lib.rs $(obj)/libsyn.rlib $(obj)/core.o FORCE + $(call if_changed_dep,rustc_procmacro) + +$(obj)/libmacros.so: $(src)/macros/lib.rs $(obj)/libsyn.rlib $(obj)/core.o FORCE $(call if_changed_dep,rustc_procmacro) quiet_cmd_rustc_library = $(if $(skip_clippy),RUSTC,$(RUSTC_OR_CLIPPY_QUIET)) L $@ @@ -420,10 +562,18 @@ $(obj)/uapi.o: $(src)/uapi/lib.rs \ $(obj)/uapi/uapi_generated.rs FORCE $(call if_changed_dep,rustc_library) +$(obj)/serde.o: private skip_clippy = 1 +$(obj)/serde.o: private skip_flags = $(serde-skip_flags) +$(obj)/serde.o: private rustc_target_flags = $(serde-flags) +$(obj)/serde.o: $(src)/serde/lib.rs $(obj)/compiler_builtins.o FORCE + $(call if_changed_dep,rustc_library) + $(obj)/kernel.o: private rustc_target_flags = --extern alloc \ - --extern build_error --extern macros --extern bindings --extern uapi + --extern build_error --extern macros --extern bindings --extern uapi \ + --extern serde --extern serde_derive $(obj)/kernel.o: $(src)/kernel/lib.rs $(obj)/alloc.o $(obj)/build_error.o \ - $(obj)/libmacros.so $(obj)/bindings.o $(obj)/uapi.o FORCE + $(obj)/libmacros.so $(obj)/bindings.o $(obj)/uapi.o \ + $(obj)/serde.o $(obj)/libserde_derive.so FORCE $(call if_changed_dep,rustc_library) endif # CONFIG_RUST diff --git a/rust/kernel/error.rs b/rust/kernel/error.rs index 5f4114b30b94c7..01dd4d2f63d27a 100644 --- a/rust/kernel/error.rs +++ b/rust/kernel/error.rs @@ -177,7 +177,7 @@ impl From for Error { /// Note that even if a function does not return anything when it succeeds, /// it should still be modeled as returning a `Result` rather than /// just an [`Error`]. -pub type Result = core::result::Result; +pub type Result = core::result::Result; /// Converts an integer as returned by a C kernel function to an error if it's negative, and /// `Ok(())` otherwise. diff --git a/rust/kernel/lib.rs b/rust/kernel/lib.rs index 676995d4e46038..8c929f776c4c1d 100644 --- a/rust/kernel/lib.rs +++ b/rust/kernel/lib.rs @@ -101,3 +101,5 @@ fn panic(info: &core::panic::PanicInfo<'_>) -> ! { // instead of `!`. See . loop {} } + +pub mod test_serde; diff --git a/rust/kernel/test_serde.rs b/rust/kernel/test_serde.rs new file mode 100644 index 00000000000000..012bf58213a81f --- /dev/null +++ b/rust/kernel/test_serde.rs @@ -0,0 +1,26 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Test `serde`. +//! +//! It contains a data format used by the `rust_serde` sample, as well +//! as a quick check that `serde_derive` works in the `kernel` crate too. + +#![allow(missing_docs)] + +mod de; +mod error; +mod ser; + +pub use de::{from_bytes, Deserializer}; +pub use error::{Error, Result}; +pub use ser::{to_vec, Serializer}; + +use serde_derive::{Deserialize, Serialize}; + +#[derive(Serialize, Deserialize, Debug)] +pub struct S { + a: (), + b: bool, + c: bool, + d: (), +} diff --git a/rust/kernel/test_serde/de.rs b/rust/kernel/test_serde/de.rs new file mode 100644 index 00000000000000..84c98d1c1c6650 --- /dev/null +++ b/rust/kernel/test_serde/de.rs @@ -0,0 +1,442 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use super::error::{Error, Result}; +use serde::de::{self, Deserialize, DeserializeSeed, SeqAccess, Visitor}; + +pub struct Deserializer<'de> { + // This string starts with the input data and characters are truncated off + // the beginning as data is parsed. + input: &'de [u8], +} + +impl<'de> Deserializer<'de> { + // By convention, `Deserializer` constructors are named like `from_xyz`. + // That way basic use cases are satisfied by something like + // `serde_json::from_str(...)` while advanced use cases that require a + // deserializer can make one with `serde_json::Deserializer::from_str(...)`. + #[allow(clippy::should_implement_trait)] + pub fn from_bytes(input: &'de [u8]) -> Self { + Deserializer { input } + } +} + +// By convention, the public API of a Serde deserializer is one or more +// `from_xyz` methods such as `from_str`, `from_bytes`, or `from_reader` +// depending on what Rust types the deserializer is able to consume as input. +// +// This basic deserializer supports only `from_str`. +pub fn from_bytes<'a, T>(s: &'a [u8]) -> Result +where + T: Deserialize<'a>, +{ + let mut deserializer = Deserializer::from_bytes(s); + let t = T::deserialize(&mut deserializer)?; + if deserializer.input.is_empty() { + Ok(t) + } else { + Err(Error::TrailingCharacters) + } +} + +// SERDE IS NOT A PARSING LIBRARY. This impl block defines a few basic parsing +// functions from scratch. More complicated formats may wish to use a dedicated +// parsing library to help implement their Serde deserializer. +impl<'de> Deserializer<'de> { + // Look at the first character in the input without consuming it. + fn peek_byte(&mut self) -> Result { + self.input.iter().next().ok_or(Error::Eof).map(|v| *v) + } + + // Consume the first character in the input. + fn next_byte(&mut self) -> Result { + let ch = self.peek_byte()?; + self.input = &self.input[1..]; + Ok(ch) + } + + // Parse the JSON identifier `true` or `false`. + fn parse_bool(&mut self) -> Result { + if self.input.starts_with(&[1]) { + self.input = &self.input[1..]; + match self.next_byte()? { + 0 => Ok(false), + 1 => Ok(true), + _ => Err(Error::InvalidBooleanValue), + } + } else { + Err(Error::ExpectedBoolean) + } + } + + // Parse a group of decimal digits as an unsigned integer of type T. + // + // This implementation is a bit too lenient, for example `001` is not + // allowed in JSON. Also the various arithmetic operations can overflow and + // panic or return bogus data. But it is good enough for example code! + fn parse_unsigned(&mut self) -> Result { + unimplemented!() + } + + // Parse a possible minus sign followed by a group of decimal digits as a + // signed integer of type T. + fn parse_signed(&mut self) -> Result { + // Optional minus sign, delegate to `parse_unsigned`, negate if negative. + unimplemented!() + } +} + +impl<'de, 'a> de::Deserializer<'de> for &'a mut Deserializer<'de> { + type Error = Error; + + // Look at the input data to decide what Serde data model type to + // deserialize as. Not all data formats are able to support this operation. + // Formats that support `deserialize_any` are known as self-describing. + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.peek_byte()? { + 0 => self.deserialize_unit(visitor), + 1 => self.deserialize_bool(visitor), + 2 => self.deserialize_map(visitor), + _ => Err(Error::Syntax), + } + } + + // Uses the `parse_bool` parsing function defined above to read the JSON + // identifier `true` or `false` from the input. + // + // Parsing refers to looking at the input and deciding that it contains the + // JSON value `true` or `false`. + // + // Deserialization refers to mapping that JSON value into Serde's data + // model by invoking one of the `Visitor` methods. In the case of JSON and + // bool that mapping is straightforward so the distinction may seem silly, + // but in other cases Deserializers sometimes perform non-obvious mappings. + // For example the TOML format has a Datetime type and Serde's data model + // does not. In the `toml` crate, a Datetime in the input is deserialized by + // mapping it to a Serde data model "struct" type with a special name and a + // single field containing the Datetime represented as a string. + fn deserialize_bool(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_bool(self.parse_bool()?) + } + + // The `parse_signed` function is generic over the integer type `T` so here + // it is invoked with `T=i8`. The next 8 methods are similar. + fn deserialize_i8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i8(self.parse_signed()?) + } + + fn deserialize_i16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i16(self.parse_signed()?) + } + + fn deserialize_i32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i32(self.parse_signed()?) + } + + fn deserialize_i64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i64(self.parse_signed()?) + } + + fn deserialize_u8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u8(self.parse_unsigned()?) + } + + fn deserialize_u16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u16(self.parse_unsigned()?) + } + + fn deserialize_u32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u32(self.parse_unsigned()?) + } + + fn deserialize_u64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u64(self.parse_unsigned()?) + } + + // The `Serializer` implementation on the previous page serialized chars as + // single-character strings so handle that representation here. + fn deserialize_char(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // Refer to the "Understanding deserializer lifetimes" page for information + // about the three deserialization flavors of strings in Serde. + fn deserialize_str(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + fn deserialize_string(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // The `Serializer` implementation on the previous page serialized byte + // arrays as JSON arrays of bytes. Handle that representation here. + fn deserialize_bytes(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + fn deserialize_byte_buf(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // An absent optional is represented as the JSON `null` and a present + // optional is represented as just the contained value. + // + // As commented in `Serializer` implementation, this is a lossy + // representation. For example the values `Some(())` and `None` both + // serialize as just `null`. Unfortunately this is typically what people + // expect when working with JSON. Other formats are encouraged to behave + // more intelligently if possible. + fn deserialize_option(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // In Serde, unit means an anonymous value containing no data. + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + if self.input.starts_with(&[0]) { + self.input = &self.input[1..]; + visitor.visit_unit() + } else { + Err(Error::ExpectedNull) + } + } + + // Unit struct means a named value containing no data. + fn deserialize_unit_struct(self, _name: &'static str, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + // As is done here, serializers are encouraged to treat newtype structs as + // insignificant wrappers around the data they contain. That means not + // parsing anything other than the contained value. + fn deserialize_newtype_struct(self, _name: &'static str, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_newtype_struct(self) + } + + // Deserialization of compound types like sequences and maps happens by + // passing the visitor an "Access" object that gives it the ability to + // iterate through the data contained in the sequence. + fn deserialize_seq(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // Tuples look just like sequences in JSON. Some formats may be able to + // represent tuples more efficiently. + // + // As indicated by the length parameter, the `Deserialize` implementation + // for a tuple in the Serde data model is required to know the length of the + // tuple before even looking at the input data. + fn deserialize_tuple(self, _len: usize, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + // Tuple structs look just like sequences in JSON. + fn deserialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + // Much like `deserialize_seq` but calls the visitors `visit_map` method + // with a `MapAccess` implementation, rather than the visitor's `visit_seq` + // method with a `SeqAccess` implementation. + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + // Parse the opening brace of the map. + if self.next_byte()? == 2 { + // Give the visitor access to each entry of the map. + let value = visitor.visit_seq(StructFieldsVisitor::new(self))?; + // Parse the closing brace of the map. + if self.next_byte()? == 3 { + Ok(value) + } else { + Err(Error::ExpectedMapEnd) + } + } else { + Err(Error::ExpectedMap) + } + } + + // Structs look just like maps in JSON. + // + // Notice the `fields` parameter - a "struct" in the Serde data model means + // that the `Deserialize` implementation is required to know what the fields + // are before even looking at the input data. Any key-value pairing in which + // the fields cannot be known ahead of time is probably a map. + fn deserialize_struct( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_map(visitor) + } + + fn deserialize_enum( + self, + _name: &'static str, + _variants: &'static [&'static str], + _visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // An identifier in Serde is the type that identifies a field of a struct or + // the variant of an enum. In JSON, struct fields and enum variants are + // represented as strings. In other formats they may be represented as + // numeric indices. + fn deserialize_identifier(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + // Like `deserialize_any` but indicates to the `Deserializer` that it makes + // no difference which `Visitor` method is called because the data is + // ignored. + // + // Some deserializers are able to implement this more efficiently than + // `deserialize_any`, for example by rapidly skipping over matched + // delimiters without paying close attention to the data in between. + // + // Some formats are not able to implement this at all. Formats that can + // implement `deserialize_any` and `deserialize_ignored_any` are known as + // self-describing. + fn deserialize_ignored_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_any(visitor) + } +} + +struct StructFieldsVisitor<'a, 'de> { + de: &'a mut Deserializer<'de>, +} + +impl<'a, 'de> StructFieldsVisitor<'a, 'de> { + fn new(de: &'a mut Deserializer<'de>) -> Self { + StructFieldsVisitor { de } + } +} + +impl<'de, 'a> SeqAccess<'de> for StructFieldsVisitor<'a, 'de> { + type Error = Error; + + fn next_element_seed(&mut self, seed: T) -> Result> + where + T: DeserializeSeed<'de>, + { + // Check if there are no more elements. + if self.de.peek_byte()? == 3 { + return Ok(None); + } + // Deserialize an array element. + seed.deserialize(&mut *self.de).map(Some) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[test] +fn test_struct() { + use serde_derive::Deserialize; + + #[derive(Deserialize, PartialEq, Debug)] + struct Test { + a: (), + b: bool, + } + + let j = &[2, 0, 1, 0, 3]; + let expected = Test { + a: (), + b: false, + }; + assert_eq!(expected, from_bytes(j).unwrap()); +} \ No newline at end of file diff --git a/rust/kernel/test_serde/error.rs b/rust/kernel/test_serde/error.rs new file mode 100644 index 00000000000000..a1eb3720ce67bc --- /dev/null +++ b/rust/kernel/test_serde/error.rs @@ -0,0 +1,73 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use core::fmt::{self, Display}; +use serde::{de, ser}; + +pub type Result = crate::error::Result; + +// This is a bare-bones implementation. A real library would provide additional +// information in its error type, for example the line and column at which the +// error occurred, the byte offset into the input, or the current key being +// processed. +#[derive(Debug)] +pub enum Error { + // One or more variants that can be created by data structures through the + // `ser::Error` and `de::Error` traits. For example the Serialize impl for + // Mutex might return an error because the mutex is poisoned, or the + // Deserialize impl for a struct may return an error because a required + // field is missing. + Message, + + // Zero or more variants that can be created directly by the Serializer and + // Deserializer without going through `ser::Error` and `de::Error`. These + // are specific to the format, in this case JSON. + Eof, + Syntax, + ExpectedBoolean, + InvalidBooleanValue, + ExpectedInteger, + ExpectedString, + ExpectedNull, + ExpectedArray, + ExpectedArrayComma, + ExpectedArrayEnd, + ExpectedMap, + ExpectedMapColon, + ExpectedMapComma, + ExpectedMapEnd, + ExpectedEnum, + TrailingCharacters, +} + +impl ser::Error for Error { + fn custom(_msg: T) -> Self { + Error::Message + } +} + +impl de::Error for Error { + fn custom(_msg: T) -> Self { + Error::Message + } +} + +impl Display for Error { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match self { + Error::Message => write!(f, "message"), + Error::Eof => f.write_str("unexpected end of input"), + /* and so forth */ + _ => unimplemented!(), + } + } +} + +//impl core::error::Error for Error {} diff --git a/rust/kernel/test_serde/ser.rs b/rust/kernel/test_serde/ser.rs new file mode 100644 index 00000000000000..56abe7095a5f26 --- /dev/null +++ b/rust/kernel/test_serde/ser.rs @@ -0,0 +1,469 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use super::error::{Error, Result}; +use alloc::vec::Vec; +use serde::ser::{self, Serialize}; + +pub struct Serializer { + // This string starts empty and JSON is appended as values are serialized. + output: Vec, +} + +// By convention, the public API of a Serde serializer is one or more `to_abc` +// functions such as `to_string`, `to_bytes`, or `to_writer` depending on what +// Rust types the serializer is able to produce as output. +// +// This basic serializer supports only `to_string`. +pub fn to_vec(value: &T) -> Result> +where + T: Serialize, +{ + let mut serializer = Serializer { output: Vec::new() }; + value.serialize(&mut serializer)?; + Ok(serializer.output) +} + +impl<'a> ser::Serializer for &'a mut Serializer { + // The output type produced by this `Serializer` during successful + // serialization. Most serializers that produce text or binary output should + // set `Ok = ()` and serialize into an `io::Write` or buffer contained + // within the `Serializer` instance, as happens here. Serializers that build + // in-memory data structures may be simplified by using `Ok` to propagate + // the data structure around. + type Ok = (); + + // The error type when some error occurs during serialization. + type Error = Error; + + // Associated types for keeping track of additional state while serializing + // compound data structures like sequences and maps. In this case no + // additional state is required beyond what is already stored in the + // Serializer struct. + type SerializeSeq = Self; + type SerializeTuple = Self; + type SerializeTupleStruct = Self; + type SerializeTupleVariant = Self; + type SerializeMap = Self; + type SerializeStruct = Self; + type SerializeStructVariant = Self; + + // Here we go with the simple methods. The following 12 methods receive one + // of the primitive types of the data model and map it to JSON by appending + // into the output string. + fn serialize_bool(self, v: bool) -> Result<()> { + self.output.try_push(1).unwrap(); + self.output.try_push(if v { 1 } else { 0 }).unwrap(); + Ok(()) + } + + // JSON does not distinguish between different sizes of integers, so all + // signed integers will be serialized the same and all unsigned integers + // will be serialized the same. Other formats, especially compact binary + // formats, may need independent logic for the different sizes. + fn serialize_i8(self, v: i8) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + fn serialize_i16(self, v: i16) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + fn serialize_i32(self, v: i32) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + // Not particularly efficient but this is example code anyway. A more + // performant approach would be to use the `itoa` crate. + fn serialize_i64(self, _v: i64) -> Result<()> { + unimplemented!(); + } + + fn serialize_u8(self, v: u8) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u16(self, v: u16) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u32(self, v: u32) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u64(self, _v: u64) -> Result<()> { + unimplemented!(); + } + + // Serialize a char as a single-character string. Other formats may + // represent this differently. + fn serialize_char(self, _v: char) -> Result<()> { + unimplemented!(); + } + + // This only works for strings that don't require escape sequences but you + // get the idea. For example it would emit invalid JSON if the input string + // contains a '"' character. + fn serialize_str(self, _v: &str) -> Result<()> { + unimplemented!(); + } + + // Serialize a byte array as an array of bytes. Could also use a base64 + // string here. Binary formats will typically represent byte arrays more + // compactly. + fn serialize_bytes(self, v: &[u8]) -> Result<()> { + use serde::ser::SerializeSeq; + let mut seq = self.serialize_seq(Some(v.len()))?; + for byte in v { + seq.serialize_element(byte)?; + } + seq.end() + } + + // An absent optional is represented as the JSON `null`. + fn serialize_none(self) -> Result<()> { + self.serialize_unit() + } + + // A present optional is represented as just the contained value. Note that + // this is a lossy representation. For example the values `Some(())` and + // `None` both serialize as just `null`. Unfortunately this is typically + // what people expect when working with JSON. Other formats are encouraged + // to behave more intelligently if possible. + fn serialize_some(self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + // In Serde, unit means an anonymous value containing no data. Map this to + // JSON as `null`. + fn serialize_unit(self) -> Result<()> { + self.output.try_push(0).unwrap(); + Ok(()) + } + + // Unit struct means a named value containing no data. Again, since there is + // no data, map this to JSON as `null`. There is no need to serialize the + // name in most formats. + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + self.serialize_unit() + } + + // When serializing a unit variant (or any other kind of variant), formats + // can choose whether to keep track of it by index or by name. Binary + // formats typically use the index of the variant and human-readable formats + // typically use the name. + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<()> { + self.serialize_str(variant) + } + + // As is done here, serializers are encouraged to treat newtype structs as + // insignificant wrappers around the data they contain. + fn serialize_newtype_struct(self, _name: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + // Note that newtype variant (and all of the other variant serialization + // methods) refer exclusively to the "externally tagged" enum + // representation. + // + // Serialize this to JSON in externally tagged form as `{ NAME: VALUE }`. + fn serialize_newtype_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + // Now we get to the serialization of compound types. + // + // The start of the sequence, each value, and the end are three separate + // method calls. This one is responsible only for serializing the start, + // which in JSON is `[`. + // + // The length of the sequence may or may not be known ahead of time. This + // doesn't make a difference in JSON because the length is not represented + // explicitly in the serialized form. Some serializers may only be able to + // support sequences for which the length is known up front. + fn serialize_seq(self, _len: Option) -> Result { + unimplemented!(); + } + + // Tuples look just like sequences in JSON. Some formats may be able to + // represent tuples more efficiently by omitting the length, since tuple + // means that the corresponding `Deserialize implementation will know the + // length without needing to look at the serialized data. + fn serialize_tuple(self, len: usize) -> Result { + self.serialize_seq(Some(len)) + } + + // Tuple structs look just like sequences in JSON. + fn serialize_tuple_struct( + self, + _name: &'static str, + len: usize, + ) -> Result { + self.serialize_seq(Some(len)) + } + + // Tuple variants are represented in JSON as `{ NAME: [DATA...] }`. Again + // this method is only responsible for the externally tagged representation. + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + unimplemented!(); + } + + // Maps are represented in JSON as `{ K: V, K: V, ... }`. + fn serialize_map(self, _len: Option) -> Result { + self.output.try_push(2).unwrap(); + Ok(self) + } + + // Structs look just like maps in JSON. In particular, JSON requires that we + // serialize the field names of the struct. Other formats may be able to + // omit the field names when serializing structs because the corresponding + // Deserialize implementation is required to know what the keys are without + // looking at the serialized data. + fn serialize_struct(self, _name: &'static str, len: usize) -> Result { + self.serialize_map(Some(len)) + } + + // Struct variants are represented in JSON as `{ NAME: { K: V, ... } }`. + // This is the externally tagged representation. + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + unimplemented!(); + } + + fn collect_str(self, _: &T) -> Result<()> + where + T: core::fmt::Display, + { + unimplemented!() + } +} + +// The following 7 impls deal with the serialization of compound types like +// sequences and maps. Serialization of such types is begun by a Serializer +// method and followed by zero or more calls to serialize individual elements of +// the compound type and one call to end the compound type. +// +// This impl is SerializeSeq so these methods are called after `serialize_seq` +// is called on the Serializer. +impl<'a> ser::SerializeSeq for &'a mut Serializer { + // Must match the `Ok` type of the serializer. + type Ok = (); + // Must match the `Error` type of the serializer. + type Error = Error; + + // Serialize a single element of the sequence. + fn serialize_element(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + // Close the sequence. + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Same thing but for tuples. +impl<'a> ser::SerializeTuple for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_element(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Same thing but for tuple structs. +impl<'a> ser::SerializeTupleStruct for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Tuple variants are a little different. Refer back to the +// `serialize_tuple_variant` method above: +// +// self.output += "{"; +// variant.serialize(&mut *self)?; +// self.output += ":["; +// +// So the `end` method in this impl is responsible for closing both the `]` and +// the `}`. +impl<'a> ser::SerializeTupleVariant for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Some `Serialize` types are not able to hold a key and value in memory at the +// same time so `SerializeMap` implementations are required to support +// `serialize_key` and `serialize_value` individually. +// +// There is a third optional method on the `SerializeMap` trait. The +// `serialize_entry` method allows serializers to optimize for the case where +// key and value are both available simultaneously. In JSON it doesn't make a +// difference so the default behavior for `serialize_entry` is fine. +impl<'a> ser::SerializeMap for &'a mut Serializer { + type Ok = (); + type Error = Error; + + // The Serde data model allows map keys to be any serializable type. JSON + // only allows string keys so the implementation below will produce invalid + // JSON if the key serializes as something other than a string. + // + // A real JSON serializer would need to validate that map keys are strings. + // This can be done by using a different Serializer to serialize the key + // (instead of `&mut **self`) and having that other serializer only + // implement `serialize_str` and return an error on any other data type. + fn serialize_key(&mut self, _key: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!() + } + + // It doesn't make a difference whether the colon is printed at the end of + // `serialize_key` or at the beginning of `serialize_value`. In this case + // the code is a bit simpler having it here. + fn serialize_value(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!() + } + + fn end(self) -> Result<()> { + self.output.try_push(3).unwrap(); + Ok(()) + } +} + +// Structs are like maps in which the keys are constrained to be compile-time +// constant strings. +impl<'a> ser::SerializeStruct for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(&mut **self) + } + + fn end(self) -> Result<()> { + self.output.try_push(3).unwrap(); + Ok(()) + } +} + +// Similar to `SerializeTupleVariant`, here the `end` method is responsible for +// closing both of the curly braces opened by `serialize_struct_variant`. +impl<'a> ser::SerializeStructVariant for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _key: &'static str, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[test] +fn test_struct() { + use serde_derive::Serialize; + + #[derive(Serialize)] + struct Test { + a: (), + b: bool, + } + + let test = Test { + a: (), + b: false, + }; + + let mut expected = Vec::new(); + expected.try_push(2).unwrap(); + expected.try_push(0).unwrap(); + expected.try_push(1).unwrap(); + expected.try_push(0).unwrap(); + expected.try_push(3).unwrap(); + assert_eq!(to_vec(&test).unwrap(), expected); +} diff --git a/rust/proc-macro2/detection.rs b/rust/proc-macro2/detection.rs new file mode 100644 index 00000000000000..3de448cb2ddeb2 --- /dev/null +++ b/rust/proc-macro2/detection.rs @@ -0,0 +1,77 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use core::sync::atomic::{AtomicUsize, Ordering}; +use std::sync::Once; + +static WORKS: AtomicUsize = AtomicUsize::new(0); +static INIT: Once = Once::new(); + +pub(crate) fn inside_proc_macro() -> bool { + match WORKS.load(Ordering::Relaxed) { + 1 => return false, + 2 => return true, + _ => {} + } + + INIT.call_once(initialize); + inside_proc_macro() +} + +pub(crate) fn force_fallback() { + WORKS.store(1, Ordering::Relaxed); +} + +pub(crate) fn unforce_fallback() { + initialize(); +} + +#[cfg(not(no_is_available))] +fn initialize() { + let available = proc_macro::is_available(); + WORKS.store(available as usize + 1, Ordering::Relaxed); +} + +// Swap in a null panic hook to avoid printing "thread panicked" to stderr, +// then use catch_unwind to determine whether the compiler's proc_macro is +// working. When proc-macro2 is used from outside of a procedural macro all +// of the proc_macro crate's APIs currently panic. +// +// The Once is to prevent the possibility of this ordering: +// +// thread 1 calls take_hook, gets the user's original hook +// thread 1 calls set_hook with the null hook +// thread 2 calls take_hook, thinks null hook is the original hook +// thread 2 calls set_hook with the null hook +// thread 1 calls set_hook with the actual original hook +// thread 2 calls set_hook with what it thinks is the original hook +// +// in which the user's hook has been lost. +// +// There is still a race condition where a panic in a different thread can +// happen during the interval that the user's original panic hook is +// unregistered such that their hook is incorrectly not called. This is +// sufficiently unlikely and less bad than printing panic messages to stderr +// on correct use of this crate. Maybe there is a libstd feature request +// here. For now, if a user needs to guarantee that this failure mode does +// not occur, they need to call e.g. `proc_macro2::Span::call_site()` from +// the main thread before launching any other threads. +#[cfg(no_is_available)] +fn initialize() { + use std::panic::{self, PanicInfo}; + + type PanicHook = dyn Fn(&PanicInfo) + Sync + Send + 'static; + + let null_hook: Box = Box::new(|_panic_info| { /* ignore */ }); + let sanity_check = &*null_hook as *const PanicHook; + let original_hook = panic::take_hook(); + panic::set_hook(null_hook); + + let works = panic::catch_unwind(proc_macro::Span::call_site).is_ok(); + WORKS.store(works as usize + 1, Ordering::Relaxed); + + let hopefully_null_hook = panic::take_hook(); + panic::set_hook(original_hook); + if sanity_check != &*hopefully_null_hook { + panic!("observed race condition in proc_macro2::inside_proc_macro"); + } +} diff --git a/rust/proc-macro2/fallback.rs b/rust/proc-macro2/fallback.rs new file mode 100644 index 00000000000000..0d576c32162db0 --- /dev/null +++ b/rust/proc-macro2/fallback.rs @@ -0,0 +1,1004 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::parse::{self, Cursor}; +use crate::rcvec::{RcVec, RcVecBuilder, RcVecIntoIter, RcVecMut}; +use crate::{Delimiter, Spacing, TokenTree}; +#[cfg(span_locations)] +use core::cell::RefCell; +#[cfg(span_locations)] +use core::cmp; +use core::fmt::{self, Debug, Display, Write}; +use core::iter::FromIterator; +use core::mem::ManuallyDrop; +use core::ops::RangeBounds; +use core::ptr; +use core::str::FromStr; +#[cfg(procmacro2_semver_exempt)] +use std::path::Path; +use std::path::PathBuf; + +/// Force use of proc-macro2's fallback implementation of the API for now, even +/// if the compiler's implementation is available. +pub fn force() { + #[cfg(wrap_proc_macro)] + crate::detection::force_fallback(); +} + +/// Resume using the compiler's implementation of the proc macro API if it is +/// available. +pub fn unforce() { + #[cfg(wrap_proc_macro)] + crate::detection::unforce_fallback(); +} + +#[derive(Clone)] +pub(crate) struct TokenStream { + inner: RcVec, +} + +#[derive(Debug)] +pub(crate) struct LexError { + pub(crate) span: Span, +} + +impl LexError { + pub(crate) fn span(&self) -> Span { + self.span + } + + fn call_site() -> Self { + LexError { + span: Span::call_site(), + } + } +} + +impl TokenStream { + pub fn new() -> Self { + TokenStream { + inner: RcVecBuilder::new().build(), + } + } + + pub fn is_empty(&self) -> bool { + self.inner.len() == 0 + } + + fn take_inner(self) -> RcVecBuilder { + let nodrop = ManuallyDrop::new(self); + unsafe { ptr::read(&nodrop.inner) }.make_owned() + } +} + +fn push_token_from_proc_macro(mut vec: RcVecMut, token: TokenTree) { + // https://github.com/dtolnay/proc-macro2/issues/235 + match token { + #[cfg(not(no_bind_by_move_pattern_guard))] + TokenTree::Literal(crate::Literal { + #[cfg(wrap_proc_macro)] + inner: crate::imp::Literal::Fallback(literal), + #[cfg(not(wrap_proc_macro))] + inner: literal, + .. + }) if literal.repr.starts_with('-') => { + push_negative_literal(vec, literal); + } + #[cfg(no_bind_by_move_pattern_guard)] + TokenTree::Literal(crate::Literal { + #[cfg(wrap_proc_macro)] + inner: crate::imp::Literal::Fallback(literal), + #[cfg(not(wrap_proc_macro))] + inner: literal, + .. + }) => { + if literal.repr.starts_with('-') { + push_negative_literal(vec, literal); + } else { + vec.push(TokenTree::Literal(crate::Literal::_new_stable(literal))); + } + } + _ => vec.push(token), + } + + #[cold] + fn push_negative_literal(mut vec: RcVecMut, mut literal: Literal) { + literal.repr.remove(0); + let mut punct = crate::Punct::new('-', Spacing::Alone); + punct.set_span(crate::Span::_new_stable(literal.span)); + vec.push(TokenTree::Punct(punct)); + vec.push(TokenTree::Literal(crate::Literal::_new_stable(literal))); + } +} + +// Nonrecursive to prevent stack overflow. +impl Drop for TokenStream { + fn drop(&mut self) { + let mut inner = match self.inner.get_mut() { + Some(inner) => inner, + None => return, + }; + while let Some(token) = inner.pop() { + let group = match token { + TokenTree::Group(group) => group.inner, + _ => continue, + }; + #[cfg(wrap_proc_macro)] + let group = match group { + crate::imp::Group::Fallback(group) => group, + crate::imp::Group::Compiler(_) => continue, + }; + inner.extend(group.stream.take_inner()); + } + } +} + +pub(crate) struct TokenStreamBuilder { + inner: RcVecBuilder, +} + +impl TokenStreamBuilder { + pub fn new() -> Self { + TokenStreamBuilder { + inner: RcVecBuilder::new(), + } + } + + pub fn with_capacity(cap: usize) -> Self { + TokenStreamBuilder { + inner: RcVecBuilder::with_capacity(cap), + } + } + + pub fn push_token_from_parser(&mut self, tt: TokenTree) { + self.inner.push(tt); + } + + pub fn build(self) -> TokenStream { + TokenStream { + inner: self.inner.build(), + } + } +} + +#[cfg(span_locations)] +fn get_cursor(src: &str) -> Cursor { + // Create a dummy file & add it to the source map + SOURCE_MAP.with(|cm| { + let mut cm = cm.borrow_mut(); + let name = format!("", cm.files.len()); + let span = cm.add_file(&name, src); + Cursor { + rest: src, + off: span.lo, + } + }) +} + +#[cfg(not(span_locations))] +fn get_cursor(src: &str) -> Cursor { + Cursor { rest: src } +} + +impl FromStr for TokenStream { + type Err = LexError; + + fn from_str(src: &str) -> Result { + // Create a dummy file & add it to the source map + let mut cursor = get_cursor(src); + + // Strip a byte order mark if present + const BYTE_ORDER_MARK: &str = "\u{feff}"; + if cursor.starts_with(BYTE_ORDER_MARK) { + cursor = cursor.advance(BYTE_ORDER_MARK.len()); + } + + parse::token_stream(cursor) + } +} + +impl Display for LexError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str("cannot parse string into token stream") + } +} + +impl Display for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let mut joint = false; + for (i, tt) in self.inner.iter().enumerate() { + if i != 0 && !joint { + write!(f, " ")?; + } + joint = false; + match tt { + TokenTree::Group(tt) => Display::fmt(tt, f), + TokenTree::Ident(tt) => Display::fmt(tt, f), + TokenTree::Punct(tt) => { + joint = tt.spacing() == Spacing::Joint; + Display::fmt(tt, f) + } + TokenTree::Literal(tt) => Display::fmt(tt, f), + }?; + } + + Ok(()) + } +} + +impl Debug for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str("TokenStream ")?; + f.debug_list().entries(self.clone()).finish() + } +} + +#[cfg(use_proc_macro)] +impl From for TokenStream { + fn from(inner: proc_macro::TokenStream) -> TokenStream { + inner + .to_string() + .parse() + .expect("compiler token stream parse failed") + } +} + +#[cfg(use_proc_macro)] +impl From for proc_macro::TokenStream { + fn from(inner: TokenStream) -> proc_macro::TokenStream { + inner + .to_string() + .parse() + .expect("failed to parse to compiler tokens") + } +} + +impl From for TokenStream { + fn from(tree: TokenTree) -> TokenStream { + let mut stream = RcVecBuilder::new(); + push_token_from_proc_macro(stream.as_mut(), tree); + TokenStream { + inner: stream.build(), + } + } +} + +impl FromIterator for TokenStream { + fn from_iter>(tokens: I) -> Self { + let mut stream = TokenStream::new(); + stream.extend(tokens); + stream + } +} + +impl FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + let mut v = RcVecBuilder::new(); + + for stream in streams { + v.extend(stream.take_inner()); + } + + TokenStream { inner: v.build() } + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, tokens: I) { + let mut vec = self.inner.make_mut(); + tokens + .into_iter() + .for_each(|token| push_token_from_proc_macro(vec.as_mut(), token)); + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + self.inner.make_mut().extend(streams.into_iter().flatten()); + } +} + +pub(crate) type TokenTreeIter = RcVecIntoIter; + +impl IntoIterator for TokenStream { + type Item = TokenTree; + type IntoIter = TokenTreeIter; + + fn into_iter(self) -> TokenTreeIter { + self.take_inner().into_iter() + } +} + +#[derive(Clone, PartialEq, Eq)] +pub(crate) struct SourceFile { + path: PathBuf, +} + +impl SourceFile { + /// Get the path to this source file as a string. + pub fn path(&self) -> PathBuf { + self.path.clone() + } + + pub fn is_real(&self) -> bool { + // XXX(nika): Support real files in the future? + false + } +} + +impl Debug for SourceFile { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.debug_struct("SourceFile") + .field("path", &self.path()) + .field("is_real", &self.is_real()) + .finish() + } +} + +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub(crate) struct LineColumn { + pub line: usize, + pub column: usize, +} + +#[cfg(span_locations)] +thread_local! { + static SOURCE_MAP: RefCell = RefCell::new(SourceMap { + // NOTE: We start with a single dummy file which all call_site() and + // def_site() spans reference. + files: vec![FileInfo { + #[cfg(procmacro2_semver_exempt)] + name: "".to_owned(), + span: Span { lo: 0, hi: 0 }, + lines: vec![0], + }], + }); +} + +#[cfg(span_locations)] +struct FileInfo { + #[cfg(procmacro2_semver_exempt)] + name: String, + span: Span, + lines: Vec, +} + +#[cfg(span_locations)] +impl FileInfo { + fn offset_line_column(&self, offset: usize) -> LineColumn { + assert!(self.span_within(Span { + lo: offset as u32, + hi: offset as u32 + })); + let offset = offset - self.span.lo as usize; + match self.lines.binary_search(&offset) { + Ok(found) => LineColumn { + line: found + 1, + column: 0, + }, + Err(idx) => LineColumn { + line: idx, + column: offset - self.lines[idx - 1], + }, + } + } + + fn span_within(&self, span: Span) -> bool { + span.lo >= self.span.lo && span.hi <= self.span.hi + } +} + +/// Computes the offsets of each line in the given source string +/// and the total number of characters +#[cfg(span_locations)] +fn lines_offsets(s: &str) -> (usize, Vec) { + let mut lines = vec![0]; + let mut total = 0; + + for ch in s.chars() { + total += 1; + if ch == '\n' { + lines.push(total); + } + } + + (total, lines) +} + +#[cfg(span_locations)] +struct SourceMap { + files: Vec, +} + +#[cfg(span_locations)] +impl SourceMap { + fn next_start_pos(&self) -> u32 { + // Add 1 so there's always space between files. + // + // We'll always have at least 1 file, as we initialize our files list + // with a dummy file. + self.files.last().unwrap().span.hi + 1 + } + + fn add_file(&mut self, name: &str, src: &str) -> Span { + let (len, lines) = lines_offsets(src); + let lo = self.next_start_pos(); + // XXX(nika): Should we bother doing a checked cast or checked add here? + let span = Span { + lo, + hi: lo + (len as u32), + }; + + self.files.push(FileInfo { + #[cfg(procmacro2_semver_exempt)] + name: name.to_owned(), + span, + lines, + }); + + #[cfg(not(procmacro2_semver_exempt))] + let _ = name; + + span + } + + fn fileinfo(&self, span: Span) -> &FileInfo { + for file in &self.files { + if file.span_within(span) { + return file; + } + } + panic!("Invalid span with no related FileInfo!"); + } +} + +#[derive(Clone, Copy, PartialEq, Eq)] +pub(crate) struct Span { + #[cfg(span_locations)] + pub(crate) lo: u32, + #[cfg(span_locations)] + pub(crate) hi: u32, +} + +impl Span { + #[cfg(not(span_locations))] + pub fn call_site() -> Self { + Span {} + } + + #[cfg(span_locations)] + pub fn call_site() -> Self { + Span { lo: 0, hi: 0 } + } + + #[cfg(not(no_hygiene))] + pub fn mixed_site() -> Self { + Span::call_site() + } + + #[cfg(procmacro2_semver_exempt)] + pub fn def_site() -> Self { + Span::call_site() + } + + pub fn resolved_at(&self, _other: Span) -> Span { + // Stable spans consist only of line/column information, so + // `resolved_at` and `located_at` only select which span the + // caller wants line/column information from. + *self + } + + pub fn located_at(&self, other: Span) -> Span { + other + } + + #[cfg(procmacro2_semver_exempt)] + pub fn source_file(&self) -> SourceFile { + SOURCE_MAP.with(|cm| { + let cm = cm.borrow(); + let fi = cm.fileinfo(*self); + SourceFile { + path: Path::new(&fi.name).to_owned(), + } + }) + } + + #[cfg(span_locations)] + pub fn start(&self) -> LineColumn { + SOURCE_MAP.with(|cm| { + let cm = cm.borrow(); + let fi = cm.fileinfo(*self); + fi.offset_line_column(self.lo as usize) + }) + } + + #[cfg(span_locations)] + pub fn end(&self) -> LineColumn { + SOURCE_MAP.with(|cm| { + let cm = cm.borrow(); + let fi = cm.fileinfo(*self); + fi.offset_line_column(self.hi as usize) + }) + } + + #[cfg(procmacro2_semver_exempt)] + pub fn before(&self) -> Span { + Span { + #[cfg(span_locations)] + lo: self.lo, + #[cfg(span_locations)] + hi: self.lo, + } + } + + #[cfg(procmacro2_semver_exempt)] + pub fn after(&self) -> Span { + Span { + #[cfg(span_locations)] + lo: self.hi, + #[cfg(span_locations)] + hi: self.hi, + } + } + + #[cfg(not(span_locations))] + pub fn join(&self, _other: Span) -> Option { + Some(Span {}) + } + + #[cfg(span_locations)] + pub fn join(&self, other: Span) -> Option { + SOURCE_MAP.with(|cm| { + let cm = cm.borrow(); + // If `other` is not within the same FileInfo as us, return None. + if !cm.fileinfo(*self).span_within(other) { + return None; + } + Some(Span { + lo: cmp::min(self.lo, other.lo), + hi: cmp::max(self.hi, other.hi), + }) + }) + } + + #[cfg(not(span_locations))] + fn first_byte(self) -> Self { + self + } + + #[cfg(span_locations)] + fn first_byte(self) -> Self { + Span { + lo: self.lo, + hi: cmp::min(self.lo.saturating_add(1), self.hi), + } + } + + #[cfg(not(span_locations))] + fn last_byte(self) -> Self { + self + } + + #[cfg(span_locations)] + fn last_byte(self) -> Self { + Span { + lo: cmp::max(self.hi.saturating_sub(1), self.lo), + hi: self.hi, + } + } +} + +impl Debug for Span { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + #[cfg(span_locations)] + return write!(f, "bytes({}..{})", self.lo, self.hi); + + #[cfg(not(span_locations))] + write!(f, "Span") + } +} + +pub(crate) fn debug_span_field_if_nontrivial(debug: &mut fmt::DebugStruct, span: Span) { + #[cfg(span_locations)] + { + if span.lo == 0 && span.hi == 0 { + return; + } + } + + if cfg!(span_locations) { + debug.field("span", &span); + } +} + +#[derive(Clone)] +pub(crate) struct Group { + delimiter: Delimiter, + stream: TokenStream, + span: Span, +} + +impl Group { + pub fn new(delimiter: Delimiter, stream: TokenStream) -> Self { + Group { + delimiter, + stream, + span: Span::call_site(), + } + } + + pub fn delimiter(&self) -> Delimiter { + self.delimiter + } + + pub fn stream(&self) -> TokenStream { + self.stream.clone() + } + + pub fn span(&self) -> Span { + self.span + } + + pub fn span_open(&self) -> Span { + self.span.first_byte() + } + + pub fn span_close(&self) -> Span { + self.span.last_byte() + } + + pub fn set_span(&mut self, span: Span) { + self.span = span; + } +} + +impl Display for Group { + // We attempt to match libproc_macro's formatting. + // Empty parens: () + // Nonempty parens: (...) + // Empty brackets: [] + // Nonempty brackets: [...] + // Empty braces: { } + // Nonempty braces: { ... } + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let (open, close) = match self.delimiter { + Delimiter::Parenthesis => ("(", ")"), + Delimiter::Brace => ("{ ", "}"), + Delimiter::Bracket => ("[", "]"), + Delimiter::None => ("", ""), + }; + + f.write_str(open)?; + Display::fmt(&self.stream, f)?; + if self.delimiter == Delimiter::Brace && !self.stream.inner.is_empty() { + f.write_str(" ")?; + } + f.write_str(close)?; + + Ok(()) + } +} + +impl Debug for Group { + fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + let mut debug = fmt.debug_struct("Group"); + debug.field("delimiter", &self.delimiter); + debug.field("stream", &self.stream); + debug_span_field_if_nontrivial(&mut debug, self.span); + debug.finish() + } +} + +#[derive(Clone)] +pub(crate) struct Ident { + sym: String, + span: Span, + raw: bool, +} + +impl Ident { + fn _new(string: &str, raw: bool, span: Span) -> Self { + validate_ident(string, raw); + + Ident { + sym: string.to_owned(), + span, + raw, + } + } + + pub fn new(string: &str, span: Span) -> Self { + Ident::_new(string, false, span) + } + + pub fn new_raw(string: &str, span: Span) -> Self { + Ident::_new(string, true, span) + } + + pub fn span(&self) -> Span { + self.span + } + + pub fn set_span(&mut self, span: Span) { + self.span = span; + } +} + +pub(crate) fn is_ident_start(c: char) -> bool { + c == '_' || c.is_ascii_alphabetic() +} + +pub(crate) fn is_ident_continue(c: char) -> bool { + c == '_' || c.is_ascii_alphanumeric() +} + +fn validate_ident(string: &str, raw: bool) { + if string.is_empty() { + panic!("Ident is not allowed to be empty; use Option"); + } + + if string.bytes().all(|digit| digit >= b'0' && digit <= b'9') { + panic!("Ident cannot be a number; use Literal instead"); + } + + fn ident_ok(string: &str) -> bool { + let mut chars = string.chars(); + let first = chars.next().unwrap(); + if !is_ident_start(first) { + return false; + } + for ch in chars { + if !is_ident_continue(ch) { + return false; + } + } + true + } + + if !ident_ok(string) { + panic!("{:?} is not a valid Ident", string); + } + + if raw { + match string { + "_" | "super" | "self" | "Self" | "crate" => { + panic!("`r#{}` cannot be a raw identifier", string); + } + _ => {} + } + } +} + +impl PartialEq for Ident { + fn eq(&self, other: &Ident) -> bool { + self.sym == other.sym && self.raw == other.raw + } +} + +impl PartialEq for Ident +where + T: ?Sized + AsRef, +{ + fn eq(&self, other: &T) -> bool { + let other = other.as_ref(); + if self.raw { + other.starts_with("r#") && self.sym == other[2..] + } else { + self.sym == other + } + } +} + +impl Display for Ident { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + if self.raw { + f.write_str("r#")?; + } + Display::fmt(&self.sym, f) + } +} + +impl Debug for Ident { + // Ident(proc_macro), Ident(r#union) + #[cfg(not(span_locations))] + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let mut debug = f.debug_tuple("Ident"); + debug.field(&format_args!("{}", self)); + debug.finish() + } + + // Ident { + // sym: proc_macro, + // span: bytes(128..138) + // } + #[cfg(span_locations)] + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let mut debug = f.debug_struct("Ident"); + debug.field("sym", &format_args!("{}", self)); + debug_span_field_if_nontrivial(&mut debug, self.span); + debug.finish() + } +} + +#[derive(Clone)] +pub(crate) struct Literal { + repr: String, + span: Span, +} + +macro_rules! suffixed_numbers { + ($($name:ident => $kind:ident,)*) => ($( + pub fn $name(n: $kind) -> Literal { + Literal::_new(format!(concat!("{}", stringify!($kind)), n)) + } + )*) +} + +macro_rules! unsuffixed_numbers { + ($($name:ident => $kind:ident,)*) => ($( + pub fn $name(n: $kind) -> Literal { + Literal::_new(n.to_string()) + } + )*) +} + +impl Literal { + pub(crate) fn _new(repr: String) -> Self { + Literal { + repr, + span: Span::call_site(), + } + } + + pub(crate) unsafe fn from_str_unchecked(repr: &str) -> Self { + Literal::_new(repr.to_owned()) + } + + suffixed_numbers! { + u8_suffixed => u8, + u16_suffixed => u16, + u32_suffixed => u32, + u64_suffixed => u64, + u128_suffixed => u128, + usize_suffixed => usize, + i8_suffixed => i8, + i16_suffixed => i16, + i32_suffixed => i32, + i64_suffixed => i64, + i128_suffixed => i128, + isize_suffixed => isize, + + f32_suffixed => f32, + f64_suffixed => f64, + } + + unsuffixed_numbers! { + u8_unsuffixed => u8, + u16_unsuffixed => u16, + u32_unsuffixed => u32, + u64_unsuffixed => u64, + u128_unsuffixed => u128, + usize_unsuffixed => usize, + i8_unsuffixed => i8, + i16_unsuffixed => i16, + i32_unsuffixed => i32, + i64_unsuffixed => i64, + i128_unsuffixed => i128, + isize_unsuffixed => isize, + } + + pub fn f32_unsuffixed(f: f32) -> Literal { + let mut s = f.to_string(); + if !s.contains('.') { + s.push_str(".0"); + } + Literal::_new(s) + } + + pub fn f64_unsuffixed(f: f64) -> Literal { + let mut s = f.to_string(); + if !s.contains('.') { + s.push_str(".0"); + } + Literal::_new(s) + } + + pub fn string(t: &str) -> Literal { + let mut repr = String::with_capacity(t.len() + 2); + repr.push('"'); + for c in t.chars() { + if c == '\'' { + // escape_debug turns this into "\'" which is unnecessary. + repr.push(c); + } else { + repr.extend(c.escape_debug()); + } + } + repr.push('"'); + Literal::_new(repr) + } + + pub fn character(t: char) -> Literal { + let mut repr = String::new(); + repr.push('\''); + if t == '"' { + // escape_debug turns this into '\"' which is unnecessary. + repr.push(t); + } else { + repr.extend(t.escape_debug()); + } + repr.push('\''); + Literal::_new(repr) + } + + pub fn byte_string(bytes: &[u8]) -> Literal { + let mut escaped = "b\"".to_string(); + for b in bytes { + #[allow(clippy::match_overlapping_arm)] + match *b { + b'\0' => escaped.push_str(r"\0"), + b'\t' => escaped.push_str(r"\t"), + b'\n' => escaped.push_str(r"\n"), + b'\r' => escaped.push_str(r"\r"), + b'"' => escaped.push_str("\\\""), + b'\\' => escaped.push_str("\\\\"), + b'\x20'..=b'\x7E' => escaped.push(*b as char), + _ => { + let _ = write!(escaped, "\\x{:02X}", b); + } + } + } + escaped.push('"'); + Literal::_new(escaped) + } + + pub fn span(&self) -> Span { + self.span + } + + pub fn set_span(&mut self, span: Span) { + self.span = span; + } + + pub fn subspan>(&self, _range: R) -> Option { + None + } +} + +impl FromStr for Literal { + type Err = LexError; + + fn from_str(mut repr: &str) -> Result { + let negative = repr.starts_with('-'); + if negative { + repr = &repr[1..]; + if !repr.starts_with(|ch: char| ch.is_ascii_digit()) { + return Err(LexError::call_site()); + } + } + let cursor = get_cursor(repr); + if let Ok((_rest, mut literal)) = parse::literal(cursor) { + if literal.repr.len() == repr.len() { + if negative { + literal.repr.insert(0, '-'); + } + return Ok(literal); + } + } + Err(LexError::call_site()) + } +} + +impl Display for Literal { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.repr, f) + } +} + +impl Debug for Literal { + fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + let mut debug = fmt.debug_struct("Literal"); + debug.field("lit", &format_args!("{}", self.repr)); + debug_span_field_if_nontrivial(&mut debug, self.span); + debug.finish() + } +} diff --git a/rust/proc-macro2/lib.rs b/rust/proc-macro2/lib.rs new file mode 100644 index 00000000000000..d3b0454a9c0fe0 --- /dev/null +++ b/rust/proc-macro2/lib.rs @@ -0,0 +1,1341 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! [![github]](https://github.com/dtolnay/proc-macro2) [![crates-io]](https://crates.io/crates/proc-macro2) [![docs-rs]](crate) +//! +//! [github]: https://img.shields.io/badge/github-8da0cb?style=for-the-badge&labelColor=555555&logo=github +//! [crates-io]: https://img.shields.io/badge/crates.io-fc8d62?style=for-the-badge&labelColor=555555&logo=rust +//! [docs-rs]: https://img.shields.io/badge/docs.rs-66c2a5?style=for-the-badge&labelColor=555555&logo=docs.rs +//! +//!
+//! +//! A wrapper around the procedural macro API of the compiler's [`proc_macro`] +//! crate. This library serves two purposes: +//! +//! [`proc_macro`]: https://doc.rust-lang.org/proc_macro/ +//! +//! - **Bring proc-macro-like functionality to other contexts like build.rs and +//! main.rs.** Types from `proc_macro` are entirely specific to procedural +//! macros and cannot ever exist in code outside of a procedural macro. +//! Meanwhile `proc_macro2` types may exist anywhere including non-macro code. +//! By developing foundational libraries like [syn] and [quote] against +//! `proc_macro2` rather than `proc_macro`, the procedural macro ecosystem +//! becomes easily applicable to many other use cases and we avoid +//! reimplementing non-macro equivalents of those libraries. +//! +//! - **Make procedural macros unit testable.** As a consequence of being +//! specific to procedural macros, nothing that uses `proc_macro` can be +//! executed from a unit test. In order for helper libraries or components of +//! a macro to be testable in isolation, they must be implemented using +//! `proc_macro2`. +//! +//! [syn]: https://github.com/dtolnay/syn +//! [quote]: https://github.com/dtolnay/quote +//! +//! # Usage +//! +//! The skeleton of a typical procedural macro typically looks like this: +//! +//! ``` +//! extern crate proc_macro; +//! +//! # const IGNORE: &str = stringify! { +//! #[proc_macro_derive(MyDerive)] +//! # }; +//! # #[cfg(wrap_proc_macro)] +//! pub fn my_derive(input: proc_macro::TokenStream) -> proc_macro::TokenStream { +//! let input = proc_macro2::TokenStream::from(input); +//! +//! let output: proc_macro2::TokenStream = { +//! /* transform input */ +//! # input +//! }; +//! +//! proc_macro::TokenStream::from(output) +//! } +//! ``` +//! +//! If parsing with [Syn], you'll use [`parse_macro_input!`] instead to +//! propagate parse errors correctly back to the compiler when parsing fails. +//! +//! [`parse_macro_input!`]: https://docs.rs/syn/1.0/syn/macro.parse_macro_input.html +//! +//! # Unstable features +//! +//! The default feature set of proc-macro2 tracks the most recent stable +//! compiler API. Functionality in `proc_macro` that is not yet stable is not +//! exposed by proc-macro2 by default. +//! +//! To opt into the additional APIs available in the most recent nightly +//! compiler, the `procmacro2_semver_exempt` config flag must be passed to +//! rustc. We will polyfill those nightly-only APIs back to Rust 1.31.0. As +//! these are unstable APIs that track the nightly compiler, minor versions of +//! proc-macro2 may make breaking changes to them at any time. +//! +//! ```sh +//! RUSTFLAGS='--cfg procmacro2_semver_exempt' cargo build +//! ``` +//! +//! Note that this must not only be done for your crate, but for any crate that +//! depends on your crate. This infectious nature is intentional, as it serves +//! as a reminder that you are outside of the normal semver guarantees. +//! +//! Semver exempt methods are marked as such in the proc-macro2 documentation. +//! +//! # Thread-Safety +//! +//! Most types in this crate are `!Sync` because the underlying compiler +//! types make use of thread-local memory, meaning they cannot be accessed from +//! a different thread. + +// Proc-macro2 types in rustdoc of other crates get linked to here. +#![doc(html_root_url = "https://docs.rs/proc-macro2/1.0.46")] +#![cfg_attr( + any(proc_macro_span, super_unstable), + feature(proc_macro_span, proc_macro_span_shrink) +)] +#![cfg_attr(super_unstable, feature(proc_macro_def_site))] +#![cfg_attr(doc_cfg, feature(doc_cfg))] +#![allow( + clippy::cast_lossless, + clippy::cast_possible_truncation, + clippy::doc_markdown, + clippy::items_after_statements, + clippy::manual_assert, + clippy::must_use_candidate, + clippy::needless_doctest_main, + clippy::return_self_not_must_use, + clippy::shadow_unrelated, + clippy::trivially_copy_pass_by_ref, + clippy::unnecessary_wraps, + clippy::unused_self, + clippy::used_underscore_binding, + clippy::vec_init_then_push +)] + +#[cfg(all(procmacro2_semver_exempt, wrap_proc_macro, not(super_unstable)))] +compile_error! {"\ + Something is not right. If you've tried to turn on \ + procmacro2_semver_exempt, you need to ensure that it \ + is turned on for the compilation of the proc-macro2 \ + build script as well. +"} + +#[cfg(use_proc_macro)] +extern crate proc_macro; + +mod marker; +mod parse; +mod rcvec; + +#[cfg(wrap_proc_macro)] +mod detection; + +// Public for proc_macro2::fallback::force() and unforce(), but those are quite +// a niche use case so we omit it from rustdoc. +#[doc(hidden)] +pub mod fallback; + +#[cfg(not(wrap_proc_macro))] +use crate::fallback as imp; +#[path = "wrapper.rs"] +#[cfg(wrap_proc_macro)] +mod imp; + +use crate::marker::Marker; +use core::cmp::Ordering; +use core::fmt::{self, Debug, Display}; +use core::hash::{Hash, Hasher}; +use core::iter::FromIterator; +use core::ops::RangeBounds; +use core::str::FromStr; +use std::error::Error; +#[cfg(procmacro2_semver_exempt)] +use std::path::PathBuf; + +/// An abstract stream of tokens, or more concretely a sequence of token trees. +/// +/// This type provides interfaces for iterating over token trees and for +/// collecting token trees into one stream. +/// +/// Token stream is both the input and output of `#[proc_macro]`, +/// `#[proc_macro_attribute]` and `#[proc_macro_derive]` definitions. +#[derive(Clone)] +pub struct TokenStream { + inner: imp::TokenStream, + _marker: Marker, +} + +/// Error returned from `TokenStream::from_str`. +pub struct LexError { + inner: imp::LexError, + _marker: Marker, +} + +impl TokenStream { + fn _new(inner: imp::TokenStream) -> Self { + TokenStream { + inner, + _marker: Marker, + } + } + + fn _new_stable(inner: fallback::TokenStream) -> Self { + TokenStream { + inner: inner.into(), + _marker: Marker, + } + } + + /// Returns an empty `TokenStream` containing no token trees. + pub fn new() -> Self { + TokenStream::_new(imp::TokenStream::new()) + } + + /// Checks if this `TokenStream` is empty. + pub fn is_empty(&self) -> bool { + self.inner.is_empty() + } +} + +/// `TokenStream::default()` returns an empty stream, +/// i.e. this is equivalent with `TokenStream::new()`. +impl Default for TokenStream { + fn default() -> Self { + TokenStream::new() + } +} + +/// Attempts to break the string into tokens and parse those tokens into a token +/// stream. +/// +/// May fail for a number of reasons, for example, if the string contains +/// unbalanced delimiters or characters not existing in the language. +/// +/// NOTE: Some errors may cause panics instead of returning `LexError`. We +/// reserve the right to change these errors into `LexError`s later. +impl FromStr for TokenStream { + type Err = LexError; + + fn from_str(src: &str) -> Result { + let e = src.parse().map_err(|e| LexError { + inner: e, + _marker: Marker, + })?; + Ok(TokenStream::_new(e)) + } +} + +#[cfg(use_proc_macro)] +impl From for TokenStream { + fn from(inner: proc_macro::TokenStream) -> TokenStream { + TokenStream::_new(inner.into()) + } +} + +#[cfg(use_proc_macro)] +impl From for proc_macro::TokenStream { + fn from(inner: TokenStream) -> proc_macro::TokenStream { + inner.inner.into() + } +} + +impl From for TokenStream { + fn from(token: TokenTree) -> Self { + TokenStream::_new(imp::TokenStream::from(token)) + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + self.inner.extend(streams); + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + self.inner + .extend(streams.into_iter().map(|stream| stream.inner)); + } +} + +/// Collects a number of token trees into a single stream. +impl FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + TokenStream::_new(streams.into_iter().collect()) + } +} +impl FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + TokenStream::_new(streams.into_iter().map(|i| i.inner).collect()) + } +} + +/// Prints the token stream as a string that is supposed to be losslessly +/// convertible back into the same token stream (modulo spans), except for +/// possibly `TokenTree::Group`s with `Delimiter::None` delimiters and negative +/// numeric literals. +impl Display for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.inner, f) + } +} + +/// Prints token in a form convenient for debugging. +impl Debug for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +impl LexError { + pub fn span(&self) -> Span { + Span::_new(self.inner.span()) + } +} + +impl Debug for LexError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +impl Display for LexError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.inner, f) + } +} + +impl Error for LexError {} + +/// The source file of a given `Span`. +/// +/// This type is semver exempt and not exposed by default. +#[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] +#[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] +#[derive(Clone, PartialEq, Eq)] +pub struct SourceFile { + inner: imp::SourceFile, + _marker: Marker, +} + +#[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] +impl SourceFile { + fn _new(inner: imp::SourceFile) -> Self { + SourceFile { + inner, + _marker: Marker, + } + } + + /// Get the path to this source file. + /// + /// ### Note + /// + /// If the code span associated with this `SourceFile` was generated by an + /// external macro, this may not be an actual path on the filesystem. Use + /// [`is_real`] to check. + /// + /// Also note that even if `is_real` returns `true`, if + /// `--remap-path-prefix` was passed on the command line, the path as given + /// may not actually be valid. + /// + /// [`is_real`]: #method.is_real + pub fn path(&self) -> PathBuf { + self.inner.path() + } + + /// Returns `true` if this source file is a real source file, and not + /// generated by an external macro's expansion. + pub fn is_real(&self) -> bool { + self.inner.is_real() + } +} + +#[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] +impl Debug for SourceFile { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +/// A line-column pair representing the start or end of a `Span`. +/// +/// This type is semver exempt and not exposed by default. +#[cfg(span_locations)] +#[cfg_attr(doc_cfg, doc(cfg(feature = "span-locations")))] +#[derive(Copy, Clone, Debug, PartialEq, Eq)] +pub struct LineColumn { + /// The 1-indexed line in the source file on which the span starts or ends + /// (inclusive). + pub line: usize, + /// The 0-indexed column (in UTF-8 characters) in the source file on which + /// the span starts or ends (inclusive). + pub column: usize, +} + +#[cfg(span_locations)] +impl Ord for LineColumn { + fn cmp(&self, other: &Self) -> Ordering { + self.line + .cmp(&other.line) + .then(self.column.cmp(&other.column)) + } +} + +#[cfg(span_locations)] +impl PartialOrd for LineColumn { + fn partial_cmp(&self, other: &Self) -> Option { + Some(self.cmp(other)) + } +} + +/// A region of source code, along with macro expansion information. +#[derive(Copy, Clone)] +pub struct Span { + inner: imp::Span, + _marker: Marker, +} + +impl Span { + fn _new(inner: imp::Span) -> Self { + Span { + inner, + _marker: Marker, + } + } + + fn _new_stable(inner: fallback::Span) -> Self { + Span { + inner: inner.into(), + _marker: Marker, + } + } + + /// The span of the invocation of the current procedural macro. + /// + /// Identifiers created with this span will be resolved as if they were + /// written directly at the macro call location (call-site hygiene) and + /// other code at the macro call site will be able to refer to them as well. + pub fn call_site() -> Self { + Span::_new(imp::Span::call_site()) + } + + /// The span located at the invocation of the procedural macro, but with + /// local variables, labels, and `$crate` resolved at the definition site + /// of the macro. This is the same hygiene behavior as `macro_rules`. + /// + /// This function requires Rust 1.45 or later. + #[cfg(not(no_hygiene))] + pub fn mixed_site() -> Self { + Span::_new(imp::Span::mixed_site()) + } + + /// A span that resolves at the macro definition site. + /// + /// This method is semver exempt and not exposed by default. + #[cfg(procmacro2_semver_exempt)] + #[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] + pub fn def_site() -> Self { + Span::_new(imp::Span::def_site()) + } + + /// Creates a new span with the same line/column information as `self` but + /// that resolves symbols as though it were at `other`. + pub fn resolved_at(&self, other: Span) -> Span { + Span::_new(self.inner.resolved_at(other.inner)) + } + + /// Creates a new span with the same name resolution behavior as `self` but + /// with the line/column information of `other`. + pub fn located_at(&self, other: Span) -> Span { + Span::_new(self.inner.located_at(other.inner)) + } + + /// Convert `proc_macro2::Span` to `proc_macro::Span`. + /// + /// This method is available when building with a nightly compiler, or when + /// building with rustc 1.29+ *without* semver exempt features. + /// + /// # Panics + /// + /// Panics if called from outside of a procedural macro. Unlike + /// `proc_macro2::Span`, the `proc_macro::Span` type can only exist within + /// the context of a procedural macro invocation. + #[cfg(wrap_proc_macro)] + pub fn unwrap(self) -> proc_macro::Span { + self.inner.unwrap() + } + + // Soft deprecated. Please use Span::unwrap. + #[cfg(wrap_proc_macro)] + #[doc(hidden)] + pub fn unstable(self) -> proc_macro::Span { + self.unwrap() + } + + /// The original source file into which this span points. + /// + /// This method is semver exempt and not exposed by default. + #[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] + #[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] + pub fn source_file(&self) -> SourceFile { + SourceFile::_new(self.inner.source_file()) + } + + /// Get the starting line/column in the source file for this span. + /// + /// This method requires the `"span-locations"` feature to be enabled. + /// + /// When executing in a procedural macro context, the returned line/column + /// are only meaningful if compiled with a nightly toolchain. The stable + /// toolchain does not have this information available. When executing + /// outside of a procedural macro, such as main.rs or build.rs, the + /// line/column are always meaningful regardless of toolchain. + #[cfg(span_locations)] + #[cfg_attr(doc_cfg, doc(cfg(feature = "span-locations")))] + pub fn start(&self) -> LineColumn { + let imp::LineColumn { line, column } = self.inner.start(); + LineColumn { line, column } + } + + /// Get the ending line/column in the source file for this span. + /// + /// This method requires the `"span-locations"` feature to be enabled. + /// + /// When executing in a procedural macro context, the returned line/column + /// are only meaningful if compiled with a nightly toolchain. The stable + /// toolchain does not have this information available. When executing + /// outside of a procedural macro, such as main.rs or build.rs, the + /// line/column are always meaningful regardless of toolchain. + #[cfg(span_locations)] + #[cfg_attr(doc_cfg, doc(cfg(feature = "span-locations")))] + pub fn end(&self) -> LineColumn { + let imp::LineColumn { line, column } = self.inner.end(); + LineColumn { line, column } + } + + /// Creates an empty span pointing to directly before this span. + /// + /// This method is semver exempt and not exposed by default. + #[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] + #[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] + pub fn before(&self) -> Span { + Span::_new(self.inner.before()) + } + + /// Creates an empty span pointing to directly after this span. + /// + /// This method is semver exempt and not exposed by default. + #[cfg(all(procmacro2_semver_exempt, any(not(wrap_proc_macro), super_unstable)))] + #[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] + pub fn after(&self) -> Span { + Span::_new(self.inner.after()) + } + + /// Create a new span encompassing `self` and `other`. + /// + /// Returns `None` if `self` and `other` are from different files. + /// + /// Warning: the underlying [`proc_macro::Span::join`] method is + /// nightly-only. When called from within a procedural macro not using a + /// nightly compiler, this method will always return `None`. + /// + /// [`proc_macro::Span::join`]: https://doc.rust-lang.org/proc_macro/struct.Span.html#method.join + pub fn join(&self, other: Span) -> Option { + self.inner.join(other.inner).map(Span::_new) + } + + /// Compares two spans to see if they're equal. + /// + /// This method is semver exempt and not exposed by default. + #[cfg(procmacro2_semver_exempt)] + #[cfg_attr(doc_cfg, doc(cfg(procmacro2_semver_exempt)))] + pub fn eq(&self, other: &Span) -> bool { + self.inner.eq(&other.inner) + } +} + +/// Prints a span in a form convenient for debugging. +impl Debug for Span { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +/// A single token or a delimited sequence of token trees (e.g. `[1, (), ..]`). +#[derive(Clone)] +pub enum TokenTree { + /// A token stream surrounded by bracket delimiters. + Group(Group), + /// An identifier. + Ident(Ident), + /// A single punctuation character (`+`, `,`, `$`, etc.). + Punct(Punct), + /// A literal character (`'a'`), string (`"hello"`), number (`2.3`), etc. + Literal(Literal), +} + +impl TokenTree { + /// Returns the span of this tree, delegating to the `span` method of + /// the contained token or a delimited stream. + pub fn span(&self) -> Span { + match self { + TokenTree::Group(t) => t.span(), + TokenTree::Ident(t) => t.span(), + TokenTree::Punct(t) => t.span(), + TokenTree::Literal(t) => t.span(), + } + } + + /// Configures the span for *only this token*. + /// + /// Note that if this token is a `Group` then this method will not configure + /// the span of each of the internal tokens, this will simply delegate to + /// the `set_span` method of each variant. + pub fn set_span(&mut self, span: Span) { + match self { + TokenTree::Group(t) => t.set_span(span), + TokenTree::Ident(t) => t.set_span(span), + TokenTree::Punct(t) => t.set_span(span), + TokenTree::Literal(t) => t.set_span(span), + } + } +} + +impl From for TokenTree { + fn from(g: Group) -> TokenTree { + TokenTree::Group(g) + } +} + +impl From for TokenTree { + fn from(g: Ident) -> TokenTree { + TokenTree::Ident(g) + } +} + +impl From for TokenTree { + fn from(g: Punct) -> TokenTree { + TokenTree::Punct(g) + } +} + +impl From for TokenTree { + fn from(g: Literal) -> TokenTree { + TokenTree::Literal(g) + } +} + +/// Prints the token tree as a string that is supposed to be losslessly +/// convertible back into the same token tree (modulo spans), except for +/// possibly `TokenTree::Group`s with `Delimiter::None` delimiters and negative +/// numeric literals. +impl Display for TokenTree { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + TokenTree::Group(t) => Display::fmt(t, f), + TokenTree::Ident(t) => Display::fmt(t, f), + TokenTree::Punct(t) => Display::fmt(t, f), + TokenTree::Literal(t) => Display::fmt(t, f), + } + } +} + +/// Prints token tree in a form convenient for debugging. +impl Debug for TokenTree { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + // Each of these has the name in the struct type in the derived debug, + // so don't bother with an extra layer of indirection + match self { + TokenTree::Group(t) => Debug::fmt(t, f), + TokenTree::Ident(t) => { + let mut debug = f.debug_struct("Ident"); + debug.field("sym", &format_args!("{}", t)); + imp::debug_span_field_if_nontrivial(&mut debug, t.span().inner); + debug.finish() + } + TokenTree::Punct(t) => Debug::fmt(t, f), + TokenTree::Literal(t) => Debug::fmt(t, f), + } + } +} + +/// A delimited token stream. +/// +/// A `Group` internally contains a `TokenStream` which is surrounded by +/// `Delimiter`s. +#[derive(Clone)] +pub struct Group { + inner: imp::Group, +} + +/// Describes how a sequence of token trees is delimited. +#[derive(Copy, Clone, Debug, Eq, PartialEq)] +pub enum Delimiter { + /// `( ... )` + Parenthesis, + /// `{ ... }` + Brace, + /// `[ ... ]` + Bracket, + /// `Ø ... Ø` + /// + /// An implicit delimiter, that may, for example, appear around tokens + /// coming from a "macro variable" `$var`. It is important to preserve + /// operator priorities in cases like `$var * 3` where `$var` is `1 + 2`. + /// Implicit delimiters may not survive roundtrip of a token stream through + /// a string. + None, +} + +impl Group { + fn _new(inner: imp::Group) -> Self { + Group { inner } + } + + fn _new_stable(inner: fallback::Group) -> Self { + Group { + inner: inner.into(), + } + } + + /// Creates a new `Group` with the given delimiter and token stream. + /// + /// This constructor will set the span for this group to + /// `Span::call_site()`. To change the span you can use the `set_span` + /// method below. + pub fn new(delimiter: Delimiter, stream: TokenStream) -> Self { + Group { + inner: imp::Group::new(delimiter, stream.inner), + } + } + + /// Returns the delimiter of this `Group` + pub fn delimiter(&self) -> Delimiter { + self.inner.delimiter() + } + + /// Returns the `TokenStream` of tokens that are delimited in this `Group`. + /// + /// Note that the returned token stream does not include the delimiter + /// returned above. + pub fn stream(&self) -> TokenStream { + TokenStream::_new(self.inner.stream()) + } + + /// Returns the span for the delimiters of this token stream, spanning the + /// entire `Group`. + /// + /// ```text + /// pub fn span(&self) -> Span { + /// ^^^^^^^ + /// ``` + pub fn span(&self) -> Span { + Span::_new(self.inner.span()) + } + + /// Returns the span pointing to the opening delimiter of this group. + /// + /// ```text + /// pub fn span_open(&self) -> Span { + /// ^ + /// ``` + pub fn span_open(&self) -> Span { + Span::_new(self.inner.span_open()) + } + + /// Returns the span pointing to the closing delimiter of this group. + /// + /// ```text + /// pub fn span_close(&self) -> Span { + /// ^ + /// ``` + pub fn span_close(&self) -> Span { + Span::_new(self.inner.span_close()) + } + + /// Configures the span for this `Group`'s delimiters, but not its internal + /// tokens. + /// + /// This method will **not** set the span of all the internal tokens spanned + /// by this group, but rather it will only set the span of the delimiter + /// tokens at the level of the `Group`. + pub fn set_span(&mut self, span: Span) { + self.inner.set_span(span.inner); + } +} + +/// Prints the group as a string that should be losslessly convertible back +/// into the same group (modulo spans), except for possibly `TokenTree::Group`s +/// with `Delimiter::None` delimiters. +impl Display for Group { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.inner, formatter) + } +} + +impl Debug for Group { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, formatter) + } +} + +/// A `Punct` is a single punctuation character like `+`, `-` or `#`. +/// +/// Multicharacter operators like `+=` are represented as two instances of +/// `Punct` with different forms of `Spacing` returned. +#[derive(Clone)] +pub struct Punct { + ch: char, + spacing: Spacing, + span: Span, +} + +/// Whether a `Punct` is followed immediately by another `Punct` or followed by +/// another token or whitespace. +#[derive(Copy, Clone, Debug, Eq, PartialEq)] +pub enum Spacing { + /// E.g. `+` is `Alone` in `+ =`, `+ident` or `+()`. + Alone, + /// E.g. `+` is `Joint` in `+=` or `'` is `Joint` in `'#`. + /// + /// Additionally, single quote `'` can join with identifiers to form + /// lifetimes `'ident`. + Joint, +} + +impl Punct { + /// Creates a new `Punct` from the given character and spacing. + /// + /// The `ch` argument must be a valid punctuation character permitted by the + /// language, otherwise the function will panic. + /// + /// The returned `Punct` will have the default span of `Span::call_site()` + /// which can be further configured with the `set_span` method below. + pub fn new(ch: char, spacing: Spacing) -> Self { + Punct { + ch, + spacing, + span: Span::call_site(), + } + } + + /// Returns the value of this punctuation character as `char`. + pub fn as_char(&self) -> char { + self.ch + } + + /// Returns the spacing of this punctuation character, indicating whether + /// it's immediately followed by another `Punct` in the token stream, so + /// they can potentially be combined into a multicharacter operator + /// (`Joint`), or it's followed by some other token or whitespace (`Alone`) + /// so the operator has certainly ended. + pub fn spacing(&self) -> Spacing { + self.spacing + } + + /// Returns the span for this punctuation character. + pub fn span(&self) -> Span { + self.span + } + + /// Configure the span for this punctuation character. + pub fn set_span(&mut self, span: Span) { + self.span = span; + } +} + +/// Prints the punctuation character as a string that should be losslessly +/// convertible back into the same character. +impl Display for Punct { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.ch, f) + } +} + +impl Debug for Punct { + fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + let mut debug = fmt.debug_struct("Punct"); + debug.field("char", &self.ch); + debug.field("spacing", &self.spacing); + imp::debug_span_field_if_nontrivial(&mut debug, self.span.inner); + debug.finish() + } +} + +/// A word of Rust code, which may be a keyword or legal variable name. +/// +/// An identifier consists of at least one Unicode code point, the first of +/// which has the XID_Start property and the rest of which have the XID_Continue +/// property. +/// +/// - The empty string is not an identifier. Use `Option`. +/// - A lifetime is not an identifier. Use `syn::Lifetime` instead. +/// +/// An identifier constructed with `Ident::new` is permitted to be a Rust +/// keyword, though parsing one through its [`Parse`] implementation rejects +/// Rust keywords. Use `input.call(Ident::parse_any)` when parsing to match the +/// behaviour of `Ident::new`. +/// +/// [`Parse`]: https://docs.rs/syn/1.0/syn/parse/trait.Parse.html +/// +/// # Examples +/// +/// A new ident can be created from a string using the `Ident::new` function. +/// A span must be provided explicitly which governs the name resolution +/// behavior of the resulting identifier. +/// +/// ``` +/// use proc_macro2::{Ident, Span}; +/// +/// fn main() { +/// let call_ident = Ident::new("calligraphy", Span::call_site()); +/// +/// println!("{}", call_ident); +/// } +/// ``` +/// +/// An ident can be interpolated into a token stream using the `quote!` macro. +/// +/// ``` +/// use proc_macro2::{Ident, Span}; +/// use quote::quote; +/// +/// fn main() { +/// let ident = Ident::new("demo", Span::call_site()); +/// +/// // Create a variable binding whose name is this ident. +/// let expanded = quote! { let #ident = 10; }; +/// +/// // Create a variable binding with a slightly different name. +/// let temp_ident = Ident::new(&format!("new_{}", ident), Span::call_site()); +/// let expanded = quote! { let #temp_ident = 10; }; +/// } +/// ``` +/// +/// A string representation of the ident is available through the `to_string()` +/// method. +/// +/// ``` +/// # use proc_macro2::{Ident, Span}; +/// # +/// # let ident = Ident::new("another_identifier", Span::call_site()); +/// # +/// // Examine the ident as a string. +/// let ident_string = ident.to_string(); +/// if ident_string.len() > 60 { +/// println!("Very long identifier: {}", ident_string) +/// } +/// ``` +#[derive(Clone)] +pub struct Ident { + inner: imp::Ident, + _marker: Marker, +} + +impl Ident { + fn _new(inner: imp::Ident) -> Self { + Ident { + inner, + _marker: Marker, + } + } + + /// Creates a new `Ident` with the given `string` as well as the specified + /// `span`. + /// + /// The `string` argument must be a valid identifier permitted by the + /// language, otherwise the function will panic. + /// + /// Note that `span`, currently in rustc, configures the hygiene information + /// for this identifier. + /// + /// As of this time `Span::call_site()` explicitly opts-in to "call-site" + /// hygiene meaning that identifiers created with this span will be resolved + /// as if they were written directly at the location of the macro call, and + /// other code at the macro call site will be able to refer to them as well. + /// + /// Later spans like `Span::def_site()` will allow to opt-in to + /// "definition-site" hygiene meaning that identifiers created with this + /// span will be resolved at the location of the macro definition and other + /// code at the macro call site will not be able to refer to them. + /// + /// Due to the current importance of hygiene this constructor, unlike other + /// tokens, requires a `Span` to be specified at construction. + /// + /// # Panics + /// + /// Panics if the input string is neither a keyword nor a legal variable + /// name. If you are not sure whether the string contains an identifier and + /// need to handle an error case, use + /// syn::parse_str::<Ident> + /// rather than `Ident::new`. + pub fn new(string: &str, span: Span) -> Self { + Ident::_new(imp::Ident::new(string, span.inner)) + } + + /// Same as `Ident::new`, but creates a raw identifier (`r#ident`). The + /// `string` argument must be a valid identifier permitted by the language + /// (including keywords, e.g. `fn`). Keywords which are usable in path + /// segments (e.g. `self`, `super`) are not supported, and will cause a + /// panic. + pub fn new_raw(string: &str, span: Span) -> Self { + Ident::_new_raw(string, span) + } + + fn _new_raw(string: &str, span: Span) -> Self { + Ident::_new(imp::Ident::new_raw(string, span.inner)) + } + + /// Returns the span of this `Ident`. + pub fn span(&self) -> Span { + Span::_new(self.inner.span()) + } + + /// Configures the span of this `Ident`, possibly changing its hygiene + /// context. + pub fn set_span(&mut self, span: Span) { + self.inner.set_span(span.inner); + } +} + +impl PartialEq for Ident { + fn eq(&self, other: &Ident) -> bool { + self.inner == other.inner + } +} + +impl PartialEq for Ident +where + T: ?Sized + AsRef, +{ + fn eq(&self, other: &T) -> bool { + self.inner == other + } +} + +impl Eq for Ident {} + +impl PartialOrd for Ident { + fn partial_cmp(&self, other: &Ident) -> Option { + Some(self.cmp(other)) + } +} + +impl Ord for Ident { + fn cmp(&self, other: &Ident) -> Ordering { + self.to_string().cmp(&other.to_string()) + } +} + +impl Hash for Ident { + fn hash(&self, hasher: &mut H) { + self.to_string().hash(hasher); + } +} + +/// Prints the identifier as a string that should be losslessly convertible back +/// into the same identifier. +impl Display for Ident { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.inner, f) + } +} + +impl Debug for Ident { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +/// A literal string (`"hello"`), byte string (`b"hello"`), character (`'a'`), +/// byte character (`b'a'`), an integer or floating point number with or without +/// a suffix (`1`, `1u8`, `2.3`, `2.3f32`). +/// +/// Boolean literals like `true` and `false` do not belong here, they are +/// `Ident`s. +#[derive(Clone)] +pub struct Literal { + inner: imp::Literal, + _marker: Marker, +} + +macro_rules! suffixed_int_literals { + ($($name:ident => $kind:ident,)*) => ($( + /// Creates a new suffixed integer literal with the specified value. + /// + /// This function will create an integer like `1u32` where the integer + /// value specified is the first part of the token and the integral is + /// also suffixed at the end. Literals created from negative numbers may + /// not survive roundtrips through `TokenStream` or strings and may be + /// broken into two tokens (`-` and positive literal). + /// + /// Literals created through this method have the `Span::call_site()` + /// span by default, which can be configured with the `set_span` method + /// below. + pub fn $name(n: $kind) -> Literal { + Literal::_new(imp::Literal::$name(n)) + } + )*) +} + +macro_rules! unsuffixed_int_literals { + ($($name:ident => $kind:ident,)*) => ($( + /// Creates a new unsuffixed integer literal with the specified value. + /// + /// This function will create an integer like `1` where the integer + /// value specified is the first part of the token. No suffix is + /// specified on this token, meaning that invocations like + /// `Literal::i8_unsuffixed(1)` are equivalent to + /// `Literal::u32_unsuffixed(1)`. Literals created from negative numbers + /// may not survive roundtrips through `TokenStream` or strings and may + /// be broken into two tokens (`-` and positive literal). + /// + /// Literals created through this method have the `Span::call_site()` + /// span by default, which can be configured with the `set_span` method + /// below. + pub fn $name(n: $kind) -> Literal { + Literal::_new(imp::Literal::$name(n)) + } + )*) +} + +impl Literal { + fn _new(inner: imp::Literal) -> Self { + Literal { + inner, + _marker: Marker, + } + } + + fn _new_stable(inner: fallback::Literal) -> Self { + Literal { + inner: inner.into(), + _marker: Marker, + } + } + + suffixed_int_literals! { + u8_suffixed => u8, + u16_suffixed => u16, + u32_suffixed => u32, + u64_suffixed => u64, + u128_suffixed => u128, + usize_suffixed => usize, + i8_suffixed => i8, + i16_suffixed => i16, + i32_suffixed => i32, + i64_suffixed => i64, + i128_suffixed => i128, + isize_suffixed => isize, + } + + unsuffixed_int_literals! { + u8_unsuffixed => u8, + u16_unsuffixed => u16, + u32_unsuffixed => u32, + u64_unsuffixed => u64, + u128_unsuffixed => u128, + usize_unsuffixed => usize, + i8_unsuffixed => i8, + i16_unsuffixed => i16, + i32_unsuffixed => i32, + i64_unsuffixed => i64, + i128_unsuffixed => i128, + isize_unsuffixed => isize, + } + + /// Creates a new unsuffixed floating-point literal. + /// + /// This constructor is similar to those like `Literal::i8_unsuffixed` where + /// the float's value is emitted directly into the token but no suffix is + /// used, so it may be inferred to be a `f64` later in the compiler. + /// Literals created from negative numbers may not survive round-trips + /// through `TokenStream` or strings and may be broken into two tokens (`-` + /// and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for example + /// if it is infinity or NaN this function will panic. + pub fn f64_unsuffixed(f: f64) -> Literal { + assert!(f.is_finite()); + Literal::_new(imp::Literal::f64_unsuffixed(f)) + } + + /// Creates a new suffixed floating-point literal. + /// + /// This constructor will create a literal like `1.0f64` where the value + /// specified is the preceding part of the token and `f64` is the suffix of + /// the token. This token will always be inferred to be an `f64` in the + /// compiler. Literals created from negative numbers may not survive + /// round-trips through `TokenStream` or strings and may be broken into two + /// tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for example + /// if it is infinity or NaN this function will panic. + pub fn f64_suffixed(f: f64) -> Literal { + assert!(f.is_finite()); + Literal::_new(imp::Literal::f64_suffixed(f)) + } + + /// Creates a new unsuffixed floating-point literal. + /// + /// This constructor is similar to those like `Literal::i8_unsuffixed` where + /// the float's value is emitted directly into the token but no suffix is + /// used, so it may be inferred to be a `f64` later in the compiler. + /// Literals created from negative numbers may not survive round-trips + /// through `TokenStream` or strings and may be broken into two tokens (`-` + /// and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for example + /// if it is infinity or NaN this function will panic. + pub fn f32_unsuffixed(f: f32) -> Literal { + assert!(f.is_finite()); + Literal::_new(imp::Literal::f32_unsuffixed(f)) + } + + /// Creates a new suffixed floating-point literal. + /// + /// This constructor will create a literal like `1.0f32` where the value + /// specified is the preceding part of the token and `f32` is the suffix of + /// the token. This token will always be inferred to be an `f32` in the + /// compiler. Literals created from negative numbers may not survive + /// round-trips through `TokenStream` or strings and may be broken into two + /// tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for example + /// if it is infinity or NaN this function will panic. + pub fn f32_suffixed(f: f32) -> Literal { + assert!(f.is_finite()); + Literal::_new(imp::Literal::f32_suffixed(f)) + } + + /// String literal. + pub fn string(string: &str) -> Literal { + Literal::_new(imp::Literal::string(string)) + } + + /// Character literal. + pub fn character(ch: char) -> Literal { + Literal::_new(imp::Literal::character(ch)) + } + + /// Byte string literal. + pub fn byte_string(s: &[u8]) -> Literal { + Literal::_new(imp::Literal::byte_string(s)) + } + + /// Returns the span encompassing this literal. + pub fn span(&self) -> Span { + Span::_new(self.inner.span()) + } + + /// Configures the span associated for this literal. + pub fn set_span(&mut self, span: Span) { + self.inner.set_span(span.inner); + } + + /// Returns a `Span` that is a subset of `self.span()` containing only + /// the source bytes in range `range`. Returns `None` if the would-be + /// trimmed span is outside the bounds of `self`. + /// + /// Warning: the underlying [`proc_macro::Literal::subspan`] method is + /// nightly-only. When called from within a procedural macro not using a + /// nightly compiler, this method will always return `None`. + /// + /// [`proc_macro::Literal::subspan`]: https://doc.rust-lang.org/proc_macro/struct.Literal.html#method.subspan + pub fn subspan>(&self, range: R) -> Option { + self.inner.subspan(range).map(Span::_new) + } + + // Intended for the `quote!` macro to use when constructing a proc-macro2 + // token out of a macro_rules $:literal token, which is already known to be + // a valid literal. This avoids reparsing/validating the literal's string + // representation. This is not public API other than for quote. + #[doc(hidden)] + pub unsafe fn from_str_unchecked(repr: &str) -> Self { + Literal::_new(imp::Literal::from_str_unchecked(repr)) + } +} + +impl FromStr for Literal { + type Err = LexError; + + fn from_str(repr: &str) -> Result { + repr.parse().map(Literal::_new).map_err(|inner| LexError { + inner, + _marker: Marker, + }) + } +} + +impl Debug for Literal { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.inner, f) + } +} + +impl Display for Literal { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.inner, f) + } +} + +/// Public implementation details for the `TokenStream` type, such as iterators. +pub mod token_stream { + use crate::marker::Marker; + use crate::{imp, TokenTree}; + use core::fmt::{self, Debug}; + + pub use crate::TokenStream; + + /// An iterator over `TokenStream`'s `TokenTree`s. + /// + /// The iteration is "shallow", e.g. the iterator doesn't recurse into + /// delimited groups, and returns whole groups as token trees. + #[derive(Clone)] + pub struct IntoIter { + inner: imp::TokenTreeIter, + _marker: Marker, + } + + impl Iterator for IntoIter { + type Item = TokenTree; + + fn next(&mut self) -> Option { + self.inner.next() + } + + fn size_hint(&self) -> (usize, Option) { + self.inner.size_hint() + } + } + + impl Debug for IntoIter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str("TokenStream ")?; + f.debug_list().entries(self.clone()).finish() + } + } + + impl IntoIterator for TokenStream { + type Item = TokenTree; + type IntoIter = IntoIter; + + fn into_iter(self) -> IntoIter { + IntoIter { + inner: self.inner.into_iter(), + _marker: Marker, + } + } + } +} diff --git a/rust/proc-macro2/marker.rs b/rust/proc-macro2/marker.rs new file mode 100644 index 00000000000000..eea17649a9b92c --- /dev/null +++ b/rust/proc-macro2/marker.rs @@ -0,0 +1,20 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use core::marker::PhantomData; +use std::panic::{RefUnwindSafe, UnwindSafe}; +use std::rc::Rc; + +// Zero sized marker with the correct set of autotrait impls we want all proc +// macro types to have. +pub(crate) type Marker = PhantomData; + +pub(crate) use self::value::*; + +mod value { + pub(crate) use core::marker::PhantomData as Marker; +} + +pub(crate) struct ProcMacroAutoTraits(Rc<()>); + +impl UnwindSafe for ProcMacroAutoTraits {} +impl RefUnwindSafe for ProcMacroAutoTraits {} diff --git a/rust/proc-macro2/parse.rs b/rust/proc-macro2/parse.rs new file mode 100644 index 00000000000000..d1721a207da47b --- /dev/null +++ b/rust/proc-macro2/parse.rs @@ -0,0 +1,874 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::fallback::{ + is_ident_continue, is_ident_start, Group, LexError, Literal, Span, TokenStream, + TokenStreamBuilder, +}; +use crate::{Delimiter, Punct, Spacing, TokenTree}; +use core::char; +use core::str::{Bytes, CharIndices, Chars}; + +#[derive(Copy, Clone, Eq, PartialEq)] +pub(crate) struct Cursor<'a> { + pub rest: &'a str, + #[cfg(span_locations)] + pub off: u32, +} + +impl<'a> Cursor<'a> { + pub fn advance(&self, bytes: usize) -> Cursor<'a> { + let (_front, rest) = self.rest.split_at(bytes); + Cursor { + rest, + #[cfg(span_locations)] + off: self.off + _front.chars().count() as u32, + } + } + + pub fn starts_with(&self, s: &str) -> bool { + self.rest.starts_with(s) + } + + fn is_empty(&self) -> bool { + self.rest.is_empty() + } + + fn len(&self) -> usize { + self.rest.len() + } + + fn as_bytes(&self) -> &'a [u8] { + self.rest.as_bytes() + } + + fn bytes(&self) -> Bytes<'a> { + self.rest.bytes() + } + + fn chars(&self) -> Chars<'a> { + self.rest.chars() + } + + fn char_indices(&self) -> CharIndices<'a> { + self.rest.char_indices() + } + + fn parse(&self, tag: &str) -> Result, Reject> { + if self.starts_with(tag) { + Ok(self.advance(tag.len())) + } else { + Err(Reject) + } + } +} + +pub(crate) struct Reject; +type PResult<'a, O> = Result<(Cursor<'a>, O), Reject>; + +fn skip_whitespace(input: Cursor) -> Cursor { + let mut s = input; + + while !s.is_empty() { + let byte = s.as_bytes()[0]; + if byte == b'/' { + if s.starts_with("//") + && (!s.starts_with("///") || s.starts_with("////")) + && !s.starts_with("//!") + { + let (cursor, _) = take_until_newline_or_eof(s); + s = cursor; + continue; + } else if s.starts_with("/**/") { + s = s.advance(4); + continue; + } else if s.starts_with("/*") + && (!s.starts_with("/**") || s.starts_with("/***")) + && !s.starts_with("/*!") + { + match block_comment(s) { + Ok((rest, _)) => { + s = rest; + continue; + } + Err(Reject) => return s, + } + } + } + match byte { + b' ' | 0x09..=0x0d => { + s = s.advance(1); + continue; + } + b if b <= 0x7f => {} + _ => { + let ch = s.chars().next().unwrap(); + if is_whitespace(ch) { + s = s.advance(ch.len_utf8()); + continue; + } + } + } + return s; + } + s +} + +fn block_comment(input: Cursor) -> PResult<&str> { + if !input.starts_with("/*") { + return Err(Reject); + } + + let mut depth = 0; + let bytes = input.as_bytes(); + let mut i = 0; + let upper = bytes.len() - 1; + + while i < upper { + if bytes[i] == b'/' && bytes[i + 1] == b'*' { + depth += 1; + i += 1; // eat '*' + } else if bytes[i] == b'*' && bytes[i + 1] == b'/' { + depth -= 1; + if depth == 0 { + return Ok((input.advance(i + 2), &input.rest[..i + 2])); + } + i += 1; // eat '/' + } + i += 1; + } + + Err(Reject) +} + +fn is_whitespace(ch: char) -> bool { + // Rust treats left-to-right mark and right-to-left mark as whitespace + ch.is_whitespace() || ch == '\u{200e}' || ch == '\u{200f}' +} + +fn word_break(input: Cursor) -> Result { + match input.chars().next() { + Some(ch) if is_ident_continue(ch) => Err(Reject), + Some(_) | None => Ok(input), + } +} + +pub(crate) fn token_stream(mut input: Cursor) -> Result { + let mut trees = TokenStreamBuilder::new(); + let mut stack = Vec::new(); + + loop { + input = skip_whitespace(input); + + if let Ok((rest, ())) = doc_comment(input, &mut trees) { + input = rest; + continue; + } + + #[cfg(span_locations)] + let lo = input.off; + + let first = match input.bytes().next() { + Some(first) => first, + None => match stack.last() { + None => return Ok(trees.build()), + #[cfg(span_locations)] + Some((lo, _frame)) => { + return Err(LexError { + span: Span { lo: *lo, hi: *lo }, + }) + } + #[cfg(not(span_locations))] + Some(_frame) => return Err(LexError { span: Span {} }), + }, + }; + + if let Some(open_delimiter) = match first { + b'(' => Some(Delimiter::Parenthesis), + b'[' => Some(Delimiter::Bracket), + b'{' => Some(Delimiter::Brace), + _ => None, + } { + input = input.advance(1); + let frame = (open_delimiter, trees); + #[cfg(span_locations)] + let frame = (lo, frame); + stack.push(frame); + trees = TokenStreamBuilder::new(); + } else if let Some(close_delimiter) = match first { + b')' => Some(Delimiter::Parenthesis), + b']' => Some(Delimiter::Bracket), + b'}' => Some(Delimiter::Brace), + _ => None, + } { + let frame = match stack.pop() { + Some(frame) => frame, + None => return Err(lex_error(input)), + }; + #[cfg(span_locations)] + let (lo, frame) = frame; + let (open_delimiter, outer) = frame; + if open_delimiter != close_delimiter { + return Err(lex_error(input)); + } + input = input.advance(1); + let mut g = Group::new(open_delimiter, trees.build()); + g.set_span(Span { + #[cfg(span_locations)] + lo, + #[cfg(span_locations)] + hi: input.off, + }); + trees = outer; + trees.push_token_from_parser(TokenTree::Group(crate::Group::_new_stable(g))); + } else { + let (rest, mut tt) = match leaf_token(input) { + Ok((rest, tt)) => (rest, tt), + Err(Reject) => return Err(lex_error(input)), + }; + tt.set_span(crate::Span::_new_stable(Span { + #[cfg(span_locations)] + lo, + #[cfg(span_locations)] + hi: rest.off, + })); + trees.push_token_from_parser(tt); + input = rest; + } + } +} + +fn lex_error(cursor: Cursor) -> LexError { + #[cfg(not(span_locations))] + let _ = cursor; + LexError { + span: Span { + #[cfg(span_locations)] + lo: cursor.off, + #[cfg(span_locations)] + hi: cursor.off, + }, + } +} + +fn leaf_token(input: Cursor) -> PResult { + if let Ok((input, l)) = literal(input) { + // must be parsed before ident + Ok((input, TokenTree::Literal(crate::Literal::_new_stable(l)))) + } else if let Ok((input, p)) = punct(input) { + Ok((input, TokenTree::Punct(p))) + } else if let Ok((input, i)) = ident(input) { + Ok((input, TokenTree::Ident(i))) + } else { + Err(Reject) + } +} + +fn ident(input: Cursor) -> PResult { + if ["r\"", "r#\"", "r##", "b\"", "b\'", "br\"", "br#"] + .iter() + .any(|prefix| input.starts_with(prefix)) + { + Err(Reject) + } else { + ident_any(input) + } +} + +fn ident_any(input: Cursor) -> PResult { + let raw = input.starts_with("r#"); + let rest = input.advance((raw as usize) << 1); + + let (rest, sym) = ident_not_raw(rest)?; + + if !raw { + let ident = crate::Ident::new(sym, crate::Span::call_site()); + return Ok((rest, ident)); + } + + match sym { + "_" | "super" | "self" | "Self" | "crate" => return Err(Reject), + _ => {} + } + + let ident = crate::Ident::_new_raw(sym, crate::Span::call_site()); + Ok((rest, ident)) +} + +fn ident_not_raw(input: Cursor) -> PResult<&str> { + let mut chars = input.char_indices(); + + match chars.next() { + Some((_, ch)) if is_ident_start(ch) => {} + _ => return Err(Reject), + } + + let mut end = input.len(); + for (i, ch) in chars { + if !is_ident_continue(ch) { + end = i; + break; + } + } + + Ok((input.advance(end), &input.rest[..end])) +} + +pub(crate) fn literal(input: Cursor) -> PResult { + let rest = literal_nocapture(input)?; + let end = input.len() - rest.len(); + Ok((rest, Literal::_new(input.rest[..end].to_string()))) +} + +fn literal_nocapture(input: Cursor) -> Result { + if let Ok(ok) = string(input) { + Ok(ok) + } else if let Ok(ok) = byte_string(input) { + Ok(ok) + } else if let Ok(ok) = byte(input) { + Ok(ok) + } else if let Ok(ok) = character(input) { + Ok(ok) + } else if let Ok(ok) = float(input) { + Ok(ok) + } else if let Ok(ok) = int(input) { + Ok(ok) + } else { + Err(Reject) + } +} + +fn literal_suffix(input: Cursor) -> Cursor { + match ident_not_raw(input) { + Ok((input, _)) => input, + Err(Reject) => input, + } +} + +fn string(input: Cursor) -> Result { + if let Ok(input) = input.parse("\"") { + cooked_string(input) + } else if let Ok(input) = input.parse("r") { + raw_string(input) + } else { + Err(Reject) + } +} + +fn cooked_string(input: Cursor) -> Result { + let mut chars = input.char_indices().peekable(); + + while let Some((i, ch)) = chars.next() { + match ch { + '"' => { + let input = input.advance(i + 1); + return Ok(literal_suffix(input)); + } + '\r' => match chars.next() { + Some((_, '\n')) => {} + _ => break, + }, + '\\' => match chars.next() { + Some((_, 'x')) => { + if !backslash_x_char(&mut chars) { + break; + } + } + Some((_, 'n')) | Some((_, 'r')) | Some((_, 't')) | Some((_, '\\')) + | Some((_, '\'')) | Some((_, '"')) | Some((_, '0')) => {} + Some((_, 'u')) => { + if !backslash_u(&mut chars) { + break; + } + } + Some((_, ch @ '\n')) | Some((_, ch @ '\r')) => { + let mut last = ch; + loop { + if last == '\r' && chars.next().map_or(true, |(_, ch)| ch != '\n') { + return Err(Reject); + } + match chars.peek() { + Some((_, ch)) if ch.is_whitespace() => { + last = *ch; + chars.next(); + } + _ => break, + } + } + } + _ => break, + }, + _ch => {} + } + } + Err(Reject) +} + +fn byte_string(input: Cursor) -> Result { + if let Ok(input) = input.parse("b\"") { + cooked_byte_string(input) + } else if let Ok(input) = input.parse("br") { + raw_string(input) + } else { + Err(Reject) + } +} + +fn cooked_byte_string(mut input: Cursor) -> Result { + let mut bytes = input.bytes().enumerate(); + while let Some((offset, b)) = bytes.next() { + match b { + b'"' => { + let input = input.advance(offset + 1); + return Ok(literal_suffix(input)); + } + b'\r' => match bytes.next() { + Some((_, b'\n')) => {} + _ => break, + }, + b'\\' => match bytes.next() { + Some((_, b'x')) => { + if !backslash_x_byte(&mut bytes) { + break; + } + } + Some((_, b'n')) | Some((_, b'r')) | Some((_, b't')) | Some((_, b'\\')) + | Some((_, b'0')) | Some((_, b'\'')) | Some((_, b'"')) => {} + Some((newline, b @ b'\n')) | Some((newline, b @ b'\r')) => { + let mut last = b as char; + let rest = input.advance(newline + 1); + let mut chars = rest.char_indices(); + loop { + if last == '\r' && chars.next().map_or(true, |(_, ch)| ch != '\n') { + return Err(Reject); + } + match chars.next() { + Some((_, ch)) if ch.is_whitespace() => last = ch, + Some((offset, _)) => { + input = rest.advance(offset); + bytes = input.bytes().enumerate(); + break; + } + None => return Err(Reject), + } + } + } + _ => break, + }, + b if b < 0x80 => {} + _ => break, + } + } + Err(Reject) +} + +fn raw_string(input: Cursor) -> Result { + let mut chars = input.char_indices(); + let mut n = 0; + for (i, ch) in &mut chars { + match ch { + '"' => { + n = i; + break; + } + '#' => {} + _ => return Err(Reject), + } + } + while let Some((i, ch)) = chars.next() { + match ch { + '"' if input.rest[i + 1..].starts_with(&input.rest[..n]) => { + let rest = input.advance(i + 1 + n); + return Ok(literal_suffix(rest)); + } + '\r' => match chars.next() { + Some((_, '\n')) => {} + _ => break, + }, + _ => {} + } + } + Err(Reject) +} + +fn byte(input: Cursor) -> Result { + let input = input.parse("b'")?; + let mut bytes = input.bytes().enumerate(); + let ok = match bytes.next().map(|(_, b)| b) { + Some(b'\\') => match bytes.next().map(|(_, b)| b) { + Some(b'x') => backslash_x_byte(&mut bytes), + Some(b'n') | Some(b'r') | Some(b't') | Some(b'\\') | Some(b'0') | Some(b'\'') + | Some(b'"') => true, + _ => false, + }, + b => b.is_some(), + }; + if !ok { + return Err(Reject); + } + let (offset, _) = bytes.next().ok_or(Reject)?; + if !input.chars().as_str().is_char_boundary(offset) { + return Err(Reject); + } + let input = input.advance(offset).parse("'")?; + Ok(literal_suffix(input)) +} + +fn character(input: Cursor) -> Result { + let input = input.parse("'")?; + let mut chars = input.char_indices(); + let ok = match chars.next().map(|(_, ch)| ch) { + Some('\\') => match chars.next().map(|(_, ch)| ch) { + Some('x') => backslash_x_char(&mut chars), + Some('u') => backslash_u(&mut chars), + Some('n') | Some('r') | Some('t') | Some('\\') | Some('0') | Some('\'') | Some('"') => { + true + } + _ => false, + }, + ch => ch.is_some(), + }; + if !ok { + return Err(Reject); + } + let (idx, _) = chars.next().ok_or(Reject)?; + let input = input.advance(idx).parse("'")?; + Ok(literal_suffix(input)) +} + +macro_rules! next_ch { + ($chars:ident @ $pat:pat $(| $rest:pat)*) => { + match $chars.next() { + Some((_, ch)) => match ch { + $pat $(| $rest)* => ch, + _ => return false, + }, + None => return false, + } + }; +} + +fn backslash_x_char(chars: &mut I) -> bool +where + I: Iterator, +{ + next_ch!(chars @ '0'..='7'); + next_ch!(chars @ '0'..='9' | 'a'..='f' | 'A'..='F'); + true +} + +fn backslash_x_byte(chars: &mut I) -> bool +where + I: Iterator, +{ + next_ch!(chars @ b'0'..=b'9' | b'a'..=b'f' | b'A'..=b'F'); + next_ch!(chars @ b'0'..=b'9' | b'a'..=b'f' | b'A'..=b'F'); + true +} + +fn backslash_u(chars: &mut I) -> bool +where + I: Iterator, +{ + next_ch!(chars @ '{'); + let mut value = 0; + let mut len = 0; + for (_, ch) in chars { + let digit = match ch { + '0'..='9' => ch as u8 - b'0', + 'a'..='f' => 10 + ch as u8 - b'a', + 'A'..='F' => 10 + ch as u8 - b'A', + '_' if len > 0 => continue, + '}' if len > 0 => return char::from_u32(value).is_some(), + _ => return false, + }; + if len == 6 { + return false; + } + value *= 0x10; + value += u32::from(digit); + len += 1; + } + false +} + +fn float(input: Cursor) -> Result { + let mut rest = float_digits(input)?; + if let Some(ch) = rest.chars().next() { + if is_ident_start(ch) { + rest = ident_not_raw(rest)?.0; + } + } + word_break(rest) +} + +fn float_digits(input: Cursor) -> Result { + let mut chars = input.chars().peekable(); + match chars.next() { + Some(ch) if ch >= '0' && ch <= '9' => {} + _ => return Err(Reject), + } + + let mut len = 1; + let mut has_dot = false; + let mut has_exp = false; + while let Some(&ch) = chars.peek() { + match ch { + '0'..='9' | '_' => { + chars.next(); + len += 1; + } + '.' => { + if has_dot { + break; + } + chars.next(); + if chars + .peek() + .map_or(false, |&ch| ch == '.' || is_ident_start(ch)) + { + return Err(Reject); + } + len += 1; + has_dot = true; + } + 'e' | 'E' => { + chars.next(); + len += 1; + has_exp = true; + break; + } + _ => break, + } + } + + if !(has_dot || has_exp) { + return Err(Reject); + } + + if has_exp { + let token_before_exp = if has_dot { + Ok(input.advance(len - 1)) + } else { + Err(Reject) + }; + let mut has_sign = false; + let mut has_exp_value = false; + while let Some(&ch) = chars.peek() { + match ch { + '+' | '-' => { + if has_exp_value { + break; + } + if has_sign { + return token_before_exp; + } + chars.next(); + len += 1; + has_sign = true; + } + '0'..='9' => { + chars.next(); + len += 1; + has_exp_value = true; + } + '_' => { + chars.next(); + len += 1; + } + _ => break, + } + } + if !has_exp_value { + return token_before_exp; + } + } + + Ok(input.advance(len)) +} + +fn int(input: Cursor) -> Result { + let mut rest = digits(input)?; + if let Some(ch) = rest.chars().next() { + if is_ident_start(ch) { + rest = ident_not_raw(rest)?.0; + } + } + word_break(rest) +} + +fn digits(mut input: Cursor) -> Result { + let base = if input.starts_with("0x") { + input = input.advance(2); + 16 + } else if input.starts_with("0o") { + input = input.advance(2); + 8 + } else if input.starts_with("0b") { + input = input.advance(2); + 2 + } else { + 10 + }; + + let mut len = 0; + let mut empty = true; + for b in input.bytes() { + match b { + b'0'..=b'9' => { + let digit = (b - b'0') as u64; + if digit >= base { + return Err(Reject); + } + } + b'a'..=b'f' => { + let digit = 10 + (b - b'a') as u64; + if digit >= base { + break; + } + } + b'A'..=b'F' => { + let digit = 10 + (b - b'A') as u64; + if digit >= base { + break; + } + } + b'_' => { + if empty && base == 10 { + return Err(Reject); + } + len += 1; + continue; + } + _ => break, + }; + len += 1; + empty = false; + } + if empty { + Err(Reject) + } else { + Ok(input.advance(len)) + } +} + +fn punct(input: Cursor) -> PResult { + let (rest, ch) = punct_char(input)?; + if ch == '\'' { + if ident_any(rest)?.0.starts_with("'") { + Err(Reject) + } else { + Ok((rest, Punct::new('\'', Spacing::Joint))) + } + } else { + let kind = match punct_char(rest) { + Ok(_) => Spacing::Joint, + Err(Reject) => Spacing::Alone, + }; + Ok((rest, Punct::new(ch, kind))) + } +} + +fn punct_char(input: Cursor) -> PResult { + if input.starts_with("//") || input.starts_with("/*") { + // Do not accept `/` of a comment as a punct. + return Err(Reject); + } + + let mut chars = input.chars(); + let first = match chars.next() { + Some(ch) => ch, + None => { + return Err(Reject); + } + }; + let recognized = "~!@#$%^&*-=+|;:,<.>/?'"; + if recognized.contains(first) { + Ok((input.advance(first.len_utf8()), first)) + } else { + Err(Reject) + } +} + +fn doc_comment<'a>(input: Cursor<'a>, trees: &mut TokenStreamBuilder) -> PResult<'a, ()> { + #[cfg(span_locations)] + let lo = input.off; + let (rest, (comment, inner)) = doc_comment_contents(input)?; + let span = crate::Span::_new_stable(Span { + #[cfg(span_locations)] + lo, + #[cfg(span_locations)] + hi: rest.off, + }); + + let mut scan_for_bare_cr = comment; + while let Some(cr) = scan_for_bare_cr.find('\r') { + let rest = &scan_for_bare_cr[cr + 1..]; + if !rest.starts_with('\n') { + return Err(Reject); + } + scan_for_bare_cr = rest; + } + + let mut pound = Punct::new('#', Spacing::Alone); + pound.set_span(span); + trees.push_token_from_parser(TokenTree::Punct(pound)); + + if inner { + let mut bang = Punct::new('!', Spacing::Alone); + bang.set_span(span); + trees.push_token_from_parser(TokenTree::Punct(bang)); + } + + let doc_ident = crate::Ident::new("doc", span); + let mut equal = Punct::new('=', Spacing::Alone); + equal.set_span(span); + let mut literal = crate::Literal::string(comment); + literal.set_span(span); + let mut bracketed = TokenStreamBuilder::with_capacity(3); + bracketed.push_token_from_parser(TokenTree::Ident(doc_ident)); + bracketed.push_token_from_parser(TokenTree::Punct(equal)); + bracketed.push_token_from_parser(TokenTree::Literal(literal)); + let group = Group::new(Delimiter::Bracket, bracketed.build()); + let mut group = crate::Group::_new_stable(group); + group.set_span(span); + trees.push_token_from_parser(TokenTree::Group(group)); + + Ok((rest, ())) +} + +fn doc_comment_contents(input: Cursor) -> PResult<(&str, bool)> { + if input.starts_with("//!") { + let input = input.advance(3); + let (input, s) = take_until_newline_or_eof(input); + Ok((input, (s, true))) + } else if input.starts_with("/*!") { + let (input, s) = block_comment(input)?; + Ok((input, (&s[3..s.len() - 2], true))) + } else if input.starts_with("///") { + let input = input.advance(3); + if input.starts_with("/") { + return Err(Reject); + } + let (input, s) = take_until_newline_or_eof(input); + Ok((input, (s, false))) + } else if input.starts_with("/**") && !input.rest[3..].starts_with('*') { + let (input, s) = block_comment(input)?; + Ok((input, (&s[3..s.len() - 2], false))) + } else { + Err(Reject) + } +} + +fn take_until_newline_or_eof(input: Cursor) -> (Cursor, &str) { + let chars = input.char_indices(); + + for (i, ch) in chars { + if ch == '\n' { + return (input.advance(i), &input.rest[..i]); + } else if ch == '\r' && input.rest[i + 1..].starts_with('\n') { + return (input.advance(i + 1), &input.rest[..i]); + } + } + + (input.advance(input.len()), input.rest) +} diff --git a/rust/proc-macro2/rcvec.rs b/rust/proc-macro2/rcvec.rs new file mode 100644 index 00000000000000..da2398c789d969 --- /dev/null +++ b/rust/proc-macro2/rcvec.rs @@ -0,0 +1,144 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use core::mem; +use core::slice; +use std::rc::Rc; +use std::vec; + +pub(crate) struct RcVec { + inner: Rc>, +} + +pub(crate) struct RcVecBuilder { + inner: Vec, +} + +pub(crate) struct RcVecMut<'a, T> { + inner: &'a mut Vec, +} + +#[derive(Clone)] +pub(crate) struct RcVecIntoIter { + inner: vec::IntoIter, +} + +impl RcVec { + pub fn is_empty(&self) -> bool { + self.inner.is_empty() + } + + pub fn len(&self) -> usize { + self.inner.len() + } + + pub fn iter(&self) -> slice::Iter { + self.inner.iter() + } + + pub fn make_mut(&mut self) -> RcVecMut + where + T: Clone, + { + RcVecMut { + inner: Rc::make_mut(&mut self.inner), + } + } + + pub fn get_mut(&mut self) -> Option> { + let inner = Rc::get_mut(&mut self.inner)?; + Some(RcVecMut { inner }) + } + + pub fn make_owned(mut self) -> RcVecBuilder + where + T: Clone, + { + let vec = if let Some(owned) = Rc::get_mut(&mut self.inner) { + mem::replace(owned, Vec::new()) + } else { + Vec::clone(&self.inner) + }; + RcVecBuilder { inner: vec } + } +} + +impl RcVecBuilder { + pub fn new() -> Self { + RcVecBuilder { inner: Vec::new() } + } + + pub fn with_capacity(cap: usize) -> Self { + RcVecBuilder { + inner: Vec::with_capacity(cap), + } + } + + pub fn push(&mut self, element: T) { + self.inner.push(element); + } + + pub fn extend(&mut self, iter: impl IntoIterator) { + self.inner.extend(iter); + } + + pub fn as_mut(&mut self) -> RcVecMut { + RcVecMut { + inner: &mut self.inner, + } + } + + pub fn build(self) -> RcVec { + RcVec { + inner: Rc::new(self.inner), + } + } +} + +impl<'a, T> RcVecMut<'a, T> { + pub fn push(&mut self, element: T) { + self.inner.push(element); + } + + pub fn extend(&mut self, iter: impl IntoIterator) { + self.inner.extend(iter); + } + + pub fn pop(&mut self) -> Option { + self.inner.pop() + } + + pub fn as_mut(&mut self) -> RcVecMut { + RcVecMut { inner: self.inner } + } +} + +impl Clone for RcVec { + fn clone(&self) -> Self { + RcVec { + inner: Rc::clone(&self.inner), + } + } +} + +impl IntoIterator for RcVecBuilder { + type Item = T; + type IntoIter = RcVecIntoIter; + + fn into_iter(self) -> Self::IntoIter { + RcVecIntoIter { + inner: self.inner.into_iter(), + } + } +} + +impl Iterator for RcVecIntoIter { + type Item = T; + + fn next(&mut self) -> Option { + self.inner.next() + } + + fn size_hint(&self) -> (usize, Option) { + self.inner.size_hint() + } +} diff --git a/rust/proc-macro2/wrapper.rs b/rust/proc-macro2/wrapper.rs new file mode 100644 index 00000000000000..dacb2e59bd04c2 --- /dev/null +++ b/rust/proc-macro2/wrapper.rs @@ -0,0 +1,996 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::detection::inside_proc_macro; +use crate::{fallback, Delimiter, Punct, Spacing, TokenTree}; +use core::fmt::{self, Debug, Display}; +use core::iter::FromIterator; +use core::ops::RangeBounds; +use core::str::FromStr; +use std::panic; +#[cfg(super_unstable)] +use std::path::PathBuf; + +#[derive(Clone)] +pub(crate) enum TokenStream { + Compiler(DeferredTokenStream), + Fallback(fallback::TokenStream), +} + +// Work around https://github.com/rust-lang/rust/issues/65080. +// In `impl Extend for TokenStream` which is used heavily by quote, +// we hold on to the appended tokens and do proc_macro::TokenStream::extend as +// late as possible to batch together consecutive uses of the Extend impl. +#[derive(Clone)] +pub(crate) struct DeferredTokenStream { + stream: proc_macro::TokenStream, + extra: Vec, +} + +pub(crate) enum LexError { + Compiler(proc_macro::LexError), + Fallback(fallback::LexError), +} + +impl LexError { + fn call_site() -> Self { + LexError::Fallback(fallback::LexError { + span: fallback::Span::call_site(), + }) + } +} + +fn mismatch() -> ! { + panic!("stable/nightly mismatch") +} + +impl DeferredTokenStream { + fn new(stream: proc_macro::TokenStream) -> Self { + DeferredTokenStream { + stream, + extra: Vec::new(), + } + } + + fn is_empty(&self) -> bool { + self.stream.is_empty() && self.extra.is_empty() + } + + fn evaluate_now(&mut self) { + // If-check provides a fast short circuit for the common case of `extra` + // being empty, which saves a round trip over the proc macro bridge. + // Improves macro expansion time in winrt by 6% in debug mode. + if !self.extra.is_empty() { + self.stream.extend(self.extra.drain(..)); + } + } + + fn into_token_stream(mut self) -> proc_macro::TokenStream { + self.evaluate_now(); + self.stream + } +} + +impl TokenStream { + pub fn new() -> Self { + if inside_proc_macro() { + TokenStream::Compiler(DeferredTokenStream::new(proc_macro::TokenStream::new())) + } else { + TokenStream::Fallback(fallback::TokenStream::new()) + } + } + + pub fn is_empty(&self) -> bool { + match self { + TokenStream::Compiler(tts) => tts.is_empty(), + TokenStream::Fallback(tts) => tts.is_empty(), + } + } + + fn unwrap_nightly(self) -> proc_macro::TokenStream { + match self { + TokenStream::Compiler(s) => s.into_token_stream(), + TokenStream::Fallback(_) => mismatch(), + } + } + + fn unwrap_stable(self) -> fallback::TokenStream { + match self { + TokenStream::Compiler(_) => mismatch(), + TokenStream::Fallback(s) => s, + } + } +} + +impl FromStr for TokenStream { + type Err = LexError; + + fn from_str(src: &str) -> Result { + if inside_proc_macro() { + Ok(TokenStream::Compiler(DeferredTokenStream::new( + proc_macro_parse(src)?, + ))) + } else { + Ok(TokenStream::Fallback(src.parse()?)) + } + } +} + +// Work around https://github.com/rust-lang/rust/issues/58736. +fn proc_macro_parse(src: &str) -> Result { + let result = panic::catch_unwind(|| src.parse().map_err(LexError::Compiler)); + result.unwrap_or_else(|_| Err(LexError::call_site())) +} + +impl Display for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + TokenStream::Compiler(tts) => Display::fmt(&tts.clone().into_token_stream(), f), + TokenStream::Fallback(tts) => Display::fmt(tts, f), + } + } +} + +impl From for TokenStream { + fn from(inner: proc_macro::TokenStream) -> TokenStream { + TokenStream::Compiler(DeferredTokenStream::new(inner)) + } +} + +impl From for proc_macro::TokenStream { + fn from(inner: TokenStream) -> proc_macro::TokenStream { + match inner { + TokenStream::Compiler(inner) => inner.into_token_stream(), + TokenStream::Fallback(inner) => inner.to_string().parse().unwrap(), + } + } +} + +impl From for TokenStream { + fn from(inner: fallback::TokenStream) -> TokenStream { + TokenStream::Fallback(inner) + } +} + +// Assumes inside_proc_macro(). +fn into_compiler_token(token: TokenTree) -> proc_macro::TokenTree { + match token { + TokenTree::Group(tt) => tt.inner.unwrap_nightly().into(), + TokenTree::Punct(tt) => { + let spacing = match tt.spacing() { + Spacing::Joint => proc_macro::Spacing::Joint, + Spacing::Alone => proc_macro::Spacing::Alone, + }; + let mut punct = proc_macro::Punct::new(tt.as_char(), spacing); + punct.set_span(tt.span().inner.unwrap_nightly()); + punct.into() + } + TokenTree::Ident(tt) => tt.inner.unwrap_nightly().into(), + TokenTree::Literal(tt) => tt.inner.unwrap_nightly().into(), + } +} + +impl From for TokenStream { + fn from(token: TokenTree) -> TokenStream { + if inside_proc_macro() { + TokenStream::Compiler(DeferredTokenStream::new(into_compiler_token(token).into())) + } else { + TokenStream::Fallback(token.into()) + } + } +} + +impl FromIterator for TokenStream { + fn from_iter>(trees: I) -> Self { + if inside_proc_macro() { + TokenStream::Compiler(DeferredTokenStream::new( + trees.into_iter().map(into_compiler_token).collect(), + )) + } else { + TokenStream::Fallback(trees.into_iter().collect()) + } + } +} + +impl FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + let mut streams = streams.into_iter(); + match streams.next() { + Some(TokenStream::Compiler(mut first)) => { + first.evaluate_now(); + first.stream.extend(streams.map(|s| match s { + TokenStream::Compiler(s) => s.into_token_stream(), + TokenStream::Fallback(_) => mismatch(), + })); + TokenStream::Compiler(first) + } + Some(TokenStream::Fallback(mut first)) => { + first.extend(streams.map(|s| match s { + TokenStream::Fallback(s) => s, + TokenStream::Compiler(_) => mismatch(), + })); + TokenStream::Fallback(first) + } + None => TokenStream::new(), + } + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, stream: I) { + match self { + TokenStream::Compiler(tts) => { + // Here is the reason for DeferredTokenStream. + for token in stream { + tts.extra.push(into_compiler_token(token)); + } + } + TokenStream::Fallback(tts) => tts.extend(stream), + } + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + match self { + TokenStream::Compiler(tts) => { + tts.evaluate_now(); + tts.stream + .extend(streams.into_iter().map(TokenStream::unwrap_nightly)); + } + TokenStream::Fallback(tts) => { + tts.extend(streams.into_iter().map(TokenStream::unwrap_stable)); + } + } + } +} + +impl Debug for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + TokenStream::Compiler(tts) => Debug::fmt(&tts.clone().into_token_stream(), f), + TokenStream::Fallback(tts) => Debug::fmt(tts, f), + } + } +} + +impl LexError { + pub(crate) fn span(&self) -> Span { + match self { + LexError::Compiler(_) => Span::call_site(), + LexError::Fallback(e) => Span::Fallback(e.span()), + } + } +} + +impl From for LexError { + fn from(e: proc_macro::LexError) -> LexError { + LexError::Compiler(e) + } +} + +impl From for LexError { + fn from(e: fallback::LexError) -> LexError { + LexError::Fallback(e) + } +} + +impl Debug for LexError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + LexError::Compiler(e) => Debug::fmt(e, f), + LexError::Fallback(e) => Debug::fmt(e, f), + } + } +} + +impl Display for LexError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + #[cfg(not(no_lexerror_display))] + LexError::Compiler(e) => Display::fmt(e, f), + #[cfg(no_lexerror_display)] + LexError::Compiler(_e) => Display::fmt( + &fallback::LexError { + span: fallback::Span::call_site(), + }, + f, + ), + LexError::Fallback(e) => Display::fmt(e, f), + } + } +} + +#[derive(Clone)] +pub(crate) enum TokenTreeIter { + Compiler(proc_macro::token_stream::IntoIter), + Fallback(fallback::TokenTreeIter), +} + +impl IntoIterator for TokenStream { + type Item = TokenTree; + type IntoIter = TokenTreeIter; + + fn into_iter(self) -> TokenTreeIter { + match self { + TokenStream::Compiler(tts) => { + TokenTreeIter::Compiler(tts.into_token_stream().into_iter()) + } + TokenStream::Fallback(tts) => TokenTreeIter::Fallback(tts.into_iter()), + } + } +} + +impl Iterator for TokenTreeIter { + type Item = TokenTree; + + fn next(&mut self) -> Option { + let token = match self { + TokenTreeIter::Compiler(iter) => iter.next()?, + TokenTreeIter::Fallback(iter) => return iter.next(), + }; + Some(match token { + proc_macro::TokenTree::Group(tt) => crate::Group::_new(Group::Compiler(tt)).into(), + proc_macro::TokenTree::Punct(tt) => { + let spacing = match tt.spacing() { + proc_macro::Spacing::Joint => Spacing::Joint, + proc_macro::Spacing::Alone => Spacing::Alone, + }; + let mut o = Punct::new(tt.as_char(), spacing); + o.set_span(crate::Span::_new(Span::Compiler(tt.span()))); + o.into() + } + proc_macro::TokenTree::Ident(s) => crate::Ident::_new(Ident::Compiler(s)).into(), + proc_macro::TokenTree::Literal(l) => crate::Literal::_new(Literal::Compiler(l)).into(), + }) + } + + fn size_hint(&self) -> (usize, Option) { + match self { + TokenTreeIter::Compiler(tts) => tts.size_hint(), + TokenTreeIter::Fallback(tts) => tts.size_hint(), + } + } +} + +#[derive(Clone, PartialEq, Eq)] +#[cfg(super_unstable)] +pub(crate) enum SourceFile { + Compiler(proc_macro::SourceFile), + Fallback(fallback::SourceFile), +} + +#[cfg(super_unstable)] +impl SourceFile { + fn nightly(sf: proc_macro::SourceFile) -> Self { + SourceFile::Compiler(sf) + } + + /// Get the path to this source file as a string. + pub fn path(&self) -> PathBuf { + match self { + SourceFile::Compiler(a) => a.path(), + SourceFile::Fallback(a) => a.path(), + } + } + + pub fn is_real(&self) -> bool { + match self { + SourceFile::Compiler(a) => a.is_real(), + SourceFile::Fallback(a) => a.is_real(), + } + } +} + +#[cfg(super_unstable)] +impl Debug for SourceFile { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + SourceFile::Compiler(a) => Debug::fmt(a, f), + SourceFile::Fallback(a) => Debug::fmt(a, f), + } + } +} + +#[cfg(any(super_unstable, feature = "span-locations"))] +pub(crate) struct LineColumn { + pub line: usize, + pub column: usize, +} + +#[derive(Copy, Clone)] +pub(crate) enum Span { + Compiler(proc_macro::Span), + Fallback(fallback::Span), +} + +impl Span { + pub fn call_site() -> Self { + if inside_proc_macro() { + Span::Compiler(proc_macro::Span::call_site()) + } else { + Span::Fallback(fallback::Span::call_site()) + } + } + + #[cfg(not(no_hygiene))] + pub fn mixed_site() -> Self { + if inside_proc_macro() { + Span::Compiler(proc_macro::Span::mixed_site()) + } else { + Span::Fallback(fallback::Span::mixed_site()) + } + } + + #[cfg(super_unstable)] + pub fn def_site() -> Self { + if inside_proc_macro() { + Span::Compiler(proc_macro::Span::def_site()) + } else { + Span::Fallback(fallback::Span::def_site()) + } + } + + pub fn resolved_at(&self, other: Span) -> Span { + match (self, other) { + #[cfg(not(no_hygiene))] + (Span::Compiler(a), Span::Compiler(b)) => Span::Compiler(a.resolved_at(b)), + + // Name resolution affects semantics, but location is only cosmetic + #[cfg(no_hygiene)] + (Span::Compiler(_), Span::Compiler(_)) => other, + + (Span::Fallback(a), Span::Fallback(b)) => Span::Fallback(a.resolved_at(b)), + _ => mismatch(), + } + } + + pub fn located_at(&self, other: Span) -> Span { + match (self, other) { + #[cfg(not(no_hygiene))] + (Span::Compiler(a), Span::Compiler(b)) => Span::Compiler(a.located_at(b)), + + // Name resolution affects semantics, but location is only cosmetic + #[cfg(no_hygiene)] + (Span::Compiler(_), Span::Compiler(_)) => *self, + + (Span::Fallback(a), Span::Fallback(b)) => Span::Fallback(a.located_at(b)), + _ => mismatch(), + } + } + + pub fn unwrap(self) -> proc_macro::Span { + match self { + Span::Compiler(s) => s, + Span::Fallback(_) => panic!("proc_macro::Span is only available in procedural macros"), + } + } + + #[cfg(super_unstable)] + pub fn source_file(&self) -> SourceFile { + match self { + Span::Compiler(s) => SourceFile::nightly(s.source_file()), + Span::Fallback(s) => SourceFile::Fallback(s.source_file()), + } + } + + #[cfg(any(super_unstable, feature = "span-locations"))] + pub fn start(&self) -> LineColumn { + match self { + #[cfg(proc_macro_span)] + Span::Compiler(s) => { + let proc_macro::LineColumn { line, column } = s.start(); + LineColumn { line, column } + } + #[cfg(not(proc_macro_span))] + Span::Compiler(_) => LineColumn { line: 0, column: 0 }, + Span::Fallback(s) => { + let fallback::LineColumn { line, column } = s.start(); + LineColumn { line, column } + } + } + } + + #[cfg(any(super_unstable, feature = "span-locations"))] + pub fn end(&self) -> LineColumn { + match self { + #[cfg(proc_macro_span)] + Span::Compiler(s) => { + let proc_macro::LineColumn { line, column } = s.end(); + LineColumn { line, column } + } + #[cfg(not(proc_macro_span))] + Span::Compiler(_) => LineColumn { line: 0, column: 0 }, + Span::Fallback(s) => { + let fallback::LineColumn { line, column } = s.end(); + LineColumn { line, column } + } + } + } + + #[cfg(super_unstable)] + pub fn before(&self) -> Span { + match self { + Span::Compiler(s) => Span::Compiler(s.before()), + Span::Fallback(s) => Span::Fallback(s.before()), + } + } + + #[cfg(super_unstable)] + pub fn after(&self) -> Span { + match self { + Span::Compiler(s) => Span::Compiler(s.after()), + Span::Fallback(s) => Span::Fallback(s.after()), + } + } + + pub fn join(&self, other: Span) -> Option { + let ret = match (self, other) { + #[cfg(proc_macro_span)] + (Span::Compiler(a), Span::Compiler(b)) => Span::Compiler(a.join(b)?), + (Span::Fallback(a), Span::Fallback(b)) => Span::Fallback(a.join(b)?), + _ => return None, + }; + Some(ret) + } + + #[cfg(super_unstable)] + pub fn eq(&self, other: &Span) -> bool { + match (self, other) { + (Span::Compiler(a), Span::Compiler(b)) => a.eq(b), + (Span::Fallback(a), Span::Fallback(b)) => a.eq(b), + _ => false, + } + } + + fn unwrap_nightly(self) -> proc_macro::Span { + match self { + Span::Compiler(s) => s, + Span::Fallback(_) => mismatch(), + } + } +} + +impl From for crate::Span { + fn from(proc_span: proc_macro::Span) -> crate::Span { + crate::Span::_new(Span::Compiler(proc_span)) + } +} + +impl From for Span { + fn from(inner: fallback::Span) -> Span { + Span::Fallback(inner) + } +} + +impl Debug for Span { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + Span::Compiler(s) => Debug::fmt(s, f), + Span::Fallback(s) => Debug::fmt(s, f), + } + } +} + +pub(crate) fn debug_span_field_if_nontrivial(debug: &mut fmt::DebugStruct, span: Span) { + match span { + Span::Compiler(s) => { + debug.field("span", &s); + } + Span::Fallback(s) => fallback::debug_span_field_if_nontrivial(debug, s), + } +} + +#[derive(Clone)] +pub(crate) enum Group { + Compiler(proc_macro::Group), + Fallback(fallback::Group), +} + +impl Group { + pub fn new(delimiter: Delimiter, stream: TokenStream) -> Self { + match stream { + TokenStream::Compiler(tts) => { + let delimiter = match delimiter { + Delimiter::Parenthesis => proc_macro::Delimiter::Parenthesis, + Delimiter::Bracket => proc_macro::Delimiter::Bracket, + Delimiter::Brace => proc_macro::Delimiter::Brace, + Delimiter::None => proc_macro::Delimiter::None, + }; + Group::Compiler(proc_macro::Group::new(delimiter, tts.into_token_stream())) + } + TokenStream::Fallback(stream) => { + Group::Fallback(fallback::Group::new(delimiter, stream)) + } + } + } + + pub fn delimiter(&self) -> Delimiter { + match self { + Group::Compiler(g) => match g.delimiter() { + proc_macro::Delimiter::Parenthesis => Delimiter::Parenthesis, + proc_macro::Delimiter::Bracket => Delimiter::Bracket, + proc_macro::Delimiter::Brace => Delimiter::Brace, + proc_macro::Delimiter::None => Delimiter::None, + }, + Group::Fallback(g) => g.delimiter(), + } + } + + pub fn stream(&self) -> TokenStream { + match self { + Group::Compiler(g) => TokenStream::Compiler(DeferredTokenStream::new(g.stream())), + Group::Fallback(g) => TokenStream::Fallback(g.stream()), + } + } + + pub fn span(&self) -> Span { + match self { + Group::Compiler(g) => Span::Compiler(g.span()), + Group::Fallback(g) => Span::Fallback(g.span()), + } + } + + pub fn span_open(&self) -> Span { + match self { + #[cfg(not(no_group_open_close))] + Group::Compiler(g) => Span::Compiler(g.span_open()), + #[cfg(no_group_open_close)] + Group::Compiler(g) => Span::Compiler(g.span()), + Group::Fallback(g) => Span::Fallback(g.span_open()), + } + } + + pub fn span_close(&self) -> Span { + match self { + #[cfg(not(no_group_open_close))] + Group::Compiler(g) => Span::Compiler(g.span_close()), + #[cfg(no_group_open_close)] + Group::Compiler(g) => Span::Compiler(g.span()), + Group::Fallback(g) => Span::Fallback(g.span_close()), + } + } + + pub fn set_span(&mut self, span: Span) { + match (self, span) { + (Group::Compiler(g), Span::Compiler(s)) => g.set_span(s), + (Group::Fallback(g), Span::Fallback(s)) => g.set_span(s), + _ => mismatch(), + } + } + + fn unwrap_nightly(self) -> proc_macro::Group { + match self { + Group::Compiler(g) => g, + Group::Fallback(_) => mismatch(), + } + } +} + +impl From for Group { + fn from(g: fallback::Group) -> Self { + Group::Fallback(g) + } +} + +impl Display for Group { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self { + Group::Compiler(group) => Display::fmt(group, formatter), + Group::Fallback(group) => Display::fmt(group, formatter), + } + } +} + +impl Debug for Group { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self { + Group::Compiler(group) => Debug::fmt(group, formatter), + Group::Fallback(group) => Debug::fmt(group, formatter), + } + } +} + +#[derive(Clone)] +pub(crate) enum Ident { + Compiler(proc_macro::Ident), + Fallback(fallback::Ident), +} + +impl Ident { + pub fn new(string: &str, span: Span) -> Self { + match span { + Span::Compiler(s) => Ident::Compiler(proc_macro::Ident::new(string, s)), + Span::Fallback(s) => Ident::Fallback(fallback::Ident::new(string, s)), + } + } + + pub fn new_raw(string: &str, span: Span) -> Self { + match span { + #[cfg(not(no_ident_new_raw))] + Span::Compiler(s) => Ident::Compiler(proc_macro::Ident::new_raw(string, s)), + #[cfg(no_ident_new_raw)] + Span::Compiler(s) => { + let _ = proc_macro::Ident::new(string, s); + // At this point the un-r#-prefixed string is known to be a + // valid identifier. Try to produce a valid raw identifier by + // running the `TokenStream` parser, and unwrapping the first + // token as an `Ident`. + let raw_prefixed = format!("r#{}", string); + if let Ok(ts) = raw_prefixed.parse::() { + let mut iter = ts.into_iter(); + if let (Some(proc_macro::TokenTree::Ident(mut id)), None) = + (iter.next(), iter.next()) + { + id.set_span(s); + return Ident::Compiler(id); + } + } + panic!("not allowed as a raw identifier: `{}`", raw_prefixed) + } + Span::Fallback(s) => Ident::Fallback(fallback::Ident::new_raw(string, s)), + } + } + + pub fn span(&self) -> Span { + match self { + Ident::Compiler(t) => Span::Compiler(t.span()), + Ident::Fallback(t) => Span::Fallback(t.span()), + } + } + + pub fn set_span(&mut self, span: Span) { + match (self, span) { + (Ident::Compiler(t), Span::Compiler(s)) => t.set_span(s), + (Ident::Fallback(t), Span::Fallback(s)) => t.set_span(s), + _ => mismatch(), + } + } + + fn unwrap_nightly(self) -> proc_macro::Ident { + match self { + Ident::Compiler(s) => s, + Ident::Fallback(_) => mismatch(), + } + } +} + +impl PartialEq for Ident { + fn eq(&self, other: &Ident) -> bool { + match (self, other) { + (Ident::Compiler(t), Ident::Compiler(o)) => t.to_string() == o.to_string(), + (Ident::Fallback(t), Ident::Fallback(o)) => t == o, + _ => mismatch(), + } + } +} + +impl PartialEq for Ident +where + T: ?Sized + AsRef, +{ + fn eq(&self, other: &T) -> bool { + let other = other.as_ref(); + match self { + Ident::Compiler(t) => t.to_string() == other, + Ident::Fallback(t) => t == other, + } + } +} + +impl Display for Ident { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + Ident::Compiler(t) => Display::fmt(t, f), + Ident::Fallback(t) => Display::fmt(t, f), + } + } +} + +impl Debug for Ident { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + Ident::Compiler(t) => Debug::fmt(t, f), + Ident::Fallback(t) => Debug::fmt(t, f), + } + } +} + +#[derive(Clone)] +pub(crate) enum Literal { + Compiler(proc_macro::Literal), + Fallback(fallback::Literal), +} + +macro_rules! suffixed_numbers { + ($($name:ident => $kind:ident,)*) => ($( + pub fn $name(n: $kind) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::$name(n)) + } else { + Literal::Fallback(fallback::Literal::$name(n)) + } + } + )*) +} + +macro_rules! unsuffixed_integers { + ($($name:ident => $kind:ident,)*) => ($( + pub fn $name(n: $kind) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::$name(n)) + } else { + Literal::Fallback(fallback::Literal::$name(n)) + } + } + )*) +} + +impl Literal { + pub unsafe fn from_str_unchecked(repr: &str) -> Self { + if inside_proc_macro() { + Literal::Compiler(compiler_literal_from_str(repr).expect("invalid literal")) + } else { + Literal::Fallback(fallback::Literal::from_str_unchecked(repr)) + } + } + + suffixed_numbers! { + u8_suffixed => u8, + u16_suffixed => u16, + u32_suffixed => u32, + u64_suffixed => u64, + u128_suffixed => u128, + usize_suffixed => usize, + i8_suffixed => i8, + i16_suffixed => i16, + i32_suffixed => i32, + i64_suffixed => i64, + i128_suffixed => i128, + isize_suffixed => isize, + + f32_suffixed => f32, + f64_suffixed => f64, + } + + unsuffixed_integers! { + u8_unsuffixed => u8, + u16_unsuffixed => u16, + u32_unsuffixed => u32, + u64_unsuffixed => u64, + u128_unsuffixed => u128, + usize_unsuffixed => usize, + i8_unsuffixed => i8, + i16_unsuffixed => i16, + i32_unsuffixed => i32, + i64_unsuffixed => i64, + i128_unsuffixed => i128, + isize_unsuffixed => isize, + } + + pub fn f32_unsuffixed(f: f32) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::f32_unsuffixed(f)) + } else { + Literal::Fallback(fallback::Literal::f32_unsuffixed(f)) + } + } + + pub fn f64_unsuffixed(f: f64) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::f64_unsuffixed(f)) + } else { + Literal::Fallback(fallback::Literal::f64_unsuffixed(f)) + } + } + + pub fn string(t: &str) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::string(t)) + } else { + Literal::Fallback(fallback::Literal::string(t)) + } + } + + pub fn character(t: char) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::character(t)) + } else { + Literal::Fallback(fallback::Literal::character(t)) + } + } + + pub fn byte_string(bytes: &[u8]) -> Literal { + if inside_proc_macro() { + Literal::Compiler(proc_macro::Literal::byte_string(bytes)) + } else { + Literal::Fallback(fallback::Literal::byte_string(bytes)) + } + } + + pub fn span(&self) -> Span { + match self { + Literal::Compiler(lit) => Span::Compiler(lit.span()), + Literal::Fallback(lit) => Span::Fallback(lit.span()), + } + } + + pub fn set_span(&mut self, span: Span) { + match (self, span) { + (Literal::Compiler(lit), Span::Compiler(s)) => lit.set_span(s), + (Literal::Fallback(lit), Span::Fallback(s)) => lit.set_span(s), + _ => mismatch(), + } + } + + pub fn subspan>(&self, range: R) -> Option { + match self { + #[cfg(proc_macro_span)] + Literal::Compiler(lit) => lit.subspan(range).map(Span::Compiler), + #[cfg(not(proc_macro_span))] + Literal::Compiler(_lit) => None, + Literal::Fallback(lit) => lit.subspan(range).map(Span::Fallback), + } + } + + fn unwrap_nightly(self) -> proc_macro::Literal { + match self { + Literal::Compiler(s) => s, + Literal::Fallback(_) => mismatch(), + } + } +} + +impl From for Literal { + fn from(s: fallback::Literal) -> Literal { + Literal::Fallback(s) + } +} + +impl FromStr for Literal { + type Err = LexError; + + fn from_str(repr: &str) -> Result { + if inside_proc_macro() { + compiler_literal_from_str(repr).map(Literal::Compiler) + } else { + let literal = fallback::Literal::from_str(repr)?; + Ok(Literal::Fallback(literal)) + } + } +} + +fn compiler_literal_from_str(repr: &str) -> Result { + #[cfg(not(no_literal_from_str))] + { + proc_macro::Literal::from_str(repr).map_err(LexError::Compiler) + } + #[cfg(no_literal_from_str)] + { + let tokens = proc_macro_parse(repr)?; + let mut iter = tokens.into_iter(); + if let (Some(proc_macro::TokenTree::Literal(literal)), None) = (iter.next(), iter.next()) { + if literal.to_string().len() == repr.len() { + return Ok(literal); + } + } + Err(LexError::call_site()) + } +} + +impl Display for Literal { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + Literal::Compiler(t) => Display::fmt(t, f), + Literal::Fallback(t) => Display::fmt(t, f), + } + } +} + +impl Debug for Literal { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + Literal::Compiler(t) => Debug::fmt(t, f), + Literal::Fallback(t) => Debug::fmt(t, f), + } + } +} diff --git a/rust/quote/ext.rs b/rust/quote/ext.rs new file mode 100644 index 00000000000000..977d2f0c591924 --- /dev/null +++ b/rust/quote/ext.rs @@ -0,0 +1,112 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::ToTokens; +use core::iter; +use proc_macro2::{TokenStream, TokenTree}; + +/// TokenStream extension trait with methods for appending tokens. +/// +/// This trait is sealed and cannot be implemented outside of the `quote` crate. +pub trait TokenStreamExt: private::Sealed { + /// For use by `ToTokens` implementations. + /// + /// Appends the token specified to this list of tokens. + fn append(&mut self, token: U) + where + U: Into; + + /// For use by `ToTokens` implementations. + /// + /// ``` + /// # use quote::{quote, TokenStreamExt, ToTokens}; + /// # use proc_macro2::TokenStream; + /// # + /// struct X; + /// + /// impl ToTokens for X { + /// fn to_tokens(&self, tokens: &mut TokenStream) { + /// tokens.append_all(&[true, false]); + /// } + /// } + /// + /// let tokens = quote!(#X); + /// assert_eq!(tokens.to_string(), "true false"); + /// ``` + fn append_all(&mut self, iter: I) + where + I: IntoIterator, + I::Item: ToTokens; + + /// For use by `ToTokens` implementations. + /// + /// Appends all of the items in the iterator `I`, separated by the tokens + /// `U`. + fn append_separated(&mut self, iter: I, op: U) + where + I: IntoIterator, + I::Item: ToTokens, + U: ToTokens; + + /// For use by `ToTokens` implementations. + /// + /// Appends all tokens in the iterator `I`, appending `U` after each + /// element, including after the last element of the iterator. + fn append_terminated(&mut self, iter: I, term: U) + where + I: IntoIterator, + I::Item: ToTokens, + U: ToTokens; +} + +impl TokenStreamExt for TokenStream { + fn append(&mut self, token: U) + where + U: Into, + { + self.extend(iter::once(token.into())); + } + + fn append_all(&mut self, iter: I) + where + I: IntoIterator, + I::Item: ToTokens, + { + for token in iter { + token.to_tokens(self); + } + } + + fn append_separated(&mut self, iter: I, op: U) + where + I: IntoIterator, + I::Item: ToTokens, + U: ToTokens, + { + for (i, token) in iter.into_iter().enumerate() { + if i > 0 { + op.to_tokens(self); + } + token.to_tokens(self); + } + } + + fn append_terminated(&mut self, iter: I, term: U) + where + I: IntoIterator, + I::Item: ToTokens, + U: ToTokens, + { + for token in iter { + token.to_tokens(self); + term.to_tokens(self); + } + } +} + +mod private { + use proc_macro2::TokenStream; + + pub trait Sealed {} + + impl Sealed for TokenStream {} +} diff --git a/rust/quote/format.rs b/rust/quote/format.rs new file mode 100644 index 00000000000000..9fbf9c5949a9f0 --- /dev/null +++ b/rust/quote/format.rs @@ -0,0 +1,170 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Formatting macro for constructing `Ident`s. +/// +///
+/// +/// # Syntax +/// +/// Syntax is copied from the [`format!`] macro, supporting both positional and +/// named arguments. +/// +/// Only a limited set of formatting traits are supported. The current mapping +/// of format types to traits is: +/// +/// * `{}` ⇒ [`IdentFragment`] +/// * `{:o}` ⇒ [`Octal`](std::fmt::Octal) +/// * `{:x}` ⇒ [`LowerHex`](std::fmt::LowerHex) +/// * `{:X}` ⇒ [`UpperHex`](std::fmt::UpperHex) +/// * `{:b}` ⇒ [`Binary`](std::fmt::Binary) +/// +/// See [`std::fmt`] for more information. +/// +///
+/// +/// # IdentFragment +/// +/// Unlike `format!`, this macro uses the [`IdentFragment`] formatting trait by +/// default. This trait is like `Display`, with a few differences: +/// +/// * `IdentFragment` is only implemented for a limited set of types, such as +/// unsigned integers and strings. +/// * [`Ident`] arguments will have their `r#` prefixes stripped, if present. +/// +/// [`IdentFragment`]: crate::IdentFragment +/// [`Ident`]: proc_macro2::Ident +/// +///
+/// +/// # Hygiene +/// +/// The [`Span`] of the first `Ident` argument is used as the span of the final +/// identifier, falling back to [`Span::call_site`] when no identifiers are +/// provided. +/// +/// ``` +/// # use quote::format_ident; +/// # let ident = format_ident!("Ident"); +/// // If `ident` is an Ident, the span of `my_ident` will be inherited from it. +/// let my_ident = format_ident!("My{}{}", ident, "IsCool"); +/// assert_eq!(my_ident, "MyIdentIsCool"); +/// ``` +/// +/// Alternatively, the span can be overridden by passing the `span` named +/// argument. +/// +/// ``` +/// # use quote::format_ident; +/// # const IGNORE_TOKENS: &'static str = stringify! { +/// let my_span = /* ... */; +/// # }; +/// # let my_span = proc_macro2::Span::call_site(); +/// format_ident!("MyIdent", span = my_span); +/// ``` +/// +/// [`Span`]: proc_macro2::Span +/// [`Span::call_site`]: proc_macro2::Span::call_site +/// +///


+/// +/// # Panics +/// +/// This method will panic if the resulting formatted string is not a valid +/// identifier. +/// +///
+/// +/// # Examples +/// +/// Composing raw and non-raw identifiers: +/// ``` +/// # use quote::format_ident; +/// let my_ident = format_ident!("My{}", "Ident"); +/// assert_eq!(my_ident, "MyIdent"); +/// +/// let raw = format_ident!("r#Raw"); +/// assert_eq!(raw, "r#Raw"); +/// +/// let my_ident_raw = format_ident!("{}Is{}", my_ident, raw); +/// assert_eq!(my_ident_raw, "MyIdentIsRaw"); +/// ``` +/// +/// Integer formatting options: +/// ``` +/// # use quote::format_ident; +/// let num: u32 = 10; +/// +/// let decimal = format_ident!("Id_{}", num); +/// assert_eq!(decimal, "Id_10"); +/// +/// let octal = format_ident!("Id_{:o}", num); +/// assert_eq!(octal, "Id_12"); +/// +/// let binary = format_ident!("Id_{:b}", num); +/// assert_eq!(binary, "Id_1010"); +/// +/// let lower_hex = format_ident!("Id_{:x}", num); +/// assert_eq!(lower_hex, "Id_a"); +/// +/// let upper_hex = format_ident!("Id_{:X}", num); +/// assert_eq!(upper_hex, "Id_A"); +/// ``` +#[macro_export] +macro_rules! format_ident { + ($fmt:expr) => { + $crate::format_ident_impl!([ + $crate::__private::Option::None, + $fmt + ]) + }; + + ($fmt:expr, $($rest:tt)*) => { + $crate::format_ident_impl!([ + $crate::__private::Option::None, + $fmt + ] $($rest)*) + }; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! format_ident_impl { + // Final state + ([$span:expr, $($fmt:tt)*]) => { + $crate::__private::mk_ident( + &$crate::__private::format!($($fmt)*), + $span, + ) + }; + + // Span argument + ([$old:expr, $($fmt:tt)*] span = $span:expr) => { + $crate::format_ident_impl!([$old, $($fmt)*] span = $span,) + }; + ([$old:expr, $($fmt:tt)*] span = $span:expr, $($rest:tt)*) => { + $crate::format_ident_impl!([ + $crate::__private::Option::Some::<$crate::__private::Span>($span), + $($fmt)* + ] $($rest)*) + }; + + // Named argument + ([$span:expr, $($fmt:tt)*] $name:ident = $arg:expr) => { + $crate::format_ident_impl!([$span, $($fmt)*] $name = $arg,) + }; + ([$span:expr, $($fmt:tt)*] $name:ident = $arg:expr, $($rest:tt)*) => { + match $crate::__private::IdentFragmentAdapter(&$arg) { + arg => $crate::format_ident_impl!([$span.or(arg.span()), $($fmt)*, $name = arg] $($rest)*), + } + }; + + // Positional argument + ([$span:expr, $($fmt:tt)*] $arg:expr) => { + $crate::format_ident_impl!([$span, $($fmt)*] $arg,) + }; + ([$span:expr, $($fmt:tt)*] $arg:expr, $($rest:tt)*) => { + match $crate::__private::IdentFragmentAdapter(&$arg) { + arg => $crate::format_ident_impl!([$span.or(arg.span()), $($fmt)*, arg] $($rest)*), + } + }; +} diff --git a/rust/quote/ident_fragment.rs b/rust/quote/ident_fragment.rs new file mode 100644 index 00000000000000..e4a8d6cd30e97a --- /dev/null +++ b/rust/quote/ident_fragment.rs @@ -0,0 +1,88 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use core::fmt; +use proc_macro2::{Ident, Span}; +use std::borrow::Cow; + +/// Specialized formatting trait used by `format_ident!`. +/// +/// [`Ident`] arguments formatted using this trait will have their `r#` prefix +/// stripped, if present. +/// +/// See [`format_ident!`] for more information. +pub trait IdentFragment { + /// Format this value as an identifier fragment. + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result; + + /// Span associated with this `IdentFragment`. + /// + /// If non-`None`, may be inherited by formatted identifiers. + fn span(&self) -> Option { + None + } +} + +impl IdentFragment for &T { + fn span(&self) -> Option { + ::span(*self) + } + + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + IdentFragment::fmt(*self, f) + } +} + +impl IdentFragment for &mut T { + fn span(&self) -> Option { + ::span(*self) + } + + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + IdentFragment::fmt(*self, f) + } +} + +impl IdentFragment for Ident { + fn span(&self) -> Option { + Some(self.span()) + } + + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let id = self.to_string(); + if id.starts_with("r#") { + fmt::Display::fmt(&id[2..], f) + } else { + fmt::Display::fmt(&id[..], f) + } + } +} + +impl IdentFragment for Cow<'_, T> +where + T: IdentFragment + ToOwned + ?Sized, +{ + fn span(&self) -> Option { + T::span(self) + } + + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + T::fmt(self, f) + } +} + +// Limited set of types which this is implemented for, as we want to avoid types +// which will often include non-identifier characters in their `Display` impl. +macro_rules! ident_fragment_display { + ($($T:ty),*) => { + $( + impl IdentFragment for $T { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + fmt::Display::fmt(self, f) + } + } + )* + }; +} + +ident_fragment_display!(bool, str, String, char); +ident_fragment_display!(u8, u16, u32, u64, u128, usize); diff --git a/rust/quote/lib.rs b/rust/quote/lib.rs new file mode 100644 index 00000000000000..ddab15fe4ebffb --- /dev/null +++ b/rust/quote/lib.rs @@ -0,0 +1,1436 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! [![github]](https://github.com/dtolnay/quote) [![crates-io]](https://crates.io/crates/quote) [![docs-rs]](https://docs.rs/quote) +//! +//! [github]: https://img.shields.io/badge/github-8da0cb?style=for-the-badge&labelColor=555555&logo=github +//! [crates-io]: https://img.shields.io/badge/crates.io-fc8d62?style=for-the-badge&labelColor=555555&logo=rust +//! [docs-rs]: https://img.shields.io/badge/docs.rs-66c2a5?style=for-the-badge&labelColor=555555&logo=docs.rs +//! +//!
+//! +//! This crate provides the [`quote!`] macro for turning Rust syntax tree data +//! structures into tokens of source code. +//! +//! [`quote!`]: macro.quote.html +//! +//! Procedural macros in Rust receive a stream of tokens as input, execute +//! arbitrary Rust code to determine how to manipulate those tokens, and produce +//! a stream of tokens to hand back to the compiler to compile into the caller's +//! crate. Quasi-quoting is a solution to one piece of that — producing +//! tokens to return to the compiler. +//! +//! The idea of quasi-quoting is that we write *code* that we treat as *data*. +//! Within the `quote!` macro, we can write what looks like code to our text +//! editor or IDE. We get all the benefits of the editor's brace matching, +//! syntax highlighting, indentation, and maybe autocompletion. But rather than +//! compiling that as code into the current crate, we can treat it as data, pass +//! it around, mutate it, and eventually hand it back to the compiler as tokens +//! to compile into the macro caller's crate. +//! +//! This crate is motivated by the procedural macro use case, but is a +//! general-purpose Rust quasi-quoting library and is not specific to procedural +//! macros. +//! +//! ```toml +//! [dependencies] +//! quote = "1.0" +//! ``` +//! +//!
+//! +//! # Example +//! +//! The following quasi-quoted block of code is something you might find in [a] +//! procedural macro having to do with data structure serialization. The `#var` +//! syntax performs interpolation of runtime variables into the quoted tokens. +//! Check out the documentation of the [`quote!`] macro for more detail about +//! the syntax. See also the [`quote_spanned!`] macro which is important for +//! implementing hygienic procedural macros. +//! +//! [a]: https://serde.rs/ +//! [`quote_spanned!`]: macro.quote_spanned.html +//! +//! ``` +//! # use quote::quote; +//! # +//! # let generics = ""; +//! # let where_clause = ""; +//! # let field_ty = ""; +//! # let item_ty = ""; +//! # let path = ""; +//! # let value = ""; +//! # +//! let tokens = quote! { +//! struct SerializeWith #generics #where_clause { +//! value: &'a #field_ty, +//! phantom: core::marker::PhantomData<#item_ty>, +//! } +//! +//! impl #generics serde::Serialize for SerializeWith #generics #where_clause { +//! fn serialize(&self, serializer: S) -> Result +//! where +//! S: serde::Serializer, +//! { +//! #path(self.value, serializer) +//! } +//! } +//! +//! SerializeWith { +//! value: #value, +//! phantom: core::marker::PhantomData::<#item_ty>, +//! } +//! }; +//! ``` + +// Quote types in rustdoc of other crates get linked to here. +#![doc(html_root_url = "https://docs.rs/quote/1.0.21")] +#![allow( + clippy::doc_markdown, + clippy::missing_errors_doc, + clippy::missing_panics_doc, + clippy::module_name_repetitions, + // false positive https://github.com/rust-lang/rust-clippy/issues/6983 + clippy::wrong_self_convention, +)] + +#[cfg(all( + not(all(target_arch = "wasm32", target_os = "unknown")), + feature = "proc-macro" +))] +extern crate proc_macro; + +mod ext; +mod format; +mod ident_fragment; +mod to_tokens; + +// Not public API. +#[doc(hidden)] +#[path = "runtime.rs"] +pub mod __private; + +pub use crate::ext::TokenStreamExt; +pub use crate::ident_fragment::IdentFragment; +pub use crate::to_tokens::ToTokens; + +// Not public API. +#[doc(hidden)] +pub mod spanned; + +/// The whole point. +/// +/// Performs variable interpolation against the input and produces it as +/// [`proc_macro2::TokenStream`]. +/// +/// Note: for returning tokens to the compiler in a procedural macro, use +/// `.into()` on the result to convert to [`proc_macro::TokenStream`]. +/// +/// [`TokenStream`]: https://docs.rs/proc-macro2/1.0/proc_macro2/struct.TokenStream.html +/// +///
+/// +/// # Interpolation +/// +/// Variable interpolation is done with `#var` (similar to `$var` in +/// `macro_rules!` macros). This grabs the `var` variable that is currently in +/// scope and inserts it in that location in the output tokens. Any type +/// implementing the [`ToTokens`] trait can be interpolated. This includes most +/// Rust primitive types as well as most of the syntax tree types from the [Syn] +/// crate. +/// +/// [`ToTokens`]: trait.ToTokens.html +/// [Syn]: https://github.com/dtolnay/syn +/// +/// Repetition is done using `#(...)*` or `#(...),*` again similar to +/// `macro_rules!`. This iterates through the elements of any variable +/// interpolated within the repetition and inserts a copy of the repetition body +/// for each one. The variables in an interpolation may be a `Vec`, slice, +/// `BTreeSet`, or any `Iterator`. +/// +/// - `#(#var)*` — no separators +/// - `#(#var),*` — the character before the asterisk is used as a separator +/// - `#( struct #var; )*` — the repetition can contain other tokens +/// - `#( #k => println!("{}", #v), )*` — even multiple interpolations +/// +///
+/// +/// # Hygiene +/// +/// Any interpolated tokens preserve the `Span` information provided by their +/// `ToTokens` implementation. Tokens that originate within the `quote!` +/// invocation are spanned with [`Span::call_site()`]. +/// +/// [`Span::call_site()`]: https://docs.rs/proc-macro2/1.0/proc_macro2/struct.Span.html#method.call_site +/// +/// A different span can be provided through the [`quote_spanned!`] macro. +/// +/// [`quote_spanned!`]: macro.quote_spanned.html +/// +///
+/// +/// # Return type +/// +/// The macro evaluates to an expression of type `proc_macro2::TokenStream`. +/// Meanwhile Rust procedural macros are expected to return the type +/// `proc_macro::TokenStream`. +/// +/// The difference between the two types is that `proc_macro` types are entirely +/// specific to procedural macros and cannot ever exist in code outside of a +/// procedural macro, while `proc_macro2` types may exist anywhere including +/// tests and non-macro code like main.rs and build.rs. This is why even the +/// procedural macro ecosystem is largely built around `proc_macro2`, because +/// that ensures the libraries are unit testable and accessible in non-macro +/// contexts. +/// +/// There is a [`From`]-conversion in both directions so returning the output of +/// `quote!` from a procedural macro usually looks like `tokens.into()` or +/// `proc_macro::TokenStream::from(tokens)`. +/// +/// [`From`]: https://doc.rust-lang.org/std/convert/trait.From.html +/// +///
+/// +/// # Examples +/// +/// ### Procedural macro +/// +/// The structure of a basic procedural macro is as follows. Refer to the [Syn] +/// crate for further useful guidance on using `quote!` as part of a procedural +/// macro. +/// +/// [Syn]: https://github.com/dtolnay/syn +/// +/// ``` +/// # #[cfg(any())] +/// extern crate proc_macro; +/// # extern crate proc_macro2; +/// +/// # #[cfg(any())] +/// use proc_macro::TokenStream; +/// # use proc_macro2::TokenStream; +/// use quote::quote; +/// +/// # const IGNORE_TOKENS: &'static str = stringify! { +/// #[proc_macro_derive(HeapSize)] +/// # }; +/// pub fn derive_heap_size(input: TokenStream) -> TokenStream { +/// // Parse the input and figure out what implementation to generate... +/// # const IGNORE_TOKENS: &'static str = stringify! { +/// let name = /* ... */; +/// let expr = /* ... */; +/// # }; +/// # +/// # let name = 0; +/// # let expr = 0; +/// +/// let expanded = quote! { +/// // The generated impl. +/// impl heapsize::HeapSize for #name { +/// fn heap_size_of_children(&self) -> usize { +/// #expr +/// } +/// } +/// }; +/// +/// // Hand the output tokens back to the compiler. +/// TokenStream::from(expanded) +/// } +/// ``` +/// +///


+/// +/// ### Combining quoted fragments +/// +/// Usually you don't end up constructing an entire final `TokenStream` in one +/// piece. Different parts may come from different helper functions. The tokens +/// produced by `quote!` themselves implement `ToTokens` and so can be +/// interpolated into later `quote!` invocations to build up a final result. +/// +/// ``` +/// # use quote::quote; +/// # +/// let type_definition = quote! {...}; +/// let methods = quote! {...}; +/// +/// let tokens = quote! { +/// #type_definition +/// #methods +/// }; +/// ``` +/// +///


+/// +/// ### Constructing identifiers +/// +/// Suppose we have an identifier `ident` which came from somewhere in a macro +/// input and we need to modify it in some way for the macro output. Let's +/// consider prepending the identifier with an underscore. +/// +/// Simply interpolating the identifier next to an underscore will not have the +/// behavior of concatenating them. The underscore and the identifier will +/// continue to be two separate tokens as if you had written `_ x`. +/// +/// ``` +/// # use proc_macro2::{self as syn, Span}; +/// # use quote::quote; +/// # +/// # let ident = syn::Ident::new("i", Span::call_site()); +/// # +/// // incorrect +/// quote! { +/// let mut _#ident = 0; +/// } +/// # ; +/// ``` +/// +/// The solution is to build a new identifier token with the correct value. As +/// this is such a common case, the [`format_ident!`] macro provides a +/// convenient utility for doing so correctly. +/// +/// ``` +/// # use proc_macro2::{Ident, Span}; +/// # use quote::{format_ident, quote}; +/// # +/// # let ident = Ident::new("i", Span::call_site()); +/// # +/// let varname = format_ident!("_{}", ident); +/// quote! { +/// let mut #varname = 0; +/// } +/// # ; +/// ``` +/// +/// Alternatively, the APIs provided by Syn and proc-macro2 can be used to +/// directly build the identifier. This is roughly equivalent to the above, but +/// will not handle `ident` being a raw identifier. +/// +/// ``` +/// # use proc_macro2::{self as syn, Span}; +/// # use quote::quote; +/// # +/// # let ident = syn::Ident::new("i", Span::call_site()); +/// # +/// let concatenated = format!("_{}", ident); +/// let varname = syn::Ident::new(&concatenated, ident.span()); +/// quote! { +/// let mut #varname = 0; +/// } +/// # ; +/// ``` +/// +///


+/// +/// ### Making method calls +/// +/// Let's say our macro requires some type specified in the macro input to have +/// a constructor called `new`. We have the type in a variable called +/// `field_type` of type `syn::Type` and want to invoke the constructor. +/// +/// ``` +/// # use quote::quote; +/// # +/// # let field_type = quote!(...); +/// # +/// // incorrect +/// quote! { +/// let value = #field_type::new(); +/// } +/// # ; +/// ``` +/// +/// This works only sometimes. If `field_type` is `String`, the expanded code +/// contains `String::new()` which is fine. But if `field_type` is something +/// like `Vec` then the expanded code is `Vec::new()` which is invalid +/// syntax. Ordinarily in handwritten Rust we would write `Vec::::new()` +/// but for macros often the following is more convenient. +/// +/// ``` +/// # use quote::quote; +/// # +/// # let field_type = quote!(...); +/// # +/// quote! { +/// let value = <#field_type>::new(); +/// } +/// # ; +/// ``` +/// +/// This expands to `>::new()` which behaves correctly. +/// +/// A similar pattern is appropriate for trait methods. +/// +/// ``` +/// # use quote::quote; +/// # +/// # let field_type = quote!(...); +/// # +/// quote! { +/// let value = <#field_type as core::default::Default>::default(); +/// } +/// # ; +/// ``` +/// +///


+/// +/// ### Interpolating text inside of doc comments +/// +/// Neither doc comments nor string literals get interpolation behavior in +/// quote: +/// +/// ```compile_fail +/// quote! { +/// /// try to interpolate: #ident +/// /// +/// /// ... +/// } +/// ``` +/// +/// ```compile_fail +/// quote! { +/// #[doc = "try to interpolate: #ident"] +/// } +/// ``` +/// +/// Instead the best way to build doc comments that involve variables is by +/// formatting the doc string literal outside of quote. +/// +/// ```rust +/// # use proc_macro2::{Ident, Span}; +/// # use quote::quote; +/// # +/// # const IGNORE: &str = stringify! { +/// let msg = format!(...); +/// # }; +/// # +/// # let ident = Ident::new("var", Span::call_site()); +/// # let msg = format!("try to interpolate: {}", ident); +/// quote! { +/// #[doc = #msg] +/// /// +/// /// ... +/// } +/// # ; +/// ``` +/// +///


+/// +/// ### Indexing into a tuple struct +/// +/// When interpolating indices of a tuple or tuple struct, we need them not to +/// appears suffixed as integer literals by interpolating them as [`syn::Index`] +/// instead. +/// +/// [`syn::Index`]: https://docs.rs/syn/1.0/syn/struct.Index.html +/// +/// ```compile_fail +/// let i = 0usize..self.fields.len(); +/// +/// // expands to 0 + self.0usize.heap_size() + self.1usize.heap_size() + ... +/// // which is not valid syntax +/// quote! { +/// 0 #( + self.#i.heap_size() )* +/// } +/// ``` +/// +/// ``` +/// # use proc_macro2::{Ident, TokenStream}; +/// # use quote::quote; +/// # +/// # mod syn { +/// # use proc_macro2::{Literal, TokenStream}; +/// # use quote::{ToTokens, TokenStreamExt}; +/// # +/// # pub struct Index(usize); +/// # +/// # impl From for Index { +/// # fn from(i: usize) -> Self { +/// # Index(i) +/// # } +/// # } +/// # +/// # impl ToTokens for Index { +/// # fn to_tokens(&self, tokens: &mut TokenStream) { +/// # tokens.append(Literal::usize_unsuffixed(self.0)); +/// # } +/// # } +/// # } +/// # +/// # struct Struct { +/// # fields: Vec, +/// # } +/// # +/// # impl Struct { +/// # fn example(&self) -> TokenStream { +/// let i = (0..self.fields.len()).map(syn::Index::from); +/// +/// // expands to 0 + self.0.heap_size() + self.1.heap_size() + ... +/// quote! { +/// 0 #( + self.#i.heap_size() )* +/// } +/// # } +/// # } +/// ``` +#[cfg(doc)] +#[macro_export] +macro_rules! quote { + ($($tt:tt)*) => { + ... + }; +} + +#[cfg(not(doc))] +#[macro_export] +macro_rules! quote { + () => { + $crate::__private::TokenStream::new() + }; + + // Special case rule for a single tt, for performance. + ($tt:tt) => {{ + let mut _s = $crate::__private::TokenStream::new(); + $crate::quote_token!{$tt _s} + _s + }}; + + // Special case rules for two tts, for performance. + (# $var:ident) => {{ + let mut _s = $crate::__private::TokenStream::new(); + $crate::ToTokens::to_tokens(&$var, &mut _s); + _s + }}; + ($tt1:tt $tt2:tt) => {{ + let mut _s = $crate::__private::TokenStream::new(); + $crate::quote_token!{$tt1 _s} + $crate::quote_token!{$tt2 _s} + _s + }}; + + // Rule for any other number of tokens. + ($($tt:tt)*) => {{ + let mut _s = $crate::__private::TokenStream::new(); + $crate::quote_each_token!{_s $($tt)*} + _s + }}; +} + +/// Same as `quote!`, but applies a given span to all tokens originating within +/// the macro invocation. +/// +///
+/// +/// # Syntax +/// +/// A span expression of type [`Span`], followed by `=>`, followed by the tokens +/// to quote. The span expression should be brief — use a variable for +/// anything more than a few characters. There should be no space before the +/// `=>` token. +/// +/// [`Span`]: https://docs.rs/proc-macro2/1.0/proc_macro2/struct.Span.html +/// +/// ``` +/// # use proc_macro2::Span; +/// # use quote::quote_spanned; +/// # +/// # const IGNORE_TOKENS: &'static str = stringify! { +/// let span = /* ... */; +/// # }; +/// # let span = Span::call_site(); +/// # let init = 0; +/// +/// // On one line, use parentheses. +/// let tokens = quote_spanned!(span=> Box::into_raw(Box::new(#init))); +/// +/// // On multiple lines, place the span at the top and use braces. +/// let tokens = quote_spanned! {span=> +/// Box::into_raw(Box::new(#init)) +/// }; +/// ``` +/// +/// The lack of space before the `=>` should look jarring to Rust programmers +/// and this is intentional. The formatting is designed to be visibly +/// off-balance and draw the eye a particular way, due to the span expression +/// being evaluated in the context of the procedural macro and the remaining +/// tokens being evaluated in the generated code. +/// +///
+/// +/// # Hygiene +/// +/// Any interpolated tokens preserve the `Span` information provided by their +/// `ToTokens` implementation. Tokens that originate within the `quote_spanned!` +/// invocation are spanned with the given span argument. +/// +///
+/// +/// # Example +/// +/// The following procedural macro code uses `quote_spanned!` to assert that a +/// particular Rust type implements the [`Sync`] trait so that references can be +/// safely shared between threads. +/// +/// [`Sync`]: https://doc.rust-lang.org/std/marker/trait.Sync.html +/// +/// ``` +/// # use quote::{quote_spanned, TokenStreamExt, ToTokens}; +/// # use proc_macro2::{Span, TokenStream}; +/// # +/// # struct Type; +/// # +/// # impl Type { +/// # fn span(&self) -> Span { +/// # Span::call_site() +/// # } +/// # } +/// # +/// # impl ToTokens for Type { +/// # fn to_tokens(&self, _tokens: &mut TokenStream) {} +/// # } +/// # +/// # let ty = Type; +/// # let call_site = Span::call_site(); +/// # +/// let ty_span = ty.span(); +/// let assert_sync = quote_spanned! {ty_span=> +/// struct _AssertSync where #ty: Sync; +/// }; +/// ``` +/// +/// If the assertion fails, the user will see an error like the following. The +/// input span of their type is highlighted in the error. +/// +/// ```text +/// error[E0277]: the trait bound `*const (): std::marker::Sync` is not satisfied +/// --> src/main.rs:10:21 +/// | +/// 10 | static ref PTR: *const () = &(); +/// | ^^^^^^^^^ `*const ()` cannot be shared between threads safely +/// ``` +/// +/// In this example it is important for the where-clause to be spanned with the +/// line/column information of the user's input type so that error messages are +/// placed appropriately by the compiler. +#[cfg(doc)] +#[macro_export] +macro_rules! quote_spanned { + ($span:expr=> $($tt:tt)*) => { + ... + }; +} + +#[cfg(not(doc))] +#[macro_export] +macro_rules! quote_spanned { + ($span:expr=>) => {{ + let _: $crate::__private::Span = $span; + $crate::__private::TokenStream::new() + }}; + + // Special case rule for a single tt, for performance. + ($span:expr=> $tt:tt) => {{ + let mut _s = $crate::__private::TokenStream::new(); + let _span: $crate::__private::Span = $span; + $crate::quote_token_spanned!{$tt _s _span} + _s + }}; + + // Special case rules for two tts, for performance. + ($span:expr=> # $var:ident) => {{ + let mut _s = $crate::__private::TokenStream::new(); + let _: $crate::__private::Span = $span; + $crate::ToTokens::to_tokens(&$var, &mut _s); + _s + }}; + ($span:expr=> $tt1:tt $tt2:tt) => {{ + let mut _s = $crate::__private::TokenStream::new(); + let _span: $crate::__private::Span = $span; + $crate::quote_token_spanned!{$tt1 _s _span} + $crate::quote_token_spanned!{$tt2 _s _span} + _s + }}; + + // Rule for any other number of tokens. + ($span:expr=> $($tt:tt)*) => {{ + let mut _s = $crate::__private::TokenStream::new(); + let _span: $crate::__private::Span = $span; + $crate::quote_each_token_spanned!{_s _span $($tt)*} + _s + }}; +} + +// Extract the names of all #metavariables and pass them to the $call macro. +// +// in: pounded_var_names!(then!(...) a #b c #( #d )* #e) +// out: then!(... b); +// then!(... d); +// then!(... e); +#[macro_export] +#[doc(hidden)] +macro_rules! pounded_var_names { + ($call:ident! $extra:tt $($tts:tt)*) => { + $crate::pounded_var_names_with_context!{$call! $extra + (@ $($tts)*) + ($($tts)* @) + } + }; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! pounded_var_names_with_context { + ($call:ident! $extra:tt ($($b1:tt)*) ($($curr:tt)*)) => { + $( + $crate::pounded_var_with_context!{$call! $extra $b1 $curr} + )* + }; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! pounded_var_with_context { + ($call:ident! $extra:tt $b1:tt ( $($inner:tt)* )) => { + $crate::pounded_var_names!{$call! $extra $($inner)*} + }; + + ($call:ident! $extra:tt $b1:tt [ $($inner:tt)* ]) => { + $crate::pounded_var_names!{$call! $extra $($inner)*} + }; + + ($call:ident! $extra:tt $b1:tt { $($inner:tt)* }) => { + $crate::pounded_var_names!{$call! $extra $($inner)*} + }; + + ($call:ident!($($extra:tt)*) # $var:ident) => { + $crate::$call!($($extra)* $var); + }; + + ($call:ident! $extra:tt $b1:tt $curr:tt) => {}; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! quote_bind_into_iter { + ($has_iter:ident $var:ident) => { + // `mut` may be unused if $var occurs multiple times in the list. + #[allow(unused_mut)] + let (mut $var, i) = $var.quote_into_iter(); + let $has_iter = $has_iter | i; + }; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! quote_bind_next_or_break { + ($var:ident) => { + let $var = match $var.next() { + Some(_x) => $crate::__private::RepInterp(_x), + None => break, + }; + }; +} + +// The obvious way to write this macro is as a tt muncher. This implementation +// does something more complex for two reasons. +// +// - With a tt muncher it's easy to hit Rust's built-in recursion_limit, which +// this implementation avoids because it isn't tail recursive. +// +// - Compile times for a tt muncher are quadratic relative to the length of +// the input. This implementation is linear, so it will be faster +// (potentially much faster) for big inputs. However, the constant factors +// of this implementation are higher than that of a tt muncher, so it is +// somewhat slower than a tt muncher if there are many invocations with +// short inputs. +// +// An invocation like this: +// +// quote_each_token!(_s a b c d e f g h i j); +// +// expands to this: +// +// quote_tokens_with_context!(_s +// (@ @ @ @ @ @ a b c d e f g h i j) +// (@ @ @ @ @ a b c d e f g h i j @) +// (@ @ @ @ a b c d e f g h i j @ @) +// (@ @ @ (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) @ @ @) +// (@ @ a b c d e f g h i j @ @ @ @) +// (@ a b c d e f g h i j @ @ @ @ @) +// (a b c d e f g h i j @ @ @ @ @ @) +// ); +// +// which gets transposed and expanded to this: +// +// quote_token_with_context!(_s @ @ @ @ @ @ a); +// quote_token_with_context!(_s @ @ @ @ @ a b); +// quote_token_with_context!(_s @ @ @ @ a b c); +// quote_token_with_context!(_s @ @ @ (a) b c d); +// quote_token_with_context!(_s @ @ a (b) c d e); +// quote_token_with_context!(_s @ a b (c) d e f); +// quote_token_with_context!(_s a b c (d) e f g); +// quote_token_with_context!(_s b c d (e) f g h); +// quote_token_with_context!(_s c d e (f) g h i); +// quote_token_with_context!(_s d e f (g) h i j); +// quote_token_with_context!(_s e f g (h) i j @); +// quote_token_with_context!(_s f g h (i) j @ @); +// quote_token_with_context!(_s g h i (j) @ @ @); +// quote_token_with_context!(_s h i j @ @ @ @); +// quote_token_with_context!(_s i j @ @ @ @ @); +// quote_token_with_context!(_s j @ @ @ @ @ @); +// +// Without having used muncher-style recursion, we get one invocation of +// quote_token_with_context for each original tt, with three tts of context on +// either side. This is enough for the longest possible interpolation form (a +// repetition with separator, as in `# (#var) , *`) to be fully represented with +// the first or last tt in the middle. +// +// The middle tt (surrounded by parentheses) is the tt being processed. +// +// - When it is a `#`, quote_token_with_context can do an interpolation. The +// interpolation kind will depend on the three subsequent tts. +// +// - When it is within a later part of an interpolation, it can be ignored +// because the interpolation has already been done. +// +// - When it is not part of an interpolation it can be pushed as a single +// token into the output. +// +// - When the middle token is an unparenthesized `@`, that call is one of the +// first 3 or last 3 calls of quote_token_with_context and does not +// correspond to one of the original input tokens, so turns into nothing. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_each_token { + ($tokens:ident $($tts:tt)*) => { + $crate::quote_tokens_with_context!{$tokens + (@ @ @ @ @ @ $($tts)*) + (@ @ @ @ @ $($tts)* @) + (@ @ @ @ $($tts)* @ @) + (@ @ @ $(($tts))* @ @ @) + (@ @ $($tts)* @ @ @ @) + (@ $($tts)* @ @ @ @ @) + ($($tts)* @ @ @ @ @ @) + } + }; +} + +// See the explanation on quote_each_token. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_each_token_spanned { + ($tokens:ident $span:ident $($tts:tt)*) => { + $crate::quote_tokens_with_context_spanned!{$tokens $span + (@ @ @ @ @ @ $($tts)*) + (@ @ @ @ @ $($tts)* @) + (@ @ @ @ $($tts)* @ @) + (@ @ @ $(($tts))* @ @ @) + (@ @ $($tts)* @ @ @ @) + (@ $($tts)* @ @ @ @ @) + ($($tts)* @ @ @ @ @ @) + } + }; +} + +// See the explanation on quote_each_token. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_tokens_with_context { + ($tokens:ident + ($($b3:tt)*) ($($b2:tt)*) ($($b1:tt)*) + ($($curr:tt)*) + ($($a1:tt)*) ($($a2:tt)*) ($($a3:tt)*) + ) => { + $( + $crate::quote_token_with_context!{$tokens $b3 $b2 $b1 $curr $a1 $a2 $a3} + )* + }; +} + +// See the explanation on quote_each_token. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_tokens_with_context_spanned { + ($tokens:ident $span:ident + ($($b3:tt)*) ($($b2:tt)*) ($($b1:tt)*) + ($($curr:tt)*) + ($($a1:tt)*) ($($a2:tt)*) ($($a3:tt)*) + ) => { + $( + $crate::quote_token_with_context_spanned!{$tokens $span $b3 $b2 $b1 $curr $a1 $a2 $a3} + )* + }; +} + +// See the explanation on quote_each_token. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_token_with_context { + // Unparenthesized `@` indicates this call does not correspond to one of the + // original input tokens. Ignore it. + ($tokens:ident $b3:tt $b2:tt $b1:tt @ $a1:tt $a2:tt $a3:tt) => {}; + + // A repetition with no separator. + ($tokens:ident $b3:tt $b2:tt $b1:tt (#) ( $($inner:tt)* ) * $a3:tt) => {{ + use $crate::__private::ext::*; + let has_iter = $crate::__private::ThereIsNoIteratorInRepetition; + $crate::pounded_var_names!{quote_bind_into_iter!(has_iter) () $($inner)*} + let _: $crate::__private::HasIterator = has_iter; + // This is `while true` instead of `loop` because if there are no + // iterators used inside of this repetition then the body would not + // contain any `break`, so the compiler would emit unreachable code + // warnings on anything below the loop. We use has_iter to detect and + // fail to compile when there are no iterators, so here we just work + // around the unneeded extra warning. + while true { + $crate::pounded_var_names!{quote_bind_next_or_break!() () $($inner)*} + $crate::quote_each_token!{$tokens $($inner)*} + } + }}; + // ... and one step later. + ($tokens:ident $b3:tt $b2:tt # (( $($inner:tt)* )) * $a2:tt $a3:tt) => {}; + // ... and one step later. + ($tokens:ident $b3:tt # ( $($inner:tt)* ) (*) $a1:tt $a2:tt $a3:tt) => {}; + + // A repetition with separator. + ($tokens:ident $b3:tt $b2:tt $b1:tt (#) ( $($inner:tt)* ) $sep:tt *) => {{ + use $crate::__private::ext::*; + let mut _i = 0usize; + let has_iter = $crate::__private::ThereIsNoIteratorInRepetition; + $crate::pounded_var_names!{quote_bind_into_iter!(has_iter) () $($inner)*} + let _: $crate::__private::HasIterator = has_iter; + while true { + $crate::pounded_var_names!{quote_bind_next_or_break!() () $($inner)*} + if _i > 0 { + $crate::quote_token!{$sep $tokens} + } + _i += 1; + $crate::quote_each_token!{$tokens $($inner)*} + } + }}; + // ... and one step later. + ($tokens:ident $b3:tt $b2:tt # (( $($inner:tt)* )) $sep:tt * $a3:tt) => {}; + // ... and one step later. + ($tokens:ident $b3:tt # ( $($inner:tt)* ) ($sep:tt) * $a2:tt $a3:tt) => {}; + // (A special case for `#(var)**`, where the first `*` is treated as the + // repetition symbol and the second `*` is treated as an ordinary token.) + ($tokens:ident # ( $($inner:tt)* ) * (*) $a1:tt $a2:tt $a3:tt) => { + // https://github.com/dtolnay/quote/issues/130 + $crate::quote_token!{* $tokens} + }; + // ... and one step later. + ($tokens:ident # ( $($inner:tt)* ) $sep:tt (*) $a1:tt $a2:tt $a3:tt) => {}; + + // A non-repetition interpolation. + ($tokens:ident $b3:tt $b2:tt $b1:tt (#) $var:ident $a2:tt $a3:tt) => { + $crate::ToTokens::to_tokens(&$var, &mut $tokens); + }; + // ... and one step later. + ($tokens:ident $b3:tt $b2:tt # ($var:ident) $a1:tt $a2:tt $a3:tt) => {}; + + // An ordinary token, not part of any interpolation. + ($tokens:ident $b3:tt $b2:tt $b1:tt ($curr:tt) $a1:tt $a2:tt $a3:tt) => { + $crate::quote_token!{$curr $tokens} + }; +} + +// See the explanation on quote_each_token, and on the individual rules of +// quote_token_with_context. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_token_with_context_spanned { + ($tokens:ident $span:ident $b3:tt $b2:tt $b1:tt @ $a1:tt $a2:tt $a3:tt) => {}; + + ($tokens:ident $span:ident $b3:tt $b2:tt $b1:tt (#) ( $($inner:tt)* ) * $a3:tt) => {{ + use $crate::__private::ext::*; + let has_iter = $crate::__private::ThereIsNoIteratorInRepetition; + $crate::pounded_var_names!{quote_bind_into_iter!(has_iter) () $($inner)*} + let _: $crate::__private::HasIterator = has_iter; + while true { + $crate::pounded_var_names!{quote_bind_next_or_break!() () $($inner)*} + $crate::quote_each_token_spanned!{$tokens $span $($inner)*} + } + }}; + ($tokens:ident $span:ident $b3:tt $b2:tt # (( $($inner:tt)* )) * $a2:tt $a3:tt) => {}; + ($tokens:ident $span:ident $b3:tt # ( $($inner:tt)* ) (*) $a1:tt $a2:tt $a3:tt) => {}; + + ($tokens:ident $span:ident $b3:tt $b2:tt $b1:tt (#) ( $($inner:tt)* ) $sep:tt *) => {{ + use $crate::__private::ext::*; + let mut _i = 0usize; + let has_iter = $crate::__private::ThereIsNoIteratorInRepetition; + $crate::pounded_var_names!{quote_bind_into_iter!(has_iter) () $($inner)*} + let _: $crate::__private::HasIterator = has_iter; + while true { + $crate::pounded_var_names!{quote_bind_next_or_break!() () $($inner)*} + if _i > 0 { + $crate::quote_token_spanned!{$sep $tokens $span} + } + _i += 1; + $crate::quote_each_token_spanned!{$tokens $span $($inner)*} + } + }}; + ($tokens:ident $span:ident $b3:tt $b2:tt # (( $($inner:tt)* )) $sep:tt * $a3:tt) => {}; + ($tokens:ident $span:ident $b3:tt # ( $($inner:tt)* ) ($sep:tt) * $a2:tt $a3:tt) => {}; + ($tokens:ident $span:ident # ( $($inner:tt)* ) * (*) $a1:tt $a2:tt $a3:tt) => { + // https://github.com/dtolnay/quote/issues/130 + $crate::quote_token_spanned!{* $tokens $span} + }; + ($tokens:ident $span:ident # ( $($inner:tt)* ) $sep:tt (*) $a1:tt $a2:tt $a3:tt) => {}; + + ($tokens:ident $span:ident $b3:tt $b2:tt $b1:tt (#) $var:ident $a2:tt $a3:tt) => { + $crate::ToTokens::to_tokens(&$var, &mut $tokens); + }; + ($tokens:ident $span:ident $b3:tt $b2:tt # ($var:ident) $a1:tt $a2:tt $a3:tt) => {}; + + ($tokens:ident $span:ident $b3:tt $b2:tt $b1:tt ($curr:tt) $a1:tt $a2:tt $a3:tt) => { + $crate::quote_token_spanned!{$curr $tokens $span} + }; +} + +// These rules are ordered by approximate token frequency, at least for the +// first 10 or so, to improve compile times. Having `ident` first is by far the +// most important because it's typically 2-3x more common than the next most +// common token. +// +// Separately, we put the token being matched in the very front so that failing +// rules may fail to match as quickly as possible. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_token { + ($ident:ident $tokens:ident) => { + $crate::__private::push_ident(&mut $tokens, stringify!($ident)); + }; + + (:: $tokens:ident) => { + $crate::__private::push_colon2(&mut $tokens); + }; + + (( $($inner:tt)* ) $tokens:ident) => { + $crate::__private::push_group( + &mut $tokens, + $crate::__private::Delimiter::Parenthesis, + $crate::quote!($($inner)*), + ); + }; + + ([ $($inner:tt)* ] $tokens:ident) => { + $crate::__private::push_group( + &mut $tokens, + $crate::__private::Delimiter::Bracket, + $crate::quote!($($inner)*), + ); + }; + + ({ $($inner:tt)* } $tokens:ident) => { + $crate::__private::push_group( + &mut $tokens, + $crate::__private::Delimiter::Brace, + $crate::quote!($($inner)*), + ); + }; + + (# $tokens:ident) => { + $crate::__private::push_pound(&mut $tokens); + }; + + (, $tokens:ident) => { + $crate::__private::push_comma(&mut $tokens); + }; + + (. $tokens:ident) => { + $crate::__private::push_dot(&mut $tokens); + }; + + (; $tokens:ident) => { + $crate::__private::push_semi(&mut $tokens); + }; + + (: $tokens:ident) => { + $crate::__private::push_colon(&mut $tokens); + }; + + (+ $tokens:ident) => { + $crate::__private::push_add(&mut $tokens); + }; + + (+= $tokens:ident) => { + $crate::__private::push_add_eq(&mut $tokens); + }; + + (& $tokens:ident) => { + $crate::__private::push_and(&mut $tokens); + }; + + (&& $tokens:ident) => { + $crate::__private::push_and_and(&mut $tokens); + }; + + (&= $tokens:ident) => { + $crate::__private::push_and_eq(&mut $tokens); + }; + + (@ $tokens:ident) => { + $crate::__private::push_at(&mut $tokens); + }; + + (! $tokens:ident) => { + $crate::__private::push_bang(&mut $tokens); + }; + + (^ $tokens:ident) => { + $crate::__private::push_caret(&mut $tokens); + }; + + (^= $tokens:ident) => { + $crate::__private::push_caret_eq(&mut $tokens); + }; + + (/ $tokens:ident) => { + $crate::__private::push_div(&mut $tokens); + }; + + (/= $tokens:ident) => { + $crate::__private::push_div_eq(&mut $tokens); + }; + + (.. $tokens:ident) => { + $crate::__private::push_dot2(&mut $tokens); + }; + + (... $tokens:ident) => { + $crate::__private::push_dot3(&mut $tokens); + }; + + (..= $tokens:ident) => { + $crate::__private::push_dot_dot_eq(&mut $tokens); + }; + + (= $tokens:ident) => { + $crate::__private::push_eq(&mut $tokens); + }; + + (== $tokens:ident) => { + $crate::__private::push_eq_eq(&mut $tokens); + }; + + (>= $tokens:ident) => { + $crate::__private::push_ge(&mut $tokens); + }; + + (> $tokens:ident) => { + $crate::__private::push_gt(&mut $tokens); + }; + + (<= $tokens:ident) => { + $crate::__private::push_le(&mut $tokens); + }; + + (< $tokens:ident) => { + $crate::__private::push_lt(&mut $tokens); + }; + + (*= $tokens:ident) => { + $crate::__private::push_mul_eq(&mut $tokens); + }; + + (!= $tokens:ident) => { + $crate::__private::push_ne(&mut $tokens); + }; + + (| $tokens:ident) => { + $crate::__private::push_or(&mut $tokens); + }; + + (|= $tokens:ident) => { + $crate::__private::push_or_eq(&mut $tokens); + }; + + (|| $tokens:ident) => { + $crate::__private::push_or_or(&mut $tokens); + }; + + (? $tokens:ident) => { + $crate::__private::push_question(&mut $tokens); + }; + + (-> $tokens:ident) => { + $crate::__private::push_rarrow(&mut $tokens); + }; + + (<- $tokens:ident) => { + $crate::__private::push_larrow(&mut $tokens); + }; + + (% $tokens:ident) => { + $crate::__private::push_rem(&mut $tokens); + }; + + (%= $tokens:ident) => { + $crate::__private::push_rem_eq(&mut $tokens); + }; + + (=> $tokens:ident) => { + $crate::__private::push_fat_arrow(&mut $tokens); + }; + + (<< $tokens:ident) => { + $crate::__private::push_shl(&mut $tokens); + }; + + (<<= $tokens:ident) => { + $crate::__private::push_shl_eq(&mut $tokens); + }; + + (>> $tokens:ident) => { + $crate::__private::push_shr(&mut $tokens); + }; + + (>>= $tokens:ident) => { + $crate::__private::push_shr_eq(&mut $tokens); + }; + + (* $tokens:ident) => { + $crate::__private::push_star(&mut $tokens); + }; + + (- $tokens:ident) => { + $crate::__private::push_sub(&mut $tokens); + }; + + (-= $tokens:ident) => { + $crate::__private::push_sub_eq(&mut $tokens); + }; + + ($lifetime:lifetime $tokens:ident) => { + $crate::__private::push_lifetime(&mut $tokens, stringify!($lifetime)); + }; + + (_ $tokens:ident) => { + $crate::__private::push_underscore(&mut $tokens); + }; + + ($other:tt $tokens:ident) => { + $crate::__private::parse(&mut $tokens, stringify!($other)); + }; +} + +// See the comment above `quote_token!` about the rule ordering. +#[macro_export] +#[doc(hidden)] +macro_rules! quote_token_spanned { + ($ident:ident $tokens:ident $span:ident) => { + $crate::__private::push_ident_spanned(&mut $tokens, $span, stringify!($ident)); + }; + + (:: $tokens:ident $span:ident) => { + $crate::__private::push_colon2_spanned(&mut $tokens, $span); + }; + + (( $($inner:tt)* ) $tokens:ident $span:ident) => { + $crate::__private::push_group_spanned( + &mut $tokens, + $span, + $crate::__private::Delimiter::Parenthesis, + $crate::quote_spanned!($span=> $($inner)*), + ); + }; + + ([ $($inner:tt)* ] $tokens:ident $span:ident) => { + $crate::__private::push_group_spanned( + &mut $tokens, + $span, + $crate::__private::Delimiter::Bracket, + $crate::quote_spanned!($span=> $($inner)*), + ); + }; + + ({ $($inner:tt)* } $tokens:ident $span:ident) => { + $crate::__private::push_group_spanned( + &mut $tokens, + $span, + $crate::__private::Delimiter::Brace, + $crate::quote_spanned!($span=> $($inner)*), + ); + }; + + (# $tokens:ident $span:ident) => { + $crate::__private::push_pound_spanned(&mut $tokens, $span); + }; + + (, $tokens:ident $span:ident) => { + $crate::__private::push_comma_spanned(&mut $tokens, $span); + }; + + (. $tokens:ident $span:ident) => { + $crate::__private::push_dot_spanned(&mut $tokens, $span); + }; + + (; $tokens:ident $span:ident) => { + $crate::__private::push_semi_spanned(&mut $tokens, $span); + }; + + (: $tokens:ident $span:ident) => { + $crate::__private::push_colon_spanned(&mut $tokens, $span); + }; + + (+ $tokens:ident $span:ident) => { + $crate::__private::push_add_spanned(&mut $tokens, $span); + }; + + (+= $tokens:ident $span:ident) => { + $crate::__private::push_add_eq_spanned(&mut $tokens, $span); + }; + + (& $tokens:ident $span:ident) => { + $crate::__private::push_and_spanned(&mut $tokens, $span); + }; + + (&& $tokens:ident $span:ident) => { + $crate::__private::push_and_and_spanned(&mut $tokens, $span); + }; + + (&= $tokens:ident $span:ident) => { + $crate::__private::push_and_eq_spanned(&mut $tokens, $span); + }; + + (@ $tokens:ident $span:ident) => { + $crate::__private::push_at_spanned(&mut $tokens, $span); + }; + + (! $tokens:ident $span:ident) => { + $crate::__private::push_bang_spanned(&mut $tokens, $span); + }; + + (^ $tokens:ident $span:ident) => { + $crate::__private::push_caret_spanned(&mut $tokens, $span); + }; + + (^= $tokens:ident $span:ident) => { + $crate::__private::push_caret_eq_spanned(&mut $tokens, $span); + }; + + (/ $tokens:ident $span:ident) => { + $crate::__private::push_div_spanned(&mut $tokens, $span); + }; + + (/= $tokens:ident $span:ident) => { + $crate::__private::push_div_eq_spanned(&mut $tokens, $span); + }; + + (.. $tokens:ident $span:ident) => { + $crate::__private::push_dot2_spanned(&mut $tokens, $span); + }; + + (... $tokens:ident $span:ident) => { + $crate::__private::push_dot3_spanned(&mut $tokens, $span); + }; + + (..= $tokens:ident $span:ident) => { + $crate::__private::push_dot_dot_eq_spanned(&mut $tokens, $span); + }; + + (= $tokens:ident $span:ident) => { + $crate::__private::push_eq_spanned(&mut $tokens, $span); + }; + + (== $tokens:ident $span:ident) => { + $crate::__private::push_eq_eq_spanned(&mut $tokens, $span); + }; + + (>= $tokens:ident $span:ident) => { + $crate::__private::push_ge_spanned(&mut $tokens, $span); + }; + + (> $tokens:ident $span:ident) => { + $crate::__private::push_gt_spanned(&mut $tokens, $span); + }; + + (<= $tokens:ident $span:ident) => { + $crate::__private::push_le_spanned(&mut $tokens, $span); + }; + + (< $tokens:ident $span:ident) => { + $crate::__private::push_lt_spanned(&mut $tokens, $span); + }; + + (*= $tokens:ident $span:ident) => { + $crate::__private::push_mul_eq_spanned(&mut $tokens, $span); + }; + + (!= $tokens:ident $span:ident) => { + $crate::__private::push_ne_spanned(&mut $tokens, $span); + }; + + (| $tokens:ident $span:ident) => { + $crate::__private::push_or_spanned(&mut $tokens, $span); + }; + + (|= $tokens:ident $span:ident) => { + $crate::__private::push_or_eq_spanned(&mut $tokens, $span); + }; + + (|| $tokens:ident $span:ident) => { + $crate::__private::push_or_or_spanned(&mut $tokens, $span); + }; + + (? $tokens:ident $span:ident) => { + $crate::__private::push_question_spanned(&mut $tokens, $span); + }; + + (-> $tokens:ident $span:ident) => { + $crate::__private::push_rarrow_spanned(&mut $tokens, $span); + }; + + (<- $tokens:ident $span:ident) => { + $crate::__private::push_larrow_spanned(&mut $tokens, $span); + }; + + (% $tokens:ident $span:ident) => { + $crate::__private::push_rem_spanned(&mut $tokens, $span); + }; + + (%= $tokens:ident $span:ident) => { + $crate::__private::push_rem_eq_spanned(&mut $tokens, $span); + }; + + (=> $tokens:ident $span:ident) => { + $crate::__private::push_fat_arrow_spanned(&mut $tokens, $span); + }; + + (<< $tokens:ident $span:ident) => { + $crate::__private::push_shl_spanned(&mut $tokens, $span); + }; + + (<<= $tokens:ident $span:ident) => { + $crate::__private::push_shl_eq_spanned(&mut $tokens, $span); + }; + + (>> $tokens:ident $span:ident) => { + $crate::__private::push_shr_spanned(&mut $tokens, $span); + }; + + (>>= $tokens:ident $span:ident) => { + $crate::__private::push_shr_eq_spanned(&mut $tokens, $span); + }; + + (* $tokens:ident $span:ident) => { + $crate::__private::push_star_spanned(&mut $tokens, $span); + }; + + (- $tokens:ident $span:ident) => { + $crate::__private::push_sub_spanned(&mut $tokens, $span); + }; + + (-= $tokens:ident $span:ident) => { + $crate::__private::push_sub_eq_spanned(&mut $tokens, $span); + }; + + ($lifetime:lifetime $tokens:ident $span:ident) => { + $crate::__private::push_lifetime_spanned(&mut $tokens, $span, stringify!($lifetime)); + }; + + (_ $tokens:ident $span:ident) => { + $crate::__private::push_underscore_spanned(&mut $tokens, $span); + }; + + ($other:tt $tokens:ident $span:ident) => { + $crate::__private::parse_spanned(&mut $tokens, $span, stringify!($other)); + }; +} diff --git a/rust/quote/runtime.rs b/rust/quote/runtime.rs new file mode 100644 index 00000000000000..966025bc44abf1 --- /dev/null +++ b/rust/quote/runtime.rs @@ -0,0 +1,440 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::{IdentFragment, ToTokens, TokenStreamExt}; +use core::fmt; +use core::iter; +use core::ops::BitOr; + +pub use core::option::Option; +pub use proc_macro2::*; +pub use std::format; + +pub struct HasIterator; // True +pub struct ThereIsNoIteratorInRepetition; // False + +impl BitOr for ThereIsNoIteratorInRepetition { + type Output = ThereIsNoIteratorInRepetition; + fn bitor(self, _rhs: ThereIsNoIteratorInRepetition) -> ThereIsNoIteratorInRepetition { + ThereIsNoIteratorInRepetition + } +} + +impl BitOr for HasIterator { + type Output = HasIterator; + fn bitor(self, _rhs: ThereIsNoIteratorInRepetition) -> HasIterator { + HasIterator + } +} + +impl BitOr for ThereIsNoIteratorInRepetition { + type Output = HasIterator; + fn bitor(self, _rhs: HasIterator) -> HasIterator { + HasIterator + } +} + +impl BitOr for HasIterator { + type Output = HasIterator; + fn bitor(self, _rhs: HasIterator) -> HasIterator { + HasIterator + } +} + +/// Extension traits used by the implementation of `quote!`. These are defined +/// in separate traits, rather than as a single trait due to ambiguity issues. +/// +/// These traits expose a `quote_into_iter` method which should allow calling +/// whichever impl happens to be applicable. Calling that method repeatedly on +/// the returned value should be idempotent. +pub mod ext { + use super::RepInterp; + use super::{HasIterator as HasIter, ThereIsNoIteratorInRepetition as DoesNotHaveIter}; + use crate::ToTokens; + use core::slice; + use std::collections::btree_set::{self, BTreeSet}; + + /// Extension trait providing the `quote_into_iter` method on iterators. + pub trait RepIteratorExt: Iterator + Sized { + fn quote_into_iter(self) -> (Self, HasIter) { + (self, HasIter) + } + } + + impl RepIteratorExt for T {} + + /// Extension trait providing the `quote_into_iter` method for + /// non-iterable types. These types interpolate the same value in each + /// iteration of the repetition. + pub trait RepToTokensExt { + /// Pretend to be an iterator for the purposes of `quote_into_iter`. + /// This allows repeated calls to `quote_into_iter` to continue + /// correctly returning DoesNotHaveIter. + fn next(&self) -> Option<&Self> { + Some(self) + } + + fn quote_into_iter(&self) -> (&Self, DoesNotHaveIter) { + (self, DoesNotHaveIter) + } + } + + impl RepToTokensExt for T {} + + /// Extension trait providing the `quote_into_iter` method for types that + /// can be referenced as an iterator. + pub trait RepAsIteratorExt<'q> { + type Iter: Iterator; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter); + } + + impl<'q, 'a, T: RepAsIteratorExt<'q> + ?Sized> RepAsIteratorExt<'q> for &'a T { + type Iter = T::Iter; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + ::quote_into_iter(*self) + } + } + + impl<'q, 'a, T: RepAsIteratorExt<'q> + ?Sized> RepAsIteratorExt<'q> for &'a mut T { + type Iter = T::Iter; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + ::quote_into_iter(*self) + } + } + + impl<'q, T: 'q> RepAsIteratorExt<'q> for [T] { + type Iter = slice::Iter<'q, T>; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + (self.iter(), HasIter) + } + } + + impl<'q, T: 'q> RepAsIteratorExt<'q> for Vec { + type Iter = slice::Iter<'q, T>; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + (self.iter(), HasIter) + } + } + + impl<'q, T: 'q> RepAsIteratorExt<'q> for BTreeSet { + type Iter = btree_set::Iter<'q, T>; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + (self.iter(), HasIter) + } + } + + impl<'q, T: RepAsIteratorExt<'q>> RepAsIteratorExt<'q> for RepInterp { + type Iter = T::Iter; + + fn quote_into_iter(&'q self) -> (Self::Iter, HasIter) { + self.0.quote_into_iter() + } + } +} + +// Helper type used within interpolations to allow for repeated binding names. +// Implements the relevant traits, and exports a dummy `next()` method. +#[derive(Copy, Clone)] +pub struct RepInterp(pub T); + +impl RepInterp { + // This method is intended to look like `Iterator::next`, and is called when + // a name is bound multiple times, as the previous binding will shadow the + // original `Iterator` object. This allows us to avoid advancing the + // iterator multiple times per iteration. + pub fn next(self) -> Option { + Some(self.0) + } +} + +impl Iterator for RepInterp { + type Item = T::Item; + + fn next(&mut self) -> Option { + self.0.next() + } +} + +impl ToTokens for RepInterp { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.0.to_tokens(tokens); + } +} + +pub fn push_group(tokens: &mut TokenStream, delimiter: Delimiter, inner: TokenStream) { + tokens.append(Group::new(delimiter, inner)); +} + +pub fn push_group_spanned( + tokens: &mut TokenStream, + span: Span, + delimiter: Delimiter, + inner: TokenStream, +) { + let mut g = Group::new(delimiter, inner); + g.set_span(span); + tokens.append(g); +} + +pub fn parse(tokens: &mut TokenStream, s: &str) { + let s: TokenStream = s.parse().expect("invalid token stream"); + tokens.extend(iter::once(s)); +} + +pub fn parse_spanned(tokens: &mut TokenStream, span: Span, s: &str) { + let s: TokenStream = s.parse().expect("invalid token stream"); + tokens.extend(s.into_iter().map(|t| respan_token_tree(t, span))); +} + +// Token tree with every span replaced by the given one. +fn respan_token_tree(mut token: TokenTree, span: Span) -> TokenTree { + match &mut token { + TokenTree::Group(g) => { + let stream = g + .stream() + .into_iter() + .map(|token| respan_token_tree(token, span)) + .collect(); + *g = Group::new(g.delimiter(), stream); + g.set_span(span); + } + other => other.set_span(span), + } + token +} + +pub fn push_ident(tokens: &mut TokenStream, s: &str) { + let span = Span::call_site(); + push_ident_spanned(tokens, span, s); +} + +pub fn push_ident_spanned(tokens: &mut TokenStream, span: Span, s: &str) { + tokens.append(ident_maybe_raw(s, span)); +} + +pub fn push_lifetime(tokens: &mut TokenStream, lifetime: &str) { + struct Lifetime<'a> { + name: &'a str, + state: u8, + } + + impl<'a> Iterator for Lifetime<'a> { + type Item = TokenTree; + + fn next(&mut self) -> Option { + match self.state { + 0 => { + self.state = 1; + Some(TokenTree::Punct(Punct::new('\'', Spacing::Joint))) + } + 1 => { + self.state = 2; + Some(TokenTree::Ident(Ident::new(self.name, Span::call_site()))) + } + _ => None, + } + } + } + + tokens.extend(Lifetime { + name: &lifetime[1..], + state: 0, + }); +} + +pub fn push_lifetime_spanned(tokens: &mut TokenStream, span: Span, lifetime: &str) { + struct Lifetime<'a> { + name: &'a str, + span: Span, + state: u8, + } + + impl<'a> Iterator for Lifetime<'a> { + type Item = TokenTree; + + fn next(&mut self) -> Option { + match self.state { + 0 => { + self.state = 1; + let mut apostrophe = Punct::new('\'', Spacing::Joint); + apostrophe.set_span(self.span); + Some(TokenTree::Punct(apostrophe)) + } + 1 => { + self.state = 2; + Some(TokenTree::Ident(Ident::new(self.name, self.span))) + } + _ => None, + } + } + } + + tokens.extend(Lifetime { + name: &lifetime[1..], + span, + state: 0, + }); +} + +macro_rules! push_punct { + ($name:ident $spanned:ident $char1:tt) => { + pub fn $name(tokens: &mut TokenStream) { + tokens.append(Punct::new($char1, Spacing::Alone)); + } + pub fn $spanned(tokens: &mut TokenStream, span: Span) { + let mut punct = Punct::new($char1, Spacing::Alone); + punct.set_span(span); + tokens.append(punct); + } + }; + ($name:ident $spanned:ident $char1:tt $char2:tt) => { + pub fn $name(tokens: &mut TokenStream) { + tokens.append(Punct::new($char1, Spacing::Joint)); + tokens.append(Punct::new($char2, Spacing::Alone)); + } + pub fn $spanned(tokens: &mut TokenStream, span: Span) { + let mut punct = Punct::new($char1, Spacing::Joint); + punct.set_span(span); + tokens.append(punct); + let mut punct = Punct::new($char2, Spacing::Alone); + punct.set_span(span); + tokens.append(punct); + } + }; + ($name:ident $spanned:ident $char1:tt $char2:tt $char3:tt) => { + pub fn $name(tokens: &mut TokenStream) { + tokens.append(Punct::new($char1, Spacing::Joint)); + tokens.append(Punct::new($char2, Spacing::Joint)); + tokens.append(Punct::new($char3, Spacing::Alone)); + } + pub fn $spanned(tokens: &mut TokenStream, span: Span) { + let mut punct = Punct::new($char1, Spacing::Joint); + punct.set_span(span); + tokens.append(punct); + let mut punct = Punct::new($char2, Spacing::Joint); + punct.set_span(span); + tokens.append(punct); + let mut punct = Punct::new($char3, Spacing::Alone); + punct.set_span(span); + tokens.append(punct); + } + }; +} + +push_punct!(push_add push_add_spanned '+'); +push_punct!(push_add_eq push_add_eq_spanned '+' '='); +push_punct!(push_and push_and_spanned '&'); +push_punct!(push_and_and push_and_and_spanned '&' '&'); +push_punct!(push_and_eq push_and_eq_spanned '&' '='); +push_punct!(push_at push_at_spanned '@'); +push_punct!(push_bang push_bang_spanned '!'); +push_punct!(push_caret push_caret_spanned '^'); +push_punct!(push_caret_eq push_caret_eq_spanned '^' '='); +push_punct!(push_colon push_colon_spanned ':'); +push_punct!(push_colon2 push_colon2_spanned ':' ':'); +push_punct!(push_comma push_comma_spanned ','); +push_punct!(push_div push_div_spanned '/'); +push_punct!(push_div_eq push_div_eq_spanned '/' '='); +push_punct!(push_dot push_dot_spanned '.'); +push_punct!(push_dot2 push_dot2_spanned '.' '.'); +push_punct!(push_dot3 push_dot3_spanned '.' '.' '.'); +push_punct!(push_dot_dot_eq push_dot_dot_eq_spanned '.' '.' '='); +push_punct!(push_eq push_eq_spanned '='); +push_punct!(push_eq_eq push_eq_eq_spanned '=' '='); +push_punct!(push_ge push_ge_spanned '>' '='); +push_punct!(push_gt push_gt_spanned '>'); +push_punct!(push_le push_le_spanned '<' '='); +push_punct!(push_lt push_lt_spanned '<'); +push_punct!(push_mul_eq push_mul_eq_spanned '*' '='); +push_punct!(push_ne push_ne_spanned '!' '='); +push_punct!(push_or push_or_spanned '|'); +push_punct!(push_or_eq push_or_eq_spanned '|' '='); +push_punct!(push_or_or push_or_or_spanned '|' '|'); +push_punct!(push_pound push_pound_spanned '#'); +push_punct!(push_question push_question_spanned '?'); +push_punct!(push_rarrow push_rarrow_spanned '-' '>'); +push_punct!(push_larrow push_larrow_spanned '<' '-'); +push_punct!(push_rem push_rem_spanned '%'); +push_punct!(push_rem_eq push_rem_eq_spanned '%' '='); +push_punct!(push_fat_arrow push_fat_arrow_spanned '=' '>'); +push_punct!(push_semi push_semi_spanned ';'); +push_punct!(push_shl push_shl_spanned '<' '<'); +push_punct!(push_shl_eq push_shl_eq_spanned '<' '<' '='); +push_punct!(push_shr push_shr_spanned '>' '>'); +push_punct!(push_shr_eq push_shr_eq_spanned '>' '>' '='); +push_punct!(push_star push_star_spanned '*'); +push_punct!(push_sub push_sub_spanned '-'); +push_punct!(push_sub_eq push_sub_eq_spanned '-' '='); + +pub fn push_underscore(tokens: &mut TokenStream) { + push_underscore_spanned(tokens, Span::call_site()); +} + +pub fn push_underscore_spanned(tokens: &mut TokenStream, span: Span) { + tokens.append(Ident::new("_", span)); +} + +// Helper method for constructing identifiers from the `format_ident!` macro, +// handling `r#` prefixes. +pub fn mk_ident(id: &str, span: Option) -> Ident { + let span = span.unwrap_or_else(Span::call_site); + ident_maybe_raw(id, span) +} + +fn ident_maybe_raw(id: &str, span: Span) -> Ident { + if id.starts_with("r#") { + Ident::new_raw(&id[2..], span) + } else { + Ident::new(id, span) + } +} + +// Adapts from `IdentFragment` to `fmt::Display` for use by the `format_ident!` +// macro, and exposes span information from these fragments. +// +// This struct also has forwarding implementations of the formatting traits +// `Octal`, `LowerHex`, `UpperHex`, and `Binary` to allow for their use within +// `format_ident!`. +#[derive(Copy, Clone)] +pub struct IdentFragmentAdapter(pub T); + +impl IdentFragmentAdapter { + pub fn span(&self) -> Option { + self.0.span() + } +} + +impl fmt::Display for IdentFragmentAdapter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + IdentFragment::fmt(&self.0, f) + } +} + +impl fmt::Octal for IdentFragmentAdapter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + fmt::Octal::fmt(&self.0, f) + } +} + +impl fmt::LowerHex for IdentFragmentAdapter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + fmt::LowerHex::fmt(&self.0, f) + } +} + +impl fmt::UpperHex for IdentFragmentAdapter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + fmt::UpperHex::fmt(&self.0, f) + } +} + +impl fmt::Binary for IdentFragmentAdapter { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + fmt::Binary::fmt(&self.0, f) + } +} diff --git a/rust/quote/spanned.rs b/rust/quote/spanned.rs new file mode 100644 index 00000000000000..aac63de0ba2ffb --- /dev/null +++ b/rust/quote/spanned.rs @@ -0,0 +1,45 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::ToTokens; +use proc_macro2::{Span, TokenStream}; + +pub trait Spanned { + fn __span(&self) -> Span; +} + +impl Spanned for Span { + fn __span(&self) -> Span { + *self + } +} + +impl Spanned for T { + fn __span(&self) -> Span { + join_spans(self.into_token_stream()) + } +} + +fn join_spans(tokens: TokenStream) -> Span { + #[cfg(not(needs_invalid_span_workaround))] + let mut iter = tokens.into_iter().map(|tt| tt.span()); + + #[cfg(needs_invalid_span_workaround)] + let mut iter = tokens.into_iter().filter_map(|tt| { + let span = tt.span(); + let debug = format!("{:?}", span); + if debug.ends_with("bytes(0..0)") { + None + } else { + Some(span) + } + }); + + let first = match iter.next() { + Some(span) => span, + None => return Span::call_site(), + }; + + iter.fold(None, |_prev, next| Some(next)) + .and_then(|last| first.join(last)) + .unwrap_or(first) +} diff --git a/rust/quote/to_tokens.rs b/rust/quote/to_tokens.rs new file mode 100644 index 00000000000000..a91ba61573e044 --- /dev/null +++ b/rust/quote/to_tokens.rs @@ -0,0 +1,211 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::TokenStreamExt; +use core::iter; +use proc_macro2::{Group, Ident, Literal, Punct, Span, TokenStream, TokenTree}; +use std::borrow::Cow; +use std::rc::Rc; + +/// Types that can be interpolated inside a `quote!` invocation. +/// +/// [`quote!`]: macro.quote.html +pub trait ToTokens { + /// Write `self` to the given `TokenStream`. + /// + /// The token append methods provided by the [`TokenStreamExt`] extension + /// trait may be useful for implementing `ToTokens`. + /// + /// [`TokenStreamExt`]: trait.TokenStreamExt.html + /// + /// # Example + /// + /// Example implementation for a struct representing Rust paths like + /// `std::cmp::PartialEq`: + /// + /// ``` + /// use proc_macro2::{TokenTree, Spacing, Span, Punct, TokenStream}; + /// use quote::{TokenStreamExt, ToTokens}; + /// + /// pub struct Path { + /// pub global: bool, + /// pub segments: Vec, + /// } + /// + /// impl ToTokens for Path { + /// fn to_tokens(&self, tokens: &mut TokenStream) { + /// for (i, segment) in self.segments.iter().enumerate() { + /// if i > 0 || self.global { + /// // Double colon `::` + /// tokens.append(Punct::new(':', Spacing::Joint)); + /// tokens.append(Punct::new(':', Spacing::Alone)); + /// } + /// segment.to_tokens(tokens); + /// } + /// } + /// } + /// # + /// # pub struct PathSegment; + /// # + /// # impl ToTokens for PathSegment { + /// # fn to_tokens(&self, tokens: &mut TokenStream) { + /// # unimplemented!() + /// # } + /// # } + /// ``` + fn to_tokens(&self, tokens: &mut TokenStream); + + /// Convert `self` directly into a `TokenStream` object. + /// + /// This method is implicitly implemented using `to_tokens`, and acts as a + /// convenience method for consumers of the `ToTokens` trait. + fn to_token_stream(&self) -> TokenStream { + let mut tokens = TokenStream::new(); + self.to_tokens(&mut tokens); + tokens + } + + /// Convert `self` directly into a `TokenStream` object. + /// + /// This method is implicitly implemented using `to_tokens`, and acts as a + /// convenience method for consumers of the `ToTokens` trait. + fn into_token_stream(self) -> TokenStream + where + Self: Sized, + { + self.to_token_stream() + } +} + +impl<'a, T: ?Sized + ToTokens> ToTokens for &'a T { + fn to_tokens(&self, tokens: &mut TokenStream) { + (**self).to_tokens(tokens); + } +} + +impl<'a, T: ?Sized + ToTokens> ToTokens for &'a mut T { + fn to_tokens(&self, tokens: &mut TokenStream) { + (**self).to_tokens(tokens); + } +} + +impl<'a, T: ?Sized + ToOwned + ToTokens> ToTokens for Cow<'a, T> { + fn to_tokens(&self, tokens: &mut TokenStream) { + (**self).to_tokens(tokens); + } +} + +impl ToTokens for Box { + fn to_tokens(&self, tokens: &mut TokenStream) { + (**self).to_tokens(tokens); + } +} + +impl ToTokens for Rc { + fn to_tokens(&self, tokens: &mut TokenStream) { + (**self).to_tokens(tokens); + } +} + +impl ToTokens for Option { + fn to_tokens(&self, tokens: &mut TokenStream) { + if let Some(ref t) = *self { + t.to_tokens(tokens); + } + } +} + +impl ToTokens for str { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(Literal::string(self)); + } +} + +impl ToTokens for String { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.as_str().to_tokens(tokens); + } +} + +macro_rules! primitive { + ($($t:ident => $name:ident)*) => { + $( + impl ToTokens for $t { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(Literal::$name(*self)); + } + } + )* + }; +} + +primitive! { + i8 => i8_suffixed + i16 => i16_suffixed + i32 => i32_suffixed + i64 => i64_suffixed + i128 => i128_suffixed + isize => isize_suffixed + + u8 => u8_suffixed + u16 => u16_suffixed + u32 => u32_suffixed + u64 => u64_suffixed + u128 => u128_suffixed + usize => usize_suffixed + + f32 => f32_suffixed + f64 => f64_suffixed +} + +impl ToTokens for char { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(Literal::character(*self)); + } +} + +impl ToTokens for bool { + fn to_tokens(&self, tokens: &mut TokenStream) { + let word = if *self { "true" } else { "false" }; + tokens.append(Ident::new(word, Span::call_site())); + } +} + +impl ToTokens for Group { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(self.clone()); + } +} + +impl ToTokens for Ident { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(self.clone()); + } +} + +impl ToTokens for Punct { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(self.clone()); + } +} + +impl ToTokens for Literal { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(self.clone()); + } +} + +impl ToTokens for TokenTree { + fn to_tokens(&self, dst: &mut TokenStream) { + dst.append(self.clone()); + } +} + +impl ToTokens for TokenStream { + fn to_tokens(&self, dst: &mut TokenStream) { + dst.extend(iter::once(self.clone())); + } + + fn into_token_stream(self) -> TokenStream { + self + } +} diff --git a/rust/serde/de/format.rs b/rust/serde/de/format.rs new file mode 100644 index 00000000000000..33233766fc27f6 --- /dev/null +++ b/rust/serde/de/format.rs @@ -0,0 +1,32 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::fmt::{self, Write}; +use lib::str; + +pub(super) struct Buf<'a> { + bytes: &'a mut [u8], + offset: usize, +} + +impl<'a> Buf<'a> { + pub fn new(bytes: &'a mut [u8]) -> Self { + Buf { bytes, offset: 0 } + } + + pub fn as_str(&self) -> &str { + let slice = &self.bytes[..self.offset]; + unsafe { str::from_utf8_unchecked(slice) } + } +} + +impl<'a> Write for Buf<'a> { + fn write_str(&mut self, s: &str) -> fmt::Result { + if self.offset + s.len() > self.bytes.len() { + Err(fmt::Error) + } else { + self.bytes[self.offset..self.offset + s.len()].copy_from_slice(s.as_bytes()); + self.offset += s.len(); + Ok(()) + } + } +} diff --git a/rust/serde/de/ignored_any.rs b/rust/serde/de/ignored_any.rs new file mode 100644 index 00000000000000..d2cc83978b50bd --- /dev/null +++ b/rust/serde/de/ignored_any.rs @@ -0,0 +1,246 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +use de::{ + Deserialize, Deserializer, EnumAccess, Error, MapAccess, SeqAccess, VariantAccess, Visitor, +}; + +/// An efficient way of discarding data from a deserializer. +/// +/// Think of this like `serde_json::Value` in that it can be deserialized from +/// any type, except that it does not store any information about the data that +/// gets deserialized. +/// +/// ```edition2018 +/// use std::fmt; +/// use std::marker::PhantomData; +/// +/// use serde::de::{ +/// self, Deserialize, DeserializeSeed, Deserializer, IgnoredAny, SeqAccess, Visitor, +/// }; +/// +/// /// A seed that can be used to deserialize only the `n`th element of a sequence +/// /// while efficiently discarding elements of any type before or after index `n`. +/// /// +/// /// For example to deserialize only the element at index 3: +/// /// +/// /// ``` +/// /// NthElement::new(3).deserialize(deserializer) +/// /// ``` +/// pub struct NthElement { +/// n: usize, +/// marker: PhantomData, +/// } +/// +/// impl NthElement { +/// pub fn new(n: usize) -> Self { +/// NthElement { +/// n: n, +/// marker: PhantomData, +/// } +/// } +/// } +/// +/// impl<'de, T> Visitor<'de> for NthElement +/// where +/// T: Deserialize<'de>, +/// { +/// type Value = T; +/// +/// fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// write!( +/// formatter, +/// "a sequence in which we care about element {}", +/// self.n +/// ) +/// } +/// +/// fn visit_seq(self, mut seq: A) -> Result +/// where +/// A: SeqAccess<'de>, +/// { +/// // Skip over the first `n` elements. +/// for i in 0..self.n { +/// // It is an error if the sequence ends before we get to element `n`. +/// if seq.next_element::()?.is_none() { +/// return Err(de::Error::invalid_length(i, &self)); +/// } +/// } +/// +/// // Deserialize the one we care about. +/// let nth = match seq.next_element()? { +/// Some(nth) => nth, +/// None => { +/// return Err(de::Error::invalid_length(self.n, &self)); +/// } +/// }; +/// +/// // Skip over any remaining elements in the sequence after `n`. +/// while let Some(IgnoredAny) = seq.next_element()? { +/// // ignore +/// } +/// +/// Ok(nth) +/// } +/// } +/// +/// impl<'de, T> DeserializeSeed<'de> for NthElement +/// where +/// T: Deserialize<'de>, +/// { +/// type Value = T; +/// +/// fn deserialize(self, deserializer: D) -> Result +/// where +/// D: Deserializer<'de>, +/// { +/// deserializer.deserialize_seq(self) +/// } +/// } +/// +/// # fn example<'de, D>(deserializer: D) -> Result<(), D::Error> +/// # where +/// # D: Deserializer<'de>, +/// # { +/// // Deserialize only the sequence element at index 3 from this deserializer. +/// // The element at index 3 is required to be a string. Elements before and +/// // after index 3 are allowed to be of any type. +/// let s: String = NthElement::new(3).deserialize(deserializer)?; +/// # Ok(()) +/// # } +/// ``` +#[derive(Copy, Clone, Debug, Default)] +pub struct IgnoredAny; + +impl<'de> Visitor<'de> for IgnoredAny { + type Value = IgnoredAny; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("anything at all") + } + + #[inline] + fn visit_bool(self, x: bool) -> Result { + let _ = x; + Ok(IgnoredAny) + } + + #[inline] + fn visit_i64(self, x: i64) -> Result { + let _ = x; + Ok(IgnoredAny) + } + + serde_if_integer128! { + #[inline] + fn visit_i128(self, x: i128) -> Result { + let _ = x; + Ok(IgnoredAny) + } + } + + #[inline] + fn visit_u64(self, x: u64) -> Result { + let _ = x; + Ok(IgnoredAny) + } + + serde_if_integer128! { + #[inline] + fn visit_u128(self, x: u128) -> Result { + let _ = x; + Ok(IgnoredAny) + } + } + + #[cfg(not(no_fp_fmt_parse))] + #[inline] + fn visit_f64(self, x: f64) -> Result { + let _ = x; + Ok(IgnoredAny) + } + + #[inline] + fn visit_str(self, s: &str) -> Result + where + E: Error, + { + let _ = s; + Ok(IgnoredAny) + } + + #[inline] + fn visit_none(self) -> Result { + Ok(IgnoredAny) + } + + #[inline] + fn visit_some(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + IgnoredAny::deserialize(deserializer) + } + + #[inline] + fn visit_newtype_struct(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + IgnoredAny::deserialize(deserializer) + } + + #[inline] + fn visit_unit(self) -> Result { + Ok(IgnoredAny) + } + + #[inline] + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + while let Some(IgnoredAny) = try!(seq.next_element()) { + // Gobble + } + Ok(IgnoredAny) + } + + #[inline] + fn visit_map(self, mut map: A) -> Result + where + A: MapAccess<'de>, + { + while let Some((IgnoredAny, IgnoredAny)) = try!(map.next_entry()) { + // Gobble + } + Ok(IgnoredAny) + } + + #[inline] + fn visit_bytes(self, bytes: &[u8]) -> Result + where + E: Error, + { + let _ = bytes; + Ok(IgnoredAny) + } + + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + try!(data.variant::()).1.newtype_variant() + } +} + +impl<'de> Deserialize<'de> for IgnoredAny { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_ignored_any(IgnoredAny) + } +} diff --git a/rust/serde/de/impls.rs b/rust/serde/de/impls.rs new file mode 100644 index 00000000000000..9377f5eee65086 --- /dev/null +++ b/rust/serde/de/impls.rs @@ -0,0 +1,2740 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +use de::{ + Deserialize, Deserializer, EnumAccess, Error, SeqAccess, Unexpected, VariantAccess, Visitor, +}; + +#[cfg(any(feature = "std", feature = "alloc", not(no_core_duration)))] +use de::MapAccess; + +use seed::InPlaceSeed; + +#[cfg(any(feature = "std", feature = "alloc"))] +use __private::size_hint; + +//////////////////////////////////////////////////////////////////////////////// + +struct UnitVisitor; + +impl<'de> Visitor<'de> for UnitVisitor { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("unit") + } + + fn visit_unit(self) -> Result + where + E: Error, + { + Ok(()) + } +} + +impl<'de> Deserialize<'de> for () { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_unit(UnitVisitor) + } +} + +#[cfg(feature = "unstable")] +impl<'de> Deserialize<'de> for ! { + fn deserialize(_deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Err(Error::custom("cannot deserialize `!`")) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +struct BoolVisitor; + +impl<'de> Visitor<'de> for BoolVisitor { + type Value = bool; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a boolean") + } + + fn visit_bool(self, v: bool) -> Result + where + E: Error, + { + Ok(v) + } +} + +impl<'de> Deserialize<'de> for bool { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_bool(BoolVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! impl_deserialize_num { + ($primitive:ident, $nonzero:ident $(cfg($($cfg:tt)*))*, $deserialize:ident $($method:ident!($($val:ident : $visit:ident)*);)*) => { + impl_deserialize_num!($primitive, $deserialize $($method!($($val : $visit)*);)*); + + #[cfg(all(not(no_num_nonzero), $($($cfg)*)*))] + impl<'de> Deserialize<'de> for num::$nonzero { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct NonZeroVisitor; + + impl<'de> Visitor<'de> for NonZeroVisitor { + type Value = num::$nonzero; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("a nonzero ", stringify!($primitive))) + } + + $($($method!(nonzero $primitive $val : $visit);)*)* + } + + deserializer.$deserialize(NonZeroVisitor) + } + } + }; + + ($primitive:ident, $deserialize:ident $($method:ident!($($val:ident : $visit:ident)*);)*) => { + impl<'de> Deserialize<'de> for $primitive { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct PrimitiveVisitor; + + impl<'de> Visitor<'de> for PrimitiveVisitor { + type Value = $primitive; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(stringify!($primitive)) + } + + $($($method!($val : $visit);)*)* + } + + deserializer.$deserialize(PrimitiveVisitor) + } + } + }; +} + +macro_rules! num_self { + ($ty:ident : $visit:ident) => { + #[inline] + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + Ok(v) + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if let Some(nonzero) = Self::Value::new(v) { + Ok(nonzero) + } else { + Err(Error::invalid_value(Unexpected::Unsigned(0), &self)) + } + } + }; +} + +macro_rules! num_as_self { + ($ty:ident : $visit:ident) => { + #[inline] + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + Ok(v as Self::Value) + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if let Some(nonzero) = Self::Value::new(v as $primitive) { + Ok(nonzero) + } else { + Err(Error::invalid_value(Unexpected::Unsigned(0), &self)) + } + } + }; +} + +macro_rules! int_to_int { + ($ty:ident : $visit:ident) => { + #[inline] + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if Self::Value::min_value() as i64 <= v as i64 + && v as i64 <= Self::Value::max_value() as i64 + { + Ok(v as Self::Value) + } else { + Err(Error::invalid_value(Unexpected::Signed(v as i64), &self)) + } + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if $primitive::min_value() as i64 <= v as i64 + && v as i64 <= $primitive::max_value() as i64 + { + if let Some(nonzero) = Self::Value::new(v as $primitive) { + return Ok(nonzero); + } + } + Err(Error::invalid_value(Unexpected::Signed(v as i64), &self)) + } + }; +} + +macro_rules! int_to_uint { + ($ty:ident : $visit:ident) => { + #[inline] + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if 0 <= v && v as u64 <= Self::Value::max_value() as u64 { + Ok(v as Self::Value) + } else { + Err(Error::invalid_value(Unexpected::Signed(v as i64), &self)) + } + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if 0 < v && v as u64 <= $primitive::max_value() as u64 { + if let Some(nonzero) = Self::Value::new(v as $primitive) { + return Ok(nonzero); + } + } + Err(Error::invalid_value(Unexpected::Signed(v as i64), &self)) + } + }; +} + +macro_rules! uint_to_self { + ($ty:ident : $visit:ident) => { + #[inline] + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if v as u64 <= Self::Value::max_value() as u64 { + Ok(v as Self::Value) + } else { + Err(Error::invalid_value(Unexpected::Unsigned(v as u64), &self)) + } + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if v as u64 <= $primitive::max_value() as u64 { + if let Some(nonzero) = Self::Value::new(v as $primitive) { + return Ok(nonzero); + } + } + Err(Error::invalid_value(Unexpected::Unsigned(v as u64), &self)) + } + }; +} + +impl_deserialize_num! { + i8, NonZeroI8 cfg(not(no_num_nonzero_signed)), deserialize_i8 + num_self!(i8:visit_i8); + int_to_int!(i16:visit_i16 i32:visit_i32 i64:visit_i64); + uint_to_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + i16, NonZeroI16 cfg(not(no_num_nonzero_signed)), deserialize_i16 + num_self!(i16:visit_i16); + num_as_self!(i8:visit_i8); + int_to_int!(i32:visit_i32 i64:visit_i64); + uint_to_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + i32, NonZeroI32 cfg(not(no_num_nonzero_signed)), deserialize_i32 + num_self!(i32:visit_i32); + num_as_self!(i8:visit_i8 i16:visit_i16); + int_to_int!(i64:visit_i64); + uint_to_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + i64, NonZeroI64 cfg(not(no_num_nonzero_signed)), deserialize_i64 + num_self!(i64:visit_i64); + num_as_self!(i8:visit_i8 i16:visit_i16 i32:visit_i32); + uint_to_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + isize, NonZeroIsize cfg(not(no_num_nonzero_signed)), deserialize_i64 + num_as_self!(i8:visit_i8 i16:visit_i16); + int_to_int!(i32:visit_i32 i64:visit_i64); + uint_to_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + u8, NonZeroU8, deserialize_u8 + num_self!(u8:visit_u8); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + uint_to_self!(u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + u16, NonZeroU16, deserialize_u16 + num_self!(u16:visit_u16); + num_as_self!(u8:visit_u8); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + uint_to_self!(u32:visit_u32 u64:visit_u64); +} + +impl_deserialize_num! { + u32, NonZeroU32, deserialize_u32 + num_self!(u32:visit_u32); + num_as_self!(u8:visit_u8 u16:visit_u16); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + uint_to_self!(u64:visit_u64); +} + +impl_deserialize_num! { + u64, NonZeroU64, deserialize_u64 + num_self!(u64:visit_u64); + num_as_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); +} + +impl_deserialize_num! { + usize, NonZeroUsize, deserialize_u64 + num_as_self!(u8:visit_u8 u16:visit_u16); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + uint_to_self!(u32:visit_u32 u64:visit_u64); +} + +#[cfg(not(no_fp_fmt_parse))] +impl_deserialize_num! { + f32, deserialize_f32 + num_self!(f32:visit_f32); + num_as_self!(f64:visit_f64); + num_as_self!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + num_as_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +#[cfg(not(no_fp_fmt_parse))] +impl_deserialize_num! { + f64, deserialize_f64 + num_self!(f64:visit_f64); + num_as_self!(f32:visit_f32); + num_as_self!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + num_as_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); +} + +serde_if_integer128! { + macro_rules! num_128 { + ($ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if v as i128 >= Self::Value::min_value() as i128 + && v as u128 <= Self::Value::max_value() as u128 + { + Ok(v as Self::Value) + } else { + Err(Error::invalid_value( + Unexpected::Other(stringify!($ty)), + &self, + )) + } + } + }; + + (nonzero $primitive:ident $ty:ident : $visit:ident) => { + fn $visit(self, v: $ty) -> Result + where + E: Error, + { + if v as i128 >= $primitive::min_value() as i128 + && v as u128 <= $primitive::max_value() as u128 + { + if let Some(nonzero) = Self::Value::new(v as $primitive) { + Ok(nonzero) + } else { + Err(Error::invalid_value(Unexpected::Unsigned(0), &self)) + } + } else { + Err(Error::invalid_value( + Unexpected::Other(stringify!($ty)), + &self, + )) + } + } + }; + } + + impl_deserialize_num! { + i128, NonZeroI128 cfg(not(no_num_nonzero_signed)), deserialize_i128 + num_self!(i128:visit_i128); + num_as_self!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + num_as_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); + num_128!(u128:visit_u128); + } + + impl_deserialize_num! { + u128, NonZeroU128, deserialize_u128 + num_self!(u128:visit_u128); + num_as_self!(u8:visit_u8 u16:visit_u16 u32:visit_u32 u64:visit_u64); + int_to_uint!(i8:visit_i8 i16:visit_i16 i32:visit_i32 i64:visit_i64); + num_128!(i128:visit_i128); + } +} + +//////////////////////////////////////////////////////////////////////////////// + +struct CharVisitor; + +impl<'de> Visitor<'de> for CharVisitor { + type Value = char; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a character") + } + + #[inline] + fn visit_char(self, v: char) -> Result + where + E: Error, + { + Ok(v) + } + + #[inline] + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + let mut iter = v.chars(); + match (iter.next(), iter.next()) { + (Some(c), None) => Ok(c), + _ => Err(Error::invalid_value(Unexpected::Str(v), &self)), + } + } +} + +impl<'de> Deserialize<'de> for char { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_char(CharVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +struct StringVisitor; +#[cfg(any(feature = "std", feature = "alloc"))] +struct StringInPlaceVisitor<'a>(&'a mut String); + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de> Visitor<'de> for StringVisitor { + type Value = String; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a string") + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + Ok(v.to_owned()) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + Ok(v) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + match str::from_utf8(v) { + Ok(s) => Ok(s.to_owned()), + Err(_) => Err(Error::invalid_value(Unexpected::Bytes(v), &self)), + } + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + match String::from_utf8(v) { + Ok(s) => Ok(s), + Err(e) => Err(Error::invalid_value( + Unexpected::Bytes(&e.into_bytes()), + &self, + )), + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de> Visitor<'de> for StringInPlaceVisitor<'a> { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a string") + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + self.0.clear(); + self.0.push_str(v); + Ok(()) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + *self.0 = v; + Ok(()) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + match str::from_utf8(v) { + Ok(s) => { + self.0.clear(); + self.0.push_str(s); + Ok(()) + } + Err(_) => Err(Error::invalid_value(Unexpected::Bytes(v), &self)), + } + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + match String::from_utf8(v) { + Ok(s) => { + *self.0 = s; + Ok(()) + } + Err(e) => Err(Error::invalid_value( + Unexpected::Bytes(&e.into_bytes()), + &self, + )), + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de> Deserialize<'de> for String { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_string(StringVisitor) + } + + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + deserializer.deserialize_string(StringInPlaceVisitor(place)) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +struct StrVisitor; + +impl<'a> Visitor<'a> for StrVisitor { + type Value = &'a str; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a borrowed string") + } + + fn visit_borrowed_str(self, v: &'a str) -> Result + where + E: Error, + { + Ok(v) // so easy + } + + fn visit_borrowed_bytes(self, v: &'a [u8]) -> Result + where + E: Error, + { + str::from_utf8(v).map_err(|_| Error::invalid_value(Unexpected::Bytes(v), &self)) + } +} + +impl<'de: 'a, 'a> Deserialize<'de> for &'a str { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(StrVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +struct BytesVisitor; + +impl<'a> Visitor<'a> for BytesVisitor { + type Value = &'a [u8]; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a borrowed byte array") + } + + fn visit_borrowed_bytes(self, v: &'a [u8]) -> Result + where + E: Error, + { + Ok(v) + } + + fn visit_borrowed_str(self, v: &'a str) -> Result + where + E: Error, + { + Ok(v.as_bytes()) + } +} + +impl<'de: 'a, 'a> Deserialize<'de> for &'a [u8] { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_bytes(BytesVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", all(not(no_core_cstr), feature = "alloc")))] +struct CStringVisitor; + +#[cfg(any(feature = "std", all(not(no_core_cstr), feature = "alloc")))] +impl<'de> Visitor<'de> for CStringVisitor { + type Value = CString; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("byte array") + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let len = size_hint::cautious(seq.size_hint()); + let mut values = Vec::with_capacity(len); + + while let Some(value) = try!(seq.next_element()) { + values.push(value); + } + + CString::new(values).map_err(Error::custom) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + CString::new(v).map_err(Error::custom) + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + CString::new(v).map_err(Error::custom) + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + CString::new(v).map_err(Error::custom) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + CString::new(v).map_err(Error::custom) + } +} + +#[cfg(any(feature = "std", all(not(no_core_cstr), feature = "alloc")))] +impl<'de> Deserialize<'de> for CString { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_byte_buf(CStringVisitor) + } +} + +macro_rules! forwarded_impl { + ( + $(#[doc = $doc:tt])* + ($($id:ident),*), $ty:ty, $func:expr + ) => { + $(#[doc = $doc])* + impl<'de $(, $id : Deserialize<'de>,)*> Deserialize<'de> for $ty { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Deserialize::deserialize(deserializer).map($func) + } + } + } +} + +#[cfg(all( + any(feature = "std", all(not(no_core_cstr), feature = "alloc")), + not(no_de_boxed_c_str) +))] +forwarded_impl!((), Box, CString::into_boxed_c_str); + +#[cfg(not(no_core_reverse))] +forwarded_impl!((T), Reverse, Reverse); + +//////////////////////////////////////////////////////////////////////////////// + +struct OptionVisitor { + marker: PhantomData, +} + +impl<'de, T> Visitor<'de> for OptionVisitor +where + T: Deserialize<'de>, +{ + type Value = Option; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("option") + } + + #[inline] + fn visit_unit(self) -> Result + where + E: Error, + { + Ok(None) + } + + #[inline] + fn visit_none(self) -> Result + where + E: Error, + { + Ok(None) + } + + #[inline] + fn visit_some(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + T::deserialize(deserializer).map(Some) + } + + fn __private_visit_untagged_option(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Ok(T::deserialize(deserializer).ok()) + } +} + +impl<'de, T> Deserialize<'de> for Option +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_option(OptionVisitor { + marker: PhantomData, + }) + } + + // The Some variant's repr is opaque, so we can't play cute tricks with its + // tag to have deserialize_in_place build the content in place unconditionally. + // + // FIXME: investigate whether branching on the old value being Some to + // deserialize_in_place the value is profitable (probably data-dependent?) +} + +//////////////////////////////////////////////////////////////////////////////// + +struct PhantomDataVisitor { + marker: PhantomData, +} + +impl<'de, T: ?Sized> Visitor<'de> for PhantomDataVisitor { + type Value = PhantomData; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("unit") + } + + #[inline] + fn visit_unit(self) -> Result + where + E: Error, + { + Ok(PhantomData) + } +} + +impl<'de, T: ?Sized> Deserialize<'de> for PhantomData { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + let visitor = PhantomDataVisitor { + marker: PhantomData, + }; + deserializer.deserialize_unit_struct("PhantomData", visitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +macro_rules! seq_impl { + ( + $ty:ident , + $access:ident, + $clear:expr, + $with_capacity:expr, + $reserve:expr, + $insert:expr + ) => { + impl<'de, T $(, $typaram)*> Deserialize<'de> for $ty + where + T: Deserialize<'de> $(+ $tbound1 $(+ $tbound2)*)*, + $($typaram: $bound1 $(+ $bound2)*,)* + { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct SeqVisitor { + marker: PhantomData<$ty>, + } + + impl<'de, T $(, $typaram)*> Visitor<'de> for SeqVisitor + where + T: Deserialize<'de> $(+ $tbound1 $(+ $tbound2)*)*, + $($typaram: $bound1 $(+ $bound2)*,)* + { + type Value = $ty; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a sequence") + } + + #[inline] + fn visit_seq(self, mut $access: A) -> Result + where + A: SeqAccess<'de>, + { + let mut values = $with_capacity; + + while let Some(value) = try!($access.next_element()) { + $insert(&mut values, value); + } + + Ok(values) + } + } + + let visitor = SeqVisitor { marker: PhantomData }; + deserializer.deserialize_seq(visitor) + } + + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + struct SeqInPlaceVisitor<'a, T: 'a $(, $typaram: 'a)*>(&'a mut $ty); + + impl<'a, 'de, T $(, $typaram)*> Visitor<'de> for SeqInPlaceVisitor<'a, T $(, $typaram)*> + where + T: Deserialize<'de> $(+ $tbound1 $(+ $tbound2)*)*, + $($typaram: $bound1 $(+ $bound2)*,)* + { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a sequence") + } + + #[inline] + fn visit_seq(mut self, mut $access: A) -> Result + where + A: SeqAccess<'de>, + { + $clear(&mut self.0); + $reserve(&mut self.0, size_hint::cautious($access.size_hint())); + + // FIXME: try to overwrite old values here? (Vec, VecDeque, LinkedList) + while let Some(value) = try!($access.next_element()) { + $insert(&mut self.0, value); + } + + Ok(()) + } + } + + deserializer.deserialize_seq(SeqInPlaceVisitor(place)) + } + } + } +} + +// Dummy impl of reserve +#[cfg(any(feature = "std", feature = "alloc"))] +fn nop_reserve(_seq: T, _n: usize) {} + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!( + BinaryHeap, + seq, + BinaryHeap::clear, + BinaryHeap::with_capacity(size_hint::cautious(seq.size_hint())), + BinaryHeap::reserve, + BinaryHeap::push +); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!( + BTreeSet, + seq, + BTreeSet::clear, + BTreeSet::new(), + nop_reserve, + BTreeSet::insert +); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!( + LinkedList, + seq, + LinkedList::clear, + LinkedList::new(), + nop_reserve, + LinkedList::push_back +); + +#[cfg(feature = "std")] +seq_impl!( + HashSet, + seq, + HashSet::clear, + HashSet::with_capacity_and_hasher(size_hint::cautious(seq.size_hint()), S::default()), + HashSet::reserve, + HashSet::insert); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!( + VecDeque, + seq, + VecDeque::clear, + VecDeque::with_capacity(size_hint::cautious(seq.size_hint())), + VecDeque::reserve, + VecDeque::push_back +); + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, T> Deserialize<'de> for Vec +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct VecVisitor { + marker: PhantomData, + } + + impl<'de, T> Visitor<'de> for VecVisitor + where + T: Deserialize<'de>, + { + type Value = Vec; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a sequence") + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let mut values = Vec::with_capacity(size_hint::cautious(seq.size_hint())); + + while let Some(value) = try!(seq.next_element()) { + values.push(value); + } + + Ok(values) + } + } + + let visitor = VecVisitor { + marker: PhantomData, + }; + deserializer.deserialize_seq(visitor) + } + + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + struct VecInPlaceVisitor<'a, T: 'a>(&'a mut Vec); + + impl<'a, 'de, T> Visitor<'de> for VecInPlaceVisitor<'a, T> + where + T: Deserialize<'de>, + { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a sequence") + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let hint = size_hint::cautious(seq.size_hint()); + if let Some(additional) = hint.checked_sub(self.0.len()) { + self.0.reserve(additional); + } + + for i in 0..self.0.len() { + let next = { + let next_place = InPlaceSeed(&mut self.0[i]); + try!(seq.next_element_seed(next_place)) + }; + if next.is_none() { + self.0.truncate(i); + return Ok(()); + } + } + + while let Some(value) = try!(seq.next_element()) { + self.0.push(value); + } + + Ok(()) + } + } + + deserializer.deserialize_seq(VecInPlaceVisitor(place)) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +struct ArrayVisitor { + marker: PhantomData, +} +struct ArrayInPlaceVisitor<'a, A: 'a>(&'a mut A); + +impl ArrayVisitor { + fn new() -> Self { + ArrayVisitor { + marker: PhantomData, + } + } +} + +impl<'de, T> Visitor<'de> for ArrayVisitor<[T; 0]> { + type Value = [T; 0]; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("an empty array") + } + + #[inline] + fn visit_seq(self, _: A) -> Result + where + A: SeqAccess<'de>, + { + Ok([]) + } +} + +// Does not require T: Deserialize<'de>. +impl<'de, T> Deserialize<'de> for [T; 0] { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_tuple(0, ArrayVisitor::<[T; 0]>::new()) + } +} + +macro_rules! array_impls { + ($($len:expr => ($($n:tt)+))+) => { + $( + impl<'de, T> Visitor<'de> for ArrayVisitor<[T; $len]> + where + T: Deserialize<'de>, + { + type Value = [T; $len]; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("an array of length ", $len)) + } + + #[inline] + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + Ok([$( + match try!(seq.next_element()) { + Some(val) => val, + None => return Err(Error::invalid_length($n, &self)), + } + ),+]) + } + } + + impl<'a, 'de, T> Visitor<'de> for ArrayInPlaceVisitor<'a, [T; $len]> + where + T: Deserialize<'de>, + { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("an array of length ", $len)) + } + + #[inline] + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let mut fail_idx = None; + for (idx, dest) in self.0[..].iter_mut().enumerate() { + if try!(seq.next_element_seed(InPlaceSeed(dest))).is_none() { + fail_idx = Some(idx); + break; + } + } + if let Some(idx) = fail_idx { + return Err(Error::invalid_length(idx, &self)); + } + Ok(()) + } + } + + impl<'de, T> Deserialize<'de> for [T; $len] + where + T: Deserialize<'de>, + { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_tuple($len, ArrayVisitor::<[T; $len]>::new()) + } + + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + deserializer.deserialize_tuple($len, ArrayInPlaceVisitor(place)) + } + } + )+ + } +} + +array_impls! { + 1 => (0) + 2 => (0 1) + 3 => (0 1 2) + 4 => (0 1 2 3) + 5 => (0 1 2 3 4) + 6 => (0 1 2 3 4 5) + 7 => (0 1 2 3 4 5 6) + 8 => (0 1 2 3 4 5 6 7) + 9 => (0 1 2 3 4 5 6 7 8) + 10 => (0 1 2 3 4 5 6 7 8 9) + 11 => (0 1 2 3 4 5 6 7 8 9 10) + 12 => (0 1 2 3 4 5 6 7 8 9 10 11) + 13 => (0 1 2 3 4 5 6 7 8 9 10 11 12) + 14 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13) + 15 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14) + 16 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15) + 17 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16) + 18 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17) + 19 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18) + 20 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19) + 21 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20) + 22 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21) + 23 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22) + 24 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23) + 25 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24) + 26 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25) + 27 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26) + 28 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27) + 29 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28) + 30 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29) + 31 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30) + 32 => (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31) +} + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! tuple_impls { + ($($len:tt => ($($n:tt $name:ident)+))+) => { + $( + impl<'de, $($name: Deserialize<'de>),+> Deserialize<'de> for ($($name,)+) { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct TupleVisitor<$($name,)+> { + marker: PhantomData<($($name,)+)>, + } + + impl<'de, $($name: Deserialize<'de>),+> Visitor<'de> for TupleVisitor<$($name,)+> { + type Value = ($($name,)+); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("a tuple of size ", $len)) + } + + #[inline] + #[allow(non_snake_case)] + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + $( + let $name = match try!(seq.next_element()) { + Some(value) => value, + None => return Err(Error::invalid_length($n, &self)), + }; + )+ + + Ok(($($name,)+)) + } + } + + deserializer.deserialize_tuple($len, TupleVisitor { marker: PhantomData }) + } + + #[inline] + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + struct TupleInPlaceVisitor<'a, $($name: 'a,)+>(&'a mut ($($name,)+)); + + impl<'a, 'de, $($name: Deserialize<'de>),+> Visitor<'de> for TupleInPlaceVisitor<'a, $($name,)+> { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("a tuple of size ", $len)) + } + + #[inline] + #[allow(non_snake_case)] + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + $( + if try!(seq.next_element_seed(InPlaceSeed(&mut (self.0).$n))).is_none() { + return Err(Error::invalid_length($n, &self)); + } + )+ + + Ok(()) + } + } + + deserializer.deserialize_tuple($len, TupleInPlaceVisitor(place)) + } + } + )+ + } +} + +tuple_impls! { + 1 => (0 T0) + 2 => (0 T0 1 T1) + 3 => (0 T0 1 T1 2 T2) + 4 => (0 T0 1 T1 2 T2 3 T3) + 5 => (0 T0 1 T1 2 T2 3 T3 4 T4) + 6 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5) + 7 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6) + 8 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7) + 9 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8) + 10 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9) + 11 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10) + 12 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11) + 13 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12) + 14 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13) + 15 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13 14 T14) + 16 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13 14 T14 15 T15) +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +macro_rules! map_impl { + ( + $ty:ident , + $access:ident, + $with_capacity:expr + ) => { + impl<'de, K, V $(, $typaram)*> Deserialize<'de> for $ty + where + K: Deserialize<'de> $(+ $kbound1 $(+ $kbound2)*)*, + V: Deserialize<'de>, + $($typaram: $bound1 $(+ $bound2)*),* + { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct MapVisitor { + marker: PhantomData<$ty>, + } + + impl<'de, K, V $(, $typaram)*> Visitor<'de> for MapVisitor + where + K: Deserialize<'de> $(+ $kbound1 $(+ $kbound2)*)*, + V: Deserialize<'de>, + $($typaram: $bound1 $(+ $bound2)*),* + { + type Value = $ty; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a map") + } + + #[inline] + fn visit_map(self, mut $access: A) -> Result + where + A: MapAccess<'de>, + { + let mut values = $with_capacity; + + while let Some((key, value)) = try!($access.next_entry()) { + values.insert(key, value); + } + + Ok(values) + } + } + + let visitor = MapVisitor { marker: PhantomData }; + deserializer.deserialize_map(visitor) + } + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +map_impl!( + BTreeMap, + map, + BTreeMap::new()); + +#[cfg(feature = "std")] +map_impl!( + HashMap, + map, + HashMap::with_capacity_and_hasher(size_hint::cautious(map.size_hint()), S::default())); + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +macro_rules! parse_ip_impl { + ($expecting:tt $ty:ty; $size:tt) => { + impl<'de> Deserialize<'de> for $ty { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + if deserializer.is_human_readable() { + deserializer.deserialize_str(FromStrVisitor::new($expecting)) + } else { + <[u8; $size]>::deserialize(deserializer).map(<$ty>::from) + } + } + } + }; +} + +#[cfg(feature = "std")] +macro_rules! variant_identifier { + ( + $name_kind:ident ($($variant:ident; $bytes:expr; $index:expr),*) + $expecting_message:expr, + $variants_name:ident + ) => { + enum $name_kind { + $($variant),* + } + + static $variants_name: &'static [&'static str] = &[$(stringify!($variant)),*]; + + impl<'de> Deserialize<'de> for $name_kind { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct KindVisitor; + + impl<'de> Visitor<'de> for KindVisitor { + type Value = $name_kind; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str($expecting_message) + } + + fn visit_u64(self, value: u64) -> Result + where + E: Error, + { + match value { + $( + $index => Ok($name_kind :: $variant), + )* + _ => Err(Error::invalid_value(Unexpected::Unsigned(value), &self),), + } + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + $( + stringify!($variant) => Ok($name_kind :: $variant), + )* + _ => Err(Error::unknown_variant(value, $variants_name)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + $( + $bytes => Ok($name_kind :: $variant), + )* + _ => { + match str::from_utf8(value) { + Ok(value) => Err(Error::unknown_variant(value, $variants_name)), + Err(_) => Err(Error::invalid_value(Unexpected::Bytes(value), &self)), + } + } + } + } + } + + deserializer.deserialize_identifier(KindVisitor) + } + } + } +} + +#[cfg(feature = "std")] +macro_rules! deserialize_enum { + ( + $name:ident $name_kind:ident ($($variant:ident; $bytes:expr; $index:expr),*) + $expecting_message:expr, + $deserializer:expr + ) => { + variant_identifier! { + $name_kind ($($variant; $bytes; $index),*) + $expecting_message, + VARIANTS + } + + struct EnumVisitor; + impl<'de> Visitor<'de> for EnumVisitor { + type Value = $name; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(concat!("a ", stringify!($name))) + } + + + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + match try!(data.variant()) { + $( + ($name_kind :: $variant, v) => v.newtype_variant().map($name :: $variant), + )* + } + } + } + $deserializer.deserialize_enum(stringify!($name), VARIANTS, EnumVisitor) + } +} + +#[cfg(feature = "std")] +impl<'de> Deserialize<'de> for net::IpAddr { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + if deserializer.is_human_readable() { + deserializer.deserialize_str(FromStrVisitor::new("IP address")) + } else { + use lib::net::IpAddr; + deserialize_enum! { + IpAddr IpAddrKind (V4; b"V4"; 0, V6; b"V6"; 1) + "`V4` or `V6`", + deserializer + } + } + } +} + +#[cfg(feature = "std")] +parse_ip_impl!("IPv4 address" net::Ipv4Addr; 4); + +#[cfg(feature = "std")] +parse_ip_impl!("IPv6 address" net::Ipv6Addr; 16); + +#[cfg(feature = "std")] +macro_rules! parse_socket_impl { + ($expecting:tt $ty:ty, $new:expr) => { + impl<'de> Deserialize<'de> for $ty { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + if deserializer.is_human_readable() { + deserializer.deserialize_str(FromStrVisitor::new($expecting)) + } else { + <(_, u16)>::deserialize(deserializer).map(|(ip, port)| $new(ip, port)) + } + } + } + }; +} + +#[cfg(feature = "std")] +impl<'de> Deserialize<'de> for net::SocketAddr { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + if deserializer.is_human_readable() { + deserializer.deserialize_str(FromStrVisitor::new("socket address")) + } else { + use lib::net::SocketAddr; + deserialize_enum! { + SocketAddr SocketAddrKind (V4; b"V4"; 0, V6; b"V6"; 1) + "`V4` or `V6`", + deserializer + } + } + } +} + +#[cfg(feature = "std")] +parse_socket_impl!("IPv4 socket address" net::SocketAddrV4, net::SocketAddrV4::new); + +#[cfg(feature = "std")] +parse_socket_impl!("IPv6 socket address" net::SocketAddrV6, |ip, port| net::SocketAddrV6::new( + ip, port, 0, 0 +)); + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +struct PathVisitor; + +#[cfg(feature = "std")] +impl<'a> Visitor<'a> for PathVisitor { + type Value = &'a Path; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a borrowed path") + } + + fn visit_borrowed_str(self, v: &'a str) -> Result + where + E: Error, + { + Ok(v.as_ref()) + } + + fn visit_borrowed_bytes(self, v: &'a [u8]) -> Result + where + E: Error, + { + str::from_utf8(v) + .map(AsRef::as_ref) + .map_err(|_| Error::invalid_value(Unexpected::Bytes(v), &self)) + } +} + +#[cfg(feature = "std")] +impl<'de: 'a, 'a> Deserialize<'de> for &'a Path { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(PathVisitor) + } +} + +#[cfg(feature = "std")] +struct PathBufVisitor; + +#[cfg(feature = "std")] +impl<'de> Visitor<'de> for PathBufVisitor { + type Value = PathBuf; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("path string") + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + Ok(From::from(v)) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + Ok(From::from(v)) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + str::from_utf8(v) + .map(From::from) + .map_err(|_| Error::invalid_value(Unexpected::Bytes(v), &self)) + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + String::from_utf8(v) + .map(From::from) + .map_err(|e| Error::invalid_value(Unexpected::Bytes(&e.into_bytes()), &self)) + } +} + +#[cfg(feature = "std")] +impl<'de> Deserialize<'de> for PathBuf { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_string(PathBufVisitor) + } +} + +#[cfg(all(feature = "std", not(no_de_boxed_path)))] +forwarded_impl!((), Box, PathBuf::into_boxed_path); + +//////////////////////////////////////////////////////////////////////////////// + +// If this were outside of the serde crate, it would just use: +// +// #[derive(Deserialize)] +// #[serde(variant_identifier)] +#[cfg(all(feature = "std", any(unix, windows)))] +variant_identifier! { + OsStringKind (Unix; b"Unix"; 0, Windows; b"Windows"; 1) + "`Unix` or `Windows`", + OSSTR_VARIANTS +} + +#[cfg(all(feature = "std", any(unix, windows)))] +struct OsStringVisitor; + +#[cfg(all(feature = "std", any(unix, windows)))] +impl<'de> Visitor<'de> for OsStringVisitor { + type Value = OsString; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("os string") + } + + #[cfg(unix)] + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + use std::os::unix::ffi::OsStringExt; + + match try!(data.variant()) { + (OsStringKind::Unix, v) => v.newtype_variant().map(OsString::from_vec), + (OsStringKind::Windows, _) => Err(Error::custom( + "cannot deserialize Windows OS string on Unix", + )), + } + } + + #[cfg(windows)] + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + use std::os::windows::ffi::OsStringExt; + + match try!(data.variant()) { + (OsStringKind::Windows, v) => v + .newtype_variant::>() + .map(|vec| OsString::from_wide(&vec)), + (OsStringKind::Unix, _) => Err(Error::custom( + "cannot deserialize Unix OS string on Windows", + )), + } + } +} + +#[cfg(all(feature = "std", any(unix, windows)))] +impl<'de> Deserialize<'de> for OsString { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_enum("OsString", OSSTR_VARIANTS, OsStringVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +forwarded_impl!((T), Box, Box::new); + +#[cfg(any(feature = "std", feature = "alloc"))] +forwarded_impl!((T), Box<[T]>, Vec::into_boxed_slice); + +#[cfg(any(feature = "std", feature = "alloc"))] +forwarded_impl!((), Box, String::into_boxed_str); + +#[cfg(all(no_de_rc_dst, feature = "rc", any(feature = "std", feature = "alloc")))] +forwarded_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Deserializing a data structure containing `Arc` will not attempt to + /// deduplicate `Arc` references to the same data. Every deserialized `Arc` + /// will end up with a strong count of 1. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + (T), Arc, Arc::new +} + +#[cfg(all(no_de_rc_dst, feature = "rc", any(feature = "std", feature = "alloc")))] +forwarded_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Deserializing a data structure containing `Rc` will not attempt to + /// deduplicate `Rc` references to the same data. Every deserialized `Rc` + /// will end up with a strong count of 1. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + (T), Rc, Rc::new +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, 'a, T: ?Sized> Deserialize<'de> for Cow<'a, T> +where + T: ToOwned, + T::Owned: Deserialize<'de>, +{ + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + T::Owned::deserialize(deserializer).map(Cow::Owned) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// This impl requires the [`"rc"`] Cargo feature of Serde. The resulting +/// `Weak` has a reference count of 0 and cannot be upgraded. +/// +/// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +impl<'de, T: ?Sized> Deserialize<'de> for RcWeak +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + try!(Option::::deserialize(deserializer)); + Ok(RcWeak::new()) + } +} + +/// This impl requires the [`"rc"`] Cargo feature of Serde. The resulting +/// `Weak` has a reference count of 0 and cannot be upgraded. +/// +/// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +impl<'de, T: ?Sized> Deserialize<'de> for ArcWeak +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + try!(Option::::deserialize(deserializer)); + Ok(ArcWeak::new()) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(all( + not(no_de_rc_dst), + feature = "rc", + any(feature = "std", feature = "alloc") +))] +macro_rules! box_forwarded_impl { + ( + $(#[doc = $doc:tt])* + $t:ident + ) => { + $(#[doc = $doc])* + impl<'de, T: ?Sized> Deserialize<'de> for $t + where + Box: Deserialize<'de>, + { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Box::deserialize(deserializer).map(Into::into) + } + } + }; +} + +#[cfg(all( + not(no_de_rc_dst), + feature = "rc", + any(feature = "std", feature = "alloc") +))] +box_forwarded_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Deserializing a data structure containing `Rc` will not attempt to + /// deduplicate `Rc` references to the same data. Every deserialized `Rc` + /// will end up with a strong count of 1. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + Rc +} + +#[cfg(all( + not(no_de_rc_dst), + feature = "rc", + any(feature = "std", feature = "alloc") +))] +box_forwarded_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Deserializing a data structure containing `Arc` will not attempt to + /// deduplicate `Arc` references to the same data. Every deserialized `Arc` + /// will end up with a strong count of 1. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + Arc +} + +//////////////////////////////////////////////////////////////////////////////// + +impl<'de, T> Deserialize<'de> for Cell +where + T: Deserialize<'de> + Copy, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + T::deserialize(deserializer).map(Cell::new) + } +} + +forwarded_impl!((T), RefCell, RefCell::new); + +#[cfg(feature = "std")] +forwarded_impl!((T), Mutex, Mutex::new); + +#[cfg(feature = "std")] +forwarded_impl!((T), RwLock, RwLock::new); + +//////////////////////////////////////////////////////////////////////////////// + +// This is a cleaned-up version of the impl generated by: +// +// #[derive(Deserialize)] +// #[serde(deny_unknown_fields)] +// struct Duration { +// secs: u64, +// nanos: u32, +// } +#[cfg(any(feature = "std", not(no_core_duration)))] +impl<'de> Deserialize<'de> for Duration { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // If this were outside of the serde crate, it would just use: + // + // #[derive(Deserialize)] + // #[serde(field_identifier, rename_all = "lowercase")] + enum Field { + Secs, + Nanos, + } + + impl<'de> Deserialize<'de> for Field { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = Field; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("`secs` or `nanos`") + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + "secs" => Ok(Field::Secs), + "nanos" => Ok(Field::Nanos), + _ => Err(Error::unknown_field(value, FIELDS)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + b"secs" => Ok(Field::Secs), + b"nanos" => Ok(Field::Nanos), + _ => { + let value = ::__private::from_utf8_lossy(value); + Err(Error::unknown_field(&*value, FIELDS)) + } + } + } + } + + deserializer.deserialize_identifier(FieldVisitor) + } + } + + fn check_overflow(secs: u64, nanos: u32) -> Result<(), E> + where + E: Error, + { + static NANOS_PER_SEC: u32 = 1_000_000_000; + match secs.checked_add((nanos / NANOS_PER_SEC) as u64) { + Some(_) => Ok(()), + None => Err(E::custom("overflow deserializing Duration")), + } + } + + struct DurationVisitor; + + impl<'de> Visitor<'de> for DurationVisitor { + type Value = Duration; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("struct Duration") + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let secs: u64 = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(0, &self)); + } + }; + let nanos: u32 = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(1, &self)); + } + }; + try!(check_overflow(secs, nanos)); + Ok(Duration::new(secs, nanos)) + } + + fn visit_map(self, mut map: A) -> Result + where + A: MapAccess<'de>, + { + let mut secs: Option = None; + let mut nanos: Option = None; + while let Some(key) = try!(map.next_key()) { + match key { + Field::Secs => { + if secs.is_some() { + return Err(::duplicate_field("secs")); + } + secs = Some(try!(map.next_value())); + } + Field::Nanos => { + if nanos.is_some() { + return Err(::duplicate_field("nanos")); + } + nanos = Some(try!(map.next_value())); + } + } + } + let secs = match secs { + Some(secs) => secs, + None => return Err(::missing_field("secs")), + }; + let nanos = match nanos { + Some(nanos) => nanos, + None => return Err(::missing_field("nanos")), + }; + try!(check_overflow(secs, nanos)); + Ok(Duration::new(secs, nanos)) + } + } + + const FIELDS: &'static [&'static str] = &["secs", "nanos"]; + deserializer.deserialize_struct("Duration", FIELDS, DurationVisitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +impl<'de> Deserialize<'de> for SystemTime { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // Reuse duration + enum Field { + Secs, + Nanos, + } + + impl<'de> Deserialize<'de> for Field { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = Field; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("`secs_since_epoch` or `nanos_since_epoch`") + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + "secs_since_epoch" => Ok(Field::Secs), + "nanos_since_epoch" => Ok(Field::Nanos), + _ => Err(Error::unknown_field(value, FIELDS)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + b"secs_since_epoch" => Ok(Field::Secs), + b"nanos_since_epoch" => Ok(Field::Nanos), + _ => { + let value = String::from_utf8_lossy(value); + Err(Error::unknown_field(&value, FIELDS)) + } + } + } + } + + deserializer.deserialize_identifier(FieldVisitor) + } + } + + fn check_overflow(secs: u64, nanos: u32) -> Result<(), E> + where + E: Error, + { + static NANOS_PER_SEC: u32 = 1_000_000_000; + match secs.checked_add((nanos / NANOS_PER_SEC) as u64) { + Some(_) => Ok(()), + None => Err(E::custom("overflow deserializing SystemTime epoch offset")), + } + } + + struct DurationVisitor; + + impl<'de> Visitor<'de> for DurationVisitor { + type Value = Duration; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("struct SystemTime") + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let secs: u64 = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(0, &self)); + } + }; + let nanos: u32 = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(1, &self)); + } + }; + try!(check_overflow(secs, nanos)); + Ok(Duration::new(secs, nanos)) + } + + fn visit_map(self, mut map: A) -> Result + where + A: MapAccess<'de>, + { + let mut secs: Option = None; + let mut nanos: Option = None; + while let Some(key) = try!(map.next_key()) { + match key { + Field::Secs => { + if secs.is_some() { + return Err(::duplicate_field( + "secs_since_epoch", + )); + } + secs = Some(try!(map.next_value())); + } + Field::Nanos => { + if nanos.is_some() { + return Err(::duplicate_field( + "nanos_since_epoch", + )); + } + nanos = Some(try!(map.next_value())); + } + } + } + let secs = match secs { + Some(secs) => secs, + None => return Err(::missing_field("secs_since_epoch")), + }; + let nanos = match nanos { + Some(nanos) => nanos, + None => return Err(::missing_field("nanos_since_epoch")), + }; + try!(check_overflow(secs, nanos)); + Ok(Duration::new(secs, nanos)) + } + } + + const FIELDS: &'static [&'static str] = &["secs_since_epoch", "nanos_since_epoch"]; + let duration = try!(deserializer.deserialize_struct("SystemTime", FIELDS, DurationVisitor)); + #[cfg(not(no_systemtime_checked_add))] + let ret = UNIX_EPOCH + .checked_add(duration) + .ok_or_else(|| D::Error::custom("overflow deserializing SystemTime")); + #[cfg(no_systemtime_checked_add)] + let ret = Ok(UNIX_EPOCH + duration); + ret + } +} + +//////////////////////////////////////////////////////////////////////////////// + +// Similar to: +// +// #[derive(Deserialize)] +// #[serde(deny_unknown_fields)] +// struct Range { +// start: u64, +// end: u32, +// } +impl<'de, Idx> Deserialize<'de> for Range +where + Idx: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + let (start, end) = try!(deserializer.deserialize_struct( + "Range", + range::FIELDS, + range::RangeVisitor { + expecting: "struct Range", + phantom: PhantomData, + }, + )); + Ok(start..end) + } +} + +#[cfg(not(no_range_inclusive))] +impl<'de, Idx> Deserialize<'de> for RangeInclusive +where + Idx: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + let (start, end) = try!(deserializer.deserialize_struct( + "RangeInclusive", + range::FIELDS, + range::RangeVisitor { + expecting: "struct RangeInclusive", + phantom: PhantomData, + }, + )); + Ok(RangeInclusive::new(start, end)) + } +} + +mod range { + use lib::*; + + use de::{Deserialize, Deserializer, Error, MapAccess, SeqAccess, Visitor}; + + pub const FIELDS: &'static [&'static str] = &["start", "end"]; + + // If this were outside of the serde crate, it would just use: + // + // #[derive(Deserialize)] + // #[serde(field_identifier, rename_all = "lowercase")] + enum Field { + Start, + End, + } + + impl<'de> Deserialize<'de> for Field { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = Field; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("`start` or `end`") + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + "start" => Ok(Field::Start), + "end" => Ok(Field::End), + _ => Err(Error::unknown_field(value, FIELDS)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + b"start" => Ok(Field::Start), + b"end" => Ok(Field::End), + _ => { + let value = ::__private::from_utf8_lossy(value); + Err(Error::unknown_field(&*value, FIELDS)) + } + } + } + } + + deserializer.deserialize_identifier(FieldVisitor) + } + } + + pub struct RangeVisitor { + pub expecting: &'static str, + pub phantom: PhantomData, + } + + impl<'de, Idx> Visitor<'de> for RangeVisitor + where + Idx: Deserialize<'de>, + { + type Value = (Idx, Idx); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(self.expecting) + } + + fn visit_seq(self, mut seq: A) -> Result + where + A: SeqAccess<'de>, + { + let start: Idx = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(0, &self)); + } + }; + let end: Idx = match try!(seq.next_element()) { + Some(value) => value, + None => { + return Err(Error::invalid_length(1, &self)); + } + }; + Ok((start, end)) + } + + fn visit_map(self, mut map: A) -> Result + where + A: MapAccess<'de>, + { + let mut start: Option = None; + let mut end: Option = None; + while let Some(key) = try!(map.next_key()) { + match key { + Field::Start => { + if start.is_some() { + return Err(::duplicate_field("start")); + } + start = Some(try!(map.next_value())); + } + Field::End => { + if end.is_some() { + return Err(::duplicate_field("end")); + } + end = Some(try!(map.next_value())); + } + } + } + let start = match start { + Some(start) => start, + None => return Err(::missing_field("start")), + }; + let end = match end { + Some(end) => end, + None => return Err(::missing_field("end")), + }; + Ok((start, end)) + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(not(no_ops_bound), all(feature = "std", not(no_collections_bound))))] +impl<'de, T> Deserialize<'de> for Bound +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + enum Field { + Unbounded, + Included, + Excluded, + } + + impl<'de> Deserialize<'de> for Field { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = Field; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("`Unbounded`, `Included` or `Excluded`") + } + + fn visit_u64(self, value: u64) -> Result + where + E: Error, + { + match value { + 0 => Ok(Field::Unbounded), + 1 => Ok(Field::Included), + 2 => Ok(Field::Excluded), + _ => Err(Error::invalid_value(Unexpected::Unsigned(value), &self)), + } + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + "Unbounded" => Ok(Field::Unbounded), + "Included" => Ok(Field::Included), + "Excluded" => Ok(Field::Excluded), + _ => Err(Error::unknown_variant(value, VARIANTS)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + b"Unbounded" => Ok(Field::Unbounded), + b"Included" => Ok(Field::Included), + b"Excluded" => Ok(Field::Excluded), + _ => match str::from_utf8(value) { + Ok(value) => Err(Error::unknown_variant(value, VARIANTS)), + Err(_) => { + Err(Error::invalid_value(Unexpected::Bytes(value), &self)) + } + }, + } + } + } + + deserializer.deserialize_identifier(FieldVisitor) + } + } + + struct BoundVisitor(PhantomData>); + + impl<'de, T> Visitor<'de> for BoundVisitor + where + T: Deserialize<'de>, + { + type Value = Bound; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("enum Bound") + } + + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + match try!(data.variant()) { + (Field::Unbounded, v) => v.unit_variant().map(|()| Bound::Unbounded), + (Field::Included, v) => v.newtype_variant().map(Bound::Included), + (Field::Excluded, v) => v.newtype_variant().map(Bound::Excluded), + } + } + } + + const VARIANTS: &'static [&'static str] = &["Unbounded", "Included", "Excluded"]; + + deserializer.deserialize_enum("Bound", VARIANTS, BoundVisitor(PhantomData)) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl<'de, T, E> Deserialize<'de> for Result +where + T: Deserialize<'de>, + E: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // If this were outside of the serde crate, it would just use: + // + // #[derive(Deserialize)] + // #[serde(variant_identifier)] + enum Field { + Ok, + Err, + } + + impl<'de> Deserialize<'de> for Field { + #[inline] + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = Field; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("`Ok` or `Err`") + } + + fn visit_u64(self, value: u64) -> Result + where + E: Error, + { + match value { + 0 => Ok(Field::Ok), + 1 => Ok(Field::Err), + _ => Err(Error::invalid_value(Unexpected::Unsigned(value), &self)), + } + } + + fn visit_str(self, value: &str) -> Result + where + E: Error, + { + match value { + "Ok" => Ok(Field::Ok), + "Err" => Ok(Field::Err), + _ => Err(Error::unknown_variant(value, VARIANTS)), + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + E: Error, + { + match value { + b"Ok" => Ok(Field::Ok), + b"Err" => Ok(Field::Err), + _ => match str::from_utf8(value) { + Ok(value) => Err(Error::unknown_variant(value, VARIANTS)), + Err(_) => { + Err(Error::invalid_value(Unexpected::Bytes(value), &self)) + } + }, + } + } + } + + deserializer.deserialize_identifier(FieldVisitor) + } + } + + struct ResultVisitor(PhantomData>); + + impl<'de, T, E> Visitor<'de> for ResultVisitor + where + T: Deserialize<'de>, + E: Deserialize<'de>, + { + type Value = Result; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("enum Result") + } + + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + match try!(data.variant()) { + (Field::Ok, v) => v.newtype_variant().map(Ok), + (Field::Err, v) => v.newtype_variant().map(Err), + } + } + } + + const VARIANTS: &'static [&'static str] = &["Ok", "Err"]; + + deserializer.deserialize_enum("Result", VARIANTS, ResultVisitor(PhantomData)) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl<'de, T> Deserialize<'de> for Wrapping +where + T: Deserialize<'de>, +{ + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Deserialize::deserialize(deserializer).map(Wrapping) + } +} + +#[cfg(all(feature = "std", not(no_std_atomic)))] +macro_rules! atomic_impl { + ($($ty:ident $size:expr)*) => { + $( + #[cfg(any(no_target_has_atomic, target_has_atomic = $size))] + impl<'de> Deserialize<'de> for $ty { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Deserialize::deserialize(deserializer).map(Self::new) + } + } + )* + }; +} + +#[cfg(all(feature = "std", not(no_std_atomic)))] +atomic_impl! { + AtomicBool "8" + AtomicI8 "8" + AtomicI16 "16" + AtomicI32 "32" + AtomicIsize "ptr" + AtomicU8 "8" + AtomicU16 "16" + AtomicU32 "32" + AtomicUsize "ptr" +} + +#[cfg(all(feature = "std", not(no_std_atomic64)))] +atomic_impl! { + AtomicI64 "64" + AtomicU64 "64" +} + +#[cfg(feature = "std")] +struct FromStrVisitor { + expecting: &'static str, + ty: PhantomData, +} + +#[cfg(feature = "std")] +impl FromStrVisitor { + fn new(expecting: &'static str) -> Self { + FromStrVisitor { + expecting: expecting, + ty: PhantomData, + } + } +} + +#[cfg(feature = "std")] +impl<'de, T> Visitor<'de> for FromStrVisitor +where + T: str::FromStr, + T::Err: fmt::Display, +{ + type Value = T; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(self.expecting) + } + + fn visit_str(self, s: &str) -> Result + where + E: Error, + { + s.parse().map_err(Error::custom) + } +} diff --git a/rust/serde/de/mod.rs b/rust/serde/de/mod.rs new file mode 100644 index 00000000000000..6da226da8ffcbb --- /dev/null +++ b/rust/serde/de/mod.rs @@ -0,0 +1,2313 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Generic data structure deserialization framework. +//! +//! The two most important traits in this module are [`Deserialize`] and +//! [`Deserializer`]. +//! +//! - **A type that implements `Deserialize` is a data structure** that can be +//! deserialized from any data format supported by Serde, and conversely +//! - **A type that implements `Deserializer` is a data format** that can +//! deserialize any data structure supported by Serde. +//! +//! # The Deserialize trait +//! +//! Serde provides [`Deserialize`] implementations for many Rust primitive and +//! standard library types. The complete list is below. All of these can be +//! deserialized using Serde out of the box. +//! +//! Additionally, Serde provides a procedural macro called [`serde_derive`] to +//! automatically generate [`Deserialize`] implementations for structs and enums +//! in your program. See the [derive section of the manual] for how to use this. +//! +//! In rare cases it may be necessary to implement [`Deserialize`] manually for +//! some type in your program. See the [Implementing `Deserialize`] section of +//! the manual for more about this. +//! +//! Third-party crates may provide [`Deserialize`] implementations for types +//! that they expose. For example the [`linked-hash-map`] crate provides a +//! [`LinkedHashMap`] type that is deserializable by Serde because the +//! crate provides an implementation of [`Deserialize`] for it. +//! +//! # The Deserializer trait +//! +//! [`Deserializer`] implementations are provided by third-party crates, for +//! example [`serde_json`], [`serde_yaml`] and [`postcard`]. +//! +//! A partial list of well-maintained formats is given on the [Serde +//! website][data formats]. +//! +//! # Implementations of Deserialize provided by Serde +//! +//! This is a slightly different set of types than what is supported for +//! serialization. Some types can be serialized by Serde but not deserialized. +//! One example is `OsStr`. +//! +//! - **Primitive types**: +//! - bool +//! - i8, i16, i32, i64, i128, isize +//! - u8, u16, u32, u64, u128, usize +//! - f32, f64 +//! - char +//! - **Compound types**: +//! - \[T; 0\] through \[T; 32\] +//! - tuples up to size 16 +//! - **Common standard library types**: +//! - String +//! - Option\ +//! - Result\ +//! - PhantomData\ +//! - **Wrapper types**: +//! - Box\ +//! - Box\<\[T\]\> +//! - Box\ +//! - Cow\<'a, T\> +//! - Cell\ +//! - RefCell\ +//! - Mutex\ +//! - RwLock\ +//! - Rc\ *(if* features = ["rc"] *is enabled)* +//! - Arc\ *(if* features = ["rc"] *is enabled)* +//! - **Collection types**: +//! - BTreeMap\ +//! - BTreeSet\ +//! - BinaryHeap\ +//! - HashMap\ +//! - HashSet\ +//! - LinkedList\ +//! - VecDeque\ +//! - Vec\ +//! - **Zero-copy types**: +//! - &str +//! - &\[u8\] +//! - **FFI types**: +//! - CString +//! - Box\ +//! - OsString +//! - **Miscellaneous standard library types**: +//! - Duration +//! - SystemTime +//! - Path +//! - PathBuf +//! - Range\ +//! - RangeInclusive\ +//! - Bound\ +//! - num::NonZero* +//! - `!` *(unstable)* +//! - **Net types**: +//! - IpAddr +//! - Ipv4Addr +//! - Ipv6Addr +//! - SocketAddr +//! - SocketAddrV4 +//! - SocketAddrV6 +//! +//! [Implementing `Deserialize`]: https://serde.rs/impl-deserialize.html +//! [`Deserialize`]: ../trait.Deserialize.html +//! [`Deserializer`]: ../trait.Deserializer.html +//! [`LinkedHashMap`]: https://docs.rs/linked-hash-map/*/linked_hash_map/struct.LinkedHashMap.html +//! [`postcard`]: https://github.com/jamesmunns/postcard +//! [`linked-hash-map`]: https://crates.io/crates/linked-hash-map +//! [`serde_derive`]: https://crates.io/crates/serde_derive +//! [`serde_json`]: https://github.com/serde-rs/json +//! [`serde_yaml`]: https://github.com/dtolnay/serde-yaml +//! [derive section of the manual]: https://serde.rs/derive.html +//! [data formats]: https://serde.rs/#data-formats + +use lib::*; + +//////////////////////////////////////////////////////////////////////////////// + +pub mod value; + +#[cfg(not(no_integer128))] +mod format; +mod ignored_any; +mod impls; +mod utf8; + +pub use self::ignored_any::IgnoredAny; + +#[cfg(feature = "std")] +#[doc(no_inline)] +pub use std::error::Error as StdError; +#[cfg(not(feature = "std"))] +#[doc(no_inline)] +pub use std_error::Error as StdError; + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! declare_error_trait { + (Error: Sized $(+ $($supertrait:ident)::+)*) => { + /// The `Error` trait allows `Deserialize` implementations to create descriptive + /// error messages belonging to the `Deserializer` against which they are + /// currently running. + /// + /// Every `Deserializer` declares an `Error` type that encompasses both + /// general-purpose deserialization errors as well as errors specific to the + /// particular deserialization format. For example the `Error` type of + /// `serde_json` can represent errors like an invalid JSON escape sequence or an + /// unterminated string literal, in addition to the error cases that are part of + /// this trait. + /// + /// Most deserializers should only need to provide the `Error::custom` method + /// and inherit the default behavior for the other methods. + /// + /// # Example implementation + /// + /// The [example data format] presented on the website shows an error + /// type appropriate for a basic JSON data format. + /// + /// [example data format]: https://serde.rs/data-format.html + pub trait Error: Sized $(+ $($supertrait)::+)* { + /// Raised when there is general error when deserializing a type. + /// + /// The message should not be capitalized and should not end with a period. + /// + /// ```edition2018 + /// # use std::str::FromStr; + /// # + /// # struct IpAddr; + /// # + /// # impl FromStr for IpAddr { + /// # type Err = String; + /// # + /// # fn from_str(_: &str) -> Result { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::de::{self, Deserialize, Deserializer}; + /// + /// impl<'de> Deserialize<'de> for IpAddr { + /// fn deserialize(deserializer: D) -> Result + /// where + /// D: Deserializer<'de>, + /// { + /// let s = String::deserialize(deserializer)?; + /// s.parse().map_err(de::Error::custom) + /// } + /// } + /// ``` + fn custom(msg: T) -> Self + where + T: Display; + + /// Raised when a `Deserialize` receives a type different from what it was + /// expecting. + /// + /// The `unexp` argument provides information about what type was received. + /// This is the type that was present in the input file or other source data + /// of the Deserializer. + /// + /// The `exp` argument provides information about what type was being + /// expected. This is the type that is written in the program. + /// + /// For example if we try to deserialize a String out of a JSON file + /// containing an integer, the unexpected type is the integer and the + /// expected type is the string. + #[cold] + fn invalid_type(unexp: Unexpected, exp: &Expected) -> Self { + Error::custom(format_args!("invalid type: {}, expected {}", unexp, exp)) + } + + /// Raised when a `Deserialize` receives a value of the right type but that + /// is wrong for some other reason. + /// + /// The `unexp` argument provides information about what value was received. + /// This is the value that was present in the input file or other source + /// data of the Deserializer. + /// + /// The `exp` argument provides information about what value was being + /// expected. This is the type that is written in the program. + /// + /// For example if we try to deserialize a String out of some binary data + /// that is not valid UTF-8, the unexpected value is the bytes and the + /// expected value is a string. + #[cold] + fn invalid_value(unexp: Unexpected, exp: &Expected) -> Self { + Error::custom(format_args!("invalid value: {}, expected {}", unexp, exp)) + } + + /// Raised when deserializing a sequence or map and the input data contains + /// too many or too few elements. + /// + /// The `len` argument is the number of elements encountered. The sequence + /// or map may have expected more arguments or fewer arguments. + /// + /// The `exp` argument provides information about what data was being + /// expected. For example `exp` might say that a tuple of size 6 was + /// expected. + #[cold] + fn invalid_length(len: usize, exp: &Expected) -> Self { + Error::custom(format_args!("invalid length {}, expected {}", len, exp)) + } + + /// Raised when a `Deserialize` enum type received a variant with an + /// unrecognized name. + #[cold] + fn unknown_variant(variant: &str, expected: &'static [&'static str]) -> Self { + if expected.is_empty() { + Error::custom(format_args!( + "unknown variant `{}`, there are no variants", + variant + )) + } else { + Error::custom(format_args!( + "unknown variant `{}`, expected {}", + variant, + OneOf { names: expected } + )) + } + } + + /// Raised when a `Deserialize` struct type received a field with an + /// unrecognized name. + #[cold] + fn unknown_field(field: &str, expected: &'static [&'static str]) -> Self { + if expected.is_empty() { + Error::custom(format_args!( + "unknown field `{}`, there are no fields", + field + )) + } else { + Error::custom(format_args!( + "unknown field `{}`, expected {}", + field, + OneOf { names: expected } + )) + } + } + + /// Raised when a `Deserialize` struct type expected to receive a required + /// field with a particular name but that field was not present in the + /// input. + #[cold] + fn missing_field(field: &'static str) -> Self { + Error::custom(format_args!("missing field `{}`", field)) + } + + /// Raised when a `Deserialize` struct type received more than one of the + /// same field. + #[cold] + fn duplicate_field(field: &'static str) -> Self { + Error::custom(format_args!("duplicate field `{}`", field)) + } + } + } +} + +#[cfg(feature = "std")] +declare_error_trait!(Error: Sized + StdError); + +#[cfg(not(feature = "std"))] +declare_error_trait!(Error: Sized + Debug + Display); + +/// `Unexpected` represents an unexpected invocation of any one of the `Visitor` +/// trait methods. +/// +/// This is used as an argument to the `invalid_type`, `invalid_value`, and +/// `invalid_length` methods of the `Error` trait to build error messages. +/// +/// ```edition2018 +/// # use std::fmt; +/// # +/// # use serde::de::{self, Unexpected, Visitor}; +/// # +/// # struct Example; +/// # +/// # impl<'de> Visitor<'de> for Example { +/// # type Value = (); +/// # +/// # fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// # write!(formatter, "definitely not a boolean") +/// # } +/// # +/// fn visit_bool(self, v: bool) -> Result +/// where +/// E: de::Error, +/// { +/// Err(de::Error::invalid_type(Unexpected::Bool(v), &self)) +/// } +/// # } +/// ``` +#[derive(Copy, Clone, PartialEq, Debug)] +pub enum Unexpected<'a> { + /// The input contained a boolean value that was not expected. + Bool(bool), + + /// The input contained an unsigned integer `u8`, `u16`, `u32` or `u64` that + /// was not expected. + Unsigned(u64), + + /// The input contained a signed integer `i8`, `i16`, `i32` or `i64` that + /// was not expected. + Signed(i64), + + /// The input contained a floating point `f32` or `f64` that was not + /// expected. + #[cfg(not(no_fp_fmt_parse))] + Float(f64), + + /// The input contained a `char` that was not expected. + Char(char), + + /// The input contained a `&str` or `String` that was not expected. + Str(&'a str), + + /// The input contained a `&[u8]` or `Vec` that was not expected. + Bytes(&'a [u8]), + + /// The input contained a unit `()` that was not expected. + Unit, + + /// The input contained an `Option` that was not expected. + Option, + + /// The input contained a newtype struct that was not expected. + NewtypeStruct, + + /// The input contained a sequence that was not expected. + Seq, + + /// The input contained a map that was not expected. + Map, + + /// The input contained an enum that was not expected. + Enum, + + /// The input contained a unit variant that was not expected. + UnitVariant, + + /// The input contained a newtype variant that was not expected. + NewtypeVariant, + + /// The input contained a tuple variant that was not expected. + TupleVariant, + + /// The input contained a struct variant that was not expected. + StructVariant, + + /// A message stating what uncategorized thing the input contained that was + /// not expected. + /// + /// The message should be a noun or noun phrase, not capitalized and without + /// a period. An example message is "unoriginal superhero". + Other(&'a str), +} + +impl<'a> fmt::Display for Unexpected<'a> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + use self::Unexpected::*; + match *self { + Bool(b) => write!(formatter, "boolean `{}`", b), + Unsigned(i) => write!(formatter, "integer `{}`", i), + Signed(i) => write!(formatter, "integer `{}`", i), + #[cfg(not(no_fp_fmt_parse))] + Float(f) => write!(formatter, "floating point `{}`", f), + Char(c) => write!(formatter, "character `{}`", c), + Str(s) => write!(formatter, "string {:?}", s), + Bytes(_) => write!(formatter, "byte array"), + Unit => write!(formatter, "unit value"), + Option => write!(formatter, "Option value"), + NewtypeStruct => write!(formatter, "newtype struct"), + Seq => write!(formatter, "sequence"), + Map => write!(formatter, "map"), + Enum => write!(formatter, "enum"), + UnitVariant => write!(formatter, "unit variant"), + NewtypeVariant => write!(formatter, "newtype variant"), + TupleVariant => write!(formatter, "tuple variant"), + StructVariant => write!(formatter, "struct variant"), + Other(other) => formatter.write_str(other), + } + } +} + +/// `Expected` represents an explanation of what data a `Visitor` was expecting +/// to receive. +/// +/// This is used as an argument to the `invalid_type`, `invalid_value`, and +/// `invalid_length` methods of the `Error` trait to build error messages. The +/// message should be a noun or noun phrase that completes the sentence "This +/// Visitor expects to receive ...", for example the message could be "an +/// integer between 0 and 64". The message should not be capitalized and should +/// not end with a period. +/// +/// Within the context of a `Visitor` implementation, the `Visitor` itself +/// (`&self`) is an implementation of this trait. +/// +/// ```edition2018 +/// # use std::fmt; +/// # +/// # use serde::de::{self, Unexpected, Visitor}; +/// # +/// # struct Example; +/// # +/// # impl<'de> Visitor<'de> for Example { +/// # type Value = (); +/// # +/// # fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// # write!(formatter, "definitely not a boolean") +/// # } +/// # +/// fn visit_bool(self, v: bool) -> Result +/// where +/// E: de::Error, +/// { +/// Err(de::Error::invalid_type(Unexpected::Bool(v), &self)) +/// } +/// # } +/// ``` +/// +/// Outside of a `Visitor`, `&"..."` can be used. +/// +/// ```edition2018 +/// # use serde::de::{self, Unexpected}; +/// # +/// # fn example() -> Result<(), E> +/// # where +/// # E: de::Error, +/// # { +/// # let v = true; +/// return Err(de::Error::invalid_type(Unexpected::Bool(v), &"a negative integer")); +/// # } +/// ``` +pub trait Expected { + /// Format an explanation of what data was being expected. Same signature as + /// the `Display` and `Debug` traits. + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result; +} + +impl<'de, T> Expected for T +where + T: Visitor<'de>, +{ + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + self.expecting(formatter) + } +} + +impl<'a> Expected for &'a str { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(self) + } +} + +impl<'a> Display for Expected + 'a { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + Expected::fmt(self, formatter) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A **data structure** that can be deserialized from any data format supported +/// by Serde. +/// +/// Serde provides `Deserialize` implementations for many Rust primitive and +/// standard library types. The complete list is [here][crate::de]. All of these +/// can be deserialized using Serde out of the box. +/// +/// Additionally, Serde provides a procedural macro called `serde_derive` to +/// automatically generate `Deserialize` implementations for structs and enums +/// in your program. See the [derive section of the manual][derive] for how to +/// use this. +/// +/// In rare cases it may be necessary to implement `Deserialize` manually for +/// some type in your program. See the [Implementing +/// `Deserialize`][impl-deserialize] section of the manual for more about this. +/// +/// Third-party crates may provide `Deserialize` implementations for types that +/// they expose. For example the `linked-hash-map` crate provides a +/// `LinkedHashMap` type that is deserializable by Serde because the crate +/// provides an implementation of `Deserialize` for it. +/// +/// [derive]: https://serde.rs/derive.html +/// [impl-deserialize]: https://serde.rs/impl-deserialize.html +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by `Self` when deserialized. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +pub trait Deserialize<'de>: Sized { + /// Deserialize this value from the given Serde deserializer. + /// + /// See the [Implementing `Deserialize`][impl-deserialize] section of the + /// manual for more information about how to implement this method. + /// + /// [impl-deserialize]: https://serde.rs/impl-deserialize.html + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>; + + /// Deserializes a value into `self` from the given Deserializer. + /// + /// The purpose of this method is to allow the deserializer to reuse + /// resources and avoid copies. As such, if this method returns an error, + /// `self` will be in an indeterminate state where some parts of the struct + /// have been overwritten. Although whatever state that is will be + /// memory-safe. + /// + /// This is generally useful when repeatedly deserializing values that + /// are processed one at a time, where the value of `self` doesn't matter + /// when the next deserialization occurs. + /// + /// If you manually implement this, your recursive deserializations should + /// use `deserialize_in_place`. + /// + /// This method is stable and an official public API, but hidden from the + /// documentation because it is almost never what newbies are looking for. + /// Showing it in rustdoc would cause it to be featured more prominently + /// than it deserves. + #[doc(hidden)] + fn deserialize_in_place(deserializer: D, place: &mut Self) -> Result<(), D::Error> + where + D: Deserializer<'de>, + { + // Default implementation just delegates to `deserialize` impl. + *place = try!(Deserialize::deserialize(deserializer)); + Ok(()) + } +} + +/// A data structure that can be deserialized without borrowing any data from +/// the deserializer. +/// +/// This is primarily useful for trait bounds on functions. For example a +/// `from_str` function may be able to deserialize a data structure that borrows +/// from the input string, but a `from_reader` function may only deserialize +/// owned data. +/// +/// ```edition2018 +/// # use serde::de::{Deserialize, DeserializeOwned}; +/// # use std::io::{Read, Result}; +/// # +/// # trait Ignore { +/// fn from_str<'a, T>(s: &'a str) -> Result +/// where +/// T: Deserialize<'a>; +/// +/// fn from_reader(rdr: R) -> Result +/// where +/// R: Read, +/// T: DeserializeOwned; +/// # } +/// ``` +/// +/// # Lifetime +/// +/// The relationship between `Deserialize` and `DeserializeOwned` in trait +/// bounds is explained in more detail on the page [Understanding deserializer +/// lifetimes]. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +pub trait DeserializeOwned: for<'de> Deserialize<'de> {} +impl DeserializeOwned for T where T: for<'de> Deserialize<'de> {} + +/// `DeserializeSeed` is the stateful form of the `Deserialize` trait. If you +/// ever find yourself looking for a way to pass data into a `Deserialize` impl, +/// this trait is the way to do it. +/// +/// As one example of stateful deserialization consider deserializing a JSON +/// array into an existing buffer. Using the `Deserialize` trait we could +/// deserialize a JSON array into a `Vec` but it would be a freshly allocated +/// `Vec`; there is no way for `Deserialize` to reuse a previously allocated +/// buffer. Using `DeserializeSeed` instead makes this possible as in the +/// example code below. +/// +/// The canonical API for stateless deserialization looks like this: +/// +/// ```edition2018 +/// # use serde::Deserialize; +/// # +/// # enum Error {} +/// # +/// fn func<'de, T: Deserialize<'de>>() -> Result +/// # { +/// # unimplemented!() +/// # } +/// ``` +/// +/// Adjusting an API like this to support stateful deserialization is a matter +/// of accepting a seed as input: +/// +/// ```edition2018 +/// # use serde::de::DeserializeSeed; +/// # +/// # enum Error {} +/// # +/// fn func_seed<'de, T: DeserializeSeed<'de>>(seed: T) -> Result +/// # { +/// # let _ = seed; +/// # unimplemented!() +/// # } +/// ``` +/// +/// In practice the majority of deserialization is stateless. An API expecting a +/// seed can be appeased by passing `std::marker::PhantomData` as a seed in the +/// case of stateless deserialization. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by `Self::Value` when deserialized. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example +/// +/// Suppose we have JSON that looks like `[[1, 2], [3, 4, 5], [6]]` and we need +/// to deserialize it into a flat representation like `vec![1, 2, 3, 4, 5, 6]`. +/// Allocating a brand new `Vec` for each subarray would be slow. Instead we +/// would like to allocate a single `Vec` and then deserialize each subarray +/// into it. This requires stateful deserialization using the `DeserializeSeed` +/// trait. +/// +/// ```edition2018 +/// use std::fmt; +/// use std::marker::PhantomData; +/// +/// use serde::de::{Deserialize, DeserializeSeed, Deserializer, SeqAccess, Visitor}; +/// +/// // A DeserializeSeed implementation that uses stateful deserialization to +/// // append array elements onto the end of an existing vector. The preexisting +/// // state ("seed") in this case is the Vec. The `deserialize` method of +/// // `ExtendVec` will be traversing the inner arrays of the JSON input and +/// // appending each integer into the existing Vec. +/// struct ExtendVec<'a, T: 'a>(&'a mut Vec); +/// +/// impl<'de, 'a, T> DeserializeSeed<'de> for ExtendVec<'a, T> +/// where +/// T: Deserialize<'de>, +/// { +/// // The return type of the `deserialize` method. This implementation +/// // appends onto an existing vector but does not create any new data +/// // structure, so the return type is (). +/// type Value = (); +/// +/// fn deserialize(self, deserializer: D) -> Result +/// where +/// D: Deserializer<'de>, +/// { +/// // Visitor implementation that will walk an inner array of the JSON +/// // input. +/// struct ExtendVecVisitor<'a, T: 'a>(&'a mut Vec); +/// +/// impl<'de, 'a, T> Visitor<'de> for ExtendVecVisitor<'a, T> +/// where +/// T: Deserialize<'de>, +/// { +/// type Value = (); +/// +/// fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// write!(formatter, "an array of integers") +/// } +/// +/// fn visit_seq(self, mut seq: A) -> Result<(), A::Error> +/// where +/// A: SeqAccess<'de>, +/// { +/// // Decrease the number of reallocations if there are many elements +/// if let Some(size_hint) = seq.size_hint() { +/// self.0.reserve(size_hint); +/// } +/// +/// // Visit each element in the inner array and push it onto +/// // the existing vector. +/// while let Some(elem) = seq.next_element()? { +/// self.0.push(elem); +/// } +/// Ok(()) +/// } +/// } +/// +/// deserializer.deserialize_seq(ExtendVecVisitor(self.0)) +/// } +/// } +/// +/// // Visitor implementation that will walk the outer array of the JSON input. +/// struct FlattenedVecVisitor(PhantomData); +/// +/// impl<'de, T> Visitor<'de> for FlattenedVecVisitor +/// where +/// T: Deserialize<'de>, +/// { +/// // This Visitor constructs a single Vec to hold the flattened +/// // contents of the inner arrays. +/// type Value = Vec; +/// +/// fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// write!(formatter, "an array of arrays") +/// } +/// +/// fn visit_seq(self, mut seq: A) -> Result, A::Error> +/// where +/// A: SeqAccess<'de>, +/// { +/// // Create a single Vec to hold the flattened contents. +/// let mut vec = Vec::new(); +/// +/// // Each iteration through this loop is one inner array. +/// while let Some(()) = seq.next_element_seed(ExtendVec(&mut vec))? { +/// // Nothing to do; inner array has been appended into `vec`. +/// } +/// +/// // Return the finished vec. +/// Ok(vec) +/// } +/// } +/// +/// # fn example<'de, D>(deserializer: D) -> Result<(), D::Error> +/// # where +/// # D: Deserializer<'de>, +/// # { +/// let visitor = FlattenedVecVisitor(PhantomData); +/// let flattened: Vec = deserializer.deserialize_seq(visitor)?; +/// # Ok(()) +/// # } +/// ``` +pub trait DeserializeSeed<'de>: Sized { + /// The type produced by using this seed. + type Value; + + /// Equivalent to the more common `Deserialize::deserialize` method, except + /// with some initial piece of data (the seed) passed in. + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>; +} + +impl<'de, T> DeserializeSeed<'de> for PhantomData +where + T: Deserialize<'de>, +{ + type Value = T; + + #[inline] + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + T::deserialize(deserializer) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A **data format** that can deserialize any data structure supported by +/// Serde. +/// +/// The role of this trait is to define the deserialization half of the [Serde +/// data model], which is a way to categorize every Rust data type into one of +/// 29 possible types. Each method of the `Deserializer` trait corresponds to one +/// of the types of the data model. +/// +/// Implementations of `Deserialize` map themselves into this data model by +/// passing to the `Deserializer` a `Visitor` implementation that can receive +/// these various types. +/// +/// The types that make up the Serde data model are: +/// +/// - **14 primitive types** +/// - bool +/// - i8, i16, i32, i64, i128 +/// - u8, u16, u32, u64, u128 +/// - f32, f64 +/// - char +/// - **string** +/// - UTF-8 bytes with a length and no null terminator. +/// - When serializing, all strings are handled equally. When deserializing, +/// there are three flavors of strings: transient, owned, and borrowed. +/// - **byte array** - \[u8\] +/// - Similar to strings, during deserialization byte arrays can be +/// transient, owned, or borrowed. +/// - **option** +/// - Either none or some value. +/// - **unit** +/// - The type of `()` in Rust. It represents an anonymous value containing +/// no data. +/// - **unit_struct** +/// - For example `struct Unit` or `PhantomData`. It represents a named +/// value containing no data. +/// - **unit_variant** +/// - For example the `E::A` and `E::B` in `enum E { A, B }`. +/// - **newtype_struct** +/// - For example `struct Millimeters(u8)`. +/// - **newtype_variant** +/// - For example the `E::N` in `enum E { N(u8) }`. +/// - **seq** +/// - A variably sized heterogeneous sequence of values, for example `Vec` +/// or `HashSet`. When serializing, the length may or may not be known +/// before iterating through all the data. When deserializing, the length +/// is determined by looking at the serialized data. +/// - **tuple** +/// - A statically sized heterogeneous sequence of values for which the +/// length will be known at deserialization time without looking at the +/// serialized data, for example `(u8,)` or `(String, u64, Vec)` or +/// `[u64; 10]`. +/// - **tuple_struct** +/// - A named tuple, for example `struct Rgb(u8, u8, u8)`. +/// - **tuple_variant** +/// - For example the `E::T` in `enum E { T(u8, u8) }`. +/// - **map** +/// - A heterogeneous key-value pairing, for example `BTreeMap`. +/// - **struct** +/// - A heterogeneous key-value pairing in which the keys are strings and +/// will be known at deserialization time without looking at the serialized +/// data, for example `struct S { r: u8, g: u8, b: u8 }`. +/// - **struct_variant** +/// - For example the `E::S` in `enum E { S { r: u8, g: u8, b: u8 } }`. +/// +/// The `Deserializer` trait supports two entry point styles which enables +/// different kinds of deserialization. +/// +/// 1. The `deserialize_any` method. Self-describing data formats like JSON are +/// able to look at the serialized data and tell what it represents. For +/// example the JSON deserializer may see an opening curly brace (`{`) and +/// know that it is seeing a map. If the data format supports +/// `Deserializer::deserialize_any`, it will drive the Visitor using whatever +/// type it sees in the input. JSON uses this approach when deserializing +/// `serde_json::Value` which is an enum that can represent any JSON +/// document. Without knowing what is in a JSON document, we can deserialize +/// it to `serde_json::Value` by going through +/// `Deserializer::deserialize_any`. +/// +/// 2. The various `deserialize_*` methods. Non-self-describing formats like +/// Postcard need to be told what is in the input in order to deserialize it. +/// The `deserialize_*` methods are hints to the deserializer for how to +/// interpret the next piece of input. Non-self-describing formats are not +/// able to deserialize something like `serde_json::Value` which relies on +/// `Deserializer::deserialize_any`. +/// +/// When implementing `Deserialize`, you should avoid relying on +/// `Deserializer::deserialize_any` unless you need to be told by the +/// Deserializer what type is in the input. Know that relying on +/// `Deserializer::deserialize_any` means your data type will be able to +/// deserialize from self-describing formats only, ruling out Postcard and many +/// others. +/// +/// [Serde data model]: https://serde.rs/data-model.html +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed from the input when deserializing. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website contains example code for +/// a basic JSON `Deserializer`. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait Deserializer<'de>: Sized { + /// The error type that can be returned if some error occurs during + /// deserialization. + type Error: Error; + + /// Require the `Deserializer` to figure out how to drive the visitor based + /// on what data type is in the input. + /// + /// When implementing `Deserialize`, you should avoid relying on + /// `Deserializer::deserialize_any` unless you need to be told by the + /// Deserializer what type is in the input. Know that relying on + /// `Deserializer::deserialize_any` means your data type will be able to + /// deserialize from self-describing formats only, ruling out Postcard and + /// many others. + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a `bool` value. + fn deserialize_bool(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an `i8` value. + fn deserialize_i8(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an `i16` value. + fn deserialize_i16(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an `i32` value. + fn deserialize_i32(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an `i64` value. + fn deserialize_i64(self, visitor: V) -> Result + where + V: Visitor<'de>; + + serde_if_integer128! { + /// Hint that the `Deserialize` type is expecting an `i128` value. + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default behavior unconditionally returns an error. + fn deserialize_i128(self, visitor: V) -> Result + where + V: Visitor<'de> + { + let _ = visitor; + Err(Error::custom("i128 is not supported")) + } + } + + /// Hint that the `Deserialize` type is expecting a `u8` value. + fn deserialize_u8(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a `u16` value. + fn deserialize_u16(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a `u32` value. + fn deserialize_u32(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a `u64` value. + fn deserialize_u64(self, visitor: V) -> Result + where + V: Visitor<'de>; + + serde_if_integer128! { + /// Hint that the `Deserialize` type is expecting an `u128` value. + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default behavior unconditionally returns an error. + fn deserialize_u128(self, visitor: V) -> Result + where + V: Visitor<'de> + { + let _ = visitor; + Err(Error::custom("u128 is not supported")) + } + } + + /// Hint that the `Deserialize` type is expecting a `f32` value. + fn deserialize_f32(self, visitor: V) -> Result + where + V: Visitor<'de> + { + let _ = visitor; + Err(Error::custom("f32 is not supported")) + } + + /// Hint that the `Deserialize` type is expecting a `f64` value. + fn deserialize_f64(self, visitor: V) -> Result + where + V: Visitor<'de> + { + let _ = visitor; + Err(Error::custom("f64 is not supported")) + } + + /// Hint that the `Deserialize` type is expecting a `char` value. + fn deserialize_char(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a string value and does + /// not benefit from taking ownership of buffered data owned by the + /// `Deserializer`. + /// + /// If the `Visitor` would benefit from taking ownership of `String` data, + /// indicate this to the `Deserializer` by using `deserialize_string` + /// instead. + fn deserialize_str(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a string value and would + /// benefit from taking ownership of buffered data owned by the + /// `Deserializer`. + /// + /// If the `Visitor` would not benefit from taking ownership of `String` + /// data, indicate that to the `Deserializer` by using `deserialize_str` + /// instead. + fn deserialize_string(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a byte array and does not + /// benefit from taking ownership of buffered data owned by the + /// `Deserializer`. + /// + /// If the `Visitor` would benefit from taking ownership of `Vec` data, + /// indicate this to the `Deserializer` by using `deserialize_byte_buf` + /// instead. + fn deserialize_bytes(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a byte array and would + /// benefit from taking ownership of buffered data owned by the + /// `Deserializer`. + /// + /// If the `Visitor` would not benefit from taking ownership of `Vec` + /// data, indicate that to the `Deserializer` by using `deserialize_bytes` + /// instead. + fn deserialize_byte_buf(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an optional value. + /// + /// This allows deserializers that encode an optional value as a nullable + /// value to convert the null value into `None` and a regular value into + /// `Some(value)`. + fn deserialize_option(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a unit value. + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a unit struct with a + /// particular name. + fn deserialize_unit_struct( + self, + name: &'static str, + visitor: V, + ) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a newtype struct with a + /// particular name. + fn deserialize_newtype_struct( + self, + name: &'static str, + visitor: V, + ) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a sequence of values. + fn deserialize_seq(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a sequence of values and + /// knows how many values there are without looking at the serialized data. + fn deserialize_tuple(self, len: usize, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a tuple struct with a + /// particular name and number of fields. + fn deserialize_tuple_struct( + self, + name: &'static str, + len: usize, + visitor: V, + ) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a map of key-value pairs. + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting a struct with a particular + /// name and fields. + fn deserialize_struct( + self, + name: &'static str, + fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting an enum value with a + /// particular name and possible variants. + fn deserialize_enum( + self, + name: &'static str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type is expecting the name of a struct + /// field or the discriminant of an enum variant. + fn deserialize_identifier(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Hint that the `Deserialize` type needs to deserialize a value whose type + /// doesn't matter because it is ignored. + /// + /// Deserializers for non-self-describing formats may not support this mode. + fn deserialize_ignored_any(self, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Determine whether `Deserialize` implementations should expect to + /// deserialize their human-readable form. + /// + /// Some types have a human-readable form that may be somewhat expensive to + /// construct, as well as a binary form that is compact and efficient. + /// Generally text-based formats like JSON and YAML will prefer to use the + /// human-readable one and binary formats like Postcard will prefer the + /// compact one. + /// + /// ```edition2018 + /// # use std::ops::Add; + /// # use std::str::FromStr; + /// # + /// # struct Timestamp; + /// # + /// # impl Timestamp { + /// # const EPOCH: Timestamp = Timestamp; + /// # } + /// # + /// # impl FromStr for Timestamp { + /// # type Err = String; + /// # fn from_str(_: &str) -> Result { + /// # unimplemented!() + /// # } + /// # } + /// # + /// # struct Duration; + /// # + /// # impl Duration { + /// # fn seconds(_: u64) -> Self { unimplemented!() } + /// # } + /// # + /// # impl Add for Timestamp { + /// # type Output = Timestamp; + /// # fn add(self, _: Duration) -> Self::Output { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::de::{self, Deserialize, Deserializer}; + /// + /// impl<'de> Deserialize<'de> for Timestamp { + /// fn deserialize(deserializer: D) -> Result + /// where + /// D: Deserializer<'de>, + /// { + /// if deserializer.is_human_readable() { + /// // Deserialize from a human-readable string like "2015-05-15T17:01:00Z". + /// let s = String::deserialize(deserializer)?; + /// Timestamp::from_str(&s).map_err(de::Error::custom) + /// } else { + /// // Deserialize from a compact binary representation, seconds since + /// // the Unix epoch. + /// let n = u64::deserialize(deserializer)?; + /// Ok(Timestamp::EPOCH + Duration::seconds(n)) + /// } + /// } + /// } + /// ``` + /// + /// The default implementation of this method returns `true`. Data formats + /// may override this to `false` to request a compact form for types that + /// support one. Note that modifying this method to change a format from + /// human-readable to compact or vice versa should be regarded as a breaking + /// change, as a value serialized in human-readable mode is not required to + /// deserialize from the same data in compact mode. + #[inline] + fn is_human_readable(&self) -> bool { + true + } + + // Not public API. + #[cfg(all(not(no_serde_derive), any(feature = "std", feature = "alloc")))] + #[doc(hidden)] + fn __deserialize_content( + self, + _: ::actually_private::T, + visitor: V, + ) -> Result<::private::de::Content<'de>, Self::Error> + where + V: Visitor<'de, Value = ::private::de::Content<'de>>, + { + self.deserialize_any(visitor) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// This trait represents a visitor that walks through a deserializer. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the requirement for lifetime of data +/// that may be borrowed by `Self::Value`. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example +/// +/// ```edition2018 +/// # use std::fmt; +/// # +/// # use serde::de::{self, Unexpected, Visitor}; +/// # +/// /// A visitor that deserializes a long string - a string containing at least +/// /// some minimum number of bytes. +/// struct LongString { +/// min: usize, +/// } +/// +/// impl<'de> Visitor<'de> for LongString { +/// type Value = String; +/// +/// fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { +/// write!(formatter, "a string containing at least {} bytes", self.min) +/// } +/// +/// fn visit_str(self, s: &str) -> Result +/// where +/// E: de::Error, +/// { +/// if s.len() >= self.min { +/// Ok(s.to_owned()) +/// } else { +/// Err(de::Error::invalid_value(Unexpected::Str(s), &self)) +/// } +/// } +/// } +/// ``` +pub trait Visitor<'de>: Sized { + /// The value produced by this visitor. + type Value; + + /// Format a message stating what data this Visitor expects to receive. + /// + /// This is used in error messages. The message should complete the sentence + /// "This Visitor expects to receive ...", for example the message could be + /// "an integer between 0 and 64". The message should not be capitalized and + /// should not end with a period. + /// + /// ```edition2018 + /// # use std::fmt; + /// # + /// # struct S { + /// # max: usize, + /// # } + /// # + /// # impl<'de> serde::de::Visitor<'de> for S { + /// # type Value = (); + /// # + /// fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + /// write!(formatter, "an integer between 0 and {}", self.max) + /// } + /// # } + /// ``` + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result; + + /// The input contains a boolean. + /// + /// The default implementation fails with a type error. + fn visit_bool(self, v: bool) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Bool(v), &self)) + } + + /// The input contains an `i8`. + /// + /// The default implementation forwards to [`visit_i64`]. + /// + /// [`visit_i64`]: #method.visit_i64 + fn visit_i8(self, v: i8) -> Result + where + E: Error, + { + self.visit_i64(v as i64) + } + + /// The input contains an `i16`. + /// + /// The default implementation forwards to [`visit_i64`]. + /// + /// [`visit_i64`]: #method.visit_i64 + fn visit_i16(self, v: i16) -> Result + where + E: Error, + { + self.visit_i64(v as i64) + } + + /// The input contains an `i32`. + /// + /// The default implementation forwards to [`visit_i64`]. + /// + /// [`visit_i64`]: #method.visit_i64 + fn visit_i32(self, v: i32) -> Result + where + E: Error, + { + self.visit_i64(v as i64) + } + + /// The input contains an `i64`. + /// + /// The default implementation fails with a type error. + fn visit_i64(self, v: i64) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Signed(v), &self)) + } + + serde_if_integer128! { + /// The input contains a `i128`. + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default implementation fails with a type error. + fn visit_i128(self, v: i128) -> Result + where + E: Error, + { + let mut buf = [0u8; 58]; + let mut writer = format::Buf::new(&mut buf); + fmt::Write::write_fmt(&mut writer, format_args!("integer `{}` as i128", v)).unwrap(); + Err(Error::invalid_type(Unexpected::Other(writer.as_str()), &self)) + } + } + + /// The input contains a `u8`. + /// + /// The default implementation forwards to [`visit_u64`]. + /// + /// [`visit_u64`]: #method.visit_u64 + fn visit_u8(self, v: u8) -> Result + where + E: Error, + { + self.visit_u64(v as u64) + } + + /// The input contains a `u16`. + /// + /// The default implementation forwards to [`visit_u64`]. + /// + /// [`visit_u64`]: #method.visit_u64 + fn visit_u16(self, v: u16) -> Result + where + E: Error, + { + self.visit_u64(v as u64) + } + + /// The input contains a `u32`. + /// + /// The default implementation forwards to [`visit_u64`]. + /// + /// [`visit_u64`]: #method.visit_u64 + fn visit_u32(self, v: u32) -> Result + where + E: Error, + { + self.visit_u64(v as u64) + } + + /// The input contains a `u64`. + /// + /// The default implementation fails with a type error. + fn visit_u64(self, v: u64) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Unsigned(v), &self)) + } + + serde_if_integer128! { + /// The input contains a `u128`. + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default implementation fails with a type error. + fn visit_u128(self, v: u128) -> Result + where + E: Error, + { + let mut buf = [0u8; 57]; + let mut writer = format::Buf::new(&mut buf); + fmt::Write::write_fmt(&mut writer, format_args!("integer `{}` as u128", v)).unwrap(); + Err(Error::invalid_type(Unexpected::Other(writer.as_str()), &self)) + } + } + + /// The input contains an `f32`. + /// + /// The default implementation forwards to [`visit_f64`]. + /// + /// [`visit_f64`]: #method.visit_f64 + #[cfg(not(no_fp_fmt_parse))] + fn visit_f32(self, v: f32) -> Result + where + E: Error, + { + self.visit_f64(v as f64) + } + + /// The input contains an `f64`. + /// + /// The default implementation fails with a type error. + #[cfg(not(no_fp_fmt_parse))] + fn visit_f64(self, v: f64) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Float(v), &self)) + } + + /// The input contains a `char`. + /// + /// The default implementation forwards to [`visit_str`] as a one-character + /// string. + /// + /// [`visit_str`]: #method.visit_str + #[inline] + fn visit_char(self, v: char) -> Result + where + E: Error, + { + self.visit_str(utf8::encode(v).as_str()) + } + + /// The input contains a string. The lifetime of the string is ephemeral and + /// it may be destroyed after this method returns. + /// + /// This method allows the `Deserializer` to avoid a copy by retaining + /// ownership of any buffered data. `Deserialize` implementations that do + /// not benefit from taking ownership of `String` data should indicate that + /// to the deserializer by using `Deserializer::deserialize_str` rather than + /// `Deserializer::deserialize_string`. + /// + /// It is never correct to implement `visit_string` without implementing + /// `visit_str`. Implement neither, both, or just `visit_str`. + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Str(v), &self)) + } + + /// The input contains a string that lives at least as long as the + /// `Deserializer`. + /// + /// This enables zero-copy deserialization of strings in some formats. For + /// example JSON input containing the JSON string `"borrowed"` can be + /// deserialized with zero copying into a `&'a str` as long as the input + /// data outlives `'a`. + /// + /// The default implementation forwards to `visit_str`. + #[inline] + fn visit_borrowed_str(self, v: &'de str) -> Result + where + E: Error, + { + self.visit_str(v) + } + + /// The input contains a string and ownership of the string is being given + /// to the `Visitor`. + /// + /// This method allows the `Visitor` to avoid a copy by taking ownership of + /// a string created by the `Deserializer`. `Deserialize` implementations + /// that benefit from taking ownership of `String` data should indicate that + /// to the deserializer by using `Deserializer::deserialize_string` rather + /// than `Deserializer::deserialize_str`, although not every deserializer + /// will honor such a request. + /// + /// It is never correct to implement `visit_string` without implementing + /// `visit_str`. Implement neither, both, or just `visit_str`. + /// + /// The default implementation forwards to `visit_str` and then drops the + /// `String`. + #[inline] + #[cfg(any(feature = "std", feature = "alloc"))] + fn visit_string(self, v: String) -> Result + where + E: Error, + { + self.visit_str(&v) + } + + /// The input contains a byte array. The lifetime of the byte array is + /// ephemeral and it may be destroyed after this method returns. + /// + /// This method allows the `Deserializer` to avoid a copy by retaining + /// ownership of any buffered data. `Deserialize` implementations that do + /// not benefit from taking ownership of `Vec` data should indicate that + /// to the deserializer by using `Deserializer::deserialize_bytes` rather + /// than `Deserializer::deserialize_byte_buf`. + /// + /// It is never correct to implement `visit_byte_buf` without implementing + /// `visit_bytes`. Implement neither, both, or just `visit_bytes`. + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + let _ = v; + Err(Error::invalid_type(Unexpected::Bytes(v), &self)) + } + + /// The input contains a byte array that lives at least as long as the + /// `Deserializer`. + /// + /// This enables zero-copy deserialization of bytes in some formats. For + /// example Postcard data containing bytes can be deserialized with zero + /// copying into a `&'a [u8]` as long as the input data outlives `'a`. + /// + /// The default implementation forwards to `visit_bytes`. + #[inline] + fn visit_borrowed_bytes(self, v: &'de [u8]) -> Result + where + E: Error, + { + self.visit_bytes(v) + } + + /// The input contains a byte array and ownership of the byte array is being + /// given to the `Visitor`. + /// + /// This method allows the `Visitor` to avoid a copy by taking ownership of + /// a byte buffer created by the `Deserializer`. `Deserialize` + /// implementations that benefit from taking ownership of `Vec` data + /// should indicate that to the deserializer by using + /// `Deserializer::deserialize_byte_buf` rather than + /// `Deserializer::deserialize_bytes`, although not every deserializer will + /// honor such a request. + /// + /// It is never correct to implement `visit_byte_buf` without implementing + /// `visit_bytes`. Implement neither, both, or just `visit_bytes`. + /// + /// The default implementation forwards to `visit_bytes` and then drops the + /// `Vec`. + #[cfg(any(feature = "std", feature = "alloc"))] + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + self.visit_bytes(&v) + } + + /// The input contains an optional that is absent. + /// + /// The default implementation fails with a type error. + fn visit_none(self) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Option, &self)) + } + + /// The input contains an optional that is present. + /// + /// The default implementation fails with a type error. + fn visit_some(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + let _ = deserializer; + Err(Error::invalid_type(Unexpected::Option, &self)) + } + + /// The input contains a unit `()`. + /// + /// The default implementation fails with a type error. + fn visit_unit(self) -> Result + where + E: Error, + { + Err(Error::invalid_type(Unexpected::Unit, &self)) + } + + /// The input contains a newtype struct. + /// + /// The content of the newtype struct may be read from the given + /// `Deserializer`. + /// + /// The default implementation fails with a type error. + fn visit_newtype_struct(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + let _ = deserializer; + Err(Error::invalid_type(Unexpected::NewtypeStruct, &self)) + } + + /// The input contains a sequence of elements. + /// + /// The default implementation fails with a type error. + fn visit_seq(self, seq: A) -> Result + where + A: SeqAccess<'de>, + { + let _ = seq; + Err(Error::invalid_type(Unexpected::Seq, &self)) + } + + /// The input contains a key-value map. + /// + /// The default implementation fails with a type error. + fn visit_map(self, map: A) -> Result + where + A: MapAccess<'de>, + { + let _ = map; + Err(Error::invalid_type(Unexpected::Map, &self)) + } + + /// The input contains an enum. + /// + /// The default implementation fails with a type error. + fn visit_enum(self, data: A) -> Result + where + A: EnumAccess<'de>, + { + let _ = data; + Err(Error::invalid_type(Unexpected::Enum, &self)) + } + + // Used when deserializing a flattened Option field. Not public API. + #[doc(hidden)] + fn __private_visit_untagged_option(self, _: D) -> Result + where + D: Deserializer<'de>, + { + Err(()) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Provides a `Visitor` access to each element of a sequence in the input. +/// +/// This is a trait that a `Deserializer` passes to a `Visitor` implementation, +/// which deserializes each item in a sequence. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by deserialized sequence elements. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SeqAccess` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SeqAccess<'de> { + /// The error type that can be returned if some error occurs during + /// deserialization. + type Error: Error; + + /// This returns `Ok(Some(value))` for the next value in the sequence, or + /// `Ok(None)` if there are no more remaining items. + /// + /// `Deserialize` implementations should typically use + /// `SeqAccess::next_element` instead. + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: DeserializeSeed<'de>; + + /// This returns `Ok(Some(value))` for the next value in the sequence, or + /// `Ok(None)` if there are no more remaining items. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `SeqAccess` implementations should not override the default behavior. + #[inline] + fn next_element(&mut self) -> Result, Self::Error> + where + T: Deserialize<'de>, + { + self.next_element_seed(PhantomData) + } + + /// Returns the number of elements remaining in the sequence, if known. + #[inline] + fn size_hint(&self) -> Option { + None + } +} + +impl<'de, 'a, A: ?Sized> SeqAccess<'de> for &'a mut A +where + A: SeqAccess<'de>, +{ + type Error = A::Error; + + #[inline] + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: DeserializeSeed<'de>, + { + (**self).next_element_seed(seed) + } + + #[inline] + fn next_element(&mut self) -> Result, Self::Error> + where + T: Deserialize<'de>, + { + (**self).next_element() + } + + #[inline] + fn size_hint(&self) -> Option { + (**self).size_hint() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Provides a `Visitor` access to each entry of a map in the input. +/// +/// This is a trait that a `Deserializer` passes to a `Visitor` implementation. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by deserialized map entries. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `MapAccess` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait MapAccess<'de> { + /// The error type that can be returned if some error occurs during + /// deserialization. + type Error: Error; + + /// This returns `Ok(Some(key))` for the next key in the map, or `Ok(None)` + /// if there are no more remaining entries. + /// + /// `Deserialize` implementations should typically use + /// `MapAccess::next_key` or `MapAccess::next_entry` instead. + fn next_key_seed(&mut self, seed: K) -> Result, Self::Error> + where + K: DeserializeSeed<'de>; + + /// This returns a `Ok(value)` for the next value in the map. + /// + /// `Deserialize` implementations should typically use + /// `MapAccess::next_value` instead. + /// + /// # Panics + /// + /// Calling `next_value_seed` before `next_key_seed` is incorrect and is + /// allowed to panic or return bogus results. + fn next_value_seed(&mut self, seed: V) -> Result + where + V: DeserializeSeed<'de>; + + /// This returns `Ok(Some((key, value)))` for the next (key-value) pair in + /// the map, or `Ok(None)` if there are no more remaining items. + /// + /// `MapAccess` implementations should override the default behavior if a + /// more efficient implementation is possible. + /// + /// `Deserialize` implementations should typically use + /// `MapAccess::next_entry` instead. + #[inline] + fn next_entry_seed( + &mut self, + kseed: K, + vseed: V, + ) -> Result, Self::Error> + where + K: DeserializeSeed<'de>, + V: DeserializeSeed<'de>, + { + match try!(self.next_key_seed(kseed)) { + Some(key) => { + let value = try!(self.next_value_seed(vseed)); + Ok(Some((key, value))) + } + None => Ok(None), + } + } + + /// This returns `Ok(Some(key))` for the next key in the map, or `Ok(None)` + /// if there are no more remaining entries. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `MapAccess` implementations should not override the default behavior. + #[inline] + fn next_key(&mut self) -> Result, Self::Error> + where + K: Deserialize<'de>, + { + self.next_key_seed(PhantomData) + } + + /// This returns a `Ok(value)` for the next value in the map. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `MapAccess` implementations should not override the default behavior. + /// + /// # Panics + /// + /// Calling `next_value` before `next_key` is incorrect and is allowed to + /// panic or return bogus results. + #[inline] + fn next_value(&mut self) -> Result + where + V: Deserialize<'de>, + { + self.next_value_seed(PhantomData) + } + + /// This returns `Ok(Some((key, value)))` for the next (key-value) pair in + /// the map, or `Ok(None)` if there are no more remaining items. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `MapAccess` implementations should not override the default behavior. + #[inline] + fn next_entry(&mut self) -> Result, Self::Error> + where + K: Deserialize<'de>, + V: Deserialize<'de>, + { + self.next_entry_seed(PhantomData, PhantomData) + } + + /// Returns the number of entries remaining in the map, if known. + #[inline] + fn size_hint(&self) -> Option { + None + } +} + +impl<'de, 'a, A: ?Sized> MapAccess<'de> for &'a mut A +where + A: MapAccess<'de>, +{ + type Error = A::Error; + + #[inline] + fn next_key_seed(&mut self, seed: K) -> Result, Self::Error> + where + K: DeserializeSeed<'de>, + { + (**self).next_key_seed(seed) + } + + #[inline] + fn next_value_seed(&mut self, seed: V) -> Result + where + V: DeserializeSeed<'de>, + { + (**self).next_value_seed(seed) + } + + #[inline] + fn next_entry_seed( + &mut self, + kseed: K, + vseed: V, + ) -> Result, Self::Error> + where + K: DeserializeSeed<'de>, + V: DeserializeSeed<'de>, + { + (**self).next_entry_seed(kseed, vseed) + } + + #[inline] + fn next_entry(&mut self) -> Result, Self::Error> + where + K: Deserialize<'de>, + V: Deserialize<'de>, + { + (**self).next_entry() + } + + #[inline] + fn next_key(&mut self) -> Result, Self::Error> + where + K: Deserialize<'de>, + { + (**self).next_key() + } + + #[inline] + fn next_value(&mut self) -> Result + where + V: Deserialize<'de>, + { + (**self).next_value() + } + + #[inline] + fn size_hint(&self) -> Option { + (**self).size_hint() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Provides a `Visitor` access to the data of an enum in the input. +/// +/// `EnumAccess` is created by the `Deserializer` and passed to the +/// `Visitor` in order to identify which variant of an enum to deserialize. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by the deserialized enum variant. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `EnumAccess` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait EnumAccess<'de>: Sized { + /// The error type that can be returned if some error occurs during + /// deserialization. + type Error: Error; + /// The `Visitor` that will be used to deserialize the content of the enum + /// variant. + type Variant: VariantAccess<'de, Error = Self::Error>; + + /// `variant` is called to identify which variant to deserialize. + /// + /// `Deserialize` implementations should typically use `EnumAccess::variant` + /// instead. + fn variant_seed(self, seed: V) -> Result<(V::Value, Self::Variant), Self::Error> + where + V: DeserializeSeed<'de>; + + /// `variant` is called to identify which variant to deserialize. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `EnumAccess` implementations should not override the default behavior. + #[inline] + fn variant(self) -> Result<(V, Self::Variant), Self::Error> + where + V: Deserialize<'de>, + { + self.variant_seed(PhantomData) + } +} + +/// `VariantAccess` is a visitor that is created by the `Deserializer` and +/// passed to the `Deserialize` to deserialize the content of a particular enum +/// variant. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed by the deserialized enum variant. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `VariantAccess` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait VariantAccess<'de>: Sized { + /// The error type that can be returned if some error occurs during + /// deserialization. Must match the error type of our `EnumAccess`. + type Error: Error; + + /// Called when deserializing a variant with no values. + /// + /// If the data contains a different type of variant, the following + /// `invalid_type` error should be constructed: + /// + /// ```edition2018 + /// # use serde::de::{self, value, DeserializeSeed, Visitor, VariantAccess, Unexpected}; + /// # + /// # struct X; + /// # + /// # impl<'de> VariantAccess<'de> for X { + /// # type Error = value::Error; + /// # + /// fn unit_variant(self) -> Result<(), Self::Error> { + /// // What the data actually contained; suppose it is a tuple variant. + /// let unexp = Unexpected::TupleVariant; + /// Err(de::Error::invalid_type(unexp, &"unit variant")) + /// } + /// # + /// # fn newtype_variant_seed(self, _: T) -> Result + /// # where + /// # T: DeserializeSeed<'de>, + /// # { unimplemented!() } + /// # + /// # fn tuple_variant(self, _: usize, _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # + /// # fn struct_variant(self, _: &[&str], _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # } + /// ``` + fn unit_variant(self) -> Result<(), Self::Error>; + + /// Called when deserializing a variant with a single value. + /// + /// `Deserialize` implementations should typically use + /// `VariantAccess::newtype_variant` instead. + /// + /// If the data contains a different type of variant, the following + /// `invalid_type` error should be constructed: + /// + /// ```edition2018 + /// # use serde::de::{self, value, DeserializeSeed, Visitor, VariantAccess, Unexpected}; + /// # + /// # struct X; + /// # + /// # impl<'de> VariantAccess<'de> for X { + /// # type Error = value::Error; + /// # + /// # fn unit_variant(self) -> Result<(), Self::Error> { + /// # unimplemented!() + /// # } + /// # + /// fn newtype_variant_seed(self, _seed: T) -> Result + /// where + /// T: DeserializeSeed<'de>, + /// { + /// // What the data actually contained; suppose it is a unit variant. + /// let unexp = Unexpected::UnitVariant; + /// Err(de::Error::invalid_type(unexp, &"newtype variant")) + /// } + /// # + /// # fn tuple_variant(self, _: usize, _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # + /// # fn struct_variant(self, _: &[&str], _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # } + /// ``` + fn newtype_variant_seed(self, seed: T) -> Result + where + T: DeserializeSeed<'de>; + + /// Called when deserializing a variant with a single value. + /// + /// This method exists as a convenience for `Deserialize` implementations. + /// `VariantAccess` implementations should not override the default + /// behavior. + #[inline] + fn newtype_variant(self) -> Result + where + T: Deserialize<'de>, + { + self.newtype_variant_seed(PhantomData) + } + + /// Called when deserializing a tuple-like variant. + /// + /// The `len` is the number of fields expected in the tuple variant. + /// + /// If the data contains a different type of variant, the following + /// `invalid_type` error should be constructed: + /// + /// ```edition2018 + /// # use serde::de::{self, value, DeserializeSeed, Visitor, VariantAccess, Unexpected}; + /// # + /// # struct X; + /// # + /// # impl<'de> VariantAccess<'de> for X { + /// # type Error = value::Error; + /// # + /// # fn unit_variant(self) -> Result<(), Self::Error> { + /// # unimplemented!() + /// # } + /// # + /// # fn newtype_variant_seed(self, _: T) -> Result + /// # where + /// # T: DeserializeSeed<'de>, + /// # { unimplemented!() } + /// # + /// fn tuple_variant( + /// self, + /// _len: usize, + /// _visitor: V, + /// ) -> Result + /// where + /// V: Visitor<'de>, + /// { + /// // What the data actually contained; suppose it is a unit variant. + /// let unexp = Unexpected::UnitVariant; + /// Err(de::Error::invalid_type(unexp, &"tuple variant")) + /// } + /// # + /// # fn struct_variant(self, _: &[&str], _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # } + /// ``` + fn tuple_variant(self, len: usize, visitor: V) -> Result + where + V: Visitor<'de>; + + /// Called when deserializing a struct-like variant. + /// + /// The `fields` are the names of the fields of the struct variant. + /// + /// If the data contains a different type of variant, the following + /// `invalid_type` error should be constructed: + /// + /// ```edition2018 + /// # use serde::de::{self, value, DeserializeSeed, Visitor, VariantAccess, Unexpected}; + /// # + /// # struct X; + /// # + /// # impl<'de> VariantAccess<'de> for X { + /// # type Error = value::Error; + /// # + /// # fn unit_variant(self) -> Result<(), Self::Error> { + /// # unimplemented!() + /// # } + /// # + /// # fn newtype_variant_seed(self, _: T) -> Result + /// # where + /// # T: DeserializeSeed<'de>, + /// # { unimplemented!() } + /// # + /// # fn tuple_variant(self, _: usize, _: V) -> Result + /// # where + /// # V: Visitor<'de>, + /// # { unimplemented!() } + /// # + /// fn struct_variant( + /// self, + /// _fields: &'static [&'static str], + /// _visitor: V, + /// ) -> Result + /// where + /// V: Visitor<'de>, + /// { + /// // What the data actually contained; suppose it is a unit variant. + /// let unexp = Unexpected::UnitVariant; + /// Err(de::Error::invalid_type(unexp, &"struct variant")) + /// } + /// # } + /// ``` + fn struct_variant( + self, + fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>; +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Converts an existing value into a `Deserializer` from which other values can +/// be deserialized. +/// +/// # Lifetime +/// +/// The `'de` lifetime of this trait is the lifetime of data that may be +/// borrowed from the resulting `Deserializer`. See the page [Understanding +/// deserializer lifetimes] for a more detailed explanation of these lifetimes. +/// +/// [Understanding deserializer lifetimes]: https://serde.rs/lifetimes.html +/// +/// # Example +/// +/// ```edition2018 +/// use std::str::FromStr; +/// use serde::Deserialize; +/// use serde::de::{value, IntoDeserializer}; +/// +/// #[derive(Deserialize)] +/// enum Setting { +/// On, +/// Off, +/// } +/// +/// impl FromStr for Setting { +/// type Err = value::Error; +/// +/// fn from_str(s: &str) -> Result { +/// Self::deserialize(s.into_deserializer()) +/// } +/// } +/// ``` +pub trait IntoDeserializer<'de, E: Error = value::Error> { + /// The type of the deserializer being converted into. + type Deserializer: Deserializer<'de, Error = E>; + + /// Convert this value into a deserializer. + fn into_deserializer(self) -> Self::Deserializer; +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Used in error messages. +/// +/// - expected `a` +/// - expected `a` or `b` +/// - expected one of `a`, `b`, `c` +/// +/// The slice of names must not be empty. +struct OneOf { + names: &'static [&'static str], +} + +impl Display for OneOf { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self.names.len() { + 0 => panic!(), // special case elsewhere + 1 => write!(formatter, "`{}`", self.names[0]), + 2 => write!(formatter, "`{}` or `{}`", self.names[0], self.names[1]), + _ => { + try!(write!(formatter, "one of ")); + for (i, alt) in self.names.iter().enumerate() { + if i > 0 { + try!(write!(formatter, ", ")); + } + try!(write!(formatter, "`{}`", alt)); + } + Ok(()) + } + } + } +} diff --git a/rust/serde/de/seed.rs b/rust/serde/de/seed.rs new file mode 100644 index 00000000000000..e0ceeddb100a26 --- /dev/null +++ b/rust/serde/de/seed.rs @@ -0,0 +1,21 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use de::{Deserialize, DeserializeSeed, Deserializer}; + +/// A DeserializeSeed helper for implementing deserialize_in_place Visitors. +/// +/// Wraps a mutable reference and calls deserialize_in_place on it. +pub struct InPlaceSeed<'a, T: 'a>(pub &'a mut T); + +impl<'a, 'de, T> DeserializeSeed<'de> for InPlaceSeed<'a, T> +where + T: Deserialize<'de>, +{ + type Value = (); + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + T::deserialize_in_place(deserializer, self.0) + } +} diff --git a/rust/serde/de/utf8.rs b/rust/serde/de/utf8.rs new file mode 100644 index 00000000000000..1fb30488967c7f --- /dev/null +++ b/rust/serde/de/utf8.rs @@ -0,0 +1,48 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +const TAG_CONT: u8 = 0b1000_0000; +const TAG_TWO_B: u8 = 0b1100_0000; +const TAG_THREE_B: u8 = 0b1110_0000; +const TAG_FOUR_B: u8 = 0b1111_0000; +const MAX_ONE_B: u32 = 0x80; +const MAX_TWO_B: u32 = 0x800; +const MAX_THREE_B: u32 = 0x10000; + +#[inline] +pub fn encode(c: char) -> Encode { + let code = c as u32; + let mut buf = [0; 4]; + let pos = if code < MAX_ONE_B { + buf[3] = code as u8; + 3 + } else if code < MAX_TWO_B { + buf[2] = (code >> 6 & 0x1F) as u8 | TAG_TWO_B; + buf[3] = (code & 0x3F) as u8 | TAG_CONT; + 2 + } else if code < MAX_THREE_B { + buf[1] = (code >> 12 & 0x0F) as u8 | TAG_THREE_B; + buf[2] = (code >> 6 & 0x3F) as u8 | TAG_CONT; + buf[3] = (code & 0x3F) as u8 | TAG_CONT; + 1 + } else { + buf[0] = (code >> 18 & 0x07) as u8 | TAG_FOUR_B; + buf[1] = (code >> 12 & 0x3F) as u8 | TAG_CONT; + buf[2] = (code >> 6 & 0x3F) as u8 | TAG_CONT; + buf[3] = (code & 0x3F) as u8 | TAG_CONT; + 0 + }; + Encode { buf: buf, pos: pos } +} + +pub struct Encode { + buf: [u8; 4], + pos: usize, +} + +impl Encode { + pub fn as_str(&self) -> &str { + str::from_utf8(&self.buf[self.pos..]).unwrap() + } +} diff --git a/rust/serde/de/value.rs b/rust/serde/de/value.rs new file mode 100644 index 00000000000000..dc7be06e92572f --- /dev/null +++ b/rust/serde/de/value.rs @@ -0,0 +1,1718 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Building blocks for deserializing basic values using the `IntoDeserializer` +//! trait. +//! +//! ```edition2018 +//! use std::str::FromStr; +//! use serde::Deserialize; +//! use serde::de::{value, IntoDeserializer}; +//! +//! #[derive(Deserialize)] +//! enum Setting { +//! On, +//! Off, +//! } +//! +//! impl FromStr for Setting { +//! type Err = value::Error; +//! +//! fn from_str(s: &str) -> Result { +//! Self::deserialize(s.into_deserializer()) +//! } +//! } +//! ``` + +use lib::*; + +use self::private::{First, Second}; +use __private::size_hint; +use de::{self, Deserializer, Expected, IntoDeserializer, SeqAccess, Visitor}; +use ser; + +//////////////////////////////////////////////////////////////////////////////// + +// For structs that contain a PhantomData. We do not want the trait +// bound `E: Clone` inferred by derive(Clone). +macro_rules! impl_copy_clone { + ($ty:ident $(<$lifetime:tt>)*) => { + impl<$($lifetime,)* E> Copy for $ty<$($lifetime,)* E> {} + + impl<$($lifetime,)* E> Clone for $ty<$($lifetime,)* E> { + fn clone(&self) -> Self { + *self + } + } + }; +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A minimal representation of all possible errors that can occur using the +/// `IntoDeserializer` trait. +#[derive(Clone, PartialEq)] +pub struct Error { + err: ErrorImpl, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +type ErrorImpl = Box; +#[cfg(not(any(feature = "std", feature = "alloc")))] +type ErrorImpl = (); + +impl de::Error for Error { + #[cfg(any(feature = "std", feature = "alloc"))] + #[cold] + fn custom(msg: T) -> Self + where + T: Display, + { + Error { + err: msg.to_string().into_boxed_str(), + } + } + + #[cfg(not(any(feature = "std", feature = "alloc")))] + #[cold] + fn custom(msg: T) -> Self + where + T: Display, + { + let _ = msg; + Error { err: () } + } +} + +impl ser::Error for Error { + #[cold] + fn custom(msg: T) -> Self + where + T: Display, + { + de::Error::custom(msg) + } +} + +impl Display for Error { + #[cfg(any(feature = "std", feature = "alloc"))] + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(&self.err) + } + + #[cfg(not(any(feature = "std", feature = "alloc")))] + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("Serde deserialization error") + } +} + +impl Debug for Error { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + let mut debug = formatter.debug_tuple("Error"); + #[cfg(any(feature = "std", feature = "alloc"))] + debug.field(&self.err); + debug.finish() + } +} + +#[cfg(feature = "std")] +impl error::Error for Error { + fn description(&self) -> &str { + &self.err + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl<'de, E> IntoDeserializer<'de, E> for () +where + E: de::Error, +{ + type Deserializer = UnitDeserializer; + + fn into_deserializer(self) -> UnitDeserializer { + UnitDeserializer::new() + } +} + +/// A deserializer holding a `()`. +pub struct UnitDeserializer { + marker: PhantomData, +} + +impl_copy_clone!(UnitDeserializer); + +impl UnitDeserializer { + #[allow(missing_docs)] + pub fn new() -> Self { + UnitDeserializer { + marker: PhantomData, + } + } +} + +impl<'de, E> de::Deserializer<'de> for UnitDeserializer +where + E: de::Error, +{ + type Error = E; + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf unit unit_struct newtype_struct seq tuple tuple_struct + map struct enum identifier ignored_any + } + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_unit() + } + + fn deserialize_option(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_none() + } +} + +impl Debug for UnitDeserializer { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.debug_struct("UnitDeserializer").finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer that cannot be instantiated. +#[cfg(feature = "unstable")] +pub struct NeverDeserializer { + never: !, + marker: PhantomData, +} + +#[cfg(feature = "unstable")] +impl<'de, E> IntoDeserializer<'de, E> for ! +where + E: de::Error, +{ + type Deserializer = NeverDeserializer; + + fn into_deserializer(self) -> Self::Deserializer { + self + } +} + +#[cfg(feature = "unstable")] +impl<'de, E> de::Deserializer<'de> for NeverDeserializer +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, _visitor: V) -> Result + where + V: de::Visitor<'de>, + { + self.never + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! primitive_deserializer { + ($ty:ty, $doc:tt, $name:ident, $method:ident $($cast:tt)*) => { + #[doc = "A deserializer holding"] + #[doc = $doc] + pub struct $name { + value: $ty, + marker: PhantomData + } + + impl_copy_clone!($name); + + impl<'de, E> IntoDeserializer<'de, E> for $ty + where + E: de::Error, + { + type Deserializer = $name; + + fn into_deserializer(self) -> $name { + $name::new(self) + } + } + + impl $name { + #[allow(missing_docs)] + pub fn new(value: $ty) -> Self { + $name { + value: value, + marker: PhantomData, + } + } + } + + impl<'de, E> de::Deserializer<'de> for $name + where + E: de::Error, + { + type Error = E; + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str + string bytes byte_buf option unit unit_struct newtype_struct seq + tuple tuple_struct map struct enum identifier ignored_any + } + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.$method(self.value $($cast)*) + } + } + + impl Debug for $name { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct(stringify!($name)) + .field("value", &self.value) + .finish() + } + } + } +} + +primitive_deserializer!(bool, "a `bool`.", BoolDeserializer, visit_bool); +primitive_deserializer!(i8, "an `i8`.", I8Deserializer, visit_i8); +primitive_deserializer!(i16, "an `i16`.", I16Deserializer, visit_i16); +primitive_deserializer!(i32, "an `i32`.", I32Deserializer, visit_i32); +primitive_deserializer!(i64, "an `i64`.", I64Deserializer, visit_i64); +primitive_deserializer!(isize, "an `isize`.", IsizeDeserializer, visit_i64 as i64); +primitive_deserializer!(u8, "a `u8`.", U8Deserializer, visit_u8); +primitive_deserializer!(u16, "a `u16`.", U16Deserializer, visit_u16); +primitive_deserializer!(u64, "a `u64`.", U64Deserializer, visit_u64); +primitive_deserializer!(usize, "a `usize`.", UsizeDeserializer, visit_u64 as u64); +#[cfg(not(no_fp_fmt_parse))] +primitive_deserializer!(f32, "an `f32`.", F32Deserializer, visit_f32); +#[cfg(not(no_fp_fmt_parse))] +primitive_deserializer!(f64, "an `f64`.", F64Deserializer, visit_f64); +primitive_deserializer!(char, "a `char`.", CharDeserializer, visit_char); + +serde_if_integer128! { + primitive_deserializer!(i128, "an `i128`.", I128Deserializer, visit_i128); + primitive_deserializer!(u128, "a `u128`.", U128Deserializer, visit_u128); +} + +/// A deserializer holding a `u32`. +pub struct U32Deserializer { + value: u32, + marker: PhantomData, +} + +impl_copy_clone!(U32Deserializer); + +impl<'de, E> IntoDeserializer<'de, E> for u32 +where + E: de::Error, +{ + type Deserializer = U32Deserializer; + + fn into_deserializer(self) -> U32Deserializer { + U32Deserializer::new(self) + } +} + +impl U32Deserializer { + #[allow(missing_docs)] + pub fn new(value: u32) -> Self { + U32Deserializer { + value: value, + marker: PhantomData, + } + } +} + +impl<'de, E> de::Deserializer<'de> for U32Deserializer +where + E: de::Error, +{ + type Error = E; + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_u32(self.value) + } + + fn deserialize_enum( + self, + name: &str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + let _ = name; + let _ = variants; + visitor.visit_enum(self) + } +} + +impl<'de, E> de::EnumAccess<'de> for U32Deserializer +where + E: de::Error, +{ + type Error = E; + type Variant = private::UnitOnly; + + fn variant_seed(self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self).map(private::unit_only) + } +} + +impl Debug for U32Deserializer { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("U32Deserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `&str`. +pub struct StrDeserializer<'a, E> { + value: &'a str, + marker: PhantomData, +} + +impl_copy_clone!(StrDeserializer<'de>); + +impl<'de, 'a, E> IntoDeserializer<'de, E> for &'a str +where + E: de::Error, +{ + type Deserializer = StrDeserializer<'a, E>; + + fn into_deserializer(self) -> StrDeserializer<'a, E> { + StrDeserializer::new(self) + } +} + +impl<'a, E> StrDeserializer<'a, E> { + #[allow(missing_docs)] + pub fn new(value: &'a str) -> Self { + StrDeserializer { + value: value, + marker: PhantomData, + } + } +} + +impl<'de, 'a, E> de::Deserializer<'de> for StrDeserializer<'a, E> +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_str(self.value) + } + + fn deserialize_enum( + self, + name: &str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + let _ = name; + let _ = variants; + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +impl<'de, 'a, E> de::EnumAccess<'de> for StrDeserializer<'a, E> +where + E: de::Error, +{ + type Error = E; + type Variant = private::UnitOnly; + + fn variant_seed(self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self).map(private::unit_only) + } +} + +impl<'a, E> Debug for StrDeserializer<'a, E> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("StrDeserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `&str` with a lifetime tied to another +/// deserializer. +pub struct BorrowedStrDeserializer<'de, E> { + value: &'de str, + marker: PhantomData, +} + +impl_copy_clone!(BorrowedStrDeserializer<'de>); + +impl<'de, E> BorrowedStrDeserializer<'de, E> { + /// Create a new borrowed deserializer from the given string. + pub fn new(value: &'de str) -> BorrowedStrDeserializer<'de, E> { + BorrowedStrDeserializer { + value: value, + marker: PhantomData, + } + } +} + +impl<'de, E> de::Deserializer<'de> for BorrowedStrDeserializer<'de, E> +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_borrowed_str(self.value) + } + + fn deserialize_enum( + self, + name: &str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + let _ = name; + let _ = variants; + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +impl<'de, E> de::EnumAccess<'de> for BorrowedStrDeserializer<'de, E> +where + E: de::Error, +{ + type Error = E; + type Variant = private::UnitOnly; + + fn variant_seed(self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self).map(private::unit_only) + } +} + +impl<'de, E> Debug for BorrowedStrDeserializer<'de, E> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("BorrowedStrDeserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `String`. +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct StringDeserializer { + value: String, + marker: PhantomData, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl Clone for StringDeserializer { + fn clone(&self) -> Self { + StringDeserializer { + value: self.value.clone(), + marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, E> IntoDeserializer<'de, E> for String +where + E: de::Error, +{ + type Deserializer = StringDeserializer; + + fn into_deserializer(self) -> StringDeserializer { + StringDeserializer::new(self) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl StringDeserializer { + #[allow(missing_docs)] + pub fn new(value: String) -> Self { + StringDeserializer { + value: value, + marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, E> de::Deserializer<'de> for StringDeserializer +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_string(self.value) + } + + fn deserialize_enum( + self, + name: &str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + let _ = name; + let _ = variants; + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, E> de::EnumAccess<'de> for StringDeserializer +where + E: de::Error, +{ + type Error = E; + type Variant = private::UnitOnly; + + fn variant_seed(self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self).map(private::unit_only) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl Debug for StringDeserializer { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("StringDeserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `Cow`. +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct CowStrDeserializer<'a, E> { + value: Cow<'a, str>, + marker: PhantomData, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, E> Clone for CowStrDeserializer<'a, E> { + fn clone(&self) -> Self { + CowStrDeserializer { + value: self.value.clone(), + marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, 'a, E> IntoDeserializer<'de, E> for Cow<'a, str> +where + E: de::Error, +{ + type Deserializer = CowStrDeserializer<'a, E>; + + fn into_deserializer(self) -> CowStrDeserializer<'a, E> { + CowStrDeserializer::new(self) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, E> CowStrDeserializer<'a, E> { + #[allow(missing_docs)] + pub fn new(value: Cow<'a, str>) -> Self { + CowStrDeserializer { + value: value, + marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, 'a, E> de::Deserializer<'de> for CowStrDeserializer<'a, E> +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + match self.value { + Cow::Borrowed(string) => visitor.visit_str(string), + Cow::Owned(string) => visitor.visit_string(string), + } + } + + fn deserialize_enum( + self, + name: &str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + let _ = name; + let _ = variants; + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, 'a, E> de::EnumAccess<'de> for CowStrDeserializer<'a, E> +where + E: de::Error, +{ + type Error = E; + type Variant = private::UnitOnly; + + fn variant_seed(self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self).map(private::unit_only) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, E> Debug for CowStrDeserializer<'a, E> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("CowStrDeserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `&[u8]`. Always calls [`Visitor::visit_bytes`]. +pub struct BytesDeserializer<'a, E> { + value: &'a [u8], + marker: PhantomData, +} + +impl<'a, E> BytesDeserializer<'a, E> { + /// Create a new deserializer from the given bytes. + pub fn new(value: &'a [u8]) -> Self { + BytesDeserializer { + value: value, + marker: PhantomData, + } + } +} + +impl_copy_clone!(BytesDeserializer<'a>); + +impl<'de, 'a, E> IntoDeserializer<'de, E> for &'a [u8] +where + E: de::Error, +{ + type Deserializer = BytesDeserializer<'a, E>; + + fn into_deserializer(self) -> BytesDeserializer<'a, E> { + BytesDeserializer::new(self) + } +} + +impl<'de, 'a, E> Deserializer<'de> for BytesDeserializer<'a, E> +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_bytes(self.value) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +impl<'a, E> Debug for BytesDeserializer<'a, E> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("BytesDeserializer") + .field("value", &self.value) + .finish() + } +} + +/// A deserializer holding a `&[u8]` with a lifetime tied to another +/// deserializer. Always calls [`Visitor::visit_borrowed_bytes`]. +pub struct BorrowedBytesDeserializer<'de, E> { + value: &'de [u8], + marker: PhantomData, +} + +impl<'de, E> BorrowedBytesDeserializer<'de, E> { + /// Create a new borrowed deserializer from the given borrowed bytes. + pub fn new(value: &'de [u8]) -> Self { + BorrowedBytesDeserializer { + value: value, + marker: PhantomData, + } + } +} + +impl_copy_clone!(BorrowedBytesDeserializer<'de>); + +impl<'de, E> Deserializer<'de> for BorrowedBytesDeserializer<'de, E> +where + E: de::Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_borrowed_bytes(self.value) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +impl<'de, E> Debug for BorrowedBytesDeserializer<'de, E> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("BorrowedBytesDeserializer") + .field("value", &self.value) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer that iterates over a sequence. +#[derive(Clone)] +pub struct SeqDeserializer { + iter: iter::Fuse, + count: usize, + marker: PhantomData, +} + +impl SeqDeserializer +where + I: Iterator, +{ + /// Construct a new `SeqDeserializer`. + pub fn new(iter: I) -> Self { + SeqDeserializer { + iter: iter.fuse(), + count: 0, + marker: PhantomData, + } + } +} + +impl SeqDeserializer +where + I: Iterator, + E: de::Error, +{ + /// Check for remaining elements after passing a `SeqDeserializer` to + /// `Visitor::visit_seq`. + pub fn end(self) -> Result<(), E> { + let remaining = self.iter.count(); + if remaining == 0 { + Ok(()) + } else { + // First argument is the number of elements in the data, second + // argument is the number of elements expected by the Deserialize. + Err(de::Error::invalid_length( + self.count + remaining, + &ExpectedInSeq(self.count), + )) + } + } +} + +impl<'de, I, T, E> de::Deserializer<'de> for SeqDeserializer +where + I: Iterator, + T: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn deserialize_any(mut self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let v = try!(visitor.visit_seq(&mut self)); + try!(self.end()); + Ok(v) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +impl<'de, I, T, E> de::SeqAccess<'de> for SeqDeserializer +where + I: Iterator, + T: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn next_element_seed(&mut self, seed: V) -> Result, Self::Error> + where + V: de::DeserializeSeed<'de>, + { + match self.iter.next() { + Some(value) => { + self.count += 1; + seed.deserialize(value.into_deserializer()).map(Some) + } + None => Ok(None), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } +} + +struct ExpectedInSeq(usize); + +impl Expected for ExpectedInSeq { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + if self.0 == 1 { + write!(formatter, "1 element in sequence") + } else { + write!(formatter, "{} elements in sequence", self.0) + } + } +} + +impl Debug for SeqDeserializer +where + I: Debug, +{ + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("SeqDeserializer") + .field("iter", &self.iter) + .field("count", &self.count) + .finish() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, T, E> IntoDeserializer<'de, E> for Vec +where + T: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Deserializer = SeqDeserializer<::IntoIter, E>; + + fn into_deserializer(self) -> Self::Deserializer { + SeqDeserializer::new(self.into_iter()) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, T, E> IntoDeserializer<'de, E> for BTreeSet +where + T: IntoDeserializer<'de, E> + Eq + Ord, + E: de::Error, +{ + type Deserializer = SeqDeserializer<::IntoIter, E>; + + fn into_deserializer(self) -> Self::Deserializer { + SeqDeserializer::new(self.into_iter()) + } +} + +#[cfg(feature = "std")] +impl<'de, T, S, E> IntoDeserializer<'de, E> for HashSet +where + T: IntoDeserializer<'de, E> + Eq + Hash, + S: BuildHasher, + E: de::Error, +{ + type Deserializer = SeqDeserializer<::IntoIter, E>; + + fn into_deserializer(self) -> Self::Deserializer { + SeqDeserializer::new(self.into_iter()) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `SeqAccess`. +#[derive(Clone, Debug)] +pub struct SeqAccessDeserializer { + seq: A, +} + +impl SeqAccessDeserializer { + /// Construct a new `SeqAccessDeserializer`. + pub fn new(seq: A) -> Self { + SeqAccessDeserializer { seq: seq } + } +} + +impl<'de, A> de::Deserializer<'de> for SeqAccessDeserializer +where + A: de::SeqAccess<'de>, +{ + type Error = A::Error; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_seq(self.seq) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer that iterates over a map. +pub struct MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, +{ + iter: iter::Fuse, + value: Option>, + count: usize, + lifetime: PhantomData<&'de ()>, + error: PhantomData, +} + +impl<'de, I, E> MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, +{ + /// Construct a new `MapDeserializer`. + pub fn new(iter: I) -> Self { + MapDeserializer { + iter: iter.fuse(), + value: None, + count: 0, + lifetime: PhantomData, + error: PhantomData, + } + } +} + +impl<'de, I, E> MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, + E: de::Error, +{ + /// Check for remaining elements after passing a `MapDeserializer` to + /// `Visitor::visit_map`. + pub fn end(self) -> Result<(), E> { + let remaining = self.iter.count(); + if remaining == 0 { + Ok(()) + } else { + // First argument is the number of elements in the data, second + // argument is the number of elements expected by the Deserialize. + Err(de::Error::invalid_length( + self.count + remaining, + &ExpectedInMap(self.count), + )) + } + } +} + +impl<'de, I, E> MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, +{ + fn next_pair(&mut self) -> Option<(First, Second)> { + match self.iter.next() { + Some(kv) => { + self.count += 1; + Some(private::Pair::split(kv)) + } + None => None, + } + } +} + +impl<'de, I, E> de::Deserializer<'de> for MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, + First: IntoDeserializer<'de, E>, + Second: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn deserialize_any(mut self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let value = try!(visitor.visit_map(&mut self)); + try!(self.end()); + Ok(value) + } + + fn deserialize_seq(mut self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let value = try!(visitor.visit_seq(&mut self)); + try!(self.end()); + Ok(value) + } + + fn deserialize_tuple(self, len: usize, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let _ = len; + self.deserialize_seq(visitor) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct tuple_struct map + struct enum identifier ignored_any + } +} + +impl<'de, I, E> de::MapAccess<'de> for MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, + First: IntoDeserializer<'de, E>, + Second: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.next_pair() { + Some((key, value)) => { + self.value = Some(value); + seed.deserialize(key.into_deserializer()).map(Some) + } + None => Ok(None), + } + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + let value = self.value.take(); + // Panic because this indicates a bug in the program rather than an + // expected failure. + let value = value.expect("MapAccess::next_value called before next_key"); + seed.deserialize(value.into_deserializer()) + } + + fn next_entry_seed( + &mut self, + kseed: TK, + vseed: TV, + ) -> Result, Self::Error> + where + TK: de::DeserializeSeed<'de>, + TV: de::DeserializeSeed<'de>, + { + match self.next_pair() { + Some((key, value)) => { + let key = try!(kseed.deserialize(key.into_deserializer())); + let value = try!(vseed.deserialize(value.into_deserializer())); + Ok(Some((key, value))) + } + None => Ok(None), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } +} + +impl<'de, I, E> de::SeqAccess<'de> for MapDeserializer<'de, I, E> +where + I: Iterator, + I::Item: private::Pair, + First: IntoDeserializer<'de, E>, + Second: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.next_pair() { + Some((k, v)) => { + let de = PairDeserializer(k, v, PhantomData); + seed.deserialize(de).map(Some) + } + None => Ok(None), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } +} + +// Cannot #[derive(Clone)] because of the bound `Second: Clone`. +impl<'de, I, E> Clone for MapDeserializer<'de, I, E> +where + I: Iterator + Clone, + I::Item: private::Pair, + Second: Clone, +{ + fn clone(&self) -> Self { + MapDeserializer { + iter: self.iter.clone(), + value: self.value.clone(), + count: self.count, + lifetime: self.lifetime, + error: self.error, + } + } +} + +impl<'de, I, E> Debug for MapDeserializer<'de, I, E> +where + I: Iterator + Debug, + I::Item: private::Pair, + Second: Debug, +{ + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("MapDeserializer") + .field("iter", &self.iter) + .field("value", &self.value) + .field("count", &self.count) + .finish() + } +} + +// Used in the `impl SeqAccess for MapDeserializer` to visit the map as a +// sequence of pairs. +struct PairDeserializer(A, B, PhantomData); + +impl<'de, A, B, E> de::Deserializer<'de> for PairDeserializer +where + A: IntoDeserializer<'de, E>, + B: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct tuple_struct map + struct enum identifier ignored_any + } + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_seq(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let mut pair_visitor = PairVisitor(Some(self.0), Some(self.1), PhantomData); + let pair = try!(visitor.visit_seq(&mut pair_visitor)); + if pair_visitor.1.is_none() { + Ok(pair) + } else { + let remaining = pair_visitor.size_hint().unwrap(); + // First argument is the number of elements in the data, second + // argument is the number of elements expected by the Deserialize. + Err(de::Error::invalid_length(2, &ExpectedInSeq(2 - remaining))) + } + } + + fn deserialize_tuple(self, len: usize, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + if len == 2 { + self.deserialize_seq(visitor) + } else { + // First argument is the number of elements in the data, second + // argument is the number of elements expected by the Deserialize. + Err(de::Error::invalid_length(2, &ExpectedInSeq(len))) + } + } +} + +struct PairVisitor(Option, Option, PhantomData); + +impl<'de, A, B, E> de::SeqAccess<'de> for PairVisitor +where + A: IntoDeserializer<'de, E>, + B: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Error = E; + + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + if let Some(k) = self.0.take() { + seed.deserialize(k.into_deserializer()).map(Some) + } else if let Some(v) = self.1.take() { + seed.deserialize(v.into_deserializer()).map(Some) + } else { + Ok(None) + } + } + + fn size_hint(&self) -> Option { + if self.0.is_some() { + Some(2) + } else if self.1.is_some() { + Some(1) + } else { + Some(0) + } + } +} + +struct ExpectedInMap(usize); + +impl Expected for ExpectedInMap { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + if self.0 == 1 { + write!(formatter, "1 element in map") + } else { + write!(formatter, "{} elements in map", self.0) + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'de, K, V, E> IntoDeserializer<'de, E> for BTreeMap +where + K: IntoDeserializer<'de, E> + Eq + Ord, + V: IntoDeserializer<'de, E>, + E: de::Error, +{ + type Deserializer = MapDeserializer<'de, ::IntoIter, E>; + + fn into_deserializer(self) -> Self::Deserializer { + MapDeserializer::new(self.into_iter()) + } +} + +#[cfg(feature = "std")] +impl<'de, K, V, S, E> IntoDeserializer<'de, E> for HashMap +where + K: IntoDeserializer<'de, E> + Eq + Hash, + V: IntoDeserializer<'de, E>, + S: BuildHasher, + E: de::Error, +{ + type Deserializer = MapDeserializer<'de, ::IntoIter, E>; + + fn into_deserializer(self) -> Self::Deserializer { + MapDeserializer::new(self.into_iter()) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding a `MapAccess`. +#[derive(Clone, Debug)] +pub struct MapAccessDeserializer { + map: A, +} + +impl MapAccessDeserializer { + /// Construct a new `MapAccessDeserializer`. + pub fn new(map: A) -> Self { + MapAccessDeserializer { map: map } + } +} + +impl<'de, A> de::Deserializer<'de> for MapAccessDeserializer +where + A: de::MapAccess<'de>, +{ + type Error = A::Error; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_map(self.map) + } + + fn deserialize_enum( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +impl<'de, A> de::EnumAccess<'de> for MapAccessDeserializer +where + A: de::MapAccess<'de>, +{ + type Error = A::Error; + type Variant = private::MapAsEnum; + + fn variant_seed(mut self, seed: T) -> Result<(T::Value, Self::Variant), Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match try!(self.map.next_key_seed(seed)) { + Some(key) => Ok((key, private::map_as_enum(self.map))), + None => Err(de::Error::invalid_type(de::Unexpected::Map, &"enum")), + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A deserializer holding an `EnumAccess`. +#[derive(Clone, Debug)] +pub struct EnumAccessDeserializer { + access: A, +} + +impl EnumAccessDeserializer { + /// Construct a new `EnumAccessDeserializer`. + pub fn new(access: A) -> Self { + EnumAccessDeserializer { access: access } + } +} + +impl<'de, A> de::Deserializer<'de> for EnumAccessDeserializer +where + A: de::EnumAccess<'de>, +{ + type Error = A::Error; + + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_enum(self.access) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +//////////////////////////////////////////////////////////////////////////////// + +mod private { + use lib::*; + + use de::{self, DeserializeSeed, Deserializer, MapAccess, Unexpected, VariantAccess, Visitor}; + + pub struct UnitOnly { + marker: PhantomData, + } + + pub fn unit_only(t: T) -> (T, UnitOnly) { + ( + t, + UnitOnly { + marker: PhantomData, + }, + ) + } + + impl<'de, E> de::VariantAccess<'de> for UnitOnly + where + E: de::Error, + { + type Error = E; + + fn unit_variant(self) -> Result<(), Self::Error> { + Ok(()) + } + + fn newtype_variant_seed(self, _seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"newtype variant", + )) + } + + fn tuple_variant(self, _len: usize, _visitor: V) -> Result + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"tuple variant", + )) + } + + fn struct_variant( + self, + _fields: &'static [&'static str], + _visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"struct variant", + )) + } + } + + pub struct MapAsEnum { + map: A, + } + + pub fn map_as_enum(map: A) -> MapAsEnum { + MapAsEnum { map: map } + } + + impl<'de, A> VariantAccess<'de> for MapAsEnum + where + A: MapAccess<'de>, + { + type Error = A::Error; + + fn unit_variant(mut self) -> Result<(), Self::Error> { + self.map.next_value() + } + + fn newtype_variant_seed(mut self, seed: T) -> Result + where + T: DeserializeSeed<'de>, + { + self.map.next_value_seed(seed) + } + + fn tuple_variant(mut self, len: usize, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.map.next_value_seed(SeedTupleVariant { + len: len, + visitor: visitor, + }) + } + + fn struct_variant( + mut self, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.map + .next_value_seed(SeedStructVariant { visitor: visitor }) + } + } + + struct SeedTupleVariant { + len: usize, + visitor: V, + } + + impl<'de, V> DeserializeSeed<'de> for SeedTupleVariant + where + V: Visitor<'de>, + { + type Value = V::Value; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_tuple(self.len, self.visitor) + } + } + + struct SeedStructVariant { + visitor: V, + } + + impl<'de, V> DeserializeSeed<'de> for SeedStructVariant + where + V: Visitor<'de>, + { + type Value = V::Value; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_map(self.visitor) + } + } + + /// Avoid having to restate the generic types on `MapDeserializer`. The + /// `Iterator::Item` contains enough information to figure out K and V. + pub trait Pair { + type First; + type Second; + fn split(self) -> (Self::First, Self::Second); + } + + impl Pair for (A, B) { + type First = A; + type Second = B; + fn split(self) -> (A, B) { + self + } + } + + pub type First = ::First; + pub type Second = ::Second; +} diff --git a/rust/serde/integer128.rs b/rust/serde/integer128.rs new file mode 100644 index 00000000000000..b47862335f28b6 --- /dev/null +++ b/rust/serde/integer128.rs @@ -0,0 +1,84 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Conditional compilation depending on whether Serde is built with support for +/// 128-bit integers. +/// +/// Data formats that wish to support Rust compiler versions older than 1.26 +/// (or targets that lack 128-bit integers) may place the i128 / u128 methods +/// of their Serializer and Deserializer behind this macro. +/// +/// Data formats that require a minimum Rust compiler version of at least 1.26, +/// or do not target platforms that lack 128-bit integers, do not need to +/// bother with this macro and may assume support for 128-bit integers. +/// +/// ```edition2018 +/// # use serde::__private::doc::Error; +/// # +/// # struct MySerializer; +/// # +/// use serde::{serde_if_integer128, Serializer}; +/// +/// impl Serializer for MySerializer { +/// type Ok = (); +/// type Error = Error; +/// +/// fn serialize_i64(self, v: i64) -> Result { +/// /* ... */ +/// # unimplemented!() +/// } +/// +/// /* ... */ +/// +/// serde_if_integer128! { +/// fn serialize_i128(self, v: i128) -> Result { +/// /* ... */ +/// # unimplemented!() +/// } +/// +/// fn serialize_u128(self, v: u128) -> Result { +/// /* ... */ +/// # unimplemented!() +/// } +/// } +/// # +/// # serde::__serialize_unimplemented! { +/// # bool i8 i16 i32 u8 u16 u32 u64 f32 f64 char str bytes none some +/// # unit unit_struct unit_variant newtype_struct newtype_variant seq +/// # tuple tuple_struct tuple_variant map struct struct_variant +/// # } +/// } +/// ``` +/// +/// When Serde is built with support for 128-bit integers, this macro expands +/// transparently into just the input tokens. +/// +/// ```edition2018 +/// macro_rules! serde_if_integer128 { +/// ($($tt:tt)*) => { +/// $($tt)* +/// }; +/// } +/// ``` +/// +/// When built without support for 128-bit integers, this macro expands to +/// nothing. +/// +/// ```edition2018 +/// macro_rules! serde_if_integer128 { +/// ($($tt:tt)*) => {}; +/// } +/// ``` +#[cfg(not(no_integer128))] +#[macro_export] +macro_rules! serde_if_integer128 { + ($($tt:tt)*) => { + $($tt)* + }; +} + +#[cfg(no_integer128)] +#[macro_export] +#[doc(hidden)] +macro_rules! serde_if_integer128 { + ($($tt:tt)*) => {}; +} diff --git a/rust/serde/lib.rs b/rust/serde/lib.rs new file mode 100644 index 00000000000000..c40cd2212d62b2 --- /dev/null +++ b/rust/serde/lib.rs @@ -0,0 +1,353 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! # Serde +//! +//! Serde is a framework for ***ser***ializing and ***de***serializing Rust data +//! structures efficiently and generically. +//! +//! The Serde ecosystem consists of data structures that know how to serialize +//! and deserialize themselves along with data formats that know how to +//! serialize and deserialize other things. Serde provides the layer by which +//! these two groups interact with each other, allowing any supported data +//! structure to be serialized and deserialized using any supported data format. +//! +//! See the Serde website for additional documentation and +//! usage examples. +//! +//! ## Design +//! +//! Where many other languages rely on runtime reflection for serializing data, +//! Serde is instead built on Rust's powerful trait system. A data structure +//! that knows how to serialize and deserialize itself is one that implements +//! Serde's `Serialize` and `Deserialize` traits (or uses Serde's derive +//! attribute to automatically generate implementations at compile time). This +//! avoids any overhead of reflection or runtime type information. In fact in +//! many situations the interaction between data structure and data format can +//! be completely optimized away by the Rust compiler, leaving Serde +//! serialization to perform the same speed as a handwritten serializer for the +//! specific selection of data structure and data format. +//! +//! ## Data formats +//! +//! The following is a partial list of data formats that have been implemented +//! for Serde by the community. +//! +//! - [JSON], the ubiquitous JavaScript Object Notation used by many HTTP APIs. +//! - [Postcard], a no\_std and embedded-systems friendly compact binary format. +//! - [CBOR], a Concise Binary Object Representation designed for small message +//! size without the need for version negotiation. +//! - [YAML], a self-proclaimed human-friendly configuration language that ain't +//! markup language. +//! - [MessagePack], an efficient binary format that resembles a compact JSON. +//! - [TOML], a minimal configuration format used by [Cargo]. +//! - [Pickle], a format common in the Python world. +//! - [RON], a Rusty Object Notation. +//! - [BSON], the data storage and network transfer format used by MongoDB. +//! - [Avro], a binary format used within Apache Hadoop, with support for schema +//! definition. +//! - [JSON5], a superset of JSON including some productions from ES5. +//! - [URL] query strings, in the x-www-form-urlencoded format. +//! - [Starlark], the format used for describing build targets by the Bazel and +//! Buck build systems. *(serialization only)* +//! - [Envy], a way to deserialize environment variables into Rust structs. +//! *(deserialization only)* +//! - [Envy Store], a way to deserialize [AWS Parameter Store] parameters into +//! Rust structs. *(deserialization only)* +//! - [S-expressions], the textual representation of code and data used by the +//! Lisp language family. +//! - [D-Bus]'s binary wire format. +//! - [FlexBuffers], the schemaless cousin of Google's FlatBuffers zero-copy +//! serialization format. +//! - [Bencode], a simple binary format used in the BitTorrent protocol. +//! - [Token streams], for processing Rust procedural macro input. +//! *(deserialization only)* +//! - [DynamoDB Items], the format used by [rusoto_dynamodb] to transfer data to +//! and from DynamoDB. +//! - [Hjson], a syntax extension to JSON designed around human reading and +//! editing. *(deserialization only)* +//! +//! [JSON]: https://github.com/serde-rs/json +//! [Postcard]: https://github.com/jamesmunns/postcard +//! [CBOR]: https://github.com/enarx/ciborium +//! [YAML]: https://github.com/dtolnay/serde-yaml +//! [MessagePack]: https://github.com/3Hren/msgpack-rust +//! [TOML]: https://docs.rs/toml +//! [Pickle]: https://github.com/birkenfeld/serde-pickle +//! [RON]: https://github.com/ron-rs/ron +//! [BSON]: https://github.com/mongodb/bson-rust +//! [Avro]: https://docs.rs/apache-avro +//! [JSON5]: https://github.com/callum-oakley/json5-rs +//! [URL]: https://docs.rs/serde_qs +//! [Starlark]: https://github.com/dtolnay/serde-starlark +//! [Envy]: https://github.com/softprops/envy +//! [Envy Store]: https://github.com/softprops/envy-store +//! [Cargo]: https://doc.rust-lang.org/cargo/reference/manifest.html +//! [AWS Parameter Store]: https://docs.aws.amazon.com/systems-manager/latest/userguide/systems-manager-parameter-store.html +//! [S-expressions]: https://github.com/rotty/lexpr-rs +//! [D-Bus]: https://docs.rs/zvariant +//! [FlexBuffers]: https://github.com/google/flatbuffers/tree/master/rust/flexbuffers +//! [Bencode]: https://github.com/P3KI/bendy +//! [Token streams]: https://github.com/oxidecomputer/serde_tokenstream +//! [DynamoDB Items]: https://docs.rs/serde_dynamo +//! [rusoto_dynamodb]: https://docs.rs/rusoto_dynamodb +//! [Hjson]: https://github.com/Canop/deser-hjson + +//////////////////////////////////////////////////////////////////////////////// + +#![feature(core_c_str)] + +// Serde types in rustdoc of other crates get linked to here. +#![doc(html_root_url = "https://docs.rs/serde/1.0.156")] +// Support using Serde without the standard library! +#![cfg_attr(not(feature = "std"), no_std)] +// Unstable functionality only if the user asks for it. For tracking and +// discussion of these features please refer to this issue: +// +// https://github.com/serde-rs/serde/issues/812 +#![cfg_attr(feature = "unstable", feature(error_in_core, never_type))] +#![allow(unknown_lints, bare_trait_objects, deprecated)] +#![cfg_attr(feature = "cargo-clippy", allow(renamed_and_removed_lints))] +// Ignored clippy and clippy_pedantic lints +#![cfg_attr( + feature = "cargo-clippy", + allow( + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/5704 + unnested_or_patterns, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/7768 + semicolon_if_nothing_returned, + // not available in our oldest supported compiler + empty_enum, + type_repetition_in_bounds, // https://github.com/rust-lang/rust-clippy/issues/8772 + // integer and float ser/de requires these sorts of casts + cast_possible_truncation, + cast_possible_wrap, + cast_sign_loss, + // things are often more readable this way + cast_lossless, + module_name_repetitions, + option_if_let_else, + single_match_else, + type_complexity, + use_self, + zero_prefixed_literal, + // correctly used + derive_partial_eq_without_eq, + enum_glob_use, + explicit_auto_deref, + let_underscore_untyped, + map_err_ignore, + new_without_default, + result_unit_err, + wildcard_imports, + // not practical + needless_pass_by_value, + similar_names, + too_many_lines, + // preference + doc_markdown, + unseparated_literal_suffix, + // false positive + needless_doctest_main, + // noisy + missing_errors_doc, + must_use_candidate, + ) +)] +// Rustc lints. +#![deny(missing_docs, unused_imports)] + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "alloc")] +extern crate alloc; + +/// A facade around all the types we need from the `std`, `core`, and `alloc` +/// crates. This avoids elaborate import wrangling having to happen in every +/// module. +mod lib { + mod core { + #[cfg(not(feature = "std"))] + pub use core::*; + #[cfg(feature = "std")] + pub use std::*; + } + + pub use self::core::{cmp, iter, mem, num, ptr, slice, str}; + pub use self::core::{f32, f64}; + pub use self::core::{i16, i32, i64, i8, isize}; + pub use self::core::{u16, u32, u64, u8, usize}; + + pub use self::core::cell::{Cell, RefCell}; + pub use self::core::clone::{self, Clone}; + pub use self::core::convert::{self, From, Into}; + pub use self::core::default::{self, Default}; + pub use self::core::fmt::{self, Debug, Display}; + pub use self::core::marker::{self, PhantomData}; + pub use self::core::num::Wrapping; + pub use self::core::ops::Range; + pub use self::core::option::{self, Option}; + pub use self::core::result::{self, Result}; + + #[cfg(all(feature = "alloc", not(feature = "std")))] + pub use alloc::borrow::{Cow, ToOwned}; + #[cfg(feature = "std")] + pub use std::borrow::{Cow, ToOwned}; + + #[cfg(all(feature = "alloc", not(feature = "std")))] + pub use alloc::string::{String, ToString}; + #[cfg(feature = "std")] + pub use std::string::{String, ToString}; + + #[cfg(all(feature = "alloc", not(feature = "std")))] + pub use alloc::vec::Vec; + #[cfg(feature = "std")] + pub use std::vec::Vec; + + #[cfg(all(feature = "alloc", not(feature = "std")))] + pub use alloc::boxed::Box; + #[cfg(feature = "std")] + pub use std::boxed::Box; + + #[cfg(all(feature = "rc", feature = "alloc", not(feature = "std")))] + pub use alloc::rc::{Rc, Weak as RcWeak}; + #[cfg(all(feature = "rc", feature = "std"))] + pub use std::rc::{Rc, Weak as RcWeak}; + + #[cfg(all(feature = "rc", feature = "alloc", not(feature = "std")))] + pub use alloc::sync::{Arc, Weak as ArcWeak}; + #[cfg(all(feature = "rc", feature = "std"))] + pub use std::sync::{Arc, Weak as ArcWeak}; + + #[cfg(all(feature = "alloc", not(feature = "std")))] + pub use alloc::collections::{BTreeMap, BTreeSet, BinaryHeap, LinkedList, VecDeque}; + #[cfg(feature = "std")] + pub use std::collections::{BTreeMap, BTreeSet, BinaryHeap, LinkedList, VecDeque}; + + #[cfg(all(not(no_core_cstr), not(feature = "std")))] + pub use core::ffi::CStr; + #[cfg(feature = "std")] + pub use std::ffi::CStr; + + #[cfg(all(not(no_core_cstr), feature = "alloc", not(feature = "std")))] + pub use alloc::ffi::CString; + #[cfg(feature = "std")] + pub use std::ffi::CString; + + #[cfg(feature = "std")] + pub use std::{error, net}; + + #[cfg(feature = "std")] + pub use std::collections::{HashMap, HashSet}; + #[cfg(feature = "std")] + pub use std::ffi::{OsStr, OsString}; + #[cfg(feature = "std")] + pub use std::hash::{BuildHasher, Hash}; + #[cfg(feature = "std")] + pub use std::io::Write; + #[cfg(feature = "std")] + pub use std::path::{Path, PathBuf}; + #[cfg(feature = "std")] + pub use std::sync::{Mutex, RwLock}; + #[cfg(feature = "std")] + pub use std::time::{SystemTime, UNIX_EPOCH}; + + #[cfg(all(feature = "std", not(no_collections_bound), no_ops_bound))] + pub use std::collections::Bound; + + #[cfg(not(no_core_reverse))] + pub use self::core::cmp::Reverse; + + #[cfg(not(no_ops_bound))] + pub use self::core::ops::Bound; + + #[cfg(not(no_range_inclusive))] + pub use self::core::ops::RangeInclusive; + + #[cfg(all(feature = "std", no_target_has_atomic, not(no_std_atomic)))] + pub use std::sync::atomic::{ + AtomicBool, AtomicI16, AtomicI32, AtomicI8, AtomicIsize, AtomicU16, AtomicU32, AtomicU8, + AtomicUsize, Ordering, + }; + #[cfg(all(feature = "std", no_target_has_atomic, not(no_std_atomic64)))] + pub use std::sync::atomic::{AtomicI64, AtomicU64}; + + #[cfg(all(feature = "std", not(no_target_has_atomic)))] + pub use std::sync::atomic::Ordering; + #[cfg(all(feature = "std", not(no_target_has_atomic), target_has_atomic = "8"))] + pub use std::sync::atomic::{AtomicBool, AtomicI8, AtomicU8}; + #[cfg(all(feature = "std", not(no_target_has_atomic), target_has_atomic = "16"))] + pub use std::sync::atomic::{AtomicI16, AtomicU16}; + #[cfg(all(feature = "std", not(no_target_has_atomic), target_has_atomic = "32"))] + pub use std::sync::atomic::{AtomicI32, AtomicU32}; + #[cfg(all(feature = "std", not(no_target_has_atomic), target_has_atomic = "64"))] + pub use std::sync::atomic::{AtomicI64, AtomicU64}; + #[cfg(all(feature = "std", not(no_target_has_atomic), target_has_atomic = "ptr"))] + pub use std::sync::atomic::{AtomicIsize, AtomicUsize}; + + #[cfg(any(feature = "std", not(no_core_duration)))] + pub use self::core::time::Duration; +} + +// None of this crate's error handling needs the `From::from` error conversion +// performed implicitly by the `?` operator or the standard library's `try!` +// macro. This simplified macro gives a 5.5% improvement in compile time +// compared to standard `try!`, and 9% improvement compared to `?`. +macro_rules! try { + ($expr:expr) => { + match $expr { + Ok(val) => val, + Err(err) => return Err(err), + } + }; +} + +//////////////////////////////////////////////////////////////////////////////// + +#[macro_use] +mod macros; + +#[macro_use] +mod integer128; + +pub mod de; +pub mod ser; + +#[doc(inline)] +pub use de::{Deserialize, Deserializer}; +#[doc(inline)] +pub use ser::{Serialize, Serializer}; + +// Used by generated code and doc tests. Not public API. +#[doc(hidden)] +#[path = "private/mod.rs"] +pub mod __private; + +#[allow(unused_imports)] +use self::__private as export; +#[allow(unused_imports)] +use self::__private as private; + +#[path = "de/seed.rs"] +mod seed; + +#[cfg(not(any(feature = "std", feature = "unstable")))] +mod std_error; + +// Re-export #[derive(Serialize, Deserialize)]. +// +// The reason re-exporting is not enabled by default is that disabling it would +// be annoying for crates that provide handwritten impls or data formats. They +// would need to disable default features and then explicitly re-enable std. +#[cfg(feature = "serde_derive")] +#[allow(unused_imports)] +#[macro_use] +extern crate serde_derive; + +/// Derive macro available if serde is built with `features = ["derive"]`. +#[cfg(feature = "serde_derive")] +pub use serde_derive::{Deserialize, Serialize}; + +#[cfg(all(not(no_serde_derive), any(feature = "std", feature = "alloc")))] +mod actually_private { + pub struct T; +} diff --git a/rust/serde/macros.rs b/rust/serde/macros.rs new file mode 100644 index 00000000000000..0b0a9ad823e280 --- /dev/null +++ b/rust/serde/macros.rs @@ -0,0 +1,238 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Super explicit first paragraph because this shows up at the top level and +// trips up people who are just looking for basic Serialize / Deserialize +// documentation. +// +/// Helper macro when implementing the `Deserializer` part of a new data format +/// for Serde. +/// +/// Some [`Deserializer`] implementations for self-describing formats do not +/// care what hint the [`Visitor`] gives them, they just want to blindly call +/// the [`Visitor`] method corresponding to the data they can tell is in the +/// input. This requires repetitive implementations of all the [`Deserializer`] +/// trait methods. +/// +/// ```edition2018 +/// # use serde::forward_to_deserialize_any; +/// # use serde::de::{value, Deserializer, Visitor}; +/// # +/// # struct MyDeserializer; +/// # +/// # impl<'de> Deserializer<'de> for MyDeserializer { +/// # type Error = value::Error; +/// # +/// # fn deserialize_any(self, _: V) -> Result +/// # where +/// # V: Visitor<'de>, +/// # { +/// # unimplemented!() +/// # } +/// # +/// #[inline] +/// fn deserialize_bool(self, visitor: V) -> Result +/// where +/// V: Visitor<'de>, +/// { +/// self.deserialize_any(visitor) +/// } +/// # +/// # forward_to_deserialize_any! { +/// # i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string +/// # bytes byte_buf option unit unit_struct newtype_struct seq tuple +/// # tuple_struct map struct enum identifier ignored_any +/// # } +/// # } +/// ``` +/// +/// The `forward_to_deserialize_any!` macro implements these simple forwarding +/// methods so that they forward directly to [`Deserializer::deserialize_any`]. +/// You can choose which methods to forward. +/// +/// ```edition2018 +/// # use serde::forward_to_deserialize_any; +/// # use serde::de::{value, Deserializer, Visitor}; +/// # +/// # struct MyDeserializer; +/// # +/// impl<'de> Deserializer<'de> for MyDeserializer { +/// # type Error = value::Error; +/// # +/// fn deserialize_any(self, visitor: V) -> Result +/// where +/// V: Visitor<'de>, +/// { +/// /* ... */ +/// # let _ = visitor; +/// # unimplemented!() +/// } +/// +/// forward_to_deserialize_any! { +/// bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string +/// bytes byte_buf option unit unit_struct newtype_struct seq tuple +/// tuple_struct map struct enum identifier ignored_any +/// } +/// } +/// ``` +/// +/// The macro assumes the convention that your `Deserializer` lifetime parameter +/// is called `'de` and that the `Visitor` type parameters on each method are +/// called `V`. A different type parameter and a different lifetime can be +/// specified explicitly if necessary. +/// +/// ```edition2018 +/// # use std::marker::PhantomData; +/// # +/// # use serde::forward_to_deserialize_any; +/// # use serde::de::{value, Deserializer, Visitor}; +/// # +/// # struct MyDeserializer(PhantomData); +/// # +/// # impl<'q, V> Deserializer<'q> for MyDeserializer { +/// # type Error = value::Error; +/// # +/// # fn deserialize_any(self, visitor: W) -> Result +/// # where +/// # W: Visitor<'q>, +/// # { +/// # unimplemented!() +/// # } +/// # +/// forward_to_deserialize_any! { +/// > +/// bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string +/// bytes byte_buf option unit unit_struct newtype_struct seq tuple +/// tuple_struct map struct enum identifier ignored_any +/// } +/// # } +/// ``` +/// +/// [`Deserializer`]: trait.Deserializer.html +/// [`Visitor`]: de/trait.Visitor.html +/// [`Deserializer::deserialize_any`]: trait.Deserializer.html#tymethod.deserialize_any +#[macro_export(local_inner_macros)] +macro_rules! forward_to_deserialize_any { + (<$visitor:ident: Visitor<$lifetime:tt>> $($func:ident)*) => { + $(forward_to_deserialize_any_helper!{$func<$lifetime, $visitor>})* + }; + // This case must be after the previous one. + ($($func:ident)*) => { + $(forward_to_deserialize_any_helper!{$func<'de, V>})* + }; +} + +#[doc(hidden)] +#[macro_export] +macro_rules! forward_to_deserialize_any_method { + ($func:ident<$l:tt, $v:ident>($($arg:ident : $ty:ty),*)) => { + #[inline] + fn $func<$v>(self, $($arg: $ty,)* visitor: $v) -> $crate::__private::Result<$v::Value, Self::Error> + where + $v: $crate::de::Visitor<$l>, + { + $( + let _ = $arg; + )* + self.deserialize_any(visitor) + } + }; +} + +#[doc(hidden)] +#[macro_export(local_inner_macros)] +macro_rules! forward_to_deserialize_any_helper { + (bool<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_bool<$l, $v>()} + }; + (i8<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_i8<$l, $v>()} + }; + (i16<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_i16<$l, $v>()} + }; + (i32<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_i32<$l, $v>()} + }; + (i64<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_i64<$l, $v>()} + }; + (i128<$l:tt, $v:ident>) => { + serde_if_integer128! { + forward_to_deserialize_any_method!{deserialize_i128<$l, $v>()} + } + }; + (u8<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_u8<$l, $v>()} + }; + (u16<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_u16<$l, $v>()} + }; + (u32<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_u32<$l, $v>()} + }; + (u64<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_u64<$l, $v>()} + }; + (u128<$l:tt, $v:ident>) => { + serde_if_integer128! { + forward_to_deserialize_any_method!{deserialize_u128<$l, $v>()} + } + }; + (f32<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_f32<$l, $v>()} + }; + (f64<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_f64<$l, $v>()} + }; + (char<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_char<$l, $v>()} + }; + (str<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_str<$l, $v>()} + }; + (string<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_string<$l, $v>()} + }; + (bytes<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_bytes<$l, $v>()} + }; + (byte_buf<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_byte_buf<$l, $v>()} + }; + (option<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_option<$l, $v>()} + }; + (unit<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_unit<$l, $v>()} + }; + (unit_struct<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_unit_struct<$l, $v>(name: &'static str)} + }; + (newtype_struct<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_newtype_struct<$l, $v>(name: &'static str)} + }; + (seq<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_seq<$l, $v>()} + }; + (tuple<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_tuple<$l, $v>(len: usize)} + }; + (tuple_struct<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_tuple_struct<$l, $v>(name: &'static str, len: usize)} + }; + (map<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_map<$l, $v>()} + }; + (struct<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_struct<$l, $v>(name: &'static str, fields: &'static [&'static str])} + }; + (enum<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_enum<$l, $v>(name: &'static str, variants: &'static [&'static str])} + }; + (identifier<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_identifier<$l, $v>()} + }; + (ignored_any<$l:tt, $v:ident>) => { + forward_to_deserialize_any_method!{deserialize_ignored_any<$l, $v>()} + }; +} diff --git a/rust/serde/private/de.rs b/rust/serde/private/de.rs new file mode 100644 index 00000000000000..707c053b7fce71 --- /dev/null +++ b/rust/serde/private/de.rs @@ -0,0 +1,2997 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +use de::value::{BorrowedBytesDeserializer, BytesDeserializer}; +use de::{Deserialize, Deserializer, Error, IntoDeserializer, Visitor}; + +#[cfg(any(feature = "std", feature = "alloc"))] +use de::{DeserializeSeed, MapAccess, Unexpected}; + +#[cfg(any(feature = "std", feature = "alloc"))] +pub use self::content::{ + Content, ContentDeserializer, ContentRefDeserializer, EnumDeserializer, + InternallyTaggedUnitVisitor, TagContentOtherField, TagContentOtherFieldVisitor, + TagOrContentField, TagOrContentFieldVisitor, TaggedContentVisitor, UntaggedUnitVisitor, +}; + +pub use seed::InPlaceSeed; + +/// If the missing field is of type `Option` then treat is as `None`, +/// otherwise it is an error. +pub fn missing_field<'de, V, E>(field: &'static str) -> Result +where + V: Deserialize<'de>, + E: Error, +{ + struct MissingFieldDeserializer(&'static str, PhantomData); + + impl<'de, E> Deserializer<'de> for MissingFieldDeserializer + where + E: Error, + { + type Error = E; + + fn deserialize_any(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + Err(Error::missing_field(self.0)) + } + + fn deserialize_option(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_none() + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } + } + + let deserializer = MissingFieldDeserializer(field, PhantomData); + Deserialize::deserialize(deserializer) +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub fn borrow_cow_str<'de: 'a, 'a, D, R>(deserializer: D) -> Result +where + D: Deserializer<'de>, + R: From>, +{ + struct CowStrVisitor; + + impl<'a> Visitor<'a> for CowStrVisitor { + type Value = Cow<'a, str>; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a string") + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + Ok(Cow::Owned(v.to_owned())) + } + + fn visit_borrowed_str(self, v: &'a str) -> Result + where + E: Error, + { + Ok(Cow::Borrowed(v)) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + Ok(Cow::Owned(v)) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + match str::from_utf8(v) { + Ok(s) => Ok(Cow::Owned(s.to_owned())), + Err(_) => Err(Error::invalid_value(Unexpected::Bytes(v), &self)), + } + } + + fn visit_borrowed_bytes(self, v: &'a [u8]) -> Result + where + E: Error, + { + match str::from_utf8(v) { + Ok(s) => Ok(Cow::Borrowed(s)), + Err(_) => Err(Error::invalid_value(Unexpected::Bytes(v), &self)), + } + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + match String::from_utf8(v) { + Ok(s) => Ok(Cow::Owned(s)), + Err(e) => Err(Error::invalid_value( + Unexpected::Bytes(&e.into_bytes()), + &self, + )), + } + } + } + + deserializer.deserialize_str(CowStrVisitor).map(From::from) +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub fn borrow_cow_bytes<'de: 'a, 'a, D, R>(deserializer: D) -> Result +where + D: Deserializer<'de>, + R: From>, +{ + struct CowBytesVisitor; + + impl<'a> Visitor<'a> for CowBytesVisitor { + type Value = Cow<'a, [u8]>; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a byte array") + } + + fn visit_str(self, v: &str) -> Result + where + E: Error, + { + Ok(Cow::Owned(v.as_bytes().to_vec())) + } + + fn visit_borrowed_str(self, v: &'a str) -> Result + where + E: Error, + { + Ok(Cow::Borrowed(v.as_bytes())) + } + + fn visit_string(self, v: String) -> Result + where + E: Error, + { + Ok(Cow::Owned(v.into_bytes())) + } + + fn visit_bytes(self, v: &[u8]) -> Result + where + E: Error, + { + Ok(Cow::Owned(v.to_vec())) + } + + fn visit_borrowed_bytes(self, v: &'a [u8]) -> Result + where + E: Error, + { + Ok(Cow::Borrowed(v)) + } + + fn visit_byte_buf(self, v: Vec) -> Result + where + E: Error, + { + Ok(Cow::Owned(v)) + } + } + + deserializer + .deserialize_bytes(CowBytesVisitor) + .map(From::from) +} + +#[cfg(any(feature = "std", feature = "alloc"))] +mod content { + // This module is private and nothing here should be used outside of + // generated code. + // + // We will iterate on the implementation for a few releases and only have to + // worry about backward compatibility for the `untagged` and `tag` attributes + // rather than for this entire mechanism. + // + // This issue is tracking making some of this stuff public: + // https://github.com/serde-rs/serde/issues/741 + + use lib::*; + + use __private::size_hint; + use actually_private; + use de::{ + self, Deserialize, DeserializeSeed, Deserializer, EnumAccess, Expected, IgnoredAny, + MapAccess, SeqAccess, Unexpected, Visitor, + }; + + /// Used from generated code to buffer the contents of the Deserializer when + /// deserializing untagged enums and internally tagged enums. + /// + /// Not public API. Use serde-value instead. + #[derive(Debug, Clone)] + pub enum Content<'de> { + Bool(bool), + + U8(u8), + U16(u16), + U32(u32), + U64(u64), + + I8(i8), + I16(i16), + I32(i32), + I64(i64), + + F32(f32), + F64(f64), + + Char(char), + String(String), + Str(&'de str), + ByteBuf(Vec), + Bytes(&'de [u8]), + + None, + Some(Box>), + + Unit, + Newtype(Box>), + Seq(Vec>), + Map(Vec<(Content<'de>, Content<'de>)>), + } + + impl<'de> Content<'de> { + pub fn as_str(&self) -> Option<&str> { + match *self { + Content::Str(x) => Some(x), + Content::String(ref x) => Some(x), + Content::Bytes(x) => str::from_utf8(x).ok(), + Content::ByteBuf(ref x) => str::from_utf8(x).ok(), + _ => None, + } + } + + #[cold] + fn unexpected(&self) -> Unexpected { + match *self { + Content::Bool(b) => Unexpected::Bool(b), + Content::U8(n) => Unexpected::Unsigned(n as u64), + Content::U16(n) => Unexpected::Unsigned(n as u64), + Content::U32(n) => Unexpected::Unsigned(n as u64), + Content::U64(n) => Unexpected::Unsigned(n), + Content::I8(n) => Unexpected::Signed(n as i64), + Content::I16(n) => Unexpected::Signed(n as i64), + Content::I32(n) => Unexpected::Signed(n as i64), + Content::I64(n) => Unexpected::Signed(n), + Content::F32(f) => Unexpected::Float(f as f64), + Content::F64(f) => Unexpected::Float(f), + Content::Char(c) => Unexpected::Char(c), + Content::String(ref s) => Unexpected::Str(s), + Content::Str(s) => Unexpected::Str(s), + Content::ByteBuf(ref b) => Unexpected::Bytes(b), + Content::Bytes(b) => Unexpected::Bytes(b), + Content::None | Content::Some(_) => Unexpected::Option, + Content::Unit => Unexpected::Unit, + Content::Newtype(_) => Unexpected::NewtypeStruct, + Content::Seq(_) => Unexpected::Seq, + Content::Map(_) => Unexpected::Map, + } + } + } + + impl<'de> Deserialize<'de> for Content<'de> { + fn deserialize(deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // Untagged and internally tagged enums are only supported in + // self-describing formats. + let visitor = ContentVisitor { value: PhantomData }; + deserializer.__deserialize_content(actually_private::T, visitor) + } + } + + struct ContentVisitor<'de> { + value: PhantomData>, + } + + impl<'de> ContentVisitor<'de> { + fn new() -> Self { + ContentVisitor { value: PhantomData } + } + } + + impl<'de> Visitor<'de> for ContentVisitor<'de> { + type Value = Content<'de>; + + fn expecting(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + fmt.write_str("any value") + } + + fn visit_bool(self, value: bool) -> Result + where + F: de::Error, + { + Ok(Content::Bool(value)) + } + + fn visit_i8(self, value: i8) -> Result + where + F: de::Error, + { + Ok(Content::I8(value)) + } + + fn visit_i16(self, value: i16) -> Result + where + F: de::Error, + { + Ok(Content::I16(value)) + } + + fn visit_i32(self, value: i32) -> Result + where + F: de::Error, + { + Ok(Content::I32(value)) + } + + fn visit_i64(self, value: i64) -> Result + where + F: de::Error, + { + Ok(Content::I64(value)) + } + + fn visit_u8(self, value: u8) -> Result + where + F: de::Error, + { + Ok(Content::U8(value)) + } + + fn visit_u16(self, value: u16) -> Result + where + F: de::Error, + { + Ok(Content::U16(value)) + } + + fn visit_u32(self, value: u32) -> Result + where + F: de::Error, + { + Ok(Content::U32(value)) + } + + fn visit_u64(self, value: u64) -> Result + where + F: de::Error, + { + Ok(Content::U64(value)) + } + + fn visit_f32(self, value: f32) -> Result + where + F: de::Error, + { + Ok(Content::F32(value)) + } + + fn visit_f64(self, value: f64) -> Result + where + F: de::Error, + { + Ok(Content::F64(value)) + } + + fn visit_char(self, value: char) -> Result + where + F: de::Error, + { + Ok(Content::Char(value)) + } + + fn visit_str(self, value: &str) -> Result + where + F: de::Error, + { + Ok(Content::String(value.into())) + } + + fn visit_borrowed_str(self, value: &'de str) -> Result + where + F: de::Error, + { + Ok(Content::Str(value)) + } + + fn visit_string(self, value: String) -> Result + where + F: de::Error, + { + Ok(Content::String(value)) + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + F: de::Error, + { + Ok(Content::ByteBuf(value.into())) + } + + fn visit_borrowed_bytes(self, value: &'de [u8]) -> Result + where + F: de::Error, + { + Ok(Content::Bytes(value)) + } + + fn visit_byte_buf(self, value: Vec) -> Result + where + F: de::Error, + { + Ok(Content::ByteBuf(value)) + } + + fn visit_unit(self) -> Result + where + F: de::Error, + { + Ok(Content::Unit) + } + + fn visit_none(self) -> Result + where + F: de::Error, + { + Ok(Content::None) + } + + fn visit_some(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Deserialize::deserialize(deserializer).map(|v| Content::Some(Box::new(v))) + } + + fn visit_newtype_struct(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + Deserialize::deserialize(deserializer).map(|v| Content::Newtype(Box::new(v))) + } + + fn visit_seq(self, mut visitor: V) -> Result + where + V: SeqAccess<'de>, + { + let mut vec = Vec::with_capacity(size_hint::cautious(visitor.size_hint())); + while let Some(e) = try!(visitor.next_element()) { + vec.push(e); + } + Ok(Content::Seq(vec)) + } + + fn visit_map(self, mut visitor: V) -> Result + where + V: MapAccess<'de>, + { + let mut vec = Vec::with_capacity(size_hint::cautious(visitor.size_hint())); + while let Some(kv) = try!(visitor.next_entry()) { + vec.push(kv); + } + Ok(Content::Map(vec)) + } + + fn visit_enum(self, _visitor: V) -> Result + where + V: EnumAccess<'de>, + { + Err(de::Error::custom( + "untagged and internally tagged enums do not support enum input", + )) + } + } + + /// This is the type of the map keys in an internally tagged enum. + /// + /// Not public API. + pub enum TagOrContent<'de> { + Tag, + Content(Content<'de>), + } + + struct TagOrContentVisitor<'de> { + name: &'static str, + value: PhantomData>, + } + + impl<'de> TagOrContentVisitor<'de> { + fn new(name: &'static str) -> Self { + TagOrContentVisitor { + name: name, + value: PhantomData, + } + } + } + + impl<'de> DeserializeSeed<'de> for TagOrContentVisitor<'de> { + type Value = TagOrContent<'de>; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // Internally tagged enums are only supported in self-describing + // formats. + deserializer.deserialize_any(self) + } + } + + impl<'de> Visitor<'de> for TagOrContentVisitor<'de> { + type Value = TagOrContent<'de>; + + fn expecting(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + write!(fmt, "a type tag `{}` or any other value", self.name) + } + + fn visit_bool(self, value: bool) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_bool(value) + .map(TagOrContent::Content) + } + + fn visit_i8(self, value: i8) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_i8(value) + .map(TagOrContent::Content) + } + + fn visit_i16(self, value: i16) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_i16(value) + .map(TagOrContent::Content) + } + + fn visit_i32(self, value: i32) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_i32(value) + .map(TagOrContent::Content) + } + + fn visit_i64(self, value: i64) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_i64(value) + .map(TagOrContent::Content) + } + + fn visit_u8(self, value: u8) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_u8(value) + .map(TagOrContent::Content) + } + + fn visit_u16(self, value: u16) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_u16(value) + .map(TagOrContent::Content) + } + + fn visit_u32(self, value: u32) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_u32(value) + .map(TagOrContent::Content) + } + + fn visit_u64(self, value: u64) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_u64(value) + .map(TagOrContent::Content) + } + + fn visit_f32(self, value: f32) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_f32(value) + .map(TagOrContent::Content) + } + + fn visit_f64(self, value: f64) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_f64(value) + .map(TagOrContent::Content) + } + + fn visit_char(self, value: char) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_char(value) + .map(TagOrContent::Content) + } + + fn visit_str(self, value: &str) -> Result + where + F: de::Error, + { + if value == self.name { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_str(value) + .map(TagOrContent::Content) + } + } + + fn visit_borrowed_str(self, value: &'de str) -> Result + where + F: de::Error, + { + if value == self.name { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_borrowed_str(value) + .map(TagOrContent::Content) + } + } + + fn visit_string(self, value: String) -> Result + where + F: de::Error, + { + if value == self.name { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_string(value) + .map(TagOrContent::Content) + } + } + + fn visit_bytes(self, value: &[u8]) -> Result + where + F: de::Error, + { + if value == self.name.as_bytes() { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_bytes(value) + .map(TagOrContent::Content) + } + } + + fn visit_borrowed_bytes(self, value: &'de [u8]) -> Result + where + F: de::Error, + { + if value == self.name.as_bytes() { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_borrowed_bytes(value) + .map(TagOrContent::Content) + } + } + + fn visit_byte_buf(self, value: Vec) -> Result + where + F: de::Error, + { + if value == self.name.as_bytes() { + Ok(TagOrContent::Tag) + } else { + ContentVisitor::new() + .visit_byte_buf(value) + .map(TagOrContent::Content) + } + } + + fn visit_unit(self) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_unit() + .map(TagOrContent::Content) + } + + fn visit_none(self) -> Result + where + F: de::Error, + { + ContentVisitor::new() + .visit_none() + .map(TagOrContent::Content) + } + + fn visit_some(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + ContentVisitor::new() + .visit_some(deserializer) + .map(TagOrContent::Content) + } + + fn visit_newtype_struct(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + ContentVisitor::new() + .visit_newtype_struct(deserializer) + .map(TagOrContent::Content) + } + + fn visit_seq(self, visitor: V) -> Result + where + V: SeqAccess<'de>, + { + ContentVisitor::new() + .visit_seq(visitor) + .map(TagOrContent::Content) + } + + fn visit_map(self, visitor: V) -> Result + where + V: MapAccess<'de>, + { + ContentVisitor::new() + .visit_map(visitor) + .map(TagOrContent::Content) + } + + fn visit_enum(self, visitor: V) -> Result + where + V: EnumAccess<'de>, + { + ContentVisitor::new() + .visit_enum(visitor) + .map(TagOrContent::Content) + } + } + + /// Used by generated code to deserialize an internally tagged enum. + /// + /// Not public API. + pub struct TaggedContent<'de, T> { + pub tag: T, + pub content: Content<'de>, + } + + /// Not public API. + pub struct TaggedContentVisitor<'de, T> { + tag_name: &'static str, + expecting: &'static str, + value: PhantomData>, + } + + impl<'de, T> TaggedContentVisitor<'de, T> { + /// Visitor for the content of an internally tagged enum with the given + /// tag name. + pub fn new(name: &'static str, expecting: &'static str) -> Self { + TaggedContentVisitor { + tag_name: name, + expecting: expecting, + value: PhantomData, + } + } + } + + impl<'de, T> DeserializeSeed<'de> for TaggedContentVisitor<'de, T> + where + T: Deserialize<'de>, + { + type Value = TaggedContent<'de, T>; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + // Internally tagged enums are only supported in self-describing + // formats. + deserializer.deserialize_any(self) + } + } + + impl<'de, T> Visitor<'de> for TaggedContentVisitor<'de, T> + where + T: Deserialize<'de>, + { + type Value = TaggedContent<'de, T>; + + fn expecting(&self, fmt: &mut fmt::Formatter) -> fmt::Result { + fmt.write_str(self.expecting) + } + + fn visit_seq(self, mut seq: S) -> Result + where + S: SeqAccess<'de>, + { + let tag = match try!(seq.next_element()) { + Some(tag) => tag, + None => { + return Err(de::Error::missing_field(self.tag_name)); + } + }; + let rest = de::value::SeqAccessDeserializer::new(seq); + Ok(TaggedContent { + tag: tag, + content: try!(Content::deserialize(rest)), + }) + } + + fn visit_map(self, mut map: M) -> Result + where + M: MapAccess<'de>, + { + let mut tag = None; + let mut vec = Vec::with_capacity(size_hint::cautious(map.size_hint())); + while let Some(k) = try!(map.next_key_seed(TagOrContentVisitor::new(self.tag_name))) { + match k { + TagOrContent::Tag => { + if tag.is_some() { + return Err(de::Error::duplicate_field(self.tag_name)); + } + tag = Some(try!(map.next_value())); + } + TagOrContent::Content(k) => { + let v = try!(map.next_value()); + vec.push((k, v)); + } + } + } + match tag { + None => Err(de::Error::missing_field(self.tag_name)), + Some(tag) => Ok(TaggedContent { + tag: tag, + content: Content::Map(vec), + }), + } + } + } + + /// Used by generated code to deserialize an adjacently tagged enum. + /// + /// Not public API. + pub enum TagOrContentField { + Tag, + Content, + } + + /// Not public API. + pub struct TagOrContentFieldVisitor { + pub tag: &'static str, + pub content: &'static str, + } + + impl<'de> DeserializeSeed<'de> for TagOrContentFieldVisitor { + type Value = TagOrContentField; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(self) + } + } + + impl<'de> Visitor<'de> for TagOrContentFieldVisitor { + type Value = TagOrContentField; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!(formatter, "{:?} or {:?}", self.tag, self.content) + } + + fn visit_str(self, field: &str) -> Result + where + E: de::Error, + { + if field == self.tag { + Ok(TagOrContentField::Tag) + } else if field == self.content { + Ok(TagOrContentField::Content) + } else { + Err(de::Error::invalid_value(Unexpected::Str(field), &self)) + } + } + } + + /// Used by generated code to deserialize an adjacently tagged enum when + /// ignoring unrelated fields is allowed. + /// + /// Not public API. + pub enum TagContentOtherField { + Tag, + Content, + Other, + } + + /// Not public API. + pub struct TagContentOtherFieldVisitor { + pub tag: &'static str, + pub content: &'static str, + } + + impl<'de> DeserializeSeed<'de> for TagContentOtherFieldVisitor { + type Value = TagContentOtherField; + + fn deserialize(self, deserializer: D) -> Result + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(self) + } + } + + impl<'de> Visitor<'de> for TagContentOtherFieldVisitor { + type Value = TagContentOtherField; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!( + formatter, + "{:?}, {:?}, or other ignored fields", + self.tag, self.content + ) + } + + fn visit_str(self, field: &str) -> Result + where + E: de::Error, + { + if field == self.tag { + Ok(TagContentOtherField::Tag) + } else if field == self.content { + Ok(TagContentOtherField::Content) + } else { + Ok(TagContentOtherField::Other) + } + } + } + + /// Not public API + pub struct ContentDeserializer<'de, E> { + content: Content<'de>, + err: PhantomData, + } + + impl<'de, E> ContentDeserializer<'de, E> + where + E: de::Error, + { + #[cold] + fn invalid_type(self, exp: &Expected) -> E { + de::Error::invalid_type(self.content.unexpected(), exp) + } + + fn deserialize_integer(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_float(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::F32(v) => visitor.visit_f32(v), + Content::F64(v) => visitor.visit_f64(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + } + + fn visit_content_seq<'de, V, E>(content: Vec>, visitor: V) -> Result + where + V: Visitor<'de>, + E: de::Error, + { + let seq = content.into_iter().map(ContentDeserializer::new); + let mut seq_visitor = de::value::SeqDeserializer::new(seq); + let value = try!(visitor.visit_seq(&mut seq_visitor)); + try!(seq_visitor.end()); + Ok(value) + } + + fn visit_content_map<'de, V, E>( + content: Vec<(Content<'de>, Content<'de>)>, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + E: de::Error, + { + let map = content + .into_iter() + .map(|(k, v)| (ContentDeserializer::new(k), ContentDeserializer::new(v))); + let mut map_visitor = de::value::MapDeserializer::new(map); + let value = try!(visitor.visit_map(&mut map_visitor)); + try!(map_visitor.end()); + Ok(value) + } + + /// Used when deserializing an internally tagged enum because the content + /// will be used exactly once. + impl<'de, E> Deserializer<'de> for ContentDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Bool(v) => visitor.visit_bool(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + Content::F32(v) => visitor.visit_f32(v), + Content::F64(v) => visitor.visit_f64(v), + Content::Char(v) => visitor.visit_char(v), + Content::String(v) => visitor.visit_string(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(v) => visitor.visit_byte_buf(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::Unit => visitor.visit_unit(), + Content::None => visitor.visit_none(), + Content::Some(v) => visitor.visit_some(ContentDeserializer::new(*v)), + Content::Newtype(v) => visitor.visit_newtype_struct(ContentDeserializer::new(*v)), + Content::Seq(v) => visit_content_seq(v, visitor), + Content::Map(v) => visit_content_map(v, visitor), + } + } + + fn deserialize_bool(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Bool(v) => visitor.visit_bool(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_i8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_f32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_float(visitor) + } + + fn deserialize_f64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_float(visitor) + } + + fn deserialize_char(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Char(v) => visitor.visit_char(v), + Content::String(v) => visitor.visit_string(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_str(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_string(visitor) + } + + fn deserialize_string(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::String(v) => visitor.visit_string(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(v) => visitor.visit_byte_buf(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_bytes(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_byte_buf(visitor) + } + + fn deserialize_byte_buf(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::String(v) => visitor.visit_string(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(v) => visitor.visit_byte_buf(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::Seq(v) => visit_content_seq(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_option(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::None => visitor.visit_none(), + Content::Some(v) => visitor.visit_some(ContentDeserializer::new(*v)), + Content::Unit => visitor.visit_unit(), + _ => visitor.visit_some(self), + } + } + + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Unit => visitor.visit_unit(), + + // Allow deserializing newtype variant containing unit. + // + // #[derive(Deserialize)] + // #[serde(tag = "result")] + // enum Response { + // Success(T), + // } + // + // We want {"result":"Success"} to deserialize into Response<()>. + Content::Map(ref v) if v.is_empty() => visitor.visit_unit(), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_unit_struct( + self, + _name: &'static str, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + match self.content { + // As a special case, allow deserializing untagged newtype + // variant containing unit struct. + // + // #[derive(Deserialize)] + // struct Info; + // + // #[derive(Deserialize)] + // #[serde(tag = "topic")] + // enum Message { + // Info(Info), + // } + // + // We want {"topic":"Info"} to deserialize even though + // ordinarily unit structs do not deserialize from empty map/seq. + Content::Map(ref v) if v.is_empty() => visitor.visit_unit(), + Content::Seq(ref v) if v.is_empty() => visitor.visit_unit(), + _ => self.deserialize_any(visitor), + } + } + + fn deserialize_newtype_struct( + self, + _name: &str, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Newtype(v) => visitor.visit_newtype_struct(ContentDeserializer::new(*v)), + _ => visitor.visit_newtype_struct(self), + } + } + + fn deserialize_seq(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Seq(v) => visit_content_seq(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_tuple(self, _len: usize, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Map(v) => visit_content_map(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_struct( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::Seq(v) => visit_content_seq(v, visitor), + Content::Map(v) => visit_content_map(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_enum( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + let (variant, value) = match self.content { + Content::Map(value) => { + let mut iter = value.into_iter(); + let (variant, value) = match iter.next() { + Some(v) => v, + None => { + return Err(de::Error::invalid_value( + de::Unexpected::Map, + &"map with a single key", + )); + } + }; + // enums are encoded in json as maps with a single key:value pair + if iter.next().is_some() { + return Err(de::Error::invalid_value( + de::Unexpected::Map, + &"map with a single key", + )); + } + (variant, Some(value)) + } + s @ Content::String(_) | s @ Content::Str(_) => (s, None), + other => { + return Err(de::Error::invalid_type( + other.unexpected(), + &"string or map", + )); + } + }; + + visitor.visit_enum(EnumDeserializer::new(variant, value)) + } + + fn deserialize_identifier(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.content { + Content::String(v) => visitor.visit_string(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(v) => visitor.visit_byte_buf(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U64(v) => visitor.visit_u64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_ignored_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + drop(self); + visitor.visit_unit() + } + + fn __deserialize_content( + self, + _: actually_private::T, + visitor: V, + ) -> Result, Self::Error> + where + V: Visitor<'de, Value = Content<'de>>, + { + let _ = visitor; + Ok(self.content) + } + } + + impl<'de, E> ContentDeserializer<'de, E> { + /// private API, don't use + pub fn new(content: Content<'de>) -> Self { + ContentDeserializer { + content: content, + err: PhantomData, + } + } + } + + pub struct EnumDeserializer<'de, E> + where + E: de::Error, + { + variant: Content<'de>, + value: Option>, + err: PhantomData, + } + + impl<'de, E> EnumDeserializer<'de, E> + where + E: de::Error, + { + pub fn new(variant: Content<'de>, value: Option>) -> EnumDeserializer<'de, E> { + EnumDeserializer { + variant: variant, + value: value, + err: PhantomData, + } + } + } + + impl<'de, E> de::EnumAccess<'de> for EnumDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + type Variant = VariantDeserializer<'de, Self::Error>; + + fn variant_seed(self, seed: V) -> Result<(V::Value, Self::Variant), E> + where + V: de::DeserializeSeed<'de>, + { + let visitor = VariantDeserializer { + value: self.value, + err: PhantomData, + }; + seed.deserialize(ContentDeserializer::new(self.variant)) + .map(|v| (v, visitor)) + } + } + + pub struct VariantDeserializer<'de, E> + where + E: de::Error, + { + value: Option>, + err: PhantomData, + } + + impl<'de, E> de::VariantAccess<'de> for VariantDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + fn unit_variant(self) -> Result<(), E> { + match self.value { + Some(value) => de::Deserialize::deserialize(ContentDeserializer::new(value)), + None => Ok(()), + } + } + + fn newtype_variant_seed(self, seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + match self.value { + Some(value) => seed.deserialize(ContentDeserializer::new(value)), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"newtype variant", + )), + } + } + + fn tuple_variant(self, _len: usize, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + match self.value { + Some(Content::Seq(v)) => { + de::Deserializer::deserialize_any(SeqDeserializer::new(v), visitor) + } + Some(other) => Err(de::Error::invalid_type( + other.unexpected(), + &"tuple variant", + )), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"tuple variant", + )), + } + } + + fn struct_variant( + self, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + match self.value { + Some(Content::Map(v)) => { + de::Deserializer::deserialize_any(MapDeserializer::new(v), visitor) + } + Some(Content::Seq(v)) => { + de::Deserializer::deserialize_any(SeqDeserializer::new(v), visitor) + } + Some(other) => Err(de::Error::invalid_type( + other.unexpected(), + &"struct variant", + )), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"struct variant", + )), + } + } + } + + struct SeqDeserializer<'de, E> + where + E: de::Error, + { + iter: > as IntoIterator>::IntoIter, + err: PhantomData, + } + + impl<'de, E> SeqDeserializer<'de, E> + where + E: de::Error, + { + fn new(vec: Vec>) -> Self { + SeqDeserializer { + iter: vec.into_iter(), + err: PhantomData, + } + } + } + + impl<'de, E> de::Deserializer<'de> for SeqDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + #[inline] + fn deserialize_any(mut self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let len = self.iter.len(); + if len == 0 { + visitor.visit_unit() + } else { + let ret = try!(visitor.visit_seq(&mut self)); + let remaining = self.iter.len(); + if remaining == 0 { + Ok(ret) + } else { + Err(de::Error::invalid_length(len, &"fewer elements in array")) + } + } + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } + } + + impl<'de, E> de::SeqAccess<'de> for SeqDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.iter.next() { + Some(value) => seed.deserialize(ContentDeserializer::new(value)).map(Some), + None => Ok(None), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } + } + + struct MapDeserializer<'de, E> + where + E: de::Error, + { + iter: , Content<'de>)> as IntoIterator>::IntoIter, + value: Option>, + err: PhantomData, + } + + impl<'de, E> MapDeserializer<'de, E> + where + E: de::Error, + { + fn new(map: Vec<(Content<'de>, Content<'de>)>) -> Self { + MapDeserializer { + iter: map.into_iter(), + value: None, + err: PhantomData, + } + } + } + + impl<'de, E> de::MapAccess<'de> for MapDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.iter.next() { + Some((key, value)) => { + self.value = Some(value); + seed.deserialize(ContentDeserializer::new(key)).map(Some) + } + None => Ok(None), + } + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + match self.value.take() { + Some(value) => seed.deserialize(ContentDeserializer::new(value)), + None => Err(de::Error::custom("value is missing")), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } + } + + impl<'de, E> de::Deserializer<'de> for MapDeserializer<'de, E> + where + E: de::Error, + { + type Error = E; + + #[inline] + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_map(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } + } + + /// Not public API. + pub struct ContentRefDeserializer<'a, 'de: 'a, E> { + content: &'a Content<'de>, + err: PhantomData, + } + + impl<'a, 'de, E> ContentRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + #[cold] + fn invalid_type(self, exp: &Expected) -> E { + de::Error::invalid_type(self.content.unexpected(), exp) + } + + fn deserialize_integer(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_float(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::F32(v) => visitor.visit_f32(v), + Content::F64(v) => visitor.visit_f64(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + } + + fn visit_content_seq_ref<'a, 'de, V, E>( + content: &'a [Content<'de>], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + E: de::Error, + { + let seq = content.iter().map(ContentRefDeserializer::new); + let mut seq_visitor = de::value::SeqDeserializer::new(seq); + let value = try!(visitor.visit_seq(&mut seq_visitor)); + try!(seq_visitor.end()); + Ok(value) + } + + fn visit_content_map_ref<'a, 'de, V, E>( + content: &'a [(Content<'de>, Content<'de>)], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + E: de::Error, + { + let map = content.iter().map(|(k, v)| { + ( + ContentRefDeserializer::new(k), + ContentRefDeserializer::new(v), + ) + }); + let mut map_visitor = de::value::MapDeserializer::new(map); + let value = try!(visitor.visit_map(&mut map_visitor)); + try!(map_visitor.end()); + Ok(value) + } + + /// Used when deserializing an untagged enum because the content may need + /// to be used more than once. + impl<'de, 'a, E> Deserializer<'de> for ContentRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Bool(v) => visitor.visit_bool(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U16(v) => visitor.visit_u16(v), + Content::U32(v) => visitor.visit_u32(v), + Content::U64(v) => visitor.visit_u64(v), + Content::I8(v) => visitor.visit_i8(v), + Content::I16(v) => visitor.visit_i16(v), + Content::I32(v) => visitor.visit_i32(v), + Content::I64(v) => visitor.visit_i64(v), + Content::F32(v) => visitor.visit_f32(v), + Content::F64(v) => visitor.visit_f64(v), + Content::Char(v) => visitor.visit_char(v), + Content::String(ref v) => visitor.visit_str(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(ref v) => visitor.visit_bytes(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::Unit => visitor.visit_unit(), + Content::None => visitor.visit_none(), + Content::Some(ref v) => visitor.visit_some(ContentRefDeserializer::new(v)), + Content::Newtype(ref v) => { + visitor.visit_newtype_struct(ContentRefDeserializer::new(v)) + } + Content::Seq(ref v) => visit_content_seq_ref(v, visitor), + Content::Map(ref v) => visit_content_map_ref(v, visitor), + } + } + + fn deserialize_bool(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Bool(v) => visitor.visit_bool(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_i8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_i64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_u64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_integer(visitor) + } + + fn deserialize_f32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_float(visitor) + } + + fn deserialize_f64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_float(visitor) + } + + fn deserialize_char(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Char(v) => visitor.visit_char(v), + Content::String(ref v) => visitor.visit_str(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_str(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::String(ref v) => visitor.visit_str(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(ref v) => visitor.visit_bytes(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_string(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_bytes(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::String(ref v) => visitor.visit_str(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(ref v) => visitor.visit_bytes(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::Seq(ref v) => visit_content_seq_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_byte_buf(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_bytes(visitor) + } + + fn deserialize_option(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::None => visitor.visit_none(), + Content::Some(ref v) => visitor.visit_some(ContentRefDeserializer::new(v)), + Content::Unit => visitor.visit_unit(), + _ => visitor.visit_some(self), + } + } + + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Unit => visitor.visit_unit(), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_unit_struct( + self, + _name: &'static str, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + fn deserialize_newtype_struct(self, _name: &str, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Newtype(ref v) => { + visitor.visit_newtype_struct(ContentRefDeserializer::new(v)) + } + _ => visitor.visit_newtype_struct(self), + } + } + + fn deserialize_seq(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Seq(ref v) => visit_content_seq_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_tuple(self, _len: usize, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Map(ref v) => visit_content_map_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_struct( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::Seq(ref v) => visit_content_seq_ref(v, visitor), + Content::Map(ref v) => visit_content_map_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_enum( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + let (variant, value) = match *self.content { + Content::Map(ref value) => { + let mut iter = value.iter(); + let (variant, value) = match iter.next() { + Some(v) => v, + None => { + return Err(de::Error::invalid_value( + de::Unexpected::Map, + &"map with a single key", + )); + } + }; + // enums are encoded in json as maps with a single key:value pair + if iter.next().is_some() { + return Err(de::Error::invalid_value( + de::Unexpected::Map, + &"map with a single key", + )); + } + (variant, Some(value)) + } + ref s @ Content::String(_) | ref s @ Content::Str(_) => (s, None), + ref other => { + return Err(de::Error::invalid_type( + other.unexpected(), + &"string or map", + )); + } + }; + + visitor.visit_enum(EnumRefDeserializer { + variant: variant, + value: value, + err: PhantomData, + }) + } + + fn deserialize_identifier(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match *self.content { + Content::String(ref v) => visitor.visit_str(v), + Content::Str(v) => visitor.visit_borrowed_str(v), + Content::ByteBuf(ref v) => visitor.visit_bytes(v), + Content::Bytes(v) => visitor.visit_borrowed_bytes(v), + Content::U8(v) => visitor.visit_u8(v), + Content::U64(v) => visitor.visit_u64(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_ignored_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_unit() + } + + fn __deserialize_content( + self, + _: actually_private::T, + visitor: V, + ) -> Result, Self::Error> + where + V: Visitor<'de, Value = Content<'de>>, + { + let _ = visitor; + Ok(self.content.clone()) + } + } + + impl<'a, 'de, E> ContentRefDeserializer<'a, 'de, E> { + /// private API, don't use + pub fn new(content: &'a Content<'de>) -> Self { + ContentRefDeserializer { + content: content, + err: PhantomData, + } + } + } + + struct EnumRefDeserializer<'a, 'de: 'a, E> + where + E: de::Error, + { + variant: &'a Content<'de>, + value: Option<&'a Content<'de>>, + err: PhantomData, + } + + impl<'de, 'a, E> de::EnumAccess<'de> for EnumRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + type Variant = VariantRefDeserializer<'a, 'de, Self::Error>; + + fn variant_seed(self, seed: V) -> Result<(V::Value, Self::Variant), Self::Error> + where + V: de::DeserializeSeed<'de>, + { + let visitor = VariantRefDeserializer { + value: self.value, + err: PhantomData, + }; + seed.deserialize(ContentRefDeserializer::new(self.variant)) + .map(|v| (v, visitor)) + } + } + + struct VariantRefDeserializer<'a, 'de: 'a, E> + where + E: de::Error, + { + value: Option<&'a Content<'de>>, + err: PhantomData, + } + + impl<'de, 'a, E> de::VariantAccess<'de> for VariantRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + fn unit_variant(self) -> Result<(), E> { + match self.value { + Some(value) => de::Deserialize::deserialize(ContentRefDeserializer::new(value)), + None => Ok(()), + } + } + + fn newtype_variant_seed(self, seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + match self.value { + Some(value) => seed.deserialize(ContentRefDeserializer::new(value)), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"newtype variant", + )), + } + } + + fn tuple_variant(self, _len: usize, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + match self.value { + Some(Content::Seq(v)) => { + de::Deserializer::deserialize_any(SeqRefDeserializer::new(v), visitor) + } + Some(other) => Err(de::Error::invalid_type( + other.unexpected(), + &"tuple variant", + )), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"tuple variant", + )), + } + } + + fn struct_variant( + self, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: de::Visitor<'de>, + { + match self.value { + Some(Content::Map(v)) => { + de::Deserializer::deserialize_any(MapRefDeserializer::new(v), visitor) + } + Some(Content::Seq(v)) => { + de::Deserializer::deserialize_any(SeqRefDeserializer::new(v), visitor) + } + Some(other) => Err(de::Error::invalid_type( + other.unexpected(), + &"struct variant", + )), + None => Err(de::Error::invalid_type( + de::Unexpected::UnitVariant, + &"struct variant", + )), + } + } + } + + struct SeqRefDeserializer<'a, 'de: 'a, E> + where + E: de::Error, + { + iter: <&'a [Content<'de>] as IntoIterator>::IntoIter, + err: PhantomData, + } + + impl<'a, 'de, E> SeqRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + fn new(slice: &'a [Content<'de>]) -> Self { + SeqRefDeserializer { + iter: slice.iter(), + err: PhantomData, + } + } + } + + impl<'de, 'a, E> de::Deserializer<'de> for SeqRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + #[inline] + fn deserialize_any(mut self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + let len = self.iter.len(); + if len == 0 { + visitor.visit_unit() + } else { + let ret = try!(visitor.visit_seq(&mut self)); + let remaining = self.iter.len(); + if remaining == 0 { + Ok(ret) + } else { + Err(de::Error::invalid_length(len, &"fewer elements in array")) + } + } + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } + } + + impl<'de, 'a, E> de::SeqAccess<'de> for SeqRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + fn next_element_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.iter.next() { + Some(value) => seed + .deserialize(ContentRefDeserializer::new(value)) + .map(Some), + None => Ok(None), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } + } + + struct MapRefDeserializer<'a, 'de: 'a, E> + where + E: de::Error, + { + iter: <&'a [(Content<'de>, Content<'de>)] as IntoIterator>::IntoIter, + value: Option<&'a Content<'de>>, + err: PhantomData, + } + + impl<'a, 'de, E> MapRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + fn new(map: &'a [(Content<'de>, Content<'de>)]) -> Self { + MapRefDeserializer { + iter: map.iter(), + value: None, + err: PhantomData, + } + } + } + + impl<'de, 'a, E> de::MapAccess<'de> for MapRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: de::DeserializeSeed<'de>, + { + match self.iter.next() { + Some((key, value)) => { + self.value = Some(value); + seed.deserialize(ContentRefDeserializer::new(key)).map(Some) + } + None => Ok(None), + } + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: de::DeserializeSeed<'de>, + { + match self.value.take() { + Some(value) => seed.deserialize(ContentRefDeserializer::new(value)), + None => Err(de::Error::custom("value is missing")), + } + } + + fn size_hint(&self) -> Option { + size_hint::from_bounds(&self.iter) + } + } + + impl<'de, 'a, E> de::Deserializer<'de> for MapRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Error = E; + + #[inline] + fn deserialize_any(self, visitor: V) -> Result + where + V: de::Visitor<'de>, + { + visitor.visit_map(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } + } + + impl<'de, E> de::IntoDeserializer<'de, E> for ContentDeserializer<'de, E> + where + E: de::Error, + { + type Deserializer = Self; + + fn into_deserializer(self) -> Self { + self + } + } + + impl<'de, 'a, E> de::IntoDeserializer<'de, E> for ContentRefDeserializer<'a, 'de, E> + where + E: de::Error, + { + type Deserializer = Self; + + fn into_deserializer(self) -> Self { + self + } + } + + /// Visitor for deserializing an internally tagged unit variant. + /// + /// Not public API. + pub struct InternallyTaggedUnitVisitor<'a> { + type_name: &'a str, + variant_name: &'a str, + } + + impl<'a> InternallyTaggedUnitVisitor<'a> { + /// Not public API. + pub fn new(type_name: &'a str, variant_name: &'a str) -> Self { + InternallyTaggedUnitVisitor { + type_name: type_name, + variant_name: variant_name, + } + } + } + + impl<'de, 'a> Visitor<'de> for InternallyTaggedUnitVisitor<'a> { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!( + formatter, + "unit variant {}::{}", + self.type_name, self.variant_name + ) + } + + fn visit_seq(self, _: S) -> Result<(), S::Error> + where + S: SeqAccess<'de>, + { + Ok(()) + } + + fn visit_map(self, mut access: M) -> Result<(), M::Error> + where + M: MapAccess<'de>, + { + while try!(access.next_entry::()).is_some() {} + Ok(()) + } + } + + /// Visitor for deserializing an untagged unit variant. + /// + /// Not public API. + pub struct UntaggedUnitVisitor<'a> { + type_name: &'a str, + variant_name: &'a str, + } + + impl<'a> UntaggedUnitVisitor<'a> { + /// Not public API. + pub fn new(type_name: &'a str, variant_name: &'a str) -> Self { + UntaggedUnitVisitor { + type_name: type_name, + variant_name: variant_name, + } + } + } + + impl<'de, 'a> Visitor<'de> for UntaggedUnitVisitor<'a> { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!( + formatter, + "unit variant {}::{}", + self.type_name, self.variant_name + ) + } + + fn visit_unit(self) -> Result<(), E> + where + E: de::Error, + { + Ok(()) + } + + fn visit_none(self) -> Result<(), E> + where + E: de::Error, + { + Ok(()) + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +// Like `IntoDeserializer` but also implemented for `&[u8]`. This is used for +// the newtype fallthrough case of `field_identifier`. +// +// #[derive(Deserialize)] +// #[serde(field_identifier)] +// enum F { +// A, +// B, +// Other(String), // deserialized using IdentifierDeserializer +// } +pub trait IdentifierDeserializer<'de, E: Error> { + type Deserializer: Deserializer<'de, Error = E>; + + fn from(self) -> Self::Deserializer; +} + +pub struct Borrowed<'de, T: 'de + ?Sized>(pub &'de T); + +impl<'de, E> IdentifierDeserializer<'de, E> for u64 +where + E: Error, +{ + type Deserializer = >::Deserializer; + + fn from(self) -> Self::Deserializer { + self.into_deserializer() + } +} + +pub struct StrDeserializer<'a, E> { + value: &'a str, + marker: PhantomData, +} + +impl<'de, 'a, E> Deserializer<'de> for StrDeserializer<'a, E> +where + E: Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_str(self.value) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +pub struct BorrowedStrDeserializer<'de, E> { + value: &'de str, + marker: PhantomData, +} + +impl<'de, E> Deserializer<'de> for BorrowedStrDeserializer<'de, E> +where + E: Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_borrowed_str(self.value) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct enum identifier ignored_any + } +} + +impl<'a, E> IdentifierDeserializer<'a, E> for &'a str +where + E: Error, +{ + type Deserializer = StrDeserializer<'a, E>; + + fn from(self) -> Self::Deserializer { + StrDeserializer { + value: self, + marker: PhantomData, + } + } +} + +impl<'de, E> IdentifierDeserializer<'de, E> for Borrowed<'de, str> +where + E: Error, +{ + type Deserializer = BorrowedStrDeserializer<'de, E>; + + fn from(self) -> Self::Deserializer { + BorrowedStrDeserializer { + value: self.0, + marker: PhantomData, + } + } +} + +impl<'a, E> IdentifierDeserializer<'a, E> for &'a [u8] +where + E: Error, +{ + type Deserializer = BytesDeserializer<'a, E>; + + fn from(self) -> Self::Deserializer { + BytesDeserializer::new(self) + } +} + +impl<'de, E> IdentifierDeserializer<'de, E> for Borrowed<'de, [u8]> +where + E: Error, +{ + type Deserializer = BorrowedBytesDeserializer<'de, E>; + + fn from(self) -> Self::Deserializer { + BorrowedBytesDeserializer::new(self.0) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapDeserializer<'a, 'de: 'a, E>( + pub &'a mut Vec, Content<'de>)>>, + pub PhantomData, +); + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> FlatMapDeserializer<'a, 'de, E> +where + E: Error, +{ + fn deserialize_other() -> Result { + Err(Error::custom("can only flatten structs and maps")) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +macro_rules! forward_to_deserialize_other { + ($($func:ident ($($arg:ty),*))*) => { + $( + fn $func(self, $(_: $arg,)* _visitor: V) -> Result + where + V: Visitor<'de>, + { + Self::deserialize_other() + } + )* + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> Deserializer<'de> for FlatMapDeserializer<'a, 'de, E> +where + E: Error, +{ + type Error = E; + + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_map(FlatInternallyTaggedAccess { + iter: self.0.iter_mut(), + pending: None, + _marker: PhantomData, + }) + } + + fn deserialize_enum( + self, + name: &'static str, + variants: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + for item in self.0.iter_mut() { + // items in the vector are nulled out when used. So we can only use + // an item if it's still filled in and if the field is one we care + // about. + let use_item = match *item { + None => false, + Some((ref c, _)) => c.as_str().map_or(false, |x| variants.contains(&x)), + }; + + if use_item { + let (key, value) = item.take().unwrap(); + return visitor.visit_enum(EnumDeserializer::new(key, Some(value))); + } + } + + Err(Error::custom(format_args!( + "no variant of enum {} found in flattened data", + name + ))) + } + + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_map(FlatMapAccess::new(self.0.iter())) + } + + fn deserialize_struct( + self, + _: &'static str, + fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + visitor.visit_map(FlatStructAccess::new(self.0.iter_mut(), fields)) + } + + fn deserialize_newtype_struct(self, _name: &str, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_newtype_struct(self) + } + + fn deserialize_option(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match visitor.__private_visit_untagged_option(self) { + Ok(value) => Ok(value), + Err(()) => Self::deserialize_other(), + } + } + + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_unit() + } + + forward_to_deserialize_other! { + deserialize_bool() + deserialize_i8() + deserialize_i16() + deserialize_i32() + deserialize_i64() + deserialize_u8() + deserialize_u16() + deserialize_u32() + deserialize_u64() + deserialize_f32() + deserialize_f64() + deserialize_char() + deserialize_str() + deserialize_string() + deserialize_bytes() + deserialize_byte_buf() + deserialize_unit_struct(&'static str) + deserialize_seq() + deserialize_tuple(usize) + deserialize_tuple_struct(&'static str, usize) + deserialize_identifier() + deserialize_ignored_any() + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapAccess<'a, 'de: 'a, E> { + iter: slice::Iter<'a, Option<(Content<'de>, Content<'de>)>>, + pending_content: Option<&'a Content<'de>>, + _marker: PhantomData, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> FlatMapAccess<'a, 'de, E> { + fn new( + iter: slice::Iter<'a, Option<(Content<'de>, Content<'de>)>>, + ) -> FlatMapAccess<'a, 'de, E> { + FlatMapAccess { + iter: iter, + pending_content: None, + _marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> MapAccess<'de> for FlatMapAccess<'a, 'de, E> +where + E: Error, +{ + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: DeserializeSeed<'de>, + { + for item in &mut self.iter { + // Items in the vector are nulled out when used by a struct. + if let Some((ref key, ref content)) = *item { + self.pending_content = Some(content); + return seed.deserialize(ContentRefDeserializer::new(key)).map(Some); + } + } + Ok(None) + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: DeserializeSeed<'de>, + { + match self.pending_content.take() { + Some(value) => seed.deserialize(ContentRefDeserializer::new(value)), + None => Err(Error::custom("value is missing")), + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatStructAccess<'a, 'de: 'a, E> { + iter: slice::IterMut<'a, Option<(Content<'de>, Content<'de>)>>, + pending_content: Option>, + fields: &'static [&'static str], + _marker: PhantomData, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> FlatStructAccess<'a, 'de, E> { + fn new( + iter: slice::IterMut<'a, Option<(Content<'de>, Content<'de>)>>, + fields: &'static [&'static str], + ) -> FlatStructAccess<'a, 'de, E> { + FlatStructAccess { + iter: iter, + pending_content: None, + fields: fields, + _marker: PhantomData, + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> MapAccess<'de> for FlatStructAccess<'a, 'de, E> +where + E: Error, +{ + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: DeserializeSeed<'de>, + { + while let Some(item) = self.iter.next() { + // items in the vector are nulled out when used. So we can only use + // an item if it's still filled in and if the field is one we care + // about. In case we do not know which fields we want, we take them all. + let use_item = match *item { + None => false, + Some((ref c, _)) => c.as_str().map_or(false, |key| self.fields.contains(&key)), + }; + + if use_item { + let (key, content) = item.take().unwrap(); + self.pending_content = Some(content); + return seed.deserialize(ContentDeserializer::new(key)).map(Some); + } + } + Ok(None) + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: DeserializeSeed<'de>, + { + match self.pending_content.take() { + Some(value) => seed.deserialize(ContentDeserializer::new(value)), + None => Err(Error::custom("value is missing")), + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatInternallyTaggedAccess<'a, 'de: 'a, E> { + iter: slice::IterMut<'a, Option<(Content<'de>, Content<'de>)>>, + pending: Option<&'a Content<'de>>, + _marker: PhantomData, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, 'de, E> MapAccess<'de> for FlatInternallyTaggedAccess<'a, 'de, E> +where + E: Error, +{ + type Error = E; + + fn next_key_seed(&mut self, seed: T) -> Result, Self::Error> + where + T: DeserializeSeed<'de>, + { + for item in &mut self.iter { + if let Some((ref key, ref content)) = *item { + // Do not take(), instead borrow this entry. The internally tagged + // enum does its own buffering so we can't tell whether this entry + // is going to be consumed. Borrowing here leaves the entry + // available for later flattened fields. + self.pending = Some(content); + return seed.deserialize(ContentRefDeserializer::new(key)).map(Some); + } + } + Ok(None) + } + + fn next_value_seed(&mut self, seed: T) -> Result + where + T: DeserializeSeed<'de>, + { + match self.pending.take() { + Some(value) => seed.deserialize(ContentRefDeserializer::new(value)), + None => panic!("value is missing"), + } + } +} diff --git a/rust/serde/private/doc.rs b/rust/serde/private/doc.rs new file mode 100644 index 00000000000000..271e0c4f7b70cb --- /dev/null +++ b/rust/serde/private/doc.rs @@ -0,0 +1,161 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Used only by Serde doc tests. Not public API. + +use lib::*; + +use ser; + +#[doc(hidden)] +#[derive(Debug)] +pub struct Error; + +impl ser::Error for Error { + fn custom(_: T) -> Self + where + T: Display, + { + unimplemented!() + } +} + +#[cfg(feature = "std")] +impl error::Error for Error { + fn description(&self) -> &str { + unimplemented!() + } +} + +impl Display for Error { + fn fmt(&self, _: &mut fmt::Formatter) -> fmt::Result { + unimplemented!() + } +} + +#[doc(hidden)] +#[macro_export] +macro_rules! __private_serialize { + () => { + trait Serialize { + fn serialize(&self, serializer: S) -> Result + where + S: $crate::Serializer; + } + }; +} + +#[doc(hidden)] +#[macro_export(local_inner_macros)] +macro_rules! __serialize_unimplemented { + ($($func:ident)*) => { + $( + __serialize_unimplemented_helper!($func); + )* + }; +} + +#[doc(hidden)] +#[macro_export] +macro_rules! __serialize_unimplemented_method { + ($func:ident $(<$t:ident>)* ($($arg:ty),*) -> $ret:ident) => { + fn $func $(<$t: ?Sized + $crate::Serialize>)* (self $(, _: $arg)*) -> $crate::__private::Result { + unimplemented!() + } + }; +} + +#[doc(hidden)] +#[macro_export(local_inner_macros)] +macro_rules! __serialize_unimplemented_helper { + (bool) => { + __serialize_unimplemented_method!(serialize_bool(bool) -> Ok); + }; + (i8) => { + __serialize_unimplemented_method!(serialize_i8(i8) -> Ok); + }; + (i16) => { + __serialize_unimplemented_method!(serialize_i16(i16) -> Ok); + }; + (i32) => { + __serialize_unimplemented_method!(serialize_i32(i32) -> Ok); + }; + (i64) => { + __serialize_unimplemented_method!(serialize_i64(i64) -> Ok); + }; + (u8) => { + __serialize_unimplemented_method!(serialize_u8(u8) -> Ok); + }; + (u16) => { + __serialize_unimplemented_method!(serialize_u16(u16) -> Ok); + }; + (u32) => { + __serialize_unimplemented_method!(serialize_u32(u32) -> Ok); + }; + (u64) => { + __serialize_unimplemented_method!(serialize_u64(u64) -> Ok); + }; + (f32) => { + __serialize_unimplemented_method!(serialize_f32(f32) -> Ok); + }; + (f64) => { + __serialize_unimplemented_method!(serialize_f64(f64) -> Ok); + }; + (char) => { + __serialize_unimplemented_method!(serialize_char(char) -> Ok); + }; + (str) => { + __serialize_unimplemented_method!(serialize_str(&str) -> Ok); + }; + (bytes) => { + __serialize_unimplemented_method!(serialize_bytes(&[u8]) -> Ok); + }; + (none) => { + __serialize_unimplemented_method!(serialize_none() -> Ok); + }; + (some) => { + __serialize_unimplemented_method!(serialize_some(&T) -> Ok); + }; + (unit) => { + __serialize_unimplemented_method!(serialize_unit() -> Ok); + }; + (unit_struct) => { + __serialize_unimplemented_method!(serialize_unit_struct(&str) -> Ok); + }; + (unit_variant) => { + __serialize_unimplemented_method!(serialize_unit_variant(&str, u32, &str) -> Ok); + }; + (newtype_struct) => { + __serialize_unimplemented_method!(serialize_newtype_struct(&str, &T) -> Ok); + }; + (newtype_variant) => { + __serialize_unimplemented_method!(serialize_newtype_variant(&str, u32, &str, &T) -> Ok); + }; + (seq) => { + type SerializeSeq = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_seq(Option) -> SerializeSeq); + }; + (tuple) => { + type SerializeTuple = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_tuple(usize) -> SerializeTuple); + }; + (tuple_struct) => { + type SerializeTupleStruct = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_tuple_struct(&str, usize) -> SerializeTupleStruct); + }; + (tuple_variant) => { + type SerializeTupleVariant = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_tuple_variant(&str, u32, &str, usize) -> SerializeTupleVariant); + }; + (map) => { + type SerializeMap = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_map(Option) -> SerializeMap); + }; + (struct) => { + type SerializeStruct = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_struct(&str, usize) -> SerializeStruct); + }; + (struct_variant) => { + type SerializeStructVariant = $crate::ser::Impossible; + __serialize_unimplemented_method!(serialize_struct_variant(&str, u32, &str, usize) -> SerializeStructVariant); + }; +} diff --git a/rust/serde/private/mod.rs b/rust/serde/private/mod.rs new file mode 100644 index 00000000000000..c05a0ea5c8e315 --- /dev/null +++ b/rust/serde/private/mod.rs @@ -0,0 +1,52 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg(not(no_serde_derive))] +pub mod de; +#[cfg(not(no_serde_derive))] +pub mod ser; + +pub mod size_hint; + +// FIXME: #[cfg(doctest)] once https://github.com/rust-lang/rust/issues/67295 is fixed. +pub mod doc; + +pub use lib::clone::Clone; +pub use lib::convert::{From, Into}; +pub use lib::default::Default; +pub use lib::fmt::{self, Formatter}; +pub use lib::marker::PhantomData; +pub use lib::option::Option::{self, None, Some}; +pub use lib::ptr; +pub use lib::result::Result::{self, Err, Ok}; + +pub use self::string::from_utf8_lossy; + +#[cfg(any(feature = "alloc", feature = "std"))] +pub use lib::{ToString, Vec}; + +#[cfg(not(no_core_try_from))] +pub use lib::convert::TryFrom; + +mod string { + use lib::*; + + #[cfg(any(feature = "std", feature = "alloc"))] + pub fn from_utf8_lossy(bytes: &[u8]) -> Cow { + String::from_utf8_lossy(bytes) + } + + // The generated code calls this like: + // + // let value = &_serde::__private::from_utf8_lossy(bytes); + // Err(_serde::de::Error::unknown_variant(value, VARIANTS)) + // + // so it is okay for the return type to be different from the std case as long + // as the above works. + #[cfg(not(any(feature = "std", feature = "alloc")))] + pub fn from_utf8_lossy(bytes: &[u8]) -> &str { + // Three unicode replacement characters if it fails. They look like a + // white-on-black question mark. The user will recognize it as invalid + // UTF-8. + str::from_utf8(bytes).unwrap_or("\u{fffd}\u{fffd}\u{fffd}") + } +} diff --git a/rust/serde/private/ser.rs b/rust/serde/private/ser.rs new file mode 100644 index 00000000000000..6efd84e5fe579d --- /dev/null +++ b/rust/serde/private/ser.rs @@ -0,0 +1,1316 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +use ser::{self, Impossible, Serialize, SerializeMap, SerializeStruct, Serializer}; + +#[cfg(any(feature = "std", feature = "alloc"))] +use self::content::{ + Content, ContentSerializer, SerializeStructVariantAsMapValue, SerializeTupleVariantAsMapValue, +}; + +/// Used to check that serde(getter) attributes return the expected type. +/// Not public API. +pub fn constrain(t: &T) -> &T { + t +} + +/// Not public API. +pub fn serialize_tagged_newtype( + serializer: S, + type_ident: &'static str, + variant_ident: &'static str, + tag: &'static str, + variant_name: &'static str, + value: &T, +) -> Result +where + S: Serializer, + T: Serialize, +{ + value.serialize(TaggedSerializer { + type_ident: type_ident, + variant_ident: variant_ident, + tag: tag, + variant_name: variant_name, + delegate: serializer, + }) +} + +struct TaggedSerializer { + type_ident: &'static str, + variant_ident: &'static str, + tag: &'static str, + variant_name: &'static str, + delegate: S, +} + +enum Unsupported { + Boolean, + Integer, + #[cfg(not(no_fp_fmt_parse))] + Float, + Char, + String, + ByteArray, + Optional, + #[cfg(any(feature = "std", feature = "alloc"))] + UnitStruct, + Sequence, + Tuple, + TupleStruct, + Enum, +} + +impl Display for Unsupported { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match *self { + Unsupported::Boolean => formatter.write_str("a boolean"), + Unsupported::Integer => formatter.write_str("an integer"), + #[cfg(not(no_fp_fmt_parse))] + Unsupported::Float => formatter.write_str("a float"), + Unsupported::Char => formatter.write_str("a char"), + Unsupported::String => formatter.write_str("a string"), + Unsupported::ByteArray => formatter.write_str("a byte array"), + Unsupported::Optional => formatter.write_str("an optional"), + #[cfg(any(feature = "std", feature = "alloc"))] + Unsupported::UnitStruct => formatter.write_str("unit struct"), + Unsupported::Sequence => formatter.write_str("a sequence"), + Unsupported::Tuple => formatter.write_str("a tuple"), + Unsupported::TupleStruct => formatter.write_str("a tuple struct"), + Unsupported::Enum => formatter.write_str("an enum"), + } + } +} + +impl TaggedSerializer +where + S: Serializer, +{ + fn bad_type(self, what: Unsupported) -> S::Error { + ser::Error::custom(format_args!( + "cannot serialize tagged newtype variant {}::{} containing {}", + self.type_ident, self.variant_ident, what + )) + } +} + +impl Serializer for TaggedSerializer +where + S: Serializer, +{ + type Ok = S::Ok; + type Error = S::Error; + + type SerializeSeq = Impossible; + type SerializeTuple = Impossible; + type SerializeTupleStruct = Impossible; + type SerializeMap = S::SerializeMap; + type SerializeStruct = S::SerializeStruct; + + #[cfg(not(any(feature = "std", feature = "alloc")))] + type SerializeTupleVariant = Impossible; + #[cfg(any(feature = "std", feature = "alloc"))] + type SerializeTupleVariant = SerializeTupleVariantAsMapValue; + + #[cfg(not(any(feature = "std", feature = "alloc")))] + type SerializeStructVariant = Impossible; + #[cfg(any(feature = "std", feature = "alloc"))] + type SerializeStructVariant = SerializeStructVariantAsMapValue; + + fn serialize_bool(self, _: bool) -> Result { + Err(self.bad_type(Unsupported::Boolean)) + } + + fn serialize_i8(self, _: i8) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_i16(self, _: i16) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_i32(self, _: i32) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_i64(self, _: i64) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_u8(self, _: u8) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_u16(self, _: u16) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_u32(self, _: u32) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + fn serialize_u64(self, _: u64) -> Result { + Err(self.bad_type(Unsupported::Integer)) + } + + #[cfg(not(no_fp_fmt_parse))] + fn serialize_f32(self, _: f32) -> Result { + Err(self.bad_type(Unsupported::Float)) + } + + #[cfg(not(no_fp_fmt_parse))] + fn serialize_f64(self, _: f64) -> Result { + Err(self.bad_type(Unsupported::Float)) + } + + fn serialize_char(self, _: char) -> Result { + Err(self.bad_type(Unsupported::Char)) + } + + fn serialize_str(self, _: &str) -> Result { + Err(self.bad_type(Unsupported::String)) + } + + fn serialize_bytes(self, _: &[u8]) -> Result { + Err(self.bad_type(Unsupported::ByteArray)) + } + + fn serialize_none(self) -> Result { + Err(self.bad_type(Unsupported::Optional)) + } + + fn serialize_some(self, _: &T) -> Result + where + T: Serialize, + { + Err(self.bad_type(Unsupported::Optional)) + } + + fn serialize_unit(self) -> Result { + let mut map = try!(self.delegate.serialize_map(Some(1))); + try!(map.serialize_entry(self.tag, self.variant_name)); + map.end() + } + + fn serialize_unit_struct(self, _: &'static str) -> Result { + let mut map = try!(self.delegate.serialize_map(Some(1))); + try!(map.serialize_entry(self.tag, self.variant_name)); + map.end() + } + + fn serialize_unit_variant( + self, + _: &'static str, + _: u32, + inner_variant: &'static str, + ) -> Result { + let mut map = try!(self.delegate.serialize_map(Some(2))); + try!(map.serialize_entry(self.tag, self.variant_name)); + try!(map.serialize_entry(inner_variant, &())); + map.end() + } + + fn serialize_newtype_struct( + self, + _: &'static str, + value: &T, + ) -> Result + where + T: Serialize, + { + value.serialize(self) + } + + fn serialize_newtype_variant( + self, + _: &'static str, + _: u32, + inner_variant: &'static str, + inner_value: &T, + ) -> Result + where + T: Serialize, + { + let mut map = try!(self.delegate.serialize_map(Some(2))); + try!(map.serialize_entry(self.tag, self.variant_name)); + try!(map.serialize_entry(inner_variant, inner_value)); + map.end() + } + + fn serialize_seq(self, _: Option) -> Result { + Err(self.bad_type(Unsupported::Sequence)) + } + + fn serialize_tuple(self, _: usize) -> Result { + Err(self.bad_type(Unsupported::Tuple)) + } + + fn serialize_tuple_struct( + self, + _: &'static str, + _: usize, + ) -> Result { + Err(self.bad_type(Unsupported::TupleStruct)) + } + + #[cfg(not(any(feature = "std", feature = "alloc")))] + fn serialize_tuple_variant( + self, + _: &'static str, + _: u32, + _: &'static str, + _: usize, + ) -> Result { + // Lack of push-based serialization means we need to buffer the content + // of the tuple variant, so it requires std. + Err(self.bad_type(Unsupported::Enum)) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + fn serialize_tuple_variant( + self, + _: &'static str, + _: u32, + inner_variant: &'static str, + len: usize, + ) -> Result { + let mut map = try!(self.delegate.serialize_map(Some(2))); + try!(map.serialize_entry(self.tag, self.variant_name)); + try!(map.serialize_key(inner_variant)); + Ok(SerializeTupleVariantAsMapValue::new( + map, + inner_variant, + len, + )) + } + + fn serialize_map(self, len: Option) -> Result { + let mut map = try!(self.delegate.serialize_map(len.map(|len| len + 1))); + try!(map.serialize_entry(self.tag, self.variant_name)); + Ok(map) + } + + fn serialize_struct( + self, + name: &'static str, + len: usize, + ) -> Result { + let mut state = try!(self.delegate.serialize_struct(name, len + 1)); + try!(state.serialize_field(self.tag, self.variant_name)); + Ok(state) + } + + #[cfg(not(any(feature = "std", feature = "alloc")))] + fn serialize_struct_variant( + self, + _: &'static str, + _: u32, + _: &'static str, + _: usize, + ) -> Result { + // Lack of push-based serialization means we need to buffer the content + // of the struct variant, so it requires std. + Err(self.bad_type(Unsupported::Enum)) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + fn serialize_struct_variant( + self, + _: &'static str, + _: u32, + inner_variant: &'static str, + len: usize, + ) -> Result { + let mut map = try!(self.delegate.serialize_map(Some(2))); + try!(map.serialize_entry(self.tag, self.variant_name)); + try!(map.serialize_key(inner_variant)); + Ok(SerializeStructVariantAsMapValue::new( + map, + inner_variant, + len, + )) + } + + #[cfg(not(any(feature = "std", feature = "alloc")))] + fn collect_str(self, _: &T) -> Result + where + T: Display, + { + Err(self.bad_type(Unsupported::String)) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +mod content { + use lib::*; + + use ser::{self, Serialize, Serializer}; + + pub struct SerializeTupleVariantAsMapValue { + map: M, + name: &'static str, + fields: Vec, + } + + impl SerializeTupleVariantAsMapValue { + pub fn new(map: M, name: &'static str, len: usize) -> Self { + SerializeTupleVariantAsMapValue { + map: map, + name: name, + fields: Vec::with_capacity(len), + } + } + } + + impl ser::SerializeTupleVariant for SerializeTupleVariantAsMapValue + where + M: ser::SerializeMap, + { + type Ok = M::Ok; + type Error = M::Error; + + fn serialize_field(&mut self, value: &T) -> Result<(), M::Error> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push(value); + Ok(()) + } + + fn end(mut self) -> Result { + try!(self + .map + .serialize_value(&Content::TupleStruct(self.name, self.fields))); + self.map.end() + } + } + + pub struct SerializeStructVariantAsMapValue { + map: M, + name: &'static str, + fields: Vec<(&'static str, Content)>, + } + + impl SerializeStructVariantAsMapValue { + pub fn new(map: M, name: &'static str, len: usize) -> Self { + SerializeStructVariantAsMapValue { + map: map, + name: name, + fields: Vec::with_capacity(len), + } + } + } + + impl ser::SerializeStructVariant for SerializeStructVariantAsMapValue + where + M: ser::SerializeMap, + { + type Ok = M::Ok; + type Error = M::Error; + + fn serialize_field( + &mut self, + key: &'static str, + value: &T, + ) -> Result<(), M::Error> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push((key, value)); + Ok(()) + } + + fn end(mut self) -> Result { + try!(self + .map + .serialize_value(&Content::Struct(self.name, self.fields))); + self.map.end() + } + } + + pub enum Content { + Bool(bool), + + U8(u8), + U16(u16), + U32(u32), + U64(u64), + + I8(i8), + I16(i16), + I32(i32), + I64(i64), + + F32(f32), + F64(f64), + + Char(char), + String(String), + Bytes(Vec), + + None, + Some(Box), + + Unit, + UnitStruct(&'static str), + UnitVariant(&'static str, u32, &'static str), + NewtypeStruct(&'static str, Box), + NewtypeVariant(&'static str, u32, &'static str, Box), + + Seq(Vec), + Tuple(Vec), + TupleStruct(&'static str, Vec), + TupleVariant(&'static str, u32, &'static str, Vec), + Map(Vec<(Content, Content)>), + Struct(&'static str, Vec<(&'static str, Content)>), + StructVariant( + &'static str, + u32, + &'static str, + Vec<(&'static str, Content)>, + ), + } + + impl Serialize for Content { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match *self { + Content::Bool(b) => serializer.serialize_bool(b), + Content::U8(u) => serializer.serialize_u8(u), + Content::U16(u) => serializer.serialize_u16(u), + Content::U32(u) => serializer.serialize_u32(u), + Content::U64(u) => serializer.serialize_u64(u), + Content::I8(i) => serializer.serialize_i8(i), + Content::I16(i) => serializer.serialize_i16(i), + Content::I32(i) => serializer.serialize_i32(i), + Content::I64(i) => serializer.serialize_i64(i), + Content::F32(f) => serializer.serialize_f32(f), + Content::F64(f) => serializer.serialize_f64(f), + Content::Char(c) => serializer.serialize_char(c), + Content::String(ref s) => serializer.serialize_str(s), + Content::Bytes(ref b) => serializer.serialize_bytes(b), + Content::None => serializer.serialize_none(), + Content::Some(ref c) => serializer.serialize_some(&**c), + Content::Unit => serializer.serialize_unit(), + Content::UnitStruct(n) => serializer.serialize_unit_struct(n), + Content::UnitVariant(n, i, v) => serializer.serialize_unit_variant(n, i, v), + Content::NewtypeStruct(n, ref c) => serializer.serialize_newtype_struct(n, &**c), + Content::NewtypeVariant(n, i, v, ref c) => { + serializer.serialize_newtype_variant(n, i, v, &**c) + } + Content::Seq(ref elements) => elements.serialize(serializer), + Content::Tuple(ref elements) => { + use ser::SerializeTuple; + let mut tuple = try!(serializer.serialize_tuple(elements.len())); + for e in elements { + try!(tuple.serialize_element(e)); + } + tuple.end() + } + Content::TupleStruct(n, ref fields) => { + use ser::SerializeTupleStruct; + let mut ts = try!(serializer.serialize_tuple_struct(n, fields.len())); + for f in fields { + try!(ts.serialize_field(f)); + } + ts.end() + } + Content::TupleVariant(n, i, v, ref fields) => { + use ser::SerializeTupleVariant; + let mut tv = try!(serializer.serialize_tuple_variant(n, i, v, fields.len())); + for f in fields { + try!(tv.serialize_field(f)); + } + tv.end() + } + Content::Map(ref entries) => { + use ser::SerializeMap; + let mut map = try!(serializer.serialize_map(Some(entries.len()))); + for (k, v) in entries { + try!(map.serialize_entry(k, v)); + } + map.end() + } + Content::Struct(n, ref fields) => { + use ser::SerializeStruct; + let mut s = try!(serializer.serialize_struct(n, fields.len())); + for &(k, ref v) in fields { + try!(s.serialize_field(k, v)); + } + s.end() + } + Content::StructVariant(n, i, v, ref fields) => { + use ser::SerializeStructVariant; + let mut sv = try!(serializer.serialize_struct_variant(n, i, v, fields.len())); + for &(k, ref v) in fields { + try!(sv.serialize_field(k, v)); + } + sv.end() + } + } + } + } + + pub struct ContentSerializer { + error: PhantomData, + } + + impl ContentSerializer { + pub fn new() -> Self { + ContentSerializer { error: PhantomData } + } + } + + impl Serializer for ContentSerializer + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + type SerializeSeq = SerializeSeq; + type SerializeTuple = SerializeTuple; + type SerializeTupleStruct = SerializeTupleStruct; + type SerializeTupleVariant = SerializeTupleVariant; + type SerializeMap = SerializeMap; + type SerializeStruct = SerializeStruct; + type SerializeStructVariant = SerializeStructVariant; + + fn serialize_bool(self, v: bool) -> Result { + Ok(Content::Bool(v)) + } + + fn serialize_i8(self, v: i8) -> Result { + Ok(Content::I8(v)) + } + + fn serialize_i16(self, v: i16) -> Result { + Ok(Content::I16(v)) + } + + fn serialize_i32(self, v: i32) -> Result { + Ok(Content::I32(v)) + } + + fn serialize_i64(self, v: i64) -> Result { + Ok(Content::I64(v)) + } + + fn serialize_u8(self, v: u8) -> Result { + Ok(Content::U8(v)) + } + + fn serialize_u16(self, v: u16) -> Result { + Ok(Content::U16(v)) + } + + fn serialize_u32(self, v: u32) -> Result { + Ok(Content::U32(v)) + } + + fn serialize_u64(self, v: u64) -> Result { + Ok(Content::U64(v)) + } + + fn serialize_f32(self, v: f32) -> Result { + Ok(Content::F32(v)) + } + + fn serialize_f64(self, v: f64) -> Result { + Ok(Content::F64(v)) + } + + fn serialize_char(self, v: char) -> Result { + Ok(Content::Char(v)) + } + + fn serialize_str(self, value: &str) -> Result { + Ok(Content::String(value.to_owned())) + } + + fn serialize_bytes(self, value: &[u8]) -> Result { + Ok(Content::Bytes(value.to_owned())) + } + + fn serialize_none(self) -> Result { + Ok(Content::None) + } + + fn serialize_some(self, value: &T) -> Result + where + T: Serialize, + { + Ok(Content::Some(Box::new(try!(value.serialize(self))))) + } + + fn serialize_unit(self) -> Result { + Ok(Content::Unit) + } + + fn serialize_unit_struct(self, name: &'static str) -> Result { + Ok(Content::UnitStruct(name)) + } + + fn serialize_unit_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + ) -> Result { + Ok(Content::UnitVariant(name, variant_index, variant)) + } + + fn serialize_newtype_struct( + self, + name: &'static str, + value: &T, + ) -> Result + where + T: Serialize, + { + Ok(Content::NewtypeStruct( + name, + Box::new(try!(value.serialize(self))), + )) + } + + fn serialize_newtype_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + value: &T, + ) -> Result + where + T: Serialize, + { + Ok(Content::NewtypeVariant( + name, + variant_index, + variant, + Box::new(try!(value.serialize(self))), + )) + } + + fn serialize_seq(self, len: Option) -> Result { + Ok(SerializeSeq { + elements: Vec::with_capacity(len.unwrap_or(0)), + error: PhantomData, + }) + } + + fn serialize_tuple(self, len: usize) -> Result { + Ok(SerializeTuple { + elements: Vec::with_capacity(len), + error: PhantomData, + }) + } + + fn serialize_tuple_struct( + self, + name: &'static str, + len: usize, + ) -> Result { + Ok(SerializeTupleStruct { + name: name, + fields: Vec::with_capacity(len), + error: PhantomData, + }) + } + + fn serialize_tuple_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result { + Ok(SerializeTupleVariant { + name: name, + variant_index: variant_index, + variant: variant, + fields: Vec::with_capacity(len), + error: PhantomData, + }) + } + + fn serialize_map(self, len: Option) -> Result { + Ok(SerializeMap { + entries: Vec::with_capacity(len.unwrap_or(0)), + key: None, + error: PhantomData, + }) + } + + fn serialize_struct( + self, + name: &'static str, + len: usize, + ) -> Result { + Ok(SerializeStruct { + name: name, + fields: Vec::with_capacity(len), + error: PhantomData, + }) + } + + fn serialize_struct_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result { + Ok(SerializeStructVariant { + name: name, + variant_index: variant_index, + variant: variant, + fields: Vec::with_capacity(len), + error: PhantomData, + }) + } + } + + pub struct SerializeSeq { + elements: Vec, + error: PhantomData, + } + + impl ser::SerializeSeq for SerializeSeq + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_element(&mut self, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.elements.push(value); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::Seq(self.elements)) + } + } + + pub struct SerializeTuple { + elements: Vec, + error: PhantomData, + } + + impl ser::SerializeTuple for SerializeTuple + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_element(&mut self, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.elements.push(value); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::Tuple(self.elements)) + } + } + + pub struct SerializeTupleStruct { + name: &'static str, + fields: Vec, + error: PhantomData, + } + + impl ser::SerializeTupleStruct for SerializeTupleStruct + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_field(&mut self, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push(value); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::TupleStruct(self.name, self.fields)) + } + } + + pub struct SerializeTupleVariant { + name: &'static str, + variant_index: u32, + variant: &'static str, + fields: Vec, + error: PhantomData, + } + + impl ser::SerializeTupleVariant for SerializeTupleVariant + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_field(&mut self, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push(value); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::TupleVariant( + self.name, + self.variant_index, + self.variant, + self.fields, + )) + } + } + + pub struct SerializeMap { + entries: Vec<(Content, Content)>, + key: Option, + error: PhantomData, + } + + impl ser::SerializeMap for SerializeMap + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_key(&mut self, key: &T) -> Result<(), E> + where + T: Serialize, + { + let key = try!(key.serialize(ContentSerializer::::new())); + self.key = Some(key); + Ok(()) + } + + fn serialize_value(&mut self, value: &T) -> Result<(), E> + where + T: Serialize, + { + let key = self + .key + .take() + .expect("serialize_value called before serialize_key"); + let value = try!(value.serialize(ContentSerializer::::new())); + self.entries.push((key, value)); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::Map(self.entries)) + } + + fn serialize_entry(&mut self, key: &K, value: &V) -> Result<(), E> + where + K: Serialize, + V: Serialize, + { + let key = try!(key.serialize(ContentSerializer::::new())); + let value = try!(value.serialize(ContentSerializer::::new())); + self.entries.push((key, value)); + Ok(()) + } + } + + pub struct SerializeStruct { + name: &'static str, + fields: Vec<(&'static str, Content)>, + error: PhantomData, + } + + impl ser::SerializeStruct for SerializeStruct + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_field(&mut self, key: &'static str, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push((key, value)); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::Struct(self.name, self.fields)) + } + } + + pub struct SerializeStructVariant { + name: &'static str, + variant_index: u32, + variant: &'static str, + fields: Vec<(&'static str, Content)>, + error: PhantomData, + } + + impl ser::SerializeStructVariant for SerializeStructVariant + where + E: ser::Error, + { + type Ok = Content; + type Error = E; + + fn serialize_field(&mut self, key: &'static str, value: &T) -> Result<(), E> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push((key, value)); + Ok(()) + } + + fn end(self) -> Result { + Ok(Content::StructVariant( + self.name, + self.variant_index, + self.variant, + self.fields, + )) + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapSerializer<'a, M: 'a>(pub &'a mut M); + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> FlatMapSerializer<'a, M> +where + M: SerializeMap + 'a, +{ + fn bad_type(what: Unsupported) -> M::Error { + ser::Error::custom(format_args!( + "can only flatten structs and maps (got {})", + what + )) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> Serializer for FlatMapSerializer<'a, M> +where + M: SerializeMap + 'a, +{ + type Ok = (); + type Error = M::Error; + + type SerializeSeq = Impossible; + type SerializeTuple = Impossible; + type SerializeTupleStruct = Impossible; + type SerializeMap = FlatMapSerializeMap<'a, M>; + type SerializeStruct = FlatMapSerializeStruct<'a, M>; + type SerializeTupleVariant = Impossible; + type SerializeStructVariant = FlatMapSerializeStructVariantAsMapValue<'a, M>; + + fn serialize_bool(self, _: bool) -> Result { + Err(Self::bad_type(Unsupported::Boolean)) + } + + fn serialize_i8(self, _: i8) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_i16(self, _: i16) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_i32(self, _: i32) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_i64(self, _: i64) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_u8(self, _: u8) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_u16(self, _: u16) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_u32(self, _: u32) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_u64(self, _: u64) -> Result { + Err(Self::bad_type(Unsupported::Integer)) + } + + fn serialize_f32(self, _: f32) -> Result { + Err(Self::bad_type(Unsupported::Float)) + } + + fn serialize_f64(self, _: f64) -> Result { + Err(Self::bad_type(Unsupported::Float)) + } + + fn serialize_char(self, _: char) -> Result { + Err(Self::bad_type(Unsupported::Char)) + } + + fn serialize_str(self, _: &str) -> Result { + Err(Self::bad_type(Unsupported::String)) + } + + fn serialize_bytes(self, _: &[u8]) -> Result { + Err(Self::bad_type(Unsupported::ByteArray)) + } + + fn serialize_none(self) -> Result { + Ok(()) + } + + fn serialize_some(self, value: &T) -> Result + where + T: Serialize, + { + value.serialize(self) + } + + fn serialize_unit(self) -> Result { + Ok(()) + } + + fn serialize_unit_struct(self, _: &'static str) -> Result { + Err(Self::bad_type(Unsupported::UnitStruct)) + } + + fn serialize_unit_variant( + self, + _: &'static str, + _: u32, + _: &'static str, + ) -> Result { + Err(Self::bad_type(Unsupported::Enum)) + } + + fn serialize_newtype_struct( + self, + _: &'static str, + value: &T, + ) -> Result + where + T: Serialize, + { + value.serialize(self) + } + + fn serialize_newtype_variant( + self, + _: &'static str, + _: u32, + variant: &'static str, + value: &T, + ) -> Result + where + T: Serialize, + { + try!(self.0.serialize_key(variant)); + self.0.serialize_value(value) + } + + fn serialize_seq(self, _: Option) -> Result { + Err(Self::bad_type(Unsupported::Sequence)) + } + + fn serialize_tuple(self, _: usize) -> Result { + Err(Self::bad_type(Unsupported::Tuple)) + } + + fn serialize_tuple_struct( + self, + _: &'static str, + _: usize, + ) -> Result { + Err(Self::bad_type(Unsupported::TupleStruct)) + } + + fn serialize_tuple_variant( + self, + _: &'static str, + _: u32, + _: &'static str, + _: usize, + ) -> Result { + Err(Self::bad_type(Unsupported::Enum)) + } + + fn serialize_map(self, _: Option) -> Result { + Ok(FlatMapSerializeMap(self.0)) + } + + fn serialize_struct( + self, + _: &'static str, + _: usize, + ) -> Result { + Ok(FlatMapSerializeStruct(self.0)) + } + + fn serialize_struct_variant( + self, + _: &'static str, + _: u32, + inner_variant: &'static str, + _: usize, + ) -> Result { + try!(self.0.serialize_key(inner_variant)); + Ok(FlatMapSerializeStructVariantAsMapValue::new( + self.0, + inner_variant, + )) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapSerializeMap<'a, M: 'a>(&'a mut M); + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> ser::SerializeMap for FlatMapSerializeMap<'a, M> +where + M: SerializeMap + 'a, +{ + type Ok = (); + type Error = M::Error; + + fn serialize_key(&mut self, key: &T) -> Result<(), Self::Error> + where + T: Serialize, + { + self.0.serialize_key(key) + } + + fn serialize_value(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize, + { + self.0.serialize_value(value) + } + + fn serialize_entry( + &mut self, + key: &K, + value: &V, + ) -> Result<(), Self::Error> + where + K: Serialize, + V: Serialize, + { + self.0.serialize_entry(key, value) + } + + fn end(self) -> Result<(), Self::Error> { + Ok(()) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapSerializeStruct<'a, M: 'a>(&'a mut M); + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> ser::SerializeStruct for FlatMapSerializeStruct<'a, M> +where + M: SerializeMap + 'a, +{ + type Ok = (); + type Error = M::Error; + + fn serialize_field( + &mut self, + key: &'static str, + value: &T, + ) -> Result<(), Self::Error> + where + T: Serialize, + { + self.0.serialize_entry(key, value) + } + + fn end(self) -> Result<(), Self::Error> { + Ok(()) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +pub struct FlatMapSerializeStructVariantAsMapValue<'a, M: 'a> { + map: &'a mut M, + name: &'static str, + fields: Vec<(&'static str, Content)>, +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> FlatMapSerializeStructVariantAsMapValue<'a, M> +where + M: SerializeMap + 'a, +{ + fn new(map: &'a mut M, name: &'static str) -> FlatMapSerializeStructVariantAsMapValue<'a, M> { + FlatMapSerializeStructVariantAsMapValue { + map: map, + name: name, + fields: Vec::new(), + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl<'a, M> ser::SerializeStructVariant for FlatMapSerializeStructVariantAsMapValue<'a, M> +where + M: SerializeMap + 'a, +{ + type Ok = (); + type Error = M::Error; + + fn serialize_field( + &mut self, + key: &'static str, + value: &T, + ) -> Result<(), Self::Error> + where + T: Serialize, + { + let value = try!(value.serialize(ContentSerializer::::new())); + self.fields.push((key, value)); + Ok(()) + } + + fn end(self) -> Result<(), Self::Error> { + try!(self + .map + .serialize_value(&Content::Struct(self.name, self.fields))); + Ok(()) + } +} diff --git a/rust/serde/private/size_hint.rs b/rust/serde/private/size_hint.rs new file mode 100644 index 00000000000000..f050fca59deb34 --- /dev/null +++ b/rust/serde/private/size_hint.rs @@ -0,0 +1,23 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +pub fn from_bounds(iter: &I) -> Option +where + I: Iterator, +{ + helper(iter.size_hint()) +} + +#[cfg(any(feature = "std", feature = "alloc"))] +#[inline] +pub fn cautious(hint: Option) -> usize { + cmp::min(hint.unwrap_or(0), 4096) +} + +fn helper(bounds: (usize, Option)) -> Option { + match bounds { + (lower, Some(upper)) if lower == upper => Some(upper), + _ => None, + } +} diff --git a/rust/serde/ser/fmt.rs b/rust/serde/ser/fmt.rs new file mode 100644 index 00000000000000..7906cf76d99fa1 --- /dev/null +++ b/rust/serde/ser/fmt.rs @@ -0,0 +1,180 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; +use ser::{Error, Impossible, Serialize, Serializer}; + +impl Error for fmt::Error { + fn custom(_msg: T) -> Self { + fmt::Error + } +} + +macro_rules! fmt_primitives { + ($($f:ident: $t:ty,)*) => { + $( + fn $f(self, v: $t) -> fmt::Result { + Display::fmt(&v, self) + } + )* + }; +} + +/// ```edition2018 +/// use serde::Serialize; +/// use std::fmt::{self, Display}; +/// +/// #[derive(Serialize)] +/// #[serde(rename_all = "kebab-case")] +/// pub enum MessageType { +/// StartRequest, +/// EndRequest, +/// } +/// +/// impl Display for MessageType { +/// fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { +/// self.serialize(f) +/// } +/// } +/// ``` +impl<'a, 'b> Serializer for &'a mut fmt::Formatter<'b> { + type Ok = (); + type Error = fmt::Error; + type SerializeSeq = Impossible<(), fmt::Error>; + type SerializeTuple = Impossible<(), fmt::Error>; + type SerializeTupleStruct = Impossible<(), fmt::Error>; + type SerializeTupleVariant = Impossible<(), fmt::Error>; + type SerializeMap = Impossible<(), fmt::Error>; + type SerializeStruct = Impossible<(), fmt::Error>; + type SerializeStructVariant = Impossible<(), fmt::Error>; + + fmt_primitives! { + serialize_bool: bool, + serialize_i8: i8, + serialize_i16: i16, + serialize_i32: i32, + serialize_i64: i64, + serialize_u8: u8, + serialize_u16: u16, + serialize_u32: u32, + serialize_u64: u64, + serialize_char: char, + serialize_str: &str, + serialize_unit_struct: &'static str, + } + + #[cfg(not(no_fp_fmt_parse))] + fmt_primitives! { + serialize_f32: f32, + serialize_f64: f64, + } + + serde_if_integer128! { + fmt_primitives! { + serialize_i128: i128, + serialize_u128: u128, + } + } + + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> fmt::Result { + Display::fmt(variant, self) + } + + fn serialize_newtype_struct(self, _name: &'static str, value: &T) -> fmt::Result + where + T: Serialize, + { + Serialize::serialize(value, self) + } + + fn serialize_bytes(self, _v: &[u8]) -> fmt::Result { + Err(fmt::Error) + } + + fn serialize_none(self) -> fmt::Result { + Err(fmt::Error) + } + + fn serialize_some(self, _value: &T) -> fmt::Result + where + T: Serialize, + { + Err(fmt::Error) + } + + fn serialize_unit(self) -> fmt::Result { + Err(fmt::Error) + } + + fn serialize_newtype_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> fmt::Result + where + T: Serialize, + { + Err(fmt::Error) + } + + fn serialize_seq(self, _len: Option) -> Result { + Err(fmt::Error) + } + + fn serialize_tuple(self, _len: usize) -> Result { + Err(fmt::Error) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result { + Err(fmt::Error) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + Err(fmt::Error) + } + + fn serialize_map(self, _len: Option) -> Result { + Err(fmt::Error) + } + + fn serialize_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result { + Err(fmt::Error) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + Err(fmt::Error) + } + + fn collect_str(self, value: &T) -> fmt::Result + where + T: Display, + { + Display::fmt(value, self) + } +} diff --git a/rust/serde/ser/impls.rs b/rust/serde/ser/impls.rs new file mode 100644 index 00000000000000..d70533eb41c3d9 --- /dev/null +++ b/rust/serde/ser/impls.rs @@ -0,0 +1,987 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::*; + +use ser::{Error, Serialize, SerializeTuple, Serializer}; + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! primitive_impl { + ($ty:ident, $method:ident $($cast:tt)*) => { + impl Serialize for $ty { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.$method(*self $($cast)*) + } + } + } +} + +primitive_impl!(bool, serialize_bool); +primitive_impl!(isize, serialize_i64 as i64); +primitive_impl!(i8, serialize_i8); +primitive_impl!(i16, serialize_i16); +primitive_impl!(i32, serialize_i32); +primitive_impl!(i64, serialize_i64); +primitive_impl!(usize, serialize_u64 as u64); +primitive_impl!(u8, serialize_u8); +primitive_impl!(u16, serialize_u16); +primitive_impl!(u32, serialize_u32); +primitive_impl!(u64, serialize_u64); +#[cfg(not(no_fp_fmt_parse))] +primitive_impl!(f32, serialize_f32); +#[cfg(not(no_fp_fmt_parse))] +primitive_impl!(f64, serialize_f64); +primitive_impl!(char, serialize_char); + +serde_if_integer128! { + primitive_impl!(i128, serialize_i128); + primitive_impl!(u128, serialize_u128); +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for str { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_str(self) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl Serialize for String { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_str(self) + } +} + +impl<'a> Serialize for fmt::Arguments<'a> { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_str(self) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", not(no_core_cstr)))] +impl Serialize for CStr { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_bytes(self.to_bytes()) + } +} + +#[cfg(any(feature = "std", all(not(no_core_cstr), feature = "alloc")))] +impl Serialize for CString { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_bytes(self.to_bytes()) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for Option +where + T: Serialize, +{ + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match *self { + Some(ref value) => serializer.serialize_some(value), + None => serializer.serialize_none(), + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for PhantomData { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_unit_struct("PhantomData") + } +} + +//////////////////////////////////////////////////////////////////////////////// + +// Does not require T: Serialize. +impl Serialize for [T; 0] { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + try!(serializer.serialize_tuple(0)).end() + } +} + +macro_rules! array_impls { + ($($len:tt)+) => { + $( + impl Serialize for [T; $len] + where + T: Serialize, + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + let mut seq = try!(serializer.serialize_tuple($len)); + for e in self { + try!(seq.serialize_element(e)); + } + seq.end() + } + } + )+ + } +} + +array_impls! { + 01 02 03 04 05 06 07 08 09 10 + 11 12 13 14 15 16 17 18 19 20 + 21 22 23 24 25 26 27 28 29 30 + 31 32 +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for [T] +where + T: Serialize, +{ + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_seq(self) + } +} + +#[cfg(all(any(feature = "std", feature = "alloc"), not(no_relaxed_trait_bounds)))] +macro_rules! seq_impl { + ($ty:ident ) => { + impl Serialize for $ty + where + T: Serialize, + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_seq(self) + } + } + } +} + +#[cfg(all(any(feature = "std", feature = "alloc"), no_relaxed_trait_bounds))] +macro_rules! seq_impl { + ($ty:ident ) => { + impl Serialize for $ty + where + T: Serialize $(+ $tbound1 $(+ $tbound2)*)*, + $($typaram: $bound,)* + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_seq(self) + } + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!(BinaryHeap); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!(BTreeSet); + +#[cfg(feature = "std")] +seq_impl!(HashSet); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!(LinkedList); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!(Vec); + +#[cfg(any(feature = "std", feature = "alloc"))] +seq_impl!(VecDeque); + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for Range +where + Idx: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use super::SerializeStruct; + let mut state = try!(serializer.serialize_struct("Range", 2)); + try!(state.serialize_field("start", &self.start)); + try!(state.serialize_field("end", &self.end)); + state.end() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(not(no_range_inclusive))] +impl Serialize for RangeInclusive +where + Idx: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use super::SerializeStruct; + let mut state = try!(serializer.serialize_struct("RangeInclusive", 2)); + try!(state.serialize_field("start", &self.start())); + try!(state.serialize_field("end", &self.end())); + state.end() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(not(no_ops_bound), all(feature = "std", not(no_collections_bound))))] +impl Serialize for Bound +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match *self { + Bound::Unbounded => serializer.serialize_unit_variant("Bound", 0, "Unbounded"), + Bound::Included(ref value) => { + serializer.serialize_newtype_variant("Bound", 1, "Included", value) + } + Bound::Excluded(ref value) => { + serializer.serialize_newtype_variant("Bound", 2, "Excluded", value) + } + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for () { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.serialize_unit() + } +} + +#[cfg(feature = "unstable")] +impl Serialize for ! { + fn serialize(&self, _serializer: S) -> Result + where + S: Serializer, + { + *self + } +} + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! tuple_impls { + ($($len:expr => ($($n:tt $name:ident)+))+) => { + $( + impl<$($name),+> Serialize for ($($name,)+) + where + $($name: Serialize,)+ + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + let mut tuple = try!(serializer.serialize_tuple($len)); + $( + try!(tuple.serialize_element(&self.$n)); + )+ + tuple.end() + } + } + )+ + } +} + +tuple_impls! { + 1 => (0 T0) + 2 => (0 T0 1 T1) + 3 => (0 T0 1 T1 2 T2) + 4 => (0 T0 1 T1 2 T2 3 T3) + 5 => (0 T0 1 T1 2 T2 3 T3 4 T4) + 6 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5) + 7 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6) + 8 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7) + 9 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8) + 10 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9) + 11 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10) + 12 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11) + 13 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12) + 14 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13) + 15 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13 14 T14) + 16 => (0 T0 1 T1 2 T2 3 T3 4 T4 5 T5 6 T6 7 T7 8 T8 9 T9 10 T10 11 T11 12 T12 13 T13 14 T14 15 T15) +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(all(any(feature = "std", feature = "alloc"), not(no_relaxed_trait_bounds)))] +macro_rules! map_impl { + ($ty:ident ) => { + impl Serialize for $ty + where + K: Serialize, + V: Serialize, + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_map(self) + } + } + } +} + +#[cfg(all(any(feature = "std", feature = "alloc"), no_relaxed_trait_bounds))] +macro_rules! map_impl { + ($ty:ident ) => { + impl Serialize for $ty + where + K: Serialize $(+ $kbound1 $(+ $kbound2)*)*, + V: Serialize, + $($typaram: $bound,)* + { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + serializer.collect_map(self) + } + } + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +map_impl!(BTreeMap); + +#[cfg(feature = "std")] +map_impl!(HashMap); + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! deref_impl { + ( + $(#[doc = $doc:tt])* + <$($desc:tt)+ + ) => { + $(#[doc = $doc])* + impl <$($desc)+ { + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + (**self).serialize(serializer) + } + } + }; +} + +deref_impl!(<'a, T: ?Sized> Serialize for &'a T where T: Serialize); +deref_impl!(<'a, T: ?Sized> Serialize for &'a mut T where T: Serialize); + +#[cfg(any(feature = "std", feature = "alloc"))] +deref_impl!( Serialize for Box where T: Serialize); + +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +deref_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Serializing a data structure containing `Rc` will serialize a copy of + /// the contents of the `Rc` each time the `Rc` is referenced within the + /// data structure. Serialization will not attempt to deduplicate these + /// repeated data. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + Serialize for Rc where T: Serialize +} + +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +deref_impl! { + /// This impl requires the [`"rc"`] Cargo feature of Serde. + /// + /// Serializing a data structure containing `Arc` will serialize a copy of + /// the contents of the `Arc` each time the `Arc` is referenced within the + /// data structure. Serialization will not attempt to deduplicate these + /// repeated data. + /// + /// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc + Serialize for Arc where T: Serialize +} + +#[cfg(any(feature = "std", feature = "alloc"))] +deref_impl!(<'a, T: ?Sized> Serialize for Cow<'a, T> where T: Serialize + ToOwned); + +//////////////////////////////////////////////////////////////////////////////// + +/// This impl requires the [`"rc"`] Cargo feature of Serde. +/// +/// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +impl Serialize for RcWeak +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.upgrade().serialize(serializer) + } +} + +/// This impl requires the [`"rc"`] Cargo feature of Serde. +/// +/// [`"rc"`]: https://serde.rs/feature-flags.html#-features-rc +#[cfg(all(feature = "rc", any(feature = "std", feature = "alloc")))] +impl Serialize for ArcWeak +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.upgrade().serialize(serializer) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! nonzero_integers { + ($($T:ident,)+) => { + $( + #[cfg(not(no_num_nonzero))] + impl Serialize for num::$T { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.get().serialize(serializer) + } + } + )+ + } +} + +nonzero_integers! { + NonZeroU8, + NonZeroU16, + NonZeroU32, + NonZeroU64, + NonZeroUsize, +} + +#[cfg(not(no_num_nonzero_signed))] +nonzero_integers! { + NonZeroI8, + NonZeroI16, + NonZeroI32, + NonZeroI64, + NonZeroIsize, +} + +// Currently 128-bit integers do not work on Emscripten targets so we need an +// additional `#[cfg]` +serde_if_integer128! { + nonzero_integers! { + NonZeroU128, + } + + #[cfg(not(no_num_nonzero_signed))] + nonzero_integers! { + NonZeroI128, + } +} + +impl Serialize for Cell +where + T: Serialize + Copy, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.get().serialize(serializer) + } +} + +impl Serialize for RefCell +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match self.try_borrow() { + Ok(value) => value.serialize(serializer), + Err(_) => Err(S::Error::custom("already mutably borrowed")), + } + } +} + +#[cfg(feature = "std")] +impl Serialize for Mutex +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match self.lock() { + Ok(locked) => locked.serialize(serializer), + Err(_) => Err(S::Error::custom("lock poison error while serializing")), + } + } +} + +#[cfg(feature = "std")] +impl Serialize for RwLock +where + T: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match self.read() { + Ok(locked) => locked.serialize(serializer), + Err(_) => Err(S::Error::custom("lock poison error while serializing")), + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for Result +where + T: Serialize, + E: Serialize, +{ + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match *self { + Result::Ok(ref value) => serializer.serialize_newtype_variant("Result", 0, "Ok", value), + Result::Err(ref value) => { + serializer.serialize_newtype_variant("Result", 1, "Err", value) + } + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(any(feature = "std", not(no_core_duration)))] +impl Serialize for Duration { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use super::SerializeStruct; + let mut state = try!(serializer.serialize_struct("Duration", 2)); + try!(state.serialize_field("secs", &self.as_secs())); + try!(state.serialize_field("nanos", &self.subsec_nanos())); + state.end() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +impl Serialize for SystemTime { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use super::SerializeStruct; + let duration_since_epoch = match self.duration_since(UNIX_EPOCH) { + Ok(duration_since_epoch) => duration_since_epoch, + Err(_) => return Err(S::Error::custom("SystemTime must be later than UNIX_EPOCH")), + }; + let mut state = try!(serializer.serialize_struct("SystemTime", 2)); + try!(state.serialize_field("secs_since_epoch", &duration_since_epoch.as_secs())); + try!(state.serialize_field("nanos_since_epoch", &duration_since_epoch.subsec_nanos())); + state.end() + } +} + +//////////////////////////////////////////////////////////////////////////////// + +/// Serialize a value that implements `Display` as a string, when that string is +/// statically known to never have more than a constant `MAX_LEN` bytes. +/// +/// Panics if the `Display` impl tries to write more than `MAX_LEN` bytes. +#[cfg(feature = "std")] +macro_rules! serialize_display_bounded_length { + ($value:expr, $max:expr, $serializer:expr) => {{ + let mut buffer = [0u8; $max]; + let remaining_len = { + let mut remaining = &mut buffer[..]; + write!(remaining, "{}", $value).unwrap(); + remaining.len() + }; + let written_len = buffer.len() - remaining_len; + let written = &buffer[..written_len]; + + // write! only provides fmt::Formatter to Display implementations, which + // has methods write_str and write_char but no method to write arbitrary + // bytes. Therefore `written` must be valid UTF-8. + let written_str = str::from_utf8(written).expect("must be valid UTF-8"); + $serializer.serialize_str(written_str) + }}; +} + +#[cfg(feature = "std")] +impl Serialize for net::IpAddr { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + match *self { + net::IpAddr::V4(ref a) => a.serialize(serializer), + net::IpAddr::V6(ref a) => a.serialize(serializer), + } + } else { + match *self { + net::IpAddr::V4(ref a) => { + serializer.serialize_newtype_variant("IpAddr", 0, "V4", a) + } + net::IpAddr::V6(ref a) => { + serializer.serialize_newtype_variant("IpAddr", 1, "V6", a) + } + } + } + } +} + +#[cfg(feature = "std")] +const DEC_DIGITS_LUT: &'static [u8] = b"\ + 0001020304050607080910111213141516171819\ + 2021222324252627282930313233343536373839\ + 4041424344454647484950515253545556575859\ + 6061626364656667686970717273747576777879\ + 8081828384858687888990919293949596979899"; + +#[cfg(feature = "std")] +#[inline] +fn format_u8(mut n: u8, out: &mut [u8]) -> usize { + if n >= 100 { + let d1 = ((n % 100) << 1) as usize; + n /= 100; + out[0] = b'0' + n; + out[1] = DEC_DIGITS_LUT[d1]; + out[2] = DEC_DIGITS_LUT[d1 + 1]; + 3 + } else if n >= 10 { + let d1 = (n << 1) as usize; + out[0] = DEC_DIGITS_LUT[d1]; + out[1] = DEC_DIGITS_LUT[d1 + 1]; + 2 + } else { + out[0] = b'0' + n; + 1 + } +} + +#[cfg(feature = "std")] +#[test] +fn test_format_u8() { + let mut i = 0u8; + + loop { + let mut buf = [0u8; 3]; + let written = format_u8(i, &mut buf); + assert_eq!(i.to_string().as_bytes(), &buf[..written]); + + match i.checked_add(1) { + Some(next) => i = next, + None => break, + } + } +} + +#[cfg(feature = "std")] +impl Serialize for net::Ipv4Addr { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + const MAX_LEN: usize = 15; + debug_assert_eq!(MAX_LEN, "101.102.103.104".len()); + let mut buf = [b'.'; MAX_LEN]; + let mut written = format_u8(self.octets()[0], &mut buf); + for oct in &self.octets()[1..] { + // Skip over delimiters that we initialized buf with + written += format_u8(*oct, &mut buf[written + 1..]) + 1; + } + // Safety: We've only written ASCII bytes to the buffer, so it is valid UTF-8 + let buf = unsafe { str::from_utf8_unchecked(&buf[..written]) }; + serializer.serialize_str(buf) + } else { + self.octets().serialize(serializer) + } + } +} + +#[cfg(feature = "std")] +impl Serialize for net::Ipv6Addr { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + const MAX_LEN: usize = 39; + debug_assert_eq!(MAX_LEN, "1001:1002:1003:1004:1005:1006:1007:1008".len()); + serialize_display_bounded_length!(self, MAX_LEN, serializer) + } else { + self.octets().serialize(serializer) + } + } +} + +#[cfg(feature = "std")] +impl Serialize for net::SocketAddr { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + match *self { + net::SocketAddr::V4(ref addr) => addr.serialize(serializer), + net::SocketAddr::V6(ref addr) => addr.serialize(serializer), + } + } else { + match *self { + net::SocketAddr::V4(ref addr) => { + serializer.serialize_newtype_variant("SocketAddr", 0, "V4", addr) + } + net::SocketAddr::V6(ref addr) => { + serializer.serialize_newtype_variant("SocketAddr", 1, "V6", addr) + } + } + } + } +} + +#[cfg(feature = "std")] +impl Serialize for net::SocketAddrV4 { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + const MAX_LEN: usize = 21; + debug_assert_eq!(MAX_LEN, "101.102.103.104:65000".len()); + serialize_display_bounded_length!(self, MAX_LEN, serializer) + } else { + (self.ip(), self.port()).serialize(serializer) + } + } +} + +#[cfg(feature = "std")] +impl Serialize for net::SocketAddrV6 { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + if serializer.is_human_readable() { + const MAX_LEN: usize = 58; + debug_assert_eq!( + MAX_LEN, + "[1001:1002:1003:1004:1005:1006:1007:1008%4294967295]:65000".len() + ); + serialize_display_bounded_length!(self, MAX_LEN, serializer) + } else { + (self.ip(), self.port()).serialize(serializer) + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +impl Serialize for Path { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + match self.to_str() { + Some(s) => s.serialize(serializer), + None => Err(Error::custom("path contains invalid UTF-8 characters")), + } + } +} + +#[cfg(feature = "std")] +impl Serialize for PathBuf { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.as_path().serialize(serializer) + } +} + +#[cfg(all(feature = "std", any(unix, windows)))] +impl Serialize for OsStr { + #[cfg(unix)] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use std::os::unix::ffi::OsStrExt; + serializer.serialize_newtype_variant("OsString", 0, "Unix", self.as_bytes()) + } + + #[cfg(windows)] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + use std::os::windows::ffi::OsStrExt; + let val = self.encode_wide().collect::>(); + serializer.serialize_newtype_variant("OsString", 1, "Windows", &val) + } +} + +#[cfg(all(feature = "std", any(unix, windows)))] +impl Serialize for OsString { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.as_os_str().serialize(serializer) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +impl Serialize for Wrapping +where + T: Serialize, +{ + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.0.serialize(serializer) + } +} + +#[cfg(not(no_core_reverse))] +impl Serialize for Reverse +where + T: Serialize, +{ + #[inline] + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + self.0.serialize(serializer) + } +} + +//////////////////////////////////////////////////////////////////////////////// + +#[cfg(all(feature = "std", not(no_std_atomic)))] +macro_rules! atomic_impl { + ($($ty:ident $size:expr)*) => { + $( + #[cfg(any(no_target_has_atomic, target_has_atomic = $size))] + impl Serialize for $ty { + fn serialize(&self, serializer: S) -> Result + where + S: Serializer, + { + // Matches the atomic ordering used in libcore for the Debug impl + self.load(Ordering::Relaxed).serialize(serializer) + } + } + )* + } +} + +#[cfg(all(feature = "std", not(no_std_atomic)))] +atomic_impl! { + AtomicBool "8" + AtomicI8 "8" + AtomicI16 "16" + AtomicI32 "32" + AtomicIsize "ptr" + AtomicU8 "8" + AtomicU16 "16" + AtomicU32 "32" + AtomicUsize "ptr" +} + +#[cfg(all(feature = "std", not(no_std_atomic64)))] +atomic_impl! { + AtomicI64 "64" + AtomicU64 "64" +} diff --git a/rust/serde/ser/impossible.rs b/rust/serde/ser/impossible.rs new file mode 100644 index 00000000000000..e25270ba70c11a --- /dev/null +++ b/rust/serde/ser/impossible.rs @@ -0,0 +1,218 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! This module contains `Impossible` serializer and its implementations. + +use lib::*; + +use ser::{ + self, Serialize, SerializeMap, SerializeSeq, SerializeStruct, SerializeStructVariant, + SerializeTuple, SerializeTupleStruct, SerializeTupleVariant, +}; + +/// Helper type for implementing a `Serializer` that does not support +/// serializing one of the compound types. +/// +/// This type cannot be instantiated, but implements every one of the traits +/// corresponding to the [`Serializer`] compound types: [`SerializeSeq`], +/// [`SerializeTuple`], [`SerializeTupleStruct`], [`SerializeTupleVariant`], +/// [`SerializeMap`], [`SerializeStruct`], and [`SerializeStructVariant`]. +/// +/// ```edition2018 +/// # use serde::ser::{Serializer, Impossible}; +/// # use serde::__private::doc::Error; +/// # +/// # struct MySerializer; +/// # +/// impl Serializer for MySerializer { +/// type Ok = (); +/// type Error = Error; +/// +/// type SerializeSeq = Impossible<(), Error>; +/// /* other associated types */ +/// +/// /// This data format does not support serializing sequences. +/// fn serialize_seq(self, +/// len: Option) +/// -> Result { +/// // Given Impossible cannot be instantiated, the only +/// // thing we can do here is to return an error. +/// # stringify! { +/// Err(...) +/// # }; +/// # unimplemented!() +/// } +/// +/// /* other Serializer methods */ +/// # serde::__serialize_unimplemented! { +/// # bool i8 i16 i32 i64 u8 u16 u32 u64 f32 f64 char str bytes none some +/// # unit unit_struct unit_variant newtype_struct newtype_variant +/// # tuple tuple_struct tuple_variant map struct struct_variant +/// # } +/// } +/// ``` +/// +/// [`Serializer`]: trait.Serializer.html +/// [`SerializeSeq`]: trait.SerializeSeq.html +/// [`SerializeTuple`]: trait.SerializeTuple.html +/// [`SerializeTupleStruct`]: trait.SerializeTupleStruct.html +/// [`SerializeTupleVariant`]: trait.SerializeTupleVariant.html +/// [`SerializeMap`]: trait.SerializeMap.html +/// [`SerializeStruct`]: trait.SerializeStruct.html +/// [`SerializeStructVariant`]: trait.SerializeStructVariant.html +pub struct Impossible { + void: Void, + ok: PhantomData, + error: PhantomData, +} + +enum Void {} + +impl SerializeSeq for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_element(&mut self, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeTuple for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_element(&mut self, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeTupleStruct for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_field(&mut self, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeTupleVariant for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_field(&mut self, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeMap for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_key(&mut self, key: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = key; + match self.void {} + } + + fn serialize_value(&mut self, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeStruct for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_field(&mut self, key: &'static str, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = key; + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} + +impl SerializeStructVariant for Impossible +where + Error: ser::Error, +{ + type Ok = Ok; + type Error = Error; + + fn serialize_field(&mut self, key: &'static str, value: &T) -> Result<(), Error> + where + T: Serialize, + { + let _ = key; + let _ = value; + match self.void {} + } + + fn end(self) -> Result { + match self.void {} + } +} diff --git a/rust/serde/ser/mod.rs b/rust/serde/ser/mod.rs new file mode 100644 index 00000000000000..9a5e47d8c5c8c3 --- /dev/null +++ b/rust/serde/ser/mod.rs @@ -0,0 +1,1992 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Generic data structure serialization framework. +//! +//! The two most important traits in this module are [`Serialize`] and +//! [`Serializer`]. +//! +//! - **A type that implements `Serialize` is a data structure** that can be +//! serialized to any data format supported by Serde, and conversely +//! - **A type that implements `Serializer` is a data format** that can +//! serialize any data structure supported by Serde. +//! +//! # The Serialize trait +//! +//! Serde provides [`Serialize`] implementations for many Rust primitive and +//! standard library types. The complete list is below. All of these can be +//! serialized using Serde out of the box. +//! +//! Additionally, Serde provides a procedural macro called [`serde_derive`] to +//! automatically generate [`Serialize`] implementations for structs and enums +//! in your program. See the [derive section of the manual] for how to use this. +//! +//! In rare cases it may be necessary to implement [`Serialize`] manually for +//! some type in your program. See the [Implementing `Serialize`] section of the +//! manual for more about this. +//! +//! Third-party crates may provide [`Serialize`] implementations for types that +//! they expose. For example the [`linked-hash-map`] crate provides a +//! [`LinkedHashMap`] type that is serializable by Serde because the crate +//! provides an implementation of [`Serialize`] for it. +//! +//! # The Serializer trait +//! +//! [`Serializer`] implementations are provided by third-party crates, for +//! example [`serde_json`], [`serde_yaml`] and [`postcard`]. +//! +//! A partial list of well-maintained formats is given on the [Serde +//! website][data formats]. +//! +//! # Implementations of Serialize provided by Serde +//! +//! - **Primitive types**: +//! - bool +//! - i8, i16, i32, i64, i128, isize +//! - u8, u16, u32, u64, u128, usize +//! - f32, f64 +//! - char +//! - str +//! - &T and &mut T +//! - **Compound types**: +//! - \[T\] +//! - \[T; 0\] through \[T; 32\] +//! - tuples up to size 16 +//! - **Common standard library types**: +//! - String +//! - Option\ +//! - Result\ +//! - PhantomData\ +//! - **Wrapper types**: +//! - Box\ +//! - Cow\<'a, T\> +//! - Cell\ +//! - RefCell\ +//! - Mutex\ +//! - RwLock\ +//! - Rc\ *(if* features = ["rc"] *is enabled)* +//! - Arc\ *(if* features = ["rc"] *is enabled)* +//! - **Collection types**: +//! - BTreeMap\ +//! - BTreeSet\ +//! - BinaryHeap\ +//! - HashMap\ +//! - HashSet\ +//! - LinkedList\ +//! - VecDeque\ +//! - Vec\ +//! - **FFI types**: +//! - CStr +//! - CString +//! - OsStr +//! - OsString +//! - **Miscellaneous standard library types**: +//! - Duration +//! - SystemTime +//! - Path +//! - PathBuf +//! - Range\ +//! - RangeInclusive\ +//! - Bound\ +//! - num::NonZero* +//! - `!` *(unstable)* +//! - **Net types**: +//! - IpAddr +//! - Ipv4Addr +//! - Ipv6Addr +//! - SocketAddr +//! - SocketAddrV4 +//! - SocketAddrV6 +//! +//! [Implementing `Serialize`]: https://serde.rs/impl-serialize.html +//! [`LinkedHashMap`]: https://docs.rs/linked-hash-map/*/linked_hash_map/struct.LinkedHashMap.html +//! [`Serialize`]: ../trait.Serialize.html +//! [`Serializer`]: ../trait.Serializer.html +//! [`postcard`]: https://github.com/jamesmunns/postcard +//! [`linked-hash-map`]: https://crates.io/crates/linked-hash-map +//! [`serde_derive`]: https://crates.io/crates/serde_derive +//! [`serde_json`]: https://github.com/serde-rs/json +//! [`serde_yaml`]: https://github.com/dtolnay/serde-yaml +//! [derive section of the manual]: https://serde.rs/derive.html +//! [data formats]: https://serde.rs/#data-formats + +use lib::*; + +mod fmt; +mod impls; +mod impossible; + +pub use self::impossible::Impossible; + +#[cfg(all(feature = "unstable", not(feature = "std")))] +#[doc(inline)] +pub use core::error::Error as StdError; +#[cfg(feature = "std")] +#[doc(no_inline)] +pub use std::error::Error as StdError; +#[cfg(not(any(feature = "std", feature = "unstable")))] +#[doc(no_inline)] +pub use std_error::Error as StdError; + +//////////////////////////////////////////////////////////////////////////////// + +macro_rules! declare_error_trait { + (Error: Sized $(+ $($supertrait:ident)::+)*) => { + /// Trait used by `Serialize` implementations to generically construct + /// errors belonging to the `Serializer` against which they are + /// currently running. + /// + /// # Example implementation + /// + /// The [example data format] presented on the website shows an error + /// type appropriate for a basic JSON data format. + /// + /// [example data format]: https://serde.rs/data-format.html + pub trait Error: Sized $(+ $($supertrait)::+)* { + /// Used when a [`Serialize`] implementation encounters any error + /// while serializing a type. + /// + /// The message should not be capitalized and should not end with a + /// period. + /// + /// For example, a filesystem [`Path`] may refuse to serialize + /// itself if it contains invalid UTF-8 data. + /// + /// ```edition2018 + /// # struct Path; + /// # + /// # impl Path { + /// # fn to_str(&self) -> Option<&str> { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::ser::{self, Serialize, Serializer}; + /// + /// impl Serialize for Path { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match self.to_str() { + /// Some(s) => serializer.serialize_str(s), + /// None => Err(ser::Error::custom("path contains invalid UTF-8 characters")), + /// } + /// } + /// } + /// ``` + /// + /// [`Path`]: https://doc.rust-lang.org/std/path/struct.Path.html + /// [`Serialize`]: ../trait.Serialize.html + fn custom(msg: T) -> Self + where + T: Display; + } + } +} + +#[cfg(feature = "std")] +declare_error_trait!(Error: Sized + StdError); + +#[cfg(not(feature = "std"))] +declare_error_trait!(Error: Sized + Debug + Display); + +//////////////////////////////////////////////////////////////////////////////// + +/// A **data structure** that can be serialized into any data format supported +/// by Serde. +/// +/// Serde provides `Serialize` implementations for many Rust primitive and +/// standard library types. The complete list is [here][crate::ser]. All of +/// these can be serialized using Serde out of the box. +/// +/// Additionally, Serde provides a procedural macro called [`serde_derive`] to +/// automatically generate `Serialize` implementations for structs and enums in +/// your program. See the [derive section of the manual] for how to use this. +/// +/// In rare cases it may be necessary to implement `Serialize` manually for some +/// type in your program. See the [Implementing `Serialize`] section of the +/// manual for more about this. +/// +/// Third-party crates may provide `Serialize` implementations for types that +/// they expose. For example the [`linked-hash-map`] crate provides a +/// [`LinkedHashMap`] type that is serializable by Serde because the crate +/// provides an implementation of `Serialize` for it. +/// +/// [Implementing `Serialize`]: https://serde.rs/impl-serialize.html +/// [`LinkedHashMap`]: https://docs.rs/linked-hash-map/*/linked_hash_map/struct.LinkedHashMap.html +/// [`linked-hash-map`]: https://crates.io/crates/linked-hash-map +/// [`serde_derive`]: https://crates.io/crates/serde_derive +/// [derive section of the manual]: https://serde.rs/derive.html +pub trait Serialize { + /// Serialize this value into the given Serde serializer. + /// + /// See the [Implementing `Serialize`] section of the manual for more + /// information about how to implement this method. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeStruct, Serializer}; + /// + /// struct Person { + /// name: String, + /// age: u8, + /// phones: Vec, + /// } + /// + /// // This is what #[derive(Serialize)] would generate. + /// impl Serialize for Person { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut s = serializer.serialize_struct("Person", 3)?; + /// s.serialize_field("name", &self.name)?; + /// s.serialize_field("age", &self.age)?; + /// s.serialize_field("phones", &self.phones)?; + /// s.end() + /// } + /// } + /// ``` + /// + /// [Implementing `Serialize`]: https://serde.rs/impl-serialize.html + fn serialize(&self, serializer: S) -> Result + where + S: Serializer; +} + +//////////////////////////////////////////////////////////////////////////////// + +/// A **data format** that can serialize any data structure supported by Serde. +/// +/// The role of this trait is to define the serialization half of the [Serde +/// data model], which is a way to categorize every Rust data structure into one +/// of 29 possible types. Each method of the `Serializer` trait corresponds to +/// one of the types of the data model. +/// +/// Implementations of `Serialize` map themselves into this data model by +/// invoking exactly one of the `Serializer` methods. +/// +/// The types that make up the Serde data model are: +/// +/// - **14 primitive types** +/// - bool +/// - i8, i16, i32, i64, i128 +/// - u8, u16, u32, u64, u128 +/// - f32, f64 +/// - char +/// - **string** +/// - UTF-8 bytes with a length and no null terminator. +/// - When serializing, all strings are handled equally. When deserializing, +/// there are three flavors of strings: transient, owned, and borrowed. +/// - **byte array** - \[u8\] +/// - Similar to strings, during deserialization byte arrays can be +/// transient, owned, or borrowed. +/// - **option** +/// - Either none or some value. +/// - **unit** +/// - The type of `()` in Rust. It represents an anonymous value containing +/// no data. +/// - **unit_struct** +/// - For example `struct Unit` or `PhantomData`. It represents a named +/// value containing no data. +/// - **unit_variant** +/// - For example the `E::A` and `E::B` in `enum E { A, B }`. +/// - **newtype_struct** +/// - For example `struct Millimeters(u8)`. +/// - **newtype_variant** +/// - For example the `E::N` in `enum E { N(u8) }`. +/// - **seq** +/// - A variably sized heterogeneous sequence of values, for example +/// `Vec` or `HashSet`. When serializing, the length may or may not +/// be known before iterating through all the data. When deserializing, +/// the length is determined by looking at the serialized data. +/// - **tuple** +/// - A statically sized heterogeneous sequence of values for which the +/// length will be known at deserialization time without looking at the +/// serialized data, for example `(u8,)` or `(String, u64, Vec)` or +/// `[u64; 10]`. +/// - **tuple_struct** +/// - A named tuple, for example `struct Rgb(u8, u8, u8)`. +/// - **tuple_variant** +/// - For example the `E::T` in `enum E { T(u8, u8) }`. +/// - **map** +/// - A heterogeneous key-value pairing, for example `BTreeMap`. +/// - **struct** +/// - A heterogeneous key-value pairing in which the keys are strings and +/// will be known at deserialization time without looking at the +/// serialized data, for example `struct S { r: u8, g: u8, b: u8 }`. +/// - **struct_variant** +/// - For example the `E::S` in `enum E { S { r: u8, g: u8, b: u8 } }`. +/// +/// Many Serde serializers produce text or binary data as output, for example +/// JSON or Postcard. This is not a requirement of the `Serializer` trait, and +/// there are serializers that do not produce text or binary output. One example +/// is the `serde_json::value::Serializer` (distinct from the main `serde_json` +/// serializer) that produces a `serde_json::Value` data structure in memory as +/// output. +/// +/// [Serde data model]: https://serde.rs/data-model.html +/// +/// # Example implementation +/// +/// The [example data format] presented on the website contains example code for +/// a basic JSON `Serializer`. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait Serializer: Sized { + /// The output type produced by this `Serializer` during successful + /// serialization. Most serializers that produce text or binary output + /// should set `Ok = ()` and serialize into an [`io::Write`] or buffer + /// contained within the `Serializer` instance. Serializers that build + /// in-memory data structures may be simplified by using `Ok` to propagate + /// the data structure around. + /// + /// [`io::Write`]: https://doc.rust-lang.org/std/io/trait.Write.html + type Ok; + + /// The error type when some error occurs during serialization. + type Error: Error; + + /// Type returned from [`serialize_seq`] for serializing the content of the + /// sequence. + /// + /// [`serialize_seq`]: #tymethod.serialize_seq + type SerializeSeq: SerializeSeq; + + /// Type returned from [`serialize_tuple`] for serializing the content of + /// the tuple. + /// + /// [`serialize_tuple`]: #tymethod.serialize_tuple + type SerializeTuple: SerializeTuple; + + /// Type returned from [`serialize_tuple_struct`] for serializing the + /// content of the tuple struct. + /// + /// [`serialize_tuple_struct`]: #tymethod.serialize_tuple_struct + type SerializeTupleStruct: SerializeTupleStruct; + + /// Type returned from [`serialize_tuple_variant`] for serializing the + /// content of the tuple variant. + /// + /// [`serialize_tuple_variant`]: #tymethod.serialize_tuple_variant + type SerializeTupleVariant: SerializeTupleVariant; + + /// Type returned from [`serialize_map`] for serializing the content of the + /// map. + /// + /// [`serialize_map`]: #tymethod.serialize_map + type SerializeMap: SerializeMap; + + /// Type returned from [`serialize_struct`] for serializing the content of + /// the struct. + /// + /// [`serialize_struct`]: #tymethod.serialize_struct + type SerializeStruct: SerializeStruct; + + /// Type returned from [`serialize_struct_variant`] for serializing the + /// content of the struct variant. + /// + /// [`serialize_struct_variant`]: #tymethod.serialize_struct_variant + type SerializeStructVariant: SerializeStructVariant; + + /// Serialize a `bool` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for bool { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_bool(*self) + /// } + /// } + /// ``` + fn serialize_bool(self, v: bool) -> Result; + + /// Serialize an `i8` value. + /// + /// If the format does not differentiate between `i8` and `i64`, a + /// reasonable implementation would be to cast the value to `i64` and + /// forward to `serialize_i64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for i8 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_i8(*self) + /// } + /// } + /// ``` + fn serialize_i8(self, v: i8) -> Result; + + /// Serialize an `i16` value. + /// + /// If the format does not differentiate between `i16` and `i64`, a + /// reasonable implementation would be to cast the value to `i64` and + /// forward to `serialize_i64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for i16 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_i16(*self) + /// } + /// } + /// ``` + fn serialize_i16(self, v: i16) -> Result; + + /// Serialize an `i32` value. + /// + /// If the format does not differentiate between `i32` and `i64`, a + /// reasonable implementation would be to cast the value to `i64` and + /// forward to `serialize_i64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for i32 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_i32(*self) + /// } + /// } + /// ``` + fn serialize_i32(self, v: i32) -> Result; + + /// Serialize an `i64` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for i64 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_i64(*self) + /// } + /// } + /// ``` + fn serialize_i64(self, v: i64) -> Result; + + serde_if_integer128! { + /// Serialize an `i128` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for i128 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_i128(*self) + /// } + /// } + /// ``` + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default behavior unconditionally returns an error. + fn serialize_i128(self, v: i128) -> Result { + let _ = v; + Err(Error::custom("i128 is not supported")) + } + } + + /// Serialize a `u8` value. + /// + /// If the format does not differentiate between `u8` and `u64`, a + /// reasonable implementation would be to cast the value to `u64` and + /// forward to `serialize_u64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for u8 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_u8(*self) + /// } + /// } + /// ``` + fn serialize_u8(self, v: u8) -> Result; + + /// Serialize a `u16` value. + /// + /// If the format does not differentiate between `u16` and `u64`, a + /// reasonable implementation would be to cast the value to `u64` and + /// forward to `serialize_u64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for u16 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_u16(*self) + /// } + /// } + /// ``` + fn serialize_u16(self, v: u16) -> Result; + + /// Serialize a `u32` value. + /// + /// If the format does not differentiate between `u32` and `u64`, a + /// reasonable implementation would be to cast the value to `u64` and + /// forward to `serialize_u64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for u32 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_u32(*self) + /// } + /// } + /// ``` + fn serialize_u32(self, v: u32) -> Result; + + /// Serialize a `u64` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for u64 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_u64(*self) + /// } + /// } + /// ``` + fn serialize_u64(self, v: u64) -> Result; + + serde_if_integer128! { + /// Serialize a `u128` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for u128 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_u128(*self) + /// } + /// } + /// ``` + /// + /// This method is available only on Rust compiler versions >=1.26. The + /// default behavior unconditionally returns an error. + fn serialize_u128(self, v: u128) -> Result { + let _ = v; + Err(Error::custom("u128 is not supported")) + } + } + + /// Serialize an `f32` value. + /// + /// If the format does not differentiate between `f32` and `f64`, a + /// reasonable implementation would be to cast the value to `f64` and + /// forward to `serialize_f64`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for f32 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_f32(*self) + /// } + /// } + /// ``` + #[cfg(not(no_fp_fmt_parse))] + fn serialize_f32(self, v: f32) -> Result; + + /// Serialize an `f64` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for f64 { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_f64(*self) + /// } + /// } + /// ``` + #[cfg(not(no_fp_fmt_parse))] + fn serialize_f64(self, v: f64) -> Result; + + /// Serialize a character. + /// + /// If the format does not support characters, it is reasonable to serialize + /// it as a single element `str` or a `u32`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for char { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_char(*self) + /// } + /// } + /// ``` + fn serialize_char(self, v: char) -> Result; + + /// Serialize a `&str`. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for str { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_str(self) + /// } + /// } + /// ``` + fn serialize_str(self, v: &str) -> Result; + + /// Serialize a chunk of raw byte data. + /// + /// Enables serializers to serialize byte slices more compactly or more + /// efficiently than other types of slices. If no efficient implementation + /// is available, a reasonable implementation would be to forward to + /// `serialize_seq`. If forwarded, the implementation looks usually just + /// like this: + /// + /// ```edition2018 + /// # use serde::ser::{Serializer, SerializeSeq}; + /// # use serde::__private::doc::Error; + /// # + /// # struct MySerializer; + /// # + /// # impl Serializer for MySerializer { + /// # type Ok = (); + /// # type Error = Error; + /// # + /// fn serialize_bytes(self, v: &[u8]) -> Result { + /// let mut seq = self.serialize_seq(Some(v.len()))?; + /// for b in v { + /// seq.serialize_element(b)?; + /// } + /// seq.end() + /// } + /// # + /// # serde::__serialize_unimplemented! { + /// # bool i8 i16 i32 i64 u8 u16 u32 u64 f32 f64 char str none some + /// # unit unit_struct unit_variant newtype_struct newtype_variant + /// # seq tuple tuple_struct tuple_variant map struct struct_variant + /// # } + /// # } + /// ``` + fn serialize_bytes(self, v: &[u8]) -> Result; + + /// Serialize a [`None`] value. + /// + /// ```edition2018 + /// # use serde::{Serialize, Serializer}; + /// # + /// # enum Option { + /// # Some(T), + /// # None, + /// # } + /// # + /// # use self::Option::{Some, None}; + /// # + /// impl Serialize for Option + /// where + /// T: Serialize, + /// { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// Some(ref value) => serializer.serialize_some(value), + /// None => serializer.serialize_none(), + /// } + /// } + /// } + /// # + /// # fn main() {} + /// ``` + /// + /// [`None`]: https://doc.rust-lang.org/std/option/enum.Option.html#variant.None + fn serialize_none(self) -> Result; + + /// Serialize a [`Some(T)`] value. + /// + /// ```edition2018 + /// # use serde::{Serialize, Serializer}; + /// # + /// # enum Option { + /// # Some(T), + /// # None, + /// # } + /// # + /// # use self::Option::{Some, None}; + /// # + /// impl Serialize for Option + /// where + /// T: Serialize, + /// { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// Some(ref value) => serializer.serialize_some(value), + /// None => serializer.serialize_none(), + /// } + /// } + /// } + /// # + /// # fn main() {} + /// ``` + /// + /// [`Some(T)`]: https://doc.rust-lang.org/std/option/enum.Option.html#variant.Some + fn serialize_some(self, value: &T) -> Result + where + T: Serialize; + + /// Serialize a `()` value. + /// + /// ```edition2018 + /// # use serde::Serializer; + /// # + /// # serde::__private_serialize!(); + /// # + /// impl Serialize for () { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_unit() + /// } + /// } + /// ``` + fn serialize_unit(self) -> Result; + + /// Serialize a unit struct like `struct Unit` or `PhantomData`. + /// + /// A reasonable implementation would be to forward to `serialize_unit`. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// + /// struct Nothing; + /// + /// impl Serialize for Nothing { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_unit_struct("Nothing") + /// } + /// } + /// ``` + fn serialize_unit_struct(self, name: &'static str) -> Result; + + /// Serialize a unit variant like `E::A` in `enum E { A, B }`. + /// + /// The `name` is the name of the enum, the `variant_index` is the index of + /// this variant within the enum, and the `variant` is the name of the + /// variant. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// + /// enum E { + /// A, + /// B, + /// } + /// + /// impl Serialize for E { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// E::A => serializer.serialize_unit_variant("E", 0, "A"), + /// E::B => serializer.serialize_unit_variant("E", 1, "B"), + /// } + /// } + /// } + /// ``` + fn serialize_unit_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + ) -> Result; + + /// Serialize a newtype struct like `struct Millimeters(u8)`. + /// + /// Serializers are encouraged to treat newtype structs as insignificant + /// wrappers around the data they contain. A reasonable implementation would + /// be to forward to `value.serialize(self)`. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// + /// struct Millimeters(u8); + /// + /// impl Serialize for Millimeters { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.serialize_newtype_struct("Millimeters", &self.0) + /// } + /// } + /// ``` + fn serialize_newtype_struct( + self, + name: &'static str, + value: &T, + ) -> Result + where + T: Serialize; + + /// Serialize a newtype variant like `E::N` in `enum E { N(u8) }`. + /// + /// The `name` is the name of the enum, the `variant_index` is the index of + /// this variant within the enum, and the `variant` is the name of the + /// variant. The `value` is the data contained within this newtype variant. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// + /// enum E { + /// M(String), + /// N(u8), + /// } + /// + /// impl Serialize for E { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// E::M(ref s) => serializer.serialize_newtype_variant("E", 0, "M", s), + /// E::N(n) => serializer.serialize_newtype_variant("E", 1, "N", &n), + /// } + /// } + /// } + /// ``` + fn serialize_newtype_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + value: &T, + ) -> Result + where + T: Serialize; + + /// Begin to serialize a variably sized sequence. This call must be + /// followed by zero or more calls to `serialize_element`, then a call to + /// `end`. + /// + /// The argument is the number of elements in the sequence, which may or may + /// not be computable before the sequence is iterated. Some serializers only + /// support sequences whose length is known up front. + /// + /// ```edition2018 + /// # use std::marker::PhantomData; + /// # + /// # struct Vec(PhantomData); + /// # + /// # impl Vec { + /// # fn len(&self) -> usize { + /// # unimplemented!() + /// # } + /// # } + /// # + /// # impl<'a, T> IntoIterator for &'a Vec { + /// # type Item = &'a T; + /// # type IntoIter = Box>; + /// # + /// # fn into_iter(self) -> Self::IntoIter { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::ser::{Serialize, Serializer, SerializeSeq}; + /// + /// impl Serialize for Vec + /// where + /// T: Serialize, + /// { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut seq = serializer.serialize_seq(Some(self.len()))?; + /// for element in self { + /// seq.serialize_element(element)?; + /// } + /// seq.end() + /// } + /// } + /// ``` + fn serialize_seq(self, len: Option) -> Result; + + /// Begin to serialize a statically sized sequence whose length will be + /// known at deserialization time without looking at the serialized data. + /// This call must be followed by zero or more calls to `serialize_element`, + /// then a call to `end`. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, Serializer, SerializeTuple}; + /// + /// # mod fool { + /// # trait Serialize {} + /// impl Serialize for (A, B, C) + /// # {} + /// # } + /// # + /// # struct Tuple3(A, B, C); + /// # + /// # impl Serialize for Tuple3 + /// where + /// A: Serialize, + /// B: Serialize, + /// C: Serialize, + /// { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut tup = serializer.serialize_tuple(3)?; + /// tup.serialize_element(&self.0)?; + /// tup.serialize_element(&self.1)?; + /// tup.serialize_element(&self.2)?; + /// tup.end() + /// } + /// } + /// ``` + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeTuple, Serializer}; + /// + /// const VRAM_SIZE: usize = 386; + /// struct Vram([u16; VRAM_SIZE]); + /// + /// impl Serialize for Vram { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut seq = serializer.serialize_tuple(VRAM_SIZE)?; + /// for element in &self.0[..] { + /// seq.serialize_element(element)?; + /// } + /// seq.end() + /// } + /// } + /// ``` + fn serialize_tuple(self, len: usize) -> Result; + + /// Begin to serialize a tuple struct like `struct Rgb(u8, u8, u8)`. This + /// call must be followed by zero or more calls to `serialize_field`, then a + /// call to `end`. + /// + /// The `name` is the name of the tuple struct and the `len` is the number + /// of data fields that will be serialized. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeTupleStruct, Serializer}; + /// + /// struct Rgb(u8, u8, u8); + /// + /// impl Serialize for Rgb { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut ts = serializer.serialize_tuple_struct("Rgb", 3)?; + /// ts.serialize_field(&self.0)?; + /// ts.serialize_field(&self.1)?; + /// ts.serialize_field(&self.2)?; + /// ts.end() + /// } + /// } + /// ``` + fn serialize_tuple_struct( + self, + name: &'static str, + len: usize, + ) -> Result; + + /// Begin to serialize a tuple variant like `E::T` in `enum E { T(u8, u8) + /// }`. This call must be followed by zero or more calls to + /// `serialize_field`, then a call to `end`. + /// + /// The `name` is the name of the enum, the `variant_index` is the index of + /// this variant within the enum, the `variant` is the name of the variant, + /// and the `len` is the number of data fields that will be serialized. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeTupleVariant, Serializer}; + /// + /// enum E { + /// T(u8, u8), + /// U(String, u32, u32), + /// } + /// + /// impl Serialize for E { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// E::T(ref a, ref b) => { + /// let mut tv = serializer.serialize_tuple_variant("E", 0, "T", 2)?; + /// tv.serialize_field(a)?; + /// tv.serialize_field(b)?; + /// tv.end() + /// } + /// E::U(ref a, ref b, ref c) => { + /// let mut tv = serializer.serialize_tuple_variant("E", 1, "U", 3)?; + /// tv.serialize_field(a)?; + /// tv.serialize_field(b)?; + /// tv.serialize_field(c)?; + /// tv.end() + /// } + /// } + /// } + /// } + /// ``` + fn serialize_tuple_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result; + + /// Begin to serialize a map. This call must be followed by zero or more + /// calls to `serialize_key` and `serialize_value`, then a call to `end`. + /// + /// The argument is the number of elements in the map, which may or may not + /// be computable before the map is iterated. Some serializers only support + /// maps whose length is known up front. + /// + /// ```edition2018 + /// # use std::marker::PhantomData; + /// # + /// # struct HashMap(PhantomData, PhantomData); + /// # + /// # impl HashMap { + /// # fn len(&self) -> usize { + /// # unimplemented!() + /// # } + /// # } + /// # + /// # impl<'a, K, V> IntoIterator for &'a HashMap { + /// # type Item = (&'a K, &'a V); + /// # type IntoIter = Box>; + /// # + /// # fn into_iter(self) -> Self::IntoIter { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::ser::{Serialize, Serializer, SerializeMap}; + /// + /// impl Serialize for HashMap + /// where + /// K: Serialize, + /// V: Serialize, + /// { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut map = serializer.serialize_map(Some(self.len()))?; + /// for (k, v) in self { + /// map.serialize_entry(k, v)?; + /// } + /// map.end() + /// } + /// } + /// ``` + fn serialize_map(self, len: Option) -> Result; + + /// Begin to serialize a struct like `struct Rgb { r: u8, g: u8, b: u8 }`. + /// This call must be followed by zero or more calls to `serialize_field`, + /// then a call to `end`. + /// + /// The `name` is the name of the struct and the `len` is the number of + /// data fields that will be serialized. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeStruct, Serializer}; + /// + /// struct Rgb { + /// r: u8, + /// g: u8, + /// b: u8, + /// } + /// + /// impl Serialize for Rgb { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// let mut rgb = serializer.serialize_struct("Rgb", 3)?; + /// rgb.serialize_field("r", &self.r)?; + /// rgb.serialize_field("g", &self.g)?; + /// rgb.serialize_field("b", &self.b)?; + /// rgb.end() + /// } + /// } + /// ``` + fn serialize_struct( + self, + name: &'static str, + len: usize, + ) -> Result; + + /// Begin to serialize a struct variant like `E::S` in `enum E { S { r: u8, + /// g: u8, b: u8 } }`. This call must be followed by zero or more calls to + /// `serialize_field`, then a call to `end`. + /// + /// The `name` is the name of the enum, the `variant_index` is the index of + /// this variant within the enum, the `variant` is the name of the variant, + /// and the `len` is the number of data fields that will be serialized. + /// + /// ```edition2018 + /// use serde::ser::{Serialize, SerializeStructVariant, Serializer}; + /// + /// enum E { + /// S { r: u8, g: u8, b: u8 }, + /// } + /// + /// impl Serialize for E { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// match *self { + /// E::S { + /// ref r, + /// ref g, + /// ref b, + /// } => { + /// let mut sv = serializer.serialize_struct_variant("E", 0, "S", 3)?; + /// sv.serialize_field("r", r)?; + /// sv.serialize_field("g", g)?; + /// sv.serialize_field("b", b)?; + /// sv.end() + /// } + /// } + /// } + /// } + /// ``` + fn serialize_struct_variant( + self, + name: &'static str, + variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result; + + /// Collect an iterator as a sequence. + /// + /// The default implementation serializes each item yielded by the iterator + /// using [`serialize_seq`]. Implementors should not need to override this + /// method. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// + /// struct SecretlyOneHigher { + /// data: Vec, + /// } + /// + /// impl Serialize for SecretlyOneHigher { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.collect_seq(self.data.iter().map(|x| x + 1)) + /// } + /// } + /// ``` + /// + /// [`serialize_seq`]: #tymethod.serialize_seq + fn collect_seq(self, iter: I) -> Result + where + I: IntoIterator, + ::Item: Serialize, + { + let iter = iter.into_iter(); + let mut serializer = try!(self.serialize_seq(iterator_len_hint(&iter))); + + #[cfg(not(no_iterator_try_fold))] + { + let mut iter = iter; + try!(iter.try_for_each(|item| serializer.serialize_element(&item))); + } + + #[cfg(no_iterator_try_fold)] + { + for item in iter { + try!(serializer.serialize_element(&item)); + } + } + + serializer.end() + } + + /// Collect an iterator as a map. + /// + /// The default implementation serializes each pair yielded by the iterator + /// using [`serialize_map`]. Implementors should not need to override this + /// method. + /// + /// ```edition2018 + /// use serde::{Serialize, Serializer}; + /// use std::collections::BTreeSet; + /// + /// struct MapToUnit { + /// keys: BTreeSet, + /// } + /// + /// // Serializes as a map in which the values are all unit. + /// impl Serialize for MapToUnit { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.collect_map(self.keys.iter().map(|k| (k, ()))) + /// } + /// } + /// ``` + /// + /// [`serialize_map`]: #tymethod.serialize_map + fn collect_map(self, iter: I) -> Result + where + K: Serialize, + V: Serialize, + I: IntoIterator, + { + let iter = iter.into_iter(); + let mut serializer = try!(self.serialize_map(iterator_len_hint(&iter))); + + #[cfg(not(no_iterator_try_fold))] + { + let mut iter = iter; + try!(iter.try_for_each(|(key, value)| serializer.serialize_entry(&key, &value))); + } + + #[cfg(no_iterator_try_fold)] + { + for (key, value) in iter { + try!(serializer.serialize_entry(&key, &value)); + } + } + + serializer.end() + } + + /// Serialize a string produced by an implementation of `Display`. + /// + /// The default implementation builds a heap-allocated [`String`] and + /// delegates to [`serialize_str`]. Serializers are encouraged to provide a + /// more efficient implementation if possible. + /// + /// ```edition2018 + /// # struct DateTime; + /// # + /// # impl DateTime { + /// # fn naive_local(&self) -> () { () } + /// # fn offset(&self) -> () { () } + /// # } + /// # + /// use serde::{Serialize, Serializer}; + /// + /// impl Serialize for DateTime { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.collect_str(&format_args!("{:?}{:?}", + /// self.naive_local(), + /// self.offset())) + /// } + /// } + /// ``` + /// + /// [`String`]: https://doc.rust-lang.org/std/string/struct.String.html + /// [`serialize_str`]: #tymethod.serialize_str + #[cfg(any(feature = "std", feature = "alloc"))] + fn collect_str(self, value: &T) -> Result + where + T: Display, + { + self.serialize_str(&value.to_string()) + } + + /// Serialize a string produced by an implementation of `Display`. + /// + /// Serializers that use `no_std` are required to provide an implementation + /// of this method. If no more sensible behavior is possible, the + /// implementation is expected to return an error. + /// + /// ```edition2018 + /// # struct DateTime; + /// # + /// # impl DateTime { + /// # fn naive_local(&self) -> () { () } + /// # fn offset(&self) -> () { () } + /// # } + /// # + /// use serde::{Serialize, Serializer}; + /// + /// impl Serialize for DateTime { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// serializer.collect_str(&format_args!("{:?}{:?}", + /// self.naive_local(), + /// self.offset())) + /// } + /// } + /// ``` + #[cfg(not(any(feature = "std", feature = "alloc")))] + fn collect_str(self, value: &T) -> Result + where + T: Display; + + /// Determine whether `Serialize` implementations should serialize in + /// human-readable form. + /// + /// Some types have a human-readable form that may be somewhat expensive to + /// construct, as well as a binary form that is compact and efficient. + /// Generally text-based formats like JSON and YAML will prefer to use the + /// human-readable one and binary formats like Postcard will prefer the + /// compact one. + /// + /// ```edition2018 + /// # use std::fmt::{self, Display}; + /// # + /// # struct Timestamp; + /// # + /// # impl Timestamp { + /// # fn seconds_since_epoch(&self) -> u64 { unimplemented!() } + /// # } + /// # + /// # impl Display for Timestamp { + /// # fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + /// # unimplemented!() + /// # } + /// # } + /// # + /// use serde::{Serialize, Serializer}; + /// + /// impl Serialize for Timestamp { + /// fn serialize(&self, serializer: S) -> Result + /// where + /// S: Serializer, + /// { + /// if serializer.is_human_readable() { + /// // Serialize to a human-readable string "2015-05-15T17:01:00Z". + /// self.to_string().serialize(serializer) + /// } else { + /// // Serialize to a compact binary representation. + /// self.seconds_since_epoch().serialize(serializer) + /// } + /// } + /// } + /// ``` + /// + /// The default implementation of this method returns `true`. Data formats + /// may override this to `false` to request a compact form for types that + /// support one. Note that modifying this method to change a format from + /// human-readable to compact or vice versa should be regarded as a breaking + /// change, as a value serialized in human-readable mode is not required to + /// deserialize from the same data in compact mode. + #[inline] + fn is_human_readable(&self) -> bool { + true + } +} + +/// Returned from `Serializer::serialize_seq`. +/// +/// # Example use +/// +/// ```edition2018 +/// # use std::marker::PhantomData; +/// # +/// # struct Vec(PhantomData); +/// # +/// # impl Vec { +/// # fn len(&self) -> usize { +/// # unimplemented!() +/// # } +/// # } +/// # +/// # impl<'a, T> IntoIterator for &'a Vec { +/// # type Item = &'a T; +/// # type IntoIter = Box>; +/// # fn into_iter(self) -> Self::IntoIter { +/// # unimplemented!() +/// # } +/// # } +/// # +/// use serde::ser::{Serialize, Serializer, SerializeSeq}; +/// +/// impl Serialize for Vec +/// where +/// T: Serialize, +/// { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut seq = serializer.serialize_seq(Some(self.len()))?; +/// for element in self { +/// seq.serialize_element(element)?; +/// } +/// seq.end() +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeSeq` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeSeq { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a sequence element. + fn serialize_element(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Finish serializing a sequence. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_tuple`. +/// +/// # Example use +/// +/// ```edition2018 +/// use serde::ser::{Serialize, Serializer, SerializeTuple}; +/// +/// # mod fool { +/// # trait Serialize {} +/// impl Serialize for (A, B, C) +/// # {} +/// # } +/// # +/// # struct Tuple3(A, B, C); +/// # +/// # impl Serialize for Tuple3 +/// where +/// A: Serialize, +/// B: Serialize, +/// C: Serialize, +/// { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut tup = serializer.serialize_tuple(3)?; +/// tup.serialize_element(&self.0)?; +/// tup.serialize_element(&self.1)?; +/// tup.serialize_element(&self.2)?; +/// tup.end() +/// } +/// } +/// ``` +/// +/// ```edition2018 +/// # use std::marker::PhantomData; +/// # +/// # struct Array(PhantomData); +/// # +/// # impl Array { +/// # fn len(&self) -> usize { +/// # unimplemented!() +/// # } +/// # } +/// # +/// # impl<'a, T> IntoIterator for &'a Array { +/// # type Item = &'a T; +/// # type IntoIter = Box>; +/// # fn into_iter(self) -> Self::IntoIter { +/// # unimplemented!() +/// # } +/// # } +/// # +/// use serde::ser::{Serialize, Serializer, SerializeTuple}; +/// +/// # mod fool { +/// # trait Serialize {} +/// impl Serialize for [T; 16] +/// # {} +/// # } +/// # +/// # impl Serialize for Array +/// where +/// T: Serialize, +/// { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut seq = serializer.serialize_tuple(16)?; +/// for element in self { +/// seq.serialize_element(element)?; +/// } +/// seq.end() +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeTuple` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeTuple { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a tuple element. + fn serialize_element(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Finish serializing a tuple. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_tuple_struct`. +/// +/// # Example use +/// +/// ```edition2018 +/// use serde::ser::{Serialize, SerializeTupleStruct, Serializer}; +/// +/// struct Rgb(u8, u8, u8); +/// +/// impl Serialize for Rgb { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut ts = serializer.serialize_tuple_struct("Rgb", 3)?; +/// ts.serialize_field(&self.0)?; +/// ts.serialize_field(&self.1)?; +/// ts.serialize_field(&self.2)?; +/// ts.end() +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeTupleStruct` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeTupleStruct { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a tuple struct field. + fn serialize_field(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Finish serializing a tuple struct. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_tuple_variant`. +/// +/// # Example use +/// +/// ```edition2018 +/// use serde::ser::{Serialize, SerializeTupleVariant, Serializer}; +/// +/// enum E { +/// T(u8, u8), +/// U(String, u32, u32), +/// } +/// +/// impl Serialize for E { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// match *self { +/// E::T(ref a, ref b) => { +/// let mut tv = serializer.serialize_tuple_variant("E", 0, "T", 2)?; +/// tv.serialize_field(a)?; +/// tv.serialize_field(b)?; +/// tv.end() +/// } +/// E::U(ref a, ref b, ref c) => { +/// let mut tv = serializer.serialize_tuple_variant("E", 1, "U", 3)?; +/// tv.serialize_field(a)?; +/// tv.serialize_field(b)?; +/// tv.serialize_field(c)?; +/// tv.end() +/// } +/// } +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeTupleVariant` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeTupleVariant { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a tuple variant field. + fn serialize_field(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Finish serializing a tuple variant. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_map`. +/// +/// # Example use +/// +/// ```edition2018 +/// # use std::marker::PhantomData; +/// # +/// # struct HashMap(PhantomData, PhantomData); +/// # +/// # impl HashMap { +/// # fn len(&self) -> usize { +/// # unimplemented!() +/// # } +/// # } +/// # +/// # impl<'a, K, V> IntoIterator for &'a HashMap { +/// # type Item = (&'a K, &'a V); +/// # type IntoIter = Box>; +/// # +/// # fn into_iter(self) -> Self::IntoIter { +/// # unimplemented!() +/// # } +/// # } +/// # +/// use serde::ser::{Serialize, Serializer, SerializeMap}; +/// +/// impl Serialize for HashMap +/// where +/// K: Serialize, +/// V: Serialize, +/// { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut map = serializer.serialize_map(Some(self.len()))?; +/// for (k, v) in self { +/// map.serialize_entry(k, v)?; +/// } +/// map.end() +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeMap` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeMap { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a map key. + /// + /// If possible, `Serialize` implementations are encouraged to use + /// `serialize_entry` instead as it may be implemented more efficiently in + /// some formats compared to a pair of calls to `serialize_key` and + /// `serialize_value`. + fn serialize_key(&mut self, key: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Serialize a map value. + /// + /// # Panics + /// + /// Calling `serialize_value` before `serialize_key` is incorrect and is + /// allowed to panic or produce bogus results. + fn serialize_value(&mut self, value: &T) -> Result<(), Self::Error> + where + T: Serialize; + + /// Serialize a map entry consisting of a key and a value. + /// + /// Some [`Serialize`] types are not able to hold a key and value in memory + /// at the same time so `SerializeMap` implementations are required to + /// support [`serialize_key`] and [`serialize_value`] individually. The + /// `serialize_entry` method allows serializers to optimize for the case + /// where key and value are both available. [`Serialize`] implementations + /// are encouraged to use `serialize_entry` if possible. + /// + /// The default implementation delegates to [`serialize_key`] and + /// [`serialize_value`]. This is appropriate for serializers that do not + /// care about performance or are not able to optimize `serialize_entry` any + /// better than this. + /// + /// [`Serialize`]: ../trait.Serialize.html + /// [`serialize_key`]: #tymethod.serialize_key + /// [`serialize_value`]: #tymethod.serialize_value + fn serialize_entry( + &mut self, + key: &K, + value: &V, + ) -> Result<(), Self::Error> + where + K: Serialize, + V: Serialize, + { + try!(self.serialize_key(key)); + self.serialize_value(value) + } + + /// Finish serializing a map. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_struct`. +/// +/// # Example use +/// +/// ```edition2018 +/// use serde::ser::{Serialize, SerializeStruct, Serializer}; +/// +/// struct Rgb { +/// r: u8, +/// g: u8, +/// b: u8, +/// } +/// +/// impl Serialize for Rgb { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// let mut rgb = serializer.serialize_struct("Rgb", 3)?; +/// rgb.serialize_field("r", &self.r)?; +/// rgb.serialize_field("g", &self.g)?; +/// rgb.serialize_field("b", &self.b)?; +/// rgb.end() +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeStruct` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeStruct { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a struct field. + fn serialize_field( + &mut self, + key: &'static str, + value: &T, + ) -> Result<(), Self::Error> + where + T: Serialize; + + /// Indicate that a struct field has been skipped. + #[inline] + fn skip_field(&mut self, key: &'static str) -> Result<(), Self::Error> { + let _ = key; + Ok(()) + } + + /// Finish serializing a struct. + fn end(self) -> Result; +} + +/// Returned from `Serializer::serialize_struct_variant`. +/// +/// # Example use +/// +/// ```edition2018 +/// use serde::ser::{Serialize, SerializeStructVariant, Serializer}; +/// +/// enum E { +/// S { r: u8, g: u8, b: u8 }, +/// } +/// +/// impl Serialize for E { +/// fn serialize(&self, serializer: S) -> Result +/// where +/// S: Serializer, +/// { +/// match *self { +/// E::S { +/// ref r, +/// ref g, +/// ref b, +/// } => { +/// let mut sv = serializer.serialize_struct_variant("E", 0, "S", 3)?; +/// sv.serialize_field("r", r)?; +/// sv.serialize_field("g", g)?; +/// sv.serialize_field("b", b)?; +/// sv.end() +/// } +/// } +/// } +/// } +/// ``` +/// +/// # Example implementation +/// +/// The [example data format] presented on the website demonstrates an +/// implementation of `SerializeStructVariant` for a basic JSON data format. +/// +/// [example data format]: https://serde.rs/data-format.html +pub trait SerializeStructVariant { + /// Must match the `Ok` type of our `Serializer`. + type Ok; + + /// Must match the `Error` type of our `Serializer`. + type Error: Error; + + /// Serialize a struct variant field. + fn serialize_field( + &mut self, + key: &'static str, + value: &T, + ) -> Result<(), Self::Error> + where + T: Serialize; + + /// Indicate that a struct variant field has been skipped. + #[inline] + fn skip_field(&mut self, key: &'static str) -> Result<(), Self::Error> { + let _ = key; + Ok(()) + } + + /// Finish serializing a struct variant. + fn end(self) -> Result; +} + +fn iterator_len_hint(iter: &I) -> Option +where + I: Iterator, +{ + match iter.size_hint() { + (lo, Some(hi)) if lo == hi => Some(lo), + _ => None, + } +} diff --git a/rust/serde/std_error.rs b/rust/serde/std_error.rs new file mode 100644 index 00000000000000..b730d38a1e07c0 --- /dev/null +++ b/rust/serde/std_error.rs @@ -0,0 +1,50 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use lib::{Debug, Display}; + +/// Either a re-export of std::error::Error or a new identical trait, depending +/// on whether Serde's "std" feature is enabled. +/// +/// Serde's error traits [`serde::ser::Error`] and [`serde::de::Error`] require +/// [`std::error::Error`] as a supertrait, but only when Serde is built with +/// "std" enabled. Data formats that don't care about no\_std support should +/// generally provide their error types with a `std::error::Error` impl +/// directly: +/// +/// ```edition2018 +/// #[derive(Debug)] +/// struct MySerError {...} +/// +/// impl serde::ser::Error for MySerError {...} +/// +/// impl std::fmt::Display for MySerError {...} +/// +/// // We don't support no_std! +/// impl std::error::Error for MySerError {} +/// ``` +/// +/// Data formats that *do* support no\_std may either have a "std" feature of +/// their own: +/// +/// ```toml +/// [features] +/// std = ["serde/std"] +/// ``` +/// +/// ```edition2018 +/// #[cfg(feature = "std")] +/// impl std::error::Error for MySerError {} +/// ``` +/// +/// ... or else provide the std Error impl unconditionally via Serde's +/// re-export: +/// +/// ```edition2018 +/// impl serde::ser::StdError for MySerError {} +/// ``` +pub trait Error: Debug + Display { + /// The underlying cause of this error, if any. + fn source(&self) -> Option<&(Error + 'static)> { + None + } +} diff --git a/rust/serde_derive/bound.rs b/rust/serde_derive/bound.rs new file mode 100644 index 00000000000000..dc6b5907ff8023 --- /dev/null +++ b/rust/serde_derive/bound.rs @@ -0,0 +1,408 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use std::collections::HashSet; + +use syn; +use syn::punctuated::{Pair, Punctuated}; + +use internals::ast::{Container, Data}; +use internals::{attr, ungroup}; + +use proc_macro2::Span; + +// Remove the default from every type parameter because in the generated impls +// they look like associated types: "error: associated type bindings are not +// allowed here". +pub fn without_defaults(generics: &syn::Generics) -> syn::Generics { + syn::Generics { + params: generics + .params + .iter() + .map(|param| match param { + syn::GenericParam::Type(param) => syn::GenericParam::Type(syn::TypeParam { + eq_token: None, + default: None, + ..param.clone() + }), + _ => param.clone(), + }) + .collect(), + ..generics.clone() + } +} + +pub fn with_where_predicates( + generics: &syn::Generics, + predicates: &[syn::WherePredicate], +) -> syn::Generics { + let mut generics = generics.clone(); + generics + .make_where_clause() + .predicates + .extend(predicates.iter().cloned()); + generics +} + +pub fn with_where_predicates_from_fields( + cont: &Container, + generics: &syn::Generics, + from_field: fn(&attr::Field) -> Option<&[syn::WherePredicate]>, +) -> syn::Generics { + let predicates = cont + .data + .all_fields() + .filter_map(|field| from_field(&field.attrs)) + .flat_map(<[syn::WherePredicate]>::to_vec); + + let mut generics = generics.clone(); + generics.make_where_clause().predicates.extend(predicates); + generics +} + +pub fn with_where_predicates_from_variants( + cont: &Container, + generics: &syn::Generics, + from_variant: fn(&attr::Variant) -> Option<&[syn::WherePredicate]>, +) -> syn::Generics { + let variants = match &cont.data { + Data::Enum(variants) => variants, + Data::Struct(_, _) => { + return generics.clone(); + } + }; + + let predicates = variants + .iter() + .filter_map(|variant| from_variant(&variant.attrs)) + .flat_map(<[syn::WherePredicate]>::to_vec); + + let mut generics = generics.clone(); + generics.make_where_clause().predicates.extend(predicates); + generics +} + +// Puts the given bound on any generic type parameters that are used in fields +// for which filter returns true. +// +// For example, the following struct needs the bound `A: Serialize, B: +// Serialize`. +// +// struct S<'b, A, B: 'b, C> { +// a: A, +// b: Option<&'b B> +// #[serde(skip_serializing)] +// c: C, +// } +pub fn with_bound( + cont: &Container, + generics: &syn::Generics, + filter: fn(&attr::Field, Option<&attr::Variant>) -> bool, + bound: &syn::Path, +) -> syn::Generics { + struct FindTyParams<'ast> { + // Set of all generic type parameters on the current struct (A, B, C in + // the example). Initialized up front. + all_type_params: HashSet, + + // Set of generic type parameters used in fields for which filter + // returns true (A and B in the example). Filled in as the visitor sees + // them. + relevant_type_params: HashSet, + + // Fields whose type is an associated type of one of the generic type + // parameters. + associated_type_usage: Vec<&'ast syn::TypePath>, + } + + impl<'ast> FindTyParams<'ast> { + fn visit_field(&mut self, field: &'ast syn::Field) { + if let syn::Type::Path(ty) = ungroup(&field.ty) { + if let Some(Pair::Punctuated(t, _)) = ty.path.segments.pairs().next() { + if self.all_type_params.contains(&t.ident) { + self.associated_type_usage.push(ty); + } + } + } + self.visit_type(&field.ty); + } + + fn visit_path(&mut self, path: &'ast syn::Path) { + if let Some(seg) = path.segments.last() { + if seg.ident == "PhantomData" { + // Hardcoded exception, because PhantomData implements + // Serialize and Deserialize whether or not T implements it. + return; + } + } + if path.leading_colon.is_none() && path.segments.len() == 1 { + let id = &path.segments[0].ident; + if self.all_type_params.contains(id) { + self.relevant_type_params.insert(id.clone()); + } + } + for segment in &path.segments { + self.visit_path_segment(segment); + } + } + + // Everything below is simply traversing the syntax tree. + + fn visit_type(&mut self, ty: &'ast syn::Type) { + match ty { + syn::Type::Array(ty) => self.visit_type(&ty.elem), + syn::Type::BareFn(ty) => { + for arg in &ty.inputs { + self.visit_type(&arg.ty); + } + self.visit_return_type(&ty.output); + } + syn::Type::Group(ty) => self.visit_type(&ty.elem), + syn::Type::ImplTrait(ty) => { + for bound in &ty.bounds { + self.visit_type_param_bound(bound); + } + } + syn::Type::Macro(ty) => self.visit_macro(&ty.mac), + syn::Type::Paren(ty) => self.visit_type(&ty.elem), + syn::Type::Path(ty) => { + if let Some(qself) = &ty.qself { + self.visit_type(&qself.ty); + } + self.visit_path(&ty.path); + } + syn::Type::Ptr(ty) => self.visit_type(&ty.elem), + syn::Type::Reference(ty) => self.visit_type(&ty.elem), + syn::Type::Slice(ty) => self.visit_type(&ty.elem), + syn::Type::TraitObject(ty) => { + for bound in &ty.bounds { + self.visit_type_param_bound(bound); + } + } + syn::Type::Tuple(ty) => { + for elem in &ty.elems { + self.visit_type(elem); + } + } + + syn::Type::Infer(_) | syn::Type::Never(_) | syn::Type::Verbatim(_) => {} + + #[cfg_attr(all(test, exhaustive), deny(non_exhaustive_omitted_patterns))] + _ => {} + } + } + + fn visit_path_segment(&mut self, segment: &'ast syn::PathSegment) { + self.visit_path_arguments(&segment.arguments); + } + + fn visit_path_arguments(&mut self, arguments: &'ast syn::PathArguments) { + match arguments { + syn::PathArguments::None => {} + syn::PathArguments::AngleBracketed(arguments) => { + for arg in &arguments.args { + match arg { + syn::GenericArgument::Type(arg) => self.visit_type(arg), + syn::GenericArgument::Binding(arg) => self.visit_type(&arg.ty), + syn::GenericArgument::Lifetime(_) + | syn::GenericArgument::Constraint(_) + | syn::GenericArgument::Const(_) => {} + } + } + } + syn::PathArguments::Parenthesized(arguments) => { + for argument in &arguments.inputs { + self.visit_type(argument); + } + self.visit_return_type(&arguments.output); + } + } + } + + fn visit_return_type(&mut self, return_type: &'ast syn::ReturnType) { + match return_type { + syn::ReturnType::Default => {} + syn::ReturnType::Type(_, output) => self.visit_type(output), + } + } + + fn visit_type_param_bound(&mut self, bound: &'ast syn::TypeParamBound) { + match bound { + syn::TypeParamBound::Trait(bound) => self.visit_path(&bound.path), + syn::TypeParamBound::Lifetime(_) => {} + } + } + + // Type parameter should not be considered used by a macro path. + // + // struct TypeMacro { + // mac: T!(), + // marker: PhantomData, + // } + fn visit_macro(&mut self, _mac: &'ast syn::Macro) {} + } + + let all_type_params = generics + .type_params() + .map(|param| param.ident.clone()) + .collect(); + + let mut visitor = FindTyParams { + all_type_params, + relevant_type_params: HashSet::new(), + associated_type_usage: Vec::new(), + }; + match &cont.data { + Data::Enum(variants) => { + for variant in variants.iter() { + let relevant_fields = variant + .fields + .iter() + .filter(|field| filter(&field.attrs, Some(&variant.attrs))); + for field in relevant_fields { + visitor.visit_field(field.original); + } + } + } + Data::Struct(_, fields) => { + for field in fields.iter().filter(|field| filter(&field.attrs, None)) { + visitor.visit_field(field.original); + } + } + } + + let relevant_type_params = visitor.relevant_type_params; + let associated_type_usage = visitor.associated_type_usage; + let new_predicates = generics + .type_params() + .map(|param| param.ident.clone()) + .filter(|id| relevant_type_params.contains(id)) + .map(|id| syn::TypePath { + qself: None, + path: id.into(), + }) + .chain(associated_type_usage.into_iter().cloned()) + .map(|bounded_ty| { + syn::WherePredicate::Type(syn::PredicateType { + lifetimes: None, + // the type parameter that is being bounded e.g. T + bounded_ty: syn::Type::Path(bounded_ty), + colon_token: ::default(), + // the bound e.g. Serialize + bounds: vec![syn::TypeParamBound::Trait(syn::TraitBound { + paren_token: None, + modifier: syn::TraitBoundModifier::None, + lifetimes: None, + path: bound.clone(), + })] + .into_iter() + .collect(), + }) + }); + + let mut generics = generics.clone(); + generics + .make_where_clause() + .predicates + .extend(new_predicates); + generics +} + +pub fn with_self_bound( + cont: &Container, + generics: &syn::Generics, + bound: &syn::Path, +) -> syn::Generics { + let mut generics = generics.clone(); + generics + .make_where_clause() + .predicates + .push(syn::WherePredicate::Type(syn::PredicateType { + lifetimes: None, + // the type that is being bounded e.g. MyStruct<'a, T> + bounded_ty: type_of_item(cont), + colon_token: ::default(), + // the bound e.g. Default + bounds: vec![syn::TypeParamBound::Trait(syn::TraitBound { + paren_token: None, + modifier: syn::TraitBoundModifier::None, + lifetimes: None, + path: bound.clone(), + })] + .into_iter() + .collect(), + })); + generics +} + +pub fn with_lifetime_bound(generics: &syn::Generics, lifetime: &str) -> syn::Generics { + let bound = syn::Lifetime::new(lifetime, Span::call_site()); + let def = syn::LifetimeDef { + attrs: Vec::new(), + lifetime: bound.clone(), + colon_token: None, + bounds: Punctuated::new(), + }; + + let params = Some(syn::GenericParam::Lifetime(def)) + .into_iter() + .chain(generics.params.iter().cloned().map(|mut param| { + match &mut param { + syn::GenericParam::Lifetime(param) => { + param.bounds.push(bound.clone()); + } + syn::GenericParam::Type(param) => { + param + .bounds + .push(syn::TypeParamBound::Lifetime(bound.clone())); + } + syn::GenericParam::Const(_) => {} + } + param + })) + .collect(); + + syn::Generics { + params, + ..generics.clone() + } +} + +fn type_of_item(cont: &Container) -> syn::Type { + syn::Type::Path(syn::TypePath { + qself: None, + path: syn::Path { + leading_colon: None, + segments: vec![syn::PathSegment { + ident: cont.ident.clone(), + arguments: syn::PathArguments::AngleBracketed( + syn::AngleBracketedGenericArguments { + colon2_token: None, + lt_token: ::default(), + args: cont + .generics + .params + .iter() + .map(|param| match param { + syn::GenericParam::Type(param) => { + syn::GenericArgument::Type(syn::Type::Path(syn::TypePath { + qself: None, + path: param.ident.clone().into(), + })) + } + syn::GenericParam::Lifetime(param) => { + syn::GenericArgument::Lifetime(param.lifetime.clone()) + } + syn::GenericParam::Const(_) => { + panic!("Serde does not support const generics yet"); + } + }) + .collect(), + gt_token: ]>::default(), + }, + ), + }] + .into_iter() + .collect(), + }, + }) +} diff --git a/rust/serde_derive/de.rs b/rust/serde_derive/de.rs new file mode 100644 index 00000000000000..06780ac192e8a4 --- /dev/null +++ b/rust/serde_derive/de.rs @@ -0,0 +1,3148 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Literal, Span, TokenStream}; +use quote::ToTokens; +use syn::punctuated::Punctuated; +use syn::spanned::Spanned; +use syn::{self, Ident, Index, Member}; + +use bound; +use dummy; +use fragment::{Expr, Fragment, Match, Stmts}; +use internals::ast::{Container, Data, Field, Style, Variant}; +use internals::{attr, replace_receiver, ungroup, Ctxt, Derive}; +use pretend; +use this; + +use std::collections::BTreeSet; +use std::ptr; + +pub fn expand_derive_deserialize( + input: &mut syn::DeriveInput, +) -> Result> { + replace_receiver(input); + + let ctxt = Ctxt::new(); + let cont = match Container::from_ast(&ctxt, input, Derive::Deserialize) { + Some(cont) => cont, + None => return Err(ctxt.check().unwrap_err()), + }; + precondition(&ctxt, &cont); + ctxt.check()?; + + let ident = &cont.ident; + let params = Parameters::new(&cont); + let (de_impl_generics, _, ty_generics, where_clause) = split_with_de_lifetime(¶ms); + let body = Stmts(deserialize_body(&cont, ¶ms)); + let delife = params.borrowed.de_lifetime(); + let serde = cont.attrs.serde_path(); + + let impl_block = if let Some(remote) = cont.attrs.remote() { + let vis = &input.vis; + let used = pretend::pretend_used(&cont, params.is_packed); + quote! { + impl #de_impl_generics #ident #ty_generics #where_clause { + #vis fn deserialize<__D>(__deserializer: __D) -> #serde::__private::Result<#remote #ty_generics, __D::Error> + where + __D: #serde::Deserializer<#delife>, + { + #used + #body + } + } + } + } else { + let fn_deserialize_in_place = deserialize_in_place_body(&cont, ¶ms); + + quote! { + #[automatically_derived] + impl #de_impl_generics #serde::Deserialize<#delife> for #ident #ty_generics #where_clause { + fn deserialize<__D>(__deserializer: __D) -> #serde::__private::Result + where + __D: #serde::Deserializer<#delife>, + { + #body + } + + #fn_deserialize_in_place + } + } + }; + + Ok(dummy::wrap_in_const( + cont.attrs.custom_serde_path(), + "DESERIALIZE", + ident, + impl_block, + )) +} + +fn precondition(cx: &Ctxt, cont: &Container) { + precondition_sized(cx, cont); + precondition_no_de_lifetime(cx, cont); +} + +fn precondition_sized(cx: &Ctxt, cont: &Container) { + if let Data::Struct(_, fields) = &cont.data { + if let Some(last) = fields.last() { + if let syn::Type::Slice(_) = ungroup(last.ty) { + cx.error_spanned_by( + cont.original, + "cannot deserialize a dynamically sized struct", + ); + } + } + } +} + +fn precondition_no_de_lifetime(cx: &Ctxt, cont: &Container) { + if let BorrowedLifetimes::Borrowed(_) = borrowed_lifetimes(cont) { + for param in cont.generics.lifetimes() { + if param.lifetime.to_string() == "'de" { + cx.error_spanned_by( + ¶m.lifetime, + "cannot deserialize when there is a lifetime parameter called 'de", + ); + return; + } + } + } +} + +struct Parameters { + /// Name of the type the `derive` is on. + local: syn::Ident, + + /// Path to the type the impl is for. Either a single `Ident` for local + /// types (does not include generic parameters) or `some::remote::Path` for + /// remote types. + this_type: syn::Path, + + /// Same as `this_type` but using `::` for generic parameters for use in + /// expression position. + this_value: syn::Path, + + /// Generics including any explicit and inferred bounds for the impl. + generics: syn::Generics, + + /// Lifetimes borrowed from the deserializer. These will become bounds on + /// the `'de` lifetime of the deserializer. + borrowed: BorrowedLifetimes, + + /// At least one field has a serde(getter) attribute, implying that the + /// remote type has a private field. + has_getter: bool, + + /// Type has a repr(packed) attribute. + is_packed: bool, +} + +impl Parameters { + fn new(cont: &Container) -> Self { + let local = cont.ident.clone(); + let this_type = this::this_type(cont); + let this_value = this::this_value(cont); + let borrowed = borrowed_lifetimes(cont); + let generics = build_generics(cont, &borrowed); + let has_getter = cont.data.has_getter(); + let is_packed = cont.attrs.is_packed(); + + Parameters { + local, + this_type, + this_value, + generics, + borrowed, + has_getter, + is_packed, + } + } + + /// Type name to use in error messages and `&'static str` arguments to + /// various Deserializer methods. + fn type_name(&self) -> String { + self.this_type.segments.last().unwrap().ident.to_string() + } +} + +// All the generics in the input, plus a bound `T: Deserialize` for each generic +// field type that will be deserialized by us, plus a bound `T: Default` for +// each generic field type that will be set to a default value. +fn build_generics(cont: &Container, borrowed: &BorrowedLifetimes) -> syn::Generics { + let generics = bound::without_defaults(cont.generics); + + let generics = bound::with_where_predicates_from_fields(cont, &generics, attr::Field::de_bound); + + let generics = + bound::with_where_predicates_from_variants(cont, &generics, attr::Variant::de_bound); + + match cont.attrs.de_bound() { + Some(predicates) => bound::with_where_predicates(&generics, predicates), + None => { + let generics = match *cont.attrs.default() { + attr::Default::Default => bound::with_self_bound( + cont, + &generics, + &parse_quote!(_serde::__private::Default), + ), + attr::Default::None | attr::Default::Path(_) => generics, + }; + + let delife = borrowed.de_lifetime(); + let generics = bound::with_bound( + cont, + &generics, + needs_deserialize_bound, + &parse_quote!(_serde::Deserialize<#delife>), + ); + + bound::with_bound( + cont, + &generics, + requires_default, + &parse_quote!(_serde::__private::Default), + ) + } + } +} + +// Fields with a `skip_deserializing` or `deserialize_with` attribute, or which +// belong to a variant with a `skip_deserializing` or `deserialize_with` +// attribute, are not deserialized by us so we do not generate a bound. Fields +// with a `bound` attribute specify their own bound so we do not generate one. +// All other fields may need a `T: Deserialize` bound where T is the type of the +// field. +fn needs_deserialize_bound(field: &attr::Field, variant: Option<&attr::Variant>) -> bool { + !field.skip_deserializing() + && field.deserialize_with().is_none() + && field.de_bound().is_none() + && variant.map_or(true, |variant| { + !variant.skip_deserializing() + && variant.deserialize_with().is_none() + && variant.de_bound().is_none() + }) +} + +// Fields with a `default` attribute (not `default=...`), and fields with a +// `skip_deserializing` attribute that do not also have `default=...`. +fn requires_default(field: &attr::Field, _variant: Option<&attr::Variant>) -> bool { + if let attr::Default::Default = *field.default() { + true + } else { + false + } +} + +enum BorrowedLifetimes { + Borrowed(BTreeSet), + Static, +} + +impl BorrowedLifetimes { + fn de_lifetime(&self) -> syn::Lifetime { + match *self { + BorrowedLifetimes::Borrowed(_) => syn::Lifetime::new("'de", Span::call_site()), + BorrowedLifetimes::Static => syn::Lifetime::new("'static", Span::call_site()), + } + } + + fn de_lifetime_def(&self) -> Option { + match self { + BorrowedLifetimes::Borrowed(bounds) => Some(syn::LifetimeDef { + attrs: Vec::new(), + lifetime: syn::Lifetime::new("'de", Span::call_site()), + colon_token: None, + bounds: bounds.iter().cloned().collect(), + }), + BorrowedLifetimes::Static => None, + } + } +} + +// The union of lifetimes borrowed by each field of the container. +// +// These turn into bounds on the `'de` lifetime of the Deserialize impl. If +// lifetimes `'a` and `'b` are borrowed but `'c` is not, the impl is: +// +// impl<'de: 'a + 'b, 'a, 'b, 'c> Deserialize<'de> for S<'a, 'b, 'c> +// +// If any borrowed lifetime is `'static`, then `'de: 'static` would be redundant +// and we use plain `'static` instead of `'de`. +fn borrowed_lifetimes(cont: &Container) -> BorrowedLifetimes { + let mut lifetimes = BTreeSet::new(); + for field in cont.data.all_fields() { + if !field.attrs.skip_deserializing() { + lifetimes.extend(field.attrs.borrowed_lifetimes().iter().cloned()); + } + } + if lifetimes.iter().any(|b| b.to_string() == "'static") { + BorrowedLifetimes::Static + } else { + BorrowedLifetimes::Borrowed(lifetimes) + } +} + +fn deserialize_body(cont: &Container, params: &Parameters) -> Fragment { + if cont.attrs.transparent() { + deserialize_transparent(cont, params) + } else if let Some(type_from) = cont.attrs.type_from() { + deserialize_from(type_from) + } else if let Some(type_try_from) = cont.attrs.type_try_from() { + deserialize_try_from(type_try_from) + } else if let attr::Identifier::No = cont.attrs.identifier() { + match &cont.data { + Data::Enum(variants) => deserialize_enum(params, variants, &cont.attrs), + Data::Struct(Style::Struct, fields) => { + deserialize_struct(None, params, fields, &cont.attrs, None, &Untagged::No) + } + Data::Struct(Style::Tuple, fields) | Data::Struct(Style::Newtype, fields) => { + deserialize_tuple(None, params, fields, &cont.attrs, None) + } + Data::Struct(Style::Unit, _) => deserialize_unit_struct(params, &cont.attrs), + } + } else { + match &cont.data { + Data::Enum(variants) => deserialize_custom_identifier(params, variants, &cont.attrs), + Data::Struct(_, _) => unreachable!("checked in serde_derive_internals"), + } + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_in_place_body(cont: &Container, params: &Parameters) -> Option { + // Only remote derives have getters, and we do not generate + // deserialize_in_place for remote derives. + assert!(!params.has_getter); + + if cont.attrs.transparent() + || cont.attrs.type_from().is_some() + || cont.attrs.type_try_from().is_some() + || cont.attrs.identifier().is_some() + || cont + .data + .all_fields() + .all(|f| f.attrs.deserialize_with().is_some()) + { + return None; + } + + let code = match &cont.data { + Data::Struct(Style::Struct, fields) => { + deserialize_struct_in_place(None, params, fields, &cont.attrs, None)? + } + Data::Struct(Style::Tuple, fields) | Data::Struct(Style::Newtype, fields) => { + deserialize_tuple_in_place(None, params, fields, &cont.attrs, None) + } + Data::Enum(_) | Data::Struct(Style::Unit, _) => { + return None; + } + }; + + let delife = params.borrowed.de_lifetime(); + let stmts = Stmts(code); + + let fn_deserialize_in_place = quote_block! { + fn deserialize_in_place<__D>(__deserializer: __D, __place: &mut Self) -> _serde::__private::Result<(), __D::Error> + where + __D: _serde::Deserializer<#delife>, + { + #stmts + } + }; + + Some(Stmts(fn_deserialize_in_place)) +} + +#[cfg(not(feature = "deserialize_in_place"))] +fn deserialize_in_place_body(_cont: &Container, _params: &Parameters) -> Option { + None +} + +fn deserialize_transparent(cont: &Container, params: &Parameters) -> Fragment { + let fields = match &cont.data { + Data::Struct(_, fields) => fields, + Data::Enum(_) => unreachable!(), + }; + + let this_value = ¶ms.this_value; + let transparent_field = fields.iter().find(|f| f.attrs.transparent()).unwrap(); + + let path = match transparent_field.attrs.deserialize_with() { + Some(path) => quote!(#path), + None => { + let span = transparent_field.original.span(); + quote_spanned!(span=> _serde::Deserialize::deserialize) + } + }; + + let assign = fields.iter().map(|field| { + let member = &field.member; + if ptr::eq(field, transparent_field) { + quote!(#member: __transparent) + } else { + let value = match field.attrs.default() { + attr::Default::Default => quote!(_serde::__private::Default::default()), + attr::Default::Path(path) => quote!(#path()), + attr::Default::None => quote!(_serde::__private::PhantomData), + }; + quote!(#member: #value) + } + }); + + quote_block! { + _serde::__private::Result::map( + #path(__deserializer), + |__transparent| #this_value { #(#assign),* }) + } +} + +fn deserialize_from(type_from: &syn::Type) -> Fragment { + quote_block! { + _serde::__private::Result::map( + <#type_from as _serde::Deserialize>::deserialize(__deserializer), + _serde::__private::From::from) + } +} + +fn deserialize_try_from(type_try_from: &syn::Type) -> Fragment { + quote_block! { + _serde::__private::Result::and_then( + <#type_try_from as _serde::Deserialize>::deserialize(__deserializer), + |v| _serde::__private::TryFrom::try_from(v).map_err(_serde::de::Error::custom)) + } +} + +fn deserialize_unit_struct(params: &Parameters, cattrs: &attr::Container) -> Fragment { + let this_type = ¶ms.this_type; + let this_value = ¶ms.this_value; + let type_name = cattrs.name().deserialize_name(); + + let expecting = format!("unit struct {}", params.type_name()); + let expecting = cattrs.expecting().unwrap_or(&expecting); + + quote_block! { + struct __Visitor; + + impl<'de> _serde::de::Visitor<'de> for __Visitor { + type Value = #this_type; + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #[inline] + fn visit_unit<__E>(self) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(#this_value) + } + } + + _serde::Deserializer::deserialize_unit_struct(__deserializer, #type_name, __Visitor) + } +} + +fn deserialize_tuple( + variant_ident: Option<&syn::Ident>, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, + deserializer: Option, +) -> Fragment { + let this_type = ¶ms.this_type; + let this_value = ¶ms.this_value; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + assert!(!cattrs.has_flatten()); + + // If there are getters (implying private fields), construct the local type + // and use an `Into` conversion to get the remote type. If there are no + // getters then construct the target type directly. + let construct = if params.has_getter { + let local = ¶ms.local; + quote!(#local) + } else { + quote!(#this_value) + }; + + let is_enum = variant_ident.is_some(); + let type_path = match variant_ident { + Some(variant_ident) => quote!(#construct::#variant_ident), + None => construct, + }; + let expecting = match variant_ident { + Some(variant_ident) => format!("tuple variant {}::{}", params.type_name(), variant_ident), + None => format!("tuple struct {}", params.type_name()), + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let nfields = fields.len(); + + let visit_newtype_struct = if !is_enum && nfields == 1 { + Some(deserialize_newtype_struct(&type_path, params, &fields[0])) + } else { + None + }; + + let visit_seq = Stmts(deserialize_seq( + &type_path, params, fields, false, cattrs, expecting, + )); + + let visitor_expr = quote! { + __Visitor { + marker: _serde::__private::PhantomData::<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData, + } + }; + let dispatch = if let Some(deserializer) = deserializer { + quote!(_serde::Deserializer::deserialize_tuple(#deserializer, #nfields, #visitor_expr)) + } else if is_enum { + quote!(_serde::de::VariantAccess::tuple_variant(__variant, #nfields, #visitor_expr)) + } else if nfields == 1 { + let type_name = cattrs.name().deserialize_name(); + quote!(_serde::Deserializer::deserialize_newtype_struct(__deserializer, #type_name, #visitor_expr)) + } else { + let type_name = cattrs.name().deserialize_name(); + quote!(_serde::Deserializer::deserialize_tuple_struct(__deserializer, #type_name, #nfields, #visitor_expr)) + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + let visitor_var = if all_skipped { + quote!(_) + } else { + quote!(mut __seq) + }; + + quote_block! { + struct __Visitor #de_impl_generics #where_clause { + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::Visitor<#delife> for __Visitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #visit_newtype_struct + + #[inline] + fn visit_seq<__A>(self, #visitor_var: __A) -> _serde::__private::Result + where + __A: _serde::de::SeqAccess<#delife>, + { + #visit_seq + } + } + + #dispatch + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_tuple_in_place( + variant_ident: Option, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, + deserializer: Option, +) -> Fragment { + let this_type = ¶ms.this_type; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + assert!(!cattrs.has_flatten()); + + let is_enum = variant_ident.is_some(); + let expecting = match variant_ident { + Some(variant_ident) => format!("tuple variant {}::{}", params.type_name(), variant_ident), + None => format!("tuple struct {}", params.type_name()), + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let nfields = fields.len(); + + let visit_newtype_struct = if !is_enum && nfields == 1 { + Some(deserialize_newtype_struct_in_place(params, &fields[0])) + } else { + None + }; + + let visit_seq = Stmts(deserialize_seq_in_place(params, fields, cattrs, expecting)); + + let visitor_expr = quote! { + __Visitor { + place: __place, + lifetime: _serde::__private::PhantomData, + } + }; + + let dispatch = if let Some(deserializer) = deserializer { + quote!(_serde::Deserializer::deserialize_tuple(#deserializer, #nfields, #visitor_expr)) + } else if is_enum { + quote!(_serde::de::VariantAccess::tuple_variant(__variant, #nfields, #visitor_expr)) + } else if nfields == 1 { + let type_name = cattrs.name().deserialize_name(); + quote!(_serde::Deserializer::deserialize_newtype_struct(__deserializer, #type_name, #visitor_expr)) + } else { + let type_name = cattrs.name().deserialize_name(); + quote!(_serde::Deserializer::deserialize_tuple_struct(__deserializer, #type_name, #nfields, #visitor_expr)) + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + let visitor_var = if all_skipped { + quote!(_) + } else { + quote!(mut __seq) + }; + + let in_place_impl_generics = de_impl_generics.in_place(); + let in_place_ty_generics = de_ty_generics.in_place(); + let place_life = place_lifetime(); + + quote_block! { + struct __Visitor #in_place_impl_generics #where_clause { + place: &#place_life mut #this_type #ty_generics, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #in_place_impl_generics _serde::de::Visitor<#delife> for __Visitor #in_place_ty_generics #where_clause { + type Value = (); + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #visit_newtype_struct + + #[inline] + fn visit_seq<__A>(self, #visitor_var: __A) -> _serde::__private::Result + where + __A: _serde::de::SeqAccess<#delife>, + { + #visit_seq + } + } + + #dispatch + } +} + +fn deserialize_seq( + type_path: &TokenStream, + params: &Parameters, + fields: &[Field], + is_struct: bool, + cattrs: &attr::Container, + expecting: &str, +) -> Fragment { + let vars = (0..fields.len()).map(field_i as fn(_) -> _); + + let deserialized_count = fields + .iter() + .filter(|field| !field.attrs.skip_deserializing()) + .count(); + let expecting = if deserialized_count == 1 { + format!("{} with 1 element", expecting) + } else { + format!("{} with {} elements", expecting, deserialized_count) + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let mut index_in_seq = 0_usize; + let let_values = vars.clone().zip(fields).map(|(var, field)| { + if field.attrs.skip_deserializing() { + let default = Expr(expr_is_missing(field, cattrs)); + quote! { + let #var = #default; + } + } else { + let visit = match field.attrs.deserialize_with() { + None => { + let field_ty = field.ty; + let span = field.original.span(); + let func = + quote_spanned!(span=> _serde::de::SeqAccess::next_element::<#field_ty>); + quote!(try!(#func(&mut __seq))) + } + Some(path) => { + let (wrapper, wrapper_ty) = wrap_deserialize_field_with(params, field.ty, path); + quote!({ + #wrapper + _serde::__private::Option::map( + try!(_serde::de::SeqAccess::next_element::<#wrapper_ty>(&mut __seq)), + |__wrap| __wrap.value) + }) + } + }; + let value_if_none = match field.attrs.default() { + attr::Default::Default => quote!(_serde::__private::Default::default()), + attr::Default::Path(path) => quote!(#path()), + attr::Default::None => quote!( + return _serde::__private::Err(_serde::de::Error::invalid_length(#index_in_seq, &#expecting)); + ), + }; + let assign = quote! { + let #var = match #visit { + _serde::__private::Some(__value) => __value, + _serde::__private::None => { + #value_if_none + } + }; + }; + index_in_seq += 1; + assign + } + }); + + let mut result = if is_struct { + let names = fields.iter().map(|f| &f.member); + quote! { + #type_path { #( #names: #vars ),* } + } + } else { + quote! { + #type_path ( #(#vars),* ) + } + }; + + if params.has_getter { + let this_type = ¶ms.this_type; + let (_, ty_generics, _) = params.generics.split_for_impl(); + result = quote! { + _serde::__private::Into::<#this_type #ty_generics>::into(#result) + }; + } + + let let_default = match cattrs.default() { + attr::Default::Default => Some(quote!( + let __default: Self::Value = _serde::__private::Default::default(); + )), + attr::Default::Path(path) => Some(quote!( + let __default: Self::Value = #path(); + )), + attr::Default::None => { + // We don't need the default value, to prevent an unused variable warning + // we'll leave the line empty. + None + } + }; + + quote_block! { + #let_default + #(#let_values)* + _serde::__private::Ok(#result) + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_seq_in_place( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, + expecting: &str, +) -> Fragment { + let deserialized_count = fields + .iter() + .filter(|field| !field.attrs.skip_deserializing()) + .count(); + let expecting = if deserialized_count == 1 { + format!("{} with 1 element", expecting) + } else { + format!("{} with {} elements", expecting, deserialized_count) + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let mut index_in_seq = 0usize; + let write_values = fields.iter().map(|field| { + let member = &field.member; + + if field.attrs.skip_deserializing() { + let default = Expr(expr_is_missing(field, cattrs)); + quote! { + self.place.#member = #default; + } + } else { + let value_if_none = match field.attrs.default() { + attr::Default::Default => quote!( + self.place.#member = _serde::__private::Default::default(); + ), + attr::Default::Path(path) => quote!( + self.place.#member = #path(); + ), + attr::Default::None => quote!( + return _serde::__private::Err(_serde::de::Error::invalid_length(#index_in_seq, &#expecting)); + ), + }; + let write = match field.attrs.deserialize_with() { + None => { + quote! { + if let _serde::__private::None = try!(_serde::de::SeqAccess::next_element_seed(&mut __seq, + _serde::__private::de::InPlaceSeed(&mut self.place.#member))) + { + #value_if_none + } + } + } + Some(path) => { + let (wrapper, wrapper_ty) = wrap_deserialize_field_with(params, field.ty, path); + quote!({ + #wrapper + match try!(_serde::de::SeqAccess::next_element::<#wrapper_ty>(&mut __seq)) { + _serde::__private::Some(__wrap) => { + self.place.#member = __wrap.value; + } + _serde::__private::None => { + #value_if_none + } + } + }) + } + }; + index_in_seq += 1; + write + } + }); + + let this_type = ¶ms.this_type; + let (_, ty_generics, _) = params.generics.split_for_impl(); + let let_default = match cattrs.default() { + attr::Default::Default => Some(quote!( + let __default: #this_type #ty_generics = _serde::__private::Default::default(); + )), + attr::Default::Path(path) => Some(quote!( + let __default: #this_type #ty_generics = #path(); + )), + attr::Default::None => { + // We don't need the default value, to prevent an unused variable warning + // we'll leave the line empty. + None + } + }; + + quote_block! { + #let_default + #(#write_values)* + _serde::__private::Ok(()) + } +} + +fn deserialize_newtype_struct( + type_path: &TokenStream, + params: &Parameters, + field: &Field, +) -> TokenStream { + let delife = params.borrowed.de_lifetime(); + let field_ty = field.ty; + + let value = match field.attrs.deserialize_with() { + None => { + let span = field.original.span(); + let func = quote_spanned!(span=> <#field_ty as _serde::Deserialize>::deserialize); + quote! { + try!(#func(__e)) + } + } + Some(path) => { + quote! { + try!(#path(__e)) + } + } + }; + + let mut result = quote!(#type_path(__field0)); + if params.has_getter { + let this_type = ¶ms.this_type; + let (_, ty_generics, _) = params.generics.split_for_impl(); + result = quote! { + _serde::__private::Into::<#this_type #ty_generics>::into(#result) + }; + } + + quote! { + #[inline] + fn visit_newtype_struct<__E>(self, __e: __E) -> _serde::__private::Result + where + __E: _serde::Deserializer<#delife>, + { + let __field0: #field_ty = #value; + _serde::__private::Ok(#result) + } + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_newtype_struct_in_place(params: &Parameters, field: &Field) -> TokenStream { + // We do not generate deserialize_in_place if every field has a + // deserialize_with. + assert!(field.attrs.deserialize_with().is_none()); + + let delife = params.borrowed.de_lifetime(); + + quote! { + #[inline] + fn visit_newtype_struct<__E>(self, __e: __E) -> _serde::__private::Result + where + __E: _serde::Deserializer<#delife>, + { + _serde::Deserialize::deserialize_in_place(__e, &mut self.place.0) + } + } +} + +enum Untagged { + Yes, + No, +} + +fn deserialize_struct( + variant_ident: Option<&syn::Ident>, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, + deserializer: Option, + untagged: &Untagged, +) -> Fragment { + let is_enum = variant_ident.is_some(); + + let this_type = ¶ms.this_type; + let this_value = ¶ms.this_value; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + // If there are getters (implying private fields), construct the local type + // and use an `Into` conversion to get the remote type. If there are no + // getters then construct the target type directly. + let construct = if params.has_getter { + let local = ¶ms.local; + quote!(#local) + } else { + quote!(#this_value) + }; + + let type_path = match variant_ident { + Some(variant_ident) => quote!(#construct::#variant_ident), + None => construct, + }; + let expecting = match variant_ident { + Some(variant_ident) => format!("struct variant {}::{}", params.type_name(), variant_ident), + None => format!("struct {}", params.type_name()), + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let visit_seq = Stmts(deserialize_seq( + &type_path, params, fields, true, cattrs, expecting, + )); + + let (field_visitor, fields_stmt, visit_map) = if cattrs.has_flatten() { + deserialize_struct_as_map_visitor(&type_path, params, fields, cattrs) + } else { + deserialize_struct_as_struct_visitor(&type_path, params, fields, cattrs) + }; + let field_visitor = Stmts(field_visitor); + let fields_stmt = fields_stmt.map(Stmts); + let visit_map = Stmts(visit_map); + + let visitor_expr = quote! { + __Visitor { + marker: _serde::__private::PhantomData::<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData, + } + }; + let need_seed = deserializer.is_none(); + let dispatch = if let Some(deserializer) = deserializer { + quote! { + _serde::Deserializer::deserialize_any(#deserializer, #visitor_expr) + } + } else if is_enum && cattrs.has_flatten() { + quote! { + _serde::de::VariantAccess::newtype_variant_seed(__variant, #visitor_expr) + } + } else if is_enum { + quote! { + _serde::de::VariantAccess::struct_variant(__variant, FIELDS, #visitor_expr) + } + } else if cattrs.has_flatten() { + quote! { + _serde::Deserializer::deserialize_map(__deserializer, #visitor_expr) + } + } else { + let type_name = cattrs.name().deserialize_name(); + quote! { + _serde::Deserializer::deserialize_struct(__deserializer, #type_name, FIELDS, #visitor_expr) + } + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + let visitor_var = if all_skipped { + quote!(_) + } else { + quote!(mut __seq) + }; + + // untagged struct variants do not get a visit_seq method. The same applies to + // structs that only have a map representation. + let visit_seq = match *untagged { + Untagged::No if !cattrs.has_flatten() => Some(quote! { + #[inline] + fn visit_seq<__A>(self, #visitor_var: __A) -> _serde::__private::Result + where + __A: _serde::de::SeqAccess<#delife>, + { + #visit_seq + } + }), + _ => None, + }; + + let visitor_seed = if need_seed && is_enum && cattrs.has_flatten() { + Some(quote! { + impl #de_impl_generics _serde::de::DeserializeSeed<#delife> for __Visitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn deserialize<__D>(self, __deserializer: __D) -> _serde::__private::Result + where + __D: _serde::Deserializer<#delife>, + { + _serde::Deserializer::deserialize_map(__deserializer, self) + } + } + }) + } else { + None + }; + + quote_block! { + #field_visitor + + struct __Visitor #de_impl_generics #where_clause { + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::Visitor<#delife> for __Visitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #visit_seq + + #[inline] + fn visit_map<__A>(self, mut __map: __A) -> _serde::__private::Result + where + __A: _serde::de::MapAccess<#delife>, + { + #visit_map + } + } + + #visitor_seed + + #fields_stmt + + #dispatch + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_struct_in_place( + variant_ident: Option, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, + deserializer: Option, +) -> Option { + let is_enum = variant_ident.is_some(); + + // for now we do not support in_place deserialization for structs that + // are represented as map. + if cattrs.has_flatten() { + return None; + } + + let this_type = ¶ms.this_type; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + let expecting = match variant_ident { + Some(variant_ident) => format!("struct variant {}::{}", params.type_name(), variant_ident), + None => format!("struct {}", params.type_name()), + }; + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let visit_seq = Stmts(deserialize_seq_in_place(params, fields, cattrs, expecting)); + + let (field_visitor, fields_stmt, visit_map) = + deserialize_struct_as_struct_in_place_visitor(params, fields, cattrs); + + let field_visitor = Stmts(field_visitor); + let fields_stmt = Stmts(fields_stmt); + let visit_map = Stmts(visit_map); + + let visitor_expr = quote! { + __Visitor { + place: __place, + lifetime: _serde::__private::PhantomData, + } + }; + let dispatch = if let Some(deserializer) = deserializer { + quote! { + _serde::Deserializer::deserialize_any(#deserializer, #visitor_expr) + } + } else if is_enum { + quote! { + _serde::de::VariantAccess::struct_variant(__variant, FIELDS, #visitor_expr) + } + } else { + let type_name = cattrs.name().deserialize_name(); + quote! { + _serde::Deserializer::deserialize_struct(__deserializer, #type_name, FIELDS, #visitor_expr) + } + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + let visitor_var = if all_skipped { + quote!(_) + } else { + quote!(mut __seq) + }; + + let visit_seq = quote! { + #[inline] + fn visit_seq<__A>(self, #visitor_var: __A) -> _serde::__private::Result + where + __A: _serde::de::SeqAccess<#delife>, + { + #visit_seq + } + }; + + let in_place_impl_generics = de_impl_generics.in_place(); + let in_place_ty_generics = de_ty_generics.in_place(); + let place_life = place_lifetime(); + + Some(quote_block! { + #field_visitor + + struct __Visitor #in_place_impl_generics #where_clause { + place: &#place_life mut #this_type #ty_generics, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #in_place_impl_generics _serde::de::Visitor<#delife> for __Visitor #in_place_ty_generics #where_clause { + type Value = (); + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #visit_seq + + #[inline] + fn visit_map<__A>(self, mut __map: __A) -> _serde::__private::Result + where + __A: _serde::de::MapAccess<#delife>, + { + #visit_map + } + } + + #fields_stmt + + #dispatch + }) +} + +fn deserialize_enum( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, +) -> Fragment { + match cattrs.tag() { + attr::TagType::External => deserialize_externally_tagged_enum(params, variants, cattrs), + attr::TagType::Internal { tag } => { + deserialize_internally_tagged_enum(params, variants, cattrs, tag) + } + attr::TagType::Adjacent { tag, content } => { + deserialize_adjacently_tagged_enum(params, variants, cattrs, tag, content) + } + attr::TagType::None => deserialize_untagged_enum(params, variants, cattrs), + } +} + +fn prepare_enum_variant_enum( + variants: &[Variant], + cattrs: &attr::Container, +) -> (TokenStream, Stmts) { + let mut deserialized_variants = variants + .iter() + .enumerate() + .filter(|&(_, variant)| !variant.attrs.skip_deserializing()); + + let variant_names_idents: Vec<_> = deserialized_variants + .clone() + .map(|(i, variant)| { + ( + variant.attrs.name().deserialize_name(), + field_i(i), + variant.attrs.aliases(), + ) + }) + .collect(); + + let other_idx = deserialized_variants.position(|(_, variant)| variant.attrs.other()); + + let variants_stmt = { + let variant_names = variant_names_idents.iter().map(|(name, _, _)| name); + quote! { + const VARIANTS: &'static [&'static str] = &[ #(#variant_names),* ]; + } + }; + + let variant_visitor = Stmts(deserialize_generated_identifier( + &variant_names_idents, + cattrs, + true, + other_idx, + )); + + (variants_stmt, variant_visitor) +} + +fn deserialize_externally_tagged_enum( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, +) -> Fragment { + let this_type = ¶ms.this_type; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + let type_name = cattrs.name().deserialize_name(); + let expecting = format!("enum {}", params.type_name()); + let expecting = cattrs.expecting().unwrap_or(&expecting); + + let (variants_stmt, variant_visitor) = prepare_enum_variant_enum(variants, cattrs); + + // Match arms to extract a variant from a string + let variant_arms = variants + .iter() + .enumerate() + .filter(|&(_, variant)| !variant.attrs.skip_deserializing()) + .map(|(i, variant)| { + let variant_name = field_i(i); + + let block = Match(deserialize_externally_tagged_variant( + params, variant, cattrs, + )); + + quote! { + (__Field::#variant_name, __variant) => #block + } + }); + + let all_skipped = variants + .iter() + .all(|variant| variant.attrs.skip_deserializing()); + let match_variant = if all_skipped { + // This is an empty enum like `enum Impossible {}` or an enum in which + // all variants have `#[serde(skip_deserializing)]`. + quote! { + // FIXME: Once feature(exhaustive_patterns) is stable: + // let _serde::__private::Err(__err) = _serde::de::EnumAccess::variant::<__Field>(__data); + // _serde::__private::Err(__err) + _serde::__private::Result::map( + _serde::de::EnumAccess::variant::<__Field>(__data), + |(__impossible, _)| match __impossible {}) + } + } else { + quote! { + match try!(_serde::de::EnumAccess::variant(__data)) { + #(#variant_arms)* + } + } + }; + + quote_block! { + #variant_visitor + + struct __Visitor #de_impl_generics #where_clause { + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::Visitor<#delife> for __Visitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + fn visit_enum<__A>(self, __data: __A) -> _serde::__private::Result + where + __A: _serde::de::EnumAccess<#delife>, + { + #match_variant + } + } + + #variants_stmt + + _serde::Deserializer::deserialize_enum( + __deserializer, + #type_name, + VARIANTS, + __Visitor { + marker: _serde::__private::PhantomData::<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData, + }, + ) + } +} + +fn deserialize_internally_tagged_enum( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, + tag: &str, +) -> Fragment { + let (variants_stmt, variant_visitor) = prepare_enum_variant_enum(variants, cattrs); + + // Match arms to extract a variant from a string + let variant_arms = variants + .iter() + .enumerate() + .filter(|&(_, variant)| !variant.attrs.skip_deserializing()) + .map(|(i, variant)| { + let variant_name = field_i(i); + + let block = Match(deserialize_internally_tagged_variant( + params, + variant, + cattrs, + quote! { + _serde::__private::de::ContentDeserializer::<__D::Error>::new(__tagged.content) + }, + )); + + quote! { + __Field::#variant_name => #block + } + }); + + let expecting = format!("internally tagged enum {}", params.type_name()); + let expecting = cattrs.expecting().unwrap_or(&expecting); + + quote_block! { + #variant_visitor + + #variants_stmt + + let __tagged = try!(_serde::Deserializer::deserialize_any( + __deserializer, + _serde::__private::de::TaggedContentVisitor::<__Field>::new(#tag, #expecting))); + + match __tagged.tag { + #(#variant_arms)* + } + } +} + +fn deserialize_adjacently_tagged_enum( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, + tag: &str, + content: &str, +) -> Fragment { + let this_type = ¶ms.this_type; + let this_value = ¶ms.this_value; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + let (variants_stmt, variant_visitor) = prepare_enum_variant_enum(variants, cattrs); + + let variant_arms: &Vec<_> = &variants + .iter() + .enumerate() + .filter(|&(_, variant)| !variant.attrs.skip_deserializing()) + .map(|(i, variant)| { + let variant_index = field_i(i); + + let block = Match(deserialize_untagged_variant( + params, + variant, + cattrs, + quote!(__deserializer), + )); + + quote! { + __Field::#variant_index => #block + } + }) + .collect(); + + let expecting = format!("adjacently tagged enum {}", params.type_name()); + let expecting = cattrs.expecting().unwrap_or(&expecting); + let type_name = cattrs.name().deserialize_name(); + let deny_unknown_fields = cattrs.deny_unknown_fields(); + + // If unknown fields are allowed, we pick the visitor that can step over + // those. Otherwise we pick the visitor that fails on unknown keys. + let field_visitor_ty = if deny_unknown_fields { + quote! { _serde::__private::de::TagOrContentFieldVisitor } + } else { + quote! { _serde::__private::de::TagContentOtherFieldVisitor } + }; + + let tag_or_content = quote! { + #field_visitor_ty { + tag: #tag, + content: #content, + } + }; + + let mut missing_content = quote! { + _serde::__private::Err(<__A::Error as _serde::de::Error>::missing_field(#content)) + }; + let mut missing_content_fallthrough = quote!(); + let missing_content_arms = variants + .iter() + .enumerate() + .filter(|&(_, variant)| !variant.attrs.skip_deserializing()) + .filter_map(|(i, variant)| { + let variant_index = field_i(i); + let variant_ident = &variant.ident; + + let arm = match variant.style { + Style::Unit => quote! { + _serde::__private::Ok(#this_value::#variant_ident) + }, + Style::Newtype if variant.attrs.deserialize_with().is_none() => { + let span = variant.original.span(); + let func = quote_spanned!(span=> _serde::__private::de::missing_field); + quote! { + #func(#content).map(#this_value::#variant_ident) + } + } + _ => { + missing_content_fallthrough = quote!(_ => #missing_content); + return None; + } + }; + Some(quote! { + __Field::#variant_index => #arm, + }) + }) + .collect::>(); + if !missing_content_arms.is_empty() { + missing_content = quote! { + match __field { + #(#missing_content_arms)* + #missing_content_fallthrough + } + }; + } + + // Advance the map by one key, returning early in case of error. + let next_key = quote! { + try!(_serde::de::MapAccess::next_key_seed(&mut __map, #tag_or_content)) + }; + + // When allowing unknown fields, we want to transparently step through keys + // we don't care about until we find `tag`, `content`, or run out of keys. + let next_relevant_key = if deny_unknown_fields { + next_key + } else { + quote!({ + let mut __rk : _serde::__private::Option<_serde::__private::de::TagOrContentField> = _serde::__private::None; + while let _serde::__private::Some(__k) = #next_key { + match __k { + _serde::__private::de::TagContentOtherField::Other => { + let _ = try!(_serde::de::MapAccess::next_value::<_serde::de::IgnoredAny>(&mut __map)); + continue; + }, + _serde::__private::de::TagContentOtherField::Tag => { + __rk = _serde::__private::Some(_serde::__private::de::TagOrContentField::Tag); + break; + } + _serde::__private::de::TagContentOtherField::Content => { + __rk = _serde::__private::Some(_serde::__private::de::TagOrContentField::Content); + break; + } + } + } + + __rk + }) + }; + + // Step through remaining keys, looking for duplicates of previously-seen + // keys. When unknown fields are denied, any key that isn't a duplicate will + // at this point immediately produce an error. + let visit_remaining_keys = quote! { + match #next_relevant_key { + _serde::__private::Some(_serde::__private::de::TagOrContentField::Tag) => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#tag)) + } + _serde::__private::Some(_serde::__private::de::TagOrContentField::Content) => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#content)) + } + _serde::__private::None => _serde::__private::Ok(__ret), + } + }; + + let finish_content_then_tag = if variant_arms.is_empty() { + quote! { + match try!(_serde::de::MapAccess::next_value::<__Field>(&mut __map)) {} + } + } else { + quote! { + let __ret = try!(match try!(_serde::de::MapAccess::next_value(&mut __map)) { + // Deserialize the buffered content now that we know the variant. + #(#variant_arms)* + }); + // Visit remaining keys, looking for duplicates. + #visit_remaining_keys + } + }; + + quote_block! { + #variant_visitor + + #variants_stmt + + struct __Seed #de_impl_generics #where_clause { + field: __Field, + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::DeserializeSeed<#delife> for __Seed #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn deserialize<__D>(self, __deserializer: __D) -> _serde::__private::Result + where + __D: _serde::Deserializer<#delife>, + { + match self.field { + #(#variant_arms)* + } + } + } + + struct __Visitor #de_impl_generics #where_clause { + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::Visitor<#delife> for __Visitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + fn visit_map<__A>(self, mut __map: __A) -> _serde::__private::Result + where + __A: _serde::de::MapAccess<#delife>, + { + // Visit the first relevant key. + match #next_relevant_key { + // First key is the tag. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Tag) => { + // Parse the tag. + let __field = try!(_serde::de::MapAccess::next_value(&mut __map)); + // Visit the second key. + match #next_relevant_key { + // Second key is a duplicate of the tag. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Tag) => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#tag)) + } + // Second key is the content. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Content) => { + let __ret = try!(_serde::de::MapAccess::next_value_seed(&mut __map, + __Seed { + field: __field, + marker: _serde::__private::PhantomData, + lifetime: _serde::__private::PhantomData, + })); + // Visit remaining keys, looking for duplicates. + #visit_remaining_keys + } + // There is no second key; might be okay if the we have a unit variant. + _serde::__private::None => #missing_content + } + } + // First key is the content. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Content) => { + // Buffer up the content. + let __content = try!(_serde::de::MapAccess::next_value::<_serde::__private::de::Content>(&mut __map)); + // Visit the second key. + match #next_relevant_key { + // Second key is the tag. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Tag) => { + let __deserializer = _serde::__private::de::ContentDeserializer::<__A::Error>::new(__content); + #finish_content_then_tag + } + // Second key is a duplicate of the content. + _serde::__private::Some(_serde::__private::de::TagOrContentField::Content) => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#content)) + } + // There is no second key. + _serde::__private::None => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::missing_field(#tag)) + } + } + } + // There is no first key. + _serde::__private::None => { + _serde::__private::Err(<__A::Error as _serde::de::Error>::missing_field(#tag)) + } + } + } + + fn visit_seq<__A>(self, mut __seq: __A) -> _serde::__private::Result + where + __A: _serde::de::SeqAccess<#delife>, + { + // Visit the first element - the tag. + match try!(_serde::de::SeqAccess::next_element(&mut __seq)) { + _serde::__private::Some(__field) => { + // Visit the second element - the content. + match try!(_serde::de::SeqAccess::next_element_seed( + &mut __seq, + __Seed { + field: __field, + marker: _serde::__private::PhantomData, + lifetime: _serde::__private::PhantomData, + }, + )) { + _serde::__private::Some(__ret) => _serde::__private::Ok(__ret), + // There is no second element. + _serde::__private::None => { + _serde::__private::Err(_serde::de::Error::invalid_length(1, &self)) + } + } + } + // There is no first element. + _serde::__private::None => { + _serde::__private::Err(_serde::de::Error::invalid_length(0, &self)) + } + } + } + } + + const FIELDS: &'static [&'static str] = &[#tag, #content]; + _serde::Deserializer::deserialize_struct( + __deserializer, + #type_name, + FIELDS, + __Visitor { + marker: _serde::__private::PhantomData::<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData, + }, + ) + } +} + +fn deserialize_untagged_enum( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, +) -> Fragment { + let attempts = variants + .iter() + .filter(|variant| !variant.attrs.skip_deserializing()) + .map(|variant| { + Expr(deserialize_untagged_variant( + params, + variant, + cattrs, + quote!( + _serde::__private::de::ContentRefDeserializer::<__D::Error>::new(&__content) + ), + )) + }); + + // TODO this message could be better by saving the errors from the failed + // attempts. The heuristic used by TOML was to count the number of fields + // processed before an error, and use the error that happened after the + // largest number of fields. I'm not sure I like that. Maybe it would be + // better to save all the errors and combine them into one message that + // explains why none of the variants matched. + let fallthrough_msg = format!( + "data did not match any variant of untagged enum {}", + params.type_name() + ); + let fallthrough_msg = cattrs.expecting().unwrap_or(&fallthrough_msg); + + quote_block! { + let __content = try!(<_serde::__private::de::Content as _serde::Deserialize>::deserialize(__deserializer)); + + #( + if let _serde::__private::Ok(__ok) = #attempts { + return _serde::__private::Ok(__ok); + } + )* + + _serde::__private::Err(_serde::de::Error::custom(#fallthrough_msg)) + } +} + +fn deserialize_externally_tagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, +) -> Fragment { + if let Some(path) = variant.attrs.deserialize_with() { + let (wrapper, wrapper_ty, unwrap_fn) = wrap_deserialize_variant_with(params, variant, path); + return quote_block! { + #wrapper + _serde::__private::Result::map( + _serde::de::VariantAccess::newtype_variant::<#wrapper_ty>(__variant), #unwrap_fn) + }; + } + + let variant_ident = &variant.ident; + + match variant.style { + Style::Unit => { + let this_value = ¶ms.this_value; + quote_block! { + try!(_serde::de::VariantAccess::unit_variant(__variant)); + _serde::__private::Ok(#this_value::#variant_ident) + } + } + Style::Newtype => deserialize_externally_tagged_newtype_variant( + variant_ident, + params, + &variant.fields[0], + cattrs, + ), + Style::Tuple => { + deserialize_tuple(Some(variant_ident), params, &variant.fields, cattrs, None) + } + Style::Struct => deserialize_struct( + Some(variant_ident), + params, + &variant.fields, + cattrs, + None, + &Untagged::No, + ), + } +} + +// Generates significant part of the visit_seq and visit_map bodies of visitors +// for the variants of internally tagged enum. +fn deserialize_internally_tagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, + deserializer: TokenStream, +) -> Fragment { + if variant.attrs.deserialize_with().is_some() { + return deserialize_untagged_variant(params, variant, cattrs, deserializer); + } + + let variant_ident = &variant.ident; + + match effective_style(variant) { + Style::Unit => { + let this_value = ¶ms.this_value; + let type_name = params.type_name(); + let variant_name = variant.ident.to_string(); + let default = variant.fields.get(0).map(|field| { + let default = Expr(expr_is_missing(field, cattrs)); + quote!((#default)) + }); + quote_block! { + try!(_serde::Deserializer::deserialize_any(#deserializer, _serde::__private::de::InternallyTaggedUnitVisitor::new(#type_name, #variant_name))); + _serde::__private::Ok(#this_value::#variant_ident #default) + } + } + Style::Newtype => deserialize_untagged_newtype_variant( + variant_ident, + params, + &variant.fields[0], + &deserializer, + ), + Style::Struct => deserialize_struct( + Some(variant_ident), + params, + &variant.fields, + cattrs, + Some(deserializer), + &Untagged::No, + ), + Style::Tuple => unreachable!("checked in serde_derive_internals"), + } +} + +fn deserialize_untagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, + deserializer: TokenStream, +) -> Fragment { + if let Some(path) = variant.attrs.deserialize_with() { + let unwrap_fn = unwrap_to_variant_closure(params, variant, false); + return quote_block! { + _serde::__private::Result::map(#path(#deserializer), #unwrap_fn) + }; + } + + let variant_ident = &variant.ident; + + match effective_style(variant) { + Style::Unit => { + let this_value = ¶ms.this_value; + let type_name = params.type_name(); + let variant_name = variant.ident.to_string(); + let default = variant.fields.get(0).map(|field| { + let default = Expr(expr_is_missing(field, cattrs)); + quote!((#default)) + }); + quote_expr! { + match _serde::Deserializer::deserialize_any( + #deserializer, + _serde::__private::de::UntaggedUnitVisitor::new(#type_name, #variant_name) + ) { + _serde::__private::Ok(()) => _serde::__private::Ok(#this_value::#variant_ident #default), + _serde::__private::Err(__err) => _serde::__private::Err(__err), + } + } + } + Style::Newtype => deserialize_untagged_newtype_variant( + variant_ident, + params, + &variant.fields[0], + &deserializer, + ), + Style::Tuple => deserialize_tuple( + Some(variant_ident), + params, + &variant.fields, + cattrs, + Some(deserializer), + ), + Style::Struct => deserialize_struct( + Some(variant_ident), + params, + &variant.fields, + cattrs, + Some(deserializer), + &Untagged::Yes, + ), + } +} + +fn deserialize_externally_tagged_newtype_variant( + variant_ident: &syn::Ident, + params: &Parameters, + field: &Field, + cattrs: &attr::Container, +) -> Fragment { + let this_value = ¶ms.this_value; + + if field.attrs.skip_deserializing() { + let default = Expr(expr_is_missing(field, cattrs)); + return quote_block! { + try!(_serde::de::VariantAccess::unit_variant(__variant)); + _serde::__private::Ok(#this_value::#variant_ident(#default)) + }; + } + + match field.attrs.deserialize_with() { + None => { + let field_ty = field.ty; + let span = field.original.span(); + let func = + quote_spanned!(span=> _serde::de::VariantAccess::newtype_variant::<#field_ty>); + quote_expr! { + _serde::__private::Result::map(#func(__variant), #this_value::#variant_ident) + } + } + Some(path) => { + let (wrapper, wrapper_ty) = wrap_deserialize_field_with(params, field.ty, path); + quote_block! { + #wrapper + _serde::__private::Result::map( + _serde::de::VariantAccess::newtype_variant::<#wrapper_ty>(__variant), + |__wrapper| #this_value::#variant_ident(__wrapper.value)) + } + } + } +} + +fn deserialize_untagged_newtype_variant( + variant_ident: &syn::Ident, + params: &Parameters, + field: &Field, + deserializer: &TokenStream, +) -> Fragment { + let this_value = ¶ms.this_value; + let field_ty = field.ty; + match field.attrs.deserialize_with() { + None => { + let span = field.original.span(); + let func = quote_spanned!(span=> <#field_ty as _serde::Deserialize>::deserialize); + quote_expr! { + _serde::__private::Result::map(#func(#deserializer), #this_value::#variant_ident) + } + } + Some(path) => { + quote_block! { + let __value: _serde::__private::Result<#field_ty, _> = #path(#deserializer); + _serde::__private::Result::map(__value, #this_value::#variant_ident) + } + } + } +} + +fn deserialize_generated_identifier( + fields: &[(String, Ident, Vec)], + cattrs: &attr::Container, + is_variant: bool, + other_idx: Option, +) -> Fragment { + let this_value = quote!(__Field); + let field_idents: &Vec<_> = &fields.iter().map(|(_, ident, _)| ident).collect(); + + let (ignore_variant, fallthrough) = if !is_variant && cattrs.has_flatten() { + let ignore_variant = quote!(__other(_serde::__private::de::Content<'de>),); + let fallthrough = quote!(_serde::__private::Ok(__Field::__other(__value))); + (Some(ignore_variant), Some(fallthrough)) + } else if let Some(other_idx) = other_idx { + let ignore_variant = fields[other_idx].1.clone(); + let fallthrough = quote!(_serde::__private::Ok(__Field::#ignore_variant)); + (None, Some(fallthrough)) + } else if is_variant || cattrs.deny_unknown_fields() { + (None, None) + } else { + let ignore_variant = quote!(__ignore,); + let fallthrough = quote!(_serde::__private::Ok(__Field::__ignore)); + (Some(ignore_variant), Some(fallthrough)) + }; + + let visitor_impl = Stmts(deserialize_identifier( + &this_value, + fields, + is_variant, + fallthrough, + None, + !is_variant && cattrs.has_flatten(), + None, + )); + + let lifetime = if !is_variant && cattrs.has_flatten() { + Some(quote!(<'de>)) + } else { + None + }; + + quote_block! { + #[allow(non_camel_case_types)] + enum __Field #lifetime { + #(#field_idents,)* + #ignore_variant + } + + struct __FieldVisitor; + + impl<'de> _serde::de::Visitor<'de> for __FieldVisitor { + type Value = __Field #lifetime; + + #visitor_impl + } + + impl<'de> _serde::Deserialize<'de> for __Field #lifetime { + #[inline] + fn deserialize<__D>(__deserializer: __D) -> _serde::__private::Result + where + __D: _serde::Deserializer<'de>, + { + _serde::Deserializer::deserialize_identifier(__deserializer, __FieldVisitor) + } + } + } +} + +// Generates `Deserialize::deserialize` body for an enum with +// `serde(field_identifier)` or `serde(variant_identifier)` attribute. +fn deserialize_custom_identifier( + params: &Parameters, + variants: &[Variant], + cattrs: &attr::Container, +) -> Fragment { + let is_variant = match cattrs.identifier() { + attr::Identifier::Variant => true, + attr::Identifier::Field => false, + attr::Identifier::No => unreachable!(), + }; + + let this_type = params.this_type.to_token_stream(); + let this_value = params.this_value.to_token_stream(); + + let (ordinary, fallthrough, fallthrough_borrowed) = if let Some(last) = variants.last() { + let last_ident = &last.ident; + if last.attrs.other() { + // Process `serde(other)` attribute. It would always be found on the + // last variant (checked in `check_identifier`), so all preceding + // are ordinary variants. + let ordinary = &variants[..variants.len() - 1]; + let fallthrough = quote!(_serde::__private::Ok(#this_value::#last_ident)); + (ordinary, Some(fallthrough), None) + } else if let Style::Newtype = last.style { + let ordinary = &variants[..variants.len() - 1]; + let fallthrough = |value| { + quote! { + _serde::__private::Result::map( + _serde::Deserialize::deserialize( + _serde::__private::de::IdentifierDeserializer::from(#value) + ), + #this_value::#last_ident) + } + }; + ( + ordinary, + Some(fallthrough(quote!(__value))), + Some(fallthrough(quote!(_serde::__private::de::Borrowed( + __value + )))), + ) + } else { + (variants, None, None) + } + } else { + (variants, None, None) + }; + + let names_idents: Vec<_> = ordinary + .iter() + .map(|variant| { + ( + variant.attrs.name().deserialize_name(), + variant.ident.clone(), + variant.attrs.aliases(), + ) + }) + .collect(); + + let names = names_idents.iter().map(|(name, _, _)| name); + + let names_const = if fallthrough.is_some() { + None + } else if is_variant { + let variants = quote! { + const VARIANTS: &'static [&'static str] = &[ #(#names),* ]; + }; + Some(variants) + } else { + let fields = quote! { + const FIELDS: &'static [&'static str] = &[ #(#names),* ]; + }; + Some(fields) + }; + + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + let visitor_impl = Stmts(deserialize_identifier( + &this_value, + &names_idents, + is_variant, + fallthrough, + fallthrough_borrowed, + false, + cattrs.expecting(), + )); + + quote_block! { + #names_const + + struct __FieldVisitor #de_impl_generics #where_clause { + marker: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::de::Visitor<#delife> for __FieldVisitor #de_ty_generics #where_clause { + type Value = #this_type #ty_generics; + + #visitor_impl + } + + let __visitor = __FieldVisitor { + marker: _serde::__private::PhantomData::<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData, + }; + _serde::Deserializer::deserialize_identifier(__deserializer, __visitor) + } +} + +fn deserialize_identifier( + this_value: &TokenStream, + fields: &[(String, Ident, Vec)], + is_variant: bool, + fallthrough: Option, + fallthrough_borrowed: Option, + collect_other_fields: bool, + expecting: Option<&str>, +) -> Fragment { + let mut flat_fields = Vec::new(); + for (_, ident, aliases) in fields { + flat_fields.extend(aliases.iter().map(|alias| (alias, ident))); + } + + let field_strs: &Vec<_> = &flat_fields.iter().map(|(name, _)| name).collect(); + let field_bytes: &Vec<_> = &flat_fields + .iter() + .map(|(name, _)| Literal::byte_string(name.as_bytes())) + .collect(); + + let constructors: &Vec<_> = &flat_fields + .iter() + .map(|(_, ident)| quote!(#this_value::#ident)) + .collect(); + let main_constructors: &Vec<_> = &fields + .iter() + .map(|(_, ident, _)| quote!(#this_value::#ident)) + .collect(); + + let expecting = expecting.unwrap_or(if is_variant { + "variant identifier" + } else { + "field identifier" + }); + + let index_expecting = if is_variant { "variant" } else { "field" }; + + let bytes_to_str = if fallthrough.is_some() || collect_other_fields { + None + } else { + Some(quote! { + let __value = &_serde::__private::from_utf8_lossy(__value); + }) + }; + + let ( + value_as_str_content, + value_as_borrowed_str_content, + value_as_bytes_content, + value_as_borrowed_bytes_content, + ) = if collect_other_fields { + ( + Some(quote! { + let __value = _serde::__private::de::Content::String(_serde::__private::ToString::to_string(__value)); + }), + Some(quote! { + let __value = _serde::__private::de::Content::Str(__value); + }), + Some(quote! { + let __value = _serde::__private::de::Content::ByteBuf(__value.to_vec()); + }), + Some(quote! { + let __value = _serde::__private::de::Content::Bytes(__value); + }), + ) + } else { + (None, None, None, None) + }; + + let fallthrough_arm_tokens; + let fallthrough_arm = if let Some(fallthrough) = &fallthrough { + fallthrough + } else if is_variant { + fallthrough_arm_tokens = quote! { + _serde::__private::Err(_serde::de::Error::unknown_variant(__value, VARIANTS)) + }; + &fallthrough_arm_tokens + } else { + fallthrough_arm_tokens = quote! { + _serde::__private::Err(_serde::de::Error::unknown_field(__value, FIELDS)) + }; + &fallthrough_arm_tokens + }; + + let u64_fallthrough_arm_tokens; + let u64_fallthrough_arm = if let Some(fallthrough) = &fallthrough { + fallthrough + } else { + let fallthrough_msg = format!("{} index 0 <= i < {}", index_expecting, fields.len()); + u64_fallthrough_arm_tokens = quote! { + _serde::__private::Err(_serde::de::Error::invalid_value( + _serde::de::Unexpected::Unsigned(__value), + &#fallthrough_msg, + )) + }; + &u64_fallthrough_arm_tokens + }; + + let variant_indices = 0_u64..; + let visit_other = if collect_other_fields { + quote! { + fn visit_bool<__E>(self, __value: bool) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::Bool(__value))) + } + + fn visit_i8<__E>(self, __value: i8) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::I8(__value))) + } + + fn visit_i16<__E>(self, __value: i16) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::I16(__value))) + } + + fn visit_i32<__E>(self, __value: i32) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::I32(__value))) + } + + fn visit_i64<__E>(self, __value: i64) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::I64(__value))) + } + + fn visit_u8<__E>(self, __value: u8) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::U8(__value))) + } + + fn visit_u16<__E>(self, __value: u16) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::U16(__value))) + } + + fn visit_u32<__E>(self, __value: u32) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::U32(__value))) + } + + fn visit_u64<__E>(self, __value: u64) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::U64(__value))) + } + + fn visit_f32<__E>(self, __value: f32) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::F32(__value))) + } + + fn visit_f64<__E>(self, __value: f64) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::F64(__value))) + } + + fn visit_char<__E>(self, __value: char) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::Char(__value))) + } + + fn visit_unit<__E>(self) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + _serde::__private::Ok(__Field::__other(_serde::__private::de::Content::Unit)) + } + } + } else { + quote! { + fn visit_u64<__E>(self, __value: u64) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + match __value { + #( + #variant_indices => _serde::__private::Ok(#main_constructors), + )* + _ => #u64_fallthrough_arm, + } + } + } + }; + + let visit_borrowed = if fallthrough_borrowed.is_some() || collect_other_fields { + let fallthrough_borrowed_arm = fallthrough_borrowed.as_ref().unwrap_or(fallthrough_arm); + Some(quote! { + fn visit_borrowed_str<__E>(self, __value: &'de str) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + match __value { + #( + #field_strs => _serde::__private::Ok(#constructors), + )* + _ => { + #value_as_borrowed_str_content + #fallthrough_borrowed_arm + } + } + } + + fn visit_borrowed_bytes<__E>(self, __value: &'de [u8]) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + match __value { + #( + #field_bytes => _serde::__private::Ok(#constructors), + )* + _ => { + #bytes_to_str + #value_as_borrowed_bytes_content + #fallthrough_borrowed_arm + } + } + } + }) + } else { + None + }; + + quote_block! { + fn expecting(&self, __formatter: &mut _serde::__private::Formatter) -> _serde::__private::fmt::Result { + _serde::__private::Formatter::write_str(__formatter, #expecting) + } + + #visit_other + + fn visit_str<__E>(self, __value: &str) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + match __value { + #( + #field_strs => _serde::__private::Ok(#constructors), + )* + _ => { + #value_as_str_content + #fallthrough_arm + } + } + } + + fn visit_bytes<__E>(self, __value: &[u8]) -> _serde::__private::Result + where + __E: _serde::de::Error, + { + match __value { + #( + #field_bytes => _serde::__private::Ok(#constructors), + )* + _ => { + #bytes_to_str + #value_as_bytes_content + #fallthrough_arm + } + } + } + + #visit_borrowed + } +} + +fn deserialize_struct_as_struct_visitor( + struct_path: &TokenStream, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> (Fragment, Option, Fragment) { + assert!(!cattrs.has_flatten()); + + let field_names_idents: Vec<_> = fields + .iter() + .enumerate() + .filter(|&(_, field)| !field.attrs.skip_deserializing()) + .map(|(i, field)| { + ( + field.attrs.name().deserialize_name(), + field_i(i), + field.attrs.aliases(), + ) + }) + .collect(); + + let fields_stmt = { + let field_names = field_names_idents + .iter() + .flat_map(|(_, _, aliases)| aliases); + + quote_block! { + const FIELDS: &'static [&'static str] = &[ #(#field_names),* ]; + } + }; + + let field_visitor = deserialize_generated_identifier(&field_names_idents, cattrs, false, None); + + let visit_map = deserialize_map(struct_path, params, fields, cattrs); + + (field_visitor, Some(fields_stmt), visit_map) +} + +fn deserialize_struct_as_map_visitor( + struct_path: &TokenStream, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> (Fragment, Option, Fragment) { + let field_names_idents: Vec<_> = fields + .iter() + .enumerate() + .filter(|&(_, field)| !field.attrs.skip_deserializing() && !field.attrs.flatten()) + .map(|(i, field)| { + ( + field.attrs.name().deserialize_name(), + field_i(i), + field.attrs.aliases(), + ) + }) + .collect(); + + let field_visitor = deserialize_generated_identifier(&field_names_idents, cattrs, false, None); + + let visit_map = deserialize_map(struct_path, params, fields, cattrs); + + (field_visitor, None, visit_map) +} + +fn deserialize_map( + struct_path: &TokenStream, + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> Fragment { + // Create the field names for the fields. + let fields_names: Vec<_> = fields + .iter() + .enumerate() + .map(|(i, field)| (field, field_i(i))) + .collect(); + + // Declare each field that will be deserialized. + let let_values = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing() && !field.attrs.flatten()) + .map(|(field, name)| { + let field_ty = field.ty; + quote! { + let mut #name: _serde::__private::Option<#field_ty> = _serde::__private::None; + } + }); + + // Collect contents for flatten fields into a buffer + let let_collect = if cattrs.has_flatten() { + Some(quote! { + let mut __collect = _serde::__private::Vec::<_serde::__private::Option<( + _serde::__private::de::Content, + _serde::__private::de::Content + )>>::new(); + }) + } else { + None + }; + + // Match arms to extract a value for a field. + let value_arms = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing() && !field.attrs.flatten()) + .map(|(field, name)| { + let deser_name = field.attrs.name().deserialize_name(); + + let visit = match field.attrs.deserialize_with() { + None => { + let field_ty = field.ty; + let span = field.original.span(); + let func = + quote_spanned!(span=> _serde::de::MapAccess::next_value::<#field_ty>); + quote! { + try!(#func(&mut __map)) + } + } + Some(path) => { + let (wrapper, wrapper_ty) = wrap_deserialize_field_with(params, field.ty, path); + quote!({ + #wrapper + match _serde::de::MapAccess::next_value::<#wrapper_ty>(&mut __map) { + _serde::__private::Ok(__wrapper) => __wrapper.value, + _serde::__private::Err(__err) => { + return _serde::__private::Err(__err); + } + } + }) + } + }; + quote! { + __Field::#name => { + if _serde::__private::Option::is_some(&#name) { + return _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#deser_name)); + } + #name = _serde::__private::Some(#visit); + } + } + }); + + // Visit ignored values to consume them + let ignored_arm = if cattrs.has_flatten() { + Some(quote! { + __Field::__other(__name) => { + __collect.push(_serde::__private::Some(( + __name, + try!(_serde::de::MapAccess::next_value(&mut __map))))); + } + }) + } else if cattrs.deny_unknown_fields() { + None + } else { + Some(quote! { + _ => { let _ = try!(_serde::de::MapAccess::next_value::<_serde::de::IgnoredAny>(&mut __map)); } + }) + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + let match_keys = if cattrs.deny_unknown_fields() && all_skipped { + quote! { + // FIXME: Once feature(exhaustive_patterns) is stable: + // let _serde::__private::None::<__Field> = try!(_serde::de::MapAccess::next_key(&mut __map)); + _serde::__private::Option::map( + try!(_serde::de::MapAccess::next_key::<__Field>(&mut __map)), + |__impossible| match __impossible {}); + } + } else { + quote! { + while let _serde::__private::Some(__key) = try!(_serde::de::MapAccess::next_key::<__Field>(&mut __map)) { + match __key { + #(#value_arms)* + #ignored_arm + } + } + } + }; + + let extract_values = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing() && !field.attrs.flatten()) + .map(|(field, name)| { + let missing_expr = Match(expr_is_missing(field, cattrs)); + + quote! { + let #name = match #name { + _serde::__private::Some(#name) => #name, + _serde::__private::None => #missing_expr + }; + } + }); + + let extract_collected = fields_names + .iter() + .filter(|&&(field, _)| field.attrs.flatten() && !field.attrs.skip_deserializing()) + .map(|(field, name)| { + let field_ty = field.ty; + let func = match field.attrs.deserialize_with() { + None => { + let span = field.original.span(); + quote_spanned!(span=> _serde::de::Deserialize::deserialize) + } + Some(path) => quote!(#path), + }; + quote! { + let #name: #field_ty = try!(#func( + _serde::__private::de::FlatMapDeserializer( + &mut __collect, + _serde::__private::PhantomData))); + } + }); + + let collected_deny_unknown_fields = if cattrs.has_flatten() && cattrs.deny_unknown_fields() { + Some(quote! { + if let _serde::__private::Some(_serde::__private::Some((__key, _))) = + __collect.into_iter().filter(_serde::__private::Option::is_some).next() + { + if let _serde::__private::Some(__key) = __key.as_str() { + return _serde::__private::Err( + _serde::de::Error::custom(format_args!("unknown field `{}`", &__key))); + } else { + return _serde::__private::Err( + _serde::de::Error::custom(format_args!("unexpected map key"))); + } + } + }) + } else { + None + }; + + let result = fields_names.iter().map(|(field, name)| { + let member = &field.member; + if field.attrs.skip_deserializing() { + let value = Expr(expr_is_missing(field, cattrs)); + quote!(#member: #value) + } else { + quote!(#member: #name) + } + }); + + let let_default = match cattrs.default() { + attr::Default::Default => Some(quote!( + let __default: Self::Value = _serde::__private::Default::default(); + )), + attr::Default::Path(path) => Some(quote!( + let __default: Self::Value = #path(); + )), + attr::Default::None => { + // We don't need the default value, to prevent an unused variable warning + // we'll leave the line empty. + None + } + }; + + let mut result = quote!(#struct_path { #(#result),* }); + if params.has_getter { + let this_type = ¶ms.this_type; + let (_, ty_generics, _) = params.generics.split_for_impl(); + result = quote! { + _serde::__private::Into::<#this_type #ty_generics>::into(#result) + }; + } + + quote_block! { + #(#let_values)* + + #let_collect + + #match_keys + + #let_default + + #(#extract_values)* + + #(#extract_collected)* + + #collected_deny_unknown_fields + + _serde::__private::Ok(#result) + } +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_struct_as_struct_in_place_visitor( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> (Fragment, Fragment, Fragment) { + assert!(!cattrs.has_flatten()); + + let field_names_idents: Vec<_> = fields + .iter() + .enumerate() + .filter(|&(_, field)| !field.attrs.skip_deserializing()) + .map(|(i, field)| { + ( + field.attrs.name().deserialize_name(), + field_i(i), + field.attrs.aliases(), + ) + }) + .collect(); + + let fields_stmt = { + let field_names = field_names_idents.iter().map(|(name, _, _)| name); + quote_block! { + const FIELDS: &'static [&'static str] = &[ #(#field_names),* ]; + } + }; + + let field_visitor = deserialize_generated_identifier(&field_names_idents, cattrs, false, None); + + let visit_map = deserialize_map_in_place(params, fields, cattrs); + + (field_visitor, fields_stmt, visit_map) +} + +#[cfg(feature = "deserialize_in_place")] +fn deserialize_map_in_place( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> Fragment { + assert!(!cattrs.has_flatten()); + + // Create the field names for the fields. + let fields_names: Vec<_> = fields + .iter() + .enumerate() + .map(|(i, field)| (field, field_i(i))) + .collect(); + + // For deserialize_in_place, declare booleans for each field that will be + // deserialized. + let let_flags = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing()) + .map(|(_, name)| { + quote! { + let mut #name: bool = false; + } + }); + + // Match arms to extract a value for a field. + let value_arms_from = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing()) + .map(|(field, name)| { + let deser_name = field.attrs.name().deserialize_name(); + let member = &field.member; + + let visit = match field.attrs.deserialize_with() { + None => { + quote! { + try!(_serde::de::MapAccess::next_value_seed(&mut __map, _serde::__private::de::InPlaceSeed(&mut self.place.#member))) + } + } + Some(path) => { + let (wrapper, wrapper_ty) = wrap_deserialize_field_with(params, field.ty, path); + quote!({ + #wrapper + self.place.#member = match _serde::de::MapAccess::next_value::<#wrapper_ty>(&mut __map) { + _serde::__private::Ok(__wrapper) => __wrapper.value, + _serde::__private::Err(__err) => { + return _serde::__private::Err(__err); + } + }; + }) + } + }; + quote! { + __Field::#name => { + if #name { + return _serde::__private::Err(<__A::Error as _serde::de::Error>::duplicate_field(#deser_name)); + } + #visit; + #name = true; + } + } + }); + + // Visit ignored values to consume them + let ignored_arm = if cattrs.deny_unknown_fields() { + None + } else { + Some(quote! { + _ => { let _ = try!(_serde::de::MapAccess::next_value::<_serde::de::IgnoredAny>(&mut __map)); } + }) + }; + + let all_skipped = fields.iter().all(|field| field.attrs.skip_deserializing()); + + let match_keys = if cattrs.deny_unknown_fields() && all_skipped { + quote! { + // FIXME: Once feature(exhaustive_patterns) is stable: + // let _serde::__private::None::<__Field> = try!(_serde::de::MapAccess::next_key(&mut __map)); + _serde::__private::Option::map( + try!(_serde::de::MapAccess::next_key::<__Field>(&mut __map)), + |__impossible| match __impossible {}); + } + } else { + quote! { + while let _serde::__private::Some(__key) = try!(_serde::de::MapAccess::next_key::<__Field>(&mut __map)) { + match __key { + #(#value_arms_from)* + #ignored_arm + } + } + } + }; + + let check_flags = fields_names + .iter() + .filter(|&&(field, _)| !field.attrs.skip_deserializing()) + .map(|(field, name)| { + let missing_expr = expr_is_missing(field, cattrs); + // If missing_expr unconditionally returns an error, don't try + // to assign its value to self.place. + if field.attrs.default().is_none() + && cattrs.default().is_none() + && field.attrs.deserialize_with().is_some() + { + let missing_expr = Stmts(missing_expr); + quote! { + if !#name { + #missing_expr; + } + } + } else { + let member = &field.member; + let missing_expr = Expr(missing_expr); + quote! { + if !#name { + self.place.#member = #missing_expr; + }; + } + } + }); + + let this_type = ¶ms.this_type; + let (_, _, ty_generics, _) = split_with_de_lifetime(params); + + let let_default = match cattrs.default() { + attr::Default::Default => Some(quote!( + let __default: #this_type #ty_generics = _serde::__private::Default::default(); + )), + attr::Default::Path(path) => Some(quote!( + let __default: #this_type #ty_generics = #path(); + )), + attr::Default::None => { + // We don't need the default value, to prevent an unused variable warning + // we'll leave the line empty. + None + } + }; + + quote_block! { + #(#let_flags)* + + #match_keys + + #let_default + + #(#check_flags)* + + _serde::__private::Ok(()) + } +} + +fn field_i(i: usize) -> Ident { + Ident::new(&format!("__field{}", i), Span::call_site()) +} + +/// This function wraps the expression in `#[serde(deserialize_with = "...")]` +/// in a trait to prevent it from accessing the internal `Deserialize` state. +fn wrap_deserialize_with( + params: &Parameters, + value_ty: &TokenStream, + deserialize_with: &syn::ExprPath, +) -> (TokenStream, TokenStream) { + let this_type = ¶ms.this_type; + let (de_impl_generics, de_ty_generics, ty_generics, where_clause) = + split_with_de_lifetime(params); + let delife = params.borrowed.de_lifetime(); + + let wrapper = quote! { + struct __DeserializeWith #de_impl_generics #where_clause { + value: #value_ty, + phantom: _serde::__private::PhantomData<#this_type #ty_generics>, + lifetime: _serde::__private::PhantomData<&#delife ()>, + } + + impl #de_impl_generics _serde::Deserialize<#delife> for __DeserializeWith #de_ty_generics #where_clause { + fn deserialize<__D>(__deserializer: __D) -> _serde::__private::Result + where + __D: _serde::Deserializer<#delife>, + { + _serde::__private::Ok(__DeserializeWith { + value: try!(#deserialize_with(__deserializer)), + phantom: _serde::__private::PhantomData, + lifetime: _serde::__private::PhantomData, + }) + } + } + }; + + let wrapper_ty = quote!(__DeserializeWith #de_ty_generics); + + (wrapper, wrapper_ty) +} + +fn wrap_deserialize_field_with( + params: &Parameters, + field_ty: &syn::Type, + deserialize_with: &syn::ExprPath, +) -> (TokenStream, TokenStream) { + wrap_deserialize_with(params, "e!(#field_ty), deserialize_with) +} + +fn wrap_deserialize_variant_with( + params: &Parameters, + variant: &Variant, + deserialize_with: &syn::ExprPath, +) -> (TokenStream, TokenStream, TokenStream) { + let field_tys = variant.fields.iter().map(|field| field.ty); + let (wrapper, wrapper_ty) = + wrap_deserialize_with(params, "e!((#(#field_tys),*)), deserialize_with); + + let unwrap_fn = unwrap_to_variant_closure(params, variant, true); + + (wrapper, wrapper_ty, unwrap_fn) +} + +// Generates closure that converts single input parameter to the final value. +fn unwrap_to_variant_closure( + params: &Parameters, + variant: &Variant, + with_wrapper: bool, +) -> TokenStream { + let this_value = ¶ms.this_value; + let variant_ident = &variant.ident; + + let (arg, wrapper) = if with_wrapper { + (quote! { __wrap }, quote! { __wrap.value }) + } else { + let field_tys = variant.fields.iter().map(|field| field.ty); + (quote! { __wrap: (#(#field_tys),*) }, quote! { __wrap }) + }; + + let field_access = (0..variant.fields.len()).map(|n| { + Member::Unnamed(Index { + index: n as u32, + span: Span::call_site(), + }) + }); + + match variant.style { + Style::Struct if variant.fields.len() == 1 => { + let member = &variant.fields[0].member; + quote! { + |#arg| #this_value::#variant_ident { #member: #wrapper } + } + } + Style::Struct => { + let members = variant.fields.iter().map(|field| &field.member); + quote! { + |#arg| #this_value::#variant_ident { #(#members: #wrapper.#field_access),* } + } + } + Style::Tuple => quote! { + |#arg| #this_value::#variant_ident(#(#wrapper.#field_access),*) + }, + Style::Newtype => quote! { + |#arg| #this_value::#variant_ident(#wrapper) + }, + Style::Unit => quote! { + |#arg| #this_value::#variant_ident + }, + } +} + +fn expr_is_missing(field: &Field, cattrs: &attr::Container) -> Fragment { + match field.attrs.default() { + attr::Default::Default => { + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::__private::Default::default); + return quote_expr!(#func()); + } + attr::Default::Path(path) => { + return quote_expr!(#path()); + } + attr::Default::None => { /* below */ } + } + + match *cattrs.default() { + attr::Default::Default | attr::Default::Path(_) => { + let member = &field.member; + return quote_expr!(__default.#member); + } + attr::Default::None => { /* below */ } + } + + let name = field.attrs.name().deserialize_name(); + match field.attrs.deserialize_with() { + None => { + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::__private::de::missing_field); + quote_expr! { + try!(#func(#name)) + } + } + Some(_) => { + quote_expr! { + return _serde::__private::Err(<__A::Error as _serde::de::Error>::missing_field(#name)) + } + } + } +} + +fn effective_style(variant: &Variant) -> Style { + match variant.style { + Style::Newtype if variant.fields[0].attrs.skip_deserializing() => Style::Unit, + other => other, + } +} + +struct DeImplGenerics<'a>(&'a Parameters); +#[cfg(feature = "deserialize_in_place")] +struct InPlaceImplGenerics<'a>(&'a Parameters); + +impl<'a> ToTokens for DeImplGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + let mut generics = self.0.generics.clone(); + if let Some(de_lifetime) = self.0.borrowed.de_lifetime_def() { + generics.params = Some(syn::GenericParam::Lifetime(de_lifetime)) + .into_iter() + .chain(generics.params) + .collect(); + } + let (impl_generics, _, _) = generics.split_for_impl(); + impl_generics.to_tokens(tokens); + } +} + +#[cfg(feature = "deserialize_in_place")] +impl<'a> ToTokens for InPlaceImplGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + let place_lifetime = place_lifetime(); + let mut generics = self.0.generics.clone(); + + // Add lifetime for `&'place mut Self, and `'a: 'place` + for param in &mut generics.params { + match param { + syn::GenericParam::Lifetime(param) => { + param.bounds.push(place_lifetime.lifetime.clone()); + } + syn::GenericParam::Type(param) => { + param.bounds.push(syn::TypeParamBound::Lifetime( + place_lifetime.lifetime.clone(), + )); + } + syn::GenericParam::Const(_) => {} + } + } + generics.params = Some(syn::GenericParam::Lifetime(place_lifetime)) + .into_iter() + .chain(generics.params) + .collect(); + if let Some(de_lifetime) = self.0.borrowed.de_lifetime_def() { + generics.params = Some(syn::GenericParam::Lifetime(de_lifetime)) + .into_iter() + .chain(generics.params) + .collect(); + } + let (impl_generics, _, _) = generics.split_for_impl(); + impl_generics.to_tokens(tokens); + } +} + +#[cfg(feature = "deserialize_in_place")] +impl<'a> DeImplGenerics<'a> { + fn in_place(self) -> InPlaceImplGenerics<'a> { + InPlaceImplGenerics(self.0) + } +} + +struct DeTypeGenerics<'a>(&'a Parameters); +#[cfg(feature = "deserialize_in_place")] +struct InPlaceTypeGenerics<'a>(&'a Parameters); + +impl<'a> ToTokens for DeTypeGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + let mut generics = self.0.generics.clone(); + if self.0.borrowed.de_lifetime_def().is_some() { + let def = syn::LifetimeDef { + attrs: Vec::new(), + lifetime: syn::Lifetime::new("'de", Span::call_site()), + colon_token: None, + bounds: Punctuated::new(), + }; + generics.params = Some(syn::GenericParam::Lifetime(def)) + .into_iter() + .chain(generics.params) + .collect(); + } + let (_, ty_generics, _) = generics.split_for_impl(); + ty_generics.to_tokens(tokens); + } +} + +#[cfg(feature = "deserialize_in_place")] +impl<'a> ToTokens for InPlaceTypeGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + let mut generics = self.0.generics.clone(); + generics.params = Some(syn::GenericParam::Lifetime(place_lifetime())) + .into_iter() + .chain(generics.params) + .collect(); + + if self.0.borrowed.de_lifetime_def().is_some() { + let def = syn::LifetimeDef { + attrs: Vec::new(), + lifetime: syn::Lifetime::new("'de", Span::call_site()), + colon_token: None, + bounds: Punctuated::new(), + }; + generics.params = Some(syn::GenericParam::Lifetime(def)) + .into_iter() + .chain(generics.params) + .collect(); + } + let (_, ty_generics, _) = generics.split_for_impl(); + ty_generics.to_tokens(tokens); + } +} + +#[cfg(feature = "deserialize_in_place")] +impl<'a> DeTypeGenerics<'a> { + fn in_place(self) -> InPlaceTypeGenerics<'a> { + InPlaceTypeGenerics(self.0) + } +} + +#[cfg(feature = "deserialize_in_place")] +fn place_lifetime() -> syn::LifetimeDef { + syn::LifetimeDef { + attrs: Vec::new(), + lifetime: syn::Lifetime::new("'place", Span::call_site()), + colon_token: None, + bounds: Punctuated::new(), + } +} + +fn split_with_de_lifetime( + params: &Parameters, +) -> ( + DeImplGenerics, + DeTypeGenerics, + syn::TypeGenerics, + Option<&syn::WhereClause>, +) { + let de_impl_generics = DeImplGenerics(params); + let de_ty_generics = DeTypeGenerics(params); + let (_, ty_generics, where_clause) = params.generics.split_for_impl(); + (de_impl_generics, de_ty_generics, ty_generics, where_clause) +} diff --git a/rust/serde_derive/dummy.rs b/rust/serde_derive/dummy.rs new file mode 100644 index 00000000000000..3530303f23f28a --- /dev/null +++ b/rust/serde_derive/dummy.rs @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Ident, TokenStream}; +use quote::format_ident; + +use syn; +use try; + +pub fn wrap_in_const( + serde_path: Option<&syn::Path>, + trait_: &str, + ty: &Ident, + code: TokenStream, +) -> TokenStream { + let try_replacement = try::replacement(); + + let dummy_const = if cfg!(no_underscore_consts) { + format_ident!("_IMPL_{}_FOR_{}", trait_, unraw(ty)) + } else { + format_ident!("_") + }; + + let use_serde = match serde_path { + Some(path) => quote! { + use #path as _serde; + }, + None => quote! { + #[allow(unused_extern_crates, clippy::useless_attribute)] + extern crate serde as _serde; + }, + }; + + quote! { + #[doc(hidden)] + #[allow(non_upper_case_globals, unused_attributes, unused_qualifications)] + const #dummy_const: () = { + #use_serde + #try_replacement + #code + }; + } +} + +fn unraw(ident: &Ident) -> String { + ident.to_string().trim_start_matches("r#").to_owned() +} diff --git a/rust/serde_derive/fragment.rs b/rust/serde_derive/fragment.rs new file mode 100644 index 00000000000000..a62ed9114224cb --- /dev/null +++ b/rust/serde_derive/fragment.rs @@ -0,0 +1,76 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::TokenStream; +use quote::ToTokens; +use syn::token; + +pub enum Fragment { + /// Tokens that can be used as an expression. + Expr(TokenStream), + /// Tokens that can be used inside a block. The surrounding curly braces are + /// not part of these tokens. + Block(TokenStream), +} + +macro_rules! quote_expr { + ($($tt:tt)*) => { + $crate::fragment::Fragment::Expr(quote!($($tt)*)) + } +} + +macro_rules! quote_block { + ($($tt:tt)*) => { + $crate::fragment::Fragment::Block(quote!($($tt)*)) + } +} + +/// Interpolate a fragment in place of an expression. This involves surrounding +/// Block fragments in curly braces. +pub struct Expr(pub Fragment); +impl ToTokens for Expr { + fn to_tokens(&self, out: &mut TokenStream) { + match &self.0 { + Fragment::Expr(expr) => expr.to_tokens(out), + Fragment::Block(block) => { + token::Brace::default().surround(out, |out| block.to_tokens(out)); + } + } + } +} + +/// Interpolate a fragment as the statements of a block. +pub struct Stmts(pub Fragment); +impl ToTokens for Stmts { + fn to_tokens(&self, out: &mut TokenStream) { + match &self.0 { + Fragment::Expr(expr) => expr.to_tokens(out), + Fragment::Block(block) => block.to_tokens(out), + } + } +} + +/// Interpolate a fragment as the value part of a `match` expression. This +/// involves putting a comma after expressions and curly braces around blocks. +pub struct Match(pub Fragment); +impl ToTokens for Match { + fn to_tokens(&self, out: &mut TokenStream) { + match &self.0 { + Fragment::Expr(expr) => { + expr.to_tokens(out); + ::default().to_tokens(out); + } + Fragment::Block(block) => { + token::Brace::default().surround(out, |out| block.to_tokens(out)); + } + } + } +} + +impl AsRef for Fragment { + fn as_ref(&self) -> &TokenStream { + match self { + Fragment::Expr(expr) => expr, + Fragment::Block(block) => block, + } + } +} diff --git a/rust/serde_derive/internals/ast.rs b/rust/serde_derive/internals/ast.rs new file mode 100644 index 00000000000000..139540571820ea --- /dev/null +++ b/rust/serde_derive/internals/ast.rs @@ -0,0 +1,204 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! A Serde ast, parsed from the Syn ast and ready to generate Rust code. + +use internals::attr; +use internals::check; +use internals::{Ctxt, Derive}; +use syn; +use syn::punctuated::Punctuated; + +/// A source data structure annotated with `#[derive(Serialize)]` and/or `#[derive(Deserialize)]`, +/// parsed into an internal representation. +pub struct Container<'a> { + /// The struct or enum name (without generics). + pub ident: syn::Ident, + /// Attributes on the structure, parsed for Serde. + pub attrs: attr::Container, + /// The contents of the struct or enum. + pub data: Data<'a>, + /// Any generics on the struct or enum. + pub generics: &'a syn::Generics, + /// Original input. + pub original: &'a syn::DeriveInput, +} + +/// The fields of a struct or enum. +/// +/// Analogous to `syn::Data`. +pub enum Data<'a> { + Enum(Vec>), + Struct(Style, Vec>), +} + +/// A variant of an enum. +pub struct Variant<'a> { + pub ident: syn::Ident, + pub attrs: attr::Variant, + pub style: Style, + pub fields: Vec>, + pub original: &'a syn::Variant, +} + +/// A field of a struct. +pub struct Field<'a> { + pub member: syn::Member, + pub attrs: attr::Field, + pub ty: &'a syn::Type, + pub original: &'a syn::Field, +} + +#[derive(Copy, Clone)] +pub enum Style { + /// Named fields. + Struct, + /// Many unnamed fields. + Tuple, + /// One unnamed field. + Newtype, + /// No fields. + Unit, +} + +impl<'a> Container<'a> { + /// Convert the raw Syn ast into a parsed container object, collecting errors in `cx`. + pub fn from_ast( + cx: &Ctxt, + item: &'a syn::DeriveInput, + derive: Derive, + ) -> Option> { + let mut attrs = attr::Container::from_ast(cx, item); + + let mut data = match &item.data { + syn::Data::Enum(data) => Data::Enum(enum_from_ast(cx, &data.variants, attrs.default())), + syn::Data::Struct(data) => { + let (style, fields) = struct_from_ast(cx, &data.fields, None, attrs.default()); + Data::Struct(style, fields) + } + syn::Data::Union(_) => { + cx.error_spanned_by(item, "Serde does not support derive for unions"); + return None; + } + }; + + let mut has_flatten = false; + match &mut data { + Data::Enum(variants) => { + for variant in variants { + variant.attrs.rename_by_rules(attrs.rename_all_rules()); + for field in &mut variant.fields { + if field.attrs.flatten() { + has_flatten = true; + } + field + .attrs + .rename_by_rules(variant.attrs.rename_all_rules()); + } + } + } + Data::Struct(_, fields) => { + for field in fields { + if field.attrs.flatten() { + has_flatten = true; + } + field.attrs.rename_by_rules(attrs.rename_all_rules()); + } + } + } + + if has_flatten { + attrs.mark_has_flatten(); + } + + let mut item = Container { + ident: item.ident.clone(), + attrs, + data, + generics: &item.generics, + original: item, + }; + check::check(cx, &mut item, derive); + Some(item) + } +} + +impl<'a> Data<'a> { + pub fn all_fields(&'a self) -> Box> + 'a> { + match self { + Data::Enum(variants) => { + Box::new(variants.iter().flat_map(|variant| variant.fields.iter())) + } + Data::Struct(_, fields) => Box::new(fields.iter()), + } + } + + pub fn has_getter(&self) -> bool { + self.all_fields().any(|f| f.attrs.getter().is_some()) + } +} + +fn enum_from_ast<'a>( + cx: &Ctxt, + variants: &'a Punctuated, + container_default: &attr::Default, +) -> Vec> { + variants + .iter() + .map(|variant| { + let attrs = attr::Variant::from_ast(cx, variant); + let (style, fields) = + struct_from_ast(cx, &variant.fields, Some(&attrs), container_default); + Variant { + ident: variant.ident.clone(), + attrs, + style, + fields, + original: variant, + } + }) + .collect() +} + +fn struct_from_ast<'a>( + cx: &Ctxt, + fields: &'a syn::Fields, + attrs: Option<&attr::Variant>, + container_default: &attr::Default, +) -> (Style, Vec>) { + match fields { + syn::Fields::Named(fields) => ( + Style::Struct, + fields_from_ast(cx, &fields.named, attrs, container_default), + ), + syn::Fields::Unnamed(fields) if fields.unnamed.len() == 1 => ( + Style::Newtype, + fields_from_ast(cx, &fields.unnamed, attrs, container_default), + ), + syn::Fields::Unnamed(fields) => ( + Style::Tuple, + fields_from_ast(cx, &fields.unnamed, attrs, container_default), + ), + syn::Fields::Unit => (Style::Unit, Vec::new()), + } +} + +fn fields_from_ast<'a>( + cx: &Ctxt, + fields: &'a Punctuated, + attrs: Option<&attr::Variant>, + container_default: &attr::Default, +) -> Vec> { + fields + .iter() + .enumerate() + .map(|(i, field)| Field { + member: match &field.ident { + Some(ident) => syn::Member::Named(ident.clone()), + None => syn::Member::Unnamed(i.into()), + }, + attrs: attr::Field::from_ast(cx, i, field, attrs, container_default), + ty: &field.ty, + original: field, + }) + .collect() +} diff --git a/rust/serde_derive/internals/attr.rs b/rust/serde_derive/internals/attr.rs new file mode 100644 index 00000000000000..d013215a5e91c5 --- /dev/null +++ b/rust/serde_derive/internals/attr.rs @@ -0,0 +1,1908 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use internals::symbol::*; +use internals::{ungroup, Ctxt}; +use proc_macro2::{Spacing, Span, TokenStream, TokenTree}; +use quote::ToTokens; +use std::borrow::Cow; +use std::collections::BTreeSet; +use std::iter::FromIterator; +use syn; +use syn::parse::ParseStream; +use syn::punctuated::Punctuated; +use syn::Meta::{List, NameValue, Path}; +use syn::NestedMeta::{Lit, Meta}; +use syn::{Ident, Lifetime}; + +// This module handles parsing of `#[serde(...)]` attributes. The entrypoints +// are `attr::Container::from_ast`, `attr::Variant::from_ast`, and +// `attr::Field::from_ast`. Each returns an instance of the corresponding +// struct. Note that none of them return a Result. Unrecognized, malformed, or +// duplicated attributes result in a span_err but otherwise are ignored. The +// user will see errors simultaneously for all bad attributes in the crate +// rather than just the first. + +pub use internals::case::RenameRule; + +struct Attr<'c, T> { + cx: &'c Ctxt, + name: Symbol, + tokens: TokenStream, + value: Option, +} + +impl<'c, T> Attr<'c, T> { + fn none(cx: &'c Ctxt, name: Symbol) -> Self { + Attr { + cx, + name, + tokens: TokenStream::new(), + value: None, + } + } + + fn set(&mut self, obj: A, value: T) { + let tokens = obj.into_token_stream(); + + if self.value.is_some() { + self.cx + .error_spanned_by(tokens, format!("duplicate serde attribute `{}`", self.name)); + } else { + self.tokens = tokens; + self.value = Some(value); + } + } + + fn set_opt(&mut self, obj: A, value: Option) { + if let Some(value) = value { + self.set(obj, value); + } + } + + fn set_if_none(&mut self, value: T) { + if self.value.is_none() { + self.value = Some(value); + } + } + + fn get(self) -> Option { + self.value + } + + fn get_with_tokens(self) -> Option<(TokenStream, T)> { + match self.value { + Some(v) => Some((self.tokens, v)), + None => None, + } + } +} + +struct BoolAttr<'c>(Attr<'c, ()>); + +impl<'c> BoolAttr<'c> { + fn none(cx: &'c Ctxt, name: Symbol) -> Self { + BoolAttr(Attr::none(cx, name)) + } + + fn set_true(&mut self, obj: A) { + self.0.set(obj, ()); + } + + fn get(&self) -> bool { + self.0.value.is_some() + } +} + +struct VecAttr<'c, T> { + cx: &'c Ctxt, + name: Symbol, + first_dup_tokens: TokenStream, + values: Vec, +} + +impl<'c, T> VecAttr<'c, T> { + fn none(cx: &'c Ctxt, name: Symbol) -> Self { + VecAttr { + cx, + name, + first_dup_tokens: TokenStream::new(), + values: Vec::new(), + } + } + + fn insert(&mut self, obj: A, value: T) { + if self.values.len() == 1 { + self.first_dup_tokens = obj.into_token_stream(); + } + self.values.push(value); + } + + fn at_most_one(mut self) -> Result, ()> { + if self.values.len() > 1 { + let dup_token = self.first_dup_tokens; + self.cx.error_spanned_by( + dup_token, + format!("duplicate serde attribute `{}`", self.name), + ); + Err(()) + } else { + Ok(self.values.pop()) + } + } + + fn get(self) -> Vec { + self.values + } +} + +pub struct Name { + serialize: String, + serialize_renamed: bool, + deserialize: String, + deserialize_renamed: bool, + deserialize_aliases: Vec, +} + +fn unraw(ident: &Ident) -> String { + ident.to_string().trim_start_matches("r#").to_owned() +} + +impl Name { + fn from_attrs( + source_name: String, + ser_name: Attr, + de_name: Attr, + de_aliases: Option>, + ) -> Name { + let deserialize_aliases = match de_aliases { + Some(de_aliases) => { + let mut alias_list = BTreeSet::new(); + for alias_name in de_aliases.get() { + alias_list.insert(alias_name); + } + alias_list.into_iter().collect() + } + None => Vec::new(), + }; + + let ser_name = ser_name.get(); + let ser_renamed = ser_name.is_some(); + let de_name = de_name.get(); + let de_renamed = de_name.is_some(); + Name { + serialize: ser_name.unwrap_or_else(|| source_name.clone()), + serialize_renamed: ser_renamed, + deserialize: de_name.unwrap_or(source_name), + deserialize_renamed: de_renamed, + deserialize_aliases, + } + } + + /// Return the container name for the container when serializing. + pub fn serialize_name(&self) -> String { + self.serialize.clone() + } + + /// Return the container name for the container when deserializing. + pub fn deserialize_name(&self) -> String { + self.deserialize.clone() + } + + fn deserialize_aliases(&self) -> Vec { + let mut aliases = self.deserialize_aliases.clone(); + let main_name = self.deserialize_name(); + if !aliases.contains(&main_name) { + aliases.push(main_name); + } + aliases + } +} + +pub struct RenameAllRules { + serialize: RenameRule, + deserialize: RenameRule, +} + +/// Represents struct or enum attribute information. +pub struct Container { + name: Name, + transparent: bool, + deny_unknown_fields: bool, + default: Default, + rename_all_rules: RenameAllRules, + ser_bound: Option>, + de_bound: Option>, + tag: TagType, + type_from: Option, + type_try_from: Option, + type_into: Option, + remote: Option, + identifier: Identifier, + has_flatten: bool, + serde_path: Option, + is_packed: bool, + /// Error message generated when type can't be deserialized + expecting: Option, +} + +/// Styles of representing an enum. +pub enum TagType { + /// The default. + /// + /// ```json + /// {"variant1": {"key1": "value1", "key2": "value2"}} + /// ``` + External, + + /// `#[serde(tag = "type")]` + /// + /// ```json + /// {"type": "variant1", "key1": "value1", "key2": "value2"} + /// ``` + Internal { tag: String }, + + /// `#[serde(tag = "t", content = "c")]` + /// + /// ```json + /// {"t": "variant1", "c": {"key1": "value1", "key2": "value2"}} + /// ``` + Adjacent { tag: String, content: String }, + + /// `#[serde(untagged)]` + /// + /// ```json + /// {"key1": "value1", "key2": "value2"} + /// ``` + None, +} + +/// Whether this enum represents the fields of a struct or the variants of an +/// enum. +#[derive(Copy, Clone)] +pub enum Identifier { + /// It does not. + No, + + /// This enum represents the fields of a struct. All of the variants must be + /// unit variants, except possibly one which is annotated with + /// `#[serde(other)]` and is a newtype variant. + Field, + + /// This enum represents the variants of an enum. All of the variants must + /// be unit variants. + Variant, +} + +impl Identifier { + #[cfg(feature = "deserialize_in_place")] + pub fn is_some(self) -> bool { + match self { + Identifier::No => false, + Identifier::Field | Identifier::Variant => true, + } + } +} + +impl Container { + /// Extract out the `#[serde(...)]` attributes from an item. + pub fn from_ast(cx: &Ctxt, item: &syn::DeriveInput) -> Self { + let mut ser_name = Attr::none(cx, RENAME); + let mut de_name = Attr::none(cx, RENAME); + let mut transparent = BoolAttr::none(cx, TRANSPARENT); + let mut deny_unknown_fields = BoolAttr::none(cx, DENY_UNKNOWN_FIELDS); + let mut default = Attr::none(cx, DEFAULT); + let mut rename_all_ser_rule = Attr::none(cx, RENAME_ALL); + let mut rename_all_de_rule = Attr::none(cx, RENAME_ALL); + let mut ser_bound = Attr::none(cx, BOUND); + let mut de_bound = Attr::none(cx, BOUND); + let mut untagged = BoolAttr::none(cx, UNTAGGED); + let mut internal_tag = Attr::none(cx, TAG); + let mut content = Attr::none(cx, CONTENT); + let mut type_from = Attr::none(cx, FROM); + let mut type_try_from = Attr::none(cx, TRY_FROM); + let mut type_into = Attr::none(cx, INTO); + let mut remote = Attr::none(cx, REMOTE); + let mut field_identifier = BoolAttr::none(cx, FIELD_IDENTIFIER); + let mut variant_identifier = BoolAttr::none(cx, VARIANT_IDENTIFIER); + let mut serde_path = Attr::none(cx, CRATE); + let mut expecting = Attr::none(cx, EXPECTING); + + for meta_item in item + .attrs + .iter() + .flat_map(|attr| get_serde_meta_items(cx, attr)) + .flatten() + { + match &meta_item { + // Parse `#[serde(rename = "foo")]` + Meta(NameValue(m)) if m.path == RENAME => { + if let Ok(s) = get_lit_str(cx, RENAME, &m.lit) { + ser_name.set(&m.path, s.value()); + de_name.set(&m.path, s.value()); + } + } + + // Parse `#[serde(rename(serialize = "foo", deserialize = "bar"))]` + Meta(List(m)) if m.path == RENAME => { + if let Ok((ser, de)) = get_renames(cx, &m.nested) { + ser_name.set_opt(&m.path, ser.map(syn::LitStr::value)); + de_name.set_opt(&m.path, de.map(syn::LitStr::value)); + } + } + + // Parse `#[serde(rename_all = "foo")]` + Meta(NameValue(m)) if m.path == RENAME_ALL => { + if let Ok(s) = get_lit_str(cx, RENAME_ALL, &m.lit) { + match RenameRule::from_str(&s.value()) { + Ok(rename_rule) => { + rename_all_ser_rule.set(&m.path, rename_rule); + rename_all_de_rule.set(&m.path, rename_rule); + } + Err(err) => cx.error_spanned_by(s, err), + } + } + } + + // Parse `#[serde(rename_all(serialize = "foo", deserialize = "bar"))]` + Meta(List(m)) if m.path == RENAME_ALL => { + if let Ok((ser, de)) = get_renames(cx, &m.nested) { + if let Some(ser) = ser { + match RenameRule::from_str(&ser.value()) { + Ok(rename_rule) => rename_all_ser_rule.set(&m.path, rename_rule), + Err(err) => cx.error_spanned_by(ser, err), + } + } + if let Some(de) = de { + match RenameRule::from_str(&de.value()) { + Ok(rename_rule) => rename_all_de_rule.set(&m.path, rename_rule), + Err(err) => cx.error_spanned_by(de, err), + } + } + } + } + + // Parse `#[serde(transparent)]` + Meta(Path(word)) if word == TRANSPARENT => { + transparent.set_true(word); + } + + // Parse `#[serde(deny_unknown_fields)]` + Meta(Path(word)) if word == DENY_UNKNOWN_FIELDS => { + deny_unknown_fields.set_true(word); + } + + // Parse `#[serde(default)]` + Meta(Path(word)) if word == DEFAULT => match &item.data { + syn::Data::Struct(syn::DataStruct { fields, .. }) => match fields { + syn::Fields::Named(_) => { + default.set(word, Default::Default); + } + syn::Fields::Unnamed(_) | syn::Fields::Unit => { + let msg = + "#[serde(default)] can only be used on structs with named fields"; + cx.error_spanned_by(fields, msg); + } + }, + syn::Data::Enum(syn::DataEnum { enum_token, .. }) => { + let msg = "#[serde(default)] can only be used on structs with named fields"; + cx.error_spanned_by(enum_token, msg); + } + syn::Data::Union(syn::DataUnion { union_token, .. }) => { + let msg = "#[serde(default)] can only be used on structs with named fields"; + cx.error_spanned_by(union_token, msg); + } + }, + + // Parse `#[serde(default = "...")]` + Meta(NameValue(m)) if m.path == DEFAULT => { + if let Ok(path) = parse_lit_into_expr_path(cx, DEFAULT, &m.lit) { + match &item.data { + syn::Data::Struct(syn::DataStruct { fields, .. }) => match fields { + syn::Fields::Named(_) => { + default.set(&m.path, Default::Path(path)); + } + syn::Fields::Unnamed(_) | syn::Fields::Unit => { + let msg = "#[serde(default = \"...\")] can only be used on structs with named fields"; + cx.error_spanned_by(fields, msg); + } + }, + syn::Data::Enum(syn::DataEnum { enum_token, .. }) => { + let msg = "#[serde(default = \"...\")] can only be used on structs with named fields"; + cx.error_spanned_by(enum_token, msg); + } + syn::Data::Union(syn::DataUnion { union_token, .. }) => { + let msg = "#[serde(default = \"...\")] can only be used on structs with named fields"; + cx.error_spanned_by(union_token, msg); + } + } + } + } + + // Parse `#[serde(bound = "T: SomeBound")]` + Meta(NameValue(m)) if m.path == BOUND => { + if let Ok(where_predicates) = parse_lit_into_where(cx, BOUND, BOUND, &m.lit) { + ser_bound.set(&m.path, where_predicates.clone()); + de_bound.set(&m.path, where_predicates); + } + } + + // Parse `#[serde(bound(serialize = "...", deserialize = "..."))]` + Meta(List(m)) if m.path == BOUND => { + if let Ok((ser, de)) = get_where_predicates(cx, &m.nested) { + ser_bound.set_opt(&m.path, ser); + de_bound.set_opt(&m.path, de); + } + } + + // Parse `#[serde(untagged)]` + Meta(Path(word)) if word == UNTAGGED => match item.data { + syn::Data::Enum(_) => { + untagged.set_true(word); + } + syn::Data::Struct(syn::DataStruct { struct_token, .. }) => { + let msg = "#[serde(untagged)] can only be used on enums"; + cx.error_spanned_by(struct_token, msg); + } + syn::Data::Union(syn::DataUnion { union_token, .. }) => { + let msg = "#[serde(untagged)] can only be used on enums"; + cx.error_spanned_by(union_token, msg); + } + }, + + // Parse `#[serde(tag = "type")]` + Meta(NameValue(m)) if m.path == TAG => { + if let Ok(s) = get_lit_str(cx, TAG, &m.lit) { + match &item.data { + syn::Data::Enum(_) => { + internal_tag.set(&m.path, s.value()); + } + syn::Data::Struct(syn::DataStruct { fields, .. }) => match fields { + syn::Fields::Named(_) => { + internal_tag.set(&m.path, s.value()); + } + syn::Fields::Unnamed(_) | syn::Fields::Unit => { + let msg = "#[serde(tag = \"...\")] can only be used on enums and structs with named fields"; + cx.error_spanned_by(fields, msg); + } + }, + syn::Data::Union(syn::DataUnion { union_token, .. }) => { + let msg = "#[serde(tag = \"...\")] can only be used on enums and structs with named fields"; + cx.error_spanned_by(union_token, msg); + } + } + } + } + + // Parse `#[serde(content = "c")]` + Meta(NameValue(m)) if m.path == CONTENT => { + if let Ok(s) = get_lit_str(cx, CONTENT, &m.lit) { + match &item.data { + syn::Data::Enum(_) => { + content.set(&m.path, s.value()); + } + syn::Data::Struct(syn::DataStruct { struct_token, .. }) => { + let msg = "#[serde(content = \"...\")] can only be used on enums"; + cx.error_spanned_by(struct_token, msg); + } + syn::Data::Union(syn::DataUnion { union_token, .. }) => { + let msg = "#[serde(content = \"...\")] can only be used on enums"; + cx.error_spanned_by(union_token, msg); + } + } + } + } + + // Parse `#[serde(from = "Type")]` + Meta(NameValue(m)) if m.path == FROM => { + if let Ok(from_ty) = parse_lit_into_ty(cx, FROM, &m.lit) { + type_from.set_opt(&m.path, Some(from_ty)); + } + } + + // Parse `#[serde(try_from = "Type")]` + Meta(NameValue(m)) if m.path == TRY_FROM => { + if let Ok(try_from_ty) = parse_lit_into_ty(cx, TRY_FROM, &m.lit) { + type_try_from.set_opt(&m.path, Some(try_from_ty)); + } + } + + // Parse `#[serde(into = "Type")]` + Meta(NameValue(m)) if m.path == INTO => { + if let Ok(into_ty) = parse_lit_into_ty(cx, INTO, &m.lit) { + type_into.set_opt(&m.path, Some(into_ty)); + } + } + + // Parse `#[serde(remote = "...")]` + Meta(NameValue(m)) if m.path == REMOTE => { + if let Ok(path) = parse_lit_into_path(cx, REMOTE, &m.lit) { + if is_primitive_path(&path, "Self") { + remote.set(&m.path, item.ident.clone().into()); + } else { + remote.set(&m.path, path); + } + } + } + + // Parse `#[serde(field_identifier)]` + Meta(Path(word)) if word == FIELD_IDENTIFIER => { + field_identifier.set_true(word); + } + + // Parse `#[serde(variant_identifier)]` + Meta(Path(word)) if word == VARIANT_IDENTIFIER => { + variant_identifier.set_true(word); + } + + // Parse `#[serde(crate = "foo")]` + Meta(NameValue(m)) if m.path == CRATE => { + if let Ok(path) = parse_lit_into_path(cx, CRATE, &m.lit) { + serde_path.set(&m.path, path); + } + } + + // Parse `#[serde(expecting = "a message")]` + Meta(NameValue(m)) if m.path == EXPECTING => { + if let Ok(s) = get_lit_str(cx, EXPECTING, &m.lit) { + expecting.set(&m.path, s.value()); + } + } + + Meta(meta_item) => { + let path = meta_item + .path() + .into_token_stream() + .to_string() + .replace(' ', ""); + let msg = format!("unknown serde container attribute `{}`", path); + cx.error_spanned_by(meta_item.path(), msg); + } + + Lit(lit) => { + let msg = "unexpected literal in serde container attribute"; + cx.error_spanned_by(lit, msg); + } + } + } + + let mut is_packed = false; + for attr in &item.attrs { + if attr.path == REPR { + let _ = attr.parse_args_with(|input: ParseStream| { + while let Some(token) = input.parse()? { + if let TokenTree::Ident(ident) = token { + is_packed |= ident == "packed"; + } + } + Ok(()) + }); + } + } + + Container { + name: Name::from_attrs(unraw(&item.ident), ser_name, de_name, None), + transparent: transparent.get(), + deny_unknown_fields: deny_unknown_fields.get(), + default: default.get().unwrap_or(Default::None), + rename_all_rules: RenameAllRules { + serialize: rename_all_ser_rule.get().unwrap_or(RenameRule::None), + deserialize: rename_all_de_rule.get().unwrap_or(RenameRule::None), + }, + ser_bound: ser_bound.get(), + de_bound: de_bound.get(), + tag: decide_tag(cx, item, untagged, internal_tag, content), + type_from: type_from.get(), + type_try_from: type_try_from.get(), + type_into: type_into.get(), + remote: remote.get(), + identifier: decide_identifier(cx, item, field_identifier, variant_identifier), + has_flatten: false, + serde_path: serde_path.get(), + is_packed, + expecting: expecting.get(), + } + } + + pub fn name(&self) -> &Name { + &self.name + } + + pub fn rename_all_rules(&self) -> &RenameAllRules { + &self.rename_all_rules + } + + pub fn transparent(&self) -> bool { + self.transparent + } + + pub fn deny_unknown_fields(&self) -> bool { + self.deny_unknown_fields + } + + pub fn default(&self) -> &Default { + &self.default + } + + pub fn ser_bound(&self) -> Option<&[syn::WherePredicate]> { + self.ser_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn de_bound(&self) -> Option<&[syn::WherePredicate]> { + self.de_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn tag(&self) -> &TagType { + &self.tag + } + + pub fn type_from(&self) -> Option<&syn::Type> { + self.type_from.as_ref() + } + + pub fn type_try_from(&self) -> Option<&syn::Type> { + self.type_try_from.as_ref() + } + + pub fn type_into(&self) -> Option<&syn::Type> { + self.type_into.as_ref() + } + + pub fn remote(&self) -> Option<&syn::Path> { + self.remote.as_ref() + } + + pub fn is_packed(&self) -> bool { + self.is_packed + } + + pub fn identifier(&self) -> Identifier { + self.identifier + } + + pub fn has_flatten(&self) -> bool { + self.has_flatten + } + + pub fn mark_has_flatten(&mut self) { + self.has_flatten = true; + } + + pub fn custom_serde_path(&self) -> Option<&syn::Path> { + self.serde_path.as_ref() + } + + pub fn serde_path(&self) -> Cow { + self.custom_serde_path() + .map_or_else(|| Cow::Owned(parse_quote!(_serde)), Cow::Borrowed) + } + + /// Error message generated when type can't be deserialized. + /// If `None`, default message will be used + pub fn expecting(&self) -> Option<&str> { + self.expecting.as_ref().map(String::as_ref) + } +} + +fn decide_tag( + cx: &Ctxt, + item: &syn::DeriveInput, + untagged: BoolAttr, + internal_tag: Attr, + content: Attr, +) -> TagType { + match ( + untagged.0.get_with_tokens(), + internal_tag.get_with_tokens(), + content.get_with_tokens(), + ) { + (None, None, None) => TagType::External, + (Some(_), None, None) => TagType::None, + (None, Some((_, tag)), None) => { + // Check that there are no tuple variants. + if let syn::Data::Enum(data) = &item.data { + for variant in &data.variants { + match &variant.fields { + syn::Fields::Named(_) | syn::Fields::Unit => {} + syn::Fields::Unnamed(fields) => { + if fields.unnamed.len() != 1 { + let msg = + "#[serde(tag = \"...\")] cannot be used with tuple variants"; + cx.error_spanned_by(variant, msg); + break; + } + } + } + } + } + TagType::Internal { tag } + } + (Some((untagged_tokens, _)), Some((tag_tokens, _)), None) => { + let msg = "enum cannot be both untagged and internally tagged"; + cx.error_spanned_by(untagged_tokens, msg); + cx.error_spanned_by(tag_tokens, msg); + TagType::External // doesn't matter, will error + } + (None, None, Some((content_tokens, _))) => { + let msg = "#[serde(tag = \"...\", content = \"...\")] must be used together"; + cx.error_spanned_by(content_tokens, msg); + TagType::External + } + (Some((untagged_tokens, _)), None, Some((content_tokens, _))) => { + let msg = "untagged enum cannot have #[serde(content = \"...\")]"; + cx.error_spanned_by(untagged_tokens, msg); + cx.error_spanned_by(content_tokens, msg); + TagType::External + } + (None, Some((_, tag)), Some((_, content))) => TagType::Adjacent { tag, content }, + (Some((untagged_tokens, _)), Some((tag_tokens, _)), Some((content_tokens, _))) => { + let msg = "untagged enum cannot have #[serde(tag = \"...\", content = \"...\")]"; + cx.error_spanned_by(untagged_tokens, msg); + cx.error_spanned_by(tag_tokens, msg); + cx.error_spanned_by(content_tokens, msg); + TagType::External + } + } +} + +fn decide_identifier( + cx: &Ctxt, + item: &syn::DeriveInput, + field_identifier: BoolAttr, + variant_identifier: BoolAttr, +) -> Identifier { + match ( + &item.data, + field_identifier.0.get_with_tokens(), + variant_identifier.0.get_with_tokens(), + ) { + (_, None, None) => Identifier::No, + (_, Some((field_identifier_tokens, _)), Some((variant_identifier_tokens, _))) => { + let msg = + "#[serde(field_identifier)] and #[serde(variant_identifier)] cannot both be set"; + cx.error_spanned_by(field_identifier_tokens, msg); + cx.error_spanned_by(variant_identifier_tokens, msg); + Identifier::No + } + (syn::Data::Enum(_), Some(_), None) => Identifier::Field, + (syn::Data::Enum(_), None, Some(_)) => Identifier::Variant, + (syn::Data::Struct(syn::DataStruct { struct_token, .. }), Some(_), None) => { + let msg = "#[serde(field_identifier)] can only be used on an enum"; + cx.error_spanned_by(struct_token, msg); + Identifier::No + } + (syn::Data::Union(syn::DataUnion { union_token, .. }), Some(_), None) => { + let msg = "#[serde(field_identifier)] can only be used on an enum"; + cx.error_spanned_by(union_token, msg); + Identifier::No + } + (syn::Data::Struct(syn::DataStruct { struct_token, .. }), None, Some(_)) => { + let msg = "#[serde(variant_identifier)] can only be used on an enum"; + cx.error_spanned_by(struct_token, msg); + Identifier::No + } + (syn::Data::Union(syn::DataUnion { union_token, .. }), None, Some(_)) => { + let msg = "#[serde(variant_identifier)] can only be used on an enum"; + cx.error_spanned_by(union_token, msg); + Identifier::No + } + } +} + +/// Represents variant attribute information +pub struct Variant { + name: Name, + rename_all_rules: RenameAllRules, + ser_bound: Option>, + de_bound: Option>, + skip_deserializing: bool, + skip_serializing: bool, + other: bool, + serialize_with: Option, + deserialize_with: Option, + borrow: Option, +} + +struct BorrowAttribute { + path: syn::Path, + lifetimes: Option>, +} + +impl Variant { + pub fn from_ast(cx: &Ctxt, variant: &syn::Variant) -> Self { + let mut ser_name = Attr::none(cx, RENAME); + let mut de_name = Attr::none(cx, RENAME); + let mut de_aliases = VecAttr::none(cx, RENAME); + let mut skip_deserializing = BoolAttr::none(cx, SKIP_DESERIALIZING); + let mut skip_serializing = BoolAttr::none(cx, SKIP_SERIALIZING); + let mut rename_all_ser_rule = Attr::none(cx, RENAME_ALL); + let mut rename_all_de_rule = Attr::none(cx, RENAME_ALL); + let mut ser_bound = Attr::none(cx, BOUND); + let mut de_bound = Attr::none(cx, BOUND); + let mut other = BoolAttr::none(cx, OTHER); + let mut serialize_with = Attr::none(cx, SERIALIZE_WITH); + let mut deserialize_with = Attr::none(cx, DESERIALIZE_WITH); + let mut borrow = Attr::none(cx, BORROW); + + for meta_item in variant + .attrs + .iter() + .flat_map(|attr| get_serde_meta_items(cx, attr)) + .flatten() + { + match &meta_item { + // Parse `#[serde(rename = "foo")]` + Meta(NameValue(m)) if m.path == RENAME => { + if let Ok(s) = get_lit_str(cx, RENAME, &m.lit) { + ser_name.set(&m.path, s.value()); + de_name.set_if_none(s.value()); + de_aliases.insert(&m.path, s.value()); + } + } + + // Parse `#[serde(rename(serialize = "foo", deserialize = "bar"))]` + Meta(List(m)) if m.path == RENAME => { + if let Ok((ser, de)) = get_multiple_renames(cx, &m.nested) { + ser_name.set_opt(&m.path, ser.map(syn::LitStr::value)); + for de_value in de { + de_name.set_if_none(de_value.value()); + de_aliases.insert(&m.path, de_value.value()); + } + } + } + + // Parse `#[serde(alias = "foo")]` + Meta(NameValue(m)) if m.path == ALIAS => { + if let Ok(s) = get_lit_str(cx, ALIAS, &m.lit) { + de_aliases.insert(&m.path, s.value()); + } + } + + // Parse `#[serde(rename_all = "foo")]` + Meta(NameValue(m)) if m.path == RENAME_ALL => { + if let Ok(s) = get_lit_str(cx, RENAME_ALL, &m.lit) { + match RenameRule::from_str(&s.value()) { + Ok(rename_rule) => { + rename_all_ser_rule.set(&m.path, rename_rule); + rename_all_de_rule.set(&m.path, rename_rule); + } + Err(err) => cx.error_spanned_by(s, err), + } + } + } + + // Parse `#[serde(rename_all(serialize = "foo", deserialize = "bar"))]` + Meta(List(m)) if m.path == RENAME_ALL => { + if let Ok((ser, de)) = get_renames(cx, &m.nested) { + if let Some(ser) = ser { + match RenameRule::from_str(&ser.value()) { + Ok(rename_rule) => rename_all_ser_rule.set(&m.path, rename_rule), + Err(err) => cx.error_spanned_by(ser, err), + } + } + if let Some(de) = de { + match RenameRule::from_str(&de.value()) { + Ok(rename_rule) => rename_all_de_rule.set(&m.path, rename_rule), + Err(err) => cx.error_spanned_by(de, err), + } + } + } + } + + // Parse `#[serde(skip)]` + Meta(Path(word)) if word == SKIP => { + skip_serializing.set_true(word); + skip_deserializing.set_true(word); + } + + // Parse `#[serde(skip_deserializing)]` + Meta(Path(word)) if word == SKIP_DESERIALIZING => { + skip_deserializing.set_true(word); + } + + // Parse `#[serde(skip_serializing)]` + Meta(Path(word)) if word == SKIP_SERIALIZING => { + skip_serializing.set_true(word); + } + + // Parse `#[serde(other)]` + Meta(Path(word)) if word == OTHER => { + other.set_true(word); + } + + // Parse `#[serde(bound = "T: SomeBound")]` + Meta(NameValue(m)) if m.path == BOUND => { + if let Ok(where_predicates) = parse_lit_into_where(cx, BOUND, BOUND, &m.lit) { + ser_bound.set(&m.path, where_predicates.clone()); + de_bound.set(&m.path, where_predicates); + } + } + + // Parse `#[serde(bound(serialize = "...", deserialize = "..."))]` + Meta(List(m)) if m.path == BOUND => { + if let Ok((ser, de)) = get_where_predicates(cx, &m.nested) { + ser_bound.set_opt(&m.path, ser); + de_bound.set_opt(&m.path, de); + } + } + + // Parse `#[serde(with = "...")]` + Meta(NameValue(m)) if m.path == WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, WITH, &m.lit) { + let mut ser_path = path.clone(); + ser_path + .path + .segments + .push(Ident::new("serialize", Span::call_site()).into()); + serialize_with.set(&m.path, ser_path); + let mut de_path = path; + de_path + .path + .segments + .push(Ident::new("deserialize", Span::call_site()).into()); + deserialize_with.set(&m.path, de_path); + } + } + + // Parse `#[serde(serialize_with = "...")]` + Meta(NameValue(m)) if m.path == SERIALIZE_WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, SERIALIZE_WITH, &m.lit) { + serialize_with.set(&m.path, path); + } + } + + // Parse `#[serde(deserialize_with = "...")]` + Meta(NameValue(m)) if m.path == DESERIALIZE_WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, DESERIALIZE_WITH, &m.lit) { + deserialize_with.set(&m.path, path); + } + } + + // Parse `#[serde(borrow)]` + Meta(Path(word)) if word == BORROW => match &variant.fields { + syn::Fields::Unnamed(fields) if fields.unnamed.len() == 1 => { + borrow.set( + word, + BorrowAttribute { + path: word.clone(), + lifetimes: None, + }, + ); + } + _ => { + let msg = "#[serde(borrow)] may only be used on newtype variants"; + cx.error_spanned_by(variant, msg); + } + }, + + // Parse `#[serde(borrow = "'a + 'b")]` + Meta(NameValue(m)) if m.path == BORROW => match &variant.fields { + syn::Fields::Unnamed(fields) if fields.unnamed.len() == 1 => { + if let Ok(lifetimes) = parse_lit_into_lifetimes(cx, &m.lit) { + borrow.set( + &m.path, + BorrowAttribute { + path: m.path.clone(), + lifetimes: Some(lifetimes), + }, + ); + } + } + _ => { + let msg = "#[serde(borrow)] may only be used on newtype variants"; + cx.error_spanned_by(variant, msg); + } + }, + + Meta(meta_item) => { + let path = meta_item + .path() + .into_token_stream() + .to_string() + .replace(' ', ""); + let msg = format!("unknown serde variant attribute `{}`", path); + cx.error_spanned_by(meta_item.path(), msg); + } + + Lit(lit) => { + let msg = "unexpected literal in serde variant attribute"; + cx.error_spanned_by(lit, msg); + } + } + } + + Variant { + name: Name::from_attrs(unraw(&variant.ident), ser_name, de_name, Some(de_aliases)), + rename_all_rules: RenameAllRules { + serialize: rename_all_ser_rule.get().unwrap_or(RenameRule::None), + deserialize: rename_all_de_rule.get().unwrap_or(RenameRule::None), + }, + ser_bound: ser_bound.get(), + de_bound: de_bound.get(), + skip_deserializing: skip_deserializing.get(), + skip_serializing: skip_serializing.get(), + other: other.get(), + serialize_with: serialize_with.get(), + deserialize_with: deserialize_with.get(), + borrow: borrow.get(), + } + } + + pub fn name(&self) -> &Name { + &self.name + } + + pub fn aliases(&self) -> Vec { + self.name.deserialize_aliases() + } + + pub fn rename_by_rules(&mut self, rules: &RenameAllRules) { + if !self.name.serialize_renamed { + self.name.serialize = rules.serialize.apply_to_variant(&self.name.serialize); + } + if !self.name.deserialize_renamed { + self.name.deserialize = rules.deserialize.apply_to_variant(&self.name.deserialize); + } + } + + pub fn rename_all_rules(&self) -> &RenameAllRules { + &self.rename_all_rules + } + + pub fn ser_bound(&self) -> Option<&[syn::WherePredicate]> { + self.ser_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn de_bound(&self) -> Option<&[syn::WherePredicate]> { + self.de_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn skip_deserializing(&self) -> bool { + self.skip_deserializing + } + + pub fn skip_serializing(&self) -> bool { + self.skip_serializing + } + + pub fn other(&self) -> bool { + self.other + } + + pub fn serialize_with(&self) -> Option<&syn::ExprPath> { + self.serialize_with.as_ref() + } + + pub fn deserialize_with(&self) -> Option<&syn::ExprPath> { + self.deserialize_with.as_ref() + } +} + +/// Represents field attribute information +pub struct Field { + name: Name, + skip_serializing: bool, + skip_deserializing: bool, + skip_serializing_if: Option, + default: Default, + serialize_with: Option, + deserialize_with: Option, + ser_bound: Option>, + de_bound: Option>, + borrowed_lifetimes: BTreeSet, + getter: Option, + flatten: bool, + transparent: bool, +} + +/// Represents the default to use for a field when deserializing. +pub enum Default { + /// Field must always be specified because it does not have a default. + None, + /// The default is given by `std::default::Default::default()`. + Default, + /// The default is given by this function. + Path(syn::ExprPath), +} + +impl Default { + pub fn is_none(&self) -> bool { + match self { + Default::None => true, + Default::Default | Default::Path(_) => false, + } + } +} + +impl Field { + /// Extract out the `#[serde(...)]` attributes from a struct field. + pub fn from_ast( + cx: &Ctxt, + index: usize, + field: &syn::Field, + attrs: Option<&Variant>, + container_default: &Default, + ) -> Self { + let mut ser_name = Attr::none(cx, RENAME); + let mut de_name = Attr::none(cx, RENAME); + let mut de_aliases = VecAttr::none(cx, RENAME); + let mut skip_serializing = BoolAttr::none(cx, SKIP_SERIALIZING); + let mut skip_deserializing = BoolAttr::none(cx, SKIP_DESERIALIZING); + let mut skip_serializing_if = Attr::none(cx, SKIP_SERIALIZING_IF); + let mut default = Attr::none(cx, DEFAULT); + let mut serialize_with = Attr::none(cx, SERIALIZE_WITH); + let mut deserialize_with = Attr::none(cx, DESERIALIZE_WITH); + let mut ser_bound = Attr::none(cx, BOUND); + let mut de_bound = Attr::none(cx, BOUND); + let mut borrowed_lifetimes = Attr::none(cx, BORROW); + let mut getter = Attr::none(cx, GETTER); + let mut flatten = BoolAttr::none(cx, FLATTEN); + + let ident = match &field.ident { + Some(ident) => unraw(ident), + None => index.to_string(), + }; + + if let Some(borrow_attribute) = attrs.and_then(|variant| variant.borrow.as_ref()) { + if let Ok(borrowable) = borrowable_lifetimes(cx, &ident, field) { + if let Some(lifetimes) = &borrow_attribute.lifetimes { + for lifetime in lifetimes { + if !borrowable.contains(lifetime) { + let msg = + format!("field `{}` does not have lifetime {}", ident, lifetime); + cx.error_spanned_by(field, msg); + } + } + borrowed_lifetimes.set(&borrow_attribute.path, lifetimes.clone()); + } else { + borrowed_lifetimes.set(&borrow_attribute.path, borrowable); + } + } + } + + for meta_item in field + .attrs + .iter() + .flat_map(|attr| get_serde_meta_items(cx, attr)) + .flatten() + { + match &meta_item { + // Parse `#[serde(rename = "foo")]` + Meta(NameValue(m)) if m.path == RENAME => { + if let Ok(s) = get_lit_str(cx, RENAME, &m.lit) { + ser_name.set(&m.path, s.value()); + de_name.set_if_none(s.value()); + de_aliases.insert(&m.path, s.value()); + } + } + + // Parse `#[serde(rename(serialize = "foo", deserialize = "bar"))]` + Meta(List(m)) if m.path == RENAME => { + if let Ok((ser, de)) = get_multiple_renames(cx, &m.nested) { + ser_name.set_opt(&m.path, ser.map(syn::LitStr::value)); + for de_value in de { + de_name.set_if_none(de_value.value()); + de_aliases.insert(&m.path, de_value.value()); + } + } + } + + // Parse `#[serde(alias = "foo")]` + Meta(NameValue(m)) if m.path == ALIAS => { + if let Ok(s) = get_lit_str(cx, ALIAS, &m.lit) { + de_aliases.insert(&m.path, s.value()); + } + } + + // Parse `#[serde(default)]` + Meta(Path(word)) if word == DEFAULT => { + default.set(word, Default::Default); + } + + // Parse `#[serde(default = "...")]` + Meta(NameValue(m)) if m.path == DEFAULT => { + if let Ok(path) = parse_lit_into_expr_path(cx, DEFAULT, &m.lit) { + default.set(&m.path, Default::Path(path)); + } + } + + // Parse `#[serde(skip_serializing)]` + Meta(Path(word)) if word == SKIP_SERIALIZING => { + skip_serializing.set_true(word); + } + + // Parse `#[serde(skip_deserializing)]` + Meta(Path(word)) if word == SKIP_DESERIALIZING => { + skip_deserializing.set_true(word); + } + + // Parse `#[serde(skip)]` + Meta(Path(word)) if word == SKIP => { + skip_serializing.set_true(word); + skip_deserializing.set_true(word); + } + + // Parse `#[serde(skip_serializing_if = "...")]` + Meta(NameValue(m)) if m.path == SKIP_SERIALIZING_IF => { + if let Ok(path) = parse_lit_into_expr_path(cx, SKIP_SERIALIZING_IF, &m.lit) { + skip_serializing_if.set(&m.path, path); + } + } + + // Parse `#[serde(serialize_with = "...")]` + Meta(NameValue(m)) if m.path == SERIALIZE_WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, SERIALIZE_WITH, &m.lit) { + serialize_with.set(&m.path, path); + } + } + + // Parse `#[serde(deserialize_with = "...")]` + Meta(NameValue(m)) if m.path == DESERIALIZE_WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, DESERIALIZE_WITH, &m.lit) { + deserialize_with.set(&m.path, path); + } + } + + // Parse `#[serde(with = "...")]` + Meta(NameValue(m)) if m.path == WITH => { + if let Ok(path) = parse_lit_into_expr_path(cx, WITH, &m.lit) { + let mut ser_path = path.clone(); + ser_path + .path + .segments + .push(Ident::new("serialize", Span::call_site()).into()); + serialize_with.set(&m.path, ser_path); + let mut de_path = path; + de_path + .path + .segments + .push(Ident::new("deserialize", Span::call_site()).into()); + deserialize_with.set(&m.path, de_path); + } + } + + // Parse `#[serde(bound = "T: SomeBound")]` + Meta(NameValue(m)) if m.path == BOUND => { + if let Ok(where_predicates) = parse_lit_into_where(cx, BOUND, BOUND, &m.lit) { + ser_bound.set(&m.path, where_predicates.clone()); + de_bound.set(&m.path, where_predicates); + } + } + + // Parse `#[serde(bound(serialize = "...", deserialize = "..."))]` + Meta(List(m)) if m.path == BOUND => { + if let Ok((ser, de)) = get_where_predicates(cx, &m.nested) { + ser_bound.set_opt(&m.path, ser); + de_bound.set_opt(&m.path, de); + } + } + + // Parse `#[serde(borrow)]` + Meta(Path(word)) if word == BORROW => { + if let Ok(borrowable) = borrowable_lifetimes(cx, &ident, field) { + borrowed_lifetimes.set(word, borrowable); + } + } + + // Parse `#[serde(borrow = "'a + 'b")]` + Meta(NameValue(m)) if m.path == BORROW => { + if let Ok(lifetimes) = parse_lit_into_lifetimes(cx, &m.lit) { + if let Ok(borrowable) = borrowable_lifetimes(cx, &ident, field) { + for lifetime in &lifetimes { + if !borrowable.contains(lifetime) { + let msg = format!( + "field `{}` does not have lifetime {}", + ident, lifetime, + ); + cx.error_spanned_by(field, msg); + } + } + borrowed_lifetimes.set(&m.path, lifetimes); + } + } + } + + // Parse `#[serde(getter = "...")]` + Meta(NameValue(m)) if m.path == GETTER => { + if let Ok(path) = parse_lit_into_expr_path(cx, GETTER, &m.lit) { + getter.set(&m.path, path); + } + } + + // Parse `#[serde(flatten)]` + Meta(Path(word)) if word == FLATTEN => { + flatten.set_true(word); + } + + Meta(meta_item) => { + let path = meta_item + .path() + .into_token_stream() + .to_string() + .replace(' ', ""); + let msg = format!("unknown serde field attribute `{}`", path); + cx.error_spanned_by(meta_item.path(), msg); + } + + Lit(lit) => { + let msg = "unexpected literal in serde field attribute"; + cx.error_spanned_by(lit, msg); + } + } + } + + // Is skip_deserializing, initialize the field to Default::default() unless a + // different default is specified by `#[serde(default = "...")]` on + // ourselves or our container (e.g. the struct we are in). + if let Default::None = *container_default { + if skip_deserializing.0.value.is_some() { + default.set_if_none(Default::Default); + } + } + + let mut borrowed_lifetimes = borrowed_lifetimes.get().unwrap_or_default(); + if !borrowed_lifetimes.is_empty() { + // Cow and Cow<[u8]> never borrow by default: + // + // impl<'de, 'a, T: ?Sized> Deserialize<'de> for Cow<'a, T> + // + // A #[serde(borrow)] attribute enables borrowing that corresponds + // roughly to these impls: + // + // impl<'de: 'a, 'a> Deserialize<'de> for Cow<'a, str> + // impl<'de: 'a, 'a> Deserialize<'de> for Cow<'a, [u8]> + if is_cow(&field.ty, is_str) { + let mut path = syn::Path { + leading_colon: None, + segments: Punctuated::new(), + }; + let span = Span::call_site(); + path.segments.push(Ident::new("_serde", span).into()); + path.segments.push(Ident::new("__private", span).into()); + path.segments.push(Ident::new("de", span).into()); + path.segments + .push(Ident::new("borrow_cow_str", span).into()); + let expr = syn::ExprPath { + attrs: Vec::new(), + qself: None, + path, + }; + deserialize_with.set_if_none(expr); + } else if is_cow(&field.ty, is_slice_u8) { + let mut path = syn::Path { + leading_colon: None, + segments: Punctuated::new(), + }; + let span = Span::call_site(); + path.segments.push(Ident::new("_serde", span).into()); + path.segments.push(Ident::new("__private", span).into()); + path.segments.push(Ident::new("de", span).into()); + path.segments + .push(Ident::new("borrow_cow_bytes", span).into()); + let expr = syn::ExprPath { + attrs: Vec::new(), + qself: None, + path, + }; + deserialize_with.set_if_none(expr); + } + } else if is_implicitly_borrowed(&field.ty) { + // Types &str and &[u8] are always implicitly borrowed. No need for + // a #[serde(borrow)]. + collect_lifetimes(&field.ty, &mut borrowed_lifetimes); + } + + Field { + name: Name::from_attrs(ident, ser_name, de_name, Some(de_aliases)), + skip_serializing: skip_serializing.get(), + skip_deserializing: skip_deserializing.get(), + skip_serializing_if: skip_serializing_if.get(), + default: default.get().unwrap_or(Default::None), + serialize_with: serialize_with.get(), + deserialize_with: deserialize_with.get(), + ser_bound: ser_bound.get(), + de_bound: de_bound.get(), + borrowed_lifetimes, + getter: getter.get(), + flatten: flatten.get(), + transparent: false, + } + } + + pub fn name(&self) -> &Name { + &self.name + } + + pub fn aliases(&self) -> Vec { + self.name.deserialize_aliases() + } + + pub fn rename_by_rules(&mut self, rules: &RenameAllRules) { + if !self.name.serialize_renamed { + self.name.serialize = rules.serialize.apply_to_field(&self.name.serialize); + } + if !self.name.deserialize_renamed { + self.name.deserialize = rules.deserialize.apply_to_field(&self.name.deserialize); + } + } + + pub fn skip_serializing(&self) -> bool { + self.skip_serializing + } + + pub fn skip_deserializing(&self) -> bool { + self.skip_deserializing + } + + pub fn skip_serializing_if(&self) -> Option<&syn::ExprPath> { + self.skip_serializing_if.as_ref() + } + + pub fn default(&self) -> &Default { + &self.default + } + + pub fn serialize_with(&self) -> Option<&syn::ExprPath> { + self.serialize_with.as_ref() + } + + pub fn deserialize_with(&self) -> Option<&syn::ExprPath> { + self.deserialize_with.as_ref() + } + + pub fn ser_bound(&self) -> Option<&[syn::WherePredicate]> { + self.ser_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn de_bound(&self) -> Option<&[syn::WherePredicate]> { + self.de_bound.as_ref().map(|vec| &vec[..]) + } + + pub fn borrowed_lifetimes(&self) -> &BTreeSet { + &self.borrowed_lifetimes + } + + pub fn getter(&self) -> Option<&syn::ExprPath> { + self.getter.as_ref() + } + + pub fn flatten(&self) -> bool { + self.flatten + } + + pub fn transparent(&self) -> bool { + self.transparent + } + + pub fn mark_transparent(&mut self) { + self.transparent = true; + } +} + +type SerAndDe = (Option, Option); + +fn get_ser_and_de<'a, 'b, T, F>( + cx: &'b Ctxt, + attr_name: Symbol, + metas: &'a Punctuated, + f: F, +) -> Result<(VecAttr<'b, T>, VecAttr<'b, T>), ()> +where + T: 'a, + F: Fn(&Ctxt, Symbol, Symbol, &'a syn::Lit) -> Result, +{ + let mut ser_meta = VecAttr::none(cx, attr_name); + let mut de_meta = VecAttr::none(cx, attr_name); + + for meta in metas { + match meta { + Meta(NameValue(meta)) if meta.path == SERIALIZE => { + if let Ok(v) = f(cx, attr_name, SERIALIZE, &meta.lit) { + ser_meta.insert(&meta.path, v); + } + } + + Meta(NameValue(meta)) if meta.path == DESERIALIZE => { + if let Ok(v) = f(cx, attr_name, DESERIALIZE, &meta.lit) { + de_meta.insert(&meta.path, v); + } + } + + _ => { + let msg = format!( + "malformed {0} attribute, expected `{0}(serialize = ..., deserialize = ...)`", + attr_name, + ); + cx.error_spanned_by(meta, msg); + return Err(()); + } + } + } + + Ok((ser_meta, de_meta)) +} + +fn get_renames<'a>( + cx: &Ctxt, + items: &'a Punctuated, +) -> Result, ()> { + let (ser, de) = get_ser_and_de(cx, RENAME, items, get_lit_str2)?; + Ok((ser.at_most_one()?, de.at_most_one()?)) +} + +fn get_multiple_renames<'a>( + cx: &Ctxt, + items: &'a Punctuated, +) -> Result<(Option<&'a syn::LitStr>, Vec<&'a syn::LitStr>), ()> { + let (ser, de) = get_ser_and_de(cx, RENAME, items, get_lit_str2)?; + Ok((ser.at_most_one()?, de.get())) +} + +fn get_where_predicates( + cx: &Ctxt, + items: &Punctuated, +) -> Result>, ()> { + let (ser, de) = get_ser_and_de(cx, BOUND, items, parse_lit_into_where)?; + Ok((ser.at_most_one()?, de.at_most_one()?)) +} + +pub fn get_serde_meta_items(cx: &Ctxt, attr: &syn::Attribute) -> Result, ()> { + if attr.path != SERDE { + return Ok(Vec::new()); + } + + match attr.parse_meta() { + Ok(List(meta)) => Ok(meta.nested.into_iter().collect()), + Ok(other) => { + cx.error_spanned_by(other, "expected #[serde(...)]"); + Err(()) + } + Err(err) => { + cx.syn_error(err); + Err(()) + } + } +} + +fn get_lit_str<'a>(cx: &Ctxt, attr_name: Symbol, lit: &'a syn::Lit) -> Result<&'a syn::LitStr, ()> { + get_lit_str2(cx, attr_name, attr_name, lit) +} + +fn get_lit_str2<'a>( + cx: &Ctxt, + attr_name: Symbol, + meta_item_name: Symbol, + lit: &'a syn::Lit, +) -> Result<&'a syn::LitStr, ()> { + if let syn::Lit::Str(lit) = lit { + Ok(lit) + } else { + let msg = format!( + "expected serde {} attribute to be a string: `{} = \"...\"`", + attr_name, meta_item_name, + ); + cx.error_spanned_by(lit, msg); + Err(()) + } +} + +fn parse_lit_into_path(cx: &Ctxt, attr_name: Symbol, lit: &syn::Lit) -> Result { + let string = get_lit_str(cx, attr_name, lit)?; + string.parse().map_err(|_| { + let msg = format!("failed to parse path: {:?}", string.value()); + cx.error_spanned_by(lit, msg); + }) +} + +fn parse_lit_into_expr_path( + cx: &Ctxt, + attr_name: Symbol, + lit: &syn::Lit, +) -> Result { + let string = get_lit_str(cx, attr_name, lit)?; + string.parse().map_err(|_| { + let msg = format!("failed to parse path: {:?}", string.value()); + cx.error_spanned_by(lit, msg); + }) +} + +fn parse_lit_into_where( + cx: &Ctxt, + attr_name: Symbol, + meta_item_name: Symbol, + lit: &syn::Lit, +) -> Result, ()> { + let string = get_lit_str2(cx, attr_name, meta_item_name, lit)?; + + string + .parse_with(Punctuated::::parse_terminated) + .map(Vec::from_iter) + .map_err(|err| cx.error_spanned_by(lit, err)) +} + +fn parse_lit_into_ty(cx: &Ctxt, attr_name: Symbol, lit: &syn::Lit) -> Result { + let string = get_lit_str(cx, attr_name, lit)?; + + string.parse().map_err(|_| { + let msg = format!("failed to parse type: {} = {:?}", attr_name, string.value()); + cx.error_spanned_by(lit, msg); + }) +} + +// Parses a string literal like "'a + 'b + 'c" containing a nonempty list of +// lifetimes separated by `+`. +fn parse_lit_into_lifetimes(cx: &Ctxt, lit: &syn::Lit) -> Result, ()> { + let string = get_lit_str(cx, BORROW, lit)?; + + if let Ok(lifetimes) = string.parse_with(|input: ParseStream| { + let mut set = BTreeSet::new(); + while !input.is_empty() { + let lifetime: Lifetime = input.parse()?; + if !set.insert(lifetime.clone()) { + let msg = format!("duplicate borrowed lifetime `{}`", lifetime); + cx.error_spanned_by(lit, msg); + } + if input.is_empty() { + break; + } + input.parse::()?; + } + Ok(set) + }) { + return if lifetimes.is_empty() { + let msg = "at least one lifetime must be borrowed"; + cx.error_spanned_by(lit, msg); + Err(()) + } else { + Ok(lifetimes) + }; + } + + let msg = format!("failed to parse borrowed lifetimes: {:?}", string.value()); + cx.error_spanned_by(lit, msg); + Err(()) +} + +fn is_implicitly_borrowed(ty: &syn::Type) -> bool { + is_implicitly_borrowed_reference(ty) || is_option(ty, is_implicitly_borrowed_reference) +} + +fn is_implicitly_borrowed_reference(ty: &syn::Type) -> bool { + is_reference(ty, is_str) || is_reference(ty, is_slice_u8) +} + +// Whether the type looks like it might be `std::borrow::Cow` where elem="T". +// This can have false negatives and false positives. +// +// False negative: +// +// use std::borrow::Cow as Pig; +// +// #[derive(Deserialize)] +// struct S<'a> { +// #[serde(borrow)] +// pig: Pig<'a, str>, +// } +// +// False positive: +// +// type str = [i16]; +// +// #[derive(Deserialize)] +// struct S<'a> { +// #[serde(borrow)] +// cow: Cow<'a, str>, +// } +fn is_cow(ty: &syn::Type, elem: fn(&syn::Type) -> bool) -> bool { + let path = match ungroup(ty) { + syn::Type::Path(ty) => &ty.path, + _ => { + return false; + } + }; + let seg = match path.segments.last() { + Some(seg) => seg, + None => { + return false; + } + }; + let args = match &seg.arguments { + syn::PathArguments::AngleBracketed(bracketed) => &bracketed.args, + _ => { + return false; + } + }; + seg.ident == "Cow" + && args.len() == 2 + && match (&args[0], &args[1]) { + (syn::GenericArgument::Lifetime(_), syn::GenericArgument::Type(arg)) => elem(arg), + _ => false, + } +} + +fn is_option(ty: &syn::Type, elem: fn(&syn::Type) -> bool) -> bool { + let path = match ungroup(ty) { + syn::Type::Path(ty) => &ty.path, + _ => { + return false; + } + }; + let seg = match path.segments.last() { + Some(seg) => seg, + None => { + return false; + } + }; + let args = match &seg.arguments { + syn::PathArguments::AngleBracketed(bracketed) => &bracketed.args, + _ => { + return false; + } + }; + seg.ident == "Option" + && args.len() == 1 + && match &args[0] { + syn::GenericArgument::Type(arg) => elem(arg), + _ => false, + } +} + +// Whether the type looks like it might be `&T` where elem="T". This can have +// false negatives and false positives. +// +// False negative: +// +// type Yarn = str; +// +// #[derive(Deserialize)] +// struct S<'a> { +// r: &'a Yarn, +// } +// +// False positive: +// +// type str = [i16]; +// +// #[derive(Deserialize)] +// struct S<'a> { +// r: &'a str, +// } +fn is_reference(ty: &syn::Type, elem: fn(&syn::Type) -> bool) -> bool { + match ungroup(ty) { + syn::Type::Reference(ty) => ty.mutability.is_none() && elem(&ty.elem), + _ => false, + } +} + +fn is_str(ty: &syn::Type) -> bool { + is_primitive_type(ty, "str") +} + +fn is_slice_u8(ty: &syn::Type) -> bool { + match ungroup(ty) { + syn::Type::Slice(ty) => is_primitive_type(&ty.elem, "u8"), + _ => false, + } +} + +fn is_primitive_type(ty: &syn::Type, primitive: &str) -> bool { + match ungroup(ty) { + syn::Type::Path(ty) => ty.qself.is_none() && is_primitive_path(&ty.path, primitive), + _ => false, + } +} + +fn is_primitive_path(path: &syn::Path, primitive: &str) -> bool { + path.leading_colon.is_none() + && path.segments.len() == 1 + && path.segments[0].ident == primitive + && path.segments[0].arguments.is_empty() +} + +// All lifetimes that this type could borrow from a Deserializer. +// +// For example a type `S<'a, 'b>` could borrow `'a` and `'b`. On the other hand +// a type `for<'a> fn(&'a str)` could not borrow `'a` from the Deserializer. +// +// This is used when there is an explicit or implicit `#[serde(borrow)]` +// attribute on the field so there must be at least one borrowable lifetime. +fn borrowable_lifetimes( + cx: &Ctxt, + name: &str, + field: &syn::Field, +) -> Result, ()> { + let mut lifetimes = BTreeSet::new(); + collect_lifetimes(&field.ty, &mut lifetimes); + if lifetimes.is_empty() { + let msg = format!("field `{}` has no lifetimes to borrow", name); + cx.error_spanned_by(field, msg); + Err(()) + } else { + Ok(lifetimes) + } +} + +fn collect_lifetimes(ty: &syn::Type, out: &mut BTreeSet) { + match ty { + syn::Type::Slice(ty) => { + collect_lifetimes(&ty.elem, out); + } + syn::Type::Array(ty) => { + collect_lifetimes(&ty.elem, out); + } + syn::Type::Ptr(ty) => { + collect_lifetimes(&ty.elem, out); + } + syn::Type::Reference(ty) => { + out.extend(ty.lifetime.iter().cloned()); + collect_lifetimes(&ty.elem, out); + } + syn::Type::Tuple(ty) => { + for elem in &ty.elems { + collect_lifetimes(elem, out); + } + } + syn::Type::Path(ty) => { + if let Some(qself) = &ty.qself { + collect_lifetimes(&qself.ty, out); + } + for seg in &ty.path.segments { + if let syn::PathArguments::AngleBracketed(bracketed) = &seg.arguments { + for arg in &bracketed.args { + match arg { + syn::GenericArgument::Lifetime(lifetime) => { + out.insert(lifetime.clone()); + } + syn::GenericArgument::Type(ty) => { + collect_lifetimes(ty, out); + } + syn::GenericArgument::Binding(binding) => { + collect_lifetimes(&binding.ty, out); + } + syn::GenericArgument::Constraint(_) + | syn::GenericArgument::Const(_) => {} + } + } + } + } + } + syn::Type::Paren(ty) => { + collect_lifetimes(&ty.elem, out); + } + syn::Type::Group(ty) => { + collect_lifetimes(&ty.elem, out); + } + syn::Type::Macro(ty) => { + collect_lifetimes_from_tokens(ty.mac.tokens.clone(), out); + } + syn::Type::BareFn(_) + | syn::Type::Never(_) + | syn::Type::TraitObject(_) + | syn::Type::ImplTrait(_) + | syn::Type::Infer(_) + | syn::Type::Verbatim(_) => {} + + #[cfg_attr(all(test, exhaustive), deny(non_exhaustive_omitted_patterns))] + _ => {} + } +} + +fn collect_lifetimes_from_tokens(tokens: TokenStream, out: &mut BTreeSet) { + let mut iter = tokens.into_iter(); + while let Some(tt) = iter.next() { + match &tt { + TokenTree::Punct(op) if op.as_char() == '\'' && op.spacing() == Spacing::Joint => { + if let Some(TokenTree::Ident(ident)) = iter.next() { + out.insert(syn::Lifetime { + apostrophe: op.span(), + ident, + }); + } + } + TokenTree::Group(group) => { + let tokens = group.stream(); + collect_lifetimes_from_tokens(tokens, out); + } + _ => {} + } + } +} diff --git a/rust/serde_derive/internals/case.rs b/rust/serde_derive/internals/case.rs new file mode 100644 index 00000000000000..6a75c7460e67b6 --- /dev/null +++ b/rust/serde_derive/internals/case.rs @@ -0,0 +1,199 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Code to convert the Rust-styled field/variant (e.g. `my_field`, `MyType`) to the +//! case of the source (e.g. `my-field`, `MY_FIELD`). + +// See https://users.rust-lang.org/t/psa-dealing-with-warning-unused-import-std-ascii-asciiext-in-today-s-nightly/13726 +#[allow(deprecated, unused_imports)] +use std::ascii::AsciiExt; + +use std::fmt::{self, Debug, Display}; + +use self::RenameRule::*; + +/// The different possible ways to change case of fields in a struct, or variants in an enum. +#[derive(Copy, Clone, PartialEq)] +pub enum RenameRule { + /// Don't apply a default rename rule. + None, + /// Rename direct children to "lowercase" style. + LowerCase, + /// Rename direct children to "UPPERCASE" style. + UpperCase, + /// Rename direct children to "PascalCase" style, as typically used for + /// enum variants. + PascalCase, + /// Rename direct children to "camelCase" style. + CamelCase, + /// Rename direct children to "snake_case" style, as commonly used for + /// fields. + SnakeCase, + /// Rename direct children to "SCREAMING_SNAKE_CASE" style, as commonly + /// used for constants. + ScreamingSnakeCase, + /// Rename direct children to "kebab-case" style. + KebabCase, + /// Rename direct children to "SCREAMING-KEBAB-CASE" style. + ScreamingKebabCase, +} + +static RENAME_RULES: &[(&str, RenameRule)] = &[ + ("lowercase", LowerCase), + ("UPPERCASE", UpperCase), + ("PascalCase", PascalCase), + ("camelCase", CamelCase), + ("snake_case", SnakeCase), + ("SCREAMING_SNAKE_CASE", ScreamingSnakeCase), + ("kebab-case", KebabCase), + ("SCREAMING-KEBAB-CASE", ScreamingKebabCase), +]; + +impl RenameRule { + pub fn from_str(rename_all_str: &str) -> Result { + for (name, rule) in RENAME_RULES { + if rename_all_str == *name { + return Ok(*rule); + } + } + Err(ParseError { + unknown: rename_all_str, + }) + } + + /// Apply a renaming rule to an enum variant, returning the version expected in the source. + pub fn apply_to_variant(&self, variant: &str) -> String { + match *self { + None | PascalCase => variant.to_owned(), + LowerCase => variant.to_ascii_lowercase(), + UpperCase => variant.to_ascii_uppercase(), + CamelCase => variant[..1].to_ascii_lowercase() + &variant[1..], + SnakeCase => { + let mut snake = String::new(); + for (i, ch) in variant.char_indices() { + if i > 0 && ch.is_uppercase() { + snake.push('_'); + } + snake.push(ch.to_ascii_lowercase()); + } + snake + } + ScreamingSnakeCase => SnakeCase.apply_to_variant(variant).to_ascii_uppercase(), + KebabCase => SnakeCase.apply_to_variant(variant).replace('_', "-"), + ScreamingKebabCase => ScreamingSnakeCase + .apply_to_variant(variant) + .replace('_', "-"), + } + } + + /// Apply a renaming rule to a struct field, returning the version expected in the source. + pub fn apply_to_field(&self, field: &str) -> String { + match *self { + None | LowerCase | SnakeCase => field.to_owned(), + UpperCase => field.to_ascii_uppercase(), + PascalCase => { + let mut pascal = String::new(); + let mut capitalize = true; + for ch in field.chars() { + if ch == '_' { + capitalize = true; + } else if capitalize { + pascal.push(ch.to_ascii_uppercase()); + capitalize = false; + } else { + pascal.push(ch); + } + } + pascal + } + CamelCase => { + let pascal = PascalCase.apply_to_field(field); + pascal[..1].to_ascii_lowercase() + &pascal[1..] + } + ScreamingSnakeCase => field.to_ascii_uppercase(), + KebabCase => field.replace('_', "-"), + ScreamingKebabCase => ScreamingSnakeCase.apply_to_field(field).replace('_', "-"), + } + } +} + +pub struct ParseError<'a> { + unknown: &'a str, +} + +impl<'a> Display for ParseError<'a> { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str("unknown rename rule `rename_all = ")?; + Debug::fmt(self.unknown, f)?; + f.write_str("`, expected one of ")?; + for (i, (name, _rule)) in RENAME_RULES.iter().enumerate() { + if i > 0 { + f.write_str(", ")?; + } + Debug::fmt(name, f)?; + } + Ok(()) + } +} + +#[test] +fn rename_variants() { + for &(original, lower, upper, camel, snake, screaming, kebab, screaming_kebab) in &[ + ( + "Outcome", "outcome", "OUTCOME", "outcome", "outcome", "OUTCOME", "outcome", "OUTCOME", + ), + ( + "VeryTasty", + "verytasty", + "VERYTASTY", + "veryTasty", + "very_tasty", + "VERY_TASTY", + "very-tasty", + "VERY-TASTY", + ), + ("A", "a", "A", "a", "a", "A", "a", "A"), + ("Z42", "z42", "Z42", "z42", "z42", "Z42", "z42", "Z42"), + ] { + assert_eq!(None.apply_to_variant(original), original); + assert_eq!(LowerCase.apply_to_variant(original), lower); + assert_eq!(UpperCase.apply_to_variant(original), upper); + assert_eq!(PascalCase.apply_to_variant(original), original); + assert_eq!(CamelCase.apply_to_variant(original), camel); + assert_eq!(SnakeCase.apply_to_variant(original), snake); + assert_eq!(ScreamingSnakeCase.apply_to_variant(original), screaming); + assert_eq!(KebabCase.apply_to_variant(original), kebab); + assert_eq!( + ScreamingKebabCase.apply_to_variant(original), + screaming_kebab + ); + } +} + +#[test] +fn rename_fields() { + for &(original, upper, pascal, camel, screaming, kebab, screaming_kebab) in &[ + ( + "outcome", "OUTCOME", "Outcome", "outcome", "OUTCOME", "outcome", "OUTCOME", + ), + ( + "very_tasty", + "VERY_TASTY", + "VeryTasty", + "veryTasty", + "VERY_TASTY", + "very-tasty", + "VERY-TASTY", + ), + ("a", "A", "A", "a", "A", "a", "A"), + ("z42", "Z42", "Z42", "z42", "Z42", "z42", "Z42"), + ] { + assert_eq!(None.apply_to_field(original), original); + assert_eq!(UpperCase.apply_to_field(original), upper); + assert_eq!(PascalCase.apply_to_field(original), pascal); + assert_eq!(CamelCase.apply_to_field(original), camel); + assert_eq!(SnakeCase.apply_to_field(original), original); + assert_eq!(ScreamingSnakeCase.apply_to_field(original), screaming); + assert_eq!(KebabCase.apply_to_field(original), kebab); + assert_eq!(ScreamingKebabCase.apply_to_field(original), screaming_kebab); + } +} diff --git a/rust/serde_derive/internals/check.rs b/rust/serde_derive/internals/check.rs new file mode 100644 index 00000000000000..1884995086845a --- /dev/null +++ b/rust/serde_derive/internals/check.rs @@ -0,0 +1,445 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use internals::ast::{Container, Data, Field, Style}; +use internals::attr::{Identifier, TagType}; +use internals::{ungroup, Ctxt, Derive}; +use syn::{Member, Type}; + +/// Cross-cutting checks that require looking at more than a single attrs +/// object. Simpler checks should happen when parsing and building the attrs. +pub fn check(cx: &Ctxt, cont: &mut Container, derive: Derive) { + check_remote_generic(cx, cont); + check_getter(cx, cont); + check_flatten(cx, cont); + check_identifier(cx, cont); + check_variant_skip_attrs(cx, cont); + check_internal_tag_field_name_conflict(cx, cont); + check_adjacent_tag_conflict(cx, cont); + check_transparent(cx, cont, derive); + check_from_and_try_from(cx, cont); +} + +/// Remote derive definition type must have either all of the generics of the +/// remote type: +/// +/// #[serde(remote = "Generic")] +/// struct Generic {…} +/// +/// or none of them, i.e. defining impls for one concrete instantiation of the +/// remote type only: +/// +/// #[serde(remote = "Generic")] +/// struct ConcreteDef {…} +/// +fn check_remote_generic(cx: &Ctxt, cont: &Container) { + if let Some(remote) = cont.attrs.remote() { + let local_has_generic = !cont.generics.params.is_empty(); + let remote_has_generic = !remote.segments.last().unwrap().arguments.is_none(); + if local_has_generic && remote_has_generic { + cx.error_spanned_by(remote, "remove generic parameters from this path"); + } + } +} + +/// Getters are only allowed inside structs (not enums) with the `remote` +/// attribute. +fn check_getter(cx: &Ctxt, cont: &Container) { + match cont.data { + Data::Enum(_) => { + if cont.data.has_getter() { + cx.error_spanned_by( + cont.original, + "#[serde(getter = \"...\")] is not allowed in an enum", + ); + } + } + Data::Struct(_, _) => { + if cont.data.has_getter() && cont.attrs.remote().is_none() { + cx.error_spanned_by( + cont.original, + "#[serde(getter = \"...\")] can only be used in structs that have #[serde(remote = \"...\")]", + ); + } + } + } +} + +/// Flattening has some restrictions we can test. +fn check_flatten(cx: &Ctxt, cont: &Container) { + match &cont.data { + Data::Enum(variants) => { + for variant in variants { + for field in &variant.fields { + check_flatten_field(cx, variant.style, field); + } + } + } + Data::Struct(style, fields) => { + for field in fields { + check_flatten_field(cx, *style, field); + } + } + } +} + +fn check_flatten_field(cx: &Ctxt, style: Style, field: &Field) { + if !field.attrs.flatten() { + return; + } + match style { + Style::Tuple => { + cx.error_spanned_by( + field.original, + "#[serde(flatten)] cannot be used on tuple structs", + ); + } + Style::Newtype => { + cx.error_spanned_by( + field.original, + "#[serde(flatten)] cannot be used on newtype structs", + ); + } + _ => {} + } +} + +/// The `other` attribute must be used at most once and it must be the last +/// variant of an enum. +/// +/// Inside a `variant_identifier` all variants must be unit variants. Inside a +/// `field_identifier` all but possibly one variant must be unit variants. The +/// last variant may be a newtype variant which is an implicit "other" case. +fn check_identifier(cx: &Ctxt, cont: &Container) { + let variants = match &cont.data { + Data::Enum(variants) => variants, + Data::Struct(_, _) => { + return; + } + }; + + for (i, variant) in variants.iter().enumerate() { + match ( + variant.style, + cont.attrs.identifier(), + variant.attrs.other(), + cont.attrs.tag(), + ) { + // The `other` attribute may not be used in a variant_identifier. + (_, Identifier::Variant, true, _) => { + cx.error_spanned_by( + variant.original, + "#[serde(other)] may not be used on a variant identifier", + ); + } + + // Variant with `other` attribute cannot appear in untagged enum + (_, Identifier::No, true, &TagType::None) => { + cx.error_spanned_by( + variant.original, + "#[serde(other)] cannot appear on untagged enum", + ); + } + + // Variant with `other` attribute must be the last one. + (Style::Unit, Identifier::Field, true, _) | (Style::Unit, Identifier::No, true, _) => { + if i < variants.len() - 1 { + cx.error_spanned_by( + variant.original, + "#[serde(other)] must be on the last variant", + ); + } + } + + // Variant with `other` attribute must be a unit variant. + (_, Identifier::Field, true, _) | (_, Identifier::No, true, _) => { + cx.error_spanned_by( + variant.original, + "#[serde(other)] must be on a unit variant", + ); + } + + // Any sort of variant is allowed if this is not an identifier. + (_, Identifier::No, false, _) => {} + + // Unit variant without `other` attribute is always fine. + (Style::Unit, _, false, _) => {} + + // The last field is allowed to be a newtype catch-all. + (Style::Newtype, Identifier::Field, false, _) => { + if i < variants.len() - 1 { + cx.error_spanned_by( + variant.original, + format!("`{}` must be the last variant", variant.ident), + ); + } + } + + (_, Identifier::Field, false, _) => { + cx.error_spanned_by( + variant.original, + "#[serde(field_identifier)] may only contain unit variants", + ); + } + + (_, Identifier::Variant, false, _) => { + cx.error_spanned_by( + variant.original, + "#[serde(variant_identifier)] may only contain unit variants", + ); + } + } + } +} + +/// Skip-(de)serializing attributes are not allowed on variants marked +/// (de)serialize_with. +fn check_variant_skip_attrs(cx: &Ctxt, cont: &Container) { + let variants = match &cont.data { + Data::Enum(variants) => variants, + Data::Struct(_, _) => { + return; + } + }; + + for variant in variants.iter() { + if variant.attrs.serialize_with().is_some() { + if variant.attrs.skip_serializing() { + cx.error_spanned_by( + variant.original, + format!( + "variant `{}` cannot have both #[serde(serialize_with)] and #[serde(skip_serializing)]", + variant.ident + ), + ); + } + + for field in &variant.fields { + let member = member_message(&field.member); + + if field.attrs.skip_serializing() { + cx.error_spanned_by( + variant.original, + format!( + "variant `{}` cannot have both #[serde(serialize_with)] and a field {} marked with #[serde(skip_serializing)]", + variant.ident, member + ), + ); + } + + if field.attrs.skip_serializing_if().is_some() { + cx.error_spanned_by( + variant.original, + format!( + "variant `{}` cannot have both #[serde(serialize_with)] and a field {} marked with #[serde(skip_serializing_if)]", + variant.ident, member + ), + ); + } + } + } + + if variant.attrs.deserialize_with().is_some() { + if variant.attrs.skip_deserializing() { + cx.error_spanned_by( + variant.original, + format!( + "variant `{}` cannot have both #[serde(deserialize_with)] and #[serde(skip_deserializing)]", + variant.ident + ), + ); + } + + for field in &variant.fields { + if field.attrs.skip_deserializing() { + let member = member_message(&field.member); + + cx.error_spanned_by( + variant.original, + format!( + "variant `{}` cannot have both #[serde(deserialize_with)] and a field {} marked with #[serde(skip_deserializing)]", + variant.ident, member + ), + ); + } + } + } + } +} + +/// The tag of an internally-tagged struct variant must not be +/// the same as either one of its fields, as this would result in +/// duplicate keys in the serialized output and/or ambiguity in +/// the to-be-deserialized input. +fn check_internal_tag_field_name_conflict(cx: &Ctxt, cont: &Container) { + let variants = match &cont.data { + Data::Enum(variants) => variants, + Data::Struct(_, _) => return, + }; + + let tag = match cont.attrs.tag() { + TagType::Internal { tag } => tag.as_str(), + TagType::External | TagType::Adjacent { .. } | TagType::None => return, + }; + + let diagnose_conflict = || { + cx.error_spanned_by( + cont.original, + format!("variant field name `{}` conflicts with internal tag", tag), + ); + }; + + for variant in variants { + match variant.style { + Style::Struct => { + for field in &variant.fields { + let check_ser = !field.attrs.skip_serializing(); + let check_de = !field.attrs.skip_deserializing(); + let name = field.attrs.name(); + let ser_name = name.serialize_name(); + + if check_ser && ser_name == tag { + diagnose_conflict(); + return; + } + + for de_name in field.attrs.aliases() { + if check_de && de_name == tag { + diagnose_conflict(); + return; + } + } + } + } + Style::Unit | Style::Newtype | Style::Tuple => {} + } + } +} + +/// In the case of adjacently-tagged enums, the type and the +/// contents tag must differ, for the same reason. +fn check_adjacent_tag_conflict(cx: &Ctxt, cont: &Container) { + let (type_tag, content_tag) = match cont.attrs.tag() { + TagType::Adjacent { tag, content } => (tag, content), + TagType::Internal { .. } | TagType::External | TagType::None => return, + }; + + if type_tag == content_tag { + cx.error_spanned_by( + cont.original, + format!( + "enum tags `{}` for type and content conflict with each other", + type_tag + ), + ); + } +} + +/// Enums and unit structs cannot be transparent. +fn check_transparent(cx: &Ctxt, cont: &mut Container, derive: Derive) { + if !cont.attrs.transparent() { + return; + } + + if cont.attrs.type_from().is_some() { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] is not allowed with #[serde(from = \"...\")]", + ); + } + + if cont.attrs.type_try_from().is_some() { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] is not allowed with #[serde(try_from = \"...\")]", + ); + } + + if cont.attrs.type_into().is_some() { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] is not allowed with #[serde(into = \"...\")]", + ); + } + + let fields = match &mut cont.data { + Data::Enum(_) => { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] is not allowed on an enum", + ); + return; + } + Data::Struct(Style::Unit, _) => { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] is not allowed on a unit struct", + ); + return; + } + Data::Struct(_, fields) => fields, + }; + + let mut transparent_field = None; + + for field in fields { + if allow_transparent(field, derive) { + if transparent_field.is_some() { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] requires struct to have at most one transparent field", + ); + return; + } + transparent_field = Some(field); + } + } + + match transparent_field { + Some(transparent_field) => transparent_field.attrs.mark_transparent(), + None => match derive { + Derive::Serialize => { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] requires at least one field that is not skipped", + ); + } + Derive::Deserialize => { + cx.error_spanned_by( + cont.original, + "#[serde(transparent)] requires at least one field that is neither skipped nor has a default", + ); + } + }, + } +} + +fn member_message(member: &Member) -> String { + match member { + Member::Named(ident) => format!("`{}`", ident), + Member::Unnamed(i) => format!("#{}", i.index), + } +} + +fn allow_transparent(field: &Field, derive: Derive) -> bool { + if let Type::Path(ty) = ungroup(field.ty) { + if let Some(seg) = ty.path.segments.last() { + if seg.ident == "PhantomData" { + return false; + } + } + } + + match derive { + Derive::Serialize => !field.attrs.skip_serializing(), + Derive::Deserialize => !field.attrs.skip_deserializing() && field.attrs.default().is_none(), + } +} + +fn check_from_and_try_from(cx: &Ctxt, cont: &mut Container) { + if cont.attrs.type_from().is_some() && cont.attrs.type_try_from().is_some() { + cx.error_spanned_by( + cont.original, + "#[serde(from = \"...\")] and #[serde(try_from = \"...\")] conflict with each other", + ); + } +} diff --git a/rust/serde_derive/internals/ctxt.rs b/rust/serde_derive/internals/ctxt.rs new file mode 100644 index 00000000000000..7a0a9f92f9924e --- /dev/null +++ b/rust/serde_derive/internals/ctxt.rs @@ -0,0 +1,64 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use quote::ToTokens; +use std::cell::RefCell; +use std::fmt::Display; +use std::thread; +use syn; + +/// A type to collect errors together and format them. +/// +/// Dropping this object will cause a panic. It must be consumed using `check`. +/// +/// References can be shared since this type uses run-time exclusive mut checking. +#[derive(Default)] +pub struct Ctxt { + // The contents will be set to `None` during checking. This is so that checking can be + // enforced. + errors: RefCell>>, +} + +impl Ctxt { + /// Create a new context object. + /// + /// This object contains no errors, but will still trigger a panic if it is not `check`ed. + pub fn new() -> Self { + Ctxt { + errors: RefCell::new(Some(Vec::new())), + } + } + + /// Add an error to the context object with a tokenenizable object. + /// + /// The object is used for spanning in error messages. + pub fn error_spanned_by(&self, obj: A, msg: T) { + self.errors + .borrow_mut() + .as_mut() + .unwrap() + // Curb monomorphization from generating too many identical methods. + .push(syn::Error::new_spanned(obj.into_token_stream(), msg)); + } + + /// Add one of Syn's parse errors. + pub fn syn_error(&self, err: syn::Error) { + self.errors.borrow_mut().as_mut().unwrap().push(err); + } + + /// Consume this object, producing a formatted error string if there are errors. + pub fn check(self) -> Result<(), Vec> { + let errors = self.errors.borrow_mut().take().unwrap(); + match errors.len() { + 0 => Ok(()), + _ => Err(errors), + } + } +} + +impl Drop for Ctxt { + fn drop(&mut self) { + if !thread::panicking() && self.errors.borrow().is_some() { + panic!("forgot to check for errors"); + } + } +} diff --git a/rust/serde_derive/internals/mod.rs b/rust/serde_derive/internals/mod.rs new file mode 100644 index 00000000000000..9de56241adf342 --- /dev/null +++ b/rust/serde_derive/internals/mod.rs @@ -0,0 +1,30 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +pub mod ast; +pub mod attr; + +mod ctxt; +pub use self::ctxt::Ctxt; + +mod receiver; +pub use self::receiver::replace_receiver; + +mod case; +mod check; +mod respan; +mod symbol; + +use syn::Type; + +#[derive(Copy, Clone)] +pub enum Derive { + Serialize, + Deserialize, +} + +pub fn ungroup(mut ty: &Type) -> &Type { + while let Type::Group(group) = ty { + ty = &group.elem; + } + ty +} diff --git a/rust/serde_derive/internals/receiver.rs b/rust/serde_derive/internals/receiver.rs new file mode 100644 index 00000000000000..fc5bc1b89db2e4 --- /dev/null +++ b/rust/serde_derive/internals/receiver.rs @@ -0,0 +1,287 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use internals::respan::respan; +use proc_macro2::Span; +use quote::ToTokens; +use std::mem; +use syn::punctuated::Punctuated; +use syn::{ + parse_quote, Data, DeriveInput, Expr, ExprPath, GenericArgument, GenericParam, Generics, Macro, + Path, PathArguments, QSelf, ReturnType, Type, TypeParamBound, TypePath, WherePredicate, +}; + +pub fn replace_receiver(input: &mut DeriveInput) { + let self_ty = { + let ident = &input.ident; + let ty_generics = input.generics.split_for_impl().1; + parse_quote!(#ident #ty_generics) + }; + let mut visitor = ReplaceReceiver(&self_ty); + visitor.visit_generics_mut(&mut input.generics); + visitor.visit_data_mut(&mut input.data); +} + +struct ReplaceReceiver<'a>(&'a TypePath); + +impl ReplaceReceiver<'_> { + fn self_ty(&self, span: Span) -> TypePath { + let tokens = self.0.to_token_stream(); + let respanned = respan(tokens, span); + syn::parse2(respanned).unwrap() + } + + fn self_to_qself(&self, qself: &mut Option, path: &mut Path) { + if path.leading_colon.is_some() || path.segments[0].ident != "Self" { + return; + } + + if path.segments.len() == 1 { + self.self_to_expr_path(path); + return; + } + + let span = path.segments[0].ident.span(); + *qself = Some(QSelf { + lt_token: Token![<](span), + ty: Box::new(Type::Path(self.self_ty(span))), + position: 0, + as_token: None, + gt_token: Token![>](span), + }); + + path.leading_colon = Some(**path.segments.pairs().next().unwrap().punct().unwrap()); + + let segments = mem::replace(&mut path.segments, Punctuated::new()); + path.segments = segments.into_pairs().skip(1).collect(); + } + + fn self_to_expr_path(&self, path: &mut Path) { + let self_ty = self.self_ty(path.segments[0].ident.span()); + let variant = mem::replace(path, self_ty.path); + for segment in &mut path.segments { + if let PathArguments::AngleBracketed(bracketed) = &mut segment.arguments { + if bracketed.colon2_token.is_none() && !bracketed.args.is_empty() { + bracketed.colon2_token = Some(::default()); + } + } + } + if variant.segments.len() > 1 { + path.segments.push_punct(::default()); + path.segments.extend(variant.segments.into_pairs().skip(1)); + } + } +} + +impl ReplaceReceiver<'_> { + // `Self` -> `Receiver` + fn visit_type_mut(&mut self, ty: &mut Type) { + let span = if let Type::Path(node) = ty { + if node.qself.is_none() && node.path.is_ident("Self") { + node.path.segments[0].ident.span() + } else { + self.visit_type_path_mut(node); + return; + } + } else { + self.visit_type_mut_impl(ty); + return; + }; + *ty = self.self_ty(span).into(); + } + + // `Self::Assoc` -> `::Assoc` + fn visit_type_path_mut(&mut self, ty: &mut TypePath) { + if ty.qself.is_none() { + self.self_to_qself(&mut ty.qself, &mut ty.path); + } + self.visit_type_path_mut_impl(ty); + } + + // `Self::method` -> `::method` + fn visit_expr_path_mut(&mut self, expr: &mut ExprPath) { + if expr.qself.is_none() { + self.self_to_qself(&mut expr.qself, &mut expr.path); + } + self.visit_expr_path_mut_impl(expr); + } + + // Everything below is simply traversing the syntax tree. + + fn visit_type_mut_impl(&mut self, ty: &mut Type) { + match ty { + Type::Array(ty) => { + self.visit_type_mut(&mut ty.elem); + self.visit_expr_mut(&mut ty.len); + } + Type::BareFn(ty) => { + for arg in &mut ty.inputs { + self.visit_type_mut(&mut arg.ty); + } + self.visit_return_type_mut(&mut ty.output); + } + Type::Group(ty) => self.visit_type_mut(&mut ty.elem), + Type::ImplTrait(ty) => { + for bound in &mut ty.bounds { + self.visit_type_param_bound_mut(bound); + } + } + Type::Macro(ty) => self.visit_macro_mut(&mut ty.mac), + Type::Paren(ty) => self.visit_type_mut(&mut ty.elem), + Type::Path(ty) => { + if let Some(qself) = &mut ty.qself { + self.visit_type_mut(&mut qself.ty); + } + self.visit_path_mut(&mut ty.path); + } + Type::Ptr(ty) => self.visit_type_mut(&mut ty.elem), + Type::Reference(ty) => self.visit_type_mut(&mut ty.elem), + Type::Slice(ty) => self.visit_type_mut(&mut ty.elem), + Type::TraitObject(ty) => { + for bound in &mut ty.bounds { + self.visit_type_param_bound_mut(bound); + } + } + Type::Tuple(ty) => { + for elem in &mut ty.elems { + self.visit_type_mut(elem); + } + } + + Type::Infer(_) | Type::Never(_) | Type::Verbatim(_) => {} + + #[cfg_attr(all(test, exhaustive), deny(non_exhaustive_omitted_patterns))] + _ => {} + } + } + + fn visit_type_path_mut_impl(&mut self, ty: &mut TypePath) { + if let Some(qself) = &mut ty.qself { + self.visit_type_mut(&mut qself.ty); + } + self.visit_path_mut(&mut ty.path); + } + + fn visit_expr_path_mut_impl(&mut self, expr: &mut ExprPath) { + if let Some(qself) = &mut expr.qself { + self.visit_type_mut(&mut qself.ty); + } + self.visit_path_mut(&mut expr.path); + } + + fn visit_path_mut(&mut self, path: &mut Path) { + for segment in &mut path.segments { + self.visit_path_arguments_mut(&mut segment.arguments); + } + } + + fn visit_path_arguments_mut(&mut self, arguments: &mut PathArguments) { + match arguments { + PathArguments::None => {} + PathArguments::AngleBracketed(arguments) => { + for arg in &mut arguments.args { + match arg { + GenericArgument::Type(arg) => self.visit_type_mut(arg), + GenericArgument::Binding(arg) => self.visit_type_mut(&mut arg.ty), + GenericArgument::Lifetime(_) + | GenericArgument::Constraint(_) + | GenericArgument::Const(_) => {} + } + } + } + PathArguments::Parenthesized(arguments) => { + for argument in &mut arguments.inputs { + self.visit_type_mut(argument); + } + self.visit_return_type_mut(&mut arguments.output); + } + } + } + + fn visit_return_type_mut(&mut self, return_type: &mut ReturnType) { + match return_type { + ReturnType::Default => {} + ReturnType::Type(_, output) => self.visit_type_mut(output), + } + } + + fn visit_type_param_bound_mut(&mut self, bound: &mut TypeParamBound) { + match bound { + TypeParamBound::Trait(bound) => self.visit_path_mut(&mut bound.path), + TypeParamBound::Lifetime(_) => {} + } + } + + fn visit_generics_mut(&mut self, generics: &mut Generics) { + for param in &mut generics.params { + match param { + GenericParam::Type(param) => { + for bound in &mut param.bounds { + self.visit_type_param_bound_mut(bound); + } + } + GenericParam::Lifetime(_) | GenericParam::Const(_) => {} + } + } + if let Some(where_clause) = &mut generics.where_clause { + for predicate in &mut where_clause.predicates { + match predicate { + WherePredicate::Type(predicate) => { + self.visit_type_mut(&mut predicate.bounded_ty); + for bound in &mut predicate.bounds { + self.visit_type_param_bound_mut(bound); + } + } + WherePredicate::Lifetime(_) | WherePredicate::Eq(_) => {} + } + } + } + } + + fn visit_data_mut(&mut self, data: &mut Data) { + match data { + Data::Struct(data) => { + for field in &mut data.fields { + self.visit_type_mut(&mut field.ty); + } + } + Data::Enum(data) => { + for variant in &mut data.variants { + for field in &mut variant.fields { + self.visit_type_mut(&mut field.ty); + } + } + } + Data::Union(_) => {} + } + } + + fn visit_expr_mut(&mut self, expr: &mut Expr) { + match expr { + Expr::Binary(expr) => { + self.visit_expr_mut(&mut expr.left); + self.visit_expr_mut(&mut expr.right); + } + Expr::Call(expr) => { + self.visit_expr_mut(&mut expr.func); + for arg in &mut expr.args { + self.visit_expr_mut(arg); + } + } + Expr::Cast(expr) => { + self.visit_expr_mut(&mut expr.expr); + self.visit_type_mut(&mut expr.ty); + } + Expr::Field(expr) => self.visit_expr_mut(&mut expr.base), + Expr::Index(expr) => { + self.visit_expr_mut(&mut expr.expr); + self.visit_expr_mut(&mut expr.index); + } + Expr::Paren(expr) => self.visit_expr_mut(&mut expr.expr), + Expr::Path(expr) => self.visit_expr_path_mut(expr), + Expr::Unary(expr) => self.visit_expr_mut(&mut expr.expr), + _ => {} + } + } + + fn visit_macro_mut(&mut self, _mac: &mut Macro) {} +} diff --git a/rust/serde_derive/internals/respan.rs b/rust/serde_derive/internals/respan.rs new file mode 100644 index 00000000000000..41aca1250c2189 --- /dev/null +++ b/rust/serde_derive/internals/respan.rs @@ -0,0 +1,18 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Group, Span, TokenStream, TokenTree}; + +pub(crate) fn respan(stream: TokenStream, span: Span) -> TokenStream { + stream + .into_iter() + .map(|token| respan_token(token, span)) + .collect() +} + +fn respan_token(mut token: TokenTree, span: Span) -> TokenTree { + if let TokenTree::Group(g) = &mut token { + *g = Group::new(g.delimiter(), respan(g.stream(), span)); + } + token.set_span(span); + token +} diff --git a/rust/serde_derive/internals/symbol.rs b/rust/serde_derive/internals/symbol.rs new file mode 100644 index 00000000000000..fb7cda06056f15 --- /dev/null +++ b/rust/serde_derive/internals/symbol.rs @@ -0,0 +1,71 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use std::fmt::{self, Display}; +use syn::{Ident, Path}; + +#[derive(Copy, Clone)] +pub struct Symbol(&'static str); + +pub const ALIAS: Symbol = Symbol("alias"); +pub const BORROW: Symbol = Symbol("borrow"); +pub const BOUND: Symbol = Symbol("bound"); +pub const CONTENT: Symbol = Symbol("content"); +pub const CRATE: Symbol = Symbol("crate"); +pub const DEFAULT: Symbol = Symbol("default"); +pub const DENY_UNKNOWN_FIELDS: Symbol = Symbol("deny_unknown_fields"); +pub const DESERIALIZE: Symbol = Symbol("deserialize"); +pub const DESERIALIZE_WITH: Symbol = Symbol("deserialize_with"); +pub const EXPECTING: Symbol = Symbol("expecting"); +pub const FIELD_IDENTIFIER: Symbol = Symbol("field_identifier"); +pub const FLATTEN: Symbol = Symbol("flatten"); +pub const FROM: Symbol = Symbol("from"); +pub const GETTER: Symbol = Symbol("getter"); +pub const INTO: Symbol = Symbol("into"); +pub const OTHER: Symbol = Symbol("other"); +pub const REMOTE: Symbol = Symbol("remote"); +pub const RENAME: Symbol = Symbol("rename"); +pub const RENAME_ALL: Symbol = Symbol("rename_all"); +pub const REPR: Symbol = Symbol("repr"); +pub const SERDE: Symbol = Symbol("serde"); +pub const SERIALIZE: Symbol = Symbol("serialize"); +pub const SERIALIZE_WITH: Symbol = Symbol("serialize_with"); +pub const SKIP: Symbol = Symbol("skip"); +pub const SKIP_DESERIALIZING: Symbol = Symbol("skip_deserializing"); +pub const SKIP_SERIALIZING: Symbol = Symbol("skip_serializing"); +pub const SKIP_SERIALIZING_IF: Symbol = Symbol("skip_serializing_if"); +pub const TAG: Symbol = Symbol("tag"); +pub const TRANSPARENT: Symbol = Symbol("transparent"); +pub const TRY_FROM: Symbol = Symbol("try_from"); +pub const UNTAGGED: Symbol = Symbol("untagged"); +pub const VARIANT_IDENTIFIER: Symbol = Symbol("variant_identifier"); +pub const WITH: Symbol = Symbol("with"); + +impl PartialEq for Ident { + fn eq(&self, word: &Symbol) -> bool { + self == word.0 + } +} + +impl<'a> PartialEq for &'a Ident { + fn eq(&self, word: &Symbol) -> bool { + *self == word.0 + } +} + +impl PartialEq for Path { + fn eq(&self, word: &Symbol) -> bool { + self.is_ident(word.0) + } +} + +impl<'a> PartialEq for &'a Path { + fn eq(&self, word: &Symbol) -> bool { + self.is_ident(word.0) + } +} + +impl Display for Symbol { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(self.0) + } +} diff --git a/rust/serde_derive/lib.rs b/rust/serde_derive/lib.rs new file mode 100644 index 00000000000000..de83885b9a4c0b --- /dev/null +++ b/rust/serde_derive/lib.rs @@ -0,0 +1,112 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! This crate provides Serde's two derive macros. +//! +//! ```edition2018 +//! # use serde_derive::{Serialize, Deserialize}; +//! # +//! #[derive(Serialize, Deserialize)] +//! # struct S; +//! # +//! # fn main() {} +//! ``` +//! +//! Please refer to [https://serde.rs/derive.html] for how to set this up. +//! +//! [https://serde.rs/derive.html]: https://serde.rs/derive.html + +#![doc(html_root_url = "https://docs.rs/serde_derive/1.0.156")] +#![allow(unknown_lints, bare_trait_objects)] +// Ignored clippy lints +#![allow( + // clippy false positive: https://github.com/rust-lang/rust-clippy/issues/7054 + clippy::branches_sharing_code, + clippy::cognitive_complexity, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/7575 + clippy::collapsible_match, + clippy::derive_partial_eq_without_eq, + clippy::enum_variant_names, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/6797 + clippy::manual_map, + clippy::match_like_matches_macro, + clippy::needless_pass_by_value, + clippy::too_many_arguments, + clippy::trivially_copy_pass_by_ref, + clippy::used_underscore_binding, + clippy::wildcard_in_or_patterns, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/5704 + clippy::unnested_or_patterns, +)] +// Ignored clippy_pedantic lints +#![allow( + clippy::cast_possible_truncation, + clippy::checked_conversions, + clippy::doc_markdown, + clippy::enum_glob_use, + clippy::indexing_slicing, + clippy::items_after_statements, + clippy::let_underscore_untyped, + clippy::manual_assert, + clippy::map_err_ignore, + clippy::match_same_arms, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/6984 + clippy::match_wildcard_for_single_variants, + clippy::module_name_repetitions, + clippy::must_use_candidate, + clippy::option_if_let_else, + clippy::similar_names, + clippy::single_match_else, + clippy::struct_excessive_bools, + clippy::too_many_lines, + clippy::unseparated_literal_suffix, + clippy::unused_self, + clippy::use_self, + clippy::wildcard_imports +)] +#![cfg_attr(all(test, exhaustive), feature(non_exhaustive_omitted_patterns_lint))] + +#[macro_use] +extern crate quote; +#[macro_use] +extern crate syn; + +extern crate proc_macro; +extern crate proc_macro2; + +mod internals; + +use proc_macro::TokenStream; +use syn::DeriveInput; + +#[macro_use] +mod bound; +#[macro_use] +mod fragment; + +mod de; +mod dummy; +mod pretend; +mod ser; +mod this; +mod try; + +#[proc_macro_derive(Serialize, attributes(serde))] +pub fn derive_serialize(input: TokenStream) -> TokenStream { + let mut input = parse_macro_input!(input as DeriveInput); + ser::expand_derive_serialize(&mut input) + .unwrap_or_else(to_compile_errors) + .into() +} + +#[proc_macro_derive(Deserialize, attributes(serde))] +pub fn derive_deserialize(input: TokenStream) -> TokenStream { + let mut input = parse_macro_input!(input as DeriveInput); + de::expand_derive_deserialize(&mut input) + .unwrap_or_else(to_compile_errors) + .into() +} + +fn to_compile_errors(errors: Vec) -> proc_macro2::TokenStream { + let compile_errors = errors.iter().map(syn::Error::to_compile_error); + quote!(#(#compile_errors)*) +} diff --git a/rust/serde_derive/pretend.rs b/rust/serde_derive/pretend.rs new file mode 100644 index 00000000000000..522fc8c2f2d93f --- /dev/null +++ b/rust/serde_derive/pretend.rs @@ -0,0 +1,203 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::TokenStream; +use quote::format_ident; + +use internals::ast::{Container, Data, Field, Style, Variant}; + +// Suppress dead_code warnings that would otherwise appear when using a remote +// derive. Other than this pretend code, a struct annotated with remote derive +// never has its fields referenced and an enum annotated with remote derive +// never has its variants constructed. +// +// warning: field is never used: `i` +// --> src/main.rs:4:20 +// | +// 4 | struct StructDef { i: i32 } +// | ^^^^^^ +// +// warning: variant is never constructed: `V` +// --> src/main.rs:8:16 +// | +// 8 | enum EnumDef { V } +// | ^ +// +pub fn pretend_used(cont: &Container, is_packed: bool) -> TokenStream { + let pretend_fields = pretend_fields_used(cont, is_packed); + let pretend_variants = pretend_variants_used(cont); + + quote! { + #pretend_fields + #pretend_variants + } +} + +// For structs with named fields, expands to: +// +// match None::<&T> { +// Some(T { a: __v0, b: __v1 }) => {} +// _ => {} +// } +// +// For packed structs on sufficiently new rustc, expands to: +// +// match None::<&T> { +// Some(__v @ T { a: _, b: _ }) => { +// let _ = addr_of!(__v.a); +// let _ = addr_of!(__v.b); +// } +// _ => {} +// } +// +// For packed structs on older rustc, we assume Sized and !Drop, and expand to: +// +// match None:: { +// Some(T { a: __v0, b: __v1 }) => {} +// _ => {} +// } +// +// For enums, expands to the following but only including struct variants: +// +// match None::<&T> { +// Some(T::A { a: __v0 }) => {} +// Some(T::B { b: __v0 }) => {} +// _ => {} +// } +// +fn pretend_fields_used(cont: &Container, is_packed: bool) -> TokenStream { + match &cont.data { + Data::Enum(variants) => pretend_fields_used_enum(cont, variants), + Data::Struct(Style::Struct, fields) => { + if is_packed { + pretend_fields_used_struct_packed(cont, fields) + } else { + pretend_fields_used_struct(cont, fields) + } + } + Data::Struct(_, _) => quote!(), + } +} + +fn pretend_fields_used_struct(cont: &Container, fields: &[Field]) -> TokenStream { + let type_ident = &cont.ident; + let (_, ty_generics, _) = cont.generics.split_for_impl(); + + let members = fields.iter().map(|field| &field.member); + let placeholders = (0usize..).map(|i| format_ident!("__v{}", i)); + + quote! { + match _serde::__private::None::<&#type_ident #ty_generics> { + _serde::__private::Some(#type_ident { #(#members: #placeholders),* }) => {} + _ => {} + } + } +} + +fn pretend_fields_used_struct_packed(cont: &Container, fields: &[Field]) -> TokenStream { + let type_ident = &cont.ident; + let (_, ty_generics, _) = cont.generics.split_for_impl(); + + let members = fields.iter().map(|field| &field.member).collect::>(); + + #[cfg(not(no_ptr_addr_of))] + { + quote! { + match _serde::__private::None::<&#type_ident #ty_generics> { + _serde::__private::Some(__v @ #type_ident { #(#members: _),* }) => { + #( + let _ = _serde::__private::ptr::addr_of!(__v.#members); + )* + } + _ => {} + } + } + } + + #[cfg(no_ptr_addr_of)] + { + let placeholders = (0usize..).map(|i| format_ident!("__v{}", i)); + + quote! { + match _serde::__private::None::<#type_ident #ty_generics> { + _serde::__private::Some(#type_ident { #(#members: #placeholders),* }) => {} + _ => {} + } + } + } +} + +fn pretend_fields_used_enum(cont: &Container, variants: &[Variant]) -> TokenStream { + let type_ident = &cont.ident; + let (_, ty_generics, _) = cont.generics.split_for_impl(); + + let patterns = variants + .iter() + .filter_map(|variant| match variant.style { + Style::Struct => { + let variant_ident = &variant.ident; + let members = variant.fields.iter().map(|field| &field.member); + let placeholders = (0usize..).map(|i| format_ident!("__v{}", i)); + Some(quote!(#type_ident::#variant_ident { #(#members: #placeholders),* })) + } + _ => None, + }) + .collect::>(); + + quote! { + match _serde::__private::None::<&#type_ident #ty_generics> { + #( + _serde::__private::Some(#patterns) => {} + )* + _ => {} + } + } +} + +// Expands to one of these per enum variant: +// +// match None { +// Some((__v0, __v1,)) => { +// let _ = E::V { a: __v0, b: __v1 }; +// } +// _ => {} +// } +// +fn pretend_variants_used(cont: &Container) -> TokenStream { + let variants = match &cont.data { + Data::Enum(variants) => variants, + Data::Struct(_, _) => { + return quote!(); + } + }; + + let type_ident = &cont.ident; + let (_, ty_generics, _) = cont.generics.split_for_impl(); + let turbofish = ty_generics.as_turbofish(); + + let cases = variants.iter().map(|variant| { + let variant_ident = &variant.ident; + let placeholders = &(0..variant.fields.len()) + .map(|i| format_ident!("__v{}", i)) + .collect::>(); + + let pat = match variant.style { + Style::Struct => { + let members = variant.fields.iter().map(|field| &field.member); + quote!({ #(#members: #placeholders),* }) + } + Style::Tuple | Style::Newtype => quote!(( #(#placeholders),* )), + Style::Unit => quote!(), + }; + + quote! { + match _serde::__private::None { + _serde::__private::Some((#(#placeholders,)*)) => { + let _ = #type_ident::#variant_ident #turbofish #pat; + } + _ => {} + } + } + }); + + quote!(#(#cases)*) +} diff --git a/rust/serde_derive/ser.rs b/rust/serde_derive/ser.rs new file mode 100644 index 00000000000000..00a667cea7fabc --- /dev/null +++ b/rust/serde_derive/ser.rs @@ -0,0 +1,1342 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Span, TokenStream}; +use syn::spanned::Spanned; +use syn::{self, Ident, Index, Member}; + +use bound; +use dummy; +use fragment::{Fragment, Match, Stmts}; +use internals::ast::{Container, Data, Field, Style, Variant}; +use internals::{attr, replace_receiver, Ctxt, Derive}; +use pretend; +use this; + +pub fn expand_derive_serialize( + input: &mut syn::DeriveInput, +) -> Result> { + replace_receiver(input); + + let ctxt = Ctxt::new(); + let cont = match Container::from_ast(&ctxt, input, Derive::Serialize) { + Some(cont) => cont, + None => return Err(ctxt.check().unwrap_err()), + }; + precondition(&ctxt, &cont); + ctxt.check()?; + + let ident = &cont.ident; + let params = Parameters::new(&cont); + let (impl_generics, ty_generics, where_clause) = params.generics.split_for_impl(); + let body = Stmts(serialize_body(&cont, ¶ms)); + let serde = cont.attrs.serde_path(); + + let impl_block = if let Some(remote) = cont.attrs.remote() { + let vis = &input.vis; + let used = pretend::pretend_used(&cont, params.is_packed); + quote! { + impl #impl_generics #ident #ty_generics #where_clause { + #vis fn serialize<__S>(__self: &#remote #ty_generics, __serializer: __S) -> #serde::__private::Result<__S::Ok, __S::Error> + where + __S: #serde::Serializer, + { + #used + #body + } + } + } + } else { + quote! { + #[automatically_derived] + impl #impl_generics #serde::Serialize for #ident #ty_generics #where_clause { + fn serialize<__S>(&self, __serializer: __S) -> #serde::__private::Result<__S::Ok, __S::Error> + where + __S: #serde::Serializer, + { + #body + } + } + } + }; + + Ok(dummy::wrap_in_const( + cont.attrs.custom_serde_path(), + "SERIALIZE", + ident, + impl_block, + )) +} + +fn precondition(cx: &Ctxt, cont: &Container) { + match cont.attrs.identifier() { + attr::Identifier::No => {} + attr::Identifier::Field => { + cx.error_spanned_by(cont.original, "field identifiers cannot be serialized"); + } + attr::Identifier::Variant => { + cx.error_spanned_by(cont.original, "variant identifiers cannot be serialized"); + } + } +} + +struct Parameters { + /// Variable holding the value being serialized. Either `self` for local + /// types or `__self` for remote types. + self_var: Ident, + + /// Path to the type the impl is for. Either a single `Ident` for local + /// types (does not include generic parameters) or `some::remote::Path` for + /// remote types. + this_type: syn::Path, + + /// Same as `this_type` but using `::` for generic parameters for use in + /// expression position. + this_value: syn::Path, + + /// Generics including any explicit and inferred bounds for the impl. + generics: syn::Generics, + + /// Type has a `serde(remote = "...")` attribute. + is_remote: bool, + + /// Type has a repr(packed) attribute. + is_packed: bool, +} + +impl Parameters { + fn new(cont: &Container) -> Self { + let is_remote = cont.attrs.remote().is_some(); + let self_var = if is_remote { + Ident::new("__self", Span::call_site()) + } else { + Ident::new("self", Span::call_site()) + }; + + let this_type = this::this_type(cont); + let this_value = this::this_value(cont); + let is_packed = cont.attrs.is_packed(); + let generics = build_generics(cont); + + Parameters { + self_var, + this_type, + this_value, + generics, + is_remote, + is_packed, + } + } + + /// Type name to use in error messages and `&'static str` arguments to + /// various Serializer methods. + fn type_name(&self) -> String { + self.this_type.segments.last().unwrap().ident.to_string() + } +} + +// All the generics in the input, plus a bound `T: Serialize` for each generic +// field type that will be serialized by us. +fn build_generics(cont: &Container) -> syn::Generics { + let generics = bound::without_defaults(cont.generics); + + let generics = + bound::with_where_predicates_from_fields(cont, &generics, attr::Field::ser_bound); + + let generics = + bound::with_where_predicates_from_variants(cont, &generics, attr::Variant::ser_bound); + + match cont.attrs.ser_bound() { + Some(predicates) => bound::with_where_predicates(&generics, predicates), + None => bound::with_bound( + cont, + &generics, + needs_serialize_bound, + &parse_quote!(_serde::Serialize), + ), + } +} + +// Fields with a `skip_serializing` or `serialize_with` attribute, or which +// belong to a variant with a 'skip_serializing` or `serialize_with` attribute, +// are not serialized by us so we do not generate a bound. Fields with a `bound` +// attribute specify their own bound so we do not generate one. All other fields +// may need a `T: Serialize` bound where T is the type of the field. +fn needs_serialize_bound(field: &attr::Field, variant: Option<&attr::Variant>) -> bool { + !field.skip_serializing() + && field.serialize_with().is_none() + && field.ser_bound().is_none() + && variant.map_or(true, |variant| { + !variant.skip_serializing() + && variant.serialize_with().is_none() + && variant.ser_bound().is_none() + }) +} + +fn serialize_body(cont: &Container, params: &Parameters) -> Fragment { + if cont.attrs.transparent() { + serialize_transparent(cont, params) + } else if let Some(type_into) = cont.attrs.type_into() { + serialize_into(params, type_into) + } else { + match &cont.data { + Data::Enum(variants) => serialize_enum(params, variants, &cont.attrs), + Data::Struct(Style::Struct, fields) => serialize_struct(params, fields, &cont.attrs), + Data::Struct(Style::Tuple, fields) => { + serialize_tuple_struct(params, fields, &cont.attrs) + } + Data::Struct(Style::Newtype, fields) => { + serialize_newtype_struct(params, &fields[0], &cont.attrs) + } + Data::Struct(Style::Unit, _) => serialize_unit_struct(&cont.attrs), + } + } +} + +fn serialize_transparent(cont: &Container, params: &Parameters) -> Fragment { + let fields = match &cont.data { + Data::Struct(_, fields) => fields, + Data::Enum(_) => unreachable!(), + }; + + let self_var = ¶ms.self_var; + let transparent_field = fields.iter().find(|f| f.attrs.transparent()).unwrap(); + let member = &transparent_field.member; + + let path = match transparent_field.attrs.serialize_with() { + Some(path) => quote!(#path), + None => { + let span = transparent_field.original.span(); + quote_spanned!(span=> _serde::Serialize::serialize) + } + }; + + quote_block! { + #path(&#self_var.#member, __serializer) + } +} + +fn serialize_into(params: &Parameters, type_into: &syn::Type) -> Fragment { + let self_var = ¶ms.self_var; + quote_block! { + _serde::Serialize::serialize( + &_serde::__private::Into::<#type_into>::into(_serde::__private::Clone::clone(#self_var)), + __serializer) + } +} + +fn serialize_unit_struct(cattrs: &attr::Container) -> Fragment { + let type_name = cattrs.name().serialize_name(); + + quote_expr! { + _serde::Serializer::serialize_unit_struct(__serializer, #type_name) + } +} + +fn serialize_newtype_struct( + params: &Parameters, + field: &Field, + cattrs: &attr::Container, +) -> Fragment { + let type_name = cattrs.name().serialize_name(); + + let mut field_expr = get_member( + params, + field, + &Member::Unnamed(Index { + index: 0, + span: Span::call_site(), + }), + ); + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::Serializer::serialize_newtype_struct); + quote_expr! { + #func(__serializer, #type_name, #field_expr) + } +} + +fn serialize_tuple_struct( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> Fragment { + let serialize_stmts = + serialize_tuple_struct_visitor(fields, params, false, &TupleTrait::SerializeTupleStruct); + + let type_name = cattrs.name().serialize_name(); + + let mut serialized_fields = fields + .iter() + .enumerate() + .filter(|(_, field)| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some()); + + let len = serialized_fields + .map(|(i, field)| match field.attrs.skip_serializing_if() { + None => quote!(1), + Some(path) => { + let index = syn::Index { + index: i as u32, + span: Span::call_site(), + }; + let field_expr = get_member(params, field, &Member::Unnamed(index)); + quote!(if #path(#field_expr) { 0 } else { 1 }) + } + }) + .fold(quote!(0), |sum, expr| quote!(#sum + #expr)); + + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_tuple_struct(__serializer, #type_name, #len)); + #(#serialize_stmts)* + _serde::ser::SerializeTupleStruct::end(__serde_state) + } +} + +fn serialize_struct(params: &Parameters, fields: &[Field], cattrs: &attr::Container) -> Fragment { + assert!(fields.len() as u64 <= u64::from(u32::max_value())); + + if cattrs.has_flatten() { + serialize_struct_as_map(params, fields, cattrs) + } else { + serialize_struct_as_struct(params, fields, cattrs) + } +} + +fn serialize_struct_tag_field(cattrs: &attr::Container, struct_trait: &StructTrait) -> TokenStream { + match cattrs.tag() { + attr::TagType::Internal { tag } => { + let type_name = cattrs.name().serialize_name(); + let func = struct_trait.serialize_field(Span::call_site()); + quote! { + try!(#func(&mut __serde_state, #tag, #type_name)); + } + } + _ => quote! {}, + } +} + +fn serialize_struct_as_struct( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> Fragment { + let serialize_fields = + serialize_struct_visitor(fields, params, false, &StructTrait::SerializeStruct); + + let type_name = cattrs.name().serialize_name(); + + let tag_field = serialize_struct_tag_field(cattrs, &StructTrait::SerializeStruct); + let tag_field_exists = !tag_field.is_empty(); + + let mut serialized_fields = fields + .iter() + .filter(|&field| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some() || tag_field_exists); + + let len = serialized_fields + .map(|field| match field.attrs.skip_serializing_if() { + None => quote!(1), + Some(path) => { + let field_expr = get_member(params, field, &field.member); + quote!(if #path(#field_expr) { 0 } else { 1 }) + } + }) + .fold( + quote!(#tag_field_exists as usize), + |sum, expr| quote!(#sum + #expr), + ); + + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_struct(__serializer, #type_name, #len)); + #tag_field + #(#serialize_fields)* + _serde::ser::SerializeStruct::end(__serde_state) + } +} + +fn serialize_struct_as_map( + params: &Parameters, + fields: &[Field], + cattrs: &attr::Container, +) -> Fragment { + let serialize_fields = + serialize_struct_visitor(fields, params, false, &StructTrait::SerializeMap); + + let tag_field = serialize_struct_tag_field(cattrs, &StructTrait::SerializeMap); + let tag_field_exists = !tag_field.is_empty(); + + let mut serialized_fields = fields + .iter() + .filter(|&field| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some() || tag_field_exists); + + let len = if cattrs.has_flatten() { + quote!(_serde::__private::None) + } else { + let len = serialized_fields + .map(|field| match field.attrs.skip_serializing_if() { + None => quote!(1), + Some(path) => { + let field_expr = get_member(params, field, &field.member); + quote!(if #path(#field_expr) { 0 } else { 1 }) + } + }) + .fold( + quote!(#tag_field_exists as usize), + |sum, expr| quote!(#sum + #expr), + ); + quote!(_serde::__private::Some(#len)) + }; + + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_map(__serializer, #len)); + #tag_field + #(#serialize_fields)* + _serde::ser::SerializeMap::end(__serde_state) + } +} + +fn serialize_enum(params: &Parameters, variants: &[Variant], cattrs: &attr::Container) -> Fragment { + assert!(variants.len() as u64 <= u64::from(u32::max_value())); + + let self_var = ¶ms.self_var; + + let arms: Vec<_> = variants + .iter() + .enumerate() + .map(|(variant_index, variant)| { + serialize_variant(params, variant, variant_index as u32, cattrs) + }) + .collect(); + + quote_expr! { + match *#self_var { + #(#arms)* + } + } +} + +fn serialize_variant( + params: &Parameters, + variant: &Variant, + variant_index: u32, + cattrs: &attr::Container, +) -> TokenStream { + let this_value = ¶ms.this_value; + let variant_ident = &variant.ident; + + if variant.attrs.skip_serializing() { + let skipped_msg = format!( + "the enum variant {}::{} cannot be serialized", + params.type_name(), + variant_ident + ); + let skipped_err = quote! { + _serde::__private::Err(_serde::ser::Error::custom(#skipped_msg)) + }; + let fields_pat = match variant.style { + Style::Unit => quote!(), + Style::Newtype | Style::Tuple => quote!((..)), + Style::Struct => quote!({ .. }), + }; + quote! { + #this_value::#variant_ident #fields_pat => #skipped_err, + } + } else { + // variant wasn't skipped + let case = match variant.style { + Style::Unit => { + quote! { + #this_value::#variant_ident + } + } + Style::Newtype => { + quote! { + #this_value::#variant_ident(ref __field0) + } + } + Style::Tuple => { + let field_names = (0..variant.fields.len()) + .map(|i| Ident::new(&format!("__field{}", i), Span::call_site())); + quote! { + #this_value::#variant_ident(#(ref #field_names),*) + } + } + Style::Struct => { + let members = variant.fields.iter().map(|f| &f.member); + quote! { + #this_value::#variant_ident { #(ref #members),* } + } + } + }; + + let body = Match(match cattrs.tag() { + attr::TagType::External => { + serialize_externally_tagged_variant(params, variant, variant_index, cattrs) + } + attr::TagType::Internal { tag } => { + serialize_internally_tagged_variant(params, variant, cattrs, tag) + } + attr::TagType::Adjacent { tag, content } => { + serialize_adjacently_tagged_variant(params, variant, cattrs, tag, content) + } + attr::TagType::None => serialize_untagged_variant(params, variant, cattrs), + }); + + quote! { + #case => #body + } + } +} + +fn serialize_externally_tagged_variant( + params: &Parameters, + variant: &Variant, + variant_index: u32, + cattrs: &attr::Container, +) -> Fragment { + let type_name = cattrs.name().serialize_name(); + let variant_name = variant.attrs.name().serialize_name(); + + if let Some(path) = variant.attrs.serialize_with() { + let ser = wrap_serialize_variant_with(params, path, variant); + return quote_expr! { + _serde::Serializer::serialize_newtype_variant( + __serializer, + #type_name, + #variant_index, + #variant_name, + #ser, + ) + }; + } + + match effective_style(variant) { + Style::Unit => { + quote_expr! { + _serde::Serializer::serialize_unit_variant( + __serializer, + #type_name, + #variant_index, + #variant_name, + ) + } + } + Style::Newtype => { + let field = &variant.fields[0]; + let mut field_expr = quote!(__field0); + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::Serializer::serialize_newtype_variant); + quote_expr! { + #func( + __serializer, + #type_name, + #variant_index, + #variant_name, + #field_expr, + ) + } + } + Style::Tuple => serialize_tuple_variant( + TupleVariant::ExternallyTagged { + type_name, + variant_index, + variant_name, + }, + params, + &variant.fields, + ), + Style::Struct => serialize_struct_variant( + StructVariant::ExternallyTagged { + variant_index, + variant_name, + }, + params, + &variant.fields, + &type_name, + ), + } +} + +fn serialize_internally_tagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, + tag: &str, +) -> Fragment { + let type_name = cattrs.name().serialize_name(); + let variant_name = variant.attrs.name().serialize_name(); + + let enum_ident_str = params.type_name(); + let variant_ident_str = variant.ident.to_string(); + + if let Some(path) = variant.attrs.serialize_with() { + let ser = wrap_serialize_variant_with(params, path, variant); + return quote_expr! { + _serde::__private::ser::serialize_tagged_newtype( + __serializer, + #enum_ident_str, + #variant_ident_str, + #tag, + #variant_name, + #ser, + ) + }; + } + + match effective_style(variant) { + Style::Unit => { + quote_block! { + let mut __struct = try!(_serde::Serializer::serialize_struct( + __serializer, #type_name, 1)); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __struct, #tag, #variant_name)); + _serde::ser::SerializeStruct::end(__struct) + } + } + Style::Newtype => { + let field = &variant.fields[0]; + let mut field_expr = quote!(__field0); + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::__private::ser::serialize_tagged_newtype); + quote_expr! { + #func( + __serializer, + #enum_ident_str, + #variant_ident_str, + #tag, + #variant_name, + #field_expr, + ) + } + } + Style::Struct => serialize_struct_variant( + StructVariant::InternallyTagged { tag, variant_name }, + params, + &variant.fields, + &type_name, + ), + Style::Tuple => unreachable!("checked in serde_derive_internals"), + } +} + +fn serialize_adjacently_tagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, + tag: &str, + content: &str, +) -> Fragment { + let this_type = ¶ms.this_type; + let type_name = cattrs.name().serialize_name(); + let variant_name = variant.attrs.name().serialize_name(); + + let inner = Stmts(if let Some(path) = variant.attrs.serialize_with() { + let ser = wrap_serialize_variant_with(params, path, variant); + quote_expr! { + _serde::Serialize::serialize(#ser, __serializer) + } + } else { + match effective_style(variant) { + Style::Unit => { + return quote_block! { + let mut __struct = try!(_serde::Serializer::serialize_struct( + __serializer, #type_name, 1)); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __struct, #tag, #variant_name)); + _serde::ser::SerializeStruct::end(__struct) + }; + } + Style::Newtype => { + let field = &variant.fields[0]; + let mut field_expr = quote!(__field0); + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::ser::SerializeStruct::serialize_field); + return quote_block! { + let mut __struct = try!(_serde::Serializer::serialize_struct( + __serializer, #type_name, 2)); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __struct, #tag, #variant_name)); + try!(#func( + &mut __struct, #content, #field_expr)); + _serde::ser::SerializeStruct::end(__struct) + }; + } + Style::Tuple => { + serialize_tuple_variant(TupleVariant::Untagged, params, &variant.fields) + } + Style::Struct => serialize_struct_variant( + StructVariant::Untagged, + params, + &variant.fields, + &variant_name, + ), + } + }); + + let fields_ty = variant.fields.iter().map(|f| &f.ty); + let fields_ident: &Vec<_> = &match variant.style { + Style::Unit => { + if variant.attrs.serialize_with().is_some() { + vec![] + } else { + unreachable!() + } + } + Style::Newtype => vec![Member::Named(Ident::new("__field0", Span::call_site()))], + Style::Tuple => (0..variant.fields.len()) + .map(|i| Member::Named(Ident::new(&format!("__field{}", i), Span::call_site()))) + .collect(), + Style::Struct => variant.fields.iter().map(|f| f.member.clone()).collect(), + }; + + let (_, ty_generics, where_clause) = params.generics.split_for_impl(); + + let wrapper_generics = if fields_ident.is_empty() { + params.generics.clone() + } else { + bound::with_lifetime_bound(¶ms.generics, "'__a") + }; + let (wrapper_impl_generics, wrapper_ty_generics, _) = wrapper_generics.split_for_impl(); + + quote_block! { + struct __AdjacentlyTagged #wrapper_generics #where_clause { + data: (#(&'__a #fields_ty,)*), + phantom: _serde::__private::PhantomData<#this_type #ty_generics>, + } + + impl #wrapper_impl_generics _serde::Serialize for __AdjacentlyTagged #wrapper_ty_generics #where_clause { + fn serialize<__S>(&self, __serializer: __S) -> _serde::__private::Result<__S::Ok, __S::Error> + where + __S: _serde::Serializer, + { + // Elements that have skip_serializing will be unused. + #[allow(unused_variables)] + let (#(#fields_ident,)*) = self.data; + #inner + } + } + + let mut __struct = try!(_serde::Serializer::serialize_struct( + __serializer, #type_name, 2)); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __struct, #tag, #variant_name)); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __struct, #content, &__AdjacentlyTagged { + data: (#(#fields_ident,)*), + phantom: _serde::__private::PhantomData::<#this_type #ty_generics>, + })); + _serde::ser::SerializeStruct::end(__struct) + } +} + +fn serialize_untagged_variant( + params: &Parameters, + variant: &Variant, + cattrs: &attr::Container, +) -> Fragment { + if let Some(path) = variant.attrs.serialize_with() { + let ser = wrap_serialize_variant_with(params, path, variant); + return quote_expr! { + _serde::Serialize::serialize(#ser, __serializer) + }; + } + + match effective_style(variant) { + Style::Unit => { + quote_expr! { + _serde::Serializer::serialize_unit(__serializer) + } + } + Style::Newtype => { + let field = &variant.fields[0]; + let mut field_expr = quote!(__field0); + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = quote_spanned!(span=> _serde::Serialize::serialize); + quote_expr! { + #func(#field_expr, __serializer) + } + } + Style::Tuple => serialize_tuple_variant(TupleVariant::Untagged, params, &variant.fields), + Style::Struct => { + let type_name = cattrs.name().serialize_name(); + serialize_struct_variant(StructVariant::Untagged, params, &variant.fields, &type_name) + } + } +} + +enum TupleVariant { + ExternallyTagged { + type_name: String, + variant_index: u32, + variant_name: String, + }, + Untagged, +} + +fn serialize_tuple_variant( + context: TupleVariant, + params: &Parameters, + fields: &[Field], +) -> Fragment { + let tuple_trait = match context { + TupleVariant::ExternallyTagged { .. } => TupleTrait::SerializeTupleVariant, + TupleVariant::Untagged => TupleTrait::SerializeTuple, + }; + + let serialize_stmts = serialize_tuple_struct_visitor(fields, params, true, &tuple_trait); + + let mut serialized_fields = fields + .iter() + .enumerate() + .filter(|(_, field)| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some()); + + let len = serialized_fields + .map(|(i, field)| match field.attrs.skip_serializing_if() { + None => quote!(1), + Some(path) => { + let field_expr = Ident::new(&format!("__field{}", i), Span::call_site()); + quote!(if #path(#field_expr) { 0 } else { 1 }) + } + }) + .fold(quote!(0), |sum, expr| quote!(#sum + #expr)); + + match context { + TupleVariant::ExternallyTagged { + type_name, + variant_index, + variant_name, + } => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_tuple_variant( + __serializer, + #type_name, + #variant_index, + #variant_name, + #len)); + #(#serialize_stmts)* + _serde::ser::SerializeTupleVariant::end(__serde_state) + } + } + TupleVariant::Untagged => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_tuple( + __serializer, + #len)); + #(#serialize_stmts)* + _serde::ser::SerializeTuple::end(__serde_state) + } + } + } +} + +enum StructVariant<'a> { + ExternallyTagged { + variant_index: u32, + variant_name: String, + }, + InternallyTagged { + tag: &'a str, + variant_name: String, + }, + Untagged, +} + +fn serialize_struct_variant( + context: StructVariant, + params: &Parameters, + fields: &[Field], + name: &str, +) -> Fragment { + if fields.iter().any(|field| field.attrs.flatten()) { + return serialize_struct_variant_with_flatten(context, params, fields, name); + } + + let struct_trait = match context { + StructVariant::ExternallyTagged { .. } => StructTrait::SerializeStructVariant, + StructVariant::InternallyTagged { .. } | StructVariant::Untagged => { + StructTrait::SerializeStruct + } + }; + + let serialize_fields = serialize_struct_visitor(fields, params, true, &struct_trait); + + let mut serialized_fields = fields + .iter() + .filter(|&field| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some()); + + let len = serialized_fields + .map(|field| { + let member = &field.member; + + match field.attrs.skip_serializing_if() { + Some(path) => quote!(if #path(#member) { 0 } else { 1 }), + None => quote!(1), + } + }) + .fold(quote!(0), |sum, expr| quote!(#sum + #expr)); + + match context { + StructVariant::ExternallyTagged { + variant_index, + variant_name, + } => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_struct_variant( + __serializer, + #name, + #variant_index, + #variant_name, + #len, + )); + #(#serialize_fields)* + _serde::ser::SerializeStructVariant::end(__serde_state) + } + } + StructVariant::InternallyTagged { tag, variant_name } => { + quote_block! { + let mut __serde_state = try!(_serde::Serializer::serialize_struct( + __serializer, + #name, + #len + 1, + )); + try!(_serde::ser::SerializeStruct::serialize_field( + &mut __serde_state, + #tag, + #variant_name, + )); + #(#serialize_fields)* + _serde::ser::SerializeStruct::end(__serde_state) + } + } + StructVariant::Untagged => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_struct( + __serializer, + #name, + #len, + )); + #(#serialize_fields)* + _serde::ser::SerializeStruct::end(__serde_state) + } + } + } +} + +fn serialize_struct_variant_with_flatten( + context: StructVariant, + params: &Parameters, + fields: &[Field], + name: &str, +) -> Fragment { + let struct_trait = StructTrait::SerializeMap; + let serialize_fields = serialize_struct_visitor(fields, params, true, &struct_trait); + + let mut serialized_fields = fields + .iter() + .filter(|&field| !field.attrs.skip_serializing()) + .peekable(); + + let let_mut = mut_if(serialized_fields.peek().is_some()); + + match context { + StructVariant::ExternallyTagged { + variant_index, + variant_name, + } => { + let this_type = ¶ms.this_type; + let fields_ty = fields.iter().map(|f| &f.ty); + let members = &fields.iter().map(|f| &f.member).collect::>(); + + let (_, ty_generics, where_clause) = params.generics.split_for_impl(); + let wrapper_generics = bound::with_lifetime_bound(¶ms.generics, "'__a"); + let (wrapper_impl_generics, wrapper_ty_generics, _) = wrapper_generics.split_for_impl(); + + quote_block! { + struct __EnumFlatten #wrapper_generics #where_clause { + data: (#(&'__a #fields_ty,)*), + phantom: _serde::__private::PhantomData<#this_type #ty_generics>, + } + + impl #wrapper_impl_generics _serde::Serialize for __EnumFlatten #wrapper_ty_generics #where_clause { + fn serialize<__S>(&self, __serializer: __S) -> _serde::__private::Result<__S::Ok, __S::Error> + where + __S: _serde::Serializer, + { + let (#(#members,)*) = self.data; + let #let_mut __serde_state = try!(_serde::Serializer::serialize_map( + __serializer, + _serde::__private::None)); + #(#serialize_fields)* + _serde::ser::SerializeMap::end(__serde_state) + } + } + + _serde::Serializer::serialize_newtype_variant( + __serializer, + #name, + #variant_index, + #variant_name, + &__EnumFlatten { + data: (#(#members,)*), + phantom: _serde::__private::PhantomData::<#this_type #ty_generics>, + }) + } + } + StructVariant::InternallyTagged { tag, variant_name } => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_map( + __serializer, + _serde::__private::None)); + try!(_serde::ser::SerializeMap::serialize_entry( + &mut __serde_state, + #tag, + #variant_name, + )); + #(#serialize_fields)* + _serde::ser::SerializeMap::end(__serde_state) + } + } + StructVariant::Untagged => { + quote_block! { + let #let_mut __serde_state = try!(_serde::Serializer::serialize_map( + __serializer, + _serde::__private::None)); + #(#serialize_fields)* + _serde::ser::SerializeMap::end(__serde_state) + } + } + } +} + +fn serialize_tuple_struct_visitor( + fields: &[Field], + params: &Parameters, + is_enum: bool, + tuple_trait: &TupleTrait, +) -> Vec { + fields + .iter() + .enumerate() + .filter(|(_, field)| !field.attrs.skip_serializing()) + .map(|(i, field)| { + let mut field_expr = if is_enum { + let id = Ident::new(&format!("__field{}", i), Span::call_site()); + quote!(#id) + } else { + get_member( + params, + field, + &Member::Unnamed(Index { + index: i as u32, + span: Span::call_site(), + }), + ) + }; + + let skip = field + .attrs + .skip_serializing_if() + .map(|path| quote!(#path(#field_expr))); + + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let func = tuple_trait.serialize_element(span); + let ser = quote! { + try!(#func(&mut __serde_state, #field_expr)); + }; + + match skip { + None => ser, + Some(skip) => quote!(if !#skip { #ser }), + } + }) + .collect() +} + +fn serialize_struct_visitor( + fields: &[Field], + params: &Parameters, + is_enum: bool, + struct_trait: &StructTrait, +) -> Vec { + fields + .iter() + .filter(|&field| !field.attrs.skip_serializing()) + .map(|field| { + let member = &field.member; + + let mut field_expr = if is_enum { + quote!(#member) + } else { + get_member(params, field, member) + }; + + let key_expr = field.attrs.name().serialize_name(); + + let skip = field + .attrs + .skip_serializing_if() + .map(|path| quote!(#path(#field_expr))); + + if let Some(path) = field.attrs.serialize_with() { + field_expr = wrap_serialize_field_with(params, field.ty, path, &field_expr); + } + + let span = field.original.span(); + let ser = if field.attrs.flatten() { + let func = quote_spanned!(span=> _serde::Serialize::serialize); + quote! { + try!(#func(&#field_expr, _serde::__private::ser::FlatMapSerializer(&mut __serde_state))); + } + } else { + let func = struct_trait.serialize_field(span); + quote! { + try!(#func(&mut __serde_state, #key_expr, #field_expr)); + } + }; + + match skip { + None => ser, + Some(skip) => { + if let Some(skip_func) = struct_trait.skip_field(span) { + quote! { + if !#skip { + #ser + } else { + try!(#skip_func(&mut __serde_state, #key_expr)); + } + } + } else { + quote! { + if !#skip { + #ser + } + } + } + } + } + }) + .collect() +} + +fn wrap_serialize_field_with( + params: &Parameters, + field_ty: &syn::Type, + serialize_with: &syn::ExprPath, + field_expr: &TokenStream, +) -> TokenStream { + wrap_serialize_with(params, serialize_with, &[field_ty], &[quote!(#field_expr)]) +} + +fn wrap_serialize_variant_with( + params: &Parameters, + serialize_with: &syn::ExprPath, + variant: &Variant, +) -> TokenStream { + let field_tys: Vec<_> = variant.fields.iter().map(|field| field.ty).collect(); + let field_exprs: Vec<_> = variant + .fields + .iter() + .map(|field| { + let id = match &field.member { + Member::Named(ident) => ident.clone(), + Member::Unnamed(member) => { + Ident::new(&format!("__field{}", member.index), Span::call_site()) + } + }; + quote!(#id) + }) + .collect(); + wrap_serialize_with( + params, + serialize_with, + field_tys.as_slice(), + field_exprs.as_slice(), + ) +} + +fn wrap_serialize_with( + params: &Parameters, + serialize_with: &syn::ExprPath, + field_tys: &[&syn::Type], + field_exprs: &[TokenStream], +) -> TokenStream { + let this_type = ¶ms.this_type; + let (_, ty_generics, where_clause) = params.generics.split_for_impl(); + + let wrapper_generics = if field_exprs.is_empty() { + params.generics.clone() + } else { + bound::with_lifetime_bound(¶ms.generics, "'__a") + }; + let (wrapper_impl_generics, wrapper_ty_generics, _) = wrapper_generics.split_for_impl(); + + let field_access = (0..field_exprs.len()).map(|n| { + Member::Unnamed(Index { + index: n as u32, + span: Span::call_site(), + }) + }); + + quote!({ + struct __SerializeWith #wrapper_impl_generics #where_clause { + values: (#(&'__a #field_tys, )*), + phantom: _serde::__private::PhantomData<#this_type #ty_generics>, + } + + impl #wrapper_impl_generics _serde::Serialize for __SerializeWith #wrapper_ty_generics #where_clause { + fn serialize<__S>(&self, __s: __S) -> _serde::__private::Result<__S::Ok, __S::Error> + where + __S: _serde::Serializer, + { + #serialize_with(#(self.values.#field_access, )* __s) + } + } + + &__SerializeWith { + values: (#(#field_exprs, )*), + phantom: _serde::__private::PhantomData::<#this_type #ty_generics>, + } + }) +} + +// Serialization of an empty struct results in code like: +// +// let mut __serde_state = try!(serializer.serialize_struct("S", 0)); +// _serde::ser::SerializeStruct::end(__serde_state) +// +// where we want to omit the `mut` to avoid a warning. +fn mut_if(is_mut: bool) -> Option { + if is_mut { + Some(quote!(mut)) + } else { + None + } +} + +fn get_member(params: &Parameters, field: &Field, member: &Member) -> TokenStream { + let self_var = ¶ms.self_var; + match (params.is_remote, field.attrs.getter()) { + (false, None) => { + if params.is_packed { + quote!(&{#self_var.#member}) + } else { + quote!(&#self_var.#member) + } + } + (true, None) => { + let inner = if params.is_packed { + quote!(&{#self_var.#member}) + } else { + quote!(&#self_var.#member) + }; + let ty = field.ty; + quote!(_serde::__private::ser::constrain::<#ty>(#inner)) + } + (true, Some(getter)) => { + let ty = field.ty; + quote!(_serde::__private::ser::constrain::<#ty>(&#getter(#self_var))) + } + (false, Some(_)) => { + unreachable!("getter is only allowed for remote impls"); + } + } +} + +fn effective_style(variant: &Variant) -> Style { + match variant.style { + Style::Newtype if variant.fields[0].attrs.skip_serializing() => Style::Unit, + other => other, + } +} + +enum StructTrait { + SerializeMap, + SerializeStruct, + SerializeStructVariant, +} + +impl StructTrait { + fn serialize_field(&self, span: Span) -> TokenStream { + match *self { + StructTrait::SerializeMap => { + quote_spanned!(span=> _serde::ser::SerializeMap::serialize_entry) + } + StructTrait::SerializeStruct => { + quote_spanned!(span=> _serde::ser::SerializeStruct::serialize_field) + } + StructTrait::SerializeStructVariant => { + quote_spanned!(span=> _serde::ser::SerializeStructVariant::serialize_field) + } + } + } + + fn skip_field(&self, span: Span) -> Option { + match *self { + StructTrait::SerializeMap => None, + StructTrait::SerializeStruct => { + Some(quote_spanned!(span=> _serde::ser::SerializeStruct::skip_field)) + } + StructTrait::SerializeStructVariant => { + Some(quote_spanned!(span=> _serde::ser::SerializeStructVariant::skip_field)) + } + } + } +} + +enum TupleTrait { + SerializeTuple, + SerializeTupleStruct, + SerializeTupleVariant, +} + +impl TupleTrait { + fn serialize_element(&self, span: Span) -> TokenStream { + match *self { + TupleTrait::SerializeTuple => { + quote_spanned!(span=> _serde::ser::SerializeTuple::serialize_element) + } + TupleTrait::SerializeTupleStruct => { + quote_spanned!(span=> _serde::ser::SerializeTupleStruct::serialize_field) + } + TupleTrait::SerializeTupleVariant => { + quote_spanned!(span=> _serde::ser::SerializeTupleVariant::serialize_field) + } + } + } +} diff --git a/rust/serde_derive/this.rs b/rust/serde_derive/this.rs new file mode 100644 index 00000000000000..7a27a7c7dbc294 --- /dev/null +++ b/rust/serde_derive/this.rs @@ -0,0 +1,34 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use internals::ast::Container; +use syn::{Path, PathArguments, Token}; + +pub fn this_type(cont: &Container) -> Path { + if let Some(remote) = cont.attrs.remote() { + let mut this = remote.clone(); + for segment in &mut this.segments { + if let PathArguments::AngleBracketed(arguments) = &mut segment.arguments { + arguments.colon2_token = None; + } + } + this + } else { + Path::from(cont.ident.clone()) + } +} + +pub fn this_value(cont: &Container) -> Path { + if let Some(remote) = cont.attrs.remote() { + let mut this = remote.clone(); + for segment in &mut this.segments { + if let PathArguments::AngleBracketed(arguments) = &mut segment.arguments { + if arguments.colon2_token.is_none() { + arguments.colon2_token = Some(Token![::](arguments.lt_token.span)); + } + } + } + this + } else { + Path::from(cont.ident.clone()) + } +} diff --git a/rust/serde_derive/try.rs b/rust/serde_derive/try.rs new file mode 100644 index 00000000000000..afed0fedb93820 --- /dev/null +++ b/rust/serde_derive/try.rs @@ -0,0 +1,26 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Punct, Spacing, TokenStream}; + +// None of our generated code requires the `From::from` error conversion +// performed by the standard library's `try!` macro. With this simplified macro +// we see a significant improvement in type checking and borrow checking time of +// the generated code and a slight improvement in binary size. +pub fn replacement() -> TokenStream { + // Cannot pass `$expr` to `quote!` prior to Rust 1.17.0 so interpolate it. + let dollar = Punct::new('$', Spacing::Alone); + + quote! { + #[allow(unused_macros)] + macro_rules! try { + (#dollar __expr:expr) => { + match #dollar __expr { + _serde::__private::Ok(__val) => __val, + _serde::__private::Err(__err) => { + return _serde::__private::Err(__err); + } + } + } + } + } +} diff --git a/rust/syn/attr.rs b/rust/syn/attr.rs new file mode 100644 index 00000000000000..cbcd158f45f6f3 --- /dev/null +++ b/rust/syn/attr.rs @@ -0,0 +1,664 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; +use proc_macro2::TokenStream; +use std::iter; +use std::slice; + +#[cfg(feature = "parsing")] +use crate::parse::{Parse, ParseBuffer, ParseStream, Parser, Result}; +#[cfg(feature = "parsing")] +use crate::punctuated::Pair; + +ast_struct! { + /// An attribute like `#[repr(transparent)]`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + ///
+ /// + /// # Syntax + /// + /// Rust has six types of attributes. + /// + /// - Outer attributes like `#[repr(transparent)]`. These appear outside or + /// in front of the item they describe. + /// - Inner attributes like `#![feature(proc_macro)]`. These appear inside + /// of the item they describe, usually a module. + /// - Outer doc comments like `/// # Example`. + /// - Inner doc comments like `//! Please file an issue`. + /// - Outer block comments `/** # Example */`. + /// - Inner block comments `/*! Please file an issue */`. + /// + /// The `style` field of type `AttrStyle` distinguishes whether an attribute + /// is outer or inner. Doc comments and block comments are promoted to + /// attributes, as this is how they are processed by the compiler and by + /// `macro_rules!` macros. + /// + /// The `path` field gives the possibly colon-delimited path against which + /// the attribute is resolved. It is equal to `"doc"` for desugared doc + /// comments. The `tokens` field contains the rest of the attribute body as + /// tokens. + /// + /// ```text + /// #[derive(Copy)] #[crate::precondition x < 5] + /// ^^^^^^~~~~~~ ^^^^^^^^^^^^^^^^^^^ ~~~~~ + /// path tokens path tokens + /// ``` + /// + ///
+ /// + /// # Parsing from tokens to Attribute + /// + /// This type does not implement the [`Parse`] trait and thus cannot be + /// parsed directly by [`ParseStream::parse`]. Instead use + /// [`ParseStream::call`] with one of the two parser functions + /// [`Attribute::parse_outer`] or [`Attribute::parse_inner`] depending on + /// which you intend to parse. + /// + /// [`Parse`]: parse::Parse + /// [`ParseStream::parse`]: parse::ParseBuffer::parse + /// [`ParseStream::call`]: parse::ParseBuffer::call + /// + /// ``` + /// use syn::{Attribute, Ident, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Parses a unit struct with attributes. + /// // + /// // #[path = "s.tmpl"] + /// // struct S; + /// struct UnitStruct { + /// attrs: Vec, + /// struct_token: Token![struct], + /// name: Ident, + /// semi_token: Token![;], + /// } + /// + /// impl Parse for UnitStruct { + /// fn parse(input: ParseStream) -> Result { + /// Ok(UnitStruct { + /// attrs: input.call(Attribute::parse_outer)?, + /// struct_token: input.parse()?, + /// name: input.parse()?, + /// semi_token: input.parse()?, + /// }) + /// } + /// } + /// ``` + /// + ///


+ /// + /// # Parsing from Attribute to structured arguments + /// + /// The grammar of attributes in Rust is very flexible, which makes the + /// syntax tree not that useful on its own. In particular, arguments of the + /// attribute are held in an arbitrary `tokens: TokenStream`. Macros are + /// expected to check the `path` of the attribute, decide whether they + /// recognize it, and then parse the remaining tokens according to whatever + /// grammar they wish to require for that kind of attribute. + /// + /// If the attribute you are parsing is expected to conform to the + /// conventional structured form of attribute, use [`parse_meta()`] to + /// obtain that structured representation. If the attribute follows some + /// other grammar of its own, use [`parse_args()`] to parse that into the + /// expected data structure. + /// + /// [`parse_meta()`]: Attribute::parse_meta + /// [`parse_args()`]: Attribute::parse_args + /// + ///


+ /// + /// # Doc comments + /// + /// The compiler transforms doc comments, such as `/// comment` and `/*! + /// comment */`, into attributes before macros are expanded. Each comment is + /// expanded into an attribute of the form `#[doc = r"comment"]`. + /// + /// As an example, the following `mod` items are expanded identically: + /// + /// ``` + /// # use syn::{ItemMod, parse_quote}; + /// let doc: ItemMod = parse_quote! { + /// /// Single line doc comments + /// /// We write so many! + /// /** + /// * Multi-line comments... + /// * May span many lines + /// */ + /// mod example { + /// //! Of course, they can be inner too + /// /*! And fit in a single line */ + /// } + /// }; + /// let attr: ItemMod = parse_quote! { + /// #[doc = r" Single line doc comments"] + /// #[doc = r" We write so many!"] + /// #[doc = r" + /// * Multi-line comments... + /// * May span many lines + /// "] + /// mod example { + /// #![doc = r" Of course, they can be inner too"] + /// #![doc = r" And fit in a single line "] + /// } + /// }; + /// assert_eq!(doc, attr); + /// ``` + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Attribute { + pub pound_token: Token![#], + pub style: AttrStyle, + pub bracket_token: token::Bracket, + pub path: Path, + pub tokens: TokenStream, + } +} + +impl Attribute { + /// Parses the content of the attribute, consisting of the path and tokens, + /// as a [`Meta`] if possible. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_meta(&self) -> Result { + fn clone_ident_segment(segment: &PathSegment) -> PathSegment { + PathSegment { + ident: segment.ident.clone(), + arguments: PathArguments::None, + } + } + + let path = Path { + leading_colon: self + .path + .leading_colon + .as_ref() + .map(|colon| Token![::](colon.spans)), + segments: self + .path + .segments + .pairs() + .map(|pair| match pair { + Pair::Punctuated(seg, punct) => { + Pair::Punctuated(clone_ident_segment(seg), Token![::](punct.spans)) + } + Pair::End(seg) => Pair::End(clone_ident_segment(seg)), + }) + .collect(), + }; + + let parser = |input: ParseStream| parsing::parse_meta_after_path(path, input); + parse::Parser::parse2(parser, self.tokens.clone()) + } + + /// Parse the arguments to the attribute as a syntax tree. + /// + /// This is similar to `syn::parse2::(attr.tokens)` except that: + /// + /// - the surrounding delimiters are *not* included in the input to the + /// parser; and + /// - the error message has a more useful span when `tokens` is empty. + /// + /// ```text + /// #[my_attr(value < 5)] + /// ^^^^^^^^^ what gets parsed + /// ``` + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_args(&self) -> Result { + self.parse_args_with(T::parse) + } + + /// Parse the arguments to the attribute using the given parser. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_args_with(&self, parser: F) -> Result { + let parser = |input: ParseStream| { + let args = enter_args(self, input)?; + parse::parse_stream(parser, &args) + }; + parser.parse2(self.tokens.clone()) + } + + /// Parses zero or more outer attributes from the stream. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_outer(input: ParseStream) -> Result> { + let mut attrs = Vec::new(); + while input.peek(Token![#]) { + attrs.push(input.call(parsing::single_parse_outer)?); + } + Ok(attrs) + } + + /// Parses zero or more inner attributes from the stream. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_inner(input: ParseStream) -> Result> { + let mut attrs = Vec::new(); + parsing::parse_inner(input, &mut attrs)?; + Ok(attrs) + } +} + +#[cfg(feature = "parsing")] +fn expected_parentheses(attr: &Attribute) -> String { + let style = match attr.style { + AttrStyle::Outer => "#", + AttrStyle::Inner(_) => "#!", + }; + + let mut path = String::new(); + for segment in &attr.path.segments { + if !path.is_empty() || attr.path.leading_colon.is_some() { + path += "::"; + } + path += &segment.ident.to_string(); + } + + format!("{}[{}(...)]", style, path) +} + +#[cfg(feature = "parsing")] +fn enter_args<'a>(attr: &Attribute, input: ParseStream<'a>) -> Result> { + if input.is_empty() { + let expected = expected_parentheses(attr); + let msg = format!("expected attribute arguments in parentheses: {}", expected); + return Err(crate::error::new2( + attr.pound_token.span, + attr.bracket_token.span, + msg, + )); + } else if input.peek(Token![=]) { + let expected = expected_parentheses(attr); + let msg = format!("expected parentheses: {}", expected); + return Err(input.error(msg)); + }; + + let content; + if input.peek(token::Paren) { + parenthesized!(content in input); + } else if input.peek(token::Bracket) { + bracketed!(content in input); + } else if input.peek(token::Brace) { + braced!(content in input); + } else { + return Err(input.error("unexpected token in attribute arguments")); + } + + if input.is_empty() { + Ok(content) + } else { + Err(input.error("unexpected token in attribute arguments")) + } +} + +ast_enum! { + /// Distinguishes between attributes that decorate an item and attributes + /// that are contained within an item. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// # Outer attributes + /// + /// - `#[repr(transparent)]` + /// - `/// # Example` + /// - `/** Please file an issue */` + /// + /// # Inner attributes + /// + /// - `#![feature(proc_macro)]` + /// - `//! # Example` + /// - `/*! Please file an issue */` + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum AttrStyle { + Outer, + Inner(Token![!]), + } +} + +ast_enum_of_structs! { + /// Content of a compile-time structured attribute. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// ## Path + /// + /// A meta path is like the `test` in `#[test]`. + /// + /// ## List + /// + /// A meta list is like the `derive(Copy)` in `#[derive(Copy)]`. + /// + /// ## NameValue + /// + /// A name-value meta is like the `path = "..."` in `#[path = + /// "sys/windows.rs"]`. + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum Meta { + Path(Path), + + /// A structured list within an attribute, like `derive(Copy, Clone)`. + List(MetaList), + + /// A name-value pair within an attribute, like `feature = "nightly"`. + NameValue(MetaNameValue), + } +} + +ast_struct! { + /// A structured list within an attribute, like `derive(Copy, Clone)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct MetaList { + pub path: Path, + pub paren_token: token::Paren, + pub nested: Punctuated, + } +} + +ast_struct! { + /// A name-value pair within an attribute, like `feature = "nightly"`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct MetaNameValue { + pub path: Path, + pub eq_token: Token![=], + pub lit: Lit, + } +} + +impl Meta { + /// Returns the identifier that begins this structured meta item. + /// + /// For example this would return the `test` in `#[test]`, the `derive` in + /// `#[derive(Copy)]`, and the `path` in `#[path = "sys/windows.rs"]`. + pub fn path(&self) -> &Path { + match self { + Meta::Path(path) => path, + Meta::List(meta) => &meta.path, + Meta::NameValue(meta) => &meta.path, + } + } +} + +ast_enum_of_structs! { + /// Element of a compile-time attribute list. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum NestedMeta { + /// A structured meta item, like the `Copy` in `#[derive(Copy)]` which + /// would be a nested `Meta::Path`. + Meta(Meta), + + /// A Rust literal, like the `"new_name"` in `#[rename("new_name")]`. + Lit(Lit), + } +} + +/// Conventional argument type associated with an invocation of an attribute +/// macro. +/// +/// For example if we are developing an attribute macro that is intended to be +/// invoked on function items as follows: +/// +/// ``` +/// # const IGNORE: &str = stringify! { +/// #[my_attribute(path = "/v1/refresh")] +/// # }; +/// pub fn refresh() { +/// /* ... */ +/// } +/// ``` +/// +/// The implementation of this macro would want to parse its attribute arguments +/// as type `AttributeArgs`. +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// use proc_macro::TokenStream; +/// use syn::{parse_macro_input, AttributeArgs, ItemFn}; +/// +/// # const IGNORE: &str = stringify! { +/// #[proc_macro_attribute] +/// # }; +/// pub fn my_attribute(args: TokenStream, input: TokenStream) -> TokenStream { +/// let args = parse_macro_input!(args as AttributeArgs); +/// let input = parse_macro_input!(input as ItemFn); +/// +/// /* ... */ +/// # "".parse().unwrap() +/// } +/// ``` +#[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] +pub type AttributeArgs = Vec; + +pub trait FilterAttrs<'a> { + type Ret: Iterator; + + fn outer(self) -> Self::Ret; + fn inner(self) -> Self::Ret; +} + +impl<'a> FilterAttrs<'a> for &'a [Attribute] { + type Ret = iter::Filter, fn(&&Attribute) -> bool>; + + fn outer(self) -> Self::Ret { + fn is_outer(attr: &&Attribute) -> bool { + match attr.style { + AttrStyle::Outer => true, + AttrStyle::Inner(_) => false, + } + } + self.iter().filter(is_outer) + } + + fn inner(self) -> Self::Ret { + fn is_inner(attr: &&Attribute) -> bool { + match attr.style { + AttrStyle::Inner(_) => true, + AttrStyle::Outer => false, + } + } + self.iter().filter(is_inner) + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::{Parse, ParseStream, Result}; + + pub fn parse_inner(input: ParseStream, attrs: &mut Vec) -> Result<()> { + while input.peek(Token![#]) && input.peek2(Token![!]) { + attrs.push(input.call(parsing::single_parse_inner)?); + } + Ok(()) + } + + pub fn single_parse_inner(input: ParseStream) -> Result { + let content; + Ok(Attribute { + pound_token: input.parse()?, + style: AttrStyle::Inner(input.parse()?), + bracket_token: bracketed!(content in input), + path: content.call(Path::parse_mod_style)?, + tokens: content.parse()?, + }) + } + + pub fn single_parse_outer(input: ParseStream) -> Result { + let content; + Ok(Attribute { + pound_token: input.parse()?, + style: AttrStyle::Outer, + bracket_token: bracketed!(content in input), + path: content.call(Path::parse_mod_style)?, + tokens: content.parse()?, + }) + } + + // Like Path::parse_mod_style but accepts keywords in the path. + fn parse_meta_path(input: ParseStream) -> Result { + Ok(Path { + leading_colon: input.parse()?, + segments: { + let mut segments = Punctuated::new(); + while input.peek(Ident::peek_any) { + let ident = Ident::parse_any(input)?; + segments.push_value(PathSegment::from(ident)); + if !input.peek(Token![::]) { + break; + } + let punct = input.parse()?; + segments.push_punct(punct); + } + if segments.is_empty() { + return Err(input.error("expected path")); + } else if segments.trailing_punct() { + return Err(input.error("expected path segment")); + } + segments + }, + }) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Meta { + fn parse(input: ParseStream) -> Result { + let path = input.call(parse_meta_path)?; + parse_meta_after_path(path, input) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for MetaList { + fn parse(input: ParseStream) -> Result { + let path = input.call(parse_meta_path)?; + parse_meta_list_after_path(path, input) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for MetaNameValue { + fn parse(input: ParseStream) -> Result { + let path = input.call(parse_meta_path)?; + parse_meta_name_value_after_path(path, input) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for NestedMeta { + fn parse(input: ParseStream) -> Result { + if input.peek(Lit) && !(input.peek(LitBool) && input.peek2(Token![=])) { + input.parse().map(NestedMeta::Lit) + } else if input.peek(Ident::peek_any) + || input.peek(Token![::]) && input.peek3(Ident::peek_any) + { + input.parse().map(NestedMeta::Meta) + } else { + Err(input.error("expected identifier or literal")) + } + } + } + + pub fn parse_meta_after_path(path: Path, input: ParseStream) -> Result { + if input.peek(token::Paren) { + parse_meta_list_after_path(path, input).map(Meta::List) + } else if input.peek(Token![=]) { + parse_meta_name_value_after_path(path, input).map(Meta::NameValue) + } else { + Ok(Meta::Path(path)) + } + } + + fn parse_meta_list_after_path(path: Path, input: ParseStream) -> Result { + let content; + Ok(MetaList { + path, + paren_token: parenthesized!(content in input), + nested: content.parse_terminated(NestedMeta::parse)?, + }) + } + + fn parse_meta_name_value_after_path(path: Path, input: ParseStream) -> Result { + Ok(MetaNameValue { + path, + eq_token: input.parse()?, + lit: input.parse()?, + }) + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::ToTokens; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Attribute { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.pound_token.to_tokens(tokens); + if let AttrStyle::Inner(b) = &self.style { + b.to_tokens(tokens); + } + self.bracket_token.surround(tokens, |tokens| { + self.path.to_tokens(tokens); + self.tokens.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for MetaList { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.path.to_tokens(tokens); + self.paren_token.surround(tokens, |tokens| { + self.nested.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for MetaNameValue { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.path.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.lit.to_tokens(tokens); + } + } +} diff --git a/rust/syn/await.rs b/rust/syn/await.rs new file mode 100644 index 00000000000000..1aae244cafd3e1 --- /dev/null +++ b/rust/syn/await.rs @@ -0,0 +1,4 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// See include!("await.rs") in token.rs. +export_token_macro! {[await]} diff --git a/rust/syn/bigint.rs b/rust/syn/bigint.rs new file mode 100644 index 00000000000000..d8f7b8ca68630e --- /dev/null +++ b/rust/syn/bigint.rs @@ -0,0 +1,68 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use std::ops::{AddAssign, MulAssign}; + +// For implementing base10_digits() accessor on LitInt. +pub struct BigInt { + digits: Vec, +} + +impl BigInt { + pub fn new() -> Self { + BigInt { digits: Vec::new() } + } + + pub fn to_string(&self) -> String { + let mut repr = String::with_capacity(self.digits.len()); + + let mut has_nonzero = false; + for digit in self.digits.iter().rev() { + has_nonzero |= *digit != 0; + if has_nonzero { + repr.push((*digit + b'0') as char); + } + } + + if repr.is_empty() { + repr.push('0'); + } + + repr + } + + fn reserve_two_digits(&mut self) { + let len = self.digits.len(); + let desired = + len + !self.digits.ends_with(&[0, 0]) as usize + !self.digits.ends_with(&[0]) as usize; + self.digits.resize(desired, 0); + } +} + +impl AddAssign for BigInt { + // Assumes increment <16. + fn add_assign(&mut self, mut increment: u8) { + self.reserve_two_digits(); + + let mut i = 0; + while increment > 0 { + let sum = self.digits[i] + increment; + self.digits[i] = sum % 10; + increment = sum / 10; + i += 1; + } + } +} + +impl MulAssign for BigInt { + // Assumes base <=16. + fn mul_assign(&mut self, base: u8) { + self.reserve_two_digits(); + + let mut carry = 0; + for digit in &mut self.digits { + let prod = *digit * base + carry; + *digit = prod % 10; + carry = prod / 10; + } + } +} diff --git a/rust/syn/buffer.rs b/rust/syn/buffer.rs new file mode 100644 index 00000000000000..9a600f527f7a50 --- /dev/null +++ b/rust/syn/buffer.rs @@ -0,0 +1,400 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! A stably addressed token buffer supporting efficient traversal based on a +//! cheaply copyable cursor. +//! +//! *This module is available only if Syn is built with the `"parsing"` feature.* + +// This module is heavily commented as it contains most of the unsafe code in +// Syn, and caution should be used when editing it. The public-facing interface +// is 100% safe but the implementation is fragile internally. + +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" +))] +use crate::proc_macro as pm; +use crate::Lifetime; +use proc_macro2::{Delimiter, Group, Ident, Literal, Punct, Spacing, Span, TokenStream, TokenTree}; +use std::cmp::Ordering; +use std::marker::PhantomData; + +/// Internal type which is used instead of `TokenTree` to represent a token tree +/// within a `TokenBuffer`. +enum Entry { + // Mimicking types from proc-macro. + // Group entries contain the offset to the matching End entry. + Group(Group, usize), + Ident(Ident), + Punct(Punct), + Literal(Literal), + // End entries contain the offset (negative) to the start of the buffer. + End(isize), +} + +/// A buffer that can be efficiently traversed multiple times, unlike +/// `TokenStream` which requires a deep copy in order to traverse more than +/// once. +/// +/// *This type is available only if Syn is built with the `"parsing"` feature.* +pub struct TokenBuffer { + // NOTE: Do not implement clone on this - while the current design could be + // cloned, other designs which could be desirable may not be cloneable. + entries: Box<[Entry]>, +} + +impl TokenBuffer { + fn recursive_new(entries: &mut Vec, stream: TokenStream) { + for tt in stream { + match tt { + TokenTree::Ident(ident) => entries.push(Entry::Ident(ident)), + TokenTree::Punct(punct) => entries.push(Entry::Punct(punct)), + TokenTree::Literal(literal) => entries.push(Entry::Literal(literal)), + TokenTree::Group(group) => { + let group_start_index = entries.len(); + entries.push(Entry::End(0)); // we replace this below + Self::recursive_new(entries, group.stream()); + let group_end_index = entries.len(); + entries.push(Entry::End(-(group_end_index as isize))); + let group_end_offset = group_end_index - group_start_index; + entries[group_start_index] = Entry::Group(group, group_end_offset); + } + } + } + } + + /// Creates a `TokenBuffer` containing all the tokens from the input + /// `proc_macro::TokenStream`. + /// + /// *This method is available only if Syn is built with both the `"parsing"` and + /// `"proc-macro"` features.* + #[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" + ))] + pub fn new(stream: pm::TokenStream) -> Self { + Self::new2(stream.into()) + } + + /// Creates a `TokenBuffer` containing all the tokens from the input + /// `proc_macro2::TokenStream`. + pub fn new2(stream: TokenStream) -> Self { + let mut entries = Vec::new(); + Self::recursive_new(&mut entries, stream); + entries.push(Entry::End(-(entries.len() as isize))); + Self { + entries: entries.into_boxed_slice(), + } + } + + /// Creates a cursor referencing the first token in the buffer and able to + /// traverse until the end of the buffer. + pub fn begin(&self) -> Cursor { + let ptr = self.entries.as_ptr(); + unsafe { Cursor::create(ptr, ptr.add(self.entries.len() - 1)) } + } +} + +/// A cheaply copyable cursor into a `TokenBuffer`. +/// +/// This cursor holds a shared reference into the immutable data which is used +/// internally to represent a `TokenStream`, and can be efficiently manipulated +/// and copied around. +/// +/// An empty `Cursor` can be created directly, or one may create a `TokenBuffer` +/// object and get a cursor to its first token with `begin()`. +/// +/// Two cursors are equal if they have the same location in the same input +/// stream, and have the same scope. +/// +/// *This type is available only if Syn is built with the `"parsing"` feature.* +pub struct Cursor<'a> { + // The current entry which the `Cursor` is pointing at. + ptr: *const Entry, + // This is the only `Entry::End` object which this cursor is allowed to + // point at. All other `End` objects are skipped over in `Cursor::create`. + scope: *const Entry, + // Cursor is covariant in 'a. This field ensures that our pointers are still + // valid. + marker: PhantomData<&'a Entry>, +} + +impl<'a> Cursor<'a> { + /// Creates a cursor referencing a static empty TokenStream. + pub fn empty() -> Self { + // It's safe in this situation for us to put an `Entry` object in global + // storage, despite it not actually being safe to send across threads + // (`Ident` is a reference into a thread-local table). This is because + // this entry never includes a `Ident` object. + // + // This wrapper struct allows us to break the rules and put a `Sync` + // object in global storage. + struct UnsafeSyncEntry(Entry); + unsafe impl Sync for UnsafeSyncEntry {} + static EMPTY_ENTRY: UnsafeSyncEntry = UnsafeSyncEntry(Entry::End(0)); + + Cursor { + ptr: &EMPTY_ENTRY.0, + scope: &EMPTY_ENTRY.0, + marker: PhantomData, + } + } + + /// This create method intelligently exits non-explicitly-entered + /// `None`-delimited scopes when the cursor reaches the end of them, + /// allowing for them to be treated transparently. + unsafe fn create(mut ptr: *const Entry, scope: *const Entry) -> Self { + // NOTE: If we're looking at a `End`, we want to advance the cursor + // past it, unless `ptr == scope`, which means that we're at the edge of + // our cursor's scope. We should only have `ptr != scope` at the exit + // from None-delimited groups entered with `ignore_none`. + while let Entry::End(_) = *ptr { + if ptr == scope { + break; + } + ptr = ptr.add(1); + } + + Cursor { + ptr, + scope, + marker: PhantomData, + } + } + + /// Get the current entry. + fn entry(self) -> &'a Entry { + unsafe { &*self.ptr } + } + + /// Bump the cursor to point at the next token after the current one. This + /// is undefined behavior if the cursor is currently looking at an + /// `Entry::End`. + /// + /// If the cursor is looking at an `Entry::Group`, the bumped cursor will + /// point at the first token in the group (with the same scope end). + unsafe fn bump_ignore_group(self) -> Cursor<'a> { + Cursor::create(self.ptr.offset(1), self.scope) + } + + /// While the cursor is looking at a `None`-delimited group, move it to look + /// at the first token inside instead. If the group is empty, this will move + /// the cursor past the `None`-delimited group. + /// + /// WARNING: This mutates its argument. + fn ignore_none(&mut self) { + while let Entry::Group(group, _) = self.entry() { + if group.delimiter() == Delimiter::None { + unsafe { *self = self.bump_ignore_group() }; + } else { + break; + } + } + } + + /// Checks whether the cursor is currently pointing at the end of its valid + /// scope. + pub fn eof(self) -> bool { + // We're at eof if we're at the end of our scope. + self.ptr == self.scope + } + + /// If the cursor is pointing at a `Group` with the given delimiter, returns + /// a cursor into that group and one pointing to the next `TokenTree`. + pub fn group(mut self, delim: Delimiter) -> Option<(Cursor<'a>, Span, Cursor<'a>)> { + // If we're not trying to enter a none-delimited group, we want to + // ignore them. We have to make sure to _not_ ignore them when we want + // to enter them, of course. For obvious reasons. + if delim != Delimiter::None { + self.ignore_none(); + } + + if let Entry::Group(group, end_offset) = self.entry() { + if group.delimiter() == delim { + let end_of_group = unsafe { self.ptr.add(*end_offset) }; + let inside_of_group = unsafe { Cursor::create(self.ptr.add(1), end_of_group) }; + let after_group = unsafe { Cursor::create(end_of_group, self.scope) }; + return Some((inside_of_group, group.span(), after_group)); + } + } + + None + } + + /// If the cursor is pointing at a `Ident`, returns it along with a cursor + /// pointing at the next `TokenTree`. + pub fn ident(mut self) -> Option<(Ident, Cursor<'a>)> { + self.ignore_none(); + match self.entry() { + Entry::Ident(ident) => Some((ident.clone(), unsafe { self.bump_ignore_group() })), + _ => None, + } + } + + /// If the cursor is pointing at a `Punct`, returns it along with a cursor + /// pointing at the next `TokenTree`. + pub fn punct(mut self) -> Option<(Punct, Cursor<'a>)> { + self.ignore_none(); + match self.entry() { + Entry::Punct(punct) if punct.as_char() != '\'' => { + Some((punct.clone(), unsafe { self.bump_ignore_group() })) + } + _ => None, + } + } + + /// If the cursor is pointing at a `Literal`, return it along with a cursor + /// pointing at the next `TokenTree`. + pub fn literal(mut self) -> Option<(Literal, Cursor<'a>)> { + self.ignore_none(); + match self.entry() { + Entry::Literal(literal) => Some((literal.clone(), unsafe { self.bump_ignore_group() })), + _ => None, + } + } + + /// If the cursor is pointing at a `Lifetime`, returns it along with a + /// cursor pointing at the next `TokenTree`. + pub fn lifetime(mut self) -> Option<(Lifetime, Cursor<'a>)> { + self.ignore_none(); + match self.entry() { + Entry::Punct(punct) if punct.as_char() == '\'' && punct.spacing() == Spacing::Joint => { + let next = unsafe { self.bump_ignore_group() }; + let (ident, rest) = next.ident()?; + let lifetime = Lifetime { + apostrophe: punct.span(), + ident, + }; + Some((lifetime, rest)) + } + _ => None, + } + } + + /// Copies all remaining tokens visible from this cursor into a + /// `TokenStream`. + pub fn token_stream(self) -> TokenStream { + let mut tts = Vec::new(); + let mut cursor = self; + while let Some((tt, rest)) = cursor.token_tree() { + tts.push(tt); + cursor = rest; + } + tts.into_iter().collect() + } + + /// If the cursor is pointing at a `TokenTree`, returns it along with a + /// cursor pointing at the next `TokenTree`. + /// + /// Returns `None` if the cursor has reached the end of its stream. + /// + /// This method does not treat `None`-delimited groups as transparent, and + /// will return a `Group(None, ..)` if the cursor is looking at one. + pub fn token_tree(self) -> Option<(TokenTree, Cursor<'a>)> { + let (tree, len) = match self.entry() { + Entry::Group(group, end_offset) => (group.clone().into(), *end_offset), + Entry::Literal(literal) => (literal.clone().into(), 1), + Entry::Ident(ident) => (ident.clone().into(), 1), + Entry::Punct(punct) => (punct.clone().into(), 1), + Entry::End(_) => return None, + }; + + let rest = unsafe { Cursor::create(self.ptr.add(len), self.scope) }; + Some((tree, rest)) + } + + /// Returns the `Span` of the current token, or `Span::call_site()` if this + /// cursor points to eof. + pub fn span(self) -> Span { + match self.entry() { + Entry::Group(group, _) => group.span(), + Entry::Literal(literal) => literal.span(), + Entry::Ident(ident) => ident.span(), + Entry::Punct(punct) => punct.span(), + Entry::End(_) => Span::call_site(), + } + } + + /// Skip over the next token without cloning it. Returns `None` if this + /// cursor points to eof. + /// + /// This method treats `'lifetimes` as a single token. + pub(crate) fn skip(self) -> Option> { + let len = match self.entry() { + Entry::End(_) => return None, + + // Treat lifetimes as a single tt for the purposes of 'skip'. + Entry::Punct(punct) if punct.as_char() == '\'' && punct.spacing() == Spacing::Joint => { + match unsafe { &*self.ptr.add(1) } { + Entry::Ident(_) => 2, + _ => 1, + } + } + + Entry::Group(_, end_offset) => *end_offset, + _ => 1, + }; + + Some(unsafe { Cursor::create(self.ptr.add(len), self.scope) }) + } +} + +impl<'a> Copy for Cursor<'a> {} + +impl<'a> Clone for Cursor<'a> { + fn clone(&self) -> Self { + *self + } +} + +impl<'a> Eq for Cursor<'a> {} + +impl<'a> PartialEq for Cursor<'a> { + fn eq(&self, other: &Self) -> bool { + self.ptr == other.ptr + } +} + +impl<'a> PartialOrd for Cursor<'a> { + fn partial_cmp(&self, other: &Self) -> Option { + if same_buffer(*self, *other) { + Some(self.ptr.cmp(&other.ptr)) + } else { + None + } + } +} + +pub(crate) fn same_scope(a: Cursor, b: Cursor) -> bool { + a.scope == b.scope +} + +pub(crate) fn same_buffer(a: Cursor, b: Cursor) -> bool { + unsafe { + match (&*a.scope, &*b.scope) { + (Entry::End(a_offset), Entry::End(b_offset)) => { + a.scope.offset(*a_offset) == b.scope.offset(*b_offset) + } + _ => unreachable!(), + } + } +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn cmp_assuming_same_buffer(a: Cursor, b: Cursor) -> Ordering { + a.ptr.cmp(&b.ptr) +} + +pub(crate) fn open_span_of_group(cursor: Cursor) -> Span { + match cursor.entry() { + Entry::Group(group, _) => group.span_open(), + _ => cursor.span(), + } +} + +pub(crate) fn close_span_of_group(cursor: Cursor) -> Span { + match cursor.entry() { + Entry::Group(group, _) => group.span_close(), + _ => cursor.span(), + } +} diff --git a/rust/syn/custom_keyword.rs b/rust/syn/custom_keyword.rs new file mode 100644 index 00000000000000..916f4c1bb8205e --- /dev/null +++ b/rust/syn/custom_keyword.rs @@ -0,0 +1,255 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Define a type that supports parsing and printing a given identifier as if it +/// were a keyword. +/// +/// # Usage +/// +/// As a convention, it is recommended that this macro be invoked within a +/// module called `kw` or `keyword` and that the resulting parser be invoked +/// with a `kw::` or `keyword::` prefix. +/// +/// ``` +/// mod kw { +/// syn::custom_keyword!(whatever); +/// } +/// ``` +/// +/// The generated syntax tree node supports the following operations just like +/// any built-in keyword token. +/// +/// - [Peeking] — `input.peek(kw::whatever)` +/// +/// - [Parsing] — `input.parse::()?` +/// +/// - [Printing] — `quote!( ... #whatever_token ... )` +/// +/// - Construction from a [`Span`] — `let whatever_token = kw::whatever(sp)` +/// +/// - Field access to its span — `let sp = whatever_token.span` +/// +/// [Peeking]: crate::parse::ParseBuffer::peek +/// [Parsing]: crate::parse::ParseBuffer::parse +/// [Printing]: quote::ToTokens +/// [`Span`]: proc_macro2::Span +/// +/// # Example +/// +/// This example parses input that looks like `bool = true` or `str = "value"`. +/// The key must be either the identifier `bool` or the identifier `str`. If +/// `bool`, the value may be either `true` or `false`. If `str`, the value may +/// be any string literal. +/// +/// The symbols `bool` and `str` are not reserved keywords in Rust so these are +/// not considered keywords in the `syn::token` module. Like any other +/// identifier that is not a keyword, these can be declared as custom keywords +/// by crates that need to use them as such. +/// +/// ``` +/// use syn::{LitBool, LitStr, Result, Token}; +/// use syn::parse::{Parse, ParseStream}; +/// +/// mod kw { +/// syn::custom_keyword!(bool); +/// syn::custom_keyword!(str); +/// } +/// +/// enum Argument { +/// Bool { +/// bool_token: kw::bool, +/// eq_token: Token![=], +/// value: LitBool, +/// }, +/// Str { +/// str_token: kw::str, +/// eq_token: Token![=], +/// value: LitStr, +/// }, +/// } +/// +/// impl Parse for Argument { +/// fn parse(input: ParseStream) -> Result { +/// let lookahead = input.lookahead1(); +/// if lookahead.peek(kw::bool) { +/// Ok(Argument::Bool { +/// bool_token: input.parse::()?, +/// eq_token: input.parse()?, +/// value: input.parse()?, +/// }) +/// } else if lookahead.peek(kw::str) { +/// Ok(Argument::Str { +/// str_token: input.parse::()?, +/// eq_token: input.parse()?, +/// value: input.parse()?, +/// }) +/// } else { +/// Err(lookahead.error()) +/// } +/// } +/// } +/// ``` +#[macro_export] +macro_rules! custom_keyword { + ($ident:ident) => { + #[allow(non_camel_case_types)] + pub struct $ident { + pub span: $crate::__private::Span, + } + + #[doc(hidden)] + #[allow(dead_code, non_snake_case)] + pub fn $ident<__S: $crate::__private::IntoSpans<[$crate::__private::Span; 1]>>( + span: __S, + ) -> $ident { + $ident { + span: $crate::__private::IntoSpans::into_spans(span)[0], + } + } + + impl $crate::__private::Default for $ident { + fn default() -> Self { + $ident { + span: $crate::__private::Span::call_site(), + } + } + } + + $crate::impl_parse_for_custom_keyword!($ident); + $crate::impl_to_tokens_for_custom_keyword!($ident); + $crate::impl_clone_for_custom_keyword!($ident); + $crate::impl_extra_traits_for_custom_keyword!($ident); + }; +} + +// Not public API. +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_parse_for_custom_keyword { + ($ident:ident) => { + // For peek. + impl $crate::token::CustomToken for $ident { + fn peek(cursor: $crate::buffer::Cursor) -> $crate::__private::bool { + if let $crate::__private::Some((ident, _rest)) = cursor.ident() { + ident == stringify!($ident) + } else { + false + } + } + + fn display() -> &'static $crate::__private::str { + concat!("`", stringify!($ident), "`") + } + } + + impl $crate::parse::Parse for $ident { + fn parse(input: $crate::parse::ParseStream) -> $crate::parse::Result<$ident> { + input.step(|cursor| { + if let $crate::__private::Some((ident, rest)) = cursor.ident() { + if ident == stringify!($ident) { + return $crate::__private::Ok(($ident { span: ident.span() }, rest)); + } + } + $crate::__private::Err(cursor.error(concat!( + "expected `", + stringify!($ident), + "`" + ))) + }) + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "parsing"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_parse_for_custom_keyword { + ($ident:ident) => {}; +} + +// Not public API. +#[cfg(feature = "printing")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_to_tokens_for_custom_keyword { + ($ident:ident) => { + impl $crate::__private::ToTokens for $ident { + fn to_tokens(&self, tokens: &mut $crate::__private::TokenStream2) { + let ident = $crate::Ident::new(stringify!($ident), self.span); + $crate::__private::TokenStreamExt::append(tokens, ident); + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "printing"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_to_tokens_for_custom_keyword { + ($ident:ident) => {}; +} + +// Not public API. +#[cfg(feature = "clone-impls")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_clone_for_custom_keyword { + ($ident:ident) => { + impl $crate::__private::Copy for $ident {} + + #[allow(clippy::expl_impl_clone_on_copy)] + impl $crate::__private::Clone for $ident { + fn clone(&self) -> Self { + *self + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "clone-impls"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_clone_for_custom_keyword { + ($ident:ident) => {}; +} + +// Not public API. +#[cfg(feature = "extra-traits")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_extra_traits_for_custom_keyword { + ($ident:ident) => { + impl $crate::__private::Debug for $ident { + fn fmt(&self, f: &mut $crate::__private::Formatter) -> $crate::__private::fmt::Result { + $crate::__private::Formatter::write_str( + f, + concat!("Keyword [", stringify!($ident), "]"), + ) + } + } + + impl $crate::__private::Eq for $ident {} + + impl $crate::__private::PartialEq for $ident { + fn eq(&self, _other: &Self) -> $crate::__private::bool { + true + } + } + + impl $crate::__private::Hash for $ident { + fn hash<__H: $crate::__private::Hasher>(&self, _state: &mut __H) {} + } + }; +} + +// Not public API. +#[cfg(not(feature = "extra-traits"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_extra_traits_for_custom_keyword { + ($ident:ident) => {}; +} diff --git a/rust/syn/custom_punctuation.rs b/rust/syn/custom_punctuation.rs new file mode 100644 index 00000000000000..5cb8281623f89f --- /dev/null +++ b/rust/syn/custom_punctuation.rs @@ -0,0 +1,302 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Define a type that supports parsing and printing a multi-character symbol +/// as if it were a punctuation token. +/// +/// # Usage +/// +/// ``` +/// syn::custom_punctuation!(LeftRightArrow, <=>); +/// ``` +/// +/// The generated syntax tree node supports the following operations just like +/// any built-in punctuation token. +/// +/// - [Peeking] — `input.peek(LeftRightArrow)` +/// +/// - [Parsing] — `input.parse::()?` +/// +/// - [Printing] — `quote!( ... #lrarrow ... )` +/// +/// - Construction from a [`Span`] — `let lrarrow = LeftRightArrow(sp)` +/// +/// - Construction from multiple [`Span`] — `let lrarrow = LeftRightArrow([sp, sp, sp])` +/// +/// - Field access to its spans — `let spans = lrarrow.spans` +/// +/// [Peeking]: crate::parse::ParseBuffer::peek +/// [Parsing]: crate::parse::ParseBuffer::parse +/// [Printing]: quote::ToTokens +/// [`Span`]: proc_macro2::Span +/// +/// # Example +/// +/// ``` +/// use proc_macro2::{TokenStream, TokenTree}; +/// use syn::parse::{Parse, ParseStream, Peek, Result}; +/// use syn::punctuated::Punctuated; +/// use syn::Expr; +/// +/// syn::custom_punctuation!(PathSeparator, ); +/// +/// // expr expr expr ... +/// struct PathSegments { +/// segments: Punctuated, +/// } +/// +/// impl Parse for PathSegments { +/// fn parse(input: ParseStream) -> Result { +/// let mut segments = Punctuated::new(); +/// +/// let first = parse_until(input, PathSeparator)?; +/// segments.push_value(syn::parse2(first)?); +/// +/// while input.peek(PathSeparator) { +/// segments.push_punct(input.parse()?); +/// +/// let next = parse_until(input, PathSeparator)?; +/// segments.push_value(syn::parse2(next)?); +/// } +/// +/// Ok(PathSegments { segments }) +/// } +/// } +/// +/// fn parse_until(input: ParseStream, end: E) -> Result { +/// let mut tokens = TokenStream::new(); +/// while !input.is_empty() && !input.peek(end) { +/// let next: TokenTree = input.parse()?; +/// tokens.extend(Some(next)); +/// } +/// Ok(tokens) +/// } +/// +/// fn main() { +/// let input = r#" a::b c::d::e "#; +/// let _: PathSegments = syn::parse_str(input).unwrap(); +/// } +/// ``` +#[macro_export] +macro_rules! custom_punctuation { + ($ident:ident, $($tt:tt)+) => { + pub struct $ident { + pub spans: $crate::custom_punctuation_repr!($($tt)+), + } + + #[doc(hidden)] + #[allow(dead_code, non_snake_case)] + pub fn $ident<__S: $crate::__private::IntoSpans<$crate::custom_punctuation_repr!($($tt)+)>>( + spans: __S, + ) -> $ident { + let _validate_len = 0 $(+ $crate::custom_punctuation_len!(strict, $tt))*; + $ident { + spans: $crate::__private::IntoSpans::into_spans(spans) + } + } + + impl $crate::__private::Default for $ident { + fn default() -> Self { + $ident($crate::__private::Span::call_site()) + } + } + + $crate::impl_parse_for_custom_punctuation!($ident, $($tt)+); + $crate::impl_to_tokens_for_custom_punctuation!($ident, $($tt)+); + $crate::impl_clone_for_custom_punctuation!($ident, $($tt)+); + $crate::impl_extra_traits_for_custom_punctuation!($ident, $($tt)+); + }; +} + +// Not public API. +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_parse_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => { + impl $crate::token::CustomToken for $ident { + fn peek(cursor: $crate::buffer::Cursor) -> bool { + $crate::token::parsing::peek_punct(cursor, $crate::stringify_punct!($($tt)+)) + } + + fn display() -> &'static $crate::__private::str { + concat!("`", $crate::stringify_punct!($($tt)+), "`") + } + } + + impl $crate::parse::Parse for $ident { + fn parse(input: $crate::parse::ParseStream) -> $crate::parse::Result<$ident> { + let spans: $crate::custom_punctuation_repr!($($tt)+) = + $crate::token::parsing::punct(input, $crate::stringify_punct!($($tt)+))?; + Ok($ident(spans)) + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "parsing"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_parse_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => {}; +} + +// Not public API. +#[cfg(feature = "printing")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_to_tokens_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => { + impl $crate::__private::ToTokens for $ident { + fn to_tokens(&self, tokens: &mut $crate::__private::TokenStream2) { + $crate::token::printing::punct($crate::stringify_punct!($($tt)+), &self.spans, tokens) + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "printing"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_to_tokens_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => {}; +} + +// Not public API. +#[cfg(feature = "clone-impls")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_clone_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => { + impl $crate::__private::Copy for $ident {} + + #[allow(clippy::expl_impl_clone_on_copy)] + impl $crate::__private::Clone for $ident { + fn clone(&self) -> Self { + *self + } + } + }; +} + +// Not public API. +#[cfg(not(feature = "clone-impls"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_clone_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => {}; +} + +// Not public API. +#[cfg(feature = "extra-traits")] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_extra_traits_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => { + impl $crate::__private::Debug for $ident { + fn fmt(&self, f: &mut $crate::__private::Formatter) -> $crate::__private::fmt::Result { + $crate::__private::Formatter::write_str(f, stringify!($ident)) + } + } + + impl $crate::__private::Eq for $ident {} + + impl $crate::__private::PartialEq for $ident { + fn eq(&self, _other: &Self) -> $crate::__private::bool { + true + } + } + + impl $crate::__private::Hash for $ident { + fn hash<__H: $crate::__private::Hasher>(&self, _state: &mut __H) {} + } + }; +} + +// Not public API. +#[cfg(not(feature = "extra-traits"))] +#[doc(hidden)] +#[macro_export] +macro_rules! impl_extra_traits_for_custom_punctuation { + ($ident:ident, $($tt:tt)+) => {}; +} + +// Not public API. +#[doc(hidden)] +#[macro_export] +macro_rules! custom_punctuation_repr { + ($($tt:tt)+) => { + [$crate::__private::Span; 0 $(+ $crate::custom_punctuation_len!(lenient, $tt))+] + }; +} + +// Not public API. +#[doc(hidden)] +#[macro_export] +#[rustfmt::skip] +macro_rules! custom_punctuation_len { + ($mode:ident, +) => { 1 }; + ($mode:ident, +=) => { 2 }; + ($mode:ident, &) => { 1 }; + ($mode:ident, &&) => { 2 }; + ($mode:ident, &=) => { 2 }; + ($mode:ident, @) => { 1 }; + ($mode:ident, !) => { 1 }; + ($mode:ident, ^) => { 1 }; + ($mode:ident, ^=) => { 2 }; + ($mode:ident, :) => { 1 }; + ($mode:ident, ::) => { 2 }; + ($mode:ident, ,) => { 1 }; + ($mode:ident, /) => { 1 }; + ($mode:ident, /=) => { 2 }; + ($mode:ident, .) => { 1 }; + ($mode:ident, ..) => { 2 }; + ($mode:ident, ...) => { 3 }; + ($mode:ident, ..=) => { 3 }; + ($mode:ident, =) => { 1 }; + ($mode:ident, ==) => { 2 }; + ($mode:ident, >=) => { 2 }; + ($mode:ident, >) => { 1 }; + ($mode:ident, <=) => { 2 }; + ($mode:ident, <) => { 1 }; + ($mode:ident, *=) => { 2 }; + ($mode:ident, !=) => { 2 }; + ($mode:ident, |) => { 1 }; + ($mode:ident, |=) => { 2 }; + ($mode:ident, ||) => { 2 }; + ($mode:ident, #) => { 1 }; + ($mode:ident, ?) => { 1 }; + ($mode:ident, ->) => { 2 }; + ($mode:ident, <-) => { 2 }; + ($mode:ident, %) => { 1 }; + ($mode:ident, %=) => { 2 }; + ($mode:ident, =>) => { 2 }; + ($mode:ident, ;) => { 1 }; + ($mode:ident, <<) => { 2 }; + ($mode:ident, <<=) => { 3 }; + ($mode:ident, >>) => { 2 }; + ($mode:ident, >>=) => { 3 }; + ($mode:ident, *) => { 1 }; + ($mode:ident, -) => { 1 }; + ($mode:ident, -=) => { 2 }; + ($mode:ident, ~) => { 1 }; + (lenient, $tt:tt) => { 0 }; + (strict, $tt:tt) => {{ $crate::custom_punctuation_unexpected!($tt); 0 }}; +} + +// Not public API. +#[doc(hidden)] +#[macro_export] +macro_rules! custom_punctuation_unexpected { + () => {}; +} + +// Not public API. +#[doc(hidden)] +#[macro_export] +macro_rules! stringify_punct { + ($($tt:tt)+) => { + concat!($(stringify!($tt)),+) + }; +} diff --git a/rust/syn/data.rs b/rust/syn/data.rs new file mode 100644 index 00000000000000..e4002648c63514 --- /dev/null +++ b/rust/syn/data.rs @@ -0,0 +1,495 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; + +ast_struct! { + /// An enum variant. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Variant { + /// Attributes tagged on the variant. + pub attrs: Vec, + + /// Name of the variant. + pub ident: Ident, + + /// Content stored in the variant. + pub fields: Fields, + + /// Explicit discriminant: `Variant = 1` + pub discriminant: Option<(Token![=], Expr)>, + } +} + +ast_enum_of_structs! { + /// Data stored within an enum variant or struct. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum Fields { + /// Named fields of a struct or struct variant such as `Point { x: f64, + /// y: f64 }`. + Named(FieldsNamed), + + /// Unnamed fields of a tuple struct or tuple variant such as `Some(T)`. + Unnamed(FieldsUnnamed), + + /// Unit struct or unit variant such as `None`. + Unit, + } +} + +ast_struct! { + /// Named fields of a struct or struct variant such as `Point { x: f64, + /// y: f64 }`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct FieldsNamed { + pub brace_token: token::Brace, + pub named: Punctuated, + } +} + +ast_struct! { + /// Unnamed fields of a tuple struct or tuple variant such as `Some(T)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct FieldsUnnamed { + pub paren_token: token::Paren, + pub unnamed: Punctuated, + } +} + +impl Fields { + /// Get an iterator over the borrowed [`Field`] items in this object. This + /// iterator can be used to iterate over a named or unnamed struct or + /// variant's fields uniformly. + pub fn iter(&self) -> punctuated::Iter { + match self { + Fields::Unit => crate::punctuated::empty_punctuated_iter(), + Fields::Named(f) => f.named.iter(), + Fields::Unnamed(f) => f.unnamed.iter(), + } + } + + /// Get an iterator over the mutably borrowed [`Field`] items in this + /// object. This iterator can be used to iterate over a named or unnamed + /// struct or variant's fields uniformly. + pub fn iter_mut(&mut self) -> punctuated::IterMut { + match self { + Fields::Unit => crate::punctuated::empty_punctuated_iter_mut(), + Fields::Named(f) => f.named.iter_mut(), + Fields::Unnamed(f) => f.unnamed.iter_mut(), + } + } + + /// Returns the number of fields. + pub fn len(&self) -> usize { + match self { + Fields::Unit => 0, + Fields::Named(f) => f.named.len(), + Fields::Unnamed(f) => f.unnamed.len(), + } + } + + /// Returns `true` if there are zero fields. + pub fn is_empty(&self) -> bool { + match self { + Fields::Unit => true, + Fields::Named(f) => f.named.is_empty(), + Fields::Unnamed(f) => f.unnamed.is_empty(), + } + } +} + +impl IntoIterator for Fields { + type Item = Field; + type IntoIter = punctuated::IntoIter; + + fn into_iter(self) -> Self::IntoIter { + match self { + Fields::Unit => Punctuated::::new().into_iter(), + Fields::Named(f) => f.named.into_iter(), + Fields::Unnamed(f) => f.unnamed.into_iter(), + } + } +} + +impl<'a> IntoIterator for &'a Fields { + type Item = &'a Field; + type IntoIter = punctuated::Iter<'a, Field>; + + fn into_iter(self) -> Self::IntoIter { + self.iter() + } +} + +impl<'a> IntoIterator for &'a mut Fields { + type Item = &'a mut Field; + type IntoIter = punctuated::IterMut<'a, Field>; + + fn into_iter(self) -> Self::IntoIter { + self.iter_mut() + } +} + +ast_struct! { + /// A field of a struct or enum variant. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Field { + /// Attributes tagged on the field. + pub attrs: Vec, + + /// Visibility of the field. + pub vis: Visibility, + + /// Name of the field, if any. + /// + /// Fields of tuple structs have no names. + pub ident: Option, + + pub colon_token: Option, + + /// Type of the field. + pub ty: Type, + } +} + +ast_enum_of_structs! { + /// The visibility level of an item: inherited or `pub` or + /// `pub(restricted)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum Visibility { + /// A public visibility level: `pub`. + Public(VisPublic), + + /// A crate-level visibility: `crate`. + Crate(VisCrate), + + /// A visibility level restricted to some path: `pub(self)` or + /// `pub(super)` or `pub(crate)` or `pub(in some::module)`. + Restricted(VisRestricted), + + /// An inherited visibility, which usually means private. + Inherited, + } +} + +ast_struct! { + /// A public visibility level: `pub`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct VisPublic { + pub pub_token: Token![pub], + } +} + +ast_struct! { + /// A crate-level visibility: `crate`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct VisCrate { + pub crate_token: Token![crate], + } +} + +ast_struct! { + /// A visibility level restricted to some path: `pub(self)` or + /// `pub(super)` or `pub(crate)` or `pub(in some::module)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct VisRestricted { + pub pub_token: Token![pub], + pub paren_token: token::Paren, + pub in_token: Option, + pub path: Box, + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::discouraged::Speculative; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Variant { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let _visibility: Visibility = input.parse()?; + let ident: Ident = input.parse()?; + let fields = if input.peek(token::Brace) { + Fields::Named(input.parse()?) + } else if input.peek(token::Paren) { + Fields::Unnamed(input.parse()?) + } else { + Fields::Unit + }; + let discriminant = if input.peek(Token![=]) { + let eq_token: Token![=] = input.parse()?; + let discriminant: Expr = input.parse()?; + Some((eq_token, discriminant)) + } else { + None + }; + Ok(Variant { + attrs, + ident, + fields, + discriminant, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for FieldsNamed { + fn parse(input: ParseStream) -> Result { + let content; + Ok(FieldsNamed { + brace_token: braced!(content in input), + named: content.parse_terminated(Field::parse_named)?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for FieldsUnnamed { + fn parse(input: ParseStream) -> Result { + let content; + Ok(FieldsUnnamed { + paren_token: parenthesized!(content in input), + unnamed: content.parse_terminated(Field::parse_unnamed)?, + }) + } + } + + impl Field { + /// Parses a named (braced struct) field. + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_named(input: ParseStream) -> Result { + Ok(Field { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + ident: Some(if input.peek(Token![_]) { + input.call(Ident::parse_any) + } else { + input.parse() + }?), + colon_token: Some(input.parse()?), + ty: input.parse()?, + }) + } + + /// Parses an unnamed (tuple struct) field. + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_unnamed(input: ParseStream) -> Result { + Ok(Field { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + ident: None, + colon_token: None, + ty: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Visibility { + fn parse(input: ParseStream) -> Result { + // Recognize an empty None-delimited group, as produced by a $:vis + // matcher that matched no tokens. + if input.peek(token::Group) { + let ahead = input.fork(); + let group = crate::group::parse_group(&ahead)?; + if group.content.is_empty() { + input.advance_to(&ahead); + return Ok(Visibility::Inherited); + } + } + + if input.peek(Token![pub]) { + Self::parse_pub(input) + } else if input.peek(Token![crate]) { + Self::parse_crate(input) + } else { + Ok(Visibility::Inherited) + } + } + } + + impl Visibility { + fn parse_pub(input: ParseStream) -> Result { + let pub_token = input.parse::()?; + + if input.peek(token::Paren) { + let ahead = input.fork(); + + let content; + let paren_token = parenthesized!(content in ahead); + if content.peek(Token![crate]) + || content.peek(Token![self]) + || content.peek(Token![super]) + { + let path = content.call(Ident::parse_any)?; + + // Ensure there are no additional tokens within `content`. + // Without explicitly checking, we may misinterpret a tuple + // field as a restricted visibility, causing a parse error. + // e.g. `pub (crate::A, crate::B)` (Issue #720). + if content.is_empty() { + input.advance_to(&ahead); + return Ok(Visibility::Restricted(VisRestricted { + pub_token, + paren_token, + in_token: None, + path: Box::new(Path::from(path)), + })); + } + } else if content.peek(Token![in]) { + let in_token: Token![in] = content.parse()?; + let path = content.call(Path::parse_mod_style)?; + + input.advance_to(&ahead); + return Ok(Visibility::Restricted(VisRestricted { + pub_token, + paren_token, + in_token: Some(in_token), + path: Box::new(path), + })); + } + } + + Ok(Visibility::Public(VisPublic { pub_token })) + } + + fn parse_crate(input: ParseStream) -> Result { + if input.peek2(Token![::]) { + Ok(Visibility::Inherited) + } else { + Ok(Visibility::Crate(VisCrate { + crate_token: input.parse()?, + })) + } + } + + #[cfg(feature = "full")] + pub(crate) fn is_some(&self) -> bool { + match self { + Visibility::Inherited => false, + _ => true, + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::print::TokensOrDefault; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Variant { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(&self.attrs); + self.ident.to_tokens(tokens); + self.fields.to_tokens(tokens); + if let Some((eq_token, disc)) = &self.discriminant { + eq_token.to_tokens(tokens); + disc.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for FieldsNamed { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.brace_token.surround(tokens, |tokens| { + self.named.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for FieldsUnnamed { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.paren_token.surround(tokens, |tokens| { + self.unnamed.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Field { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(&self.attrs); + self.vis.to_tokens(tokens); + if let Some(ident) = &self.ident { + ident.to_tokens(tokens); + TokensOrDefault(&self.colon_token).to_tokens(tokens); + } + self.ty.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for VisPublic { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.pub_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for VisCrate { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.crate_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for VisRestricted { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.pub_token.to_tokens(tokens); + self.paren_token.surround(tokens, |tokens| { + // TODO: If we have a path which is not "self" or "super" or + // "crate", automatically add the "in" token. + self.in_token.to_tokens(tokens); + self.path.to_tokens(tokens); + }); + } + } +} diff --git a/rust/syn/derive.rs b/rust/syn/derive.rs new file mode 100644 index 00000000000000..455e72938067c4 --- /dev/null +++ b/rust/syn/derive.rs @@ -0,0 +1,276 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; + +ast_struct! { + /// Data structure sent to a `proc_macro_derive` macro. + /// + /// *This type is available only if Syn is built with the `"derive"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "derive")))] + pub struct DeriveInput { + /// Attributes tagged on the whole struct or enum. + pub attrs: Vec, + + /// Visibility of the struct or enum. + pub vis: Visibility, + + /// Name of the struct or enum. + pub ident: Ident, + + /// Generics required to complete the definition. + pub generics: Generics, + + /// Data within the struct or enum. + pub data: Data, + } +} + +ast_enum_of_structs! { + /// The storage of a struct, enum or union data structure. + /// + /// *This type is available only if Syn is built with the `"derive"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "derive")))] + pub enum Data { + /// A struct input to a `proc_macro_derive` macro. + Struct(DataStruct), + + /// An enum input to a `proc_macro_derive` macro. + Enum(DataEnum), + + /// An untagged union input to a `proc_macro_derive` macro. + Union(DataUnion), + } + + do_not_generate_to_tokens +} + +ast_struct! { + /// A struct input to a `proc_macro_derive` macro. + /// + /// *This type is available only if Syn is built with the `"derive"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "derive")))] + pub struct DataStruct { + pub struct_token: Token![struct], + pub fields: Fields, + pub semi_token: Option, + } +} + +ast_struct! { + /// An enum input to a `proc_macro_derive` macro. + /// + /// *This type is available only if Syn is built with the `"derive"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "derive")))] + pub struct DataEnum { + pub enum_token: Token![enum], + pub brace_token: token::Brace, + pub variants: Punctuated, + } +} + +ast_struct! { + /// An untagged union input to a `proc_macro_derive` macro. + /// + /// *This type is available only if Syn is built with the `"derive"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "derive")))] + pub struct DataUnion { + pub union_token: Token![union], + pub fields: FieldsNamed, + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for DeriveInput { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis = input.parse::()?; + + let lookahead = input.lookahead1(); + if lookahead.peek(Token![struct]) { + let struct_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, fields, semi) = data_struct(input)?; + Ok(DeriveInput { + attrs, + vis, + ident, + generics: Generics { + where_clause, + ..generics + }, + data: Data::Struct(DataStruct { + struct_token, + fields, + semi_token: semi, + }), + }) + } else if lookahead.peek(Token![enum]) { + let enum_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, brace, variants) = data_enum(input)?; + Ok(DeriveInput { + attrs, + vis, + ident, + generics: Generics { + where_clause, + ..generics + }, + data: Data::Enum(DataEnum { + enum_token, + brace_token: brace, + variants, + }), + }) + } else if lookahead.peek(Token![union]) { + let union_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, fields) = data_union(input)?; + Ok(DeriveInput { + attrs, + vis, + ident, + generics: Generics { + where_clause, + ..generics + }, + data: Data::Union(DataUnion { + union_token, + fields, + }), + }) + } else { + Err(lookahead.error()) + } + } + } + + pub fn data_struct( + input: ParseStream, + ) -> Result<(Option, Fields, Option)> { + let mut lookahead = input.lookahead1(); + let mut where_clause = None; + if lookahead.peek(Token![where]) { + where_clause = Some(input.parse()?); + lookahead = input.lookahead1(); + } + + if where_clause.is_none() && lookahead.peek(token::Paren) { + let fields = input.parse()?; + + lookahead = input.lookahead1(); + if lookahead.peek(Token![where]) { + where_clause = Some(input.parse()?); + lookahead = input.lookahead1(); + } + + if lookahead.peek(Token![;]) { + let semi = input.parse()?; + Ok((where_clause, Fields::Unnamed(fields), Some(semi))) + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(token::Brace) { + let fields = input.parse()?; + Ok((where_clause, Fields::Named(fields), None)) + } else if lookahead.peek(Token![;]) { + let semi = input.parse()?; + Ok((where_clause, Fields::Unit, Some(semi))) + } else { + Err(lookahead.error()) + } + } + + pub fn data_enum( + input: ParseStream, + ) -> Result<( + Option, + token::Brace, + Punctuated, + )> { + let where_clause = input.parse()?; + + let content; + let brace = braced!(content in input); + let variants = content.parse_terminated(Variant::parse)?; + + Ok((where_clause, brace, variants)) + } + + pub fn data_union(input: ParseStream) -> Result<(Option, FieldsNamed)> { + let where_clause = input.parse()?; + let fields = input.parse()?; + Ok((where_clause, fields)) + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::attr::FilterAttrs; + use crate::print::TokensOrDefault; + use proc_macro2::TokenStream; + use quote::ToTokens; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for DeriveInput { + fn to_tokens(&self, tokens: &mut TokenStream) { + for attr in self.attrs.outer() { + attr.to_tokens(tokens); + } + self.vis.to_tokens(tokens); + match &self.data { + Data::Struct(d) => d.struct_token.to_tokens(tokens), + Data::Enum(d) => d.enum_token.to_tokens(tokens), + Data::Union(d) => d.union_token.to_tokens(tokens), + } + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + match &self.data { + Data::Struct(data) => match &data.fields { + Fields::Named(fields) => { + self.generics.where_clause.to_tokens(tokens); + fields.to_tokens(tokens); + } + Fields::Unnamed(fields) => { + fields.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + TokensOrDefault(&data.semi_token).to_tokens(tokens); + } + Fields::Unit => { + self.generics.where_clause.to_tokens(tokens); + TokensOrDefault(&data.semi_token).to_tokens(tokens); + } + }, + Data::Enum(data) => { + self.generics.where_clause.to_tokens(tokens); + data.brace_token.surround(tokens, |tokens| { + data.variants.to_tokens(tokens); + }); + } + Data::Union(data) => { + self.generics.where_clause.to_tokens(tokens); + data.fields.to_tokens(tokens); + } + } + } + } +} diff --git a/rust/syn/discouraged.rs b/rust/syn/discouraged.rs new file mode 100644 index 00000000000000..9843c82d54d213 --- /dev/null +++ b/rust/syn/discouraged.rs @@ -0,0 +1,196 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Extensions to the parsing API with niche applicability. + +use super::*; + +/// Extensions to the `ParseStream` API to support speculative parsing. +pub trait Speculative { + /// Advance this parse stream to the position of a forked parse stream. + /// + /// This is the opposite operation to [`ParseStream::fork`]. You can fork a + /// parse stream, perform some speculative parsing, then join the original + /// stream to the fork to "commit" the parsing from the fork to the main + /// stream. + /// + /// If you can avoid doing this, you should, as it limits the ability to + /// generate useful errors. That said, it is often the only way to parse + /// syntax of the form `A* B*` for arbitrary syntax `A` and `B`. The problem + /// is that when the fork fails to parse an `A`, it's impossible to tell + /// whether that was because of a syntax error and the user meant to provide + /// an `A`, or that the `A`s are finished and it's time to start parsing + /// `B`s. Use with care. + /// + /// Also note that if `A` is a subset of `B`, `A* B*` can be parsed by + /// parsing `B*` and removing the leading members of `A` from the + /// repetition, bypassing the need to involve the downsides associated with + /// speculative parsing. + /// + /// [`ParseStream::fork`]: ParseBuffer::fork + /// + /// # Example + /// + /// There has been chatter about the possibility of making the colons in the + /// turbofish syntax like `path::to::` no longer required by accepting + /// `path::to` in expression position. Specifically, according to [RFC + /// 2544], [`PathSegment`] parsing should always try to consume a following + /// `<` token as the start of generic arguments, and reset to the `<` if + /// that fails (e.g. the token is acting as a less-than operator). + /// + /// This is the exact kind of parsing behavior which requires the "fork, + /// try, commit" behavior that [`ParseStream::fork`] discourages. With + /// `advance_to`, we can avoid having to parse the speculatively parsed + /// content a second time. + /// + /// This change in behavior can be implemented in syn by replacing just the + /// `Parse` implementation for `PathSegment`: + /// + /// ``` + /// # use syn::ext::IdentExt; + /// use syn::parse::discouraged::Speculative; + /// # use syn::parse::{Parse, ParseStream}; + /// # use syn::{Ident, PathArguments, Result, Token}; + /// + /// pub struct PathSegment { + /// pub ident: Ident, + /// pub arguments: PathArguments, + /// } + /// # + /// # impl From for PathSegment + /// # where + /// # T: Into, + /// # { + /// # fn from(ident: T) -> Self { + /// # PathSegment { + /// # ident: ident.into(), + /// # arguments: PathArguments::None, + /// # } + /// # } + /// # } + /// + /// impl Parse for PathSegment { + /// fn parse(input: ParseStream) -> Result { + /// if input.peek(Token![super]) + /// || input.peek(Token![self]) + /// || input.peek(Token![Self]) + /// || input.peek(Token![crate]) + /// { + /// let ident = input.call(Ident::parse_any)?; + /// return Ok(PathSegment::from(ident)); + /// } + /// + /// let ident = input.parse()?; + /// if input.peek(Token![::]) && input.peek3(Token![<]) { + /// return Ok(PathSegment { + /// ident, + /// arguments: PathArguments::AngleBracketed(input.parse()?), + /// }); + /// } + /// if input.peek(Token![<]) && !input.peek(Token![<=]) { + /// let fork = input.fork(); + /// if let Ok(arguments) = fork.parse() { + /// input.advance_to(&fork); + /// return Ok(PathSegment { + /// ident, + /// arguments: PathArguments::AngleBracketed(arguments), + /// }); + /// } + /// } + /// Ok(PathSegment::from(ident)) + /// } + /// } + /// + /// # syn::parse_str::("a").unwrap(); + /// ``` + /// + /// # Drawbacks + /// + /// The main drawback of this style of speculative parsing is in error + /// presentation. Even if the lookahead is the "correct" parse, the error + /// that is shown is that of the "fallback" parse. To use the same example + /// as the turbofish above, take the following unfinished "turbofish": + /// + /// ```text + /// let _ = f<&'a fn(), for<'a> serde::>(); + /// ``` + /// + /// If this is parsed as generic arguments, we can provide the error message + /// + /// ```text + /// error: expected identifier + /// --> src.rs:L:C + /// | + /// L | let _ = f<&'a fn(), for<'a> serde::>(); + /// | ^ + /// ``` + /// + /// but if parsed using the above speculative parsing, it falls back to + /// assuming that the `<` is a less-than when it fails to parse the generic + /// arguments, and tries to interpret the `&'a` as the start of a labelled + /// loop, resulting in the much less helpful error + /// + /// ```text + /// error: expected `:` + /// --> src.rs:L:C + /// | + /// L | let _ = f<&'a fn(), for<'a> serde::>(); + /// | ^^ + /// ``` + /// + /// This can be mitigated with various heuristics (two examples: show both + /// forks' parse errors, or show the one that consumed more tokens), but + /// when you can control the grammar, sticking to something that can be + /// parsed LL(3) and without the LL(*) speculative parsing this makes + /// possible, displaying reasonable errors becomes much more simple. + /// + /// [RFC 2544]: https://github.com/rust-lang/rfcs/pull/2544 + /// [`PathSegment`]: crate::PathSegment + /// + /// # Performance + /// + /// This method performs a cheap fixed amount of work that does not depend + /// on how far apart the two streams are positioned. + /// + /// # Panics + /// + /// The forked stream in the argument of `advance_to` must have been + /// obtained by forking `self`. Attempting to advance to any other stream + /// will cause a panic. + fn advance_to(&self, fork: &Self); +} + +impl<'a> Speculative for ParseBuffer<'a> { + fn advance_to(&self, fork: &Self) { + if !crate::buffer::same_scope(self.cursor(), fork.cursor()) { + panic!("Fork was not derived from the advancing parse stream"); + } + + let (self_unexp, self_sp) = inner_unexpected(self); + let (fork_unexp, fork_sp) = inner_unexpected(fork); + if !Rc::ptr_eq(&self_unexp, &fork_unexp) { + match (fork_sp, self_sp) { + // Unexpected set on the fork, but not on `self`, copy it over. + (Some(span), None) => { + self_unexp.set(Unexpected::Some(span)); + } + // Unexpected unset. Use chain to propagate errors from fork. + (None, None) => { + fork_unexp.set(Unexpected::Chain(self_unexp)); + + // Ensure toplevel 'unexpected' tokens from the fork don't + // bubble up the chain by replacing the root `unexpected` + // pointer, only 'unexpected' tokens from existing group + // parsers should bubble. + fork.unexpected + .set(Some(Rc::new(Cell::new(Unexpected::None)))); + } + // Unexpected has been set on `self`. No changes needed. + (_, Some(_)) => {} + } + } + + // See comment on `cell` in the struct definition. + self.cell + .set(unsafe { mem::transmute::>(fork.cursor()) }); + } +} diff --git a/rust/syn/error.rs b/rust/syn/error.rs new file mode 100644 index 00000000000000..262f3ab8cab758 --- /dev/null +++ b/rust/syn/error.rs @@ -0,0 +1,430 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg(feature = "parsing")] +use crate::buffer::Cursor; +use crate::thread::ThreadBound; +use proc_macro2::{ + Delimiter, Group, Ident, LexError, Literal, Punct, Spacing, Span, TokenStream, TokenTree, +}; +#[cfg(feature = "printing")] +use quote::ToTokens; +use std::fmt::{self, Debug, Display}; +use std::iter::FromIterator; +use std::slice; +use std::vec; + +/// The result of a Syn parser. +pub type Result = std::result::Result; + +/// Error returned when a Syn parser cannot parse the input tokens. +/// +/// # Error reporting in proc macros +/// +/// The correct way to report errors back to the compiler from a procedural +/// macro is by emitting an appropriately spanned invocation of +/// [`compile_error!`] in the generated code. This produces a better diagnostic +/// message than simply panicking the macro. +/// +/// [`compile_error!`]: std::compile_error! +/// +/// When parsing macro input, the [`parse_macro_input!`] macro handles the +/// conversion to `compile_error!` automatically. +/// +/// [`parse_macro_input!`]: crate::parse_macro_input! +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// use proc_macro::TokenStream; +/// use syn::{parse_macro_input, AttributeArgs, ItemFn}; +/// +/// # const IGNORE: &str = stringify! { +/// #[proc_macro_attribute] +/// # }; +/// pub fn my_attr(args: TokenStream, input: TokenStream) -> TokenStream { +/// let args = parse_macro_input!(args as AttributeArgs); +/// let input = parse_macro_input!(input as ItemFn); +/// +/// /* ... */ +/// # TokenStream::new() +/// } +/// ``` +/// +/// For errors that arise later than the initial parsing stage, the +/// [`.to_compile_error()`] or [`.into_compile_error()`] methods can be used to +/// perform an explicit conversion to `compile_error!`. +/// +/// [`.to_compile_error()`]: Error::to_compile_error +/// [`.into_compile_error()`]: Error::into_compile_error +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// # use proc_macro::TokenStream; +/// # use syn::{parse_macro_input, DeriveInput}; +/// # +/// # const IGNORE: &str = stringify! { +/// #[proc_macro_derive(MyDerive)] +/// # }; +/// pub fn my_derive(input: TokenStream) -> TokenStream { +/// let input = parse_macro_input!(input as DeriveInput); +/// +/// // fn(DeriveInput) -> syn::Result +/// expand::my_derive(input) +/// .unwrap_or_else(syn::Error::into_compile_error) +/// .into() +/// } +/// # +/// # mod expand { +/// # use proc_macro2::TokenStream; +/// # use syn::{DeriveInput, Result}; +/// # +/// # pub fn my_derive(input: DeriveInput) -> Result { +/// # unimplemented!() +/// # } +/// # } +/// ``` +pub struct Error { + messages: Vec, +} + +struct ErrorMessage { + // Span is implemented as an index into a thread-local interner to keep the + // size small. It is not safe to access from a different thread. We want + // errors to be Send and Sync to play nicely with the Failure crate, so pin + // the span we're given to its original thread and assume it is + // Span::call_site if accessed from any other thread. + start_span: ThreadBound, + end_span: ThreadBound, + message: String, +} + +#[cfg(test)] +struct _Test +where + Error: Send + Sync; + +impl Error { + /// Usually the [`ParseStream::error`] method will be used instead, which + /// automatically uses the correct span from the current position of the + /// parse stream. + /// + /// Use `Error::new` when the error needs to be triggered on some span other + /// than where the parse stream is currently positioned. + /// + /// [`ParseStream::error`]: crate::parse::ParseBuffer::error + /// + /// # Example + /// + /// ``` + /// use syn::{Error, Ident, LitStr, Result, Token}; + /// use syn::parse::ParseStream; + /// + /// // Parses input that looks like `name = "string"` where the key must be + /// // the identifier `name` and the value may be any string literal. + /// // Returns the string literal. + /// fn parse_name(input: ParseStream) -> Result { + /// let name_token: Ident = input.parse()?; + /// if name_token != "name" { + /// // Trigger an error not on the current position of the stream, + /// // but on the position of the unexpected identifier. + /// return Err(Error::new(name_token.span(), "expected `name`")); + /// } + /// input.parse::()?; + /// let s: LitStr = input.parse()?; + /// Ok(s) + /// } + /// ``` + pub fn new(span: Span, message: T) -> Self { + return new(span, message.to_string()); + + fn new(span: Span, message: String) -> Error { + Error { + messages: vec![ErrorMessage { + start_span: ThreadBound::new(span), + end_span: ThreadBound::new(span), + message, + }], + } + } + } + + /// Creates an error with the specified message spanning the given syntax + /// tree node. + /// + /// Unlike the `Error::new` constructor, this constructor takes an argument + /// `tokens` which is a syntax tree node. This allows the resulting `Error` + /// to attempt to span all tokens inside of `tokens`. While you would + /// typically be able to use the `Spanned` trait with the above `Error::new` + /// constructor, implementation limitations today mean that + /// `Error::new_spanned` may provide a higher-quality error message on + /// stable Rust. + /// + /// When in doubt it's recommended to stick to `Error::new` (or + /// `ParseStream::error`)! + #[cfg(feature = "printing")] + pub fn new_spanned(tokens: T, message: U) -> Self { + return new_spanned(tokens.into_token_stream(), message.to_string()); + + fn new_spanned(tokens: TokenStream, message: String) -> Error { + let mut iter = tokens.into_iter(); + let start = iter.next().map_or_else(Span::call_site, |t| t.span()); + let end = iter.last().map_or(start, |t| t.span()); + Error { + messages: vec![ErrorMessage { + start_span: ThreadBound::new(start), + end_span: ThreadBound::new(end), + message, + }], + } + } + } + + /// The source location of the error. + /// + /// Spans are not thread-safe so this function returns `Span::call_site()` + /// if called from a different thread than the one on which the `Error` was + /// originally created. + pub fn span(&self) -> Span { + let start = match self.messages[0].start_span.get() { + Some(span) => *span, + None => return Span::call_site(), + }; + let end = match self.messages[0].end_span.get() { + Some(span) => *span, + None => return Span::call_site(), + }; + start.join(end).unwrap_or(start) + } + + /// Render the error as an invocation of [`compile_error!`]. + /// + /// The [`parse_macro_input!`] macro provides a convenient way to invoke + /// this method correctly in a procedural macro. + /// + /// [`compile_error!`]: std::compile_error! + /// [`parse_macro_input!`]: crate::parse_macro_input! + pub fn to_compile_error(&self) -> TokenStream { + self.messages + .iter() + .map(ErrorMessage::to_compile_error) + .collect() + } + + /// Render the error as an invocation of [`compile_error!`]. + /// + /// [`compile_error!`]: std::compile_error! + /// + /// # Example + /// + /// ``` + /// # extern crate proc_macro; + /// # + /// use proc_macro::TokenStream; + /// use syn::{parse_macro_input, DeriveInput, Error}; + /// + /// # const _: &str = stringify! { + /// #[proc_macro_derive(MyTrait)] + /// # }; + /// pub fn derive_my_trait(input: TokenStream) -> TokenStream { + /// let input = parse_macro_input!(input as DeriveInput); + /// my_trait::expand(input) + /// .unwrap_or_else(Error::into_compile_error) + /// .into() + /// } + /// + /// mod my_trait { + /// use proc_macro2::TokenStream; + /// use syn::{DeriveInput, Result}; + /// + /// pub(crate) fn expand(input: DeriveInput) -> Result { + /// /* ... */ + /// # unimplemented!() + /// } + /// } + /// ``` + pub fn into_compile_error(self) -> TokenStream { + self.to_compile_error() + } + + /// Add another error message to self such that when `to_compile_error()` is + /// called, both errors will be emitted together. + pub fn combine(&mut self, another: Error) { + self.messages.extend(another.messages); + } +} + +impl ErrorMessage { + fn to_compile_error(&self) -> TokenStream { + let start = self + .start_span + .get() + .cloned() + .unwrap_or_else(Span::call_site); + let end = self.end_span.get().cloned().unwrap_or_else(Span::call_site); + + // compile_error!($message) + TokenStream::from_iter(vec![ + TokenTree::Ident(Ident::new("compile_error", start)), + TokenTree::Punct({ + let mut punct = Punct::new('!', Spacing::Alone); + punct.set_span(start); + punct + }), + TokenTree::Group({ + let mut group = Group::new(Delimiter::Brace, { + TokenStream::from_iter(vec![TokenTree::Literal({ + let mut string = Literal::string(&self.message); + string.set_span(end); + string + })]) + }); + group.set_span(end); + group + }), + ]) + } +} + +#[cfg(feature = "parsing")] +pub fn new_at(scope: Span, cursor: Cursor, message: T) -> Error { + if cursor.eof() { + Error::new(scope, format!("unexpected end of input, {}", message)) + } else { + let span = crate::buffer::open_span_of_group(cursor); + Error::new(span, message) + } +} + +#[cfg(all(feature = "parsing", any(feature = "full", feature = "derive")))] +pub fn new2(start: Span, end: Span, message: T) -> Error { + return new2(start, end, message.to_string()); + + fn new2(start: Span, end: Span, message: String) -> Error { + Error { + messages: vec![ErrorMessage { + start_span: ThreadBound::new(start), + end_span: ThreadBound::new(end), + message, + }], + } + } +} + +impl Debug for Error { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + if self.messages.len() == 1 { + formatter + .debug_tuple("Error") + .field(&self.messages[0]) + .finish() + } else { + formatter + .debug_tuple("Error") + .field(&self.messages) + .finish() + } + } +} + +impl Debug for ErrorMessage { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.message, formatter) + } +} + +impl Display for Error { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str(&self.messages[0].message) + } +} + +impl Clone for Error { + fn clone(&self) -> Self { + Error { + messages: self.messages.clone(), + } + } +} + +impl Clone for ErrorMessage { + fn clone(&self) -> Self { + let start = self + .start_span + .get() + .cloned() + .unwrap_or_else(Span::call_site); + let end = self.end_span.get().cloned().unwrap_or_else(Span::call_site); + ErrorMessage { + start_span: ThreadBound::new(start), + end_span: ThreadBound::new(end), + message: self.message.clone(), + } + } +} + +impl std::error::Error for Error {} + +impl From for Error { + fn from(err: LexError) -> Self { + Error::new(err.span(), "lex error") + } +} + +impl IntoIterator for Error { + type Item = Error; + type IntoIter = IntoIter; + + fn into_iter(self) -> Self::IntoIter { + IntoIter { + messages: self.messages.into_iter(), + } + } +} + +pub struct IntoIter { + messages: vec::IntoIter, +} + +impl Iterator for IntoIter { + type Item = Error; + + fn next(&mut self) -> Option { + Some(Error { + messages: vec![self.messages.next()?], + }) + } +} + +impl<'a> IntoIterator for &'a Error { + type Item = Error; + type IntoIter = Iter<'a>; + + fn into_iter(self) -> Self::IntoIter { + Iter { + messages: self.messages.iter(), + } + } +} + +pub struct Iter<'a> { + messages: slice::Iter<'a, ErrorMessage>, +} + +impl<'a> Iterator for Iter<'a> { + type Item = Error; + + fn next(&mut self) -> Option { + Some(Error { + messages: vec![self.messages.next()?.clone()], + }) + } +} + +impl Extend for Error { + fn extend>(&mut self, iter: T) { + for err in iter { + self.combine(err); + } + } +} diff --git a/rust/syn/export.rs b/rust/syn/export.rs new file mode 100644 index 00000000000000..5ea4dd75eb4ae1 --- /dev/null +++ b/rust/syn/export.rs @@ -0,0 +1,41 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +pub use std::clone::Clone; +pub use std::cmp::{Eq, PartialEq}; +pub use std::default::Default; +pub use std::fmt::{self, Debug, Formatter}; +pub use std::hash::{Hash, Hasher}; +pub use std::marker::Copy; +pub use std::option::Option::{None, Some}; +pub use std::result::Result::{Err, Ok}; + +#[cfg(feature = "printing")] +pub extern crate quote; + +pub use proc_macro2::{Span, TokenStream as TokenStream2}; + +#[cfg(feature = "parsing")] +pub use crate::group::{parse_braces, parse_brackets, parse_parens}; + +pub use crate::span::IntoSpans; + +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" +))] +pub use proc_macro::TokenStream; + +#[cfg(feature = "printing")] +pub use quote::{ToTokens, TokenStreamExt}; + +#[allow(non_camel_case_types)] +pub type bool = help::Bool; +#[allow(non_camel_case_types)] +pub type str = help::Str; + +mod help { + pub type Bool = bool; + pub type Str = str; +} + +pub struct private(pub(crate) ()); diff --git a/rust/syn/expr.rs b/rust/syn/expr.rs new file mode 100644 index 00000000000000..54155892fe0405 --- /dev/null +++ b/rust/syn/expr.rs @@ -0,0 +1,3560 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; +#[cfg(feature = "full")] +use crate::reserved::Reserved; +use proc_macro2::{Span, TokenStream}; +#[cfg(feature = "printing")] +use quote::IdentFragment; +#[cfg(feature = "printing")] +use std::fmt::{self, Display}; +use std::hash::{Hash, Hasher}; +#[cfg(feature = "parsing")] +use std::mem; + +ast_enum_of_structs! { + /// A Rust expression. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature, but most of the variants are not available unless "full" is enabled.* + /// + /// # Syntax tree enums + /// + /// This type is a syntax tree enum. In Syn this and other syntax tree enums + /// are designed to be traversed using the following rebinding idiom. + /// + /// ``` + /// # use syn::Expr; + /// # + /// # fn example(expr: Expr) { + /// # const IGNORE: &str = stringify! { + /// let expr: Expr = /* ... */; + /// # }; + /// match expr { + /// Expr::MethodCall(expr) => { + /// /* ... */ + /// } + /// Expr::Cast(expr) => { + /// /* ... */ + /// } + /// Expr::If(expr) => { + /// /* ... */ + /// } + /// + /// /* ... */ + /// # _ => {} + /// # } + /// # } + /// ``` + /// + /// We begin with a variable `expr` of type `Expr` that has no fields + /// (because it is an enum), and by matching on it and rebinding a variable + /// with the same name `expr` we effectively imbue our variable with all of + /// the data fields provided by the variant that it turned out to be. So for + /// example above if we ended up in the `MethodCall` case then we get to use + /// `expr.receiver`, `expr.args` etc; if we ended up in the `If` case we get + /// to use `expr.cond`, `expr.then_branch`, `expr.else_branch`. + /// + /// This approach avoids repeating the variant names twice on every line. + /// + /// ``` + /// # use syn::{Expr, ExprMethodCall}; + /// # + /// # fn example(expr: Expr) { + /// // Repetitive; recommend not doing this. + /// match expr { + /// Expr::MethodCall(ExprMethodCall { method, args, .. }) => { + /// # } + /// # _ => {} + /// # } + /// # } + /// ``` + /// + /// In general, the name to which a syntax tree enum variant is bound should + /// be a suitable name for the complete syntax tree enum type. + /// + /// ``` + /// # use syn::{Expr, ExprField}; + /// # + /// # fn example(discriminant: ExprField) { + /// // Binding is called `base` which is the name I would use if I were + /// // assigning `*discriminant.base` without an `if let`. + /// if let Expr::Tuple(base) = *discriminant.base { + /// # } + /// # } + /// ``` + /// + /// A sign that you may not be choosing the right variable names is if you + /// see names getting repeated in your code, like accessing + /// `receiver.receiver` or `pat.pat` or `cond.cond`. + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum Expr { + /// A slice literal expression: `[a, b, c, d]`. + Array(ExprArray), + + /// An assignment expression: `a = compute()`. + Assign(ExprAssign), + + /// A compound assignment expression: `counter += 1`. + AssignOp(ExprAssignOp), + + /// An async block: `async { ... }`. + Async(ExprAsync), + + /// An await expression: `fut.await`. + Await(ExprAwait), + + /// A binary operation: `a + b`, `a * b`. + Binary(ExprBinary), + + /// A blocked scope: `{ ... }`. + Block(ExprBlock), + + /// A box expression: `box f`. + Box(ExprBox), + + /// A `break`, with an optional label to break and an optional + /// expression. + Break(ExprBreak), + + /// A function call expression: `invoke(a, b)`. + Call(ExprCall), + + /// A cast expression: `foo as f64`. + Cast(ExprCast), + + /// A closure expression: `|a, b| a + b`. + Closure(ExprClosure), + + /// A `continue`, with an optional label. + Continue(ExprContinue), + + /// Access of a named struct field (`obj.k`) or unnamed tuple struct + /// field (`obj.0`). + Field(ExprField), + + /// A for loop: `for pat in expr { ... }`. + ForLoop(ExprForLoop), + + /// An expression contained within invisible delimiters. + /// + /// This variant is important for faithfully representing the precedence + /// of expressions and is related to `None`-delimited spans in a + /// `TokenStream`. + Group(ExprGroup), + + /// An `if` expression with an optional `else` block: `if expr { ... } + /// else { ... }`. + /// + /// The `else` branch expression may only be an `If` or `Block` + /// expression, not any of the other types of expression. + If(ExprIf), + + /// A square bracketed indexing expression: `vector[2]`. + Index(ExprIndex), + + /// A `let` guard: `let Some(x) = opt`. + Let(ExprLet), + + /// A literal in place of an expression: `1`, `"foo"`. + Lit(ExprLit), + + /// Conditionless loop: `loop { ... }`. + Loop(ExprLoop), + + /// A macro invocation expression: `format!("{}", q)`. + Macro(ExprMacro), + + /// A `match` expression: `match n { Some(n) => {}, None => {} }`. + Match(ExprMatch), + + /// A method call expression: `x.foo::(a, b)`. + MethodCall(ExprMethodCall), + + /// A parenthesized expression: `(a + b)`. + Paren(ExprParen), + + /// A path like `std::mem::replace` possibly containing generic + /// parameters and a qualified self-type. + /// + /// A plain identifier like `x` is a path of length 1. + Path(ExprPath), + + /// A range expression: `1..2`, `1..`, `..2`, `1..=2`, `..=2`. + Range(ExprRange), + + /// A referencing operation: `&a` or `&mut a`. + Reference(ExprReference), + + /// An array literal constructed from one repeated element: `[0u8; N]`. + Repeat(ExprRepeat), + + /// A `return`, with an optional value to be returned. + Return(ExprReturn), + + /// A struct literal expression: `Point { x: 1, y: 1 }`. + /// + /// The `rest` provides the value of the remaining fields as in `S { a: + /// 1, b: 1, ..rest }`. + Struct(ExprStruct), + + /// A try-expression: `expr?`. + Try(ExprTry), + + /// A try block: `try { ... }`. + TryBlock(ExprTryBlock), + + /// A tuple expression: `(a, b, c, d)`. + Tuple(ExprTuple), + + /// A type ascription expression: `foo: f64`. + Type(ExprType), + + /// A unary operation: `!x`, `*x`. + Unary(ExprUnary), + + /// An unsafe block: `unsafe { ... }`. + Unsafe(ExprUnsafe), + + /// Tokens in expression position not interpreted by Syn. + Verbatim(TokenStream), + + /// A while loop: `while expr { ... }`. + While(ExprWhile), + + /// A yield expression: `yield expr`. + Yield(ExprYield), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match expr { + // Expr::Array(expr) => {...} + // Expr::Assign(expr) => {...} + // ... + // Expr::Yield(expr) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// A slice literal expression: `[a, b, c, d]`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprArray #full { + pub attrs: Vec, + pub bracket_token: token::Bracket, + pub elems: Punctuated, + } +} + +ast_struct! { + /// An assignment expression: `a = compute()`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprAssign #full { + pub attrs: Vec, + pub left: Box, + pub eq_token: Token![=], + pub right: Box, + } +} + +ast_struct! { + /// A compound assignment expression: `counter += 1`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprAssignOp #full { + pub attrs: Vec, + pub left: Box, + pub op: BinOp, + pub right: Box, + } +} + +ast_struct! { + /// An async block: `async { ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprAsync #full { + pub attrs: Vec, + pub async_token: Token![async], + pub capture: Option, + pub block: Block, + } +} + +ast_struct! { + /// An await expression: `fut.await`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprAwait #full { + pub attrs: Vec, + pub base: Box, + pub dot_token: Token![.], + pub await_token: token::Await, + } +} + +ast_struct! { + /// A binary operation: `a + b`, `a * b`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct ExprBinary { + pub attrs: Vec, + pub left: Box, + pub op: BinOp, + pub right: Box, + } +} + +ast_struct! { + /// A blocked scope: `{ ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ExprBlock #full { + pub attrs: Vec, + pub label: OptionTypeParam> + /// over the type parameters in `self.params`. + pub fn type_params(&self) -> TypeParams { + TypeParams(self.params.iter()) + } + + /// Returns an + /// Iterator<Item = &mut TypeParam> + /// over the type parameters in `self.params`. + pub fn type_params_mut(&mut self) -> TypeParamsMut { + TypeParamsMut(self.params.iter_mut()) + } + + /// Returns an + /// Iterator<Item = &LifetimeDef> + /// over the lifetime parameters in `self.params`. + pub fn lifetimes(&self) -> Lifetimes { + Lifetimes(self.params.iter()) + } + + /// Returns an + /// Iterator<Item = &mut LifetimeDef> + /// over the lifetime parameters in `self.params`. + pub fn lifetimes_mut(&mut self) -> LifetimesMut { + LifetimesMut(self.params.iter_mut()) + } + + /// Returns an + /// Iterator<Item = &ConstParam> + /// over the constant parameters in `self.params`. + pub fn const_params(&self) -> ConstParams { + ConstParams(self.params.iter()) + } + + /// Returns an + /// Iterator<Item = &mut ConstParam> + /// over the constant parameters in `self.params`. + pub fn const_params_mut(&mut self) -> ConstParamsMut { + ConstParamsMut(self.params.iter_mut()) + } + + /// Initializes an empty `where`-clause if there is not one present already. + pub fn make_where_clause(&mut self) -> &mut WhereClause { + self.where_clause.get_or_insert_with(|| WhereClause { + where_token: ::default(), + predicates: Punctuated::new(), + }) + } +} + +pub struct TypeParams<'a>(Iter<'a, GenericParam>); + +impl<'a> Iterator for TypeParams<'a> { + type Item = &'a TypeParam; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Type(type_param) = next { + Some(type_param) + } else { + self.next() + } + } +} + +pub struct TypeParamsMut<'a>(IterMut<'a, GenericParam>); + +impl<'a> Iterator for TypeParamsMut<'a> { + type Item = &'a mut TypeParam; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Type(type_param) = next { + Some(type_param) + } else { + self.next() + } + } +} + +pub struct Lifetimes<'a>(Iter<'a, GenericParam>); + +impl<'a> Iterator for Lifetimes<'a> { + type Item = &'a LifetimeDef; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Lifetime(lifetime) = next { + Some(lifetime) + } else { + self.next() + } + } +} + +pub struct LifetimesMut<'a>(IterMut<'a, GenericParam>); + +impl<'a> Iterator for LifetimesMut<'a> { + type Item = &'a mut LifetimeDef; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Lifetime(lifetime) = next { + Some(lifetime) + } else { + self.next() + } + } +} + +pub struct ConstParams<'a>(Iter<'a, GenericParam>); + +impl<'a> Iterator for ConstParams<'a> { + type Item = &'a ConstParam; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Const(const_param) = next { + Some(const_param) + } else { + self.next() + } + } +} + +pub struct ConstParamsMut<'a>(IterMut<'a, GenericParam>); + +impl<'a> Iterator for ConstParamsMut<'a> { + type Item = &'a mut ConstParam; + + fn next(&mut self) -> Option { + let next = match self.0.next() { + Some(item) => item, + None => return None, + }; + if let GenericParam::Const(const_param) = next { + Some(const_param) + } else { + self.next() + } + } +} + +/// Returned by `Generics::split_for_impl`. +/// +/// *This type is available only if Syn is built with the `"derive"` or `"full"` +/// feature and the `"printing"` feature.* +#[cfg(feature = "printing")] +#[cfg_attr( + doc_cfg, + doc(cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))) +)] +pub struct ImplGenerics<'a>(&'a Generics); + +/// Returned by `Generics::split_for_impl`. +/// +/// *This type is available only if Syn is built with the `"derive"` or `"full"` +/// feature and the `"printing"` feature.* +#[cfg(feature = "printing")] +#[cfg_attr( + doc_cfg, + doc(cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))) +)] +pub struct TypeGenerics<'a>(&'a Generics); + +/// Returned by `TypeGenerics::as_turbofish`. +/// +/// *This type is available only if Syn is built with the `"derive"` or `"full"` +/// feature and the `"printing"` feature.* +#[cfg(feature = "printing")] +#[cfg_attr( + doc_cfg, + doc(cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))) +)] +pub struct Turbofish<'a>(&'a Generics); + +#[cfg(feature = "printing")] +impl Generics { + /// Split a type's generics into the pieces required for impl'ing a trait + /// for that type. + /// + /// ``` + /// # use proc_macro2::{Span, Ident}; + /// # use quote::quote; + /// # + /// # let generics: syn::Generics = Default::default(); + /// # let name = Ident::new("MyType", Span::call_site()); + /// # + /// let (impl_generics, ty_generics, where_clause) = generics.split_for_impl(); + /// quote! { + /// impl #impl_generics MyTrait for #name #ty_generics #where_clause { + /// // ... + /// } + /// } + /// # ; + /// ``` + /// + /// *This method is available only if Syn is built with the `"derive"` or + /// `"full"` feature and the `"printing"` feature.* + #[cfg_attr( + doc_cfg, + doc(cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))) + )] + pub fn split_for_impl(&self) -> (ImplGenerics, TypeGenerics, Option<&WhereClause>) { + ( + ImplGenerics(self), + TypeGenerics(self), + self.where_clause.as_ref(), + ) + } +} + +#[cfg(feature = "printing")] +macro_rules! generics_wrapper_impls { + ($ty:ident) => { + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl<'a> Clone for $ty<'a> { + fn clone(&self) -> Self { + $ty(self.0) + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl<'a> Debug for $ty<'a> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_tuple(stringify!($ty)) + .field(self.0) + .finish() + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl<'a> Eq for $ty<'a> {} + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl<'a> PartialEq for $ty<'a> { + fn eq(&self, other: &Self) -> bool { + self.0 == other.0 + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl<'a> Hash for $ty<'a> { + fn hash(&self, state: &mut H) { + self.0.hash(state); + } + } + }; +} + +#[cfg(feature = "printing")] +generics_wrapper_impls!(ImplGenerics); +#[cfg(feature = "printing")] +generics_wrapper_impls!(TypeGenerics); +#[cfg(feature = "printing")] +generics_wrapper_impls!(Turbofish); + +#[cfg(feature = "printing")] +impl<'a> TypeGenerics<'a> { + /// Turn a type's generics like `` into a turbofish like `::`. + /// + /// *This method is available only if Syn is built with the `"derive"` or + /// `"full"` feature and the `"printing"` feature.* + pub fn as_turbofish(&self) -> Turbofish { + Turbofish(self.0) + } +} + +ast_struct! { + /// A set of bound lifetimes: `for<'a, 'b, 'c>`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct BoundLifetimes { + pub for_token: Token![for], + pub lt_token: Token![<], + pub lifetimes: Punctuated, + pub gt_token: Token![>], + } +} + +impl Default for BoundLifetimes { + fn default() -> Self { + BoundLifetimes { + for_token: Default::default(), + lt_token: Default::default(), + lifetimes: Punctuated::new(), + gt_token: Default::default(), + } + } +} + +impl LifetimeDef { + pub fn new(lifetime: Lifetime) -> Self { + LifetimeDef { + attrs: Vec::new(), + lifetime, + colon_token: None, + bounds: Punctuated::new(), + } + } +} + +impl From for TypeParam { + fn from(ident: Ident) -> Self { + TypeParam { + attrs: vec![], + ident, + colon_token: None, + bounds: Punctuated::new(), + eq_token: None, + default: None, + } + } +} + +ast_enum_of_structs! { + /// A trait or lifetime used as a bound on a type parameter. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum TypeParamBound { + Trait(TraitBound), + Lifetime(Lifetime), + } +} + +ast_struct! { + /// A trait used as a bound on a type parameter. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TraitBound { + pub paren_token: Option, + pub modifier: TraitBoundModifier, + /// The `for<'a>` in `for<'a> Foo<&'a T>` + pub lifetimes: Option, + /// The `Foo<&'a T>` in `for<'a> Foo<&'a T>` + pub path: Path, + } +} + +ast_enum! { + /// A modifier on a trait bound, currently only used for the `?` in + /// `?Sized`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum TraitBoundModifier { + None, + Maybe(Token![?]), + } +} + +ast_struct! { + /// A `where` clause in a definition: `where T: Deserialize<'de>, D: + /// 'static`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct WhereClause { + pub where_token: Token![where], + pub predicates: Punctuated, + } +} + +ast_enum_of_structs! { + /// A single predicate in a `where` clause: `T: Deserialize<'de>`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum WherePredicate { + /// A type predicate in a `where` clause: `for<'c> Foo<'c>: Trait<'c>`. + Type(PredicateType), + + /// A lifetime predicate in a `where` clause: `'a: 'b + 'c`. + Lifetime(PredicateLifetime), + + /// An equality predicate in a `where` clause (unsupported). + Eq(PredicateEq), + } +} + +ast_struct! { + /// A type predicate in a `where` clause: `for<'c> Foo<'c>: Trait<'c>`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct PredicateType { + /// Any lifetimes from a `for` binding + pub lifetimes: Option, + /// The type being bounded + pub bounded_ty: Type, + pub colon_token: Token![:], + /// Trait and lifetime bounds (`Clone+Send+'static`) + pub bounds: Punctuated, + } +} + +ast_struct! { + /// A lifetime predicate in a `where` clause: `'a: 'b + 'c`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct PredicateLifetime { + pub lifetime: Lifetime, + pub colon_token: Token![:], + pub bounds: Punctuated, + } +} + +ast_struct! { + /// An equality predicate in a `where` clause (unsupported). + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct PredicateEq { + pub lhs_ty: Type, + pub eq_token: Token![=], + pub rhs_ty: Type, + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Generics { + fn parse(input: ParseStream) -> Result { + if !input.peek(Token![<]) { + return Ok(Generics::default()); + } + + let lt_token: Token![<] = input.parse()?; + + let mut params = Punctuated::new(); + loop { + if input.peek(Token![>]) { + break; + } + + let attrs = input.call(Attribute::parse_outer)?; + let lookahead = input.lookahead1(); + if lookahead.peek(Lifetime) { + params.push_value(GenericParam::Lifetime(LifetimeDef { + attrs, + ..input.parse()? + })); + } else if lookahead.peek(Ident) { + params.push_value(GenericParam::Type(TypeParam { + attrs, + ..input.parse()? + })); + } else if lookahead.peek(Token![const]) { + params.push_value(GenericParam::Const(ConstParam { + attrs, + ..input.parse()? + })); + } else if input.peek(Token![_]) { + params.push_value(GenericParam::Type(TypeParam { + attrs, + ident: input.call(Ident::parse_any)?, + colon_token: None, + bounds: Punctuated::new(), + eq_token: None, + default: None, + })); + } else { + return Err(lookahead.error()); + } + + if input.peek(Token![>]) { + break; + } + let punct = input.parse()?; + params.push_punct(punct); + } + + let gt_token: Token![>] = input.parse()?; + + Ok(Generics { + lt_token: Some(lt_token), + params, + gt_token: Some(gt_token), + where_clause: None, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for GenericParam { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) { + Ok(GenericParam::Type(TypeParam { + attrs, + ..input.parse()? + })) + } else if lookahead.peek(Lifetime) { + Ok(GenericParam::Lifetime(LifetimeDef { + attrs, + ..input.parse()? + })) + } else if lookahead.peek(Token![const]) { + Ok(GenericParam::Const(ConstParam { + attrs, + ..input.parse()? + })) + } else { + Err(lookahead.error()) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LifetimeDef { + fn parse(input: ParseStream) -> Result { + let has_colon; + Ok(LifetimeDef { + attrs: input.call(Attribute::parse_outer)?, + lifetime: input.parse()?, + colon_token: { + if input.peek(Token![:]) { + has_colon = true; + Some(input.parse()?) + } else { + has_colon = false; + None + } + }, + bounds: { + let mut bounds = Punctuated::new(); + if has_colon { + loop { + if input.peek(Token![,]) || input.peek(Token![>]) { + break; + } + let value = input.parse()?; + bounds.push_value(value); + if !input.peek(Token![+]) { + break; + } + let punct = input.parse()?; + bounds.push_punct(punct); + } + } + bounds + }, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for BoundLifetimes { + fn parse(input: ParseStream) -> Result { + Ok(BoundLifetimes { + for_token: input.parse()?, + lt_token: input.parse()?, + lifetimes: { + let mut lifetimes = Punctuated::new(); + while !input.peek(Token![>]) { + lifetimes.push_value(input.parse()?); + if input.peek(Token![>]) { + break; + } + lifetimes.push_punct(input.parse()?); + } + lifetimes + }, + gt_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Option { + fn parse(input: ParseStream) -> Result { + if input.peek(Token![for]) { + input.parse().map(Some) + } else { + Ok(None) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeParam { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let ident: Ident = input.parse()?; + let colon_token: Option = input.parse()?; + + let begin_bound = input.fork(); + let mut is_maybe_const = false; + let mut bounds = Punctuated::new(); + if colon_token.is_some() { + loop { + if input.peek(Token![,]) || input.peek(Token![>]) || input.peek(Token![=]) { + break; + } + if input.peek(Token![~]) && input.peek2(Token![const]) { + input.parse::()?; + input.parse::()?; + is_maybe_const = true; + } + let value: TypeParamBound = input.parse()?; + bounds.push_value(value); + if !input.peek(Token![+]) { + break; + } + let punct: Token![+] = input.parse()?; + bounds.push_punct(punct); + } + } + + let mut eq_token: Option = input.parse()?; + let mut default = if eq_token.is_some() { + Some(input.parse::()?) + } else { + None + }; + + if is_maybe_const { + bounds.clear(); + eq_token = None; + default = Some(Type::Verbatim(verbatim::between(begin_bound, input))); + } + + Ok(TypeParam { + attrs, + ident, + colon_token, + bounds, + eq_token, + default, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeParamBound { + fn parse(input: ParseStream) -> Result { + if input.peek(Lifetime) { + return input.parse().map(TypeParamBound::Lifetime); + } + + if input.peek(token::Paren) { + let content; + let paren_token = parenthesized!(content in input); + let mut bound: TraitBound = content.parse()?; + bound.paren_token = Some(paren_token); + return Ok(TypeParamBound::Trait(bound)); + } + + input.parse().map(TypeParamBound::Trait) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitBound { + fn parse(input: ParseStream) -> Result { + #[cfg(feature = "full")] + let tilde_const = if input.peek(Token![~]) && input.peek2(Token![const]) { + let tilde_token = input.parse::()?; + let const_token = input.parse::()?; + Some((tilde_token, const_token)) + } else { + None + }; + + let modifier: TraitBoundModifier = input.parse()?; + let lifetimes: Option = input.parse()?; + + let mut path: Path = input.parse()?; + if path.segments.last().unwrap().arguments.is_empty() + && (input.peek(token::Paren) || input.peek(Token![::]) && input.peek3(token::Paren)) + { + input.parse::>()?; + let args: ParenthesizedGenericArguments = input.parse()?; + let parenthesized = PathArguments::Parenthesized(args); + path.segments.last_mut().unwrap().arguments = parenthesized; + } + + #[cfg(feature = "full")] + { + if let Some((tilde_token, const_token)) = tilde_const { + path.segments.insert( + 0, + PathSegment { + ident: Ident::new("const", const_token.span), + arguments: PathArguments::None, + }, + ); + let (_const, punct) = path.segments.pairs_mut().next().unwrap().into_tuple(); + *punct.unwrap() = Token![::](tilde_token.span); + } + } + + Ok(TraitBound { + paren_token: None, + modifier, + lifetimes, + path, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitBoundModifier { + fn parse(input: ParseStream) -> Result { + if input.peek(Token![?]) { + input.parse().map(TraitBoundModifier::Maybe) + } else { + Ok(TraitBoundModifier::None) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ConstParam { + fn parse(input: ParseStream) -> Result { + let mut default = None; + Ok(ConstParam { + attrs: input.call(Attribute::parse_outer)?, + const_token: input.parse()?, + ident: input.parse()?, + colon_token: input.parse()?, + ty: input.parse()?, + eq_token: { + if input.peek(Token![=]) { + let eq_token = input.parse()?; + default = Some(path::parsing::const_argument(input)?); + Some(eq_token) + } else { + None + } + }, + default, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for WhereClause { + fn parse(input: ParseStream) -> Result { + Ok(WhereClause { + where_token: input.parse()?, + predicates: { + let mut predicates = Punctuated::new(); + loop { + if input.is_empty() + || input.peek(token::Brace) + || input.peek(Token![,]) + || input.peek(Token![;]) + || input.peek(Token![:]) && !input.peek(Token![::]) + || input.peek(Token![=]) + { + break; + } + let value = input.parse()?; + predicates.push_value(value); + if !input.peek(Token![,]) { + break; + } + let punct = input.parse()?; + predicates.push_punct(punct); + } + predicates + }, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Option { + fn parse(input: ParseStream) -> Result { + if input.peek(Token![where]) { + input.parse().map(Some) + } else { + Ok(None) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for WherePredicate { + fn parse(input: ParseStream) -> Result { + if input.peek(Lifetime) && input.peek2(Token![:]) { + Ok(WherePredicate::Lifetime(PredicateLifetime { + lifetime: input.parse()?, + colon_token: input.parse()?, + bounds: { + let mut bounds = Punctuated::new(); + loop { + if input.is_empty() + || input.peek(token::Brace) + || input.peek(Token![,]) + || input.peek(Token![;]) + || input.peek(Token![:]) + || input.peek(Token![=]) + { + break; + } + let value = input.parse()?; + bounds.push_value(value); + if !input.peek(Token![+]) { + break; + } + let punct = input.parse()?; + bounds.push_punct(punct); + } + bounds + }, + })) + } else { + Ok(WherePredicate::Type(PredicateType { + lifetimes: input.parse()?, + bounded_ty: input.parse()?, + colon_token: input.parse()?, + bounds: { + let mut bounds = Punctuated::new(); + loop { + if input.is_empty() + || input.peek(token::Brace) + || input.peek(Token![,]) + || input.peek(Token![;]) + || input.peek(Token![:]) && !input.peek(Token![::]) + || input.peek(Token![=]) + { + break; + } + let value = input.parse()?; + bounds.push_value(value); + if !input.peek(Token![+]) { + break; + } + let punct = input.parse()?; + bounds.push_punct(punct); + } + bounds + }, + })) + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::attr::FilterAttrs; + use crate::print::TokensOrDefault; + #[cfg(feature = "full")] + use crate::punctuated::Pair; + use proc_macro2::TokenStream; + #[cfg(feature = "full")] + use proc_macro2::TokenTree; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Generics { + fn to_tokens(&self, tokens: &mut TokenStream) { + if self.params.is_empty() { + return; + } + + TokensOrDefault(&self.lt_token).to_tokens(tokens); + + // Print lifetimes before types and consts, regardless of their + // order in self.params. + // + // TODO: ordering rules for const parameters vs type parameters have + // not been settled yet. https://github.com/rust-lang/rust/issues/44580 + let mut trailing_or_empty = true; + for param in self.params.pairs() { + if let GenericParam::Lifetime(_) = **param.value() { + param.to_tokens(tokens); + trailing_or_empty = param.punct().is_some(); + } + } + for param in self.params.pairs() { + match **param.value() { + GenericParam::Type(_) | GenericParam::Const(_) => { + if !trailing_or_empty { + ::default().to_tokens(tokens); + trailing_or_empty = true; + } + param.to_tokens(tokens); + } + GenericParam::Lifetime(_) => {} + } + } + + TokensOrDefault(&self.gt_token).to_tokens(tokens); + } + } + + impl<'a> ToTokens for ImplGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + if self.0.params.is_empty() { + return; + } + + TokensOrDefault(&self.0.lt_token).to_tokens(tokens); + + // Print lifetimes before types and consts, regardless of their + // order in self.params. + // + // TODO: ordering rules for const parameters vs type parameters have + // not been settled yet. https://github.com/rust-lang/rust/issues/44580 + let mut trailing_or_empty = true; + for param in self.0.params.pairs() { + if let GenericParam::Lifetime(_) = **param.value() { + param.to_tokens(tokens); + trailing_or_empty = param.punct().is_some(); + } + } + for param in self.0.params.pairs() { + if let GenericParam::Lifetime(_) = **param.value() { + continue; + } + if !trailing_or_empty { + ::default().to_tokens(tokens); + trailing_or_empty = true; + } + match *param.value() { + GenericParam::Lifetime(_) => unreachable!(), + GenericParam::Type(param) => { + // Leave off the type parameter defaults + tokens.append_all(param.attrs.outer()); + param.ident.to_tokens(tokens); + if !param.bounds.is_empty() { + TokensOrDefault(¶m.colon_token).to_tokens(tokens); + param.bounds.to_tokens(tokens); + } + } + GenericParam::Const(param) => { + // Leave off the const parameter defaults + tokens.append_all(param.attrs.outer()); + param.const_token.to_tokens(tokens); + param.ident.to_tokens(tokens); + param.colon_token.to_tokens(tokens); + param.ty.to_tokens(tokens); + } + } + param.punct().to_tokens(tokens); + } + + TokensOrDefault(&self.0.gt_token).to_tokens(tokens); + } + } + + impl<'a> ToTokens for TypeGenerics<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + if self.0.params.is_empty() { + return; + } + + TokensOrDefault(&self.0.lt_token).to_tokens(tokens); + + // Print lifetimes before types and consts, regardless of their + // order in self.params. + // + // TODO: ordering rules for const parameters vs type parameters have + // not been settled yet. https://github.com/rust-lang/rust/issues/44580 + let mut trailing_or_empty = true; + for param in self.0.params.pairs() { + if let GenericParam::Lifetime(def) = *param.value() { + // Leave off the lifetime bounds and attributes + def.lifetime.to_tokens(tokens); + param.punct().to_tokens(tokens); + trailing_or_empty = param.punct().is_some(); + } + } + for param in self.0.params.pairs() { + if let GenericParam::Lifetime(_) = **param.value() { + continue; + } + if !trailing_or_empty { + ::default().to_tokens(tokens); + trailing_or_empty = true; + } + match *param.value() { + GenericParam::Lifetime(_) => unreachable!(), + GenericParam::Type(param) => { + // Leave off the type parameter defaults + param.ident.to_tokens(tokens); + } + GenericParam::Const(param) => { + // Leave off the const parameter defaults + param.ident.to_tokens(tokens); + } + } + param.punct().to_tokens(tokens); + } + + TokensOrDefault(&self.0.gt_token).to_tokens(tokens); + } + } + + impl<'a> ToTokens for Turbofish<'a> { + fn to_tokens(&self, tokens: &mut TokenStream) { + if !self.0.params.is_empty() { + ::default().to_tokens(tokens); + TypeGenerics(self.0).to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for BoundLifetimes { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.for_token.to_tokens(tokens); + self.lt_token.to_tokens(tokens); + self.lifetimes.to_tokens(tokens); + self.gt_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LifetimeDef { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.lifetime.to_tokens(tokens); + if !self.bounds.is_empty() { + TokensOrDefault(&self.colon_token).to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeParam { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.ident.to_tokens(tokens); + if !self.bounds.is_empty() { + TokensOrDefault(&self.colon_token).to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + if let Some(default) = &self.default { + #[cfg(feature = "full")] + { + if self.eq_token.is_none() { + if let Type::Verbatim(default) = default { + let mut iter = default.clone().into_iter().peekable(); + while let Some(token) = iter.next() { + if let TokenTree::Punct(q) = token { + if q.as_char() == '~' { + if let Some(TokenTree::Ident(c)) = iter.peek() { + if c == "const" { + if self.bounds.is_empty() { + TokensOrDefault(&self.colon_token) + .to_tokens(tokens); + } + return default.to_tokens(tokens); + } + } + } + } + } + } + } + } + TokensOrDefault(&self.eq_token).to_tokens(tokens); + default.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitBound { + fn to_tokens(&self, tokens: &mut TokenStream) { + let to_tokens = |tokens: &mut TokenStream| { + #[cfg(feature = "full")] + let skip = match self.path.segments.pairs().next() { + Some(Pair::Punctuated(t, p)) if t.ident == "const" => { + Token![~](p.spans[0]).to_tokens(tokens); + t.to_tokens(tokens); + 1 + } + _ => 0, + }; + self.modifier.to_tokens(tokens); + self.lifetimes.to_tokens(tokens); + #[cfg(feature = "full")] + { + self.path.leading_colon.to_tokens(tokens); + tokens.append_all(self.path.segments.pairs().skip(skip)); + } + #[cfg(not(feature = "full"))] + { + self.path.to_tokens(tokens); + } + }; + match &self.paren_token { + Some(paren) => paren.surround(tokens, to_tokens), + None => to_tokens(tokens), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitBoundModifier { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + TraitBoundModifier::None => {} + TraitBoundModifier::Maybe(t) => t.to_tokens(tokens), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ConstParam { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.const_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + if let Some(default) = &self.default { + TokensOrDefault(&self.eq_token).to_tokens(tokens); + default.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for WhereClause { + fn to_tokens(&self, tokens: &mut TokenStream) { + if !self.predicates.is_empty() { + self.where_token.to_tokens(tokens); + self.predicates.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PredicateType { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.lifetimes.to_tokens(tokens); + self.bounded_ty.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PredicateLifetime { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.lifetime.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PredicateEq { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.lhs_ty.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.rhs_ty.to_tokens(tokens); + } + } +} diff --git a/rust/syn/group.rs b/rust/syn/group.rs new file mode 100644 index 00000000000000..b02e79db2c2082 --- /dev/null +++ b/rust/syn/group.rs @@ -0,0 +1,284 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::error::Result; +use crate::parse::ParseBuffer; +use crate::token; +use proc_macro2::{Delimiter, Span}; + +// Not public API. +#[doc(hidden)] +pub struct Parens<'a> { + pub token: token::Paren, + pub content: ParseBuffer<'a>, +} + +// Not public API. +#[doc(hidden)] +pub struct Braces<'a> { + pub token: token::Brace, + pub content: ParseBuffer<'a>, +} + +// Not public API. +#[doc(hidden)] +pub struct Brackets<'a> { + pub token: token::Bracket, + pub content: ParseBuffer<'a>, +} + +// Not public API. +#[cfg(any(feature = "full", feature = "derive"))] +#[doc(hidden)] +pub struct Group<'a> { + pub token: token::Group, + pub content: ParseBuffer<'a>, +} + +// Not public API. +#[doc(hidden)] +pub fn parse_parens<'a>(input: &ParseBuffer<'a>) -> Result> { + parse_delimited(input, Delimiter::Parenthesis).map(|(span, content)| Parens { + token: token::Paren(span), + content, + }) +} + +// Not public API. +#[doc(hidden)] +pub fn parse_braces<'a>(input: &ParseBuffer<'a>) -> Result> { + parse_delimited(input, Delimiter::Brace).map(|(span, content)| Braces { + token: token::Brace(span), + content, + }) +} + +// Not public API. +#[doc(hidden)] +pub fn parse_brackets<'a>(input: &ParseBuffer<'a>) -> Result> { + parse_delimited(input, Delimiter::Bracket).map(|(span, content)| Brackets { + token: token::Bracket(span), + content, + }) +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn parse_group<'a>(input: &ParseBuffer<'a>) -> Result> { + parse_delimited(input, Delimiter::None).map(|(span, content)| Group { + token: token::Group(span), + content, + }) +} + +fn parse_delimited<'a>( + input: &ParseBuffer<'a>, + delimiter: Delimiter, +) -> Result<(Span, ParseBuffer<'a>)> { + input.step(|cursor| { + if let Some((content, span, rest)) = cursor.group(delimiter) { + let scope = crate::buffer::close_span_of_group(*cursor); + let nested = crate::parse::advance_step_cursor(cursor, content); + let unexpected = crate::parse::get_unexpected(input); + let content = crate::parse::new_parse_buffer(scope, nested, unexpected); + Ok(((span, content), rest)) + } else { + let message = match delimiter { + Delimiter::Parenthesis => "expected parentheses", + Delimiter::Brace => "expected curly braces", + Delimiter::Bracket => "expected square brackets", + Delimiter::None => "expected invisible group", + }; + Err(cursor.error(message)) + } + }) +} + +/// Parse a set of parentheses and expose their content to subsequent parsers. +/// +/// # Example +/// +/// ``` +/// # use quote::quote; +/// # +/// use syn::{parenthesized, token, Ident, Result, Token, Type}; +/// use syn::parse::{Parse, ParseStream}; +/// use syn::punctuated::Punctuated; +/// +/// // Parse a simplified tuple struct syntax like: +/// // +/// // struct S(A, B); +/// struct TupleStruct { +/// struct_token: Token![struct], +/// ident: Ident, +/// paren_token: token::Paren, +/// fields: Punctuated, +/// semi_token: Token![;], +/// } +/// +/// impl Parse for TupleStruct { +/// fn parse(input: ParseStream) -> Result { +/// let content; +/// Ok(TupleStruct { +/// struct_token: input.parse()?, +/// ident: input.parse()?, +/// paren_token: parenthesized!(content in input), +/// fields: content.parse_terminated(Type::parse)?, +/// semi_token: input.parse()?, +/// }) +/// } +/// } +/// # +/// # fn main() { +/// # let input = quote! { +/// # struct S(A, B); +/// # }; +/// # syn::parse2::(input).unwrap(); +/// # } +/// ``` +#[macro_export] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +macro_rules! parenthesized { + ($content:ident in $cursor:expr) => { + match $crate::__private::parse_parens(&$cursor) { + $crate::__private::Ok(parens) => { + $content = parens.content; + parens.token + } + $crate::__private::Err(error) => { + return $crate::__private::Err(error); + } + } + }; +} + +/// Parse a set of curly braces and expose their content to subsequent parsers. +/// +/// # Example +/// +/// ``` +/// # use quote::quote; +/// # +/// use syn::{braced, token, Ident, Result, Token, Type}; +/// use syn::parse::{Parse, ParseStream}; +/// use syn::punctuated::Punctuated; +/// +/// // Parse a simplified struct syntax like: +/// // +/// // struct S { +/// // a: A, +/// // b: B, +/// // } +/// struct Struct { +/// struct_token: Token![struct], +/// ident: Ident, +/// brace_token: token::Brace, +/// fields: Punctuated, +/// } +/// +/// struct Field { +/// name: Ident, +/// colon_token: Token![:], +/// ty: Type, +/// } +/// +/// impl Parse for Struct { +/// fn parse(input: ParseStream) -> Result { +/// let content; +/// Ok(Struct { +/// struct_token: input.parse()?, +/// ident: input.parse()?, +/// brace_token: braced!(content in input), +/// fields: content.parse_terminated(Field::parse)?, +/// }) +/// } +/// } +/// +/// impl Parse for Field { +/// fn parse(input: ParseStream) -> Result { +/// Ok(Field { +/// name: input.parse()?, +/// colon_token: input.parse()?, +/// ty: input.parse()?, +/// }) +/// } +/// } +/// # +/// # fn main() { +/// # let input = quote! { +/// # struct S { +/// # a: A, +/// # b: B, +/// # } +/// # }; +/// # syn::parse2::(input).unwrap(); +/// # } +/// ``` +#[macro_export] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +macro_rules! braced { + ($content:ident in $cursor:expr) => { + match $crate::__private::parse_braces(&$cursor) { + $crate::__private::Ok(braces) => { + $content = braces.content; + braces.token + } + $crate::__private::Err(error) => { + return $crate::__private::Err(error); + } + } + }; +} + +/// Parse a set of square brackets and expose their content to subsequent +/// parsers. +/// +/// # Example +/// +/// ``` +/// # use quote::quote; +/// # +/// use proc_macro2::TokenStream; +/// use syn::{bracketed, token, Result, Token}; +/// use syn::parse::{Parse, ParseStream}; +/// +/// // Parse an outer attribute like: +/// // +/// // #[repr(C, packed)] +/// struct OuterAttribute { +/// pound_token: Token![#], +/// bracket_token: token::Bracket, +/// content: TokenStream, +/// } +/// +/// impl Parse for OuterAttribute { +/// fn parse(input: ParseStream) -> Result { +/// let content; +/// Ok(OuterAttribute { +/// pound_token: input.parse()?, +/// bracket_token: bracketed!(content in input), +/// content: content.parse()?, +/// }) +/// } +/// } +/// # +/// # fn main() { +/// # let input = quote! { +/// # #[repr(C, packed)] +/// # }; +/// # syn::parse2::(input).unwrap(); +/// # } +/// ``` +#[macro_export] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +macro_rules! bracketed { + ($content:ident in $cursor:expr) => { + match $crate::__private::parse_brackets(&$cursor) { + $crate::__private::Ok(brackets) => { + $content = brackets.content; + brackets.token + } + $crate::__private::Err(error) => { + return $crate::__private::Err(error); + } + } + }; +} diff --git a/rust/syn/ident.rs b/rust/syn/ident.rs new file mode 100644 index 00000000000000..9ae2f0cbd83e94 --- /dev/null +++ b/rust/syn/ident.rs @@ -0,0 +1,103 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg(feature = "parsing")] +use crate::buffer::Cursor; +#[cfg(feature = "parsing")] +use crate::lookahead; +#[cfg(feature = "parsing")] +use crate::parse::{Parse, ParseStream, Result}; +#[cfg(feature = "parsing")] +use crate::token::Token; + +pub use proc_macro2::Ident; + +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[allow(non_snake_case)] +pub fn Ident(marker: lookahead::TokenMarker) -> Ident { + match marker {} +} + +#[cfg(feature = "parsing")] +fn accept_as_ident(ident: &Ident) -> bool { + match ident.to_string().as_str() { + "_" | + // Based on https://doc.rust-lang.org/grammar.html#keywords + // and https://github.com/rust-lang/rfcs/blob/master/text/2421-unreservations-2018.md + // and https://github.com/rust-lang/rfcs/blob/master/text/2420-unreserve-proc.md + "abstract" | "as" | "become" | "box" | "break" | "const" | "continue" | + "crate" | "do" | "else" | "enum" | "extern" | "false" | "final" | "fn" | + "for" | "if" | "impl" | "in" | "let" | "loop" | "macro" | "match" | + "mod" | "move" | "mut" | "override" | "priv" | "pub" | "ref" | + "return" | "Self" | "self" | "static" | "struct" | "super" | "trait" | + "true" | "type" | "typeof" | "unsafe" | "unsized" | "use" | "virtual" | + "where" | "while" | "yield" => false, + _ => true, + } +} + +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Ident { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| { + if let Some((ident, rest)) = cursor.ident() { + if accept_as_ident(&ident) { + return Ok((ident, rest)); + } + } + Err(cursor.error("expected identifier")) + }) + } +} + +#[cfg(feature = "parsing")] +impl Token for Ident { + fn peek(cursor: Cursor) -> bool { + if let Some((ident, _rest)) = cursor.ident() { + accept_as_ident(&ident) + } else { + false + } + } + + fn display() -> &'static str { + "identifier" + } +} + +macro_rules! ident_from_token { + ($token:ident) => { + impl From for Ident { + fn from(token: Token![$token]) -> Ident { + Ident::new(stringify!($token), token.span) + } + } + }; +} + +ident_from_token!(self); +ident_from_token!(Self); +ident_from_token!(super); +ident_from_token!(crate); +ident_from_token!(extern); + +impl From for Ident { + fn from(token: Token![_]) -> Ident { + Ident::new("_", token.span) + } +} + +pub fn xid_ok(symbol: &str) -> bool { + let mut chars = symbol.chars(); + let first = chars.next().unwrap(); + if !(first == '_' || first.is_ascii_alphabetic()) { + return false; + } + for ch in chars { + if !(ch == '_' || ch.is_ascii_alphanumeric()) { + return false; + } + } + true +} diff --git a/rust/syn/item.rs b/rust/syn/item.rs new file mode 100644 index 00000000000000..b256d94a25e2b8 --- /dev/null +++ b/rust/syn/item.rs @@ -0,0 +1,3315 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::derive::{Data, DataEnum, DataStruct, DataUnion, DeriveInput}; +use crate::punctuated::Punctuated; +use proc_macro2::TokenStream; + +#[cfg(feature = "parsing")] +use std::mem; + +ast_enum_of_structs! { + /// Things that can appear directly inside of a module or scope. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum Item { + /// A constant item: `const MAX: u16 = 65535`. + Const(ItemConst), + + /// An enum definition: `enum Foo { A(A), B(B) }`. + Enum(ItemEnum), + + /// An `extern crate` item: `extern crate serde`. + ExternCrate(ItemExternCrate), + + /// A free-standing function: `fn process(n: usize) -> Result<()> { ... + /// }`. + Fn(ItemFn), + + /// A block of foreign items: `extern "C" { ... }`. + ForeignMod(ItemForeignMod), + + /// An impl block providing trait or associated items: `impl Trait + /// for Data { ... }`. + Impl(ItemImpl), + + /// A macro invocation, which includes `macro_rules!` definitions. + Macro(ItemMacro), + + /// A 2.0-style declarative macro introduced by the `macro` keyword. + Macro2(ItemMacro2), + + /// A module or module declaration: `mod m` or `mod m { ... }`. + Mod(ItemMod), + + /// A static item: `static BIKE: Shed = Shed(42)`. + Static(ItemStatic), + + /// A struct definition: `struct Foo { x: A }`. + Struct(ItemStruct), + + /// A trait definition: `pub trait Iterator { ... }`. + Trait(ItemTrait), + + /// A trait alias: `pub trait SharableIterator = Iterator + Sync`. + TraitAlias(ItemTraitAlias), + + /// A type alias: `type Result = std::result::Result`. + Type(ItemType), + + /// A union definition: `union Foo { x: A, y: B }`. + Union(ItemUnion), + + /// A use declaration: `use std::collections::HashMap`. + Use(ItemUse), + + /// Tokens forming an item not interpreted by Syn. + Verbatim(TokenStream), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match item { + // Item::Const(item) => {...} + // Item::Enum(item) => {...} + // ... + // Item::Verbatim(item) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// A constant item: `const MAX: u16 = 65535`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemConst { + pub attrs: Vec, + pub vis: Visibility, + pub const_token: Token![const], + pub ident: Ident, + pub colon_token: Token![:], + pub ty: Box, + pub eq_token: Token![=], + pub expr: Box, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// An enum definition: `enum Foo { A(A), B(B) }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemEnum { + pub attrs: Vec, + pub vis: Visibility, + pub enum_token: Token![enum], + pub ident: Ident, + pub generics: Generics, + pub brace_token: token::Brace, + pub variants: Punctuated, + } +} + +ast_struct! { + /// An `extern crate` item: `extern crate serde`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemExternCrate { + pub attrs: Vec, + pub vis: Visibility, + pub extern_token: Token![extern], + pub crate_token: Token![crate], + pub ident: Ident, + pub rename: Option<(Token![as], Ident)>, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A free-standing function: `fn process(n: usize) -> Result<()> { ... + /// }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemFn { + pub attrs: Vec, + pub vis: Visibility, + pub sig: Signature, + pub block: Box, + } +} + +ast_struct! { + /// A block of foreign items: `extern "C" { ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemForeignMod { + pub attrs: Vec, + pub abi: Abi, + pub brace_token: token::Brace, + pub items: Vec, + } +} + +ast_struct! { + /// An impl block providing trait or associated items: `impl Trait + /// for Data { ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemImpl { + pub attrs: Vec, + pub defaultness: Option, + pub unsafety: Option, + pub impl_token: Token![impl], + pub generics: Generics, + /// Trait this impl implements. + pub trait_: Option<(Option, Path, Token![for])>, + /// The Self type of the impl. + pub self_ty: Box, + pub brace_token: token::Brace, + pub items: Vec, + } +} + +ast_struct! { + /// A macro invocation, which includes `macro_rules!` definitions. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemMacro { + pub attrs: Vec, + /// The `example` in `macro_rules! example { ... }`. + pub ident: Option, + pub mac: Macro, + pub semi_token: Option, + } +} + +ast_struct! { + /// A 2.0-style declarative macro introduced by the `macro` keyword. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemMacro2 { + pub attrs: Vec, + pub vis: Visibility, + pub macro_token: Token![macro], + pub ident: Ident, + pub rules: TokenStream, + } +} + +ast_struct! { + /// A module or module declaration: `mod m` or `mod m { ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemMod { + pub attrs: Vec, + pub vis: Visibility, + pub mod_token: Token![mod], + pub ident: Ident, + pub content: Option<(token::Brace, Vec)>, + pub semi: Option, + } +} + +ast_struct! { + /// A static item: `static BIKE: Shed = Shed(42)`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemStatic { + pub attrs: Vec, + pub vis: Visibility, + pub static_token: Token![static], + pub mutability: Option, + pub ident: Ident, + pub colon_token: Token![:], + pub ty: Box, + pub eq_token: Token![=], + pub expr: Box, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A struct definition: `struct Foo { x: A }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemStruct { + pub attrs: Vec, + pub vis: Visibility, + pub struct_token: Token![struct], + pub ident: Ident, + pub generics: Generics, + pub fields: Fields, + pub semi_token: Option, + } +} + +ast_struct! { + /// A trait definition: `pub trait Iterator { ... }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemTrait { + pub attrs: Vec, + pub vis: Visibility, + pub unsafety: Option, + pub auto_token: Option, + pub trait_token: Token![trait], + pub ident: Ident, + pub generics: Generics, + pub colon_token: Option, + pub supertraits: Punctuated, + pub brace_token: token::Brace, + pub items: Vec, + } +} + +ast_struct! { + /// A trait alias: `pub trait SharableIterator = Iterator + Sync`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemTraitAlias { + pub attrs: Vec, + pub vis: Visibility, + pub trait_token: Token![trait], + pub ident: Ident, + pub generics: Generics, + pub eq_token: Token![=], + pub bounds: Punctuated, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A type alias: `type Result = std::result::Result`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemType { + pub attrs: Vec, + pub vis: Visibility, + pub type_token: Token![type], + pub ident: Ident, + pub generics: Generics, + pub eq_token: Token![=], + pub ty: Box, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A union definition: `union Foo { x: A, y: B }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemUnion { + pub attrs: Vec, + pub vis: Visibility, + pub union_token: Token![union], + pub ident: Ident, + pub generics: Generics, + pub fields: FieldsNamed, + } +} + +ast_struct! { + /// A use declaration: `use std::collections::HashMap`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ItemUse { + pub attrs: Vec, + pub vis: Visibility, + pub use_token: Token![use], + pub leading_colon: Option, + pub tree: UseTree, + pub semi_token: Token![;], + } +} + +impl Item { + #[cfg(feature = "parsing")] + pub(crate) fn replace_attrs(&mut self, new: Vec) -> Vec { + match self { + Item::ExternCrate(ItemExternCrate { attrs, .. }) + | Item::Use(ItemUse { attrs, .. }) + | Item::Static(ItemStatic { attrs, .. }) + | Item::Const(ItemConst { attrs, .. }) + | Item::Fn(ItemFn { attrs, .. }) + | Item::Mod(ItemMod { attrs, .. }) + | Item::ForeignMod(ItemForeignMod { attrs, .. }) + | Item::Type(ItemType { attrs, .. }) + | Item::Struct(ItemStruct { attrs, .. }) + | Item::Enum(ItemEnum { attrs, .. }) + | Item::Union(ItemUnion { attrs, .. }) + | Item::Trait(ItemTrait { attrs, .. }) + | Item::TraitAlias(ItemTraitAlias { attrs, .. }) + | Item::Impl(ItemImpl { attrs, .. }) + | Item::Macro(ItemMacro { attrs, .. }) + | Item::Macro2(ItemMacro2 { attrs, .. }) => mem::replace(attrs, new), + Item::Verbatim(_) => Vec::new(), + + #[cfg(syn_no_non_exhaustive)] + _ => unreachable!(), + } + } +} + +impl From for Item { + fn from(input: DeriveInput) -> Item { + match input.data { + Data::Struct(data) => Item::Struct(ItemStruct { + attrs: input.attrs, + vis: input.vis, + struct_token: data.struct_token, + ident: input.ident, + generics: input.generics, + fields: data.fields, + semi_token: data.semi_token, + }), + Data::Enum(data) => Item::Enum(ItemEnum { + attrs: input.attrs, + vis: input.vis, + enum_token: data.enum_token, + ident: input.ident, + generics: input.generics, + brace_token: data.brace_token, + variants: data.variants, + }), + Data::Union(data) => Item::Union(ItemUnion { + attrs: input.attrs, + vis: input.vis, + union_token: data.union_token, + ident: input.ident, + generics: input.generics, + fields: data.fields, + }), + } + } +} + +impl From for DeriveInput { + fn from(input: ItemStruct) -> DeriveInput { + DeriveInput { + attrs: input.attrs, + vis: input.vis, + ident: input.ident, + generics: input.generics, + data: Data::Struct(DataStruct { + struct_token: input.struct_token, + fields: input.fields, + semi_token: input.semi_token, + }), + } + } +} + +impl From for DeriveInput { + fn from(input: ItemEnum) -> DeriveInput { + DeriveInput { + attrs: input.attrs, + vis: input.vis, + ident: input.ident, + generics: input.generics, + data: Data::Enum(DataEnum { + enum_token: input.enum_token, + brace_token: input.brace_token, + variants: input.variants, + }), + } + } +} + +impl From for DeriveInput { + fn from(input: ItemUnion) -> DeriveInput { + DeriveInput { + attrs: input.attrs, + vis: input.vis, + ident: input.ident, + generics: input.generics, + data: Data::Union(DataUnion { + union_token: input.union_token, + fields: input.fields, + }), + } + } +} + +ast_enum_of_structs! { + /// A suffix of an import tree in a `use` item: `Type as Renamed` or `*`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub enum UseTree { + /// A path prefix of imports in a `use` item: `std::...`. + Path(UsePath), + + /// An identifier imported by a `use` item: `HashMap`. + Name(UseName), + + /// An renamed identifier imported by a `use` item: `HashMap as Map`. + Rename(UseRename), + + /// A glob import in a `use` item: `*`. + Glob(UseGlob), + + /// A braced group of imports in a `use` item: `{A, B, C}`. + Group(UseGroup), + } +} + +ast_struct! { + /// A path prefix of imports in a `use` item: `std::...`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct UsePath { + pub ident: Ident, + pub colon2_token: Token![::], + pub tree: Box, + } +} + +ast_struct! { + /// An identifier imported by a `use` item: `HashMap`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct UseName { + pub ident: Ident, + } +} + +ast_struct! { + /// An renamed identifier imported by a `use` item: `HashMap as Map`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct UseRename { + pub ident: Ident, + pub as_token: Token![as], + pub rename: Ident, + } +} + +ast_struct! { + /// A glob import in a `use` item: `*`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct UseGlob { + pub star_token: Token![*], + } +} + +ast_struct! { + /// A braced group of imports in a `use` item: `{A, B, C}`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct UseGroup { + pub brace_token: token::Brace, + pub items: Punctuated, + } +} + +ast_enum_of_structs! { + /// An item within an `extern` block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum ForeignItem { + /// A foreign function in an `extern` block. + Fn(ForeignItemFn), + + /// A foreign static item in an `extern` block: `static ext: u8`. + Static(ForeignItemStatic), + + /// A foreign type in an `extern` block: `type void`. + Type(ForeignItemType), + + /// A macro invocation within an extern block. + Macro(ForeignItemMacro), + + /// Tokens in an `extern` block not interpreted by Syn. + Verbatim(TokenStream), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match item { + // ForeignItem::Fn(item) => {...} + // ForeignItem::Static(item) => {...} + // ... + // ForeignItem::Verbatim(item) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// A foreign function in an `extern` block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ForeignItemFn { + pub attrs: Vec, + pub vis: Visibility, + pub sig: Signature, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A foreign static item in an `extern` block: `static ext: u8`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ForeignItemStatic { + pub attrs: Vec, + pub vis: Visibility, + pub static_token: Token![static], + pub mutability: Option, + pub ident: Ident, + pub colon_token: Token![:], + pub ty: Box, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A foreign type in an `extern` block: `type void`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ForeignItemType { + pub attrs: Vec, + pub vis: Visibility, + pub type_token: Token![type], + pub ident: Ident, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A macro invocation within an extern block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ForeignItemMacro { + pub attrs: Vec, + pub mac: Macro, + pub semi_token: Option, + } +} + +ast_enum_of_structs! { + /// An item declaration within the definition of a trait. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum TraitItem { + /// An associated constant within the definition of a trait. + Const(TraitItemConst), + + /// A trait method within the definition of a trait. + Method(TraitItemMethod), + + /// An associated type within the definition of a trait. + Type(TraitItemType), + + /// A macro invocation within the definition of a trait. + Macro(TraitItemMacro), + + /// Tokens within the definition of a trait not interpreted by Syn. + Verbatim(TokenStream), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match item { + // TraitItem::Const(item) => {...} + // TraitItem::Method(item) => {...} + // ... + // TraitItem::Verbatim(item) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// An associated constant within the definition of a trait. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct TraitItemConst { + pub attrs: Vec, + pub const_token: Token![const], + pub ident: Ident, + pub colon_token: Token![:], + pub ty: Type, + pub default: Option<(Token![=], Expr)>, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A trait method within the definition of a trait. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct TraitItemMethod { + pub attrs: Vec, + pub sig: Signature, + pub default: Option, + pub semi_token: Option, + } +} + +ast_struct! { + /// An associated type within the definition of a trait. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct TraitItemType { + pub attrs: Vec, + pub type_token: Token![type], + pub ident: Ident, + pub generics: Generics, + pub colon_token: Option, + pub bounds: Punctuated, + pub default: Option<(Token![=], Type)>, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A macro invocation within the definition of a trait. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct TraitItemMacro { + pub attrs: Vec, + pub mac: Macro, + pub semi_token: Option, + } +} + +ast_enum_of_structs! { + /// An item within an impl block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum ImplItem { + /// An associated constant within an impl block. + Const(ImplItemConst), + + /// A method within an impl block. + Method(ImplItemMethod), + + /// An associated type within an impl block. + Type(ImplItemType), + + /// A macro invocation within an impl block. + Macro(ImplItemMacro), + + /// Tokens within an impl block not interpreted by Syn. + Verbatim(TokenStream), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match item { + // ImplItem::Const(item) => {...} + // ImplItem::Method(item) => {...} + // ... + // ImplItem::Verbatim(item) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// An associated constant within an impl block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ImplItemConst { + pub attrs: Vec, + pub vis: Visibility, + pub defaultness: Option, + pub const_token: Token![const], + pub ident: Ident, + pub colon_token: Token![:], + pub ty: Type, + pub eq_token: Token![=], + pub expr: Expr, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A method within an impl block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ImplItemMethod { + pub attrs: Vec, + pub vis: Visibility, + pub defaultness: Option, + pub sig: Signature, + pub block: Block, + } +} + +ast_struct! { + /// An associated type within an impl block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ImplItemType { + pub attrs: Vec, + pub vis: Visibility, + pub defaultness: Option, + pub type_token: Token![type], + pub ident: Ident, + pub generics: Generics, + pub eq_token: Token![=], + pub ty: Type, + pub semi_token: Token![;], + } +} + +ast_struct! { + /// A macro invocation within an impl block. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct ImplItemMacro { + pub attrs: Vec, + pub mac: Macro, + pub semi_token: Option, + } +} + +ast_struct! { + /// A function signature in a trait or implementation: `unsafe fn + /// initialize(&self)`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct Signature { + pub constness: Option, + pub asyncness: Option, + pub unsafety: Option, + pub abi: Option, + pub fn_token: Token![fn], + pub ident: Ident, + pub generics: Generics, + pub paren_token: token::Paren, + pub inputs: Punctuated, + pub variadic: Option, + pub output: ReturnType, + } +} + +impl Signature { + /// A method's `self` receiver, such as `&self` or `self: Box`. + pub fn receiver(&self) -> Option<&FnArg> { + let arg = self.inputs.first()?; + match arg { + FnArg::Receiver(_) => Some(arg), + FnArg::Typed(PatType { pat, .. }) => { + if let Pat::Ident(PatIdent { ident, .. }) = &**pat { + if ident == "self" { + return Some(arg); + } + } + None + } + } + } +} + +ast_enum_of_structs! { + /// An argument in a function signature: the `n: usize` in `fn f(n: usize)`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub enum FnArg { + /// The `self` argument of an associated method, whether taken by value + /// or by reference. + /// + /// Note that `self` receivers with a specified type, such as `self: + /// Box`, are parsed as a `FnArg::Typed`. + Receiver(Receiver), + + /// A function argument accepted by pattern and type. + Typed(PatType), + } +} + +ast_struct! { + /// The `self` argument of an associated method, whether taken by value + /// or by reference. + /// + /// Note that `self` receivers with a specified type, such as `self: + /// Box`, are parsed as a `FnArg::Typed`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct Receiver { + pub attrs: Vec, + pub reference: Option<(Token![&], Option)>, + pub mutability: Option, + pub self_token: Token![self], + } +} + +impl Receiver { + pub fn lifetime(&self) -> Option<&Lifetime> { + self.reference.as_ref()?.1.as_ref() + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::discouraged::Speculative; + use crate::parse::{Parse, ParseBuffer, ParseStream, Result}; + use crate::token::Brace; + use proc_macro2::{Delimiter, Group, Punct, Spacing, TokenTree}; + use std::iter::{self, FromIterator}; + + crate::custom_keyword!(macro_rules); + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Item { + fn parse(input: ParseStream) -> Result { + let begin = input.fork(); + let mut attrs = input.call(Attribute::parse_outer)?; + let ahead = input.fork(); + let vis: Visibility = ahead.parse()?; + + let lookahead = ahead.lookahead1(); + let mut item = if lookahead.peek(Token![fn]) || peek_signature(&ahead) { + let vis: Visibility = input.parse()?; + let sig: Signature = input.parse()?; + if input.peek(Token![;]) { + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + parse_rest_of_fn(input, Vec::new(), vis, sig).map(Item::Fn) + } + } else if lookahead.peek(Token![extern]) { + ahead.parse::()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(Token![crate]) { + input.parse().map(Item::ExternCrate) + } else if lookahead.peek(token::Brace) { + input.parse().map(Item::ForeignMod) + } else if lookahead.peek(LitStr) { + ahead.parse::()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(token::Brace) { + input.parse().map(Item::ForeignMod) + } else { + Err(lookahead.error()) + } + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(Token![use]) { + input.parse().map(Item::Use) + } else if lookahead.peek(Token![static]) { + let vis = input.parse()?; + let static_token = input.parse()?; + let mutability = input.parse()?; + let ident = input.parse()?; + if input.peek(Token![=]) { + input.parse::()?; + input.parse::()?; + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + let colon_token = input.parse()?; + let ty = input.parse()?; + if input.peek(Token![;]) { + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + Ok(Item::Static(ItemStatic { + attrs: Vec::new(), + vis, + static_token, + mutability, + ident, + colon_token, + ty, + eq_token: input.parse()?, + expr: input.parse()?, + semi_token: input.parse()?, + })) + } + } + } else if lookahead.peek(Token![const]) { + ahead.parse::()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + let vis = input.parse()?; + let const_token = input.parse()?; + let ident = { + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.call(Ident::parse_any)? + } else { + return Err(lookahead.error()); + } + }; + let colon_token = input.parse()?; + let ty = input.parse()?; + if input.peek(Token![;]) { + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + Ok(Item::Const(ItemConst { + attrs: Vec::new(), + vis, + const_token, + ident, + colon_token, + ty, + eq_token: input.parse()?, + expr: input.parse()?, + semi_token: input.parse()?, + })) + } + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(Token![unsafe]) { + ahead.parse::()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(Token![trait]) + || lookahead.peek(Token![auto]) && ahead.peek2(Token![trait]) + { + input.parse().map(Item::Trait) + } else if lookahead.peek(Token![impl]) { + let allow_verbatim_impl = true; + if let Some(item) = parse_impl(input, allow_verbatim_impl)? { + Ok(Item::Impl(item)) + } else { + Ok(Item::Verbatim(verbatim::between(begin, input))) + } + } else if lookahead.peek(Token![extern]) { + input.parse::()?; + input.parse::()?; + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else if lookahead.peek(Token![mod]) { + input.parse::()?; + input.parse::()?; + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(Token![mod]) { + input.parse().map(Item::Mod) + } else if lookahead.peek(Token![type]) { + parse_item_type(begin, input) + } else if lookahead.peek(Token![struct]) { + input.parse().map(Item::Struct) + } else if lookahead.peek(Token![enum]) { + input.parse().map(Item::Enum) + } else if lookahead.peek(Token![union]) && ahead.peek2(Ident) { + input.parse().map(Item::Union) + } else if lookahead.peek(Token![trait]) { + input.call(parse_trait_or_trait_alias) + } else if lookahead.peek(Token![auto]) && ahead.peek2(Token![trait]) { + input.parse().map(Item::Trait) + } else if lookahead.peek(Token![impl]) + || lookahead.peek(Token![default]) && !ahead.peek2(Token![!]) + { + let allow_verbatim_impl = true; + if let Some(item) = parse_impl(input, allow_verbatim_impl)? { + Ok(Item::Impl(item)) + } else { + Ok(Item::Verbatim(verbatim::between(begin, input))) + } + } else if lookahead.peek(Token![macro]) { + input.parse().map(Item::Macro2) + } else if vis.is_inherited() + && (lookahead.peek(Ident) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + || lookahead.peek(Token![::])) + { + input.parse().map(Item::Macro) + } else if ahead.peek(macro_rules) { + input.advance_to(&ahead); + input.parse::()?; + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + Err(lookahead.error()) + }?; + + attrs.extend(item.replace_attrs(Vec::new())); + item.replace_attrs(attrs); + Ok(item) + } + } + + struct FlexibleItemType { + vis: Visibility, + defaultness: Option, + type_token: Token![type], + ident: Ident, + generics: Generics, + colon_token: Option, + bounds: Punctuated, + ty: Option<(Token![=], Type)>, + semi_token: Token![;], + } + + enum WhereClauseLocation { + // type Ty where T: 'static = T; + BeforeEq, + // type Ty = T where T: 'static; + #[allow(dead_code)] + AfterEq, + // TODO: goes away once the migration period on rust-lang/rust#89122 is over + Both, + } + + impl FlexibleItemType { + fn parse(input: ParseStream, where_clause_location: WhereClauseLocation) -> Result { + let vis: Visibility = input.parse()?; + let defaultness: Option = input.parse()?; + let type_token: Token![type] = input.parse()?; + let ident: Ident = input.parse()?; + let mut generics: Generics = input.parse()?; + let colon_token: Option = input.parse()?; + + let mut bounds = Punctuated::new(); + if colon_token.is_some() { + loop { + if input.peek(Token![where]) || input.peek(Token![=]) || input.peek(Token![;]) { + break; + } + bounds.push_value(input.parse::()?); + if input.peek(Token![where]) || input.peek(Token![=]) || input.peek(Token![;]) { + break; + } + bounds.push_punct(input.parse::()?); + } + } + + match where_clause_location { + WhereClauseLocation::BeforeEq | WhereClauseLocation::Both => { + generics.where_clause = input.parse()?; + } + _ => {} + } + + let ty = if let Some(eq_token) = input.parse()? { + Some((eq_token, input.parse::()?)) + } else { + None + }; + + match where_clause_location { + WhereClauseLocation::AfterEq | WhereClauseLocation::Both + if generics.where_clause.is_none() => + { + generics.where_clause = input.parse()?; + } + _ => {} + } + + let semi_token: Token![;] = input.parse()?; + + Ok(FlexibleItemType { + vis, + defaultness, + type_token, + ident, + generics, + colon_token, + bounds, + ty, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemMacro { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let path = input.call(Path::parse_mod_style)?; + let bang_token: Token![!] = input.parse()?; + let ident: Option = input.parse()?; + let (delimiter, tokens) = input.call(mac::parse_delimiter)?; + let semi_token: Option = if !delimiter.is_brace() { + Some(input.parse()?) + } else { + None + }; + Ok(ItemMacro { + attrs, + ident, + mac: Macro { + path, + bang_token, + delimiter, + tokens, + }, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemMacro2 { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let macro_token: Token![macro] = input.parse()?; + let ident: Ident = input.parse()?; + let mut rules = TokenStream::new(); + + let mut lookahead = input.lookahead1(); + if lookahead.peek(token::Paren) { + let paren_content; + let paren_token = parenthesized!(paren_content in input); + let args: TokenStream = paren_content.parse()?; + let mut args = Group::new(Delimiter::Parenthesis, args); + args.set_span(paren_token.span); + rules.extend(iter::once(TokenTree::Group(args))); + lookahead = input.lookahead1(); + } + + if lookahead.peek(token::Brace) { + let brace_content; + let brace_token = braced!(brace_content in input); + let body: TokenStream = brace_content.parse()?; + let mut body = Group::new(Delimiter::Brace, body); + body.set_span(brace_token.span); + rules.extend(iter::once(TokenTree::Group(body))); + } else { + return Err(lookahead.error()); + } + + Ok(ItemMacro2 { + attrs, + vis, + macro_token, + ident, + rules, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemExternCrate { + fn parse(input: ParseStream) -> Result { + Ok(ItemExternCrate { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + extern_token: input.parse()?, + crate_token: input.parse()?, + ident: { + if input.peek(Token![self]) { + input.call(Ident::parse_any)? + } else { + input.parse()? + } + }, + rename: { + if input.peek(Token![as]) { + let as_token: Token![as] = input.parse()?; + let rename: Ident = if input.peek(Token![_]) { + Ident::from(input.parse::()?) + } else { + input.parse()? + }; + Some((as_token, rename)) + } else { + None + } + }, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemUse { + fn parse(input: ParseStream) -> Result { + Ok(ItemUse { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + use_token: input.parse()?, + leading_colon: input.parse()?, + tree: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for UseTree { + fn parse(input: ParseStream) -> Result { + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + { + let ident = input.call(Ident::parse_any)?; + if input.peek(Token![::]) { + Ok(UseTree::Path(UsePath { + ident, + colon2_token: input.parse()?, + tree: Box::new(input.parse()?), + })) + } else if input.peek(Token![as]) { + Ok(UseTree::Rename(UseRename { + ident, + as_token: input.parse()?, + rename: { + if input.peek(Ident) { + input.parse()? + } else if input.peek(Token![_]) { + Ident::from(input.parse::()?) + } else { + return Err(input.error("expected identifier or underscore")); + } + }, + })) + } else { + Ok(UseTree::Name(UseName { ident })) + } + } else if lookahead.peek(Token![*]) { + Ok(UseTree::Glob(UseGlob { + star_token: input.parse()?, + })) + } else if lookahead.peek(token::Brace) { + let content; + Ok(UseTree::Group(UseGroup { + brace_token: braced!(content in input), + items: content.parse_terminated(UseTree::parse)?, + })) + } else { + Err(lookahead.error()) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemStatic { + fn parse(input: ParseStream) -> Result { + Ok(ItemStatic { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + static_token: input.parse()?, + mutability: input.parse()?, + ident: input.parse()?, + colon_token: input.parse()?, + ty: input.parse()?, + eq_token: input.parse()?, + expr: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemConst { + fn parse(input: ParseStream) -> Result { + Ok(ItemConst { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + const_token: input.parse()?, + ident: { + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.call(Ident::parse_any)? + } else { + return Err(lookahead.error()); + } + }, + colon_token: input.parse()?, + ty: input.parse()?, + eq_token: input.parse()?, + expr: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + fn pop_variadic(args: &mut Punctuated) -> Option { + let trailing_punct = args.trailing_punct(); + + let last = match args.last_mut()? { + FnArg::Typed(last) => last, + _ => return None, + }; + + let ty = match last.ty.as_ref() { + Type::Verbatim(ty) => ty, + _ => return None, + }; + + let mut variadic = Variadic { + attrs: Vec::new(), + dots: parse2(ty.clone()).ok()?, + }; + + if let Pat::Verbatim(pat) = last.pat.as_ref() { + if pat.to_string() == "..." && !trailing_punct { + variadic.attrs = mem::replace(&mut last.attrs, Vec::new()); + args.pop(); + } + } + + Some(variadic) + } + + fn variadic_to_tokens(dots: &Token![...]) -> TokenStream { + TokenStream::from_iter(vec![ + TokenTree::Punct({ + let mut dot = Punct::new('.', Spacing::Joint); + dot.set_span(dots.spans[0]); + dot + }), + TokenTree::Punct({ + let mut dot = Punct::new('.', Spacing::Joint); + dot.set_span(dots.spans[1]); + dot + }), + TokenTree::Punct({ + let mut dot = Punct::new('.', Spacing::Alone); + dot.set_span(dots.spans[2]); + dot + }), + ]) + } + + fn peek_signature(input: ParseStream) -> bool { + let fork = input.fork(); + fork.parse::>().is_ok() + && fork.parse::>().is_ok() + && fork.parse::>().is_ok() + && fork.parse::>().is_ok() + && fork.peek(Token![fn]) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Signature { + fn parse(input: ParseStream) -> Result { + let constness: Option = input.parse()?; + let asyncness: Option = input.parse()?; + let unsafety: Option = input.parse()?; + let abi: Option = input.parse()?; + let fn_token: Token![fn] = input.parse()?; + let ident: Ident = input.parse()?; + let mut generics: Generics = input.parse()?; + + let content; + let paren_token = parenthesized!(content in input); + let mut inputs = parse_fn_args(&content)?; + let variadic = pop_variadic(&mut inputs); + + let output: ReturnType = input.parse()?; + generics.where_clause = input.parse()?; + + Ok(Signature { + constness, + asyncness, + unsafety, + abi, + fn_token, + ident, + generics, + paren_token, + inputs, + variadic, + output, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemFn { + fn parse(input: ParseStream) -> Result { + let outer_attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let sig: Signature = input.parse()?; + parse_rest_of_fn(input, outer_attrs, vis, sig) + } + } + + fn parse_rest_of_fn( + input: ParseStream, + mut attrs: Vec, + vis: Visibility, + sig: Signature, + ) -> Result { + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + let stmts = content.call(Block::parse_within)?; + + Ok(ItemFn { + attrs, + vis, + sig, + block: Box::new(Block { brace_token, stmts }), + }) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for FnArg { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + + let ahead = input.fork(); + if let Ok(mut receiver) = ahead.parse::() { + if !ahead.peek(Token![:]) { + input.advance_to(&ahead); + receiver.attrs = attrs; + return Ok(FnArg::Receiver(receiver)); + } + } + + let mut typed = input.call(fn_arg_typed)?; + typed.attrs = attrs; + Ok(FnArg::Typed(typed)) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Receiver { + fn parse(input: ParseStream) -> Result { + Ok(Receiver { + attrs: Vec::new(), + reference: { + if input.peek(Token![&]) { + Some((input.parse()?, input.parse()?)) + } else { + None + } + }, + mutability: input.parse()?, + self_token: input.parse()?, + }) + } + } + + fn parse_fn_args(input: ParseStream) -> Result> { + let mut args = Punctuated::new(); + let mut has_receiver = false; + + while !input.is_empty() { + let attrs = input.call(Attribute::parse_outer)?; + + let arg = if let Some(dots) = input.parse::>()? { + FnArg::Typed(PatType { + attrs, + pat: Box::new(Pat::Verbatim(variadic_to_tokens(&dots))), + colon_token: Token![:](dots.spans[0]), + ty: Box::new(Type::Verbatim(variadic_to_tokens(&dots))), + }) + } else { + let mut arg: FnArg = input.parse()?; + match &mut arg { + FnArg::Receiver(receiver) if has_receiver => { + return Err(Error::new( + receiver.self_token.span, + "unexpected second method receiver", + )); + } + FnArg::Receiver(receiver) if !args.is_empty() => { + return Err(Error::new( + receiver.self_token.span, + "unexpected method receiver", + )); + } + FnArg::Receiver(receiver) => { + has_receiver = true; + receiver.attrs = attrs; + } + FnArg::Typed(arg) => arg.attrs = attrs, + } + arg + }; + args.push_value(arg); + + if input.is_empty() { + break; + } + + let comma: Token![,] = input.parse()?; + args.push_punct(comma); + } + + Ok(args) + } + + fn fn_arg_typed(input: ParseStream) -> Result { + // Hack to parse pre-2018 syntax in + // test/ui/rfc-2565-param-attrs/param-attrs-pretty.rs + // because the rest of the test case is valuable. + if input.peek(Ident) && input.peek2(Token![<]) { + let span = input.fork().parse::()?.span(); + return Ok(PatType { + attrs: Vec::new(), + pat: Box::new(Pat::Wild(PatWild { + attrs: Vec::new(), + underscore_token: Token![_](span), + })), + colon_token: Token![:](span), + ty: input.parse()?, + }); + } + + Ok(PatType { + attrs: Vec::new(), + pat: Box::new(pat::parsing::multi_pat(input)?), + colon_token: input.parse()?, + ty: Box::new(match input.parse::>()? { + Some(dot3) => Type::Verbatim(variadic_to_tokens(&dot3)), + None => input.parse()?, + }), + }) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemMod { + fn parse(input: ParseStream) -> Result { + let mut attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let mod_token: Token![mod] = input.parse()?; + let ident: Ident = input.parse()?; + + let lookahead = input.lookahead1(); + if lookahead.peek(Token![;]) { + Ok(ItemMod { + attrs, + vis, + mod_token, + ident, + content: None, + semi: Some(input.parse()?), + }) + } else if lookahead.peek(token::Brace) { + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + + let mut items = Vec::new(); + while !content.is_empty() { + items.push(content.parse()?); + } + + Ok(ItemMod { + attrs, + vis, + mod_token, + ident, + content: Some((brace_token, items)), + semi: None, + }) + } else { + Err(lookahead.error()) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemForeignMod { + fn parse(input: ParseStream) -> Result { + let mut attrs = input.call(Attribute::parse_outer)?; + let abi: Abi = input.parse()?; + + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + let mut items = Vec::new(); + while !content.is_empty() { + items.push(content.parse()?); + } + + Ok(ItemForeignMod { + attrs, + abi, + brace_token, + items, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ForeignItem { + fn parse(input: ParseStream) -> Result { + let begin = input.fork(); + let mut attrs = input.call(Attribute::parse_outer)?; + let ahead = input.fork(); + let vis: Visibility = ahead.parse()?; + + let lookahead = ahead.lookahead1(); + let mut item = if lookahead.peek(Token![fn]) || peek_signature(&ahead) { + let vis: Visibility = input.parse()?; + let sig: Signature = input.parse()?; + if input.peek(token::Brace) { + let content; + braced!(content in input); + content.call(Attribute::parse_inner)?; + content.call(Block::parse_within)?; + + Ok(ForeignItem::Verbatim(verbatim::between(begin, input))) + } else { + Ok(ForeignItem::Fn(ForeignItemFn { + attrs: Vec::new(), + vis, + sig, + semi_token: input.parse()?, + })) + } + } else if lookahead.peek(Token![static]) { + let vis = input.parse()?; + let static_token = input.parse()?; + let mutability = input.parse()?; + let ident = input.parse()?; + let colon_token = input.parse()?; + let ty = input.parse()?; + if input.peek(Token![=]) { + input.parse::()?; + input.parse::()?; + input.parse::()?; + Ok(ForeignItem::Verbatim(verbatim::between(begin, input))) + } else { + Ok(ForeignItem::Static(ForeignItemStatic { + attrs: Vec::new(), + vis, + static_token, + mutability, + ident, + colon_token, + ty, + semi_token: input.parse()?, + })) + } + } else if lookahead.peek(Token![type]) { + parse_foreign_item_type(begin, input) + } else if vis.is_inherited() + && (lookahead.peek(Ident) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + || lookahead.peek(Token![::])) + { + input.parse().map(ForeignItem::Macro) + } else { + Err(lookahead.error()) + }?; + + let item_attrs = match &mut item { + ForeignItem::Fn(item) => &mut item.attrs, + ForeignItem::Static(item) => &mut item.attrs, + ForeignItem::Type(item) => &mut item.attrs, + ForeignItem::Macro(item) => &mut item.attrs, + ForeignItem::Verbatim(_) => return Ok(item), + + #[cfg(syn_no_non_exhaustive)] + _ => unreachable!(), + }; + attrs.append(item_attrs); + *item_attrs = attrs; + + Ok(item) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ForeignItemFn { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let sig: Signature = input.parse()?; + let semi_token: Token![;] = input.parse()?; + Ok(ForeignItemFn { + attrs, + vis, + sig, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ForeignItemStatic { + fn parse(input: ParseStream) -> Result { + Ok(ForeignItemStatic { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + static_token: input.parse()?, + mutability: input.parse()?, + ident: input.parse()?, + colon_token: input.parse()?, + ty: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ForeignItemType { + fn parse(input: ParseStream) -> Result { + Ok(ForeignItemType { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + type_token: input.parse()?, + ident: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + fn parse_foreign_item_type(begin: ParseBuffer, input: ParseStream) -> Result { + let FlexibleItemType { + vis, + defaultness, + type_token, + ident, + generics, + colon_token, + bounds: _, + ty, + semi_token, + } = FlexibleItemType::parse(input, WhereClauseLocation::BeforeEq)?; + + if defaultness.is_some() + || generics.lt_token.is_some() + || generics.where_clause.is_some() + || colon_token.is_some() + || ty.is_some() + { + Ok(ForeignItem::Verbatim(verbatim::between(begin, input))) + } else { + Ok(ForeignItem::Type(ForeignItemType { + attrs: Vec::new(), + vis, + type_token, + ident, + semi_token, + })) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ForeignItemMacro { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let mac: Macro = input.parse()?; + let semi_token: Option = if mac.delimiter.is_brace() { + None + } else { + Some(input.parse()?) + }; + Ok(ForeignItemMacro { + attrs, + mac, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemType { + fn parse(input: ParseStream) -> Result { + Ok(ItemType { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + type_token: input.parse()?, + ident: input.parse()?, + generics: { + let mut generics: Generics = input.parse()?; + generics.where_clause = input.parse()?; + generics + }, + eq_token: input.parse()?, + ty: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + fn parse_item_type(begin: ParseBuffer, input: ParseStream) -> Result { + let FlexibleItemType { + vis, + defaultness, + type_token, + ident, + generics, + colon_token, + bounds: _, + ty, + semi_token, + } = FlexibleItemType::parse(input, WhereClauseLocation::BeforeEq)?; + + if defaultness.is_some() || colon_token.is_some() || ty.is_none() { + Ok(Item::Verbatim(verbatim::between(begin, input))) + } else { + let (eq_token, ty) = ty.unwrap(); + Ok(Item::Type(ItemType { + attrs: Vec::new(), + vis, + type_token, + ident, + generics, + eq_token, + ty: Box::new(ty), + semi_token, + })) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemStruct { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis = input.parse::()?; + let struct_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, fields, semi_token) = derive::parsing::data_struct(input)?; + Ok(ItemStruct { + attrs, + vis, + struct_token, + ident, + generics: Generics { + where_clause, + ..generics + }, + fields, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemEnum { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis = input.parse::()?; + let enum_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, brace_token, variants) = derive::parsing::data_enum(input)?; + Ok(ItemEnum { + attrs, + vis, + enum_token, + ident, + generics: Generics { + where_clause, + ..generics + }, + brace_token, + variants, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemUnion { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis = input.parse::()?; + let union_token = input.parse::()?; + let ident = input.parse::()?; + let generics = input.parse::()?; + let (where_clause, fields) = derive::parsing::data_union(input)?; + Ok(ItemUnion { + attrs, + vis, + union_token, + ident, + generics: Generics { + where_clause, + ..generics + }, + fields, + }) + } + } + + fn parse_trait_or_trait_alias(input: ParseStream) -> Result { + let (attrs, vis, trait_token, ident, generics) = parse_start_of_trait_alias(input)?; + let lookahead = input.lookahead1(); + if lookahead.peek(token::Brace) + || lookahead.peek(Token![:]) + || lookahead.peek(Token![where]) + { + let unsafety = None; + let auto_token = None; + parse_rest_of_trait( + input, + attrs, + vis, + unsafety, + auto_token, + trait_token, + ident, + generics, + ) + .map(Item::Trait) + } else if lookahead.peek(Token![=]) { + parse_rest_of_trait_alias(input, attrs, vis, trait_token, ident, generics) + .map(Item::TraitAlias) + } else { + Err(lookahead.error()) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemTrait { + fn parse(input: ParseStream) -> Result { + let outer_attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let unsafety: Option = input.parse()?; + let auto_token: Option = input.parse()?; + let trait_token: Token![trait] = input.parse()?; + let ident: Ident = input.parse()?; + let generics: Generics = input.parse()?; + parse_rest_of_trait( + input, + outer_attrs, + vis, + unsafety, + auto_token, + trait_token, + ident, + generics, + ) + } + } + + fn parse_rest_of_trait( + input: ParseStream, + mut attrs: Vec, + vis: Visibility, + unsafety: Option, + auto_token: Option, + trait_token: Token![trait], + ident: Ident, + mut generics: Generics, + ) -> Result { + let colon_token: Option = input.parse()?; + + let mut supertraits = Punctuated::new(); + if colon_token.is_some() { + loop { + if input.peek(Token![where]) || input.peek(token::Brace) { + break; + } + supertraits.push_value(input.parse()?); + if input.peek(Token![where]) || input.peek(token::Brace) { + break; + } + supertraits.push_punct(input.parse()?); + } + } + + generics.where_clause = input.parse()?; + + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + let mut items = Vec::new(); + while !content.is_empty() { + items.push(content.parse()?); + } + + Ok(ItemTrait { + attrs, + vis, + unsafety, + auto_token, + trait_token, + ident, + generics, + colon_token, + supertraits, + brace_token, + items, + }) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemTraitAlias { + fn parse(input: ParseStream) -> Result { + let (attrs, vis, trait_token, ident, generics) = parse_start_of_trait_alias(input)?; + parse_rest_of_trait_alias(input, attrs, vis, trait_token, ident, generics) + } + } + + fn parse_start_of_trait_alias( + input: ParseStream, + ) -> Result<(Vec, Visibility, Token![trait], Ident, Generics)> { + let attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let trait_token: Token![trait] = input.parse()?; + let ident: Ident = input.parse()?; + let generics: Generics = input.parse()?; + Ok((attrs, vis, trait_token, ident, generics)) + } + + fn parse_rest_of_trait_alias( + input: ParseStream, + attrs: Vec, + vis: Visibility, + trait_token: Token![trait], + ident: Ident, + mut generics: Generics, + ) -> Result { + let eq_token: Token![=] = input.parse()?; + + let mut bounds = Punctuated::new(); + loop { + if input.peek(Token![where]) || input.peek(Token![;]) { + break; + } + bounds.push_value(input.parse()?); + if input.peek(Token![where]) || input.peek(Token![;]) { + break; + } + bounds.push_punct(input.parse()?); + } + + generics.where_clause = input.parse()?; + let semi_token: Token![;] = input.parse()?; + + Ok(ItemTraitAlias { + attrs, + vis, + trait_token, + ident, + generics, + eq_token, + bounds, + semi_token, + }) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitItem { + fn parse(input: ParseStream) -> Result { + let begin = input.fork(); + let mut attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let defaultness: Option = input.parse()?; + let ahead = input.fork(); + + let lookahead = ahead.lookahead1(); + let mut item = if lookahead.peek(Token![fn]) || peek_signature(&ahead) { + input.parse().map(TraitItem::Method) + } else if lookahead.peek(Token![const]) { + ahead.parse::()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.parse().map(TraitItem::Const) + } else if lookahead.peek(Token![async]) + || lookahead.peek(Token![unsafe]) + || lookahead.peek(Token![extern]) + || lookahead.peek(Token![fn]) + { + input.parse().map(TraitItem::Method) + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(Token![type]) { + parse_trait_item_type(begin.fork(), input) + } else if lookahead.peek(Ident) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + || lookahead.peek(Token![::]) + { + input.parse().map(TraitItem::Macro) + } else { + Err(lookahead.error()) + }?; + + match (vis, defaultness) { + (Visibility::Inherited, None) => {} + _ => return Ok(TraitItem::Verbatim(verbatim::between(begin, input))), + } + + let item_attrs = match &mut item { + TraitItem::Const(item) => &mut item.attrs, + TraitItem::Method(item) => &mut item.attrs, + TraitItem::Type(item) => &mut item.attrs, + TraitItem::Macro(item) => &mut item.attrs, + TraitItem::Verbatim(_) => unreachable!(), + + #[cfg(syn_no_non_exhaustive)] + _ => unreachable!(), + }; + attrs.append(item_attrs); + *item_attrs = attrs; + Ok(item) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitItemConst { + fn parse(input: ParseStream) -> Result { + Ok(TraitItemConst { + attrs: input.call(Attribute::parse_outer)?, + const_token: input.parse()?, + ident: { + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.call(Ident::parse_any)? + } else { + return Err(lookahead.error()); + } + }, + colon_token: input.parse()?, + ty: input.parse()?, + default: { + if input.peek(Token![=]) { + let eq_token: Token![=] = input.parse()?; + let default: Expr = input.parse()?; + Some((eq_token, default)) + } else { + None + } + }, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitItemMethod { + fn parse(input: ParseStream) -> Result { + let mut attrs = input.call(Attribute::parse_outer)?; + let sig: Signature = input.parse()?; + + let lookahead = input.lookahead1(); + let (brace_token, stmts, semi_token) = if lookahead.peek(token::Brace) { + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + let stmts = content.call(Block::parse_within)?; + (Some(brace_token), stmts, None) + } else if lookahead.peek(Token![;]) { + let semi_token: Token![;] = input.parse()?; + (None, Vec::new(), Some(semi_token)) + } else { + return Err(lookahead.error()); + }; + + Ok(TraitItemMethod { + attrs, + sig, + default: brace_token.map(|brace_token| Block { brace_token, stmts }), + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitItemType { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let type_token: Token![type] = input.parse()?; + let ident: Ident = input.parse()?; + let mut generics: Generics = input.parse()?; + let colon_token: Option = input.parse()?; + + let mut bounds = Punctuated::new(); + if colon_token.is_some() { + while !input.peek(Token![where]) && !input.peek(Token![=]) && !input.peek(Token![;]) + { + if !bounds.is_empty() { + bounds.push_punct(input.parse()?); + } + bounds.push_value(input.parse()?); + } + } + + let default = if input.peek(Token![=]) { + let eq_token: Token![=] = input.parse()?; + let default: Type = input.parse()?; + Some((eq_token, default)) + } else { + None + }; + + generics.where_clause = input.parse()?; + let semi_token: Token![;] = input.parse()?; + + Ok(TraitItemType { + attrs, + type_token, + ident, + generics, + colon_token, + bounds, + default, + semi_token, + }) + } + } + + fn parse_trait_item_type(begin: ParseBuffer, input: ParseStream) -> Result { + let FlexibleItemType { + vis, + defaultness, + type_token, + ident, + generics, + colon_token, + bounds, + ty, + semi_token, + } = FlexibleItemType::parse(input, WhereClauseLocation::Both)?; + + if defaultness.is_some() || vis.is_some() { + Ok(TraitItem::Verbatim(verbatim::between(begin, input))) + } else { + Ok(TraitItem::Type(TraitItemType { + attrs: Vec::new(), + type_token, + ident, + generics, + colon_token, + bounds, + default: ty, + semi_token, + })) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TraitItemMacro { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let mac: Macro = input.parse()?; + let semi_token: Option = if mac.delimiter.is_brace() { + None + } else { + Some(input.parse()?) + }; + Ok(TraitItemMacro { + attrs, + mac, + semi_token, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ItemImpl { + fn parse(input: ParseStream) -> Result { + let allow_verbatim_impl = false; + parse_impl(input, allow_verbatim_impl).map(Option::unwrap) + } + } + + fn parse_impl(input: ParseStream, allow_verbatim_impl: bool) -> Result> { + let mut attrs = input.call(Attribute::parse_outer)?; + let has_visibility = allow_verbatim_impl && input.parse::()?.is_some(); + let defaultness: Option = input.parse()?; + let unsafety: Option = input.parse()?; + let impl_token: Token![impl] = input.parse()?; + + let has_generics = input.peek(Token![<]) + && (input.peek2(Token![>]) + || input.peek2(Token![#]) + || (input.peek2(Ident) || input.peek2(Lifetime)) + && (input.peek3(Token![:]) + || input.peek3(Token![,]) + || input.peek3(Token![>]) + || input.peek3(Token![=])) + || input.peek2(Token![const])); + let mut generics: Generics = if has_generics { + input.parse()? + } else { + Generics::default() + }; + + let is_const_impl = allow_verbatim_impl + && (input.peek(Token![const]) || input.peek(Token![?]) && input.peek2(Token![const])); + if is_const_impl { + input.parse::>()?; + input.parse::()?; + } + + let begin = input.fork(); + let polarity = if input.peek(Token![!]) && !input.peek2(token::Brace) { + Some(input.parse::()?) + } else { + None + }; + + #[cfg(not(feature = "printing"))] + let first_ty_span = input.span(); + let mut first_ty: Type = input.parse()?; + let self_ty: Type; + let trait_; + + let is_impl_for = input.peek(Token![for]); + if is_impl_for { + let for_token: Token![for] = input.parse()?; + let mut first_ty_ref = &first_ty; + while let Type::Group(ty) = first_ty_ref { + first_ty_ref = &ty.elem; + } + if let Type::Path(TypePath { qself: None, .. }) = first_ty_ref { + while let Type::Group(ty) = first_ty { + first_ty = *ty.elem; + } + if let Type::Path(TypePath { qself: None, path }) = first_ty { + trait_ = Some((polarity, path, for_token)); + } else { + unreachable!(); + } + } else if !allow_verbatim_impl { + #[cfg(feature = "printing")] + return Err(Error::new_spanned(first_ty_ref, "expected trait path")); + #[cfg(not(feature = "printing"))] + return Err(Error::new(first_ty_span, "expected trait path")); + } else { + trait_ = None; + } + self_ty = input.parse()?; + } else { + trait_ = None; + self_ty = if polarity.is_none() { + first_ty + } else { + Type::Verbatim(verbatim::between(begin, input)) + }; + } + + generics.where_clause = input.parse()?; + + let content; + let brace_token = braced!(content in input); + attr::parsing::parse_inner(&content, &mut attrs)?; + + let mut items = Vec::new(); + while !content.is_empty() { + items.push(content.parse()?); + } + + if has_visibility || is_const_impl || is_impl_for && trait_.is_none() { + Ok(None) + } else { + Ok(Some(ItemImpl { + attrs, + defaultness, + unsafety, + impl_token, + generics, + trait_, + self_ty: Box::new(self_ty), + brace_token, + items, + })) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ImplItem { + fn parse(input: ParseStream) -> Result { + let begin = input.fork(); + let mut attrs = input.call(Attribute::parse_outer)?; + let ahead = input.fork(); + let vis: Visibility = ahead.parse()?; + + let mut lookahead = ahead.lookahead1(); + let defaultness = if lookahead.peek(Token![default]) && !ahead.peek2(Token![!]) { + let defaultness: Token![default] = ahead.parse()?; + lookahead = ahead.lookahead1(); + Some(defaultness) + } else { + None + }; + + let mut item = if lookahead.peek(Token![fn]) || peek_signature(&ahead) { + input.parse().map(ImplItem::Method) + } else if lookahead.peek(Token![const]) { + let const_token: Token![const] = ahead.parse()?; + let lookahead = ahead.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.advance_to(&ahead); + let ident: Ident = input.call(Ident::parse_any)?; + let colon_token: Token![:] = input.parse()?; + let ty: Type = input.parse()?; + if let Some(eq_token) = input.parse()? { + return Ok(ImplItem::Const(ImplItemConst { + attrs, + vis, + defaultness, + const_token, + ident, + colon_token, + ty, + eq_token, + expr: input.parse()?, + semi_token: input.parse()?, + })); + } else { + input.parse::()?; + return Ok(ImplItem::Verbatim(verbatim::between(begin, input))); + } + } else { + Err(lookahead.error()) + } + } else if lookahead.peek(Token![type]) { + parse_impl_item_type(begin, input) + } else if vis.is_inherited() + && defaultness.is_none() + && (lookahead.peek(Ident) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + || lookahead.peek(Token![::])) + { + input.parse().map(ImplItem::Macro) + } else { + Err(lookahead.error()) + }?; + + { + let item_attrs = match &mut item { + ImplItem::Const(item) => &mut item.attrs, + ImplItem::Method(item) => &mut item.attrs, + ImplItem::Type(item) => &mut item.attrs, + ImplItem::Macro(item) => &mut item.attrs, + ImplItem::Verbatim(_) => return Ok(item), + + #[cfg(syn_no_non_exhaustive)] + _ => unreachable!(), + }; + attrs.append(item_attrs); + *item_attrs = attrs; + } + + Ok(item) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ImplItemConst { + fn parse(input: ParseStream) -> Result { + Ok(ImplItemConst { + attrs: input.call(Attribute::parse_outer)?, + vis: input.parse()?, + defaultness: input.parse()?, + const_token: input.parse()?, + ident: { + let lookahead = input.lookahead1(); + if lookahead.peek(Ident) || lookahead.peek(Token![_]) { + input.call(Ident::parse_any)? + } else { + return Err(lookahead.error()); + } + }, + colon_token: input.parse()?, + ty: input.parse()?, + eq_token: input.parse()?, + expr: input.parse()?, + semi_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ImplItemMethod { + fn parse(input: ParseStream) -> Result { + let mut attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let defaultness: Option = input.parse()?; + let sig: Signature = input.parse()?; + + let block = if let Some(semi) = input.parse::>()? { + // Accept methods without a body in an impl block because + // rustc's *parser* does not reject them (the compilation error + // is emitted later than parsing) and it can be useful for macro + // DSLs. + let mut punct = Punct::new(';', Spacing::Alone); + punct.set_span(semi.span); + let tokens = TokenStream::from_iter(vec![TokenTree::Punct(punct)]); + Block { + brace_token: Brace { span: semi.span }, + stmts: vec![Stmt::Item(Item::Verbatim(tokens))], + } + } else { + let content; + let brace_token = braced!(content in input); + attrs.extend(content.call(Attribute::parse_inner)?); + Block { + brace_token, + stmts: content.call(Block::parse_within)?, + } + }; + + Ok(ImplItemMethod { + attrs, + vis, + defaultness, + sig, + block, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ImplItemType { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let vis: Visibility = input.parse()?; + let defaultness: Option = input.parse()?; + let type_token: Token![type] = input.parse()?; + let ident: Ident = input.parse()?; + let mut generics: Generics = input.parse()?; + let eq_token: Token![=] = input.parse()?; + let ty: Type = input.parse()?; + generics.where_clause = input.parse()?; + let semi_token: Token![;] = input.parse()?; + Ok(ImplItemType { + attrs, + vis, + defaultness, + type_token, + ident, + generics, + eq_token, + ty, + semi_token, + }) + } + } + + fn parse_impl_item_type(begin: ParseBuffer, input: ParseStream) -> Result { + let FlexibleItemType { + vis, + defaultness, + type_token, + ident, + generics, + colon_token, + bounds: _, + ty, + semi_token, + } = FlexibleItemType::parse(input, WhereClauseLocation::Both)?; + + if colon_token.is_some() || ty.is_none() { + Ok(ImplItem::Verbatim(verbatim::between(begin, input))) + } else { + let (eq_token, ty) = ty.unwrap(); + Ok(ImplItem::Type(ImplItemType { + attrs: Vec::new(), + vis, + defaultness, + type_token, + ident, + generics, + eq_token, + ty, + semi_token, + })) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ImplItemMacro { + fn parse(input: ParseStream) -> Result { + let attrs = input.call(Attribute::parse_outer)?; + let mac: Macro = input.parse()?; + let semi_token: Option = if mac.delimiter.is_brace() { + None + } else { + Some(input.parse()?) + }; + Ok(ImplItemMacro { + attrs, + mac, + semi_token, + }) + } + } + + impl Visibility { + fn is_inherited(&self) -> bool { + match *self { + Visibility::Inherited => true, + _ => false, + } + } + } + + impl MacroDelimiter { + fn is_brace(&self) -> bool { + match *self { + MacroDelimiter::Brace(_) => true, + MacroDelimiter::Paren(_) | MacroDelimiter::Bracket(_) => false, + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::attr::FilterAttrs; + use crate::print::TokensOrDefault; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemExternCrate { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.extern_token.to_tokens(tokens); + self.crate_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + if let Some((as_token, rename)) = &self.rename { + as_token.to_tokens(tokens); + rename.to_tokens(tokens); + } + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemUse { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.use_token.to_tokens(tokens); + self.leading_colon.to_tokens(tokens); + self.tree.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemStatic { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.static_token.to_tokens(tokens); + self.mutability.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.expr.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemConst { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.const_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.expr.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemFn { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.sig.to_tokens(tokens); + self.block.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&self.block.stmts); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemMod { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.mod_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + if let Some((brace, items)) = &self.content { + brace.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(items); + }); + } else { + TokensOrDefault(&self.semi).to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemForeignMod { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.abi.to_tokens(tokens); + self.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&self.items); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemType { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.type_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemEnum { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.enum_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.brace_token.surround(tokens, |tokens| { + self.variants.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemStruct { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.struct_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + match &self.fields { + Fields::Named(fields) => { + self.generics.where_clause.to_tokens(tokens); + fields.to_tokens(tokens); + } + Fields::Unnamed(fields) => { + fields.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + TokensOrDefault(&self.semi_token).to_tokens(tokens); + } + Fields::Unit => { + self.generics.where_clause.to_tokens(tokens); + TokensOrDefault(&self.semi_token).to_tokens(tokens); + } + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemUnion { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.union_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.fields.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemTrait { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.unsafety.to_tokens(tokens); + self.auto_token.to_tokens(tokens); + self.trait_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + if !self.supertraits.is_empty() { + TokensOrDefault(&self.colon_token).to_tokens(tokens); + self.supertraits.to_tokens(tokens); + } + self.generics.where_clause.to_tokens(tokens); + self.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&self.items); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemTraitAlias { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.trait_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemImpl { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.defaultness.to_tokens(tokens); + self.unsafety.to_tokens(tokens); + self.impl_token.to_tokens(tokens); + self.generics.to_tokens(tokens); + if let Some((polarity, path, for_token)) = &self.trait_ { + polarity.to_tokens(tokens); + path.to_tokens(tokens); + for_token.to_tokens(tokens); + } + self.self_ty.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&self.items); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.mac.path.to_tokens(tokens); + self.mac.bang_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + match &self.mac.delimiter { + MacroDelimiter::Paren(paren) => { + paren.surround(tokens, |tokens| self.mac.tokens.to_tokens(tokens)); + } + MacroDelimiter::Brace(brace) => { + brace.surround(tokens, |tokens| self.mac.tokens.to_tokens(tokens)); + } + MacroDelimiter::Bracket(bracket) => { + bracket.surround(tokens, |tokens| self.mac.tokens.to_tokens(tokens)); + } + } + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ItemMacro2 { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.macro_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.rules.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UsePath { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + self.colon2_token.to_tokens(tokens); + self.tree.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UseName { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UseRename { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + self.as_token.to_tokens(tokens); + self.rename.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UseGlob { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.star_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UseGroup { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.brace_token.surround(tokens, |tokens| { + self.items.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitItemConst { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.const_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + if let Some((eq_token, default)) = &self.default { + eq_token.to_tokens(tokens); + default.to_tokens(tokens); + } + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitItemMethod { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.sig.to_tokens(tokens); + match &self.default { + Some(block) => { + block.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&block.stmts); + }); + } + None => { + TokensOrDefault(&self.semi_token).to_tokens(tokens); + } + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitItemType { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.type_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + if !self.bounds.is_empty() { + TokensOrDefault(&self.colon_token).to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + if let Some((eq_token, default)) = &self.default { + eq_token.to_tokens(tokens); + default.to_tokens(tokens); + } + self.generics.where_clause.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TraitItemMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.mac.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ImplItemConst { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.defaultness.to_tokens(tokens); + self.const_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.expr.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ImplItemMethod { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.defaultness.to_tokens(tokens); + self.sig.to_tokens(tokens); + if self.block.stmts.len() == 1 { + if let Stmt::Item(Item::Verbatim(verbatim)) = &self.block.stmts[0] { + if verbatim.to_string() == ";" { + verbatim.to_tokens(tokens); + return; + } + } + } + self.block.brace_token.surround(tokens, |tokens| { + tokens.append_all(self.attrs.inner()); + tokens.append_all(&self.block.stmts); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ImplItemType { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.defaultness.to_tokens(tokens); + self.type_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ImplItemMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.mac.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ForeignItemFn { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.sig.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ForeignItemStatic { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.static_token.to_tokens(tokens); + self.mutability.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ForeignItemType { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.vis.to_tokens(tokens); + self.type_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ForeignItemMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.mac.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + } + } + + fn maybe_variadic_to_tokens(arg: &FnArg, tokens: &mut TokenStream) -> bool { + let arg = match arg { + FnArg::Typed(arg) => arg, + FnArg::Receiver(receiver) => { + receiver.to_tokens(tokens); + return false; + } + }; + + match arg.ty.as_ref() { + Type::Verbatim(ty) if ty.to_string() == "..." => { + match arg.pat.as_ref() { + Pat::Verbatim(pat) if pat.to_string() == "..." => { + tokens.append_all(arg.attrs.outer()); + pat.to_tokens(tokens); + } + _ => arg.to_tokens(tokens), + } + true + } + _ => { + arg.to_tokens(tokens); + false + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Signature { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.constness.to_tokens(tokens); + self.asyncness.to_tokens(tokens); + self.unsafety.to_tokens(tokens); + self.abi.to_tokens(tokens); + self.fn_token.to_tokens(tokens); + self.ident.to_tokens(tokens); + self.generics.to_tokens(tokens); + self.paren_token.surround(tokens, |tokens| { + let mut last_is_variadic = false; + for pair in self.inputs.pairs() { + last_is_variadic = maybe_variadic_to_tokens(pair.value(), tokens); + pair.punct().to_tokens(tokens); + } + if self.variadic.is_some() && !last_is_variadic { + if !self.inputs.empty_or_trailing() { + ::default().to_tokens(tokens); + } + self.variadic.to_tokens(tokens); + } + }); + self.output.to_tokens(tokens); + self.generics.where_clause.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Receiver { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + if let Some((ampersand, lifetime)) = &self.reference { + ampersand.to_tokens(tokens); + lifetime.to_tokens(tokens); + } + self.mutability.to_tokens(tokens); + self.self_token.to_tokens(tokens); + } + } +} diff --git a/rust/syn/lib.rs b/rust/syn/lib.rs new file mode 100644 index 00000000000000..c32d6b3cb8cbf9 --- /dev/null +++ b/rust/syn/lib.rs @@ -0,0 +1,985 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! [![github]](https://github.com/dtolnay/syn) [![crates-io]](https://crates.io/crates/syn) [![docs-rs]](crate) +//! +//! [github]: https://img.shields.io/badge/github-8da0cb?style=for-the-badge&labelColor=555555&logo=github +//! [crates-io]: https://img.shields.io/badge/crates.io-fc8d62?style=for-the-badge&labelColor=555555&logo=rust +//! [docs-rs]: https://img.shields.io/badge/docs.rs-66c2a5?style=for-the-badge&labelColor=555555&logo=docs.rs +//! +//!
+//! +//! Syn is a parsing library for parsing a stream of Rust tokens into a syntax +//! tree of Rust source code. +//! +//! Currently this library is geared toward use in Rust procedural macros, but +//! contains some APIs that may be useful more generally. +//! +//! - **Data structures** — Syn provides a complete syntax tree that can +//! represent any valid Rust source code. The syntax tree is rooted at +//! [`syn::File`] which represents a full source file, but there are other +//! entry points that may be useful to procedural macros including +//! [`syn::Item`], [`syn::Expr`] and [`syn::Type`]. +//! +//! - **Derives** — Of particular interest to derive macros is +//! [`syn::DeriveInput`] which is any of the three legal input items to a +//! derive macro. An example below shows using this type in a library that can +//! derive implementations of a user-defined trait. +//! +//! - **Parsing** — Parsing in Syn is built around [parser functions] with the +//! signature `fn(ParseStream) -> Result`. Every syntax tree node defined +//! by Syn is individually parsable and may be used as a building block for +//! custom syntaxes, or you may dream up your own brand new syntax without +//! involving any of our syntax tree types. +//! +//! - **Location information** — Every token parsed by Syn is associated with a +//! `Span` that tracks line and column information back to the source of that +//! token. These spans allow a procedural macro to display detailed error +//! messages pointing to all the right places in the user's code. There is an +//! example of this below. +//! +//! - **Feature flags** — Functionality is aggressively feature gated so your +//! procedural macros enable only what they need, and do not pay in compile +//! time for all the rest. +//! +//! [`syn::File`]: File +//! [`syn::Item`]: Item +//! [`syn::Expr`]: Expr +//! [`syn::Type`]: Type +//! [`syn::DeriveInput`]: DeriveInput +//! [parser functions]: mod@parse +//! +//!
+//! +//! # Example of a derive macro +//! +//! The canonical derive macro using Syn looks like this. We write an ordinary +//! Rust function tagged with a `proc_macro_derive` attribute and the name of +//! the trait we are deriving. Any time that derive appears in the user's code, +//! the Rust compiler passes their data structure as tokens into our macro. We +//! get to execute arbitrary Rust code to figure out what to do with those +//! tokens, then hand some tokens back to the compiler to compile into the +//! user's crate. +//! +//! [`TokenStream`]: proc_macro::TokenStream +//! +//! ```toml +//! [dependencies] +//! syn = "1.0" +//! quote = "1.0" +//! +//! [lib] +//! proc-macro = true +//! ``` +//! +//! ``` +//! # extern crate proc_macro; +//! # +//! use proc_macro::TokenStream; +//! use quote::quote; +//! use syn::{parse_macro_input, DeriveInput}; +//! +//! # const IGNORE_TOKENS: &str = stringify! { +//! #[proc_macro_derive(MyMacro)] +//! # }; +//! pub fn my_macro(input: TokenStream) -> TokenStream { +//! // Parse the input tokens into a syntax tree +//! let input = parse_macro_input!(input as DeriveInput); +//! +//! // Build the output, possibly using quasi-quotation +//! let expanded = quote! { +//! // ... +//! }; +//! +//! // Hand the output tokens back to the compiler +//! TokenStream::from(expanded) +//! } +//! ``` +//! +//! The [`heapsize`] example directory shows a complete working implementation +//! of a derive macro. It works on any Rust compiler 1.31+. The example derives +//! a `HeapSize` trait which computes an estimate of the amount of heap memory +//! owned by a value. +//! +//! [`heapsize`]: https://github.com/dtolnay/syn/tree/master/examples/heapsize +//! +//! ``` +//! pub trait HeapSize { +//! /// Total number of bytes of heap memory owned by `self`. +//! fn heap_size_of_children(&self) -> usize; +//! } +//! ``` +//! +//! The derive macro allows users to write `#[derive(HeapSize)]` on data +//! structures in their program. +//! +//! ``` +//! # const IGNORE_TOKENS: &str = stringify! { +//! #[derive(HeapSize)] +//! # }; +//! struct Demo<'a, T: ?Sized> { +//! a: Box, +//! b: u8, +//! c: &'a str, +//! d: String, +//! } +//! ``` +//! +//!


+//! +//! # Spans and error reporting +//! +//! The token-based procedural macro API provides great control over where the +//! compiler's error messages are displayed in user code. Consider the error the +//! user sees if one of their field types does not implement `HeapSize`. +//! +//! ``` +//! # const IGNORE_TOKENS: &str = stringify! { +//! #[derive(HeapSize)] +//! # }; +//! struct Broken { +//! ok: String, +//! bad: std::thread::Thread, +//! } +//! ``` +//! +//! By tracking span information all the way through the expansion of a +//! procedural macro as shown in the `heapsize` example, token-based macros in +//! Syn are able to trigger errors that directly pinpoint the source of the +//! problem. +//! +//! ```text +//! error[E0277]: the trait bound `std::thread::Thread: HeapSize` is not satisfied +//! --> src/main.rs:7:5 +//! | +//! 7 | bad: std::thread::Thread, +//! | ^^^^^^^^^^^^^^^^^^^^^^^^ the trait `HeapSize` is not implemented for `Thread` +//! ``` +//! +//!
+//! +//! # Parsing a custom syntax +//! +//! The [`lazy-static`] example directory shows the implementation of a +//! `functionlike!(...)` procedural macro in which the input tokens are parsed +//! using Syn's parsing API. +//! +//! [`lazy-static`]: https://github.com/dtolnay/syn/tree/master/examples/lazy-static +//! +//! The example reimplements the popular `lazy_static` crate from crates.io as a +//! procedural macro. +//! +//! ``` +//! # macro_rules! lazy_static { +//! # ($($tt:tt)*) => {} +//! # } +//! # +//! lazy_static! { +//! static ref USERNAME: Regex = Regex::new("^[a-z0-9_-]{3,16}$").unwrap(); +//! } +//! ``` +//! +//! The implementation shows how to trigger custom warnings and error messages +//! on the macro input. +//! +//! ```text +//! warning: come on, pick a more creative name +//! --> src/main.rs:10:16 +//! | +//! 10 | static ref FOO: String = "lazy_static".to_owned(); +//! | ^^^ +//! ``` +//! +//!
+//! +//! # Testing +//! +//! When testing macros, we often care not just that the macro can be used +//! successfully but also that when the macro is provided with invalid input it +//! produces maximally helpful error messages. Consider using the [`trybuild`] +//! crate to write tests for errors that are emitted by your macro or errors +//! detected by the Rust compiler in the expanded code following misuse of the +//! macro. Such tests help avoid regressions from later refactors that +//! mistakenly make an error no longer trigger or be less helpful than it used +//! to be. +//! +//! [`trybuild`]: https://github.com/dtolnay/trybuild +//! +//!
+//! +//! # Debugging +//! +//! When developing a procedural macro it can be helpful to look at what the +//! generated code looks like. Use `cargo rustc -- -Zunstable-options +//! --pretty=expanded` or the [`cargo expand`] subcommand. +//! +//! [`cargo expand`]: https://github.com/dtolnay/cargo-expand +//! +//! To show the expanded code for some crate that uses your procedural macro, +//! run `cargo expand` from that crate. To show the expanded code for one of +//! your own test cases, run `cargo expand --test the_test_case` where the last +//! argument is the name of the test file without the `.rs` extension. +//! +//! This write-up by Brandon W Maister discusses debugging in more detail: +//! [Debugging Rust's new Custom Derive system][debugging]. +//! +//! [debugging]: https://quodlibetor.github.io/posts/debugging-rusts-new-custom-derive-system/ +//! +//!
+//! +//! # Optional features +//! +//! Syn puts a lot of functionality behind optional features in order to +//! optimize compile time for the most common use cases. The following features +//! are available. +//! +//! - **`derive`** *(enabled by default)* — Data structures for representing the +//! possible input to a derive macro, including structs and enums and types. +//! - **`full`** — Data structures for representing the syntax tree of all valid +//! Rust source code, including items and expressions. +//! - **`parsing`** *(enabled by default)* — Ability to parse input tokens into +//! a syntax tree node of a chosen type. +//! - **`printing`** *(enabled by default)* — Ability to print a syntax tree +//! node as tokens of Rust source code. +//! - **`visit`** — Trait for traversing a syntax tree. +//! - **`visit-mut`** — Trait for traversing and mutating in place a syntax +//! tree. +//! - **`fold`** — Trait for transforming an owned syntax tree. +//! - **`clone-impls`** *(enabled by default)* — Clone impls for all syntax tree +//! types. +//! - **`extra-traits`** — Debug, Eq, PartialEq, Hash impls for all syntax tree +//! types. +//! - **`proc-macro`** *(enabled by default)* — Runtime dependency on the +//! dynamic library libproc_macro from rustc toolchain. + +// Syn types in rustdoc of other crates get linked to here. +#![doc(html_root_url = "https://docs.rs/syn/1.0.104")] +#![cfg_attr(doc_cfg, feature(doc_cfg))] +#![allow(non_camel_case_types)] +#![allow( + clippy::bool_to_int_with_if, + clippy::cast_lossless, + clippy::cast_possible_truncation, + clippy::cast_possible_wrap, + clippy::cast_ptr_alignment, + clippy::default_trait_access, + clippy::doc_markdown, + clippy::expl_impl_clone_on_copy, + clippy::explicit_auto_deref, + clippy::if_not_else, + clippy::inherent_to_string, + clippy::items_after_statements, + clippy::large_enum_variant, + clippy::manual_assert, + clippy::match_on_vec_items, + clippy::match_same_arms, + clippy::match_wildcard_for_single_variants, // clippy bug: https://github.com/rust-lang/rust-clippy/issues/6984 + clippy::missing_errors_doc, + clippy::missing_panics_doc, + clippy::module_name_repetitions, + clippy::must_use_candidate, + clippy::needless_doctest_main, + clippy::needless_pass_by_value, + clippy::never_loop, + clippy::redundant_else, + clippy::return_self_not_must_use, + clippy::similar_names, + clippy::single_match_else, + clippy::too_many_arguments, + clippy::too_many_lines, + clippy::trivially_copy_pass_by_ref, + clippy::unnecessary_unwrap, + clippy::used_underscore_binding, + clippy::wildcard_imports +)] + +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" +))] +extern crate proc_macro; +extern crate proc_macro2; + +#[cfg(feature = "printing")] +extern crate quote; + +#[macro_use] +mod macros; + +#[cfg(feature = "parsing")] +#[macro_use] +mod group; + +#[macro_use] +pub mod token; + +mod ident; +pub use crate::ident::Ident; + +#[cfg(any(feature = "full", feature = "derive"))] +mod attr; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::attr::{ + AttrStyle, Attribute, AttributeArgs, Meta, MetaList, MetaNameValue, NestedMeta, +}; + +mod bigint; + +#[cfg(any(feature = "full", feature = "derive"))] +mod data; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::data::{ + Field, Fields, FieldsNamed, FieldsUnnamed, Variant, VisCrate, VisPublic, VisRestricted, + Visibility, +}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod expr; +#[cfg(feature = "full")] +pub use crate::expr::{ + Arm, FieldValue, GenericMethodArgument, Label, MethodTurbofish, RangeLimits, +}; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::expr::{ + Expr, ExprArray, ExprAssign, ExprAssignOp, ExprAsync, ExprAwait, ExprBinary, ExprBlock, + ExprBox, ExprBreak, ExprCall, ExprCast, ExprClosure, ExprContinue, ExprField, ExprForLoop, + ExprGroup, ExprIf, ExprIndex, ExprLet, ExprLit, ExprLoop, ExprMacro, ExprMatch, ExprMethodCall, + ExprParen, ExprPath, ExprRange, ExprReference, ExprRepeat, ExprReturn, ExprStruct, ExprTry, + ExprTryBlock, ExprTuple, ExprType, ExprUnary, ExprUnsafe, ExprWhile, ExprYield, Index, Member, +}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod generics; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::generics::{ + BoundLifetimes, ConstParam, GenericParam, Generics, LifetimeDef, PredicateEq, + PredicateLifetime, PredicateType, TraitBound, TraitBoundModifier, TypeParam, TypeParamBound, + WhereClause, WherePredicate, +}; +#[cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))] +pub use crate::generics::{ImplGenerics, Turbofish, TypeGenerics}; + +#[cfg(feature = "full")] +mod item; +#[cfg(feature = "full")] +pub use crate::item::{ + FnArg, ForeignItem, ForeignItemFn, ForeignItemMacro, ForeignItemStatic, ForeignItemType, + ImplItem, ImplItemConst, ImplItemMacro, ImplItemMethod, ImplItemType, Item, ItemConst, + ItemEnum, ItemExternCrate, ItemFn, ItemForeignMod, ItemImpl, ItemMacro, ItemMacro2, ItemMod, + ItemStatic, ItemStruct, ItemTrait, ItemTraitAlias, ItemType, ItemUnion, ItemUse, Receiver, + Signature, TraitItem, TraitItemConst, TraitItemMacro, TraitItemMethod, TraitItemType, UseGlob, + UseGroup, UseName, UsePath, UseRename, UseTree, +}; + +#[cfg(feature = "full")] +mod file; +#[cfg(feature = "full")] +pub use crate::file::File; + +mod lifetime; +pub use crate::lifetime::Lifetime; + +mod lit; +pub use crate::lit::{ + Lit, LitBool, LitByte, LitByteStr, LitChar, LitFloat, LitInt, LitStr, StrStyle, +}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod mac; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::mac::{Macro, MacroDelimiter}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod derive; +#[cfg(feature = "derive")] +pub use crate::derive::{Data, DataEnum, DataStruct, DataUnion, DeriveInput}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod op; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::op::{BinOp, UnOp}; + +#[cfg(feature = "full")] +mod stmt; +#[cfg(feature = "full")] +pub use crate::stmt::{Block, Local, Stmt}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod ty; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::ty::{ + Abi, BareFnArg, ReturnType, Type, TypeArray, TypeBareFn, TypeGroup, TypeImplTrait, TypeInfer, + TypeMacro, TypeNever, TypeParen, TypePath, TypePtr, TypeReference, TypeSlice, TypeTraitObject, + TypeTuple, Variadic, +}; + +#[cfg(feature = "full")] +mod pat; +#[cfg(feature = "full")] +pub use crate::pat::{ + FieldPat, Pat, PatBox, PatIdent, PatLit, PatMacro, PatOr, PatPath, PatRange, PatReference, + PatRest, PatSlice, PatStruct, PatTuple, PatTupleStruct, PatType, PatWild, +}; + +#[cfg(any(feature = "full", feature = "derive"))] +mod path; +#[cfg(any(feature = "full", feature = "derive"))] +pub use crate::path::{ + AngleBracketedGenericArguments, Binding, Constraint, GenericArgument, + ParenthesizedGenericArguments, Path, PathArguments, PathSegment, QSelf, +}; + +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +pub mod buffer; +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +pub mod ext; +pub mod punctuated; +#[cfg(all(any(feature = "full", feature = "derive"), feature = "extra-traits"))] +mod tt; + +// Not public API except the `parse_quote!` macro. +#[cfg(feature = "parsing")] +#[doc(hidden)] +pub mod parse_quote; + +// Not public API except the `parse_macro_input!` macro. +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "parsing", + feature = "proc-macro" +))] +#[doc(hidden)] +pub mod parse_macro_input; + +#[cfg(all(feature = "parsing", feature = "printing"))] +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "printing"))))] +pub mod spanned; + +#[cfg(all(feature = "parsing", feature = "full"))] +mod whitespace; + +mod gen { + /// Syntax tree traversal to walk a shared borrow of a syntax tree. + /// + /// Each method of the [`Visit`] trait is a hook that can be overridden to + /// customize the behavior when visiting the corresponding type of node. By + /// default, every method recursively visits the substructure of the input + /// by invoking the right visitor method of each of its fields. + /// + /// [`Visit`]: visit::Visit + /// + /// ``` + /// # use syn::{Attribute, BinOp, Expr, ExprBinary}; + /// # + /// pub trait Visit<'ast> { + /// /* ... */ + /// + /// fn visit_expr_binary(&mut self, node: &'ast ExprBinary) { + /// visit_expr_binary(self, node); + /// } + /// + /// /* ... */ + /// # fn visit_attribute(&mut self, node: &'ast Attribute); + /// # fn visit_expr(&mut self, node: &'ast Expr); + /// # fn visit_bin_op(&mut self, node: &'ast BinOp); + /// } + /// + /// pub fn visit_expr_binary<'ast, V>(v: &mut V, node: &'ast ExprBinary) + /// where + /// V: Visit<'ast> + ?Sized, + /// { + /// for attr in &node.attrs { + /// v.visit_attribute(attr); + /// } + /// v.visit_expr(&*node.left); + /// v.visit_bin_op(&node.op); + /// v.visit_expr(&*node.right); + /// } + /// + /// /* ... */ + /// ``` + /// + /// *This module is available only if Syn is built with the `"visit"` feature.* + /// + ///
+ /// + /// # Example + /// + /// This visitor will print the name of every freestanding function in the + /// syntax tree, including nested functions. + /// + /// ``` + /// // [dependencies] + /// // quote = "1.0" + /// // syn = { version = "1.0", features = ["full", "visit"] } + /// + /// use quote::quote; + /// use syn::visit::{self, Visit}; + /// use syn::{File, ItemFn}; + /// + /// struct FnVisitor; + /// + /// impl<'ast> Visit<'ast> for FnVisitor { + /// fn visit_item_fn(&mut self, node: &'ast ItemFn) { + /// println!("Function with name={}", node.sig.ident); + /// + /// // Delegate to the default impl to visit any nested functions. + /// visit::visit_item_fn(self, node); + /// } + /// } + /// + /// fn main() { + /// let code = quote! { + /// pub fn f() { + /// fn g() {} + /// } + /// }; + /// + /// let syntax_tree: File = syn::parse2(code).unwrap(); + /// FnVisitor.visit_file(&syntax_tree); + /// } + /// ``` + /// + /// The `'ast` lifetime on the input references means that the syntax tree + /// outlives the complete recursive visit call, so the visitor is allowed to + /// hold on to references into the syntax tree. + /// + /// ``` + /// use quote::quote; + /// use syn::visit::{self, Visit}; + /// use syn::{File, ItemFn}; + /// + /// struct FnVisitor<'ast> { + /// functions: Vec<&'ast ItemFn>, + /// } + /// + /// impl<'ast> Visit<'ast> for FnVisitor<'ast> { + /// fn visit_item_fn(&mut self, node: &'ast ItemFn) { + /// self.functions.push(node); + /// visit::visit_item_fn(self, node); + /// } + /// } + /// + /// fn main() { + /// let code = quote! { + /// pub fn f() { + /// fn g() {} + /// } + /// }; + /// + /// let syntax_tree: File = syn::parse2(code).unwrap(); + /// let mut visitor = FnVisitor { functions: Vec::new() }; + /// visitor.visit_file(&syntax_tree); + /// for f in visitor.functions { + /// println!("Function with name={}", f.sig.ident); + /// } + /// } + /// ``` + #[cfg(feature = "visit")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "visit")))] + #[rustfmt::skip] + pub mod visit; + + /// Syntax tree traversal to mutate an exclusive borrow of a syntax tree in + /// place. + /// + /// Each method of the [`VisitMut`] trait is a hook that can be overridden + /// to customize the behavior when mutating the corresponding type of node. + /// By default, every method recursively visits the substructure of the + /// input by invoking the right visitor method of each of its fields. + /// + /// [`VisitMut`]: visit_mut::VisitMut + /// + /// ``` + /// # use syn::{Attribute, BinOp, Expr, ExprBinary}; + /// # + /// pub trait VisitMut { + /// /* ... */ + /// + /// fn visit_expr_binary_mut(&mut self, node: &mut ExprBinary) { + /// visit_expr_binary_mut(self, node); + /// } + /// + /// /* ... */ + /// # fn visit_attribute_mut(&mut self, node: &mut Attribute); + /// # fn visit_expr_mut(&mut self, node: &mut Expr); + /// # fn visit_bin_op_mut(&mut self, node: &mut BinOp); + /// } + /// + /// pub fn visit_expr_binary_mut(v: &mut V, node: &mut ExprBinary) + /// where + /// V: VisitMut + ?Sized, + /// { + /// for attr in &mut node.attrs { + /// v.visit_attribute_mut(attr); + /// } + /// v.visit_expr_mut(&mut *node.left); + /// v.visit_bin_op_mut(&mut node.op); + /// v.visit_expr_mut(&mut *node.right); + /// } + /// + /// /* ... */ + /// ``` + /// + /// *This module is available only if Syn is built with the `"visit-mut"` + /// feature.* + /// + ///
+ /// + /// # Example + /// + /// This mut visitor replace occurrences of u256 suffixed integer literals + /// like `999u256` with a macro invocation `bigint::u256!(999)`. + /// + /// ``` + /// // [dependencies] + /// // quote = "1.0" + /// // syn = { version = "1.0", features = ["full", "visit-mut"] } + /// + /// use quote::quote; + /// use syn::visit_mut::{self, VisitMut}; + /// use syn::{parse_quote, Expr, File, Lit, LitInt}; + /// + /// struct BigintReplace; + /// + /// impl VisitMut for BigintReplace { + /// fn visit_expr_mut(&mut self, node: &mut Expr) { + /// if let Expr::Lit(expr) = &node { + /// if let Lit::Int(int) = &expr.lit { + /// if int.suffix() == "u256" { + /// let digits = int.base10_digits(); + /// let unsuffixed: LitInt = syn::parse_str(digits).unwrap(); + /// *node = parse_quote!(bigint::u256!(#unsuffixed)); + /// return; + /// } + /// } + /// } + /// + /// // Delegate to the default impl to visit nested expressions. + /// visit_mut::visit_expr_mut(self, node); + /// } + /// } + /// + /// fn main() { + /// let code = quote! { + /// fn main() { + /// let _ = 999u256; + /// } + /// }; + /// + /// let mut syntax_tree: File = syn::parse2(code).unwrap(); + /// BigintReplace.visit_file_mut(&mut syntax_tree); + /// println!("{}", quote!(#syntax_tree)); + /// } + /// ``` + #[cfg(feature = "visit-mut")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "visit-mut")))] + #[rustfmt::skip] + pub mod visit_mut; + + /// Syntax tree traversal to transform the nodes of an owned syntax tree. + /// + /// Each method of the [`Fold`] trait is a hook that can be overridden to + /// customize the behavior when transforming the corresponding type of node. + /// By default, every method recursively visits the substructure of the + /// input by invoking the right visitor method of each of its fields. + /// + /// [`Fold`]: fold::Fold + /// + /// ``` + /// # use syn::{Attribute, BinOp, Expr, ExprBinary}; + /// # + /// pub trait Fold { + /// /* ... */ + /// + /// fn fold_expr_binary(&mut self, node: ExprBinary) -> ExprBinary { + /// fold_expr_binary(self, node) + /// } + /// + /// /* ... */ + /// # fn fold_attribute(&mut self, node: Attribute) -> Attribute; + /// # fn fold_expr(&mut self, node: Expr) -> Expr; + /// # fn fold_bin_op(&mut self, node: BinOp) -> BinOp; + /// } + /// + /// pub fn fold_expr_binary(v: &mut V, node: ExprBinary) -> ExprBinary + /// where + /// V: Fold + ?Sized, + /// { + /// ExprBinary { + /// attrs: node + /// .attrs + /// .into_iter() + /// .map(|attr| v.fold_attribute(attr)) + /// .collect(), + /// left: Box::new(v.fold_expr(*node.left)), + /// op: v.fold_bin_op(node.op), + /// right: Box::new(v.fold_expr(*node.right)), + /// } + /// } + /// + /// /* ... */ + /// ``` + /// + /// *This module is available only if Syn is built with the `"fold"` feature.* + /// + ///
+ /// + /// # Example + /// + /// This fold inserts parentheses to fully parenthesizes any expression. + /// + /// ``` + /// // [dependencies] + /// // quote = "1.0" + /// // syn = { version = "1.0", features = ["fold", "full"] } + /// + /// use quote::quote; + /// use syn::fold::{fold_expr, Fold}; + /// use syn::{token, Expr, ExprParen}; + /// + /// struct ParenthesizeEveryExpr; + /// + /// impl Fold for ParenthesizeEveryExpr { + /// fn fold_expr(&mut self, expr: Expr) -> Expr { + /// Expr::Paren(ExprParen { + /// attrs: Vec::new(), + /// expr: Box::new(fold_expr(self, expr)), + /// paren_token: token::Paren::default(), + /// }) + /// } + /// } + /// + /// fn main() { + /// let code = quote! { a() + b(1) * c.d }; + /// let expr: Expr = syn::parse2(code).unwrap(); + /// let parenthesized = ParenthesizeEveryExpr.fold_expr(expr); + /// println!("{}", quote!(#parenthesized)); + /// + /// // Output: (((a)()) + (((b)((1))) * ((c).d))) + /// } + /// ``` + #[cfg(feature = "fold")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "fold")))] + #[rustfmt::skip] + pub mod fold; + + #[cfg(feature = "clone-impls")] + #[rustfmt::skip] + mod clone; + + #[cfg(feature = "extra-traits")] + #[rustfmt::skip] + mod eq; + + #[cfg(feature = "extra-traits")] + #[rustfmt::skip] + mod hash; + + #[cfg(feature = "extra-traits")] + #[rustfmt::skip] + mod debug; + + #[cfg(any(feature = "full", feature = "derive"))] + #[path = "../gen_helper.rs"] + mod helper; +} +pub use crate::gen::*; + +// Not public API. +#[doc(hidden)] +#[path = "export.rs"] +pub mod __private; + +mod custom_keyword; +mod custom_punctuation; +mod sealed; +mod span; +mod thread; + +#[cfg(feature = "parsing")] +mod lookahead; + +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +pub mod parse; + +#[cfg(feature = "full")] +mod reserved; + +#[cfg(all(any(feature = "full", feature = "derive"), feature = "parsing"))] +mod verbatim; + +#[cfg(all(any(feature = "full", feature = "derive"), feature = "printing"))] +mod print; + +//////////////////////////////////////////////////////////////////////////////// + +mod error; +pub use crate::error::{Error, Result}; + +/// Parse tokens of source code into the chosen syntax tree node. +/// +/// This is preferred over parsing a string because tokens are able to preserve +/// information about where in the user's code they were originally written (the +/// "span" of the token), possibly allowing the compiler to produce better error +/// messages. +/// +/// This function parses a `proc_macro::TokenStream` which is the type used for +/// interop with the compiler in a procedural macro. To parse a +/// `proc_macro2::TokenStream`, use [`syn::parse2`] instead. +/// +/// [`syn::parse2`]: parse2 +/// +/// *This function is available only if Syn is built with both the `"parsing"` and +/// `"proc-macro"` features.* +/// +/// # Examples +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// use proc_macro::TokenStream; +/// use quote::quote; +/// use syn::DeriveInput; +/// +/// # const IGNORE_TOKENS: &str = stringify! { +/// #[proc_macro_derive(MyMacro)] +/// # }; +/// pub fn my_macro(input: TokenStream) -> TokenStream { +/// // Parse the tokens into a syntax tree +/// let ast: DeriveInput = syn::parse(input).unwrap(); +/// +/// // Build the output, possibly using quasi-quotation +/// let expanded = quote! { +/// /* ... */ +/// }; +/// +/// // Convert into a token stream and return it +/// expanded.into() +/// } +/// ``` +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "parsing", + feature = "proc-macro" +))] +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "proc-macro"))))] +pub fn parse(tokens: proc_macro::TokenStream) -> Result { + parse::Parser::parse(T::parse, tokens) +} + +/// Parse a proc-macro2 token stream into the chosen syntax tree node. +/// +/// This function will check that the input is fully parsed. If there are +/// any unparsed tokens at the end of the stream, an error is returned. +/// +/// This function parses a `proc_macro2::TokenStream` which is commonly useful +/// when the input comes from a node of the Syn syntax tree, for example the +/// body tokens of a [`Macro`] node. When in a procedural macro parsing the +/// `proc_macro::TokenStream` provided by the compiler, use [`syn::parse`] +/// instead. +/// +/// [`syn::parse`]: parse() +/// +/// *This function is available only if Syn is built with the `"parsing"` feature.* +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +pub fn parse2(tokens: proc_macro2::TokenStream) -> Result { + parse::Parser::parse2(T::parse, tokens) +} + +/// Parse a string of Rust code into the chosen syntax tree node. +/// +/// *This function is available only if Syn is built with the `"parsing"` feature.* +/// +/// # Hygiene +/// +/// Every span in the resulting syntax tree will be set to resolve at the macro +/// call site. +/// +/// # Examples +/// +/// ``` +/// use syn::{Expr, Result}; +/// +/// fn run() -> Result<()> { +/// let code = "assert_eq!(u8::max_value(), 255)"; +/// let expr = syn::parse_str::(code)?; +/// println!("{:#?}", expr); +/// Ok(()) +/// } +/// # +/// # run().unwrap(); +/// ``` +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +pub fn parse_str(s: &str) -> Result { + parse::Parser::parse_str(T::parse, s) +} + +// FIXME the name parse_file makes it sound like you might pass in a path to a +// file, rather than the content. +/// Parse the content of a file of Rust code. +/// +/// This is different from `syn::parse_str::(content)` in two ways: +/// +/// - It discards a leading byte order mark `\u{FEFF}` if the file has one. +/// - It preserves the shebang line of the file, such as `#!/usr/bin/env rustx`. +/// +/// If present, either of these would be an error using `from_str`. +/// +/// *This function is available only if Syn is built with the `"parsing"` and +/// `"full"` features.* +/// +/// # Examples +/// +/// ```no_run +/// use std::error::Error; +/// use std::fs::File; +/// use std::io::Read; +/// +/// fn run() -> Result<(), Box> { +/// let mut file = File::open("path/to/code.rs")?; +/// let mut content = String::new(); +/// file.read_to_string(&mut content)?; +/// +/// let ast = syn::parse_file(&content)?; +/// if let Some(shebang) = ast.shebang { +/// println!("{}", shebang); +/// } +/// println!("{} items", ast.items.len()); +/// +/// Ok(()) +/// } +/// # +/// # run().unwrap(); +/// ``` +#[cfg(all(feature = "parsing", feature = "full"))] +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "full"))))] +pub fn parse_file(mut content: &str) -> Result { + // Strip the BOM if it is present + const BOM: &str = "\u{feff}"; + if content.starts_with(BOM) { + content = &content[BOM.len()..]; + } + + let mut shebang = None; + if content.starts_with("#!") { + let rest = whitespace::skip(&content[2..]); + if !rest.starts_with('[') { + if let Some(idx) = content.find('\n') { + shebang = Some(content[..idx].to_string()); + content = &content[idx..]; + } else { + shebang = Some(content.to_string()); + content = ""; + } + } + } + + let mut file: File = parse_str(content)?; + file.shebang = shebang; + Ok(file) +} diff --git a/rust/syn/lifetime.rs b/rust/syn/lifetime.rs new file mode 100644 index 00000000000000..23b644e8ea728c --- /dev/null +++ b/rust/syn/lifetime.rs @@ -0,0 +1,156 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Ident, Span}; +use std::cmp::Ordering; +use std::fmt::{self, Display}; +use std::hash::{Hash, Hasher}; + +#[cfg(feature = "parsing")] +use crate::lookahead; + +/// A Rust lifetime: `'a`. +/// +/// Lifetime names must conform to the following rules: +/// +/// - Must start with an apostrophe. +/// - Must not consist of just an apostrophe: `'`. +/// - Character after the apostrophe must be `_` or a Unicode code point with +/// the XID_Start property. +/// - All following characters must be Unicode code points with the XID_Continue +/// property. +pub struct Lifetime { + pub apostrophe: Span, + pub ident: Ident, +} + +impl Lifetime { + /// # Panics + /// + /// Panics if the lifetime does not conform to the bulleted rules above. + /// + /// # Invocation + /// + /// ``` + /// # use proc_macro2::Span; + /// # use syn::Lifetime; + /// # + /// # fn f() -> Lifetime { + /// Lifetime::new("'a", Span::call_site()) + /// # } + /// ``` + pub fn new(symbol: &str, span: Span) -> Self { + if !symbol.starts_with('\'') { + panic!( + "lifetime name must start with apostrophe as in \"'a\", got {:?}", + symbol + ); + } + + if symbol == "'" { + panic!("lifetime name must not be empty"); + } + + if !crate::ident::xid_ok(&symbol[1..]) { + panic!("{:?} is not a valid lifetime name", symbol); + } + + Lifetime { + apostrophe: span, + ident: Ident::new(&symbol[1..], span), + } + } + + pub fn span(&self) -> Span { + self.apostrophe + .join(self.ident.span()) + .unwrap_or(self.apostrophe) + } + + pub fn set_span(&mut self, span: Span) { + self.apostrophe = span; + self.ident.set_span(span); + } +} + +impl Display for Lifetime { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + "'".fmt(formatter)?; + self.ident.fmt(formatter) + } +} + +impl Clone for Lifetime { + fn clone(&self) -> Self { + Lifetime { + apostrophe: self.apostrophe, + ident: self.ident.clone(), + } + } +} + +impl PartialEq for Lifetime { + fn eq(&self, other: &Lifetime) -> bool { + self.ident.eq(&other.ident) + } +} + +impl Eq for Lifetime {} + +impl PartialOrd for Lifetime { + fn partial_cmp(&self, other: &Lifetime) -> Option { + Some(self.cmp(other)) + } +} + +impl Ord for Lifetime { + fn cmp(&self, other: &Lifetime) -> Ordering { + self.ident.cmp(&other.ident) + } +} + +impl Hash for Lifetime { + fn hash(&self, h: &mut H) { + self.ident.hash(h); + } +} + +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[allow(non_snake_case)] +pub fn Lifetime(marker: lookahead::TokenMarker) -> Lifetime { + match marker {} +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Lifetime { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| { + cursor + .lifetime() + .ok_or_else(|| cursor.error("expected lifetime")) + }) + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::{Punct, Spacing, TokenStream}; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Lifetime { + fn to_tokens(&self, tokens: &mut TokenStream) { + let mut apostrophe = Punct::new('\'', Spacing::Joint); + apostrophe.set_span(self.apostrophe); + tokens.append(apostrophe); + self.ident.to_tokens(tokens); + } + } +} diff --git a/rust/syn/lit.rs b/rust/syn/lit.rs new file mode 100644 index 00000000000000..51229eb165f108 --- /dev/null +++ b/rust/syn/lit.rs @@ -0,0 +1,1602 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg(feature = "parsing")] +use crate::lookahead; +#[cfg(feature = "parsing")] +use crate::parse::{Parse, Parser}; +use crate::{Error, Result}; +use proc_macro2::{Ident, Literal, Span}; +#[cfg(feature = "parsing")] +use proc_macro2::{TokenStream, TokenTree}; +use std::fmt::{self, Display}; +#[cfg(feature = "extra-traits")] +use std::hash::{Hash, Hasher}; +use std::str::{self, FromStr}; + +ast_enum_of_structs! { + /// A Rust literal such as a string or integer or boolean. + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: crate::Expr#syntax-tree-enums + pub enum Lit { + /// A UTF-8 string literal: `"foo"`. + Str(LitStr), + + /// A byte string literal: `b"foo"`. + ByteStr(LitByteStr), + + /// A byte literal: `b'f'`. + Byte(LitByte), + + /// A character literal: `'a'`. + Char(LitChar), + + /// An integer literal: `1` or `1u16`. + Int(LitInt), + + /// A floating point literal: `1f64` or `1.0e10f64`. + /// + /// Must be finite. May not be infinite or NaN. + Float(LitFloat), + + /// A boolean literal: `true` or `false`. + Bool(LitBool), + + /// A raw token literal not interpreted by Syn. + Verbatim(Literal), + } +} + +ast_struct! { + /// A UTF-8 string literal: `"foo"`. + pub struct LitStr { + repr: Box, + } +} + +ast_struct! { + /// A byte string literal: `b"foo"`. + pub struct LitByteStr { + repr: Box, + } +} + +ast_struct! { + /// A byte literal: `b'f'`. + pub struct LitByte { + repr: Box, + } +} + +ast_struct! { + /// A character literal: `'a'`. + pub struct LitChar { + repr: Box, + } +} + +struct LitRepr { + token: Literal, + suffix: Box, +} + +ast_struct! { + /// An integer literal: `1` or `1u16`. + pub struct LitInt { + repr: Box, + } +} + +struct LitIntRepr { + token: Literal, + digits: Box, + suffix: Box, +} + +ast_struct! { + /// A floating point literal: `1f64` or `1.0e10f64`. + /// + /// Must be finite. May not be infinite or NaN. + pub struct LitFloat { + repr: Box, + } +} + +struct LitFloatRepr { + token: Literal, + digits: Box, + suffix: Box, +} + +ast_struct! { + /// A boolean literal: `true` or `false`. + pub struct LitBool { + pub value: bool, + pub span: Span, + } +} + +impl LitStr { + pub fn new(value: &str, span: Span) -> Self { + let mut token = Literal::string(value); + token.set_span(span); + LitStr { + repr: Box::new(LitRepr { + token, + suffix: Box::::default(), + }), + } + } + + pub fn value(&self) -> String { + let repr = self.repr.token.to_string(); + let (value, _suffix) = value::parse_lit_str(&repr); + String::from(value) + } + + /// Parse a syntax tree node from the content of this string literal. + /// + /// All spans in the syntax tree will point to the span of this `LitStr`. + /// + /// # Example + /// + /// ``` + /// use proc_macro2::Span; + /// use syn::{Attribute, Error, Ident, Lit, Meta, MetaNameValue, Path, Result}; + /// + /// // Parses the path from an attribute that looks like: + /// // + /// // #[path = "a::b::c"] + /// // + /// // or returns `None` if the input is some other attribute. + /// fn get_path(attr: &Attribute) -> Result> { + /// if !attr.path.is_ident("path") { + /// return Ok(None); + /// } + /// + /// match attr.parse_meta()? { + /// Meta::NameValue(MetaNameValue { lit: Lit::Str(lit_str), .. }) => { + /// lit_str.parse().map(Some) + /// } + /// _ => { + /// let message = "expected #[path = \"...\"]"; + /// Err(Error::new_spanned(attr, message)) + /// } + /// } + /// } + /// ``` + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse(&self) -> Result { + self.parse_with(T::parse) + } + + /// Invoke parser on the content of this string literal. + /// + /// All spans in the syntax tree will point to the span of this `LitStr`. + /// + /// # Example + /// + /// ``` + /// # use proc_macro2::Span; + /// # use syn::{LitStr, Result}; + /// # + /// # fn main() -> Result<()> { + /// # let lit_str = LitStr::new("a::b::c", Span::call_site()); + /// # + /// # const IGNORE: &str = stringify! { + /// let lit_str: LitStr = /* ... */; + /// # }; + /// + /// // Parse a string literal like "a::b::c" into a Path, not allowing + /// // generic arguments on any of the path segments. + /// let basic_path = lit_str.parse_with(syn::Path::parse_mod_style)?; + /// # + /// # Ok(()) + /// # } + /// ``` + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_with(&self, parser: F) -> Result { + use proc_macro2::Group; + + // Token stream with every span replaced by the given one. + fn respan_token_stream(stream: TokenStream, span: Span) -> TokenStream { + stream + .into_iter() + .map(|token| respan_token_tree(token, span)) + .collect() + } + + // Token tree with every span replaced by the given one. + fn respan_token_tree(mut token: TokenTree, span: Span) -> TokenTree { + match &mut token { + TokenTree::Group(g) => { + let stream = respan_token_stream(g.stream(), span); + *g = Group::new(g.delimiter(), stream); + g.set_span(span); + } + other => other.set_span(span), + } + token + } + + // Parse string literal into a token stream with every span equal to the + // original literal's span. + let mut tokens = TokenStream::from_str(&self.value())?; + tokens = respan_token_stream(tokens, self.span()); + + parser.parse2(tokens) + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl LitByteStr { + pub fn new(value: &[u8], span: Span) -> Self { + let mut token = Literal::byte_string(value); + token.set_span(span); + LitByteStr { + repr: Box::new(LitRepr { + token, + suffix: Box::::default(), + }), + } + } + + pub fn value(&self) -> Vec { + let repr = self.repr.token.to_string(); + let (value, _suffix) = value::parse_lit_byte_str(&repr); + value + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl LitByte { + pub fn new(value: u8, span: Span) -> Self { + let mut token = Literal::u8_suffixed(value); + token.set_span(span); + LitByte { + repr: Box::new(LitRepr { + token, + suffix: Box::::default(), + }), + } + } + + pub fn value(&self) -> u8 { + let repr = self.repr.token.to_string(); + let (value, _suffix) = value::parse_lit_byte(&repr); + value + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl LitChar { + pub fn new(value: char, span: Span) -> Self { + let mut token = Literal::character(value); + token.set_span(span); + LitChar { + repr: Box::new(LitRepr { + token, + suffix: Box::::default(), + }), + } + } + + pub fn value(&self) -> char { + let repr = self.repr.token.to_string(); + let (value, _suffix) = value::parse_lit_char(&repr); + value + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl LitInt { + pub fn new(repr: &str, span: Span) -> Self { + let (digits, suffix) = match value::parse_lit_int(repr) { + Some(parse) => parse, + None => panic!("Not an integer literal: `{}`", repr), + }; + + let mut token = match value::to_literal(repr, &digits, &suffix) { + Some(token) => token, + None => panic!("Unsupported integer literal: `{}`", repr), + }; + + token.set_span(span); + LitInt { + repr: Box::new(LitIntRepr { + token, + digits, + suffix, + }), + } + } + + pub fn base10_digits(&self) -> &str { + &self.repr.digits + } + + /// Parses the literal into a selected number type. + /// + /// This is equivalent to `lit.base10_digits().parse()` except that the + /// resulting errors will be correctly spanned to point to the literal token + /// in the macro input. + /// + /// ``` + /// use syn::LitInt; + /// use syn::parse::{Parse, ParseStream, Result}; + /// + /// struct Port { + /// value: u16, + /// } + /// + /// impl Parse for Port { + /// fn parse(input: ParseStream) -> Result { + /// let lit: LitInt = input.parse()?; + /// let value = lit.base10_parse::()?; + /// Ok(Port { value }) + /// } + /// } + /// ``` + pub fn base10_parse(&self) -> Result + where + N: FromStr, + N::Err: Display, + { + self.base10_digits() + .parse() + .map_err(|err| Error::new(self.span(), err)) + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl From for LitInt { + fn from(token: Literal) -> Self { + let repr = token.to_string(); + if let Some((digits, suffix)) = value::parse_lit_int(&repr) { + LitInt { + repr: Box::new(LitIntRepr { + token, + digits, + suffix, + }), + } + } else { + panic!("Not an integer literal: `{}`", repr); + } + } +} + +impl Display for LitInt { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + self.repr.token.fmt(formatter) + } +} + +impl LitFloat { + pub fn new(repr: &str, span: Span) -> Self { + let (digits, suffix) = match value::parse_lit_float(repr) { + Some(parse) => parse, + None => panic!("Not a float literal: `{}`", repr), + }; + + let mut token = match value::to_literal(repr, &digits, &suffix) { + Some(token) => token, + None => panic!("Unsupported float literal: `{}`", repr), + }; + + token.set_span(span); + LitFloat { + repr: Box::new(LitFloatRepr { + token, + digits, + suffix, + }), + } + } + + pub fn base10_digits(&self) -> &str { + &self.repr.digits + } + + pub fn base10_parse(&self) -> Result + where + N: FromStr, + N::Err: Display, + { + self.base10_digits() + .parse() + .map_err(|err| Error::new(self.span(), err)) + } + + pub fn suffix(&self) -> &str { + &self.repr.suffix + } + + pub fn span(&self) -> Span { + self.repr.token.span() + } + + pub fn set_span(&mut self, span: Span) { + self.repr.token.set_span(span); + } + + pub fn token(&self) -> Literal { + self.repr.token.clone() + } +} + +impl From for LitFloat { + fn from(token: Literal) -> Self { + let repr = token.to_string(); + if let Some((digits, suffix)) = value::parse_lit_float(&repr) { + LitFloat { + repr: Box::new(LitFloatRepr { + token, + digits, + suffix, + }), + } + } else { + panic!("Not a float literal: `{}`", repr); + } + } +} + +impl Display for LitFloat { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + self.repr.token.fmt(formatter) + } +} + +impl LitBool { + pub fn new(value: bool, span: Span) -> Self { + LitBool { value, span } + } + + pub fn value(&self) -> bool { + self.value + } + + pub fn span(&self) -> Span { + self.span + } + + pub fn set_span(&mut self, span: Span) { + self.span = span; + } + + pub fn token(&self) -> Ident { + let s = if self.value { "true" } else { "false" }; + Ident::new(s, self.span) + } +} + +#[cfg(feature = "extra-traits")] +mod debug_impls { + use super::*; + use std::fmt::{self, Debug}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitStr { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitStr") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitByteStr { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitByteStr") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitByte { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitByte") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitChar { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitChar") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitInt { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitInt") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitFloat { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitFloat") + .field("token", &format_args!("{}", self.repr.token)) + .finish() + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for LitBool { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_struct("LitBool") + .field("value", &self.value) + .finish() + } + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for LitRepr { + fn clone(&self) -> Self { + LitRepr { + token: self.token.clone(), + suffix: self.suffix.clone(), + } + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for LitIntRepr { + fn clone(&self) -> Self { + LitIntRepr { + token: self.token.clone(), + digits: self.digits.clone(), + suffix: self.suffix.clone(), + } + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for LitFloatRepr { + fn clone(&self) -> Self { + LitFloatRepr { + token: self.token.clone(), + digits: self.digits.clone(), + suffix: self.suffix.clone(), + } + } +} + +macro_rules! lit_extra_traits { + ($ty:ident) => { + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Clone for $ty { + fn clone(&self) -> Self { + $ty { + repr: self.repr.clone(), + } + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl PartialEq for $ty { + fn eq(&self, other: &Self) -> bool { + self.repr.token.to_string() == other.repr.token.to_string() + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Hash for $ty { + fn hash(&self, state: &mut H) + where + H: Hasher, + { + self.repr.token.to_string().hash(state); + } + } + + #[cfg(feature = "parsing")] + #[doc(hidden)] + #[allow(non_snake_case)] + pub fn $ty(marker: lookahead::TokenMarker) -> $ty { + match marker {} + } + }; +} + +lit_extra_traits!(LitStr); +lit_extra_traits!(LitByteStr); +lit_extra_traits!(LitByte); +lit_extra_traits!(LitChar); +lit_extra_traits!(LitInt); +lit_extra_traits!(LitFloat); + +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[allow(non_snake_case)] +pub fn LitBool(marker: lookahead::TokenMarker) -> LitBool { + match marker {} +} + +ast_enum! { + /// The style of a string literal, either plain quoted or a raw string like + /// `r##"data"##`. + pub enum StrStyle #no_visit { + /// An ordinary string like `"data"`. + Cooked, + /// A raw string like `r##"data"##`. + /// + /// The unsigned integer is the number of `#` symbols used. + Raw(usize), + } +} + +#[cfg(feature = "parsing")] +#[doc(hidden)] +#[allow(non_snake_case)] +pub fn Lit(marker: lookahead::TokenMarker) -> Lit { + match marker {} +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::buffer::Cursor; + use crate::parse::{Parse, ParseStream, Result}; + use proc_macro2::Punct; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Lit { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| { + if let Some((lit, rest)) = cursor.literal() { + return Ok((Lit::new(lit), rest)); + } + + if let Some((ident, rest)) = cursor.ident() { + let value = ident == "true"; + if value || ident == "false" { + let lit_bool = LitBool { + value, + span: ident.span(), + }; + return Ok((Lit::Bool(lit_bool), rest)); + } + } + + if let Some((punct, rest)) = cursor.punct() { + if punct.as_char() == '-' { + if let Some((lit, rest)) = parse_negative_lit(punct, rest) { + return Ok((lit, rest)); + } + } + } + + Err(cursor.error("expected literal")) + }) + } + } + + fn parse_negative_lit(neg: Punct, cursor: Cursor) -> Option<(Lit, Cursor)> { + let (lit, rest) = cursor.literal()?; + + let mut span = neg.span(); + span = span.join(lit.span()).unwrap_or(span); + + let mut repr = lit.to_string(); + repr.insert(0, '-'); + + if let Some((digits, suffix)) = value::parse_lit_int(&repr) { + if let Some(mut token) = value::to_literal(&repr, &digits, &suffix) { + token.set_span(span); + return Some(( + Lit::Int(LitInt { + repr: Box::new(LitIntRepr { + token, + digits, + suffix, + }), + }), + rest, + )); + } + } + + let (digits, suffix) = value::parse_lit_float(&repr)?; + let mut token = value::to_literal(&repr, &digits, &suffix)?; + token.set_span(span); + Some(( + Lit::Float(LitFloat { + repr: Box::new(LitFloatRepr { + token, + digits, + suffix, + }), + }), + rest, + )) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitStr { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Str(lit)) => Ok(lit), + _ => Err(head.error("expected string literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitByteStr { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::ByteStr(lit)) => Ok(lit), + _ => Err(head.error("expected byte string literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitByte { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Byte(lit)) => Ok(lit), + _ => Err(head.error("expected byte literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitChar { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Char(lit)) => Ok(lit), + _ => Err(head.error("expected character literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitInt { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Int(lit)) => Ok(lit), + _ => Err(head.error("expected integer literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitFloat { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Float(lit)) => Ok(lit), + _ => Err(head.error("expected floating point literal")), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for LitBool { + fn parse(input: ParseStream) -> Result { + let head = input.fork(); + match input.parse() { + Ok(Lit::Bool(lit)) => Ok(lit), + _ => Err(head.error("expected boolean literal")), + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitStr { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitByteStr { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitByte { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitChar { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitInt { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitFloat { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.repr.token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for LitBool { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(self.token()); + } + } +} + +mod value { + use super::*; + use crate::bigint::BigInt; + use std::char; + use std::ops::{Index, RangeFrom}; + + impl Lit { + /// Interpret a Syn literal from a proc-macro2 literal. + pub fn new(token: Literal) -> Self { + let repr = token.to_string(); + + match byte(&repr, 0) { + b'"' | b'r' => { + let (_, suffix) = parse_lit_str(&repr); + return Lit::Str(LitStr { + repr: Box::new(LitRepr { token, suffix }), + }); + } + b'b' => match byte(&repr, 1) { + b'"' | b'r' => { + let (_, suffix) = parse_lit_byte_str(&repr); + return Lit::ByteStr(LitByteStr { + repr: Box::new(LitRepr { token, suffix }), + }); + } + b'\'' => { + let (_, suffix) = parse_lit_byte(&repr); + return Lit::Byte(LitByte { + repr: Box::new(LitRepr { token, suffix }), + }); + } + _ => {} + }, + b'\'' => { + let (_, suffix) = parse_lit_char(&repr); + return Lit::Char(LitChar { + repr: Box::new(LitRepr { token, suffix }), + }); + } + b'0'..=b'9' | b'-' => { + if let Some((digits, suffix)) = parse_lit_int(&repr) { + return Lit::Int(LitInt { + repr: Box::new(LitIntRepr { + token, + digits, + suffix, + }), + }); + } + if let Some((digits, suffix)) = parse_lit_float(&repr) { + return Lit::Float(LitFloat { + repr: Box::new(LitFloatRepr { + token, + digits, + suffix, + }), + }); + } + } + b't' | b'f' => { + if repr == "true" || repr == "false" { + return Lit::Bool(LitBool { + value: repr == "true", + span: token.span(), + }); + } + } + _ => {} + } + + panic!("Unrecognized literal: `{}`", repr); + } + + pub fn suffix(&self) -> &str { + match self { + Lit::Str(lit) => lit.suffix(), + Lit::ByteStr(lit) => lit.suffix(), + Lit::Byte(lit) => lit.suffix(), + Lit::Char(lit) => lit.suffix(), + Lit::Int(lit) => lit.suffix(), + Lit::Float(lit) => lit.suffix(), + Lit::Bool(_) | Lit::Verbatim(_) => "", + } + } + + pub fn span(&self) -> Span { + match self { + Lit::Str(lit) => lit.span(), + Lit::ByteStr(lit) => lit.span(), + Lit::Byte(lit) => lit.span(), + Lit::Char(lit) => lit.span(), + Lit::Int(lit) => lit.span(), + Lit::Float(lit) => lit.span(), + Lit::Bool(lit) => lit.span, + Lit::Verbatim(lit) => lit.span(), + } + } + + pub fn set_span(&mut self, span: Span) { + match self { + Lit::Str(lit) => lit.set_span(span), + Lit::ByteStr(lit) => lit.set_span(span), + Lit::Byte(lit) => lit.set_span(span), + Lit::Char(lit) => lit.set_span(span), + Lit::Int(lit) => lit.set_span(span), + Lit::Float(lit) => lit.set_span(span), + Lit::Bool(lit) => lit.span = span, + Lit::Verbatim(lit) => lit.set_span(span), + } + } + } + + /// Get the byte at offset idx, or a default of `b'\0'` if we're looking + /// past the end of the input buffer. + pub fn byte + ?Sized>(s: &S, idx: usize) -> u8 { + let s = s.as_ref(); + if idx < s.len() { + s[idx] + } else { + 0 + } + } + + fn next_chr(s: &str) -> char { + s.chars().next().unwrap_or('\0') + } + + // Returns (content, suffix). + pub fn parse_lit_str(s: &str) -> (Box, Box) { + match byte(s, 0) { + b'"' => parse_lit_str_cooked(s), + b'r' => parse_lit_str_raw(s), + _ => unreachable!(), + } + } + + // Clippy false positive + // https://github.com/rust-lang-nursery/rust-clippy/issues/2329 + #[allow(clippy::needless_continue)] + fn parse_lit_str_cooked(mut s: &str) -> (Box, Box) { + assert_eq!(byte(s, 0), b'"'); + s = &s[1..]; + + let mut content = String::new(); + 'outer: loop { + let ch = match byte(s, 0) { + b'"' => break, + b'\\' => { + let b = byte(s, 1); + s = &s[2..]; + match b { + b'x' => { + let (byte, rest) = backslash_x(s); + s = rest; + assert!(byte <= 0x80, "Invalid \\x byte in string literal"); + char::from_u32(u32::from(byte)).unwrap() + } + b'u' => { + let (chr, rest) = backslash_u(s); + s = rest; + chr + } + b'n' => '\n', + b'r' => '\r', + b't' => '\t', + b'\\' => '\\', + b'0' => '\0', + b'\'' => '\'', + b'"' => '"', + b'\r' | b'\n' => loop { + let ch = next_chr(s); + if ch.is_whitespace() { + s = &s[ch.len_utf8()..]; + } else { + continue 'outer; + } + }, + b => panic!("unexpected byte {:?} after \\ character in byte literal", b), + } + } + b'\r' => { + assert_eq!(byte(s, 1), b'\n', "Bare CR not allowed in string"); + s = &s[2..]; + '\n' + } + _ => { + let ch = next_chr(s); + s = &s[ch.len_utf8()..]; + ch + } + }; + content.push(ch); + } + + assert!(s.starts_with('"')); + let content = content.into_boxed_str(); + let suffix = s[1..].to_owned().into_boxed_str(); + (content, suffix) + } + + fn parse_lit_str_raw(mut s: &str) -> (Box, Box) { + assert_eq!(byte(s, 0), b'r'); + s = &s[1..]; + + let mut pounds = 0; + while byte(s, pounds) == b'#' { + pounds += 1; + } + assert_eq!(byte(s, pounds), b'"'); + let close = s.rfind('"').unwrap(); + for end in s[close + 1..close + 1 + pounds].bytes() { + assert_eq!(end, b'#'); + } + + let content = s[pounds + 1..close].to_owned().into_boxed_str(); + let suffix = s[close + 1 + pounds..].to_owned().into_boxed_str(); + (content, suffix) + } + + // Returns (content, suffix). + pub fn parse_lit_byte_str(s: &str) -> (Vec, Box) { + assert_eq!(byte(s, 0), b'b'); + match byte(s, 1) { + b'"' => parse_lit_byte_str_cooked(s), + b'r' => parse_lit_byte_str_raw(s), + _ => unreachable!(), + } + } + + // Clippy false positive + // https://github.com/rust-lang-nursery/rust-clippy/issues/2329 + #[allow(clippy::needless_continue)] + fn parse_lit_byte_str_cooked(mut s: &str) -> (Vec, Box) { + assert_eq!(byte(s, 0), b'b'); + assert_eq!(byte(s, 1), b'"'); + s = &s[2..]; + + // We're going to want to have slices which don't respect codepoint boundaries. + let mut v = s.as_bytes(); + + let mut out = Vec::new(); + 'outer: loop { + let byte = match byte(v, 0) { + b'"' => break, + b'\\' => { + let b = byte(v, 1); + v = &v[2..]; + match b { + b'x' => { + let (b, rest) = backslash_x(v); + v = rest; + b + } + b'n' => b'\n', + b'r' => b'\r', + b't' => b'\t', + b'\\' => b'\\', + b'0' => b'\0', + b'\'' => b'\'', + b'"' => b'"', + b'\r' | b'\n' => loop { + let byte = byte(v, 0); + let ch = char::from_u32(u32::from(byte)).unwrap(); + if ch.is_whitespace() { + v = &v[1..]; + } else { + continue 'outer; + } + }, + b => panic!("unexpected byte {:?} after \\ character in byte literal", b), + } + } + b'\r' => { + assert_eq!(byte(v, 1), b'\n', "Bare CR not allowed in string"); + v = &v[2..]; + b'\n' + } + b => { + v = &v[1..]; + b + } + }; + out.push(byte); + } + + assert_eq!(byte(v, 0), b'"'); + let suffix = s[s.len() - v.len() + 1..].to_owned().into_boxed_str(); + (out, suffix) + } + + fn parse_lit_byte_str_raw(s: &str) -> (Vec, Box) { + assert_eq!(byte(s, 0), b'b'); + let (value, suffix) = parse_lit_str_raw(&s[1..]); + (String::from(value).into_bytes(), suffix) + } + + // Returns (value, suffix). + pub fn parse_lit_byte(s: &str) -> (u8, Box) { + assert_eq!(byte(s, 0), b'b'); + assert_eq!(byte(s, 1), b'\''); + + // We're going to want to have slices which don't respect codepoint boundaries. + let mut v = s[2..].as_bytes(); + + let b = match byte(v, 0) { + b'\\' => { + let b = byte(v, 1); + v = &v[2..]; + match b { + b'x' => { + let (b, rest) = backslash_x(v); + v = rest; + b + } + b'n' => b'\n', + b'r' => b'\r', + b't' => b'\t', + b'\\' => b'\\', + b'0' => b'\0', + b'\'' => b'\'', + b'"' => b'"', + b => panic!("unexpected byte {:?} after \\ character in byte literal", b), + } + } + b => { + v = &v[1..]; + b + } + }; + + assert_eq!(byte(v, 0), b'\''); + let suffix = s[s.len() - v.len() + 1..].to_owned().into_boxed_str(); + (b, suffix) + } + + // Returns (value, suffix). + pub fn parse_lit_char(mut s: &str) -> (char, Box) { + assert_eq!(byte(s, 0), b'\''); + s = &s[1..]; + + let ch = match byte(s, 0) { + b'\\' => { + let b = byte(s, 1); + s = &s[2..]; + match b { + b'x' => { + let (byte, rest) = backslash_x(s); + s = rest; + assert!(byte <= 0x80, "Invalid \\x byte in string literal"); + char::from_u32(u32::from(byte)).unwrap() + } + b'u' => { + let (chr, rest) = backslash_u(s); + s = rest; + chr + } + b'n' => '\n', + b'r' => '\r', + b't' => '\t', + b'\\' => '\\', + b'0' => '\0', + b'\'' => '\'', + b'"' => '"', + b => panic!("unexpected byte {:?} after \\ character in byte literal", b), + } + } + _ => { + let ch = next_chr(s); + s = &s[ch.len_utf8()..]; + ch + } + }; + assert_eq!(byte(s, 0), b'\''); + let suffix = s[1..].to_owned().into_boxed_str(); + (ch, suffix) + } + + fn backslash_x(s: &S) -> (u8, &S) + where + S: Index, Output = S> + AsRef<[u8]> + ?Sized, + { + let mut ch = 0; + let b0 = byte(s, 0); + let b1 = byte(s, 1); + ch += 0x10 + * match b0 { + b'0'..=b'9' => b0 - b'0', + b'a'..=b'f' => 10 + (b0 - b'a'), + b'A'..=b'F' => 10 + (b0 - b'A'), + _ => panic!("unexpected non-hex character after \\x"), + }; + ch += match b1 { + b'0'..=b'9' => b1 - b'0', + b'a'..=b'f' => 10 + (b1 - b'a'), + b'A'..=b'F' => 10 + (b1 - b'A'), + _ => panic!("unexpected non-hex character after \\x"), + }; + (ch, &s[2..]) + } + + fn backslash_u(mut s: &str) -> (char, &str) { + if byte(s, 0) != b'{' { + panic!("{}", "expected { after \\u"); + } + s = &s[1..]; + + let mut ch = 0; + let mut digits = 0; + loop { + let b = byte(s, 0); + let digit = match b { + b'0'..=b'9' => b - b'0', + b'a'..=b'f' => 10 + b - b'a', + b'A'..=b'F' => 10 + b - b'A', + b'_' if digits > 0 => { + s = &s[1..]; + continue; + } + b'}' if digits == 0 => panic!("invalid empty unicode escape"), + b'}' => break, + _ => panic!("unexpected non-hex character after \\u"), + }; + if digits == 6 { + panic!("overlong unicode escape (must have at most 6 hex digits)"); + } + ch *= 0x10; + ch += u32::from(digit); + digits += 1; + s = &s[1..]; + } + assert!(byte(s, 0) == b'}'); + s = &s[1..]; + + if let Some(ch) = char::from_u32(ch) { + (ch, s) + } else { + panic!("character code {:x} is not a valid unicode character", ch); + } + } + + // Returns base 10 digits and suffix. + pub fn parse_lit_int(mut s: &str) -> Option<(Box, Box)> { + let negative = byte(s, 0) == b'-'; + if negative { + s = &s[1..]; + } + + let base = match (byte(s, 0), byte(s, 1)) { + (b'0', b'x') => { + s = &s[2..]; + 16 + } + (b'0', b'o') => { + s = &s[2..]; + 8 + } + (b'0', b'b') => { + s = &s[2..]; + 2 + } + (b'0'..=b'9', _) => 10, + _ => return None, + }; + + let mut value = BigInt::new(); + 'outer: loop { + let b = byte(s, 0); + let digit = match b { + b'0'..=b'9' => b - b'0', + b'a'..=b'f' if base > 10 => b - b'a' + 10, + b'A'..=b'F' if base > 10 => b - b'A' + 10, + b'_' => { + s = &s[1..]; + continue; + } + // If looking at a floating point literal, we don't want to + // consider it an integer. + b'.' if base == 10 => return None, + b'e' | b'E' if base == 10 => { + let mut has_exp = false; + for (i, b) in s[1..].bytes().enumerate() { + match b { + b'_' => {} + b'-' | b'+' => return None, + b'0'..=b'9' => has_exp = true, + _ => { + let suffix = &s[1 + i..]; + if has_exp && crate::ident::xid_ok(suffix) { + return None; + } else { + break 'outer; + } + } + } + } + if has_exp { + return None; + } else { + break; + } + } + _ => break, + }; + + if digit >= base { + return None; + } + + value *= base; + value += digit; + s = &s[1..]; + } + + let suffix = s; + if suffix.is_empty() || crate::ident::xid_ok(suffix) { + let mut repr = value.to_string(); + if negative { + repr.insert(0, '-'); + } + Some((repr.into_boxed_str(), suffix.to_owned().into_boxed_str())) + } else { + None + } + } + + // Returns base 10 digits and suffix. + pub fn parse_lit_float(input: &str) -> Option<(Box, Box)> { + // Rust's floating point literals are very similar to the ones parsed by + // the standard library, except that rust's literals can contain + // ignorable underscores. Let's remove those underscores. + + let mut bytes = input.to_owned().into_bytes(); + + let start = (*bytes.first()? == b'-') as usize; + match bytes.get(start)? { + b'0'..=b'9' => {} + _ => return None, + } + + let mut read = start; + let mut write = start; + let mut has_dot = false; + let mut has_e = false; + let mut has_sign = false; + let mut has_exponent = false; + while read < bytes.len() { + match bytes[read] { + b'_' => { + // Don't increase write + read += 1; + continue; + } + b'0'..=b'9' => { + if has_e { + has_exponent = true; + } + bytes[write] = bytes[read]; + } + b'.' => { + if has_e || has_dot { + return None; + } + has_dot = true; + bytes[write] = b'.'; + } + b'e' | b'E' => { + match bytes[read + 1..] + .iter() + .find(|b| **b != b'_') + .unwrap_or(&b'\0') + { + b'-' | b'+' | b'0'..=b'9' => {} + _ => break, + } + if has_e { + if has_exponent { + break; + } else { + return None; + } + } + has_e = true; + bytes[write] = b'e'; + } + b'-' | b'+' => { + if has_sign || has_exponent || !has_e { + return None; + } + has_sign = true; + if bytes[read] == b'-' { + bytes[write] = bytes[read]; + } else { + // Omit '+' + read += 1; + continue; + } + } + _ => break, + } + read += 1; + write += 1; + } + + if has_e && !has_exponent { + return None; + } + + let mut digits = String::from_utf8(bytes).unwrap(); + let suffix = digits.split_off(read); + digits.truncate(write); + if suffix.is_empty() || crate::ident::xid_ok(&suffix) { + Some((digits.into_boxed_str(), suffix.into_boxed_str())) + } else { + None + } + } + + #[allow(clippy::unnecessary_wraps)] + pub fn to_literal(repr: &str, digits: &str, suffix: &str) -> Option { + #[cfg(syn_no_negative_literal_parse)] + { + // Rustc older than https://github.com/rust-lang/rust/pull/87262. + if repr.starts_with('-') { + let f64_parse_finite = || digits.parse().ok().filter(|x: &f64| x.is_finite()); + let f32_parse_finite = || digits.parse().ok().filter(|x: &f32| x.is_finite()); + return if suffix == "f64" { + f64_parse_finite().map(Literal::f64_suffixed) + } else if suffix == "f32" { + f32_parse_finite().map(Literal::f32_suffixed) + } else if suffix == "i64" { + digits.parse().ok().map(Literal::i64_suffixed) + } else if suffix == "i32" { + digits.parse().ok().map(Literal::i32_suffixed) + } else if suffix == "i16" { + digits.parse().ok().map(Literal::i16_suffixed) + } else if suffix == "i8" { + digits.parse().ok().map(Literal::i8_suffixed) + } else if !suffix.is_empty() { + None + } else if digits.contains('.') { + f64_parse_finite().map(Literal::f64_unsuffixed) + } else { + digits.parse().ok().map(Literal::i64_unsuffixed) + }; + } + } + let _ = digits; + let _ = suffix; + Some(repr.parse::().unwrap()) + } +} diff --git a/rust/syn/lookahead.rs b/rust/syn/lookahead.rs new file mode 100644 index 00000000000000..118fdbbe11ff95 --- /dev/null +++ b/rust/syn/lookahead.rs @@ -0,0 +1,171 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::buffer::Cursor; +use crate::error::{self, Error}; +use crate::sealed::lookahead::Sealed; +use crate::span::IntoSpans; +use crate::token::Token; +use proc_macro2::{Delimiter, Span}; +use std::cell::RefCell; + +/// Support for checking the next token in a stream to decide how to parse. +/// +/// An important advantage over [`ParseStream::peek`] is that here we +/// automatically construct an appropriate error message based on the token +/// alternatives that get peeked. If you are producing your own error message, +/// go ahead and use `ParseStream::peek` instead. +/// +/// Use [`ParseStream::lookahead1`] to construct this object. +/// +/// [`ParseStream::peek`]: crate::parse::ParseBuffer::peek +/// [`ParseStream::lookahead1`]: crate::parse::ParseBuffer::lookahead1 +/// +/// Consuming tokens from the source stream after constructing a lookahead +/// object does not also advance the lookahead object. +/// +/// # Example +/// +/// ``` +/// use syn::{ConstParam, Ident, Lifetime, LifetimeDef, Result, Token, TypeParam}; +/// use syn::parse::{Parse, ParseStream}; +/// +/// // A generic parameter, a single one of the comma-separated elements inside +/// // angle brackets in: +/// // +/// // fn f() { ... } +/// // +/// // On invalid input, lookahead gives us a reasonable error message. +/// // +/// // error: expected one of: identifier, lifetime, `const` +/// // | +/// // 5 | fn f() {} +/// // | ^ +/// enum GenericParam { +/// Type(TypeParam), +/// Lifetime(LifetimeDef), +/// Const(ConstParam), +/// } +/// +/// impl Parse for GenericParam { +/// fn parse(input: ParseStream) -> Result { +/// let lookahead = input.lookahead1(); +/// if lookahead.peek(Ident) { +/// input.parse().map(GenericParam::Type) +/// } else if lookahead.peek(Lifetime) { +/// input.parse().map(GenericParam::Lifetime) +/// } else if lookahead.peek(Token![const]) { +/// input.parse().map(GenericParam::Const) +/// } else { +/// Err(lookahead.error()) +/// } +/// } +/// } +/// ``` +pub struct Lookahead1<'a> { + scope: Span, + cursor: Cursor<'a>, + comparisons: RefCell>, +} + +pub fn new(scope: Span, cursor: Cursor) -> Lookahead1 { + Lookahead1 { + scope, + cursor, + comparisons: RefCell::new(Vec::new()), + } +} + +fn peek_impl( + lookahead: &Lookahead1, + peek: fn(Cursor) -> bool, + display: fn() -> &'static str, +) -> bool { + if peek(lookahead.cursor) { + return true; + } + lookahead.comparisons.borrow_mut().push(display()); + false +} + +impl<'a> Lookahead1<'a> { + /// Looks at the next token in the parse stream to determine whether it + /// matches the requested type of token. + /// + /// # Syntax + /// + /// Note that this method does not use turbofish syntax. Pass the peek type + /// inside of parentheses. + /// + /// - `input.peek(Token![struct])` + /// - `input.peek(Token![==])` + /// - `input.peek(Ident)` *(does not accept keywords)* + /// - `input.peek(Ident::peek_any)` + /// - `input.peek(Lifetime)` + /// - `input.peek(token::Brace)` + pub fn peek(&self, token: T) -> bool { + let _ = token; + peek_impl(self, T::Token::peek, T::Token::display) + } + + /// Triggers an error at the current position of the parse stream. + /// + /// The error message will identify all of the expected token types that + /// have been peeked against this lookahead instance. + pub fn error(self) -> Error { + let comparisons = self.comparisons.borrow(); + match comparisons.len() { + 0 => { + if self.cursor.eof() { + Error::new(self.scope, "unexpected end of input") + } else { + Error::new(self.cursor.span(), "unexpected token") + } + } + 1 => { + let message = format!("expected {}", comparisons[0]); + error::new_at(self.scope, self.cursor, message) + } + 2 => { + let message = format!("expected {} or {}", comparisons[0], comparisons[1]); + error::new_at(self.scope, self.cursor, message) + } + _ => { + let join = comparisons.join(", "); + let message = format!("expected one of: {}", join); + error::new_at(self.scope, self.cursor, message) + } + } + } +} + +/// Types that can be parsed by looking at just one token. +/// +/// Use [`ParseStream::peek`] to peek one of these types in a parse stream +/// without consuming it from the stream. +/// +/// This trait is sealed and cannot be implemented for types outside of Syn. +/// +/// [`ParseStream::peek`]: crate::parse::ParseBuffer::peek +pub trait Peek: Sealed { + // Not public API. + #[doc(hidden)] + type Token: Token; +} + +impl T, T: Token> Peek for F { + type Token = T; +} + +pub enum TokenMarker {} + +impl IntoSpans for TokenMarker { + fn into_spans(self) -> S { + match self {} + } +} + +pub fn is_delimiter(cursor: Cursor, delimiter: Delimiter) -> bool { + cursor.group(delimiter).is_some() +} + +impl T, T: Token> Sealed for F {} diff --git a/rust/syn/mac.rs b/rust/syn/mac.rs new file mode 100644 index 00000000000000..99dd6accd95d24 --- /dev/null +++ b/rust/syn/mac.rs @@ -0,0 +1,221 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::token::{Brace, Bracket, Paren}; +use proc_macro2::TokenStream; +#[cfg(feature = "parsing")] +use proc_macro2::{Delimiter, Group, Span, TokenTree}; + +#[cfg(feature = "parsing")] +use crate::parse::{Parse, ParseStream, Parser, Result}; + +ast_struct! { + /// A macro invocation: `println!("{}", mac)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Macro { + pub path: Path, + pub bang_token: Token![!], + pub delimiter: MacroDelimiter, + pub tokens: TokenStream, + } +} + +ast_enum! { + /// A grouping token that surrounds a macro body: `m!(...)` or `m!{...}` or `m![...]`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum MacroDelimiter { + Paren(Paren), + Brace(Brace), + Bracket(Bracket), + } +} + +#[cfg(feature = "parsing")] +fn delimiter_span_close(macro_delimiter: &MacroDelimiter) -> Span { + let delimiter = match macro_delimiter { + MacroDelimiter::Paren(_) => Delimiter::Parenthesis, + MacroDelimiter::Brace(_) => Delimiter::Brace, + MacroDelimiter::Bracket(_) => Delimiter::Bracket, + }; + let mut group = Group::new(delimiter, TokenStream::new()); + group.set_span(match macro_delimiter { + MacroDelimiter::Paren(token) => token.span, + MacroDelimiter::Brace(token) => token.span, + MacroDelimiter::Bracket(token) => token.span, + }); + group.span_close() +} + +impl Macro { + /// Parse the tokens within the macro invocation's delimiters into a syntax + /// tree. + /// + /// This is equivalent to `syn::parse2::(mac.tokens)` except that it + /// produces a more useful span when `tokens` is empty. + /// + /// # Example + /// + /// ``` + /// use syn::{parse_quote, Expr, ExprLit, Ident, Lit, LitStr, Macro, Token}; + /// use syn::ext::IdentExt; + /// use syn::parse::{Error, Parse, ParseStream, Result}; + /// use syn::punctuated::Punctuated; + /// + /// // The arguments expected by libcore's format_args macro, and as a + /// // result most other formatting and printing macros like println. + /// // + /// // println!("{} is {number:.prec$}", "x", prec=5, number=0.01) + /// struct FormatArgs { + /// format_string: Expr, + /// positional_args: Vec, + /// named_args: Vec<(Ident, Expr)>, + /// } + /// + /// impl Parse for FormatArgs { + /// fn parse(input: ParseStream) -> Result { + /// let format_string: Expr; + /// let mut positional_args = Vec::new(); + /// let mut named_args = Vec::new(); + /// + /// format_string = input.parse()?; + /// while !input.is_empty() { + /// input.parse::()?; + /// if input.is_empty() { + /// break; + /// } + /// if input.peek(Ident::peek_any) && input.peek2(Token![=]) { + /// while !input.is_empty() { + /// let name: Ident = input.call(Ident::parse_any)?; + /// input.parse::()?; + /// let value: Expr = input.parse()?; + /// named_args.push((name, value)); + /// if input.is_empty() { + /// break; + /// } + /// input.parse::()?; + /// } + /// break; + /// } + /// positional_args.push(input.parse()?); + /// } + /// + /// Ok(FormatArgs { + /// format_string, + /// positional_args, + /// named_args, + /// }) + /// } + /// } + /// + /// // Extract the first argument, the format string literal, from an + /// // invocation of a formatting or printing macro. + /// fn get_format_string(m: &Macro) -> Result { + /// let args: FormatArgs = m.parse_body()?; + /// match args.format_string { + /// Expr::Lit(ExprLit { lit: Lit::Str(lit), .. }) => Ok(lit), + /// other => { + /// // First argument was not a string literal expression. + /// // Maybe something like: println!(concat!(...), ...) + /// Err(Error::new_spanned(other, "format string must be a string literal")) + /// } + /// } + /// } + /// + /// fn main() { + /// let invocation = parse_quote! { + /// println!("{:?}", Instant::now()) + /// }; + /// let lit = get_format_string(&invocation).unwrap(); + /// assert_eq!(lit.value(), "{:?}"); + /// } + /// ``` + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_body(&self) -> Result { + self.parse_body_with(T::parse) + } + + /// Parse the tokens within the macro invocation's delimiters using the + /// given parser. + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_body_with(&self, parser: F) -> Result { + let scope = delimiter_span_close(&self.delimiter); + crate::parse::parse_scoped(parser, scope, self.tokens.clone()) + } +} + +#[cfg(feature = "parsing")] +pub fn parse_delimiter(input: ParseStream) -> Result<(MacroDelimiter, TokenStream)> { + input.step(|cursor| { + if let Some((TokenTree::Group(g), rest)) = cursor.token_tree() { + let span = g.span(); + let delimiter = match g.delimiter() { + Delimiter::Parenthesis => MacroDelimiter::Paren(Paren(span)), + Delimiter::Brace => MacroDelimiter::Brace(Brace(span)), + Delimiter::Bracket => MacroDelimiter::Bracket(Bracket(span)), + Delimiter::None => { + return Err(cursor.error("expected delimiter")); + } + }; + Ok(((delimiter, g.stream()), rest)) + } else { + Err(cursor.error("expected delimiter")) + } + }) +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Macro { + fn parse(input: ParseStream) -> Result { + let tokens; + Ok(Macro { + path: input.call(Path::parse_mod_style)?, + bang_token: input.parse()?, + delimiter: { + let (delimiter, content) = parse_delimiter(input)?; + tokens = content; + delimiter + }, + tokens, + }) + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::ToTokens; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Macro { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.path.to_tokens(tokens); + self.bang_token.to_tokens(tokens); + match &self.delimiter { + MacroDelimiter::Paren(paren) => { + paren.surround(tokens, |tokens| self.tokens.to_tokens(tokens)); + } + MacroDelimiter::Brace(brace) => { + brace.surround(tokens, |tokens| self.tokens.to_tokens(tokens)); + } + MacroDelimiter::Bracket(bracket) => { + bracket.surround(tokens, |tokens| self.tokens.to_tokens(tokens)); + } + } + } + } +} diff --git a/rust/syn/macros.rs b/rust/syn/macros.rs new file mode 100644 index 00000000000000..86e64dc38992c6 --- /dev/null +++ b/rust/syn/macros.rs @@ -0,0 +1,179 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg_attr( + not(any(feature = "full", feature = "derive")), + allow(unknown_lints, unused_macro_rules) +)] +macro_rules! ast_struct { + ( + [$($attrs_pub:tt)*] + struct $name:ident #full $($rest:tt)* + ) => { + #[cfg(feature = "full")] + $($attrs_pub)* struct $name $($rest)* + + #[cfg(not(feature = "full"))] + $($attrs_pub)* struct $name { + _noconstruct: ::std::marker::PhantomData<::proc_macro2::Span>, + } + + #[cfg(all(not(feature = "full"), feature = "printing"))] + impl ::quote::ToTokens for $name { + fn to_tokens(&self, _: &mut ::proc_macro2::TokenStream) { + unreachable!() + } + } + }; + + ( + [$($attrs_pub:tt)*] + struct $name:ident $($rest:tt)* + ) => { + $($attrs_pub)* struct $name $($rest)* + }; + + ($($t:tt)*) => { + strip_attrs_pub!(ast_struct!($($t)*)); + }; +} + +macro_rules! ast_enum { + // Drop the `#no_visit` attribute, if present. + ( + [$($attrs_pub:tt)*] + enum $name:ident #no_visit $($rest:tt)* + ) => ( + ast_enum!([$($attrs_pub)*] enum $name $($rest)*); + ); + + ( + [$($attrs_pub:tt)*] + enum $name:ident $($rest:tt)* + ) => ( + $($attrs_pub)* enum $name $($rest)* + ); + + ($($t:tt)*) => { + strip_attrs_pub!(ast_enum!($($t)*)); + }; +} + +macro_rules! ast_enum_of_structs { + ( + $(#[$enum_attr:meta])* + $pub:ident $enum:ident $name:ident $body:tt + $($remaining:tt)* + ) => { + ast_enum!($(#[$enum_attr])* $pub $enum $name $body); + ast_enum_of_structs_impl!($pub $enum $name $body $($remaining)*); + }; +} + +macro_rules! ast_enum_of_structs_impl { + ( + $pub:ident $enum:ident $name:ident { + $( + $(#[cfg $cfg_attr:tt])* + $(#[doc $($doc_attr:tt)*])* + $variant:ident $( ($($member:ident)::+) )*, + )* + } + + $($remaining:tt)* + ) => { + check_keyword_matches!(pub $pub); + check_keyword_matches!(enum $enum); + + $($( + ast_enum_from_struct!($name::$variant, $($member)::+); + )*)* + + #[cfg(feature = "printing")] + generate_to_tokens! { + $($remaining)* + () + tokens + $name { + $( + $(#[cfg $cfg_attr])* + $(#[doc $($doc_attr)*])* + $variant $($($member)::+)*, + )* + } + } + }; +} + +macro_rules! ast_enum_from_struct { + // No From for verbatim variants. + ($name:ident::Verbatim, $member:ident) => {}; + + ($name:ident::$variant:ident, $member:ident) => { + impl From<$member> for $name { + fn from(e: $member) -> $name { + $name::$variant(e) + } + } + }; +} + +#[cfg(feature = "printing")] +#[cfg_attr( + not(any(feature = "full", feature = "derive")), + allow(unknown_lints, unused_macro_rules) +)] +macro_rules! generate_to_tokens { + (do_not_generate_to_tokens $($foo:tt)*) => (); + + ( + ($($arms:tt)*) $tokens:ident $name:ident { + $(#[cfg $cfg_attr:tt])* + $(#[doc $($doc_attr:tt)*])* + $variant:ident, + $($next:tt)* + } + ) => { + generate_to_tokens!( + ($($arms)* $(#[cfg $cfg_attr])* $name::$variant => {}) + $tokens $name { $($next)* } + ); + }; + + ( + ($($arms:tt)*) $tokens:ident $name:ident { + $(#[cfg $cfg_attr:tt])* + $(#[doc $($doc_attr:tt)*])* + $variant:ident $member:ident, + $($next:tt)* + } + ) => { + generate_to_tokens!( + ($($arms)* $(#[cfg $cfg_attr])* $name::$variant(_e) => _e.to_tokens($tokens),) + $tokens $name { $($next)* } + ); + }; + + (($($arms:tt)*) $tokens:ident $name:ident {}) => { + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ::quote::ToTokens for $name { + fn to_tokens(&self, $tokens: &mut ::proc_macro2::TokenStream) { + match self { + $($arms)* + } + } + } + }; +} + +macro_rules! strip_attrs_pub { + ($mac:ident!($(#[$m:meta])* $pub:ident $($t:tt)*)) => { + check_keyword_matches!(pub $pub); + + $mac!([$(#[$m])* $pub] $($t)*); + }; +} + +macro_rules! check_keyword_matches { + (enum enum) => {}; + (pub pub) => {}; +} diff --git a/rust/syn/op.rs b/rust/syn/op.rs new file mode 100644 index 00000000000000..96ccd7d3e1f3d1 --- /dev/null +++ b/rust/syn/op.rs @@ -0,0 +1,236 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +ast_enum! { + /// A binary operator: `+`, `+=`, `&`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum BinOp { + /// The `+` operator (addition) + Add(Token![+]), + /// The `-` operator (subtraction) + Sub(Token![-]), + /// The `*` operator (multiplication) + Mul(Token![*]), + /// The `/` operator (division) + Div(Token![/]), + /// The `%` operator (modulus) + Rem(Token![%]), + /// The `&&` operator (logical and) + And(Token![&&]), + /// The `||` operator (logical or) + Or(Token![||]), + /// The `^` operator (bitwise xor) + BitXor(Token![^]), + /// The `&` operator (bitwise and) + BitAnd(Token![&]), + /// The `|` operator (bitwise or) + BitOr(Token![|]), + /// The `<<` operator (shift left) + Shl(Token![<<]), + /// The `>>` operator (shift right) + Shr(Token![>>]), + /// The `==` operator (equality) + Eq(Token![==]), + /// The `<` operator (less than) + Lt(Token![<]), + /// The `<=` operator (less than or equal to) + Le(Token![<=]), + /// The `!=` operator (not equal to) + Ne(Token![!=]), + /// The `>=` operator (greater than or equal to) + Ge(Token![>=]), + /// The `>` operator (greater than) + Gt(Token![>]), + /// The `+=` operator + AddEq(Token![+=]), + /// The `-=` operator + SubEq(Token![-=]), + /// The `*=` operator + MulEq(Token![*=]), + /// The `/=` operator + DivEq(Token![/=]), + /// The `%=` operator + RemEq(Token![%=]), + /// The `^=` operator + BitXorEq(Token![^=]), + /// The `&=` operator + BitAndEq(Token![&=]), + /// The `|=` operator + BitOrEq(Token![|=]), + /// The `<<=` operator + ShlEq(Token![<<=]), + /// The `>>=` operator + ShrEq(Token![>>=]), + } +} + +ast_enum! { + /// A unary operator: `*`, `!`, `-`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum UnOp { + /// The `*` operator for dereferencing + Deref(Token![*]), + /// The `!` operator for logical inversion + Not(Token![!]), + /// The `-` operator for negation + Neg(Token![-]), + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::parse::{Parse, ParseStream, Result}; + + fn parse_binop(input: ParseStream) -> Result { + if input.peek(Token![&&]) { + input.parse().map(BinOp::And) + } else if input.peek(Token![||]) { + input.parse().map(BinOp::Or) + } else if input.peek(Token![<<]) { + input.parse().map(BinOp::Shl) + } else if input.peek(Token![>>]) { + input.parse().map(BinOp::Shr) + } else if input.peek(Token![==]) { + input.parse().map(BinOp::Eq) + } else if input.peek(Token![<=]) { + input.parse().map(BinOp::Le) + } else if input.peek(Token![!=]) { + input.parse().map(BinOp::Ne) + } else if input.peek(Token![>=]) { + input.parse().map(BinOp::Ge) + } else if input.peek(Token![+]) { + input.parse().map(BinOp::Add) + } else if input.peek(Token![-]) { + input.parse().map(BinOp::Sub) + } else if input.peek(Token![*]) { + input.parse().map(BinOp::Mul) + } else if input.peek(Token![/]) { + input.parse().map(BinOp::Div) + } else if input.peek(Token![%]) { + input.parse().map(BinOp::Rem) + } else if input.peek(Token![^]) { + input.parse().map(BinOp::BitXor) + } else if input.peek(Token![&]) { + input.parse().map(BinOp::BitAnd) + } else if input.peek(Token![|]) { + input.parse().map(BinOp::BitOr) + } else if input.peek(Token![<]) { + input.parse().map(BinOp::Lt) + } else if input.peek(Token![>]) { + input.parse().map(BinOp::Gt) + } else { + Err(input.error("expected binary operator")) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for BinOp { + #[cfg(not(feature = "full"))] + fn parse(input: ParseStream) -> Result { + parse_binop(input) + } + + #[cfg(feature = "full")] + fn parse(input: ParseStream) -> Result { + if input.peek(Token![+=]) { + input.parse().map(BinOp::AddEq) + } else if input.peek(Token![-=]) { + input.parse().map(BinOp::SubEq) + } else if input.peek(Token![*=]) { + input.parse().map(BinOp::MulEq) + } else if input.peek(Token![/=]) { + input.parse().map(BinOp::DivEq) + } else if input.peek(Token![%=]) { + input.parse().map(BinOp::RemEq) + } else if input.peek(Token![^=]) { + input.parse().map(BinOp::BitXorEq) + } else if input.peek(Token![&=]) { + input.parse().map(BinOp::BitAndEq) + } else if input.peek(Token![|=]) { + input.parse().map(BinOp::BitOrEq) + } else if input.peek(Token![<<=]) { + input.parse().map(BinOp::ShlEq) + } else if input.peek(Token![>>=]) { + input.parse().map(BinOp::ShrEq) + } else { + parse_binop(input) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for UnOp { + fn parse(input: ParseStream) -> Result { + let lookahead = input.lookahead1(); + if lookahead.peek(Token![*]) { + input.parse().map(UnOp::Deref) + } else if lookahead.peek(Token![!]) { + input.parse().map(UnOp::Not) + } else if lookahead.peek(Token![-]) { + input.parse().map(UnOp::Neg) + } else { + Err(lookahead.error()) + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::ToTokens; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for BinOp { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + BinOp::Add(t) => t.to_tokens(tokens), + BinOp::Sub(t) => t.to_tokens(tokens), + BinOp::Mul(t) => t.to_tokens(tokens), + BinOp::Div(t) => t.to_tokens(tokens), + BinOp::Rem(t) => t.to_tokens(tokens), + BinOp::And(t) => t.to_tokens(tokens), + BinOp::Or(t) => t.to_tokens(tokens), + BinOp::BitXor(t) => t.to_tokens(tokens), + BinOp::BitAnd(t) => t.to_tokens(tokens), + BinOp::BitOr(t) => t.to_tokens(tokens), + BinOp::Shl(t) => t.to_tokens(tokens), + BinOp::Shr(t) => t.to_tokens(tokens), + BinOp::Eq(t) => t.to_tokens(tokens), + BinOp::Lt(t) => t.to_tokens(tokens), + BinOp::Le(t) => t.to_tokens(tokens), + BinOp::Ne(t) => t.to_tokens(tokens), + BinOp::Ge(t) => t.to_tokens(tokens), + BinOp::Gt(t) => t.to_tokens(tokens), + BinOp::AddEq(t) => t.to_tokens(tokens), + BinOp::SubEq(t) => t.to_tokens(tokens), + BinOp::MulEq(t) => t.to_tokens(tokens), + BinOp::DivEq(t) => t.to_tokens(tokens), + BinOp::RemEq(t) => t.to_tokens(tokens), + BinOp::BitXorEq(t) => t.to_tokens(tokens), + BinOp::BitAndEq(t) => t.to_tokens(tokens), + BinOp::BitOrEq(t) => t.to_tokens(tokens), + BinOp::ShlEq(t) => t.to_tokens(tokens), + BinOp::ShrEq(t) => t.to_tokens(tokens), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for UnOp { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + UnOp::Deref(t) => t.to_tokens(tokens), + UnOp::Not(t) => t.to_tokens(tokens), + UnOp::Neg(t) => t.to_tokens(tokens), + } + } + } +} diff --git a/rust/syn/parse.rs b/rust/syn/parse.rs new file mode 100644 index 00000000000000..aa07edc885dcb0 --- /dev/null +++ b/rust/syn/parse.rs @@ -0,0 +1,1316 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Parsing interface for parsing a token stream into a syntax tree node. +//! +//! Parsing in Syn is built on parser functions that take in a [`ParseStream`] +//! and produce a [`Result`] where `T` is some syntax tree node. Underlying +//! these parser functions is a lower level mechanism built around the +//! [`Cursor`] type. `Cursor` is a cheaply copyable cursor over a range of +//! tokens in a token stream. +//! +//! [`Result`]: Result +//! [`Cursor`]: crate::buffer::Cursor +//! +//! # Example +//! +//! Here is a snippet of parsing code to get a feel for the style of the +//! library. We define data structures for a subset of Rust syntax including +//! enums (not shown) and structs, then provide implementations of the [`Parse`] +//! trait to parse these syntax tree data structures from a token stream. +//! +//! Once `Parse` impls have been defined, they can be called conveniently from a +//! procedural macro through [`parse_macro_input!`] as shown at the bottom of +//! the snippet. If the caller provides syntactically invalid input to the +//! procedural macro, they will receive a helpful compiler error message +//! pointing out the exact token that triggered the failure to parse. +//! +//! [`parse_macro_input!`]: crate::parse_macro_input! +//! +//! ``` +//! # extern crate proc_macro; +//! # +//! use proc_macro::TokenStream; +//! use syn::{braced, parse_macro_input, token, Field, Ident, Result, Token}; +//! use syn::parse::{Parse, ParseStream}; +//! use syn::punctuated::Punctuated; +//! +//! enum Item { +//! Struct(ItemStruct), +//! Enum(ItemEnum), +//! } +//! +//! struct ItemStruct { +//! struct_token: Token![struct], +//! ident: Ident, +//! brace_token: token::Brace, +//! fields: Punctuated, +//! } +//! # +//! # enum ItemEnum {} +//! +//! impl Parse for Item { +//! fn parse(input: ParseStream) -> Result { +//! let lookahead = input.lookahead1(); +//! if lookahead.peek(Token![struct]) { +//! input.parse().map(Item::Struct) +//! } else if lookahead.peek(Token![enum]) { +//! input.parse().map(Item::Enum) +//! } else { +//! Err(lookahead.error()) +//! } +//! } +//! } +//! +//! impl Parse for ItemStruct { +//! fn parse(input: ParseStream) -> Result { +//! let content; +//! Ok(ItemStruct { +//! struct_token: input.parse()?, +//! ident: input.parse()?, +//! brace_token: braced!(content in input), +//! fields: content.parse_terminated(Field::parse_named)?, +//! }) +//! } +//! } +//! # +//! # impl Parse for ItemEnum { +//! # fn parse(input: ParseStream) -> Result { +//! # unimplemented!() +//! # } +//! # } +//! +//! # const IGNORE: &str = stringify! { +//! #[proc_macro] +//! # }; +//! pub fn my_macro(tokens: TokenStream) -> TokenStream { +//! let input = parse_macro_input!(tokens as Item); +//! +//! /* ... */ +//! # "".parse().unwrap() +//! } +//! ``` +//! +//! # The `syn::parse*` functions +//! +//! The [`syn::parse`], [`syn::parse2`], and [`syn::parse_str`] functions serve +//! as an entry point for parsing syntax tree nodes that can be parsed in an +//! obvious default way. These functions can return any syntax tree node that +//! implements the [`Parse`] trait, which includes most types in Syn. +//! +//! [`syn::parse`]: crate::parse() +//! [`syn::parse2`]: crate::parse2() +//! [`syn::parse_str`]: crate::parse_str() +//! +//! ``` +//! use syn::Type; +//! +//! # fn run_parser() -> syn::Result<()> { +//! let t: Type = syn::parse_str("std::collections::HashMap")?; +//! # Ok(()) +//! # } +//! # +//! # run_parser().unwrap(); +//! ``` +//! +//! The [`parse_quote!`] macro also uses this approach. +//! +//! [`parse_quote!`]: crate::parse_quote! +//! +//! # The `Parser` trait +//! +//! Some types can be parsed in several ways depending on context. For example +//! an [`Attribute`] can be either "outer" like `#[...]` or "inner" like +//! `#![...]` and parsing the wrong one would be a bug. Similarly [`Punctuated`] +//! may or may not allow trailing punctuation, and parsing it the wrong way +//! would either reject valid input or accept invalid input. +//! +//! [`Attribute`]: crate::Attribute +//! [`Punctuated`]: crate::punctuated +//! +//! The `Parse` trait is not implemented in these cases because there is no good +//! behavior to consider the default. +//! +//! ```compile_fail +//! # extern crate proc_macro; +//! # +//! # use syn::punctuated::Punctuated; +//! # use syn::{PathSegment, Result, Token}; +//! # +//! # fn f(tokens: proc_macro::TokenStream) -> Result<()> { +//! # +//! // Can't parse `Punctuated` without knowing whether trailing punctuation +//! // should be allowed in this context. +//! let path: Punctuated = syn::parse(tokens)?; +//! # +//! # Ok(()) +//! # } +//! ``` +//! +//! In these cases the types provide a choice of parser functions rather than a +//! single `Parse` implementation, and those parser functions can be invoked +//! through the [`Parser`] trait. +//! +//! +//! ``` +//! # extern crate proc_macro; +//! # +//! use proc_macro::TokenStream; +//! use syn::parse::Parser; +//! use syn::punctuated::Punctuated; +//! use syn::{Attribute, Expr, PathSegment, Result, Token}; +//! +//! fn call_some_parser_methods(input: TokenStream) -> Result<()> { +//! // Parse a nonempty sequence of path segments separated by `::` punctuation +//! // with no trailing punctuation. +//! let tokens = input.clone(); +//! let parser = Punctuated::::parse_separated_nonempty; +//! let _path = parser.parse(tokens)?; +//! +//! // Parse a possibly empty sequence of expressions terminated by commas with +//! // an optional trailing punctuation. +//! let tokens = input.clone(); +//! let parser = Punctuated::::parse_terminated; +//! let _args = parser.parse(tokens)?; +//! +//! // Parse zero or more outer attributes but not inner attributes. +//! let tokens = input.clone(); +//! let parser = Attribute::parse_outer; +//! let _attrs = parser.parse(tokens)?; +//! +//! Ok(()) +//! } +//! ``` +//! +//! --- +//! +//! *This module is available only if Syn is built with the `"parsing"` feature.* + +#[path = "discouraged.rs"] +pub mod discouraged; + +use crate::buffer::{Cursor, TokenBuffer}; +use crate::error; +use crate::lookahead; +#[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" +))] +use crate::proc_macro; +use crate::punctuated::Punctuated; +use crate::token::Token; +use proc_macro2::{self, Delimiter, Group, Literal, Punct, Span, TokenStream, TokenTree}; +use std::cell::Cell; +use std::fmt::{self, Debug, Display}; +#[cfg(feature = "extra-traits")] +use std::hash::{Hash, Hasher}; +use std::marker::PhantomData; +use std::mem; +use std::ops::Deref; +use std::rc::Rc; +use std::str::FromStr; + +pub use crate::error::{Error, Result}; +pub use crate::lookahead::{Lookahead1, Peek}; + +/// Parsing interface implemented by all types that can be parsed in a default +/// way from a token stream. +/// +/// Refer to the [module documentation] for details about implementing and using +/// the `Parse` trait. +/// +/// [module documentation]: self +pub trait Parse: Sized { + fn parse(input: ParseStream) -> Result; +} + +/// Input to a Syn parser function. +/// +/// See the methods of this type under the documentation of [`ParseBuffer`]. For +/// an overview of parsing in Syn, refer to the [module documentation]. +/// +/// [module documentation]: self +pub type ParseStream<'a> = &'a ParseBuffer<'a>; + +/// Cursor position within a buffered token stream. +/// +/// This type is more commonly used through the type alias [`ParseStream`] which +/// is an alias for `&ParseBuffer`. +/// +/// `ParseStream` is the input type for all parser functions in Syn. They have +/// the signature `fn(ParseStream) -> Result`. +/// +/// ## Calling a parser function +/// +/// There is no public way to construct a `ParseBuffer`. Instead, if you are +/// looking to invoke a parser function that requires `ParseStream` as input, +/// you will need to go through one of the public parsing entry points. +/// +/// - The [`parse_macro_input!`] macro if parsing input of a procedural macro; +/// - One of [the `syn::parse*` functions][syn-parse]; or +/// - A method of the [`Parser`] trait. +/// +/// [`parse_macro_input!`]: crate::parse_macro_input! +/// [syn-parse]: self#the-synparse-functions +pub struct ParseBuffer<'a> { + scope: Span, + // Instead of Cell> so that ParseBuffer<'a> is covariant in 'a. + // The rest of the code in this module needs to be careful that only a + // cursor derived from this `cell` is ever assigned to this `cell`. + // + // Cell> cannot be covariant in 'a because then we could take a + // ParseBuffer<'a>, upcast to ParseBuffer<'short> for some lifetime shorter + // than 'a, and then assign a Cursor<'short> into the Cell. + // + // By extension, it would not be safe to expose an API that accepts a + // Cursor<'a> and trusts that it lives as long as the cursor currently in + // the cell. + cell: Cell>, + marker: PhantomData>, + unexpected: Cell>>>, +} + +impl<'a> Drop for ParseBuffer<'a> { + fn drop(&mut self) { + if let Some(unexpected_span) = span_of_unexpected_ignoring_nones(self.cursor()) { + let (inner, old_span) = inner_unexpected(self); + if old_span.is_none() { + inner.set(Unexpected::Some(unexpected_span)); + } + } + } +} + +impl<'a> Display for ParseBuffer<'a> { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.cursor().token_stream(), f) + } +} + +impl<'a> Debug for ParseBuffer<'a> { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Debug::fmt(&self.cursor().token_stream(), f) + } +} + +/// Cursor state associated with speculative parsing. +/// +/// This type is the input of the closure provided to [`ParseStream::step`]. +/// +/// [`ParseStream::step`]: ParseBuffer::step +/// +/// # Example +/// +/// ``` +/// use proc_macro2::TokenTree; +/// use syn::Result; +/// use syn::parse::ParseStream; +/// +/// // This function advances the stream past the next occurrence of `@`. If +/// // no `@` is present in the stream, the stream position is unchanged and +/// // an error is returned. +/// fn skip_past_next_at(input: ParseStream) -> Result<()> { +/// input.step(|cursor| { +/// let mut rest = *cursor; +/// while let Some((tt, next)) = rest.token_tree() { +/// match &tt { +/// TokenTree::Punct(punct) if punct.as_char() == '@' => { +/// return Ok(((), next)); +/// } +/// _ => rest = next, +/// } +/// } +/// Err(cursor.error("no `@` was found after this point")) +/// }) +/// } +/// # +/// # fn remainder_after_skipping_past_next_at( +/// # input: ParseStream, +/// # ) -> Result { +/// # skip_past_next_at(input)?; +/// # input.parse() +/// # } +/// # +/// # use syn::parse::Parser; +/// # let remainder = remainder_after_skipping_past_next_at +/// # .parse_str("a @ b c") +/// # .unwrap(); +/// # assert_eq!(remainder.to_string(), "b c"); +/// ``` +pub struct StepCursor<'c, 'a> { + scope: Span, + // This field is covariant in 'c. + cursor: Cursor<'c>, + // This field is contravariant in 'c. Together these make StepCursor + // invariant in 'c. Also covariant in 'a. The user cannot cast 'c to a + // different lifetime but can upcast into a StepCursor with a shorter + // lifetime 'a. + // + // As long as we only ever construct a StepCursor for which 'c outlives 'a, + // this means if ever a StepCursor<'c, 'a> exists we are guaranteed that 'c + // outlives 'a. + marker: PhantomData) -> Cursor<'a>>, +} + +impl<'c, 'a> Deref for StepCursor<'c, 'a> { + type Target = Cursor<'c>; + + fn deref(&self) -> &Self::Target { + &self.cursor + } +} + +impl<'c, 'a> Copy for StepCursor<'c, 'a> {} + +impl<'c, 'a> Clone for StepCursor<'c, 'a> { + fn clone(&self) -> Self { + *self + } +} + +impl<'c, 'a> StepCursor<'c, 'a> { + /// Triggers an error at the current position of the parse stream. + /// + /// The `ParseStream::step` invocation will return this same error without + /// advancing the stream state. + pub fn error(self, message: T) -> Error { + error::new_at(self.scope, self.cursor, message) + } +} + +pub(crate) fn advance_step_cursor<'c, 'a>(proof: StepCursor<'c, 'a>, to: Cursor<'c>) -> Cursor<'a> { + // Refer to the comments within the StepCursor definition. We use the + // fact that a StepCursor<'c, 'a> exists as proof that 'c outlives 'a. + // Cursor is covariant in its lifetime parameter so we can cast a + // Cursor<'c> to one with the shorter lifetime Cursor<'a>. + let _ = proof; + unsafe { mem::transmute::, Cursor<'a>>(to) } +} + +pub(crate) fn new_parse_buffer( + scope: Span, + cursor: Cursor, + unexpected: Rc>, +) -> ParseBuffer { + ParseBuffer { + scope, + // See comment on `cell` in the struct definition. + cell: Cell::new(unsafe { mem::transmute::>(cursor) }), + marker: PhantomData, + unexpected: Cell::new(Some(unexpected)), + } +} + +pub(crate) enum Unexpected { + None, + Some(Span), + Chain(Rc>), +} + +impl Default for Unexpected { + fn default() -> Self { + Unexpected::None + } +} + +impl Clone for Unexpected { + fn clone(&self) -> Self { + match self { + Unexpected::None => Unexpected::None, + Unexpected::Some(span) => Unexpected::Some(*span), + Unexpected::Chain(next) => Unexpected::Chain(next.clone()), + } + } +} + +// We call this on Cell and Cell> where temporarily +// swapping in a None is cheap. +fn cell_clone(cell: &Cell) -> T { + let prev = cell.take(); + let ret = prev.clone(); + cell.set(prev); + ret +} + +fn inner_unexpected(buffer: &ParseBuffer) -> (Rc>, Option) { + let mut unexpected = get_unexpected(buffer); + loop { + match cell_clone(&unexpected) { + Unexpected::None => return (unexpected, None), + Unexpected::Some(span) => return (unexpected, Some(span)), + Unexpected::Chain(next) => unexpected = next, + } + } +} + +pub(crate) fn get_unexpected(buffer: &ParseBuffer) -> Rc> { + cell_clone(&buffer.unexpected).unwrap() +} + +fn span_of_unexpected_ignoring_nones(mut cursor: Cursor) -> Option { + if cursor.eof() { + return None; + } + while let Some((inner, _span, rest)) = cursor.group(Delimiter::None) { + if let Some(unexpected) = span_of_unexpected_ignoring_nones(inner) { + return Some(unexpected); + } + cursor = rest; + } + if cursor.eof() { + None + } else { + Some(cursor.span()) + } +} + +impl<'a> ParseBuffer<'a> { + /// Parses a syntax tree node of type `T`, advancing the position of our + /// parse stream past it. + pub fn parse(&self) -> Result { + T::parse(self) + } + + /// Calls the given parser function to parse a syntax tree node of type `T` + /// from this stream. + /// + /// # Example + /// + /// The parser below invokes [`Attribute::parse_outer`] to parse a vector of + /// zero or more outer attributes. + /// + /// [`Attribute::parse_outer`]: crate::Attribute::parse_outer + /// + /// ``` + /// use syn::{Attribute, Ident, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Parses a unit struct with attributes. + /// // + /// // #[path = "s.tmpl"] + /// // struct S; + /// struct UnitStruct { + /// attrs: Vec, + /// struct_token: Token![struct], + /// name: Ident, + /// semi_token: Token![;], + /// } + /// + /// impl Parse for UnitStruct { + /// fn parse(input: ParseStream) -> Result { + /// Ok(UnitStruct { + /// attrs: input.call(Attribute::parse_outer)?, + /// struct_token: input.parse()?, + /// name: input.parse()?, + /// semi_token: input.parse()?, + /// }) + /// } + /// } + /// ``` + pub fn call(&self, function: fn(ParseStream) -> Result) -> Result { + function(self) + } + + /// Looks at the next token in the parse stream to determine whether it + /// matches the requested type of token. + /// + /// Does not advance the position of the parse stream. + /// + /// # Syntax + /// + /// Note that this method does not use turbofish syntax. Pass the peek type + /// inside of parentheses. + /// + /// - `input.peek(Token![struct])` + /// - `input.peek(Token![==])` + /// - `input.peek(Ident)` *(does not accept keywords)* + /// - `input.peek(Ident::peek_any)` + /// - `input.peek(Lifetime)` + /// - `input.peek(token::Brace)` + /// + /// # Example + /// + /// In this example we finish parsing the list of supertraits when the next + /// token in the input is either `where` or an opening curly brace. + /// + /// ``` + /// use syn::{braced, token, Generics, Ident, Result, Token, TypeParamBound}; + /// use syn::parse::{Parse, ParseStream}; + /// use syn::punctuated::Punctuated; + /// + /// // Parses a trait definition containing no associated items. + /// // + /// // trait Marker<'de, T>: A + B<'de> where Box: Clone {} + /// struct MarkerTrait { + /// trait_token: Token![trait], + /// ident: Ident, + /// generics: Generics, + /// colon_token: Option, + /// supertraits: Punctuated, + /// brace_token: token::Brace, + /// } + /// + /// impl Parse for MarkerTrait { + /// fn parse(input: ParseStream) -> Result { + /// let trait_token: Token![trait] = input.parse()?; + /// let ident: Ident = input.parse()?; + /// let mut generics: Generics = input.parse()?; + /// let colon_token: Option = input.parse()?; + /// + /// let mut supertraits = Punctuated::new(); + /// if colon_token.is_some() { + /// loop { + /// supertraits.push_value(input.parse()?); + /// if input.peek(Token![where]) || input.peek(token::Brace) { + /// break; + /// } + /// supertraits.push_punct(input.parse()?); + /// } + /// } + /// + /// generics.where_clause = input.parse()?; + /// let content; + /// let empty_brace_token = braced!(content in input); + /// + /// Ok(MarkerTrait { + /// trait_token, + /// ident, + /// generics, + /// colon_token, + /// supertraits, + /// brace_token: empty_brace_token, + /// }) + /// } + /// } + /// ``` + pub fn peek(&self, token: T) -> bool { + let _ = token; + T::Token::peek(self.cursor()) + } + + /// Looks at the second-next token in the parse stream. + /// + /// This is commonly useful as a way to implement contextual keywords. + /// + /// # Example + /// + /// This example needs to use `peek2` because the symbol `union` is not a + /// keyword in Rust. We can't use just `peek` and decide to parse a union if + /// the very next token is `union`, because someone is free to write a `mod + /// union` and a macro invocation that looks like `union::some_macro! { ... + /// }`. In other words `union` is a contextual keyword. + /// + /// ``` + /// use syn::{Ident, ItemUnion, Macro, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Parses either a union or a macro invocation. + /// enum UnionOrMacro { + /// // union MaybeUninit { uninit: (), value: T } + /// Union(ItemUnion), + /// // lazy_static! { ... } + /// Macro(Macro), + /// } + /// + /// impl Parse for UnionOrMacro { + /// fn parse(input: ParseStream) -> Result { + /// if input.peek(Token![union]) && input.peek2(Ident) { + /// input.parse().map(UnionOrMacro::Union) + /// } else { + /// input.parse().map(UnionOrMacro::Macro) + /// } + /// } + /// } + /// ``` + pub fn peek2(&self, token: T) -> bool { + fn peek2(buffer: &ParseBuffer, peek: fn(Cursor) -> bool) -> bool { + if let Some(group) = buffer.cursor().group(Delimiter::None) { + if group.0.skip().map_or(false, peek) { + return true; + } + } + buffer.cursor().skip().map_or(false, peek) + } + + let _ = token; + peek2(self, T::Token::peek) + } + + /// Looks at the third-next token in the parse stream. + pub fn peek3(&self, token: T) -> bool { + fn peek3(buffer: &ParseBuffer, peek: fn(Cursor) -> bool) -> bool { + if let Some(group) = buffer.cursor().group(Delimiter::None) { + if group.0.skip().and_then(Cursor::skip).map_or(false, peek) { + return true; + } + } + buffer + .cursor() + .skip() + .and_then(Cursor::skip) + .map_or(false, peek) + } + + let _ = token; + peek3(self, T::Token::peek) + } + + /// Parses zero or more occurrences of `T` separated by punctuation of type + /// `P`, with optional trailing punctuation. + /// + /// Parsing continues until the end of this parse stream. The entire content + /// of this parse stream must consist of `T` and `P`. + /// + /// # Example + /// + /// ``` + /// # use quote::quote; + /// # + /// use syn::{parenthesized, token, Ident, Result, Token, Type}; + /// use syn::parse::{Parse, ParseStream}; + /// use syn::punctuated::Punctuated; + /// + /// // Parse a simplified tuple struct syntax like: + /// // + /// // struct S(A, B); + /// struct TupleStruct { + /// struct_token: Token![struct], + /// ident: Ident, + /// paren_token: token::Paren, + /// fields: Punctuated, + /// semi_token: Token![;], + /// } + /// + /// impl Parse for TupleStruct { + /// fn parse(input: ParseStream) -> Result { + /// let content; + /// Ok(TupleStruct { + /// struct_token: input.parse()?, + /// ident: input.parse()?, + /// paren_token: parenthesized!(content in input), + /// fields: content.parse_terminated(Type::parse)?, + /// semi_token: input.parse()?, + /// }) + /// } + /// } + /// # + /// # let input = quote! { + /// # struct S(A, B); + /// # }; + /// # syn::parse2::(input).unwrap(); + /// ``` + pub fn parse_terminated( + &self, + parser: fn(ParseStream) -> Result, + ) -> Result> { + Punctuated::parse_terminated_with(self, parser) + } + + /// Returns whether there are tokens remaining in this stream. + /// + /// This method returns true at the end of the content of a set of + /// delimiters, as well as at the very end of the complete macro input. + /// + /// # Example + /// + /// ``` + /// use syn::{braced, token, Ident, Item, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Parses a Rust `mod m { ... }` containing zero or more items. + /// struct Mod { + /// mod_token: Token![mod], + /// name: Ident, + /// brace_token: token::Brace, + /// items: Vec, + /// } + /// + /// impl Parse for Mod { + /// fn parse(input: ParseStream) -> Result { + /// let content; + /// Ok(Mod { + /// mod_token: input.parse()?, + /// name: input.parse()?, + /// brace_token: braced!(content in input), + /// items: { + /// let mut items = Vec::new(); + /// while !content.is_empty() { + /// items.push(content.parse()?); + /// } + /// items + /// }, + /// }) + /// } + /// } + /// ``` + pub fn is_empty(&self) -> bool { + self.cursor().eof() + } + + /// Constructs a helper for peeking at the next token in this stream and + /// building an error message if it is not one of a set of expected tokens. + /// + /// # Example + /// + /// ``` + /// use syn::{ConstParam, Ident, Lifetime, LifetimeDef, Result, Token, TypeParam}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // A generic parameter, a single one of the comma-separated elements inside + /// // angle brackets in: + /// // + /// // fn f() { ... } + /// // + /// // On invalid input, lookahead gives us a reasonable error message. + /// // + /// // error: expected one of: identifier, lifetime, `const` + /// // | + /// // 5 | fn f() {} + /// // | ^ + /// enum GenericParam { + /// Type(TypeParam), + /// Lifetime(LifetimeDef), + /// Const(ConstParam), + /// } + /// + /// impl Parse for GenericParam { + /// fn parse(input: ParseStream) -> Result { + /// let lookahead = input.lookahead1(); + /// if lookahead.peek(Ident) { + /// input.parse().map(GenericParam::Type) + /// } else if lookahead.peek(Lifetime) { + /// input.parse().map(GenericParam::Lifetime) + /// } else if lookahead.peek(Token![const]) { + /// input.parse().map(GenericParam::Const) + /// } else { + /// Err(lookahead.error()) + /// } + /// } + /// } + /// ``` + pub fn lookahead1(&self) -> Lookahead1<'a> { + lookahead::new(self.scope, self.cursor()) + } + + /// Forks a parse stream so that parsing tokens out of either the original + /// or the fork does not advance the position of the other. + /// + /// # Performance + /// + /// Forking a parse stream is a cheap fixed amount of work and does not + /// involve copying token buffers. Where you might hit performance problems + /// is if your macro ends up parsing a large amount of content more than + /// once. + /// + /// ``` + /// # use syn::{Expr, Result}; + /// # use syn::parse::ParseStream; + /// # + /// # fn bad(input: ParseStream) -> Result { + /// // Do not do this. + /// if input.fork().parse::().is_ok() { + /// return input.parse::(); + /// } + /// # unimplemented!() + /// # } + /// ``` + /// + /// As a rule, avoid parsing an unbounded amount of tokens out of a forked + /// parse stream. Only use a fork when the amount of work performed against + /// the fork is small and bounded. + /// + /// When complex speculative parsing against the forked stream is + /// unavoidable, use [`parse::discouraged::Speculative`] to advance the + /// original stream once the fork's parse is determined to have been + /// successful. + /// + /// For a lower level way to perform speculative parsing at the token level, + /// consider using [`ParseStream::step`] instead. + /// + /// [`parse::discouraged::Speculative`]: discouraged::Speculative + /// [`ParseStream::step`]: ParseBuffer::step + /// + /// # Example + /// + /// The parse implementation shown here parses possibly restricted `pub` + /// visibilities. + /// + /// - `pub` + /// - `pub(crate)` + /// - `pub(self)` + /// - `pub(super)` + /// - `pub(in some::path)` + /// + /// To handle the case of visibilities inside of tuple structs, the parser + /// needs to distinguish parentheses that specify visibility restrictions + /// from parentheses that form part of a tuple type. + /// + /// ``` + /// # struct A; + /// # struct B; + /// # struct C; + /// # + /// struct S(pub(crate) A, pub (B, C)); + /// ``` + /// + /// In this example input the first tuple struct element of `S` has + /// `pub(crate)` visibility while the second tuple struct element has `pub` + /// visibility; the parentheses around `(B, C)` are part of the type rather + /// than part of a visibility restriction. + /// + /// The parser uses a forked parse stream to check the first token inside of + /// parentheses after the `pub` keyword. This is a small bounded amount of + /// work performed against the forked parse stream. + /// + /// ``` + /// use syn::{parenthesized, token, Ident, Path, Result, Token}; + /// use syn::ext::IdentExt; + /// use syn::parse::{Parse, ParseStream}; + /// + /// struct PubVisibility { + /// pub_token: Token![pub], + /// restricted: Option, + /// } + /// + /// struct Restricted { + /// paren_token: token::Paren, + /// in_token: Option, + /// path: Path, + /// } + /// + /// impl Parse for PubVisibility { + /// fn parse(input: ParseStream) -> Result { + /// let pub_token: Token![pub] = input.parse()?; + /// + /// if input.peek(token::Paren) { + /// let ahead = input.fork(); + /// let mut content; + /// parenthesized!(content in ahead); + /// + /// if content.peek(Token![crate]) + /// || content.peek(Token![self]) + /// || content.peek(Token![super]) + /// { + /// return Ok(PubVisibility { + /// pub_token, + /// restricted: Some(Restricted { + /// paren_token: parenthesized!(content in input), + /// in_token: None, + /// path: Path::from(content.call(Ident::parse_any)?), + /// }), + /// }); + /// } else if content.peek(Token![in]) { + /// return Ok(PubVisibility { + /// pub_token, + /// restricted: Some(Restricted { + /// paren_token: parenthesized!(content in input), + /// in_token: Some(content.parse()?), + /// path: content.call(Path::parse_mod_style)?, + /// }), + /// }); + /// } + /// } + /// + /// Ok(PubVisibility { + /// pub_token, + /// restricted: None, + /// }) + /// } + /// } + /// ``` + pub fn fork(&self) -> Self { + ParseBuffer { + scope: self.scope, + cell: self.cell.clone(), + marker: PhantomData, + // Not the parent's unexpected. Nothing cares whether the clone + // parses all the way unless we `advance_to`. + unexpected: Cell::new(Some(Rc::new(Cell::new(Unexpected::None)))), + } + } + + /// Triggers an error at the current position of the parse stream. + /// + /// # Example + /// + /// ``` + /// use syn::{Expr, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Some kind of loop: `while` or `for` or `loop`. + /// struct Loop { + /// expr: Expr, + /// } + /// + /// impl Parse for Loop { + /// fn parse(input: ParseStream) -> Result { + /// if input.peek(Token![while]) + /// || input.peek(Token![for]) + /// || input.peek(Token![loop]) + /// { + /// Ok(Loop { + /// expr: input.parse()?, + /// }) + /// } else { + /// Err(input.error("expected some kind of loop")) + /// } + /// } + /// } + /// ``` + pub fn error(&self, message: T) -> Error { + error::new_at(self.scope, self.cursor(), message) + } + + /// Speculatively parses tokens from this parse stream, advancing the + /// position of this stream only if parsing succeeds. + /// + /// This is a powerful low-level API used for defining the `Parse` impls of + /// the basic built-in token types. It is not something that will be used + /// widely outside of the Syn codebase. + /// + /// # Example + /// + /// ``` + /// use proc_macro2::TokenTree; + /// use syn::Result; + /// use syn::parse::ParseStream; + /// + /// // This function advances the stream past the next occurrence of `@`. If + /// // no `@` is present in the stream, the stream position is unchanged and + /// // an error is returned. + /// fn skip_past_next_at(input: ParseStream) -> Result<()> { + /// input.step(|cursor| { + /// let mut rest = *cursor; + /// while let Some((tt, next)) = rest.token_tree() { + /// match &tt { + /// TokenTree::Punct(punct) if punct.as_char() == '@' => { + /// return Ok(((), next)); + /// } + /// _ => rest = next, + /// } + /// } + /// Err(cursor.error("no `@` was found after this point")) + /// }) + /// } + /// # + /// # fn remainder_after_skipping_past_next_at( + /// # input: ParseStream, + /// # ) -> Result { + /// # skip_past_next_at(input)?; + /// # input.parse() + /// # } + /// # + /// # use syn::parse::Parser; + /// # let remainder = remainder_after_skipping_past_next_at + /// # .parse_str("a @ b c") + /// # .unwrap(); + /// # assert_eq!(remainder.to_string(), "b c"); + /// ``` + pub fn step(&self, function: F) -> Result + where + F: for<'c> FnOnce(StepCursor<'c, 'a>) -> Result<(R, Cursor<'c>)>, + { + // Since the user's function is required to work for any 'c, we know + // that the Cursor<'c> they return is either derived from the input + // StepCursor<'c, 'a> or from a Cursor<'static>. + // + // It would not be legal to write this function without the invariant + // lifetime 'c in StepCursor<'c, 'a>. If this function were written only + // in terms of 'a, the user could take our ParseBuffer<'a>, upcast it to + // a ParseBuffer<'short> which some shorter lifetime than 'a, invoke + // `step` on their ParseBuffer<'short> with a closure that returns + // Cursor<'short>, and we would wrongly write that Cursor<'short> into + // the Cell intended to hold Cursor<'a>. + // + // In some cases it may be necessary for R to contain a Cursor<'a>. + // Within Syn we solve this using `advance_step_cursor` which uses the + // existence of a StepCursor<'c, 'a> as proof that it is safe to cast + // from Cursor<'c> to Cursor<'a>. If needed outside of Syn, it would be + // safe to expose that API as a method on StepCursor. + let (node, rest) = function(StepCursor { + scope: self.scope, + cursor: self.cell.get(), + marker: PhantomData, + })?; + self.cell.set(rest); + Ok(node) + } + + /// Returns the `Span` of the next token in the parse stream, or + /// `Span::call_site()` if this parse stream has completely exhausted its + /// input `TokenStream`. + pub fn span(&self) -> Span { + let cursor = self.cursor(); + if cursor.eof() { + self.scope + } else { + crate::buffer::open_span_of_group(cursor) + } + } + + /// Provides low-level access to the token representation underlying this + /// parse stream. + /// + /// Cursors are immutable so no operations you perform against the cursor + /// will affect the state of this parse stream. + pub fn cursor(&self) -> Cursor<'a> { + self.cell.get() + } + + fn check_unexpected(&self) -> Result<()> { + match inner_unexpected(self).1 { + Some(span) => Err(Error::new(span, "unexpected token")), + None => Ok(()), + } + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Box { + fn parse(input: ParseStream) -> Result { + input.parse().map(Box::new) + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Option { + fn parse(input: ParseStream) -> Result { + if T::peek(input.cursor()) { + Ok(Some(input.parse()?)) + } else { + Ok(None) + } + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for TokenStream { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| Ok((cursor.token_stream(), Cursor::empty()))) + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for TokenTree { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| match cursor.token_tree() { + Some((tt, rest)) => Ok((tt, rest)), + None => Err(cursor.error("expected token tree")), + }) + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Group { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| { + for delim in &[Delimiter::Parenthesis, Delimiter::Brace, Delimiter::Bracket] { + if let Some((inside, span, rest)) = cursor.group(*delim) { + let mut group = Group::new(*delim, inside.token_stream()); + group.set_span(span); + return Ok((group, rest)); + } + } + Err(cursor.error("expected group token")) + }) + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Punct { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| match cursor.punct() { + Some((punct, rest)) => Ok((punct, rest)), + None => Err(cursor.error("expected punctuation token")), + }) + } +} + +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Literal { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| match cursor.literal() { + Some((literal, rest)) => Ok((literal, rest)), + None => Err(cursor.error("expected literal token")), + }) + } +} + +/// Parser that can parse Rust tokens into a particular syntax tree node. +/// +/// Refer to the [module documentation] for details about parsing in Syn. +/// +/// [module documentation]: self +/// +/// *This trait is available only if Syn is built with the `"parsing"` feature.* +pub trait Parser: Sized { + type Output; + + /// Parse a proc-macro2 token stream into the chosen syntax tree node. + /// + /// This function will check that the input is fully parsed. If there are + /// any unparsed tokens at the end of the stream, an error is returned. + fn parse2(self, tokens: TokenStream) -> Result; + + /// Parse tokens of source code into the chosen syntax tree node. + /// + /// This function will check that the input is fully parsed. If there are + /// any unparsed tokens at the end of the stream, an error is returned. + /// + /// *This method is available only if Syn is built with both the `"parsing"` and + /// `"proc-macro"` features.* + #[cfg(all( + not(all(target_arch = "wasm32", any(target_os = "unknown", target_os = "wasi"))), + feature = "proc-macro" + ))] + fn parse(self, tokens: proc_macro::TokenStream) -> Result { + self.parse2(proc_macro2::TokenStream::from(tokens)) + } + + /// Parse a string of Rust code into the chosen syntax tree node. + /// + /// This function will check that the input is fully parsed. If there are + /// any unparsed tokens at the end of the string, an error is returned. + /// + /// # Hygiene + /// + /// Every span in the resulting syntax tree will be set to resolve at the + /// macro call site. + fn parse_str(self, s: &str) -> Result { + self.parse2(proc_macro2::TokenStream::from_str(s)?) + } + + // Not public API. + #[doc(hidden)] + #[cfg(any(feature = "full", feature = "derive"))] + fn __parse_scoped(self, scope: Span, tokens: TokenStream) -> Result { + let _ = scope; + self.parse2(tokens) + } + + // Not public API. + #[doc(hidden)] + #[cfg(any(feature = "full", feature = "derive"))] + fn __parse_stream(self, input: ParseStream) -> Result { + input.parse().and_then(|tokens| self.parse2(tokens)) + } +} + +fn tokens_to_parse_buffer(tokens: &TokenBuffer) -> ParseBuffer { + let scope = Span::call_site(); + let cursor = tokens.begin(); + let unexpected = Rc::new(Cell::new(Unexpected::None)); + new_parse_buffer(scope, cursor, unexpected) +} + +impl Parser for F +where + F: FnOnce(ParseStream) -> Result, +{ + type Output = T; + + fn parse2(self, tokens: TokenStream) -> Result { + let buf = TokenBuffer::new2(tokens); + let state = tokens_to_parse_buffer(&buf); + let node = self(&state)?; + state.check_unexpected()?; + if let Some(unexpected_span) = span_of_unexpected_ignoring_nones(state.cursor()) { + Err(Error::new(unexpected_span, "unexpected token")) + } else { + Ok(node) + } + } + + #[cfg(any(feature = "full", feature = "derive"))] + fn __parse_scoped(self, scope: Span, tokens: TokenStream) -> Result { + let buf = TokenBuffer::new2(tokens); + let cursor = buf.begin(); + let unexpected = Rc::new(Cell::new(Unexpected::None)); + let state = new_parse_buffer(scope, cursor, unexpected); + let node = self(&state)?; + state.check_unexpected()?; + if let Some(unexpected_span) = span_of_unexpected_ignoring_nones(state.cursor()) { + Err(Error::new(unexpected_span, "unexpected token")) + } else { + Ok(node) + } + } + + #[cfg(any(feature = "full", feature = "derive"))] + fn __parse_stream(self, input: ParseStream) -> Result { + self(input) + } +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn parse_scoped(f: F, scope: Span, tokens: TokenStream) -> Result { + f.__parse_scoped(scope, tokens) +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn parse_stream(f: F, input: ParseStream) -> Result { + f.__parse_stream(input) +} + +/// An empty syntax tree node that consumes no tokens when parsed. +/// +/// This is useful for attribute macros that want to ensure they are not +/// provided any attribute args. +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// use proc_macro::TokenStream; +/// use syn::parse_macro_input; +/// use syn::parse::Nothing; +/// +/// # const IGNORE: &str = stringify! { +/// #[proc_macro_attribute] +/// # }; +/// pub fn my_attr(args: TokenStream, input: TokenStream) -> TokenStream { +/// parse_macro_input!(args as Nothing); +/// +/// /* ... */ +/// # "".parse().unwrap() +/// } +/// ``` +/// +/// ```text +/// error: unexpected token +/// --> src/main.rs:3:19 +/// | +/// 3 | #[my_attr(asdf)] +/// | ^^^^ +/// ``` +pub struct Nothing; + +impl Parse for Nothing { + fn parse(_input: ParseStream) -> Result { + Ok(Nothing) + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Debug for Nothing { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str("Nothing") + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Eq for Nothing {} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl PartialEq for Nothing { + fn eq(&self, _other: &Self) -> bool { + true + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Hash for Nothing { + fn hash(&self, _state: &mut H) {} +} diff --git a/rust/syn/parse_macro_input.rs b/rust/syn/parse_macro_input.rs new file mode 100644 index 00000000000000..8c68fdda2f717f --- /dev/null +++ b/rust/syn/parse_macro_input.rs @@ -0,0 +1,181 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Parse the input TokenStream of a macro, triggering a compile error if the +/// tokens fail to parse. +/// +/// Refer to the [`parse` module] documentation for more details about parsing +/// in Syn. +/// +/// [`parse` module]: mod@crate::parse +/// +///
+/// +/// # Intended usage +/// +/// This macro must be called from a function that returns +/// `proc_macro::TokenStream`. Usually this will be your proc macro entry point, +/// the function that has the #\[proc_macro\] / #\[proc_macro_derive\] / +/// #\[proc_macro_attribute\] attribute. +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// use proc_macro::TokenStream; +/// use syn::{parse_macro_input, Result}; +/// use syn::parse::{Parse, ParseStream}; +/// +/// struct MyMacroInput { +/// /* ... */ +/// } +/// +/// impl Parse for MyMacroInput { +/// fn parse(input: ParseStream) -> Result { +/// /* ... */ +/// # Ok(MyMacroInput {}) +/// } +/// } +/// +/// # const IGNORE: &str = stringify! { +/// #[proc_macro] +/// # }; +/// pub fn my_macro(tokens: TokenStream) -> TokenStream { +/// let input = parse_macro_input!(tokens as MyMacroInput); +/// +/// /* ... */ +/// # "".parse().unwrap() +/// } +/// ``` +/// +///
+/// +/// # Usage with Parser +/// +/// This macro can also be used with the [`Parser` trait] for types that have +/// multiple ways that they can be parsed. +/// +/// [`Parser` trait]: crate::parse::Parser +/// +/// ``` +/// # extern crate proc_macro; +/// # +/// # use proc_macro::TokenStream; +/// # use syn::{parse_macro_input, Result}; +/// # use syn::parse::ParseStream; +/// # +/// # struct MyMacroInput {} +/// # +/// impl MyMacroInput { +/// fn parse_alternate(input: ParseStream) -> Result { +/// /* ... */ +/// # Ok(MyMacroInput {}) +/// } +/// } +/// +/// # const IGNORE: &str = stringify! { +/// #[proc_macro] +/// # }; +/// pub fn my_macro(tokens: TokenStream) -> TokenStream { +/// let input = parse_macro_input!(tokens with MyMacroInput::parse_alternate); +/// +/// /* ... */ +/// # "".parse().unwrap() +/// } +/// ``` +/// +///
+/// +/// # Expansion +/// +/// `parse_macro_input!($variable as $Type)` expands to something like: +/// +/// ```no_run +/// # extern crate proc_macro; +/// # +/// # macro_rules! doc_test { +/// # ($variable:ident as $Type:ty) => { +/// match syn::parse::<$Type>($variable) { +/// Ok(syntax_tree) => syntax_tree, +/// Err(err) => return proc_macro::TokenStream::from(err.to_compile_error()), +/// } +/// # }; +/// # } +/// # +/// # fn test(input: proc_macro::TokenStream) -> proc_macro::TokenStream { +/// # let _ = doc_test!(input as syn::Ident); +/// # proc_macro::TokenStream::new() +/// # } +/// ``` +#[macro_export] +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "proc-macro"))))] +macro_rules! parse_macro_input { + ($tokenstream:ident as $ty:ty) => { + match $crate::parse_macro_input::parse::<$ty>($tokenstream) { + $crate::__private::Ok(data) => data, + $crate::__private::Err(err) => { + return $crate::__private::TokenStream::from(err.to_compile_error()); + } + } + }; + ($tokenstream:ident with $parser:path) => { + match $crate::parse::Parser::parse($parser, $tokenstream) { + $crate::__private::Ok(data) => data, + $crate::__private::Err(err) => { + return $crate::__private::TokenStream::from(err.to_compile_error()); + } + } + }; + ($tokenstream:ident) => { + $crate::parse_macro_input!($tokenstream as _) + }; +} + +//////////////////////////////////////////////////////////////////////////////// +// Can parse any type that implements Parse. + +use crate::parse::{Parse, ParseStream, Parser, Result}; +use proc_macro::TokenStream; + +// Not public API. +#[doc(hidden)] +pub fn parse(token_stream: TokenStream) -> Result { + T::parse.parse(token_stream) +} + +// Not public API. +#[doc(hidden)] +pub trait ParseMacroInput: Sized { + fn parse(input: ParseStream) -> Result; +} + +impl ParseMacroInput for T { + fn parse(input: ParseStream) -> Result { + ::parse(input) + } +} + +//////////////////////////////////////////////////////////////////////////////// +// Any other types that we want `parse_macro_input!` to be able to parse. + +#[cfg(any(feature = "full", feature = "derive"))] +use crate::AttributeArgs; + +#[cfg(any(feature = "full", feature = "derive"))] +impl ParseMacroInput for AttributeArgs { + fn parse(input: ParseStream) -> Result { + let mut metas = Vec::new(); + + loop { + if input.is_empty() { + break; + } + let value = input.parse()?; + metas.push(value); + if input.is_empty() { + break; + } + input.parse::()?; + } + + Ok(metas) + } +} diff --git a/rust/syn/parse_quote.rs b/rust/syn/parse_quote.rs new file mode 100644 index 00000000000000..4eb592d099f631 --- /dev/null +++ b/rust/syn/parse_quote.rs @@ -0,0 +1,169 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +/// Quasi-quotation macro that accepts input like the [`quote!`] macro but uses +/// type inference to figure out a return type for those tokens. +/// +/// [`quote!`]: https://docs.rs/quote/1.0/quote/index.html +/// +/// The return type can be any syntax tree node that implements the [`Parse`] +/// trait. +/// +/// [`Parse`]: crate::parse::Parse +/// +/// ``` +/// use quote::quote; +/// use syn::{parse_quote, Stmt}; +/// +/// fn main() { +/// let name = quote!(v); +/// let ty = quote!(u8); +/// +/// let stmt: Stmt = parse_quote! { +/// let #name: #ty = Default::default(); +/// }; +/// +/// println!("{:#?}", stmt); +/// } +/// ``` +/// +/// *This macro is available only if Syn is built with the `"parsing"` feature, +/// although interpolation of syntax tree nodes into the quoted tokens is only +/// supported if Syn is built with the `"printing"` feature as well.* +/// +/// # Example +/// +/// The following helper function adds a bound `T: HeapSize` to every type +/// parameter `T` in the input generics. +/// +/// ``` +/// use syn::{parse_quote, Generics, GenericParam}; +/// +/// // Add a bound `T: HeapSize` to every type parameter T. +/// fn add_trait_bounds(mut generics: Generics) -> Generics { +/// for param in &mut generics.params { +/// if let GenericParam::Type(type_param) = param { +/// type_param.bounds.push(parse_quote!(HeapSize)); +/// } +/// } +/// generics +/// } +/// ``` +/// +/// # Special cases +/// +/// This macro can parse the following additional types as a special case even +/// though they do not implement the `Parse` trait. +/// +/// - [`Attribute`] — parses one attribute, allowing either outer like `#[...]` +/// or inner like `#![...]` +/// - [`Punctuated`] — parses zero or more `T` separated by punctuation +/// `P` with optional trailing punctuation +/// - [`Vec`] — parses the same as `Block::parse_within` +/// +/// [`Vec`]: Block::parse_within +/// +/// # Panics +/// +/// Panics if the tokens fail to parse as the expected syntax tree type. The +/// caller is responsible for ensuring that the input tokens are syntactically +/// valid. +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "printing"))))] +#[macro_export] +macro_rules! parse_quote { + ($($tt:tt)*) => { + $crate::parse_quote::parse($crate::__private::quote::quote!($($tt)*)) + }; +} + +/// This macro is [`parse_quote!`] + [`quote_spanned!`][quote::quote_spanned]. +/// +/// Please refer to each of their documentation. +/// +/// # Example +/// +/// ``` +/// use quote::{quote, quote_spanned}; +/// use syn::spanned::Spanned; +/// use syn::{parse_quote_spanned, ReturnType, Signature}; +/// +/// // Changes `fn()` to `fn() -> Pin>>`, +/// // and `fn() -> T` to `fn() -> Pin>>`, +/// // without introducing any call_site() spans. +/// fn make_ret_pinned_future(sig: &mut Signature) { +/// let ret = match &sig.output { +/// ReturnType::Default => quote_spanned!(sig.paren_token.span=> ()), +/// ReturnType::Type(_, ret) => quote!(#ret), +/// }; +/// sig.output = parse_quote_spanned! {ret.span()=> +/// -> ::std::pin::Pin<::std::boxed::Box>> +/// }; +/// } +/// ``` +#[cfg_attr(doc_cfg, doc(cfg(all(feature = "parsing", feature = "printing"))))] +#[macro_export] +macro_rules! parse_quote_spanned { + ($span:expr=> $($tt:tt)*) => { + $crate::parse_quote::parse($crate::__private::quote::quote_spanned!($span=> $($tt)*)) + }; +} + +//////////////////////////////////////////////////////////////////////////////// +// Can parse any type that implements Parse. + +use crate::parse::{Parse, ParseStream, Parser, Result}; +use proc_macro2::TokenStream; + +// Not public API. +#[doc(hidden)] +pub fn parse(token_stream: TokenStream) -> T { + let parser = T::parse; + match parser.parse2(token_stream) { + Ok(t) => t, + Err(err) => panic!("{}", err), + } +} + +// Not public API. +#[doc(hidden)] +pub trait ParseQuote: Sized { + fn parse(input: ParseStream) -> Result; +} + +impl ParseQuote for T { + fn parse(input: ParseStream) -> Result { + ::parse(input) + } +} + +//////////////////////////////////////////////////////////////////////////////// +// Any other types that we want `parse_quote!` to be able to parse. + +use crate::punctuated::Punctuated; +#[cfg(any(feature = "full", feature = "derive"))] +use crate::{attr, Attribute}; +#[cfg(feature = "full")] +use crate::{Block, Stmt}; + +#[cfg(any(feature = "full", feature = "derive"))] +impl ParseQuote for Attribute { + fn parse(input: ParseStream) -> Result { + if input.peek(Token![#]) && input.peek2(Token![!]) { + attr::parsing::single_parse_inner(input) + } else { + attr::parsing::single_parse_outer(input) + } + } +} + +impl ParseQuote for Punctuated { + fn parse(input: ParseStream) -> Result { + Self::parse_terminated(input) + } +} + +#[cfg(feature = "full")] +impl ParseQuote for Vec { + fn parse(input: ParseStream) -> Result { + Block::parse_within(input) + } +} diff --git a/rust/syn/pat.rs b/rust/syn/pat.rs new file mode 100644 index 00000000000000..a2c2004e2b6400 --- /dev/null +++ b/rust/syn/pat.rs @@ -0,0 +1,929 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; +use proc_macro2::TokenStream; + +ast_enum_of_structs! { + /// A pattern in a local binding, function signature, match expression, or + /// various other places. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum Pat { + /// A box pattern: `box v`. + Box(PatBox), + + /// A pattern that binds a new variable: `ref mut binding @ SUBPATTERN`. + Ident(PatIdent), + + /// A literal pattern: `0`. + /// + /// This holds an `Expr` rather than a `Lit` because negative numbers + /// are represented as an `Expr::Unary`. + Lit(PatLit), + + /// A macro in pattern position. + Macro(PatMacro), + + /// A pattern that matches any one of a set of cases. + Or(PatOr), + + /// A path pattern like `Color::Red`, optionally qualified with a + /// self-type. + /// + /// Unqualified path patterns can legally refer to variants, structs, + /// constants or associated constants. Qualified path patterns like + /// `::B::C` and `::B::C` can only legally refer to + /// associated constants. + Path(PatPath), + + /// A range pattern: `1..=2`. + Range(PatRange), + + /// A reference pattern: `&mut var`. + Reference(PatReference), + + /// The dots in a tuple or slice pattern: `[0, 1, ..]` + Rest(PatRest), + + /// A dynamically sized slice pattern: `[a, b, ref i @ .., y, z]`. + Slice(PatSlice), + + /// A struct or struct variant pattern: `Variant { x, y, .. }`. + Struct(PatStruct), + + /// A tuple pattern: `(a, b)`. + Tuple(PatTuple), + + /// A tuple struct or tuple variant pattern: `Variant(x, y, .., z)`. + TupleStruct(PatTupleStruct), + + /// A type ascription pattern: `foo: f64`. + Type(PatType), + + /// Tokens in pattern position not interpreted by Syn. + Verbatim(TokenStream), + + /// A pattern that matches any value: `_`. + Wild(PatWild), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match pat { + // Pat::Box(pat) => {...} + // Pat::Ident(pat) => {...} + // ... + // Pat::Wild(pat) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// A box pattern: `box v`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatBox { + pub attrs: Vec, + pub box_token: Token![box], + pub pat: Box, + } +} + +ast_struct! { + /// A pattern that binds a new variable: `ref mut binding @ SUBPATTERN`. + /// + /// It may also be a unit struct or struct variant (e.g. `None`), or a + /// constant; these cannot be distinguished syntactically. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatIdent { + pub attrs: Vec, + pub by_ref: Option, + pub mutability: Option, + pub ident: Ident, + pub subpat: Option<(Token![@], Box)>, + } +} + +ast_struct! { + /// A literal pattern: `0`. + /// + /// This holds an `Expr` rather than a `Lit` because negative numbers + /// are represented as an `Expr::Unary`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatLit { + pub attrs: Vec, + pub expr: Box, + } +} + +ast_struct! { + /// A macro in pattern position. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatMacro { + pub attrs: Vec, + pub mac: Macro, + } +} + +ast_struct! { + /// A pattern that matches any one of a set of cases. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatOr { + pub attrs: Vec, + pub leading_vert: Option, + pub cases: Punctuated, + } +} + +ast_struct! { + /// A path pattern like `Color::Red`, optionally qualified with a + /// self-type. + /// + /// Unqualified path patterns can legally refer to variants, structs, + /// constants or associated constants. Qualified path patterns like + /// `::B::C` and `::B::C` can only legally refer to + /// associated constants. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatPath { + pub attrs: Vec, + pub qself: Option, + pub path: Path, + } +} + +ast_struct! { + /// A range pattern: `1..=2`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatRange { + pub attrs: Vec, + pub lo: Box, + pub limits: RangeLimits, + pub hi: Box, + } +} + +ast_struct! { + /// A reference pattern: `&mut var`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatReference { + pub attrs: Vec, + pub and_token: Token![&], + pub mutability: Option, + pub pat: Box, + } +} + +ast_struct! { + /// The dots in a tuple or slice pattern: `[0, 1, ..]` + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatRest { + pub attrs: Vec, + pub dot2_token: Token![..], + } +} + +ast_struct! { + /// A dynamically sized slice pattern: `[a, b, ref i @ .., y, z]`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatSlice { + pub attrs: Vec, + pub bracket_token: token::Bracket, + pub elems: Punctuated, + } +} + +ast_struct! { + /// A struct or struct variant pattern: `Variant { x, y, .. }`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatStruct { + pub attrs: Vec, + pub path: Path, + pub brace_token: token::Brace, + pub fields: Punctuated, + pub dot2_token: Option, + } +} + +ast_struct! { + /// A tuple pattern: `(a, b)`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatTuple { + pub attrs: Vec, + pub paren_token: token::Paren, + pub elems: Punctuated, + } +} + +ast_struct! { + /// A tuple struct or tuple variant pattern: `Variant(x, y, .., z)`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatTupleStruct { + pub attrs: Vec, + pub path: Path, + pub pat: PatTuple, + } +} + +ast_struct! { + /// A type ascription pattern: `foo: f64`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatType { + pub attrs: Vec, + pub pat: Box, + pub colon_token: Token![:], + pub ty: Box, + } +} + +ast_struct! { + /// A pattern that matches any value: `_`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct PatWild { + pub attrs: Vec, + pub underscore_token: Token![_], + } +} + +ast_struct! { + /// A single field in a struct pattern. + /// + /// Patterns like the fields of Foo `{ x, ref y, ref mut z }` are treated + /// the same as `x: x, y: ref y, z: ref mut z` but there is no colon token. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct FieldPat { + pub attrs: Vec, + pub member: Member, + pub colon_token: Option, + pub pat: Box, + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::{Parse, ParseBuffer, ParseStream, Result}; + use crate::path; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Pat { + fn parse(input: ParseStream) -> Result { + let begin = input.fork(); + let lookahead = input.lookahead1(); + if { + let ahead = input.fork(); + ahead.parse::>()?.is_some() + && (ahead.peek(Token![::]) + || ahead.peek(Token![!]) + || ahead.peek(token::Brace) + || ahead.peek(token::Paren) + || ahead.peek(Token![..]) + && ahead.parse::().is_ok() + && !(ahead.is_empty() || ahead.peek(Token![,]))) + } || { + let ahead = input.fork(); + ahead.parse::>()?.is_some() && ahead.peek(Token![::]) + } || lookahead.peek(Token![::]) + || lookahead.peek(Token![<]) + || input.peek(Token![Self]) + || input.peek(Token![super]) + || input.peek(Token![crate]) + { + pat_path_or_macro_or_struct_or_range(input) + } else if lookahead.peek(Token![_]) { + input.call(pat_wild).map(Pat::Wild) + } else if input.peek(Token![box]) { + input.call(pat_box).map(Pat::Box) + } else if input.peek(Token![-]) || lookahead.peek(Lit) || lookahead.peek(Token![const]) + { + pat_lit_or_range(input) + } else if lookahead.peek(Token![ref]) + || lookahead.peek(Token![mut]) + || input.peek(Token![self]) + || input.peek(Ident) + { + input.call(pat_ident).map(Pat::Ident) + } else if lookahead.peek(Token![&]) { + input.call(pat_reference).map(Pat::Reference) + } else if lookahead.peek(token::Paren) { + input.call(pat_tuple).map(Pat::Tuple) + } else if lookahead.peek(token::Bracket) { + input.call(pat_slice).map(Pat::Slice) + } else if lookahead.peek(Token![..]) && !input.peek(Token![...]) { + pat_range_half_open(input, begin) + } else if lookahead.peek(Token![const]) { + input.call(pat_const).map(Pat::Verbatim) + } else { + Err(lookahead.error()) + } + } + } + + fn pat_path_or_macro_or_struct_or_range(input: ParseStream) -> Result { + let begin = input.fork(); + let (qself, path) = path::parsing::qpath(input, true)?; + + if qself.is_none() && input.peek(Token![!]) && !input.peek(Token![!=]) { + let mut contains_arguments = false; + for segment in &path.segments { + match segment.arguments { + PathArguments::None => {} + PathArguments::AngleBracketed(_) | PathArguments::Parenthesized(_) => { + contains_arguments = true; + } + } + } + + if !contains_arguments { + let bang_token: Token![!] = input.parse()?; + let (delimiter, tokens) = mac::parse_delimiter(input)?; + return Ok(Pat::Macro(PatMacro { + attrs: Vec::new(), + mac: Macro { + path, + bang_token, + delimiter, + tokens, + }, + })); + } + } + + if input.peek(token::Brace) { + let pat = pat_struct(begin.fork(), input, path)?; + if qself.is_some() { + Ok(Pat::Verbatim(verbatim::between(begin, input))) + } else { + Ok(pat) + } + } else if input.peek(token::Paren) { + let pat = pat_tuple_struct(input, path)?; + if qself.is_some() { + Ok(Pat::Verbatim(verbatim::between(begin, input))) + } else { + Ok(Pat::TupleStruct(pat)) + } + } else if input.peek(Token![..]) { + pat_range(input, begin, qself, path) + } else { + Ok(Pat::Path(PatPath { + attrs: Vec::new(), + qself, + path, + })) + } + } + + fn pat_wild(input: ParseStream) -> Result { + Ok(PatWild { + attrs: Vec::new(), + underscore_token: input.parse()?, + }) + } + + fn pat_box(input: ParseStream) -> Result { + Ok(PatBox { + attrs: Vec::new(), + box_token: input.parse()?, + pat: input.parse()?, + }) + } + + fn pat_ident(input: ParseStream) -> Result { + Ok(PatIdent { + attrs: Vec::new(), + by_ref: input.parse()?, + mutability: input.parse()?, + ident: input.call(Ident::parse_any)?, + subpat: { + if input.peek(Token![@]) { + let at_token: Token![@] = input.parse()?; + let subpat: Pat = input.parse()?; + Some((at_token, Box::new(subpat))) + } else { + None + } + }, + }) + } + + fn pat_tuple_struct(input: ParseStream, path: Path) -> Result { + Ok(PatTupleStruct { + attrs: Vec::new(), + path, + pat: input.call(pat_tuple)?, + }) + } + + fn pat_struct(begin: ParseBuffer, input: ParseStream, path: Path) -> Result { + let content; + let brace_token = braced!(content in input); + + let mut fields = Punctuated::new(); + let mut dot2_token = None; + while !content.is_empty() { + let attrs = content.call(Attribute::parse_outer)?; + if content.peek(Token![..]) { + dot2_token = Some(content.parse()?); + if !attrs.is_empty() { + return Ok(Pat::Verbatim(verbatim::between(begin, input))); + } + break; + } + let mut value = content.call(field_pat)?; + value.attrs = attrs; + fields.push_value(value); + if content.is_empty() { + break; + } + let punct: Token![,] = content.parse()?; + fields.push_punct(punct); + } + + Ok(Pat::Struct(PatStruct { + attrs: Vec::new(), + path, + brace_token, + fields, + dot2_token, + })) + } + + impl Member { + fn is_unnamed(&self) -> bool { + match *self { + Member::Named(_) => false, + Member::Unnamed(_) => true, + } + } + } + + fn field_pat(input: ParseStream) -> Result { + let boxed: Option = input.parse()?; + let by_ref: Option = input.parse()?; + let mutability: Option = input.parse()?; + let member: Member = input.parse()?; + + if boxed.is_none() && by_ref.is_none() && mutability.is_none() && input.peek(Token![:]) + || member.is_unnamed() + { + return Ok(FieldPat { + attrs: Vec::new(), + member, + colon_token: input.parse()?, + pat: Box::new(multi_pat_with_leading_vert(input)?), + }); + } + + let ident = match member { + Member::Named(ident) => ident, + Member::Unnamed(_) => unreachable!(), + }; + + let mut pat = Pat::Ident(PatIdent { + attrs: Vec::new(), + by_ref, + mutability, + ident: ident.clone(), + subpat: None, + }); + + if let Some(boxed) = boxed { + pat = Pat::Box(PatBox { + attrs: Vec::new(), + box_token: boxed, + pat: Box::new(pat), + }); + } + + Ok(FieldPat { + attrs: Vec::new(), + member: Member::Named(ident), + colon_token: None, + pat: Box::new(pat), + }) + } + + fn pat_range( + input: ParseStream, + begin: ParseBuffer, + qself: Option, + path: Path, + ) -> Result { + let limits: RangeLimits = input.parse()?; + let hi = input.call(pat_lit_expr)?; + if let Some(hi) = hi { + Ok(Pat::Range(PatRange { + attrs: Vec::new(), + lo: Box::new(Expr::Path(ExprPath { + attrs: Vec::new(), + qself, + path, + })), + limits, + hi, + })) + } else { + Ok(Pat::Verbatim(verbatim::between(begin, input))) + } + } + + fn pat_range_half_open(input: ParseStream, begin: ParseBuffer) -> Result { + let limits: RangeLimits = input.parse()?; + let hi = input.call(pat_lit_expr)?; + if hi.is_some() { + Ok(Pat::Verbatim(verbatim::between(begin, input))) + } else { + match limits { + RangeLimits::HalfOpen(dot2_token) => Ok(Pat::Rest(PatRest { + attrs: Vec::new(), + dot2_token, + })), + RangeLimits::Closed(_) => Err(input.error("expected range upper bound")), + } + } + } + + fn pat_tuple(input: ParseStream) -> Result { + let content; + let paren_token = parenthesized!(content in input); + + let mut elems = Punctuated::new(); + while !content.is_empty() { + let value = multi_pat_with_leading_vert(&content)?; + elems.push_value(value); + if content.is_empty() { + break; + } + let punct = content.parse()?; + elems.push_punct(punct); + } + + Ok(PatTuple { + attrs: Vec::new(), + paren_token, + elems, + }) + } + + fn pat_reference(input: ParseStream) -> Result { + Ok(PatReference { + attrs: Vec::new(), + and_token: input.parse()?, + mutability: input.parse()?, + pat: input.parse()?, + }) + } + + fn pat_lit_or_range(input: ParseStream) -> Result { + let begin = input.fork(); + let lo = input.call(pat_lit_expr)?.unwrap(); + if input.peek(Token![..]) { + let limits: RangeLimits = input.parse()?; + let hi = input.call(pat_lit_expr)?; + if let Some(hi) = hi { + Ok(Pat::Range(PatRange { + attrs: Vec::new(), + lo, + limits, + hi, + })) + } else { + Ok(Pat::Verbatim(verbatim::between(begin, input))) + } + } else if let Expr::Verbatim(verbatim) = *lo { + Ok(Pat::Verbatim(verbatim)) + } else { + Ok(Pat::Lit(PatLit { + attrs: Vec::new(), + expr: lo, + })) + } + } + + fn pat_lit_expr(input: ParseStream) -> Result>> { + if input.is_empty() + || input.peek(Token![|]) + || input.peek(Token![=]) + || input.peek(Token![:]) && !input.peek(Token![::]) + || input.peek(Token![,]) + || input.peek(Token![;]) + { + return Ok(None); + } + + let neg: Option = input.parse()?; + + let lookahead = input.lookahead1(); + let expr = if lookahead.peek(Lit) { + Expr::Lit(input.parse()?) + } else if lookahead.peek(Ident) + || lookahead.peek(Token![::]) + || lookahead.peek(Token![<]) + || lookahead.peek(Token![self]) + || lookahead.peek(Token![Self]) + || lookahead.peek(Token![super]) + || lookahead.peek(Token![crate]) + { + Expr::Path(input.parse()?) + } else if lookahead.peek(Token![const]) { + Expr::Verbatim(input.call(expr::parsing::expr_const)?) + } else { + return Err(lookahead.error()); + }; + + Ok(Some(Box::new(if let Some(neg) = neg { + Expr::Unary(ExprUnary { + attrs: Vec::new(), + op: UnOp::Neg(neg), + expr: Box::new(expr), + }) + } else { + expr + }))) + } + + fn pat_slice(input: ParseStream) -> Result { + let content; + let bracket_token = bracketed!(content in input); + + let mut elems = Punctuated::new(); + while !content.is_empty() { + let value = multi_pat_with_leading_vert(&content)?; + elems.push_value(value); + if content.is_empty() { + break; + } + let punct = content.parse()?; + elems.push_punct(punct); + } + + Ok(PatSlice { + attrs: Vec::new(), + bracket_token, + elems, + }) + } + + fn pat_const(input: ParseStream) -> Result { + let begin = input.fork(); + input.parse::()?; + + let content; + braced!(content in input); + content.call(Attribute::parse_inner)?; + content.call(Block::parse_within)?; + + Ok(verbatim::between(begin, input)) + } + + pub fn multi_pat(input: ParseStream) -> Result { + multi_pat_impl(input, None) + } + + pub fn multi_pat_with_leading_vert(input: ParseStream) -> Result { + let leading_vert: Option = input.parse()?; + multi_pat_impl(input, leading_vert) + } + + fn multi_pat_impl(input: ParseStream, leading_vert: Option) -> Result { + let mut pat: Pat = input.parse()?; + if leading_vert.is_some() + || input.peek(Token![|]) && !input.peek(Token![||]) && !input.peek(Token![|=]) + { + let mut cases = Punctuated::new(); + cases.push_value(pat); + while input.peek(Token![|]) && !input.peek(Token![||]) && !input.peek(Token![|=]) { + let punct = input.parse()?; + cases.push_punct(punct); + let pat: Pat = input.parse()?; + cases.push_value(pat); + } + pat = Pat::Or(PatOr { + attrs: Vec::new(), + leading_vert, + cases, + }); + } + Ok(pat) + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::attr::FilterAttrs; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatWild { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.underscore_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatIdent { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.by_ref.to_tokens(tokens); + self.mutability.to_tokens(tokens); + self.ident.to_tokens(tokens); + if let Some((at_token, subpat)) = &self.subpat { + at_token.to_tokens(tokens); + subpat.to_tokens(tokens); + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatStruct { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.path.to_tokens(tokens); + self.brace_token.surround(tokens, |tokens| { + self.fields.to_tokens(tokens); + // NOTE: We need a comma before the dot2 token if it is present. + if !self.fields.empty_or_trailing() && self.dot2_token.is_some() { + ::default().to_tokens(tokens); + } + self.dot2_token.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatTupleStruct { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.path.to_tokens(tokens); + self.pat.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatType { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.pat.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatPath { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + path::printing::print_path(tokens, &self.qself, &self.path); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatTuple { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.paren_token.surround(tokens, |tokens| { + self.elems.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatBox { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.box_token.to_tokens(tokens); + self.pat.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatReference { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.and_token.to_tokens(tokens); + self.mutability.to_tokens(tokens); + self.pat.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatRest { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.dot2_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatLit { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.expr.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatRange { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.lo.to_tokens(tokens); + self.limits.to_tokens(tokens); + self.hi.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatSlice { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.bracket_token.surround(tokens, |tokens| { + self.elems.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.mac.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PatOr { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.leading_vert.to_tokens(tokens); + self.cases.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for FieldPat { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + if let Some(colon_token) = &self.colon_token { + self.member.to_tokens(tokens); + colon_token.to_tokens(tokens); + } + self.pat.to_tokens(tokens); + } + } +} diff --git a/rust/syn/path.rs b/rust/syn/path.rs new file mode 100644 index 00000000000000..b69c9ceb11d9d0 --- /dev/null +++ b/rust/syn/path.rs @@ -0,0 +1,856 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; + +ast_struct! { + /// A path at which a named item is exported (e.g. `std::collections::HashMap`). + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Path { + pub leading_colon: Option, + pub segments: Punctuated, + } +} + +impl From for Path +where + T: Into, +{ + fn from(segment: T) -> Self { + let mut path = Path { + leading_colon: None, + segments: Punctuated::new(), + }; + path.segments.push_value(segment.into()); + path + } +} + +ast_struct! { + /// A segment of a path together with any path arguments on that segment. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct PathSegment { + pub ident: Ident, + pub arguments: PathArguments, + } +} + +impl From for PathSegment +where + T: Into, +{ + fn from(ident: T) -> Self { + PathSegment { + ident: ident.into(), + arguments: PathArguments::None, + } + } +} + +ast_enum! { + /// Angle bracketed or parenthesized arguments of a path segment. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// ## Angle bracketed + /// + /// The `<'a, T>` in `std::slice::iter<'a, T>`. + /// + /// ## Parenthesized + /// + /// The `(A, B) -> C` in `Fn(A, B) -> C`. + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum PathArguments { + None, + /// The `<'a, T>` in `std::slice::iter<'a, T>`. + AngleBracketed(AngleBracketedGenericArguments), + /// The `(A, B) -> C` in `Fn(A, B) -> C`. + Parenthesized(ParenthesizedGenericArguments), + } +} + +impl Default for PathArguments { + fn default() -> Self { + PathArguments::None + } +} + +impl PathArguments { + pub fn is_empty(&self) -> bool { + match self { + PathArguments::None => true, + PathArguments::AngleBracketed(bracketed) => bracketed.args.is_empty(), + PathArguments::Parenthesized(_) => false, + } + } + + pub fn is_none(&self) -> bool { + match self { + PathArguments::None => true, + PathArguments::AngleBracketed(_) | PathArguments::Parenthesized(_) => false, + } + } +} + +ast_enum! { + /// An individual generic argument, like `'a`, `T`, or `Item = T`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum GenericArgument { + /// A lifetime argument. + Lifetime(Lifetime), + /// A type argument. + Type(Type), + /// A const expression. Must be inside of a block. + /// + /// NOTE: Identity expressions are represented as Type arguments, as + /// they are indistinguishable syntactically. + Const(Expr), + /// A binding (equality constraint) on an associated type: the `Item = + /// u8` in `Iterator`. + Binding(Binding), + /// An associated type bound: `Iterator`. + Constraint(Constraint), + } +} + +ast_struct! { + /// Angle bracketed arguments of a path segment: the `` in `HashMap`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct AngleBracketedGenericArguments { + pub colon2_token: Option, + pub lt_token: Token![<], + pub args: Punctuated, + pub gt_token: Token![>], + } +} + +ast_struct! { + /// A binding (equality constraint) on an associated type: `Item = u8`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Binding { + pub ident: Ident, + pub eq_token: Token![=], + pub ty: Type, + } +} + +ast_struct! { + /// An associated type bound: `Iterator`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Constraint { + pub ident: Ident, + pub colon_token: Token![:], + pub bounds: Punctuated, + } +} + +ast_struct! { + /// Arguments of a function path segment: the `(A, B) -> C` in `Fn(A,B) -> + /// C`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct ParenthesizedGenericArguments { + pub paren_token: token::Paren, + /// `(A, B)` + pub inputs: Punctuated, + /// `C` + pub output: ReturnType, + } +} + +ast_struct! { + /// The explicit Self type in a qualified path: the `T` in `::fmt`. + /// + /// The actual path, including the trait and the associated item, is stored + /// separately. The `position` field represents the index of the associated + /// item qualified with this Self type. + /// + /// ```text + /// as a::b::Trait>::AssociatedItem + /// ^~~~~~ ~~~~~~~~~~~~~~^ + /// ty position = 3 + /// + /// >::AssociatedItem + /// ^~~~~~ ^ + /// ty position = 0 + /// ``` + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct QSelf { + pub lt_token: Token![<], + pub ty: Box, + pub position: usize, + pub as_token: Option, + pub gt_token: Token![>], + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + + use crate::ext::IdentExt; + use crate::parse::{Parse, ParseStream, Result}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Path { + fn parse(input: ParseStream) -> Result { + Self::parse_helper(input, false) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for GenericArgument { + fn parse(input: ParseStream) -> Result { + if input.peek(Lifetime) && !input.peek2(Token![+]) { + return Ok(GenericArgument::Lifetime(input.parse()?)); + } + + if input.peek(Ident) && input.peek2(Token![=]) { + let ident: Ident = input.parse()?; + let eq_token: Token![=] = input.parse()?; + + let ty = if input.peek(Lit) { + let begin = input.fork(); + input.parse::()?; + Type::Verbatim(verbatim::between(begin, input)) + } else if input.peek(token::Brace) { + let begin = input.fork(); + + #[cfg(feature = "full")] + { + input.parse::()?; + } + + #[cfg(not(feature = "full"))] + { + let content; + braced!(content in input); + content.parse::()?; + } + + Type::Verbatim(verbatim::between(begin, input)) + } else { + input.parse()? + }; + + return Ok(GenericArgument::Binding(Binding { + ident, + eq_token, + ty, + })); + } + + #[cfg(feature = "full")] + { + if input.peek(Ident) && input.peek2(Token![:]) && !input.peek2(Token![::]) { + return Ok(GenericArgument::Constraint(input.parse()?)); + } + } + + if input.peek(Lit) || input.peek(token::Brace) { + return const_argument(input).map(GenericArgument::Const); + } + + #[cfg(feature = "full")] + let begin = input.fork(); + + let argument: Type = input.parse()?; + + #[cfg(feature = "full")] + { + if match &argument { + Type::Path(argument) + if argument.qself.is_none() + && argument.path.leading_colon.is_none() + && argument.path.segments.len() == 1 => + { + match argument.path.segments[0].arguments { + PathArguments::AngleBracketed(_) => true, + _ => false, + } + } + _ => false, + } && if input.peek(Token![=]) { + input.parse::()?; + input.parse::()?; + true + } else if input.peek(Token![:]) { + input.parse::()?; + input.call(constraint_bounds)?; + true + } else { + false + } { + let verbatim = verbatim::between(begin, input); + return Ok(GenericArgument::Type(Type::Verbatim(verbatim))); + } + } + + Ok(GenericArgument::Type(argument)) + } + } + + pub fn const_argument(input: ParseStream) -> Result { + let lookahead = input.lookahead1(); + + if input.peek(Lit) { + let lit = input.parse()?; + return Ok(Expr::Lit(lit)); + } + + #[cfg(feature = "full")] + { + if input.peek(Ident) { + let ident: Ident = input.parse()?; + return Ok(Expr::Path(ExprPath { + attrs: Vec::new(), + qself: None, + path: Path::from(ident), + })); + } + } + + if input.peek(token::Brace) { + #[cfg(feature = "full")] + { + let block: ExprBlock = input.parse()?; + return Ok(Expr::Block(block)); + } + + #[cfg(not(feature = "full"))] + { + let begin = input.fork(); + let content; + braced!(content in input); + content.parse::()?; + let verbatim = verbatim::between(begin, input); + return Ok(Expr::Verbatim(verbatim)); + } + } + + Err(lookahead.error()) + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for AngleBracketedGenericArguments { + fn parse(input: ParseStream) -> Result { + Ok(AngleBracketedGenericArguments { + colon2_token: input.parse()?, + lt_token: input.parse()?, + args: { + let mut args = Punctuated::new(); + loop { + if input.peek(Token![>]) { + break; + } + let value = input.parse()?; + args.push_value(value); + if input.peek(Token![>]) { + break; + } + let punct = input.parse()?; + args.push_punct(punct); + } + args + }, + gt_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ParenthesizedGenericArguments { + fn parse(input: ParseStream) -> Result { + let content; + Ok(ParenthesizedGenericArguments { + paren_token: parenthesized!(content in input), + inputs: content.parse_terminated(Type::parse)?, + output: input.call(ReturnType::without_plus)?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for PathSegment { + fn parse(input: ParseStream) -> Result { + Self::parse_helper(input, false) + } + } + + impl PathSegment { + fn parse_helper(input: ParseStream, expr_style: bool) -> Result { + if input.peek(Token![super]) || input.peek(Token![self]) || input.peek(Token![crate]) { + let ident = input.call(Ident::parse_any)?; + return Ok(PathSegment::from(ident)); + } + + let ident = if input.peek(Token![Self]) { + input.call(Ident::parse_any)? + } else { + input.parse()? + }; + + if !expr_style && input.peek(Token![<]) && !input.peek(Token![<=]) + || input.peek(Token![::]) && input.peek3(Token![<]) + { + Ok(PathSegment { + ident, + arguments: PathArguments::AngleBracketed(input.parse()?), + }) + } else { + Ok(PathSegment::from(ident)) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Binding { + fn parse(input: ParseStream) -> Result { + Ok(Binding { + ident: input.parse()?, + eq_token: input.parse()?, + ty: input.parse()?, + }) + } + } + + #[cfg(feature = "full")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Constraint { + fn parse(input: ParseStream) -> Result { + Ok(Constraint { + ident: input.parse()?, + colon_token: input.parse()?, + bounds: constraint_bounds(input)?, + }) + } + } + + #[cfg(feature = "full")] + fn constraint_bounds(input: ParseStream) -> Result> { + let mut bounds = Punctuated::new(); + loop { + if input.peek(Token![,]) || input.peek(Token![>]) { + break; + } + let value = input.parse()?; + bounds.push_value(value); + if !input.peek(Token![+]) { + break; + } + let punct = input.parse()?; + bounds.push_punct(punct); + } + Ok(bounds) + } + + impl Path { + /// Parse a `Path` containing no path arguments on any of its segments. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + /// + /// # Example + /// + /// ``` + /// use syn::{Path, Result, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // A simplified single `use` statement like: + /// // + /// // use std::collections::HashMap; + /// // + /// // Note that generic parameters are not allowed in a `use` statement + /// // so the following must not be accepted. + /// // + /// // use a::::c; + /// struct SingleUse { + /// use_token: Token![use], + /// path: Path, + /// } + /// + /// impl Parse for SingleUse { + /// fn parse(input: ParseStream) -> Result { + /// Ok(SingleUse { + /// use_token: input.parse()?, + /// path: input.call(Path::parse_mod_style)?, + /// }) + /// } + /// } + /// ``` + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_mod_style(input: ParseStream) -> Result { + Ok(Path { + leading_colon: input.parse()?, + segments: { + let mut segments = Punctuated::new(); + loop { + if !input.peek(Ident) + && !input.peek(Token![super]) + && !input.peek(Token![self]) + && !input.peek(Token![Self]) + && !input.peek(Token![crate]) + { + break; + } + let ident = Ident::parse_any(input)?; + segments.push_value(PathSegment::from(ident)); + if !input.peek(Token![::]) { + break; + } + let punct = input.parse()?; + segments.push_punct(punct); + } + if segments.is_empty() { + return Err(input.error("expected path")); + } else if segments.trailing_punct() { + return Err(input.error("expected path segment")); + } + segments + }, + }) + } + + /// Determines whether this is a path of length 1 equal to the given + /// ident. + /// + /// For them to compare equal, it must be the case that: + /// + /// - the path has no leading colon, + /// - the number of path segments is 1, + /// - the first path segment has no angle bracketed or parenthesized + /// path arguments, and + /// - the ident of the first path segment is equal to the given one. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + /// + /// # Example + /// + /// ``` + /// use syn::{Attribute, Error, Meta, NestedMeta, Result}; + /// # use std::iter::FromIterator; + /// + /// fn get_serde_meta_items(attr: &Attribute) -> Result> { + /// if attr.path.is_ident("serde") { + /// match attr.parse_meta()? { + /// Meta::List(meta) => Ok(Vec::from_iter(meta.nested)), + /// bad => Err(Error::new_spanned(bad, "unrecognized attribute")), + /// } + /// } else { + /// Ok(Vec::new()) + /// } + /// } + /// ``` + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn is_ident(&self, ident: &I) -> bool + where + Ident: PartialEq, + { + match self.get_ident() { + Some(id) => id == ident, + None => false, + } + } + + /// If this path consists of a single ident, returns the ident. + /// + /// A path is considered an ident if: + /// + /// - the path has no leading colon, + /// - the number of path segments is 1, and + /// - the first path segment has no angle bracketed or parenthesized + /// path arguments. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn get_ident(&self) -> Option<&Ident> { + if self.leading_colon.is_none() + && self.segments.len() == 1 + && self.segments[0].arguments.is_none() + { + Some(&self.segments[0].ident) + } else { + None + } + } + + pub(crate) fn parse_helper(input: ParseStream, expr_style: bool) -> Result { + let mut path = Path { + leading_colon: input.parse()?, + segments: { + let mut segments = Punctuated::new(); + let value = PathSegment::parse_helper(input, expr_style)?; + segments.push_value(value); + segments + }, + }; + Path::parse_rest(input, &mut path, expr_style)?; + Ok(path) + } + + pub(crate) fn parse_rest( + input: ParseStream, + path: &mut Self, + expr_style: bool, + ) -> Result<()> { + while input.peek(Token![::]) && !input.peek3(token::Paren) { + let punct: Token![::] = input.parse()?; + path.segments.push_punct(punct); + let value = PathSegment::parse_helper(input, expr_style)?; + path.segments.push_value(value); + } + Ok(()) + } + } + + pub fn qpath(input: ParseStream, expr_style: bool) -> Result<(Option, Path)> { + if input.peek(Token![<]) { + let lt_token: Token![<] = input.parse()?; + let this: Type = input.parse()?; + let path = if input.peek(Token![as]) { + let as_token: Token![as] = input.parse()?; + let path: Path = input.parse()?; + Some((as_token, path)) + } else { + None + }; + let gt_token: Token![>] = input.parse()?; + let colon2_token: Token![::] = input.parse()?; + let mut rest = Punctuated::new(); + loop { + let path = PathSegment::parse_helper(input, expr_style)?; + rest.push_value(path); + if !input.peek(Token![::]) { + break; + } + let punct: Token![::] = input.parse()?; + rest.push_punct(punct); + } + let (position, as_token, path) = match path { + Some((as_token, mut path)) => { + let pos = path.segments.len(); + path.segments.push_punct(colon2_token); + path.segments.extend(rest.into_pairs()); + (pos, Some(as_token), path) + } + None => { + let path = Path { + leading_colon: Some(colon2_token), + segments: rest, + }; + (0, None, path) + } + }; + let qself = QSelf { + lt_token, + ty: Box::new(this), + position, + as_token, + gt_token, + }; + Ok((Some(qself), path)) + } else { + let path = Path::parse_helper(input, expr_style)?; + Ok((None, path)) + } + } +} + +#[cfg(feature = "printing")] +pub(crate) mod printing { + use super::*; + use crate::print::TokensOrDefault; + use proc_macro2::TokenStream; + use quote::ToTokens; + use std::cmp; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Path { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.leading_colon.to_tokens(tokens); + self.segments.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PathSegment { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + self.arguments.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for PathArguments { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + PathArguments::None => {} + PathArguments::AngleBracketed(arguments) => { + arguments.to_tokens(tokens); + } + PathArguments::Parenthesized(arguments) => { + arguments.to_tokens(tokens); + } + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for GenericArgument { + #[allow(clippy::match_same_arms)] + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + GenericArgument::Lifetime(lt) => lt.to_tokens(tokens), + GenericArgument::Type(ty) => ty.to_tokens(tokens), + GenericArgument::Const(e) => match *e { + Expr::Lit(_) => e.to_tokens(tokens), + + // NOTE: We should probably support parsing blocks with only + // expressions in them without the full feature for const + // generics. + #[cfg(feature = "full")] + Expr::Block(_) => e.to_tokens(tokens), + + // ERROR CORRECTION: Add braces to make sure that the + // generated code is valid. + _ => token::Brace::default().surround(tokens, |tokens| { + e.to_tokens(tokens); + }), + }, + GenericArgument::Binding(tb) => tb.to_tokens(tokens), + GenericArgument::Constraint(tc) => tc.to_tokens(tokens), + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for AngleBracketedGenericArguments { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.colon2_token.to_tokens(tokens); + self.lt_token.to_tokens(tokens); + + // Print lifetimes before types/consts/bindings, regardless of their + // order in self.args. + let mut trailing_or_empty = true; + for param in self.args.pairs() { + match **param.value() { + GenericArgument::Lifetime(_) => { + param.to_tokens(tokens); + trailing_or_empty = param.punct().is_some(); + } + GenericArgument::Type(_) + | GenericArgument::Const(_) + | GenericArgument::Binding(_) + | GenericArgument::Constraint(_) => {} + } + } + for param in self.args.pairs() { + match **param.value() { + GenericArgument::Type(_) + | GenericArgument::Const(_) + | GenericArgument::Binding(_) + | GenericArgument::Constraint(_) => { + if !trailing_or_empty { + ::default().to_tokens(tokens); + } + param.to_tokens(tokens); + trailing_or_empty = param.punct().is_some(); + } + GenericArgument::Lifetime(_) => {} + } + } + + self.gt_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Binding { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + self.eq_token.to_tokens(tokens); + self.ty.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Constraint { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.ident.to_tokens(tokens); + self.colon_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ParenthesizedGenericArguments { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.paren_token.surround(tokens, |tokens| { + self.inputs.to_tokens(tokens); + }); + self.output.to_tokens(tokens); + } + } + + pub(crate) fn print_path(tokens: &mut TokenStream, qself: &Option, path: &Path) { + let qself = match qself { + Some(qself) => qself, + None => { + path.to_tokens(tokens); + return; + } + }; + qself.lt_token.to_tokens(tokens); + qself.ty.to_tokens(tokens); + + let pos = cmp::min(qself.position, path.segments.len()); + let mut segments = path.segments.pairs(); + if pos > 0 { + TokensOrDefault(&qself.as_token).to_tokens(tokens); + path.leading_colon.to_tokens(tokens); + for (i, segment) in segments.by_ref().take(pos).enumerate() { + if i + 1 == pos { + segment.value().to_tokens(tokens); + qself.gt_token.to_tokens(tokens); + segment.punct().to_tokens(tokens); + } else { + segment.to_tokens(tokens); + } + } + } else { + qself.gt_token.to_tokens(tokens); + path.leading_colon.to_tokens(tokens); + } + for segment in segments { + segment.to_tokens(tokens); + } + } +} diff --git a/rust/syn/print.rs b/rust/syn/print.rs new file mode 100644 index 00000000000000..abc287ee950180 --- /dev/null +++ b/rust/syn/print.rs @@ -0,0 +1,18 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::TokenStream; +use quote::ToTokens; + +pub struct TokensOrDefault<'a, T: 'a>(pub &'a Option); + +impl<'a, T> ToTokens for TokensOrDefault<'a, T> +where + T: ToTokens + Default, +{ + fn to_tokens(&self, tokens: &mut TokenStream) { + match self.0 { + Some(t) => t.to_tokens(tokens), + None => T::default().to_tokens(tokens), + } + } +} diff --git a/rust/syn/punctuated.rs b/rust/syn/punctuated.rs new file mode 100644 index 00000000000000..b1ad84b301faed --- /dev/null +++ b/rust/syn/punctuated.rs @@ -0,0 +1,1070 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! A punctuated sequence of syntax tree nodes separated by punctuation. +//! +//! Lots of things in Rust are punctuated sequences. +//! +//! - The fields of a struct are `Punctuated`. +//! - The segments of a path are `Punctuated`. +//! - The bounds on a generic parameter are `Punctuated`. +//! - The arguments to a function call are `Punctuated`. +//! +//! This module provides a common representation for these punctuated sequences +//! in the form of the [`Punctuated`] type. We store a vector of pairs of +//! syntax tree node + punctuation, where every node in the sequence is followed +//! by punctuation except for possibly the final one. +//! +//! [`Punctuated`]: Punctuated +//! +//! ```text +//! a_function_call(arg1, arg2, arg3); +//! ~~~~^ ~~~~^ ~~~~ +//! ``` + +#[cfg(feature = "extra-traits")] +use std::fmt::{self, Debug}; +#[cfg(feature = "extra-traits")] +use std::hash::{Hash, Hasher}; +#[cfg(any(feature = "full", feature = "derive"))] +use std::iter; +use std::iter::FromIterator; +use std::ops::{Index, IndexMut}; +use std::option; +use std::slice; +use std::vec; + +#[cfg(feature = "parsing")] +use crate::parse::{Parse, ParseStream, Result}; +#[cfg(feature = "parsing")] +use crate::token::Token; + +/// A punctuated sequence of syntax tree nodes of type `T` separated by +/// punctuation of type `P`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct Punctuated { + inner: Vec<(T, P)>, + last: Option>, +} + +impl Punctuated { + /// Creates an empty punctuated sequence. + #[cfg(not(syn_no_const_vec_new))] + pub const fn new() -> Self { + Punctuated { + inner: Vec::new(), + last: None, + } + } + + /// Creates an empty punctuated sequence. + #[cfg(syn_no_const_vec_new)] + pub fn new() -> Self { + Punctuated { + inner: Vec::new(), + last: None, + } + } + + /// Determines whether this punctuated sequence is empty, meaning it + /// contains no syntax tree nodes or punctuation. + pub fn is_empty(&self) -> bool { + self.inner.len() == 0 && self.last.is_none() + } + + /// Returns the number of syntax tree nodes in this punctuated sequence. + /// + /// This is the number of nodes of type `T`, not counting the punctuation of + /// type `P`. + pub fn len(&self) -> usize { + self.inner.len() + if self.last.is_some() { 1 } else { 0 } + } + + /// Borrows the first element in this sequence. + pub fn first(&self) -> Option<&T> { + self.iter().next() + } + + /// Mutably borrows the first element in this sequence. + pub fn first_mut(&mut self) -> Option<&mut T> { + self.iter_mut().next() + } + + /// Borrows the last element in this sequence. + pub fn last(&self) -> Option<&T> { + self.iter().next_back() + } + + /// Mutably borrows the last element in this sequence. + pub fn last_mut(&mut self) -> Option<&mut T> { + self.iter_mut().next_back() + } + + /// Returns an iterator over borrowed syntax tree nodes of type `&T`. + pub fn iter(&self) -> Iter { + Iter { + inner: Box::new(PrivateIter { + inner: self.inner.iter(), + last: self.last.as_ref().map(Box::as_ref).into_iter(), + }), + } + } + + /// Returns an iterator over mutably borrowed syntax tree nodes of type + /// `&mut T`. + pub fn iter_mut(&mut self) -> IterMut { + IterMut { + inner: Box::new(PrivateIterMut { + inner: self.inner.iter_mut(), + last: self.last.as_mut().map(Box::as_mut).into_iter(), + }), + } + } + + /// Returns an iterator over the contents of this sequence as borrowed + /// punctuated pairs. + pub fn pairs(&self) -> Pairs { + Pairs { + inner: self.inner.iter(), + last: self.last.as_ref().map(Box::as_ref).into_iter(), + } + } + + /// Returns an iterator over the contents of this sequence as mutably + /// borrowed punctuated pairs. + pub fn pairs_mut(&mut self) -> PairsMut { + PairsMut { + inner: self.inner.iter_mut(), + last: self.last.as_mut().map(Box::as_mut).into_iter(), + } + } + + /// Returns an iterator over the contents of this sequence as owned + /// punctuated pairs. + pub fn into_pairs(self) -> IntoPairs { + IntoPairs { + inner: self.inner.into_iter(), + last: self.last.map(|t| *t).into_iter(), + } + } + + /// Appends a syntax tree node onto the end of this punctuated sequence. The + /// sequence must previously have a trailing punctuation. + /// + /// Use [`push`] instead if the punctuated sequence may or may not already + /// have trailing punctuation. + /// + /// [`push`]: Punctuated::push + /// + /// # Panics + /// + /// Panics if the sequence does not already have a trailing punctuation when + /// this method is called. + pub fn push_value(&mut self, value: T) { + assert!( + self.empty_or_trailing(), + "Punctuated::push_value: cannot push value if Punctuated is missing trailing punctuation", + ); + + self.last = Some(Box::new(value)); + } + + /// Appends a trailing punctuation onto the end of this punctuated sequence. + /// The sequence must be non-empty and must not already have trailing + /// punctuation. + /// + /// # Panics + /// + /// Panics if the sequence is empty or already has a trailing punctuation. + pub fn push_punct(&mut self, punctuation: P) { + assert!( + self.last.is_some(), + "Punctuated::push_punct: cannot push punctuation if Punctuated is empty or already has trailing punctuation", + ); + + let last = self.last.take().unwrap(); + self.inner.push((*last, punctuation)); + } + + /// Removes the last punctuated pair from this sequence, or `None` if the + /// sequence is empty. + pub fn pop(&mut self) -> Option> { + if self.last.is_some() { + self.last.take().map(|t| Pair::End(*t)) + } else { + self.inner.pop().map(|(t, p)| Pair::Punctuated(t, p)) + } + } + + /// Determines whether this punctuated sequence ends with a trailing + /// punctuation. + pub fn trailing_punct(&self) -> bool { + self.last.is_none() && !self.is_empty() + } + + /// Returns true if either this `Punctuated` is empty, or it has a trailing + /// punctuation. + /// + /// Equivalent to `punctuated.is_empty() || punctuated.trailing_punct()`. + pub fn empty_or_trailing(&self) -> bool { + self.last.is_none() + } + + /// Appends a syntax tree node onto the end of this punctuated sequence. + /// + /// If there is not a trailing punctuation in this sequence when this method + /// is called, the default value of punctuation type `P` is inserted before + /// the given value of type `T`. + pub fn push(&mut self, value: T) + where + P: Default, + { + if !self.empty_or_trailing() { + self.push_punct(Default::default()); + } + self.push_value(value); + } + + /// Inserts an element at position `index`. + /// + /// # Panics + /// + /// Panics if `index` is greater than the number of elements previously in + /// this punctuated sequence. + pub fn insert(&mut self, index: usize, value: T) + where + P: Default, + { + assert!( + index <= self.len(), + "Punctuated::insert: index out of range", + ); + + if index == self.len() { + self.push(value); + } else { + self.inner.insert(index, (value, Default::default())); + } + } + + /// Clears the sequence of all values and punctuation, making it empty. + pub fn clear(&mut self) { + self.inner.clear(); + self.last = None; + } + + /// Parses zero or more occurrences of `T` separated by punctuation of type + /// `P`, with optional trailing punctuation. + /// + /// Parsing continues until the end of this parse stream. The entire content + /// of this parse stream must consist of `T` and `P`. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_terminated(input: ParseStream) -> Result + where + T: Parse, + P: Parse, + { + Self::parse_terminated_with(input, T::parse) + } + + /// Parses zero or more occurrences of `T` using the given parse function, + /// separated by punctuation of type `P`, with optional trailing + /// punctuation. + /// + /// Like [`parse_terminated`], the entire content of this stream is expected + /// to be parsed. + /// + /// [`parse_terminated`]: Punctuated::parse_terminated + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_terminated_with( + input: ParseStream, + parser: fn(ParseStream) -> Result, + ) -> Result + where + P: Parse, + { + let mut punctuated = Punctuated::new(); + + loop { + if input.is_empty() { + break; + } + let value = parser(input)?; + punctuated.push_value(value); + if input.is_empty() { + break; + } + let punct = input.parse()?; + punctuated.push_punct(punct); + } + + Ok(punctuated) + } + + /// Parses one or more occurrences of `T` separated by punctuation of type + /// `P`, not accepting trailing punctuation. + /// + /// Parsing continues as long as punctuation `P` is present at the head of + /// the stream. This method returns upon parsing a `T` and observing that it + /// is not followed by a `P`, even if there are remaining tokens in the + /// stream. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_separated_nonempty(input: ParseStream) -> Result + where + T: Parse, + P: Token + Parse, + { + Self::parse_separated_nonempty_with(input, T::parse) + } + + /// Parses one or more occurrences of `T` using the given parse function, + /// separated by punctuation of type `P`, not accepting trailing + /// punctuation. + /// + /// Like [`parse_separated_nonempty`], may complete early without parsing + /// the entire content of this stream. + /// + /// [`parse_separated_nonempty`]: Punctuated::parse_separated_nonempty + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_separated_nonempty_with( + input: ParseStream, + parser: fn(ParseStream) -> Result, + ) -> Result + where + P: Token + Parse, + { + let mut punctuated = Punctuated::new(); + + loop { + let value = parser(input)?; + punctuated.push_value(value); + if !P::peek(input.cursor()) { + break; + } + let punct = input.parse()?; + punctuated.push_punct(punct); + } + + Ok(punctuated) + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for Punctuated +where + T: Clone, + P: Clone, +{ + fn clone(&self) -> Self { + Punctuated { + inner: self.inner.clone(), + last: self.last.clone(), + } + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Eq for Punctuated +where + T: Eq, + P: Eq, +{ +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl PartialEq for Punctuated +where + T: PartialEq, + P: PartialEq, +{ + fn eq(&self, other: &Self) -> bool { + let Punctuated { inner, last } = self; + *inner == other.inner && *last == other.last + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Hash for Punctuated +where + T: Hash, + P: Hash, +{ + fn hash(&self, state: &mut H) { + let Punctuated { inner, last } = self; + inner.hash(state); + last.hash(state); + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Debug for Punctuated { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + let mut list = f.debug_list(); + for (t, p) in &self.inner { + list.entry(t); + list.entry(p); + } + if let Some(last) = &self.last { + list.entry(last); + } + list.finish() + } +} + +impl FromIterator for Punctuated +where + P: Default, +{ + fn from_iter>(i: I) -> Self { + let mut ret = Punctuated::new(); + ret.extend(i); + ret + } +} + +impl Extend for Punctuated +where + P: Default, +{ + fn extend>(&mut self, i: I) { + for value in i { + self.push(value); + } + } +} + +impl FromIterator> for Punctuated { + fn from_iter>>(i: I) -> Self { + let mut ret = Punctuated::new(); + ret.extend(i); + ret + } +} + +impl Extend> for Punctuated { + fn extend>>(&mut self, i: I) { + assert!( + self.empty_or_trailing(), + "Punctuated::extend: Punctuated is not empty or does not have a trailing punctuation", + ); + + let mut nomore = false; + for pair in i { + if nomore { + panic!("Punctuated extended with items after a Pair::End"); + } + match pair { + Pair::Punctuated(a, b) => self.inner.push((a, b)), + Pair::End(a) => { + self.last = Some(Box::new(a)); + nomore = true; + } + } + } + } +} + +impl IntoIterator for Punctuated { + type Item = T; + type IntoIter = IntoIter; + + fn into_iter(self) -> Self::IntoIter { + let mut elements = Vec::with_capacity(self.len()); + elements.extend(self.inner.into_iter().map(|pair| pair.0)); + elements.extend(self.last.map(|t| *t)); + + IntoIter { + inner: elements.into_iter(), + } + } +} + +impl<'a, T, P> IntoIterator for &'a Punctuated { + type Item = &'a T; + type IntoIter = Iter<'a, T>; + + fn into_iter(self) -> Self::IntoIter { + Punctuated::iter(self) + } +} + +impl<'a, T, P> IntoIterator for &'a mut Punctuated { + type Item = &'a mut T; + type IntoIter = IterMut<'a, T>; + + fn into_iter(self) -> Self::IntoIter { + Punctuated::iter_mut(self) + } +} + +impl Default for Punctuated { + fn default() -> Self { + Punctuated::new() + } +} + +/// An iterator over borrowed pairs of type `Pair<&T, &P>`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct Pairs<'a, T: 'a, P: 'a> { + inner: slice::Iter<'a, (T, P)>, + last: option::IntoIter<&'a T>, +} + +impl<'a, T, P> Iterator for Pairs<'a, T, P> { + type Item = Pair<&'a T, &'a P>; + + fn next(&mut self) -> Option { + self.inner + .next() + .map(|(t, p)| Pair::Punctuated(t, p)) + .or_else(|| self.last.next().map(Pair::End)) + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl<'a, T, P> DoubleEndedIterator for Pairs<'a, T, P> { + fn next_back(&mut self) -> Option { + self.last + .next() + .map(Pair::End) + .or_else(|| self.inner.next_back().map(|(t, p)| Pair::Punctuated(t, p))) + } +} + +impl<'a, T, P> ExactSizeIterator for Pairs<'a, T, P> { + fn len(&self) -> usize { + self.inner.len() + self.last.len() + } +} + +// No Clone bound on T or P. +impl<'a, T, P> Clone for Pairs<'a, T, P> { + fn clone(&self) -> Self { + Pairs { + inner: self.inner.clone(), + last: self.last.clone(), + } + } +} + +/// An iterator over mutably borrowed pairs of type `Pair<&mut T, &mut P>`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct PairsMut<'a, T: 'a, P: 'a> { + inner: slice::IterMut<'a, (T, P)>, + last: option::IntoIter<&'a mut T>, +} + +impl<'a, T, P> Iterator for PairsMut<'a, T, P> { + type Item = Pair<&'a mut T, &'a mut P>; + + fn next(&mut self) -> Option { + self.inner + .next() + .map(|(t, p)| Pair::Punctuated(t, p)) + .or_else(|| self.last.next().map(Pair::End)) + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl<'a, T, P> DoubleEndedIterator for PairsMut<'a, T, P> { + fn next_back(&mut self) -> Option { + self.last + .next() + .map(Pair::End) + .or_else(|| self.inner.next_back().map(|(t, p)| Pair::Punctuated(t, p))) + } +} + +impl<'a, T, P> ExactSizeIterator for PairsMut<'a, T, P> { + fn len(&self) -> usize { + self.inner.len() + self.last.len() + } +} + +/// An iterator over owned pairs of type `Pair`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct IntoPairs { + inner: vec::IntoIter<(T, P)>, + last: option::IntoIter, +} + +impl Iterator for IntoPairs { + type Item = Pair; + + fn next(&mut self) -> Option { + self.inner + .next() + .map(|(t, p)| Pair::Punctuated(t, p)) + .or_else(|| self.last.next().map(Pair::End)) + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl DoubleEndedIterator for IntoPairs { + fn next_back(&mut self) -> Option { + self.last + .next() + .map(Pair::End) + .or_else(|| self.inner.next_back().map(|(t, p)| Pair::Punctuated(t, p))) + } +} + +impl ExactSizeIterator for IntoPairs { + fn len(&self) -> usize { + self.inner.len() + self.last.len() + } +} + +impl Clone for IntoPairs +where + T: Clone, + P: Clone, +{ + fn clone(&self) -> Self { + IntoPairs { + inner: self.inner.clone(), + last: self.last.clone(), + } + } +} + +/// An iterator over owned values of type `T`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct IntoIter { + inner: vec::IntoIter, +} + +impl Iterator for IntoIter { + type Item = T; + + fn next(&mut self) -> Option { + self.inner.next() + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl DoubleEndedIterator for IntoIter { + fn next_back(&mut self) -> Option { + self.inner.next_back() + } +} + +impl ExactSizeIterator for IntoIter { + fn len(&self) -> usize { + self.inner.len() + } +} + +impl Clone for IntoIter +where + T: Clone, +{ + fn clone(&self) -> Self { + IntoIter { + inner: self.inner.clone(), + } + } +} + +/// An iterator over borrowed values of type `&T`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct Iter<'a, T: 'a> { + // The `Item = &'a T` needs to be specified to support rustc 1.31 and older. + // On modern compilers we would be able to write just IterTrait<'a, T> where + // Item can be inferred unambiguously from the supertrait. + inner: Box + 'a>, +} + +trait IterTrait<'a, T: 'a>: + DoubleEndedIterator + ExactSizeIterator +{ + fn clone_box(&self) -> Box + 'a>; +} + +struct PrivateIter<'a, T: 'a, P: 'a> { + inner: slice::Iter<'a, (T, P)>, + last: option::IntoIter<&'a T>, +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn empty_punctuated_iter<'a, T>() -> Iter<'a, T> { + Iter { + inner: Box::new(iter::empty()), + } +} + +// No Clone bound on T. +impl<'a, T> Clone for Iter<'a, T> { + fn clone(&self) -> Self { + Iter { + inner: self.inner.clone_box(), + } + } +} + +impl<'a, T> Iterator for Iter<'a, T> { + type Item = &'a T; + + fn next(&mut self) -> Option { + self.inner.next() + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl<'a, T> DoubleEndedIterator for Iter<'a, T> { + fn next_back(&mut self) -> Option { + self.inner.next_back() + } +} + +impl<'a, T> ExactSizeIterator for Iter<'a, T> { + fn len(&self) -> usize { + self.inner.len() + } +} + +impl<'a, T, P> Iterator for PrivateIter<'a, T, P> { + type Item = &'a T; + + fn next(&mut self) -> Option { + self.inner + .next() + .map(|pair| &pair.0) + .or_else(|| self.last.next()) + } +} + +impl<'a, T, P> DoubleEndedIterator for PrivateIter<'a, T, P> { + fn next_back(&mut self) -> Option { + self.last + .next() + .or_else(|| self.inner.next_back().map(|pair| &pair.0)) + } +} + +impl<'a, T, P> ExactSizeIterator for PrivateIter<'a, T, P> { + fn len(&self) -> usize { + self.inner.len() + self.last.len() + } +} + +// No Clone bound on T or P. +impl<'a, T, P> Clone for PrivateIter<'a, T, P> { + fn clone(&self) -> Self { + PrivateIter { + inner: self.inner.clone(), + last: self.last.clone(), + } + } +} + +impl<'a, T, I> IterTrait<'a, T> for I +where + T: 'a, + I: DoubleEndedIterator + ExactSizeIterator + Clone + 'a, +{ + fn clone_box(&self) -> Box + 'a> { + Box::new(self.clone()) + } +} + +/// An iterator over mutably borrowed values of type `&mut T`. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub struct IterMut<'a, T: 'a> { + inner: Box + 'a>, +} + +trait IterMutTrait<'a, T: 'a>: + DoubleEndedIterator + ExactSizeIterator +{ +} + +struct PrivateIterMut<'a, T: 'a, P: 'a> { + inner: slice::IterMut<'a, (T, P)>, + last: option::IntoIter<&'a mut T>, +} + +#[cfg(any(feature = "full", feature = "derive"))] +pub(crate) fn empty_punctuated_iter_mut<'a, T>() -> IterMut<'a, T> { + IterMut { + inner: Box::new(iter::empty()), + } +} + +impl<'a, T> Iterator for IterMut<'a, T> { + type Item = &'a mut T; + + fn next(&mut self) -> Option { + self.inner.next() + } + + fn size_hint(&self) -> (usize, Option) { + (self.len(), Some(self.len())) + } +} + +impl<'a, T> DoubleEndedIterator for IterMut<'a, T> { + fn next_back(&mut self) -> Option { + self.inner.next_back() + } +} + +impl<'a, T> ExactSizeIterator for IterMut<'a, T> { + fn len(&self) -> usize { + self.inner.len() + } +} + +impl<'a, T, P> Iterator for PrivateIterMut<'a, T, P> { + type Item = &'a mut T; + + fn next(&mut self) -> Option { + self.inner + .next() + .map(|pair| &mut pair.0) + .or_else(|| self.last.next()) + } +} + +impl<'a, T, P> DoubleEndedIterator for PrivateIterMut<'a, T, P> { + fn next_back(&mut self) -> Option { + self.last + .next() + .or_else(|| self.inner.next_back().map(|pair| &mut pair.0)) + } +} + +impl<'a, T, P> ExactSizeIterator for PrivateIterMut<'a, T, P> { + fn len(&self) -> usize { + self.inner.len() + self.last.len() + } +} + +impl<'a, T, I> IterMutTrait<'a, T> for I +where + T: 'a, + I: DoubleEndedIterator + ExactSizeIterator + 'a, +{ +} + +/// A single syntax tree node of type `T` followed by its trailing punctuation +/// of type `P` if any. +/// +/// Refer to the [module documentation] for details about punctuated sequences. +/// +/// [module documentation]: self +pub enum Pair { + Punctuated(T, P), + End(T), +} + +impl Pair { + /// Extracts the syntax tree node from this punctuated pair, discarding the + /// following punctuation. + pub fn into_value(self) -> T { + match self { + Pair::Punctuated(t, _) | Pair::End(t) => t, + } + } + + /// Borrows the syntax tree node from this punctuated pair. + pub fn value(&self) -> &T { + match self { + Pair::Punctuated(t, _) | Pair::End(t) => t, + } + } + + /// Mutably borrows the syntax tree node from this punctuated pair. + pub fn value_mut(&mut self) -> &mut T { + match self { + Pair::Punctuated(t, _) | Pair::End(t) => t, + } + } + + /// Borrows the punctuation from this punctuated pair, unless this pair is + /// the final one and there is no trailing punctuation. + pub fn punct(&self) -> Option<&P> { + match self { + Pair::Punctuated(_, p) => Some(p), + Pair::End(_) => None, + } + } + + /// Mutably borrows the punctuation from this punctuated pair, unless the + /// pair is the final one and there is no trailing punctuation. + /// + /// # Example + /// + /// ``` + /// # use proc_macro2::Span; + /// # use syn::punctuated::Punctuated; + /// # use syn::{parse_quote, Token, TypeParamBound}; + /// # + /// # let mut punctuated = Punctuated::::new(); + /// # let span = Span::call_site(); + /// # + /// punctuated.insert(0, parse_quote!('lifetime)); + /// if let Some(punct) = punctuated.pairs_mut().next().unwrap().punct_mut() { + /// punct.span = span; + /// } + /// ``` + pub fn punct_mut(&mut self) -> Option<&mut P> { + match self { + Pair::Punctuated(_, p) => Some(p), + Pair::End(_) => None, + } + } + + /// Creates a punctuated pair out of a syntax tree node and an optional + /// following punctuation. + pub fn new(t: T, p: Option

) -> Self { + match p { + Some(p) => Pair::Punctuated(t, p), + None => Pair::End(t), + } + } + + /// Produces this punctuated pair as a tuple of syntax tree node and + /// optional following punctuation. + pub fn into_tuple(self) -> (T, Option

) { + match self { + Pair::Punctuated(t, p) => (t, Some(p)), + Pair::End(t) => (t, None), + } + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for Pair +where + T: Clone, + P: Clone, +{ + fn clone(&self) -> Self { + match self { + Pair::Punctuated(t, p) => Pair::Punctuated(t.clone(), p.clone()), + Pair::End(t) => Pair::End(t.clone()), + } + } +} + +impl Index for Punctuated { + type Output = T; + + fn index(&self, index: usize) -> &Self::Output { + if index == self.len() - 1 { + match &self.last { + Some(t) => t, + None => &self.inner[index].0, + } + } else { + &self.inner[index].0 + } + } +} + +impl IndexMut for Punctuated { + fn index_mut(&mut self, index: usize) -> &mut Self::Output { + if index == self.len() - 1 { + match &mut self.last { + Some(t) => t, + None => &mut self.inner[index].0, + } + } else { + &mut self.inner[index].0 + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Punctuated + where + T: ToTokens, + P: ToTokens, + { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.pairs()); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Pair + where + T: ToTokens, + P: ToTokens, + { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + Pair::Punctuated(a, b) => { + a.to_tokens(tokens); + b.to_tokens(tokens); + } + Pair::End(a) => a.to_tokens(tokens), + } + } + } +} diff --git a/rust/syn/reserved.rs b/rust/syn/reserved.rs new file mode 100644 index 00000000000000..f6bb508a6907e8 --- /dev/null +++ b/rust/syn/reserved.rs @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Type for a syntax tree node that is reserved for future use. +// +// For example ExprReference contains a field `raw` of type Reserved. If `&raw +// place` syntax becomes a thing as per https://github.com/rust-lang/rfcs/pull/2582, +// we can backward compatibly change `raw`'s type to Option without +// the possibility of breaking any code. + +use proc_macro2::Span; +use std::marker::PhantomData; + +#[cfg(feature = "extra-traits")] +use std::fmt::{self, Debug}; + +ast_struct! { + pub struct Reserved { + _private: PhantomData, + } +} + +impl Default for Reserved { + fn default() -> Self { + Reserved { + _private: PhantomData, + } + } +} + +#[cfg(feature = "clone-impls")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] +impl Clone for Reserved { + fn clone(&self) -> Self { + Reserved { + _private: self._private, + } + } +} + +#[cfg(feature = "extra-traits")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] +impl Debug for Reserved { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.debug_struct("Reserved").finish() + } +} diff --git a/rust/syn/sealed.rs b/rust/syn/sealed.rs new file mode 100644 index 00000000000000..02abcb1cd90dfe --- /dev/null +++ b/rust/syn/sealed.rs @@ -0,0 +1,6 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +#[cfg(feature = "parsing")] +pub mod lookahead { + pub trait Sealed: Copy {} +} diff --git a/rust/syn/span.rs b/rust/syn/span.rs new file mode 100644 index 00000000000000..0574c49d783519 --- /dev/null +++ b/rust/syn/span.rs @@ -0,0 +1,69 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::Span; + +pub trait IntoSpans { + fn into_spans(self) -> S; +} + +impl IntoSpans<[Span; 1]> for Span { + fn into_spans(self) -> [Span; 1] { + [self] + } +} + +impl IntoSpans<[Span; 2]> for Span { + fn into_spans(self) -> [Span; 2] { + [self, self] + } +} + +impl IntoSpans<[Span; 3]> for Span { + fn into_spans(self) -> [Span; 3] { + [self, self, self] + } +} + +impl IntoSpans<[Span; 1]> for [Span; 1] { + fn into_spans(self) -> [Span; 1] { + self + } +} + +impl IntoSpans<[Span; 2]> for [Span; 2] { + fn into_spans(self) -> [Span; 2] { + self + } +} + +impl IntoSpans<[Span; 3]> for [Span; 3] { + fn into_spans(self) -> [Span; 3] { + self + } +} + +#[cfg(feature = "parsing")] +pub trait FromSpans: Sized { + fn from_spans(spans: &[Span]) -> Self; +} + +#[cfg(feature = "parsing")] +impl FromSpans for [Span; 1] { + fn from_spans(spans: &[Span]) -> Self { + [spans[0]] + } +} + +#[cfg(feature = "parsing")] +impl FromSpans for [Span; 2] { + fn from_spans(spans: &[Span]) -> Self { + [spans[0], spans[1]] + } +} + +#[cfg(feature = "parsing")] +impl FromSpans for [Span; 3] { + fn from_spans(spans: &[Span]) -> Self { + [spans[0], spans[1], spans[2]] + } +} diff --git a/rust/syn/spanned.rs b/rust/syn/spanned.rs new file mode 100644 index 00000000000000..ee65ea85841a05 --- /dev/null +++ b/rust/syn/spanned.rs @@ -0,0 +1,116 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! A trait that can provide the `Span` of the complete contents of a syntax +//! tree node. +//! +//! *This module is available only if Syn is built with both the `"parsing"` and +//! `"printing"` features.* +//! +//!
+//! +//! # Example +//! +//! Suppose in a procedural macro we have a [`Type`] that we want to assert +//! implements the [`Sync`] trait. Maybe this is the type of one of the fields +//! of a struct for which we are deriving a trait implementation, and we need to +//! be able to pass a reference to one of those fields across threads. +//! +//! [`Type`]: crate::Type +//! [`Sync`]: std::marker::Sync +//! +//! If the field type does *not* implement `Sync` as required, we want the +//! compiler to report an error pointing out exactly which type it was. +//! +//! The following macro code takes a variable `ty` of type `Type` and produces a +//! static assertion that `Sync` is implemented for that type. +//! +//! ``` +//! # extern crate proc_macro; +//! # +//! use proc_macro::TokenStream; +//! use proc_macro2::Span; +//! use quote::quote_spanned; +//! use syn::Type; +//! use syn::spanned::Spanned; +//! +//! # const IGNORE_TOKENS: &str = stringify! { +//! #[proc_macro_derive(MyMacro)] +//! # }; +//! pub fn my_macro(input: TokenStream) -> TokenStream { +//! # let ty = get_a_type(); +//! /* ... */ +//! +//! let assert_sync = quote_spanned! {ty.span()=> +//! struct _AssertSync where #ty: Sync; +//! }; +//! +//! /* ... */ +//! # input +//! } +//! # +//! # fn get_a_type() -> Type { +//! # unimplemented!() +//! # } +//! ``` +//! +//! By inserting this `assert_sync` fragment into the output code generated by +//! our macro, the user's code will fail to compile if `ty` does not implement +//! `Sync`. The errors they would see look like the following. +//! +//! ```text +//! error[E0277]: the trait bound `*const i32: std::marker::Sync` is not satisfied +//! --> src/main.rs:10:21 +//! | +//! 10 | bad_field: *const i32, +//! | ^^^^^^^^^^ `*const i32` cannot be shared between threads safely +//! ``` +//! +//! In this technique, using the `Type`'s span for the error message makes the +//! error appear in the correct place underlining the right type. +//! +//!
+//! +//! # Limitations +//! +//! The underlying [`proc_macro::Span::join`] method is nightly-only. When +//! called from within a procedural macro in a nightly compiler, `Spanned` will +//! use `join` to produce the intended span. When not using a nightly compiler, +//! only the span of the *first token* of the syntax tree node is returned. +//! +//! In the common case of wanting to use the joined span as the span of a +//! `syn::Error`, consider instead using [`syn::Error::new_spanned`] which is +//! able to span the error correctly under the complete syntax tree node without +//! needing the unstable `join`. +//! +//! [`syn::Error::new_spanned`]: crate::Error::new_spanned + +use proc_macro2::Span; +use quote::spanned::Spanned as ToTokens; + +/// A trait that can provide the `Span` of the complete contents of a syntax +/// tree node. +/// +/// This trait is automatically implemented for all types that implement +/// [`ToTokens`] from the `quote` crate, as well as for `Span` itself. +/// +/// [`ToTokens`]: quote::ToTokens +/// +/// See the [module documentation] for an example. +/// +/// [module documentation]: self +/// +/// *This trait is available only if Syn is built with both the `"parsing"` and +/// `"printing"` features.* +pub trait Spanned { + /// Returns a `Span` covering the complete contents of this syntax tree + /// node, or [`Span::call_site()`] if this node is empty. + /// + /// [`Span::call_site()`]: proc_macro2::Span::call_site + fn span(&self) -> Span; +} + +impl Spanned for T { + fn span(&self) -> Span { + self.__span() + } +} diff --git a/rust/syn/stmt.rs b/rust/syn/stmt.rs new file mode 100644 index 00000000000000..41b9378537ba7b --- /dev/null +++ b/rust/syn/stmt.rs @@ -0,0 +1,351 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; + +ast_struct! { + /// A braced block containing Rust statements. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct Block { + pub brace_token: token::Brace, + /// Statements in a block + pub stmts: Vec, + } +} + +ast_enum! { + /// A statement, usually ending in a semicolon. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub enum Stmt { + /// A local (let) binding. + Local(Local), + + /// An item definition. + Item(Item), + + /// Expr without trailing semicolon. + Expr(Expr), + + /// Expression with trailing semicolon. + Semi(Expr, Token![;]), + } +} + +ast_struct! { + /// A local `let` binding: `let x: u64 = s.parse()?`. + /// + /// *This type is available only if Syn is built with the `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(feature = "full")))] + pub struct Local { + pub attrs: Vec, + pub let_token: Token![let], + pub pat: Pat, + pub init: Option<(Token![=], Box)>, + pub semi_token: Token![;], + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::parse::discouraged::Speculative; + use crate::parse::{Parse, ParseBuffer, ParseStream, Result}; + use proc_macro2::TokenStream; + + impl Block { + /// Parse the body of a block as zero or more statements, possibly + /// including one trailing expression. + /// + /// *This function is available only if Syn is built with the `"parsing"` + /// feature.* + /// + /// # Example + /// + /// ``` + /// use syn::{braced, token, Attribute, Block, Ident, Result, Stmt, Token}; + /// use syn::parse::{Parse, ParseStream}; + /// + /// // Parse a function with no generics or parameter list. + /// // + /// // fn playground { + /// // let mut x = 1; + /// // x += 1; + /// // println!("{}", x); + /// // } + /// struct MiniFunction { + /// attrs: Vec, + /// fn_token: Token![fn], + /// name: Ident, + /// brace_token: token::Brace, + /// stmts: Vec, + /// } + /// + /// impl Parse for MiniFunction { + /// fn parse(input: ParseStream) -> Result { + /// let outer_attrs = input.call(Attribute::parse_outer)?; + /// let fn_token: Token![fn] = input.parse()?; + /// let name: Ident = input.parse()?; + /// + /// let content; + /// let brace_token = braced!(content in input); + /// let inner_attrs = content.call(Attribute::parse_inner)?; + /// let stmts = content.call(Block::parse_within)?; + /// + /// Ok(MiniFunction { + /// attrs: { + /// let mut attrs = outer_attrs; + /// attrs.extend(inner_attrs); + /// attrs + /// }, + /// fn_token, + /// name, + /// brace_token, + /// stmts, + /// }) + /// } + /// } + /// ``` + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn parse_within(input: ParseStream) -> Result> { + let mut stmts = Vec::new(); + loop { + while let Some(semi) = input.parse::>()? { + stmts.push(Stmt::Semi(Expr::Verbatim(TokenStream::new()), semi)); + } + if input.is_empty() { + break; + } + let s = parse_stmt(input, true)?; + let requires_semicolon = if let Stmt::Expr(s) = &s { + expr::requires_terminator(s) + } else { + false + }; + stmts.push(s); + if input.is_empty() { + break; + } else if requires_semicolon { + return Err(input.error("unexpected token")); + } + } + Ok(stmts) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Block { + fn parse(input: ParseStream) -> Result { + let content; + Ok(Block { + brace_token: braced!(content in input), + stmts: content.call(Block::parse_within)?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Stmt { + fn parse(input: ParseStream) -> Result { + parse_stmt(input, false) + } + } + + fn parse_stmt(input: ParseStream, allow_nosemi: bool) -> Result { + let begin = input.fork(); + let mut attrs = input.call(Attribute::parse_outer)?; + + // brace-style macros; paren and bracket macros get parsed as + // expression statements. + let ahead = input.fork(); + if let Ok(path) = ahead.call(Path::parse_mod_style) { + if ahead.peek(Token![!]) + && (ahead.peek2(token::Brace) + && !(ahead.peek3(Token![.]) || ahead.peek3(Token![?])) + || ahead.peek2(Ident)) + { + input.advance_to(&ahead); + return stmt_mac(input, attrs, path); + } + } + + if input.peek(Token![let]) { + stmt_local(input, attrs, begin) + } else if input.peek(Token![pub]) + || input.peek(Token![crate]) && !input.peek2(Token![::]) + || input.peek(Token![extern]) + || input.peek(Token![use]) + || input.peek(Token![static]) + && (input.peek2(Token![mut]) + || input.peek2(Ident) + && !(input.peek2(Token![async]) + && (input.peek3(Token![move]) || input.peek3(Token![|])))) + || input.peek(Token![const]) && !input.peek2(token::Brace) + || input.peek(Token![unsafe]) && !input.peek2(token::Brace) + || input.peek(Token![async]) + && (input.peek2(Token![unsafe]) + || input.peek2(Token![extern]) + || input.peek2(Token![fn])) + || input.peek(Token![fn]) + || input.peek(Token![mod]) + || input.peek(Token![type]) + || input.peek(Token![struct]) + || input.peek(Token![enum]) + || input.peek(Token![union]) && input.peek2(Ident) + || input.peek(Token![auto]) && input.peek2(Token![trait]) + || input.peek(Token![trait]) + || input.peek(Token![default]) + && (input.peek2(Token![unsafe]) || input.peek2(Token![impl])) + || input.peek(Token![impl]) + || input.peek(Token![macro]) + { + let mut item: Item = input.parse()?; + attrs.extend(item.replace_attrs(Vec::new())); + item.replace_attrs(attrs); + Ok(Stmt::Item(item)) + } else { + stmt_expr(input, allow_nosemi, attrs) + } + } + + fn stmt_mac(input: ParseStream, attrs: Vec, path: Path) -> Result { + let bang_token: Token![!] = input.parse()?; + let ident: Option = input.parse()?; + let (delimiter, tokens) = mac::parse_delimiter(input)?; + let semi_token: Option = input.parse()?; + + Ok(Stmt::Item(Item::Macro(ItemMacro { + attrs, + ident, + mac: Macro { + path, + bang_token, + delimiter, + tokens, + }, + semi_token, + }))) + } + + fn stmt_local(input: ParseStream, attrs: Vec, begin: ParseBuffer) -> Result { + let let_token: Token![let] = input.parse()?; + + let mut pat: Pat = pat::parsing::multi_pat_with_leading_vert(input)?; + if input.peek(Token![:]) { + let colon_token: Token![:] = input.parse()?; + let ty: Type = input.parse()?; + pat = Pat::Type(PatType { + attrs: Vec::new(), + pat: Box::new(pat), + colon_token, + ty: Box::new(ty), + }); + } + + let init = if input.peek(Token![=]) { + let eq_token: Token![=] = input.parse()?; + let init: Expr = input.parse()?; + + if input.peek(Token![else]) { + input.parse::()?; + let content; + braced!(content in input); + content.call(Block::parse_within)?; + let verbatim = Expr::Verbatim(verbatim::between(begin, input)); + let semi_token: Token![;] = input.parse()?; + return Ok(Stmt::Semi(verbatim, semi_token)); + } + + Some((eq_token, Box::new(init))) + } else { + None + }; + + let semi_token: Token![;] = input.parse()?; + + Ok(Stmt::Local(Local { + attrs, + let_token, + pat, + init, + semi_token, + })) + } + + fn stmt_expr( + input: ParseStream, + allow_nosemi: bool, + mut attrs: Vec, + ) -> Result { + let mut e = expr::parsing::expr_early(input)?; + + let mut attr_target = &mut e; + loop { + attr_target = match attr_target { + Expr::Assign(e) => &mut e.left, + Expr::AssignOp(e) => &mut e.left, + Expr::Binary(e) => &mut e.left, + _ => break, + }; + } + attrs.extend(attr_target.replace_attrs(Vec::new())); + attr_target.replace_attrs(attrs); + + if input.peek(Token![;]) { + return Ok(Stmt::Semi(e, input.parse()?)); + } + + if allow_nosemi || !expr::requires_terminator(&e) { + Ok(Stmt::Expr(e)) + } else { + Err(input.error("expected semicolon")) + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Block { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.brace_token.surround(tokens, |tokens| { + tokens.append_all(&self.stmts); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Stmt { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + Stmt::Local(local) => local.to_tokens(tokens), + Stmt::Item(item) => item.to_tokens(tokens), + Stmt::Expr(expr) => expr.to_tokens(tokens), + Stmt::Semi(expr, semi) => { + expr.to_tokens(tokens); + semi.to_tokens(tokens); + } + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Local { + fn to_tokens(&self, tokens: &mut TokenStream) { + expr::printing::outer_attrs_to_tokens(&self.attrs, tokens); + self.let_token.to_tokens(tokens); + self.pat.to_tokens(tokens); + if let Some((eq_token, init)) = &self.init { + eq_token.to_tokens(tokens); + init.to_tokens(tokens); + } + self.semi_token.to_tokens(tokens); + } + } +} diff --git a/rust/syn/thread.rs b/rust/syn/thread.rs new file mode 100644 index 00000000000000..d535c913a9aa29 --- /dev/null +++ b/rust/syn/thread.rs @@ -0,0 +1,43 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use std::fmt::{self, Debug}; +use std::thread::{self, ThreadId}; + +/// ThreadBound is a Sync-maker and Send-maker that allows accessing a value +/// of type T only from the original thread on which the ThreadBound was +/// constructed. +pub struct ThreadBound { + value: T, + thread_id: ThreadId, +} + +unsafe impl Sync for ThreadBound {} + +// Send bound requires Copy, as otherwise Drop could run in the wrong place. +unsafe impl Send for ThreadBound {} + +impl ThreadBound { + pub fn new(value: T) -> Self { + ThreadBound { + value, + thread_id: thread::current().id(), + } + } + + pub fn get(&self) -> Option<&T> { + if thread::current().id() == self.thread_id { + Some(&self.value) + } else { + None + } + } +} + +impl Debug for ThreadBound { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self.get() { + Some(value) => Debug::fmt(value, formatter), + None => formatter.write_str("unknown"), + } + } +} diff --git a/rust/syn/token.rs b/rust/syn/token.rs new file mode 100644 index 00000000000000..b9984c5782fb1a --- /dev/null +++ b/rust/syn/token.rs @@ -0,0 +1,1015 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +//! Tokens representing Rust punctuation, keywords, and delimiters. +//! +//! The type names in this module can be difficult to keep straight, so we +//! prefer to use the [`Token!`] macro instead. This is a type-macro that +//! expands to the token type of the given token. +//! +//! [`Token!`]: crate::Token +//! +//! # Example +//! +//! The [`ItemStatic`] syntax tree node is defined like this. +//! +//! [`ItemStatic`]: crate::ItemStatic +//! +//! ``` +//! # use syn::{Attribute, Expr, Ident, Token, Type, Visibility}; +//! # +//! pub struct ItemStatic { +//! pub attrs: Vec, +//! pub vis: Visibility, +//! pub static_token: Token![static], +//! pub mutability: Option, +//! pub ident: Ident, +//! pub colon_token: Token![:], +//! pub ty: Box, +//! pub eq_token: Token![=], +//! pub expr: Box, +//! pub semi_token: Token![;], +//! } +//! ``` +//! +//! # Parsing +//! +//! Keywords and punctuation can be parsed through the [`ParseStream::parse`] +//! method. Delimiter tokens are parsed using the [`parenthesized!`], +//! [`bracketed!`] and [`braced!`] macros. +//! +//! [`ParseStream::parse`]: crate::parse::ParseBuffer::parse() +//! [`parenthesized!`]: crate::parenthesized! +//! [`bracketed!`]: crate::bracketed! +//! [`braced!`]: crate::braced! +//! +//! ``` +//! use syn::{Attribute, Result}; +//! use syn::parse::{Parse, ParseStream}; +//! # +//! # enum ItemStatic {} +//! +//! // Parse the ItemStatic struct shown above. +//! impl Parse for ItemStatic { +//! fn parse(input: ParseStream) -> Result { +//! # use syn::ItemStatic; +//! # fn parse(input: ParseStream) -> Result { +//! Ok(ItemStatic { +//! attrs: input.call(Attribute::parse_outer)?, +//! vis: input.parse()?, +//! static_token: input.parse()?, +//! mutability: input.parse()?, +//! ident: input.parse()?, +//! colon_token: input.parse()?, +//! ty: input.parse()?, +//! eq_token: input.parse()?, +//! expr: input.parse()?, +//! semi_token: input.parse()?, +//! }) +//! # } +//! # unimplemented!() +//! } +//! } +//! ``` +//! +//! # Other operations +//! +//! Every keyword and punctuation token supports the following operations. +//! +//! - [Peeking] — `input.peek(Token![...])` +//! +//! - [Parsing] — `input.parse::()?` +//! +//! - [Printing] — `quote!( ... #the_token ... )` +//! +//! - Construction from a [`Span`] — `let the_token = Token![...](sp)` +//! +//! - Field access to its span — `let sp = the_token.span` +//! +//! [Peeking]: crate::parse::ParseBuffer::peek() +//! [Parsing]: crate::parse::ParseBuffer::parse() +//! [Printing]: https://docs.rs/quote/1.0/quote/trait.ToTokens.html +//! [`Span`]: https://docs.rs/proc-macro2/1.0/proc_macro2/struct.Span.html + +use self::private::WithSpan; +#[cfg(feature = "parsing")] +use crate::buffer::Cursor; +#[cfg(feature = "parsing")] +use crate::error::Result; +#[cfg(feature = "parsing")] +use crate::lifetime::Lifetime; +#[cfg(feature = "parsing")] +use crate::lit::{Lit, LitBool, LitByte, LitByteStr, LitChar, LitFloat, LitInt, LitStr}; +#[cfg(feature = "parsing")] +use crate::lookahead; +#[cfg(feature = "parsing")] +use crate::parse::{Parse, ParseStream}; +use crate::span::IntoSpans; +#[cfg(any(feature = "parsing", feature = "printing"))] +use proc_macro2::Ident; +use proc_macro2::Span; +#[cfg(feature = "printing")] +use proc_macro2::TokenStream; +#[cfg(feature = "parsing")] +use proc_macro2::{Delimiter, Literal, Punct, TokenTree}; +#[cfg(feature = "printing")] +use quote::{ToTokens, TokenStreamExt}; +#[cfg(feature = "extra-traits")] +use std::cmp; +#[cfg(feature = "extra-traits")] +use std::fmt::{self, Debug}; +#[cfg(feature = "extra-traits")] +use std::hash::{Hash, Hasher}; +use std::ops::{Deref, DerefMut}; + +/// Marker trait for types that represent single tokens. +/// +/// This trait is sealed and cannot be implemented for types outside of Syn. +#[cfg(feature = "parsing")] +pub trait Token: private::Sealed { + // Not public API. + #[doc(hidden)] + fn peek(cursor: Cursor) -> bool; + + // Not public API. + #[doc(hidden)] + fn display() -> &'static str; +} + +mod private { + use proc_macro2::Span; + + #[cfg(feature = "parsing")] + pub trait Sealed {} + + /// Support writing `token.span` rather than `token.spans[0]` on tokens that + /// hold a single span. + #[repr(C)] + pub struct WithSpan { + pub span: Span, + } +} + +#[cfg(feature = "parsing")] +impl private::Sealed for Ident {} + +#[cfg(feature = "parsing")] +fn peek_impl(cursor: Cursor, peek: fn(ParseStream) -> bool) -> bool { + use crate::parse::Unexpected; + use std::cell::Cell; + use std::rc::Rc; + + let scope = Span::call_site(); + let unexpected = Rc::new(Cell::new(Unexpected::None)); + let buffer = crate::parse::new_parse_buffer(scope, cursor, unexpected); + peek(&buffer) +} + +macro_rules! impl_token { + ($display:tt $name:ty) => { + #[cfg(feature = "parsing")] + impl Token for $name { + fn peek(cursor: Cursor) -> bool { + fn peek(input: ParseStream) -> bool { + <$name as Parse>::parse(input).is_ok() + } + peek_impl(cursor, peek) + } + + fn display() -> &'static str { + $display + } + } + + #[cfg(feature = "parsing")] + impl private::Sealed for $name {} + }; +} + +impl_token!("lifetime" Lifetime); +impl_token!("literal" Lit); +impl_token!("string literal" LitStr); +impl_token!("byte string literal" LitByteStr); +impl_token!("byte literal" LitByte); +impl_token!("character literal" LitChar); +impl_token!("integer literal" LitInt); +impl_token!("floating point literal" LitFloat); +impl_token!("boolean literal" LitBool); +impl_token!("group token" proc_macro2::Group); + +macro_rules! impl_low_level_token { + ($display:tt $ty:ident $get:ident) => { + #[cfg(feature = "parsing")] + impl Token for $ty { + fn peek(cursor: Cursor) -> bool { + cursor.$get().is_some() + } + + fn display() -> &'static str { + $display + } + } + + #[cfg(feature = "parsing")] + impl private::Sealed for $ty {} + }; +} + +impl_low_level_token!("punctuation token" Punct punct); +impl_low_level_token!("literal" Literal literal); +impl_low_level_token!("token" TokenTree token_tree); + +// Not public API. +#[doc(hidden)] +#[cfg(feature = "parsing")] +pub trait CustomToken { + fn peek(cursor: Cursor) -> bool; + fn display() -> &'static str; +} + +#[cfg(feature = "parsing")] +impl private::Sealed for T {} + +#[cfg(feature = "parsing")] +impl Token for T { + fn peek(cursor: Cursor) -> bool { + ::peek(cursor) + } + + fn display() -> &'static str { + ::display() + } +} + +macro_rules! define_keywords { + ($($token:tt pub struct $name:ident #[$doc:meta])*) => { + $( + #[$doc] + /// + /// Don't try to remember the name of this type — use the + /// [`Token!`] macro instead. + /// + /// [`Token!`]: crate::token + pub struct $name { + pub span: Span, + } + + #[doc(hidden)] + #[allow(non_snake_case)] + pub fn $name>(span: S) -> $name { + $name { + span: span.into_spans()[0], + } + } + + impl std::default::Default for $name { + fn default() -> Self { + $name { + span: Span::call_site(), + } + } + } + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Copy for $name {} + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Clone for $name { + fn clone(&self) -> Self { + *self + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for $name { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str(stringify!($name)) + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl cmp::Eq for $name {} + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl PartialEq for $name { + fn eq(&self, _other: &$name) -> bool { + true + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Hash for $name { + fn hash(&self, _state: &mut H) {} + } + + #[cfg(feature = "printing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for $name { + fn to_tokens(&self, tokens: &mut TokenStream) { + printing::keyword($token, self.span, tokens); + } + } + + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for $name { + fn parse(input: ParseStream) -> Result { + Ok($name { + span: parsing::keyword(input, $token)?, + }) + } + } + + #[cfg(feature = "parsing")] + impl Token for $name { + fn peek(cursor: Cursor) -> bool { + parsing::peek_keyword(cursor, $token) + } + + fn display() -> &'static str { + concat!("`", $token, "`") + } + } + + #[cfg(feature = "parsing")] + impl private::Sealed for $name {} + )* + }; +} + +macro_rules! impl_deref_if_len_is_1 { + ($name:ident/1) => { + impl Deref for $name { + type Target = WithSpan; + + fn deref(&self) -> &Self::Target { + unsafe { &*(self as *const Self as *const WithSpan) } + } + } + + impl DerefMut for $name { + fn deref_mut(&mut self) -> &mut Self::Target { + unsafe { &mut *(self as *mut Self as *mut WithSpan) } + } + } + }; + + ($name:ident/$len:tt) => {}; +} + +macro_rules! define_punctuation_structs { + ($($token:tt pub struct $name:ident/$len:tt #[$doc:meta])*) => { + $( + #[repr(C)] + #[$doc] + /// + /// Don't try to remember the name of this type — use the + /// [`Token!`] macro instead. + /// + /// [`Token!`]: crate::token + pub struct $name { + pub spans: [Span; $len], + } + + #[doc(hidden)] + #[allow(non_snake_case)] + pub fn $name>(spans: S) -> $name { + $name { + spans: spans.into_spans(), + } + } + + impl std::default::Default for $name { + fn default() -> Self { + $name { + spans: [Span::call_site(); $len], + } + } + } + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Copy for $name {} + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Clone for $name { + fn clone(&self) -> Self { + *self + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for $name { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str(stringify!($name)) + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl cmp::Eq for $name {} + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl PartialEq for $name { + fn eq(&self, _other: &$name) -> bool { + true + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Hash for $name { + fn hash(&self, _state: &mut H) {} + } + + impl_deref_if_len_is_1!($name/$len); + )* + }; +} + +macro_rules! define_punctuation { + ($($token:tt pub struct $name:ident/$len:tt #[$doc:meta])*) => { + $( + define_punctuation_structs! { + $token pub struct $name/$len #[$doc] + } + + #[cfg(feature = "printing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for $name { + fn to_tokens(&self, tokens: &mut TokenStream) { + printing::punct($token, &self.spans, tokens); + } + } + + #[cfg(feature = "parsing")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for $name { + fn parse(input: ParseStream) -> Result { + Ok($name { + spans: parsing::punct(input, $token)?, + }) + } + } + + #[cfg(feature = "parsing")] + impl Token for $name { + fn peek(cursor: Cursor) -> bool { + parsing::peek_punct(cursor, $token) + } + + fn display() -> &'static str { + concat!("`", $token, "`") + } + } + + #[cfg(feature = "parsing")] + impl private::Sealed for $name {} + )* + }; +} + +macro_rules! define_delimiters { + ($($token:tt pub struct $name:ident #[$doc:meta])*) => { + $( + #[$doc] + pub struct $name { + pub span: Span, + } + + #[doc(hidden)] + #[allow(non_snake_case)] + pub fn $name>(span: S) -> $name { + $name { + span: span.into_spans()[0], + } + } + + impl std::default::Default for $name { + fn default() -> Self { + $name { + span: Span::call_site(), + } + } + } + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Copy for $name {} + + #[cfg(feature = "clone-impls")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "clone-impls")))] + impl Clone for $name { + fn clone(&self) -> Self { + *self + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Debug for $name { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str(stringify!($name)) + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl cmp::Eq for $name {} + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl PartialEq for $name { + fn eq(&self, _other: &$name) -> bool { + true + } + } + + #[cfg(feature = "extra-traits")] + #[cfg_attr(doc_cfg, doc(cfg(feature = "extra-traits")))] + impl Hash for $name { + fn hash(&self, _state: &mut H) {} + } + + impl $name { + #[cfg(feature = "printing")] + pub fn surround(&self, tokens: &mut TokenStream, f: F) + where + F: FnOnce(&mut TokenStream), + { + printing::delim($token, self.span, tokens, f); + } + } + + #[cfg(feature = "parsing")] + impl private::Sealed for $name {} + )* + }; +} + +define_punctuation_structs! { + "_" pub struct Underscore/1 /// `_` +} + +#[cfg(feature = "printing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] +impl ToTokens for Underscore { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append(Ident::new("_", self.span)); + } +} + +#[cfg(feature = "parsing")] +#[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] +impl Parse for Underscore { + fn parse(input: ParseStream) -> Result { + input.step(|cursor| { + if let Some((ident, rest)) = cursor.ident() { + if ident == "_" { + return Ok((Underscore(ident.span()), rest)); + } + } + if let Some((punct, rest)) = cursor.punct() { + if punct.as_char() == '_' { + return Ok((Underscore(punct.span()), rest)); + } + } + Err(cursor.error("expected `_`")) + }) + } +} + +#[cfg(feature = "parsing")] +impl Token for Underscore { + fn peek(cursor: Cursor) -> bool { + if let Some((ident, _rest)) = cursor.ident() { + return ident == "_"; + } + if let Some((punct, _rest)) = cursor.punct() { + return punct.as_char() == '_'; + } + false + } + + fn display() -> &'static str { + "`_`" + } +} + +#[cfg(feature = "parsing")] +impl private::Sealed for Underscore {} + +#[cfg(feature = "parsing")] +impl Token for Paren { + fn peek(cursor: Cursor) -> bool { + lookahead::is_delimiter(cursor, Delimiter::Parenthesis) + } + + fn display() -> &'static str { + "parentheses" + } +} + +#[cfg(feature = "parsing")] +impl Token for Brace { + fn peek(cursor: Cursor) -> bool { + lookahead::is_delimiter(cursor, Delimiter::Brace) + } + + fn display() -> &'static str { + "curly braces" + } +} + +#[cfg(feature = "parsing")] +impl Token for Bracket { + fn peek(cursor: Cursor) -> bool { + lookahead::is_delimiter(cursor, Delimiter::Bracket) + } + + fn display() -> &'static str { + "square brackets" + } +} + +#[cfg(feature = "parsing")] +impl Token for Group { + fn peek(cursor: Cursor) -> bool { + lookahead::is_delimiter(cursor, Delimiter::None) + } + + fn display() -> &'static str { + "invisible group" + } +} + +define_keywords! { + "abstract" pub struct Abstract /// `abstract` + "as" pub struct As /// `as` + "async" pub struct Async /// `async` + "auto" pub struct Auto /// `auto` + "await" pub struct Await /// `await` + "become" pub struct Become /// `become` + "box" pub struct Box /// `box` + "break" pub struct Break /// `break` + "const" pub struct Const /// `const` + "continue" pub struct Continue /// `continue` + "crate" pub struct Crate /// `crate` + "default" pub struct Default /// `default` + "do" pub struct Do /// `do` + "dyn" pub struct Dyn /// `dyn` + "else" pub struct Else /// `else` + "enum" pub struct Enum /// `enum` + "extern" pub struct Extern /// `extern` + "final" pub struct Final /// `final` + "fn" pub struct Fn /// `fn` + "for" pub struct For /// `for` + "if" pub struct If /// `if` + "impl" pub struct Impl /// `impl` + "in" pub struct In /// `in` + "let" pub struct Let /// `let` + "loop" pub struct Loop /// `loop` + "macro" pub struct Macro /// `macro` + "match" pub struct Match /// `match` + "mod" pub struct Mod /// `mod` + "move" pub struct Move /// `move` + "mut" pub struct Mut /// `mut` + "override" pub struct Override /// `override` + "priv" pub struct Priv /// `priv` + "pub" pub struct Pub /// `pub` + "ref" pub struct Ref /// `ref` + "return" pub struct Return /// `return` + "Self" pub struct SelfType /// `Self` + "self" pub struct SelfValue /// `self` + "static" pub struct Static /// `static` + "struct" pub struct Struct /// `struct` + "super" pub struct Super /// `super` + "trait" pub struct Trait /// `trait` + "try" pub struct Try /// `try` + "type" pub struct Type /// `type` + "typeof" pub struct Typeof /// `typeof` + "union" pub struct Union /// `union` + "unsafe" pub struct Unsafe /// `unsafe` + "unsized" pub struct Unsized /// `unsized` + "use" pub struct Use /// `use` + "virtual" pub struct Virtual /// `virtual` + "where" pub struct Where /// `where` + "while" pub struct While /// `while` + "yield" pub struct Yield /// `yield` +} + +define_punctuation! { + "+" pub struct Add/1 /// `+` + "+=" pub struct AddEq/2 /// `+=` + "&" pub struct And/1 /// `&` + "&&" pub struct AndAnd/2 /// `&&` + "&=" pub struct AndEq/2 /// `&=` + "@" pub struct At/1 /// `@` + "!" pub struct Bang/1 /// `!` + "^" pub struct Caret/1 /// `^` + "^=" pub struct CaretEq/2 /// `^=` + ":" pub struct Colon/1 /// `:` + "::" pub struct Colon2/2 /// `::` + "," pub struct Comma/1 /// `,` + "/" pub struct Div/1 /// `/` + "/=" pub struct DivEq/2 /// `/=` + "$" pub struct Dollar/1 /// `$` + "." pub struct Dot/1 /// `.` + ".." pub struct Dot2/2 /// `..` + "..." pub struct Dot3/3 /// `...` + "..=" pub struct DotDotEq/3 /// `..=` + "=" pub struct Eq/1 /// `=` + "==" pub struct EqEq/2 /// `==` + ">=" pub struct Ge/2 /// `>=` + ">" pub struct Gt/1 /// `>` + "<=" pub struct Le/2 /// `<=` + "<" pub struct Lt/1 /// `<` + "*=" pub struct MulEq/2 /// `*=` + "!=" pub struct Ne/2 /// `!=` + "|" pub struct Or/1 /// `|` + "|=" pub struct OrEq/2 /// `|=` + "||" pub struct OrOr/2 /// `||` + "#" pub struct Pound/1 /// `#` + "?" pub struct Question/1 /// `?` + "->" pub struct RArrow/2 /// `->` + "<-" pub struct LArrow/2 /// `<-` + "%" pub struct Rem/1 /// `%` + "%=" pub struct RemEq/2 /// `%=` + "=>" pub struct FatArrow/2 /// `=>` + ";" pub struct Semi/1 /// `;` + "<<" pub struct Shl/2 /// `<<` + "<<=" pub struct ShlEq/3 /// `<<=` + ">>" pub struct Shr/2 /// `>>` + ">>=" pub struct ShrEq/3 /// `>>=` + "*" pub struct Star/1 /// `*` + "-" pub struct Sub/1 /// `-` + "-=" pub struct SubEq/2 /// `-=` + "~" pub struct Tilde/1 /// `~` +} + +define_delimiters! { + "{" pub struct Brace /// `{...}` + "[" pub struct Bracket /// `[...]` + "(" pub struct Paren /// `(...)` + " " pub struct Group /// None-delimited group +} + +macro_rules! export_token_macro { + ($($await_rule:tt)*) => { + /// A type-macro that expands to the name of the Rust type representation of a + /// given token. + /// + /// See the [token module] documentation for details and examples. + /// + /// [token module]: crate::token + // Unfortunate duplication due to a rustdoc bug. + // https://github.com/rust-lang/rust/issues/45939 + #[macro_export] + macro_rules! Token { + [abstract] => { $crate::token::Abstract }; + [as] => { $crate::token::As }; + [async] => { $crate::token::Async }; + [auto] => { $crate::token::Auto }; + $($await_rule => { $crate::token::Await };)* + [become] => { $crate::token::Become }; + [box] => { $crate::token::Box }; + [break] => { $crate::token::Break }; + [const] => { $crate::token::Const }; + [continue] => { $crate::token::Continue }; + [crate] => { $crate::token::Crate }; + [default] => { $crate::token::Default }; + [do] => { $crate::token::Do }; + [dyn] => { $crate::token::Dyn }; + [else] => { $crate::token::Else }; + [enum] => { $crate::token::Enum }; + [extern] => { $crate::token::Extern }; + [final] => { $crate::token::Final }; + [fn] => { $crate::token::Fn }; + [for] => { $crate::token::For }; + [if] => { $crate::token::If }; + [impl] => { $crate::token::Impl }; + [in] => { $crate::token::In }; + [let] => { $crate::token::Let }; + [loop] => { $crate::token::Loop }; + [macro] => { $crate::token::Macro }; + [match] => { $crate::token::Match }; + [mod] => { $crate::token::Mod }; + [move] => { $crate::token::Move }; + [mut] => { $crate::token::Mut }; + [override] => { $crate::token::Override }; + [priv] => { $crate::token::Priv }; + [pub] => { $crate::token::Pub }; + [ref] => { $crate::token::Ref }; + [return] => { $crate::token::Return }; + [Self] => { $crate::token::SelfType }; + [self] => { $crate::token::SelfValue }; + [static] => { $crate::token::Static }; + [struct] => { $crate::token::Struct }; + [super] => { $crate::token::Super }; + [trait] => { $crate::token::Trait }; + [try] => { $crate::token::Try }; + [type] => { $crate::token::Type }; + [typeof] => { $crate::token::Typeof }; + [union] => { $crate::token::Union }; + [unsafe] => { $crate::token::Unsafe }; + [unsized] => { $crate::token::Unsized }; + [use] => { $crate::token::Use }; + [virtual] => { $crate::token::Virtual }; + [where] => { $crate::token::Where }; + [while] => { $crate::token::While }; + [yield] => { $crate::token::Yield }; + [+] => { $crate::token::Add }; + [+=] => { $crate::token::AddEq }; + [&] => { $crate::token::And }; + [&&] => { $crate::token::AndAnd }; + [&=] => { $crate::token::AndEq }; + [@] => { $crate::token::At }; + [!] => { $crate::token::Bang }; + [^] => { $crate::token::Caret }; + [^=] => { $crate::token::CaretEq }; + [:] => { $crate::token::Colon }; + [::] => { $crate::token::Colon2 }; + [,] => { $crate::token::Comma }; + [/] => { $crate::token::Div }; + [/=] => { $crate::token::DivEq }; + [$] => { $crate::token::Dollar }; + [.] => { $crate::token::Dot }; + [..] => { $crate::token::Dot2 }; + [...] => { $crate::token::Dot3 }; + [..=] => { $crate::token::DotDotEq }; + [=] => { $crate::token::Eq }; + [==] => { $crate::token::EqEq }; + [>=] => { $crate::token::Ge }; + [>] => { $crate::token::Gt }; + [<=] => { $crate::token::Le }; + [<] => { $crate::token::Lt }; + [*=] => { $crate::token::MulEq }; + [!=] => { $crate::token::Ne }; + [|] => { $crate::token::Or }; + [|=] => { $crate::token::OrEq }; + [||] => { $crate::token::OrOr }; + [#] => { $crate::token::Pound }; + [?] => { $crate::token::Question }; + [->] => { $crate::token::RArrow }; + [<-] => { $crate::token::LArrow }; + [%] => { $crate::token::Rem }; + [%=] => { $crate::token::RemEq }; + [=>] => { $crate::token::FatArrow }; + [;] => { $crate::token::Semi }; + [<<] => { $crate::token::Shl }; + [<<=] => { $crate::token::ShlEq }; + [>>] => { $crate::token::Shr }; + [>>=] => { $crate::token::ShrEq }; + [*] => { $crate::token::Star }; + [-] => { $crate::token::Sub }; + [-=] => { $crate::token::SubEq }; + [~] => { $crate::token::Tilde }; + [_] => { $crate::token::Underscore }; + } + }; +} + +// Old rustc does not permit `await` appearing anywhere in the source file. +// https://github.com/rust-lang/rust/issues/57919 +// We put the Token![await] rule in a place that is not lexed by old rustc. +#[cfg(not(syn_omit_await_from_token_macro))] +include!("await.rs"); // export_token_macro! {[await]} +#[cfg(syn_omit_await_from_token_macro)] +export_token_macro! {} + +// Not public API. +#[doc(hidden)] +#[cfg(feature = "parsing")] +pub mod parsing { + use crate::buffer::Cursor; + use crate::error::{Error, Result}; + use crate::parse::ParseStream; + use crate::span::FromSpans; + use proc_macro2::{Spacing, Span}; + + pub fn keyword(input: ParseStream, token: &str) -> Result { + input.step(|cursor| { + if let Some((ident, rest)) = cursor.ident() { + if ident == token { + return Ok((ident.span(), rest)); + } + } + Err(cursor.error(format!("expected `{}`", token))) + }) + } + + pub fn peek_keyword(cursor: Cursor, token: &str) -> bool { + if let Some((ident, _rest)) = cursor.ident() { + ident == token + } else { + false + } + } + + pub fn punct(input: ParseStream, token: &str) -> Result { + let mut spans = [input.span(); 3]; + punct_helper(input, token, &mut spans)?; + Ok(S::from_spans(&spans)) + } + + fn punct_helper(input: ParseStream, token: &str, spans: &mut [Span; 3]) -> Result<()> { + input.step(|cursor| { + let mut cursor = *cursor; + assert!(token.len() <= spans.len()); + + for (i, ch) in token.chars().enumerate() { + match cursor.punct() { + Some((punct, rest)) => { + spans[i] = punct.span(); + if punct.as_char() != ch { + break; + } else if i == token.len() - 1 { + return Ok(((), rest)); + } else if punct.spacing() != Spacing::Joint { + break; + } + cursor = rest; + } + None => break, + } + } + + Err(Error::new(spans[0], format!("expected `{}`", token))) + }) + } + + pub fn peek_punct(mut cursor: Cursor, token: &str) -> bool { + for (i, ch) in token.chars().enumerate() { + match cursor.punct() { + Some((punct, rest)) => { + if punct.as_char() != ch { + break; + } else if i == token.len() - 1 { + return true; + } else if punct.spacing() != Spacing::Joint { + break; + } + cursor = rest; + } + None => break, + } + } + false + } +} + +// Not public API. +#[doc(hidden)] +#[cfg(feature = "printing")] +pub mod printing { + use proc_macro2::{Delimiter, Group, Ident, Punct, Spacing, Span, TokenStream}; + use quote::TokenStreamExt; + + pub fn punct(s: &str, spans: &[Span], tokens: &mut TokenStream) { + assert_eq!(s.len(), spans.len()); + + let mut chars = s.chars(); + let mut spans = spans.iter(); + let ch = chars.next_back().unwrap(); + let span = spans.next_back().unwrap(); + for (ch, span) in chars.zip(spans) { + let mut op = Punct::new(ch, Spacing::Joint); + op.set_span(*span); + tokens.append(op); + } + + let mut op = Punct::new(ch, Spacing::Alone); + op.set_span(*span); + tokens.append(op); + } + + pub fn keyword(s: &str, span: Span, tokens: &mut TokenStream) { + tokens.append(Ident::new(s, span)); + } + + pub fn delim(s: &str, span: Span, tokens: &mut TokenStream, f: F) + where + F: FnOnce(&mut TokenStream), + { + let delim = match s { + "(" => Delimiter::Parenthesis, + "[" => Delimiter::Bracket, + "{" => Delimiter::Brace, + " " => Delimiter::None, + _ => panic!("unknown delimiter: {}", s), + }; + let mut inner = TokenStream::new(); + f(&mut inner); + let mut g = Group::new(delim, inner); + g.set_span(span); + tokens.append(g); + } +} diff --git a/rust/syn/tt.rs b/rust/syn/tt.rs new file mode 100644 index 00000000000000..b4a9a3876f7d6e --- /dev/null +++ b/rust/syn/tt.rs @@ -0,0 +1,109 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use proc_macro2::{Delimiter, TokenStream, TokenTree}; +use std::hash::{Hash, Hasher}; + +pub struct TokenTreeHelper<'a>(pub &'a TokenTree); + +impl<'a> PartialEq for TokenTreeHelper<'a> { + fn eq(&self, other: &Self) -> bool { + use proc_macro2::Spacing; + + match (self.0, other.0) { + (TokenTree::Group(g1), TokenTree::Group(g2)) => { + match (g1.delimiter(), g2.delimiter()) { + (Delimiter::Parenthesis, Delimiter::Parenthesis) + | (Delimiter::Brace, Delimiter::Brace) + | (Delimiter::Bracket, Delimiter::Bracket) + | (Delimiter::None, Delimiter::None) => {} + _ => return false, + } + + let s1 = g1.stream().into_iter(); + let mut s2 = g2.stream().into_iter(); + + for item1 in s1 { + let item2 = match s2.next() { + Some(item) => item, + None => return false, + }; + if TokenTreeHelper(&item1) != TokenTreeHelper(&item2) { + return false; + } + } + s2.next().is_none() + } + (TokenTree::Punct(o1), TokenTree::Punct(o2)) => { + o1.as_char() == o2.as_char() + && match (o1.spacing(), o2.spacing()) { + (Spacing::Alone, Spacing::Alone) | (Spacing::Joint, Spacing::Joint) => true, + _ => false, + } + } + (TokenTree::Literal(l1), TokenTree::Literal(l2)) => l1.to_string() == l2.to_string(), + (TokenTree::Ident(s1), TokenTree::Ident(s2)) => s1 == s2, + _ => false, + } + } +} + +impl<'a> Hash for TokenTreeHelper<'a> { + fn hash(&self, h: &mut H) { + use proc_macro2::Spacing; + + match self.0 { + TokenTree::Group(g) => { + 0u8.hash(h); + match g.delimiter() { + Delimiter::Parenthesis => 0u8.hash(h), + Delimiter::Brace => 1u8.hash(h), + Delimiter::Bracket => 2u8.hash(h), + Delimiter::None => 3u8.hash(h), + } + + for item in g.stream() { + TokenTreeHelper(&item).hash(h); + } + 0xffu8.hash(h); // terminator w/ a variant we don't normally hash + } + TokenTree::Punct(op) => { + 1u8.hash(h); + op.as_char().hash(h); + match op.spacing() { + Spacing::Alone => 0u8.hash(h), + Spacing::Joint => 1u8.hash(h), + } + } + TokenTree::Literal(lit) => (2u8, lit.to_string()).hash(h), + TokenTree::Ident(word) => (3u8, word).hash(h), + } + } +} + +pub struct TokenStreamHelper<'a>(pub &'a TokenStream); + +impl<'a> PartialEq for TokenStreamHelper<'a> { + fn eq(&self, other: &Self) -> bool { + let left = self.0.clone().into_iter().collect::>(); + let right = other.0.clone().into_iter().collect::>(); + if left.len() != right.len() { + return false; + } + for (a, b) in left.into_iter().zip(right) { + if TokenTreeHelper(&a) != TokenTreeHelper(&b) { + return false; + } + } + true + } +} + +impl<'a> Hash for TokenStreamHelper<'a> { + fn hash(&self, state: &mut H) { + let tts = self.0.clone().into_iter().collect::>(); + tts.len().hash(state); + for tt in tts { + TokenTreeHelper(&tt).hash(state); + } + } +} diff --git a/rust/syn/ty.rs b/rust/syn/ty.rs new file mode 100644 index 00000000000000..ecdf93b952a509 --- /dev/null +++ b/rust/syn/ty.rs @@ -0,0 +1,1288 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use super::*; +use crate::punctuated::Punctuated; +use proc_macro2::TokenStream; + +ast_enum_of_structs! { + /// The possible types that a Rust value could have. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + /// + /// # Syntax tree enum + /// + /// This type is a [syntax tree enum]. + /// + /// [syntax tree enum]: Expr#syntax-tree-enums + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + #[cfg_attr(not(syn_no_non_exhaustive), non_exhaustive)] + pub enum Type { + /// A fixed size array type: `[T; n]`. + Array(TypeArray), + + /// A bare function type: `fn(usize) -> bool`. + BareFn(TypeBareFn), + + /// A type contained within invisible delimiters. + Group(TypeGroup), + + /// An `impl Bound1 + Bound2 + Bound3` type where `Bound` is a trait or + /// a lifetime. + ImplTrait(TypeImplTrait), + + /// Indication that a type should be inferred by the compiler: `_`. + Infer(TypeInfer), + + /// A macro in the type position. + Macro(TypeMacro), + + /// The never type: `!`. + Never(TypeNever), + + /// A parenthesized type equivalent to the inner type. + Paren(TypeParen), + + /// A path like `std::slice::Iter`, optionally qualified with a + /// self-type as in ` as SomeTrait>::Associated`. + Path(TypePath), + + /// A raw pointer type: `*const T` or `*mut T`. + Ptr(TypePtr), + + /// A reference type: `&'a T` or `&'a mut T`. + Reference(TypeReference), + + /// A dynamically sized slice type: `[T]`. + Slice(TypeSlice), + + /// A trait object type `dyn Bound1 + Bound2 + Bound3` where `Bound` is a + /// trait or a lifetime. + TraitObject(TypeTraitObject), + + /// A tuple type: `(A, B, C, String)`. + Tuple(TypeTuple), + + /// Tokens in type position not interpreted by Syn. + Verbatim(TokenStream), + + // Not public API. + // + // For testing exhaustiveness in downstream code, use the following idiom: + // + // match ty { + // Type::Array(ty) => {...} + // Type::BareFn(ty) => {...} + // ... + // Type::Verbatim(ty) => {...} + // + // #[cfg_attr(test, deny(non_exhaustive_omitted_patterns))] + // _ => { /* some sane fallback */ } + // } + // + // This way we fail your tests but don't break your library when adding + // a variant. You will be notified by a test failure when a variant is + // added, so that you can add code to handle it, but your library will + // continue to compile and work for downstream users in the interim. + #[cfg(syn_no_non_exhaustive)] + #[doc(hidden)] + __NonExhaustive, + } +} + +ast_struct! { + /// A fixed size array type: `[T; n]`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeArray { + pub bracket_token: token::Bracket, + pub elem: Box, + pub semi_token: Token![;], + pub len: Expr, + } +} + +ast_struct! { + /// A bare function type: `fn(usize) -> bool`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeBareFn { + pub lifetimes: Option, + pub unsafety: Option, + pub abi: Option, + pub fn_token: Token![fn], + pub paren_token: token::Paren, + pub inputs: Punctuated, + pub variadic: Option, + pub output: ReturnType, + } +} + +ast_struct! { + /// A type contained within invisible delimiters. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeGroup { + pub group_token: token::Group, + pub elem: Box, + } +} + +ast_struct! { + /// An `impl Bound1 + Bound2 + Bound3` type where `Bound` is a trait or + /// a lifetime. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeImplTrait { + pub impl_token: Token![impl], + pub bounds: Punctuated, + } +} + +ast_struct! { + /// Indication that a type should be inferred by the compiler: `_`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeInfer { + pub underscore_token: Token![_], + } +} + +ast_struct! { + /// A macro in the type position. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeMacro { + pub mac: Macro, + } +} + +ast_struct! { + /// The never type: `!`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeNever { + pub bang_token: Token![!], + } +} + +ast_struct! { + /// A parenthesized type equivalent to the inner type. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeParen { + pub paren_token: token::Paren, + pub elem: Box, + } +} + +ast_struct! { + /// A path like `std::slice::Iter`, optionally qualified with a + /// self-type as in ` as SomeTrait>::Associated`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypePath { + pub qself: Option, + pub path: Path, + } +} + +ast_struct! { + /// A raw pointer type: `*const T` or `*mut T`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypePtr { + pub star_token: Token![*], + pub const_token: Option, + pub mutability: Option, + pub elem: Box, + } +} + +ast_struct! { + /// A reference type: `&'a T` or `&'a mut T`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeReference { + pub and_token: Token![&], + pub lifetime: Option, + pub mutability: Option, + pub elem: Box, + } +} + +ast_struct! { + /// A dynamically sized slice type: `[T]`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeSlice { + pub bracket_token: token::Bracket, + pub elem: Box, + } +} + +ast_struct! { + /// A trait object type `dyn Bound1 + Bound2 + Bound3` where `Bound` is a + /// trait or a lifetime. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeTraitObject { + pub dyn_token: Option, + pub bounds: Punctuated, + } +} + +ast_struct! { + /// A tuple type: `(A, B, C, String)`. + /// + /// *This type is available only if Syn is built with the `"derive"` or + /// `"full"` feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct TypeTuple { + pub paren_token: token::Paren, + pub elems: Punctuated, + } +} + +ast_struct! { + /// The binary interface of a function: `extern "C"`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Abi { + pub extern_token: Token![extern], + pub name: Option, + } +} + +ast_struct! { + /// An argument in a function type: the `usize` in `fn(usize) -> bool`. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct BareFnArg { + pub attrs: Vec, + pub name: Option<(Ident, Token![:])>, + pub ty: Type, + } +} + +ast_struct! { + /// The variadic argument of a foreign function. + /// + /// ```rust + /// # struct c_char; + /// # struct c_int; + /// # + /// extern "C" { + /// fn printf(format: *const c_char, ...) -> c_int; + /// // ^^^ + /// } + /// ``` + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub struct Variadic { + pub attrs: Vec, + pub dots: Token![...], + } +} + +ast_enum! { + /// Return type of a function signature. + /// + /// *This type is available only if Syn is built with the `"derive"` or `"full"` + /// feature.* + #[cfg_attr(doc_cfg, doc(cfg(any(feature = "full", feature = "derive"))))] + pub enum ReturnType { + /// Return type is not specified. + /// + /// Functions default to `()` and closures default to type inference. + Default, + /// A particular type is returned. + Type(Token![->], Box), + } +} + +#[cfg(feature = "parsing")] +pub mod parsing { + use super::*; + use crate::ext::IdentExt; + use crate::parse::{Parse, ParseStream, Result}; + use crate::path; + use proc_macro2::{Punct, Spacing, TokenTree}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Type { + fn parse(input: ParseStream) -> Result { + let allow_plus = true; + let allow_group_generic = true; + ambig_ty(input, allow_plus, allow_group_generic) + } + } + + impl Type { + /// In some positions, types may not contain the `+` character, to + /// disambiguate them. For example in the expression `1 as T`, T may not + /// contain a `+` character. + /// + /// This parser does not allow a `+`, while the default parser does. + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn without_plus(input: ParseStream) -> Result { + let allow_plus = false; + let allow_group_generic = true; + ambig_ty(input, allow_plus, allow_group_generic) + } + } + + pub(crate) fn ambig_ty( + input: ParseStream, + allow_plus: bool, + allow_group_generic: bool, + ) -> Result { + let begin = input.fork(); + + if input.peek(token::Group) { + let mut group: TypeGroup = input.parse()?; + if input.peek(Token![::]) && input.peek3(Ident::peek_any) { + if let Type::Path(mut ty) = *group.elem { + Path::parse_rest(input, &mut ty.path, false)?; + return Ok(Type::Path(ty)); + } else { + return Ok(Type::Path(TypePath { + qself: Some(QSelf { + lt_token: Token![<](group.group_token.span), + position: 0, + as_token: None, + gt_token: Token![>](group.group_token.span), + ty: group.elem, + }), + path: Path::parse_helper(input, false)?, + })); + } + } else if input.peek(Token![<]) && allow_group_generic + || input.peek(Token![::]) && input.peek3(Token![<]) + { + if let Type::Path(mut ty) = *group.elem { + let arguments = &mut ty.path.segments.last_mut().unwrap().arguments; + if let PathArguments::None = arguments { + *arguments = PathArguments::AngleBracketed(input.parse()?); + Path::parse_rest(input, &mut ty.path, false)?; + return Ok(Type::Path(ty)); + } else { + group.elem = Box::new(Type::Path(ty)); + } + } + } + return Ok(Type::Group(group)); + } + + let mut lifetimes = None::; + let mut lookahead = input.lookahead1(); + if lookahead.peek(Token![for]) { + lifetimes = input.parse()?; + lookahead = input.lookahead1(); + if !lookahead.peek(Ident) + && !lookahead.peek(Token![fn]) + && !lookahead.peek(Token![unsafe]) + && !lookahead.peek(Token![extern]) + && !lookahead.peek(Token![super]) + && !lookahead.peek(Token![self]) + && !lookahead.peek(Token![Self]) + && !lookahead.peek(Token![crate]) + || input.peek(Token![dyn]) + { + return Err(lookahead.error()); + } + } + + if lookahead.peek(token::Paren) { + let content; + let paren_token = parenthesized!(content in input); + if content.is_empty() { + return Ok(Type::Tuple(TypeTuple { + paren_token, + elems: Punctuated::new(), + })); + } + if content.peek(Lifetime) { + return Ok(Type::Paren(TypeParen { + paren_token, + elem: Box::new(Type::TraitObject(content.parse()?)), + })); + } + if content.peek(Token![?]) { + return Ok(Type::TraitObject(TypeTraitObject { + dyn_token: None, + bounds: { + let mut bounds = Punctuated::new(); + bounds.push_value(TypeParamBound::Trait(TraitBound { + paren_token: Some(paren_token), + ..content.parse()? + })); + while let Some(plus) = input.parse()? { + bounds.push_punct(plus); + bounds.push_value(input.parse()?); + } + bounds + }, + })); + } + let mut first: Type = content.parse()?; + if content.peek(Token![,]) { + return Ok(Type::Tuple(TypeTuple { + paren_token, + elems: { + let mut elems = Punctuated::new(); + elems.push_value(first); + elems.push_punct(content.parse()?); + while !content.is_empty() { + elems.push_value(content.parse()?); + if content.is_empty() { + break; + } + elems.push_punct(content.parse()?); + } + elems + }, + })); + } + if allow_plus && input.peek(Token![+]) { + loop { + let first = match first { + Type::Path(TypePath { qself: None, path }) => { + TypeParamBound::Trait(TraitBound { + paren_token: Some(paren_token), + modifier: TraitBoundModifier::None, + lifetimes: None, + path, + }) + } + Type::TraitObject(TypeTraitObject { + dyn_token: None, + bounds, + }) => { + if bounds.len() > 1 || bounds.trailing_punct() { + first = Type::TraitObject(TypeTraitObject { + dyn_token: None, + bounds, + }); + break; + } + match bounds.into_iter().next().unwrap() { + TypeParamBound::Trait(trait_bound) => { + TypeParamBound::Trait(TraitBound { + paren_token: Some(paren_token), + ..trait_bound + }) + } + other @ TypeParamBound::Lifetime(_) => other, + } + } + _ => break, + }; + return Ok(Type::TraitObject(TypeTraitObject { + dyn_token: None, + bounds: { + let mut bounds = Punctuated::new(); + bounds.push_value(first); + while let Some(plus) = input.parse()? { + bounds.push_punct(plus); + bounds.push_value(input.parse()?); + } + bounds + }, + })); + } + } + Ok(Type::Paren(TypeParen { + paren_token, + elem: Box::new(first), + })) + } else if lookahead.peek(Token![fn]) + || lookahead.peek(Token![unsafe]) + || lookahead.peek(Token![extern]) + { + let allow_mut_self = true; + if let Some(mut bare_fn) = parse_bare_fn(input, allow_mut_self)? { + bare_fn.lifetimes = lifetimes; + Ok(Type::BareFn(bare_fn)) + } else { + Ok(Type::Verbatim(verbatim::between(begin, input))) + } + } else if lookahead.peek(Ident) + || input.peek(Token![super]) + || input.peek(Token![self]) + || input.peek(Token![Self]) + || input.peek(Token![crate]) + || lookahead.peek(Token![::]) + || lookahead.peek(Token![<]) + { + let dyn_token: Option = input.parse()?; + if dyn_token.is_some() { + let star_token: Option = input.parse()?; + let bounds = TypeTraitObject::parse_bounds(input, allow_plus)?; + return Ok(if star_token.is_some() { + Type::Verbatim(verbatim::between(begin, input)) + } else { + Type::TraitObject(TypeTraitObject { dyn_token, bounds }) + }); + } + + let ty: TypePath = input.parse()?; + if ty.qself.is_some() { + return Ok(Type::Path(ty)); + } + + if input.peek(Token![!]) && !input.peek(Token![!=]) { + let mut contains_arguments = false; + for segment in &ty.path.segments { + match segment.arguments { + PathArguments::None => {} + PathArguments::AngleBracketed(_) | PathArguments::Parenthesized(_) => { + contains_arguments = true; + } + } + } + + if !contains_arguments { + let bang_token: Token![!] = input.parse()?; + let (delimiter, tokens) = mac::parse_delimiter(input)?; + return Ok(Type::Macro(TypeMacro { + mac: Macro { + path: ty.path, + bang_token, + delimiter, + tokens, + }, + })); + } + } + + if lifetimes.is_some() || allow_plus && input.peek(Token![+]) { + let mut bounds = Punctuated::new(); + bounds.push_value(TypeParamBound::Trait(TraitBound { + paren_token: None, + modifier: TraitBoundModifier::None, + lifetimes, + path: ty.path, + })); + if allow_plus { + while input.peek(Token![+]) { + bounds.push_punct(input.parse()?); + if !(input.peek(Ident::peek_any) + || input.peek(Token![::]) + || input.peek(Token![?]) + || input.peek(Lifetime) + || input.peek(token::Paren)) + { + break; + } + bounds.push_value(input.parse()?); + } + } + return Ok(Type::TraitObject(TypeTraitObject { + dyn_token: None, + bounds, + })); + } + + Ok(Type::Path(ty)) + } else if lookahead.peek(token::Bracket) { + let content; + let bracket_token = bracketed!(content in input); + let elem: Type = content.parse()?; + if content.peek(Token![;]) { + Ok(Type::Array(TypeArray { + bracket_token, + elem: Box::new(elem), + semi_token: content.parse()?, + len: content.parse()?, + })) + } else { + Ok(Type::Slice(TypeSlice { + bracket_token, + elem: Box::new(elem), + })) + } + } else if lookahead.peek(Token![*]) { + input.parse().map(Type::Ptr) + } else if lookahead.peek(Token![&]) { + input.parse().map(Type::Reference) + } else if lookahead.peek(Token![!]) && !input.peek(Token![=]) { + input.parse().map(Type::Never) + } else if lookahead.peek(Token![impl]) { + TypeImplTrait::parse(input, allow_plus).map(Type::ImplTrait) + } else if lookahead.peek(Token![_]) { + input.parse().map(Type::Infer) + } else if lookahead.peek(Lifetime) { + input.parse().map(Type::TraitObject) + } else { + Err(lookahead.error()) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeSlice { + fn parse(input: ParseStream) -> Result { + let content; + Ok(TypeSlice { + bracket_token: bracketed!(content in input), + elem: content.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeArray { + fn parse(input: ParseStream) -> Result { + let content; + Ok(TypeArray { + bracket_token: bracketed!(content in input), + elem: content.parse()?, + semi_token: content.parse()?, + len: content.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypePtr { + fn parse(input: ParseStream) -> Result { + let star_token: Token![*] = input.parse()?; + + let lookahead = input.lookahead1(); + let (const_token, mutability) = if lookahead.peek(Token![const]) { + (Some(input.parse()?), None) + } else if lookahead.peek(Token![mut]) { + (None, Some(input.parse()?)) + } else { + return Err(lookahead.error()); + }; + + Ok(TypePtr { + star_token, + const_token, + mutability, + elem: Box::new(input.call(Type::without_plus)?), + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeReference { + fn parse(input: ParseStream) -> Result { + Ok(TypeReference { + and_token: input.parse()?, + lifetime: input.parse()?, + mutability: input.parse()?, + // & binds tighter than +, so we don't allow + here. + elem: Box::new(input.call(Type::without_plus)?), + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeBareFn { + fn parse(input: ParseStream) -> Result { + let allow_mut_self = false; + parse_bare_fn(input, allow_mut_self).map(Option::unwrap) + } + } + + fn parse_bare_fn(input: ParseStream, allow_mut_self: bool) -> Result> { + let args; + let mut variadic = None; + let mut has_mut_self = false; + + let bare_fn = TypeBareFn { + lifetimes: input.parse()?, + unsafety: input.parse()?, + abi: input.parse()?, + fn_token: input.parse()?, + paren_token: parenthesized!(args in input), + inputs: { + let mut inputs = Punctuated::new(); + + while !args.is_empty() { + let attrs = args.call(Attribute::parse_outer)?; + + if inputs.empty_or_trailing() && args.peek(Token![...]) { + variadic = Some(Variadic { + attrs, + dots: args.parse()?, + }); + break; + } + + if let Some(arg) = parse_bare_fn_arg(&args, allow_mut_self)? { + inputs.push_value(BareFnArg { attrs, ..arg }); + } else { + has_mut_self = true; + } + if args.is_empty() { + break; + } + + let comma = args.parse()?; + if !has_mut_self { + inputs.push_punct(comma); + } + } + + inputs + }, + variadic, + output: input.call(ReturnType::without_plus)?, + }; + + if has_mut_self { + Ok(None) + } else { + Ok(Some(bare_fn)) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeNever { + fn parse(input: ParseStream) -> Result { + Ok(TypeNever { + bang_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeInfer { + fn parse(input: ParseStream) -> Result { + Ok(TypeInfer { + underscore_token: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeTuple { + fn parse(input: ParseStream) -> Result { + let content; + let paren_token = parenthesized!(content in input); + + if content.is_empty() { + return Ok(TypeTuple { + paren_token, + elems: Punctuated::new(), + }); + } + + let first: Type = content.parse()?; + Ok(TypeTuple { + paren_token, + elems: { + let mut elems = Punctuated::new(); + elems.push_value(first); + elems.push_punct(content.parse()?); + while !content.is_empty() { + elems.push_value(content.parse()?); + if content.is_empty() { + break; + } + elems.push_punct(content.parse()?); + } + elems + }, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeMacro { + fn parse(input: ParseStream) -> Result { + Ok(TypeMacro { + mac: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypePath { + fn parse(input: ParseStream) -> Result { + let expr_style = false; + let (qself, mut path) = path::parsing::qpath(input, expr_style)?; + + while path.segments.last().unwrap().arguments.is_empty() + && (input.peek(token::Paren) || input.peek(Token![::]) && input.peek3(token::Paren)) + { + input.parse::>()?; + let args: ParenthesizedGenericArguments = input.parse()?; + let allow_associated_type = cfg!(feature = "full") + && match &args.output { + ReturnType::Default => true, + ReturnType::Type(_, ty) => match **ty { + // TODO: probably some of the other kinds allow this too. + Type::Paren(_) => true, + _ => false, + }, + }; + let parenthesized = PathArguments::Parenthesized(args); + path.segments.last_mut().unwrap().arguments = parenthesized; + if allow_associated_type { + Path::parse_rest(input, &mut path, expr_style)?; + } + } + + Ok(TypePath { qself, path }) + } + } + + impl ReturnType { + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn without_plus(input: ParseStream) -> Result { + let allow_plus = false; + Self::parse(input, allow_plus) + } + + pub(crate) fn parse(input: ParseStream, allow_plus: bool) -> Result { + if input.peek(Token![->]) { + let arrow = input.parse()?; + let allow_group_generic = true; + let ty = ambig_ty(input, allow_plus, allow_group_generic)?; + Ok(ReturnType::Type(arrow, Box::new(ty))) + } else { + Ok(ReturnType::Default) + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for ReturnType { + fn parse(input: ParseStream) -> Result { + let allow_plus = true; + Self::parse(input, allow_plus) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeTraitObject { + fn parse(input: ParseStream) -> Result { + let allow_plus = true; + Self::parse(input, allow_plus) + } + } + + fn at_least_one_type(bounds: &Punctuated) -> bool { + for bound in bounds { + if let TypeParamBound::Trait(_) = *bound { + return true; + } + } + false + } + + impl TypeTraitObject { + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn without_plus(input: ParseStream) -> Result { + let allow_plus = false; + Self::parse(input, allow_plus) + } + + // Only allow multiple trait references if allow_plus is true. + pub(crate) fn parse(input: ParseStream, allow_plus: bool) -> Result { + Ok(TypeTraitObject { + dyn_token: input.parse()?, + bounds: Self::parse_bounds(input, allow_plus)?, + }) + } + + fn parse_bounds( + input: ParseStream, + allow_plus: bool, + ) -> Result> { + let mut bounds = Punctuated::new(); + loop { + bounds.push_value(input.parse()?); + if !(allow_plus && input.peek(Token![+])) { + break; + } + bounds.push_punct(input.parse()?); + if !(input.peek(Ident::peek_any) + || input.peek(Token![::]) + || input.peek(Token![?]) + || input.peek(Lifetime) + || input.peek(token::Paren)) + { + break; + } + } + // Just lifetimes like `'a + 'b` is not a TraitObject. + if !at_least_one_type(&bounds) { + return Err(input.error("expected at least one type")); + } + Ok(bounds) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeImplTrait { + fn parse(input: ParseStream) -> Result { + let allow_plus = true; + Self::parse(input, allow_plus) + } + } + + impl TypeImplTrait { + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + pub fn without_plus(input: ParseStream) -> Result { + let allow_plus = false; + Self::parse(input, allow_plus) + } + + pub(crate) fn parse(input: ParseStream, allow_plus: bool) -> Result { + Ok(TypeImplTrait { + impl_token: input.parse()?, + bounds: TypeTraitObject::parse_bounds(input, allow_plus)?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeGroup { + fn parse(input: ParseStream) -> Result { + let group = crate::group::parse_group(input)?; + Ok(TypeGroup { + group_token: group.token, + elem: group.content.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for TypeParen { + fn parse(input: ParseStream) -> Result { + let allow_plus = false; + Self::parse(input, allow_plus) + } + } + + impl TypeParen { + fn parse(input: ParseStream, allow_plus: bool) -> Result { + let content; + Ok(TypeParen { + paren_token: parenthesized!(content in input), + elem: Box::new({ + let allow_group_generic = true; + ambig_ty(&content, allow_plus, allow_group_generic)? + }), + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for BareFnArg { + fn parse(input: ParseStream) -> Result { + let allow_mut_self = false; + parse_bare_fn_arg(input, allow_mut_self).map(Option::unwrap) + } + } + + fn parse_bare_fn_arg( + input: ParseStream, + mut allow_mut_self: bool, + ) -> Result> { + let mut has_mut_self = false; + let arg = BareFnArg { + attrs: input.call(Attribute::parse_outer)?, + name: { + if (input.peek(Ident) || input.peek(Token![_]) || input.peek(Token![self])) + && input.peek2(Token![:]) + && !input.peek2(Token![::]) + { + let name = input.call(Ident::parse_any)?; + let colon: Token![:] = input.parse()?; + Some((name, colon)) + } else if allow_mut_self + && input.peek(Token![mut]) + && input.peek2(Token![self]) + && input.peek3(Token![:]) + && !input.peek3(Token![::]) + { + has_mut_self = true; + allow_mut_self = false; + input.parse::()?; + input.parse::()?; + input.parse::()?; + None + } else { + None + } + }, + ty: if !has_mut_self && input.peek(Token![...]) { + let dot3 = input.parse::()?; + let args = vec![ + TokenTree::Punct(Punct::new('.', Spacing::Joint)), + TokenTree::Punct(Punct::new('.', Spacing::Joint)), + TokenTree::Punct(Punct::new('.', Spacing::Alone)), + ]; + let tokens: TokenStream = args + .into_iter() + .zip(&dot3.spans) + .map(|(mut arg, span)| { + arg.set_span(*span); + arg + }) + .collect(); + Type::Verbatim(tokens) + } else if allow_mut_self && input.peek(Token![mut]) && input.peek2(Token![self]) { + has_mut_self = true; + input.parse::()?; + Type::Path(TypePath { + qself: None, + path: input.parse::()?.into(), + }) + } else { + input.parse()? + }, + }; + + if has_mut_self { + Ok(None) + } else { + Ok(Some(arg)) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Abi { + fn parse(input: ParseStream) -> Result { + Ok(Abi { + extern_token: input.parse()?, + name: input.parse()?, + }) + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "parsing")))] + impl Parse for Option { + fn parse(input: ParseStream) -> Result { + if input.peek(Token![extern]) { + input.parse().map(Some) + } else { + Ok(None) + } + } + } +} + +#[cfg(feature = "printing")] +mod printing { + use super::*; + use crate::attr::FilterAttrs; + use crate::print::TokensOrDefault; + use proc_macro2::TokenStream; + use quote::{ToTokens, TokenStreamExt}; + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeSlice { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.bracket_token.surround(tokens, |tokens| { + self.elem.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeArray { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.bracket_token.surround(tokens, |tokens| { + self.elem.to_tokens(tokens); + self.semi_token.to_tokens(tokens); + self.len.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypePtr { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.star_token.to_tokens(tokens); + match &self.mutability { + Some(tok) => tok.to_tokens(tokens), + None => { + TokensOrDefault(&self.const_token).to_tokens(tokens); + } + } + self.elem.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeReference { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.and_token.to_tokens(tokens); + self.lifetime.to_tokens(tokens); + self.mutability.to_tokens(tokens); + self.elem.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeBareFn { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.lifetimes.to_tokens(tokens); + self.unsafety.to_tokens(tokens); + self.abi.to_tokens(tokens); + self.fn_token.to_tokens(tokens); + self.paren_token.surround(tokens, |tokens| { + self.inputs.to_tokens(tokens); + if let Some(variadic) = &self.variadic { + if !self.inputs.empty_or_trailing() { + let span = variadic.dots.spans[0]; + Token![,](span).to_tokens(tokens); + } + variadic.to_tokens(tokens); + } + }); + self.output.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeNever { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.bang_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeTuple { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.paren_token.surround(tokens, |tokens| { + self.elems.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypePath { + fn to_tokens(&self, tokens: &mut TokenStream) { + path::printing::print_path(tokens, &self.qself, &self.path); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeTraitObject { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.dyn_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeImplTrait { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.impl_token.to_tokens(tokens); + self.bounds.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeGroup { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.group_token.surround(tokens, |tokens| { + self.elem.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeParen { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.paren_token.surround(tokens, |tokens| { + self.elem.to_tokens(tokens); + }); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeInfer { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.underscore_token.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for TypeMacro { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.mac.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for ReturnType { + fn to_tokens(&self, tokens: &mut TokenStream) { + match self { + ReturnType::Default => {} + ReturnType::Type(arrow, ty) => { + arrow.to_tokens(tokens); + ty.to_tokens(tokens); + } + } + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for BareFnArg { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + if let Some((name, colon)) = &self.name { + name.to_tokens(tokens); + colon.to_tokens(tokens); + } + self.ty.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Variadic { + fn to_tokens(&self, tokens: &mut TokenStream) { + tokens.append_all(self.attrs.outer()); + self.dots.to_tokens(tokens); + } + } + + #[cfg_attr(doc_cfg, doc(cfg(feature = "printing")))] + impl ToTokens for Abi { + fn to_tokens(&self, tokens: &mut TokenStream) { + self.extern_token.to_tokens(tokens); + self.name.to_tokens(tokens); + } + } +} diff --git a/rust/syn/verbatim.rs b/rust/syn/verbatim.rs new file mode 100644 index 00000000000000..15e651164a02e8 --- /dev/null +++ b/rust/syn/verbatim.rs @@ -0,0 +1,35 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +use crate::parse::{ParseBuffer, ParseStream}; +use proc_macro2::{Delimiter, TokenStream}; +use std::cmp::Ordering; +use std::iter; + +pub fn between<'a>(begin: ParseBuffer<'a>, end: ParseStream<'a>) -> TokenStream { + let end = end.cursor(); + let mut cursor = begin.cursor(); + assert!(crate::buffer::same_buffer(end, cursor)); + + let mut tokens = TokenStream::new(); + while cursor != end { + let (tt, next) = cursor.token_tree().unwrap(); + + if crate::buffer::cmp_assuming_same_buffer(end, next) == Ordering::Less { + // A syntax node can cross the boundary of a None-delimited group + // due to such groups being transparent to the parser in most cases. + // Any time this occurs the group is known to be semantically + // irrelevant. https://github.com/dtolnay/syn/issues/1235 + if let Some((inside, _span, after)) = cursor.group(Delimiter::None) { + assert!(next == after); + cursor = inside; + continue; + } else { + panic!("verbatim end must not be inside a delimited group"); + } + } + + tokens.extend(iter::once(tt)); + cursor = next; + } + tokens +} diff --git a/rust/syn/whitespace.rs b/rust/syn/whitespace.rs new file mode 100644 index 00000000000000..99c2d3ab28f916 --- /dev/null +++ b/rust/syn/whitespace.rs @@ -0,0 +1,67 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +pub fn skip(mut s: &str) -> &str { + 'skip: while !s.is_empty() { + let byte = s.as_bytes()[0]; + if byte == b'/' { + if s.starts_with("//") + && (!s.starts_with("///") || s.starts_with("////")) + && !s.starts_with("//!") + { + if let Some(i) = s.find('\n') { + s = &s[i + 1..]; + continue; + } else { + return ""; + } + } else if s.starts_with("/**/") { + s = &s[4..]; + continue; + } else if s.starts_with("/*") + && (!s.starts_with("/**") || s.starts_with("/***")) + && !s.starts_with("/*!") + { + let mut depth = 0; + let bytes = s.as_bytes(); + let mut i = 0; + let upper = bytes.len() - 1; + while i < upper { + if bytes[i] == b'/' && bytes[i + 1] == b'*' { + depth += 1; + i += 1; // eat '*' + } else if bytes[i] == b'*' && bytes[i + 1] == b'/' { + depth -= 1; + if depth == 0 { + s = &s[i + 2..]; + continue 'skip; + } + i += 1; // eat '/' + } + i += 1; + } + return s; + } + } + match byte { + b' ' | 0x09..=0x0d => { + s = &s[1..]; + continue; + } + b if b <= 0x7f => {} + _ => { + let ch = s.chars().next().unwrap(); + if is_whitespace(ch) { + s = &s[ch.len_utf8()..]; + continue; + } + } + } + return s; + } + s +} + +fn is_whitespace(ch: char) -> bool { + // Rust treats left-to-right mark and right-to-left mark as whitespace + ch.is_whitespace() || ch == '\u{200e}' || ch == '\u{200f}' +} diff --git a/samples/rust/Kconfig b/samples/rust/Kconfig index b0f74a81c8f9ad..9151045ebbfe6a 100644 --- a/samples/rust/Kconfig +++ b/samples/rust/Kconfig @@ -30,6 +30,16 @@ config SAMPLE_RUST_PRINT If unsure, say N. +config SAMPLE_RUST_SERDE + tristate "Serde" + help + This option builds the Rust `serde` sample. + + To compile this as a module, choose M here: + the module will be called rust_serde. + + If unsure, say N. + config SAMPLE_RUST_HOSTPROGS bool "Host programs" help diff --git a/samples/rust/Makefile b/samples/rust/Makefile index 03086dabbea44f..11fc58df91f197 100644 --- a/samples/rust/Makefile +++ b/samples/rust/Makefile @@ -2,5 +2,6 @@ obj-$(CONFIG_SAMPLE_RUST_MINIMAL) += rust_minimal.o obj-$(CONFIG_SAMPLE_RUST_PRINT) += rust_print.o +obj-$(CONFIG_SAMPLE_RUST_SERDE) += rust_serde.o subdir-$(CONFIG_SAMPLE_RUST_HOSTPROGS) += hostprogs diff --git a/samples/rust/local_data_format/de.rs b/samples/rust/local_data_format/de.rs new file mode 100644 index 00000000000000..32cfc53f98b3f8 --- /dev/null +++ b/samples/rust/local_data_format/de.rs @@ -0,0 +1,422 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use super::error::{Error, Result}; +use serde::de::{self, Deserialize, DeserializeSeed, SeqAccess, Visitor}; + +pub struct Deserializer<'de> { + // This string starts with the input data and characters are truncated off + // the beginning as data is parsed. + input: &'de [u8], +} + +impl<'de> Deserializer<'de> { + // By convention, `Deserializer` constructors are named like `from_xyz`. + // That way basic use cases are satisfied by something like + // `serde_json::from_str(...)` while advanced use cases that require a + // deserializer can make one with `serde_json::Deserializer::from_str(...)`. + #[allow(clippy::should_implement_trait)] + pub fn from_bytes(input: &'de [u8]) -> Self { + Deserializer { input } + } +} + +// By convention, the public API of a Serde deserializer is one or more +// `from_xyz` methods such as `from_str`, `from_bytes`, or `from_reader` +// depending on what Rust types the deserializer is able to consume as input. +// +// This basic deserializer supports only `from_str`. +pub fn from_bytes<'a, T>(s: &'a [u8]) -> Result +where + T: Deserialize<'a>, +{ + let mut deserializer = Deserializer::from_bytes(s); + let t = T::deserialize(&mut deserializer)?; + if deserializer.input.is_empty() { + Ok(t) + } else { + Err(Error::TrailingCharacters) + } +} + +// SERDE IS NOT A PARSING LIBRARY. This impl block defines a few basic parsing +// functions from scratch. More complicated formats may wish to use a dedicated +// parsing library to help implement their Serde deserializer. +impl<'de> Deserializer<'de> { + // Look at the first character in the input without consuming it. + fn peek_byte(&mut self) -> Result { + self.input.iter().next().ok_or(Error::Eof).map(|v| *v) + } + + // Consume the first character in the input. + fn next_byte(&mut self) -> Result { + let ch = self.peek_byte()?; + self.input = &self.input[1..]; + Ok(ch) + } + + // Parse the JSON identifier `true` or `false`. + fn parse_bool(&mut self) -> Result { + if self.input.starts_with(&[1]) { + self.input = &self.input[1..]; + match self.next_byte()? { + 42 => Ok(false), + 43 => Ok(true), + _ => Err(Error::InvalidBooleanValue), + } + } else { + Err(Error::ExpectedBoolean) + } + } + + // Parse a group of decimal digits as an unsigned integer of type T. + // + // This implementation is a bit too lenient, for example `001` is not + // allowed in JSON. Also the various arithmetic operations can overflow and + // panic or return bogus data. But it is good enough for example code! + fn parse_unsigned(&mut self) -> Result { + unimplemented!() + } + + // Parse a possible minus sign followed by a group of decimal digits as a + // signed integer of type T. + fn parse_signed(&mut self) -> Result { + // Optional minus sign, delegate to `parse_unsigned`, negate if negative. + unimplemented!() + } +} + +impl<'de, 'a> de::Deserializer<'de> for &'a mut Deserializer<'de> { + type Error = Error; + + // Look at the input data to decide what Serde data model type to + // deserialize as. Not all data formats are able to support this operation. + // Formats that support `deserialize_any` are known as self-describing. + fn deserialize_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + match self.peek_byte()? { + 0 => self.deserialize_unit(visitor), + 1 => self.deserialize_bool(visitor), + 2 => self.deserialize_map(visitor), + _ => Err(Error::Syntax), + } + } + + // Uses the `parse_bool` parsing function defined above to read the JSON + // identifier `true` or `false` from the input. + // + // Parsing refers to looking at the input and deciding that it contains the + // JSON value `true` or `false`. + // + // Deserialization refers to mapping that JSON value into Serde's data + // model by invoking one of the `Visitor` methods. In the case of JSON and + // bool that mapping is straightforward so the distinction may seem silly, + // but in other cases Deserializers sometimes perform non-obvious mappings. + // For example the TOML format has a Datetime type and Serde's data model + // does not. In the `toml` crate, a Datetime in the input is deserialized by + // mapping it to a Serde data model "struct" type with a special name and a + // single field containing the Datetime represented as a string. + fn deserialize_bool(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_bool(self.parse_bool()?) + } + + // The `parse_signed` function is generic over the integer type `T` so here + // it is invoked with `T=i8`. The next 8 methods are similar. + fn deserialize_i8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i8(self.parse_signed()?) + } + + fn deserialize_i16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i16(self.parse_signed()?) + } + + fn deserialize_i32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i32(self.parse_signed()?) + } + + fn deserialize_i64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_i64(self.parse_signed()?) + } + + fn deserialize_u8(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u8(self.parse_unsigned()?) + } + + fn deserialize_u16(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u16(self.parse_unsigned()?) + } + + fn deserialize_u32(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u32(self.parse_unsigned()?) + } + + fn deserialize_u64(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_u64(self.parse_unsigned()?) + } + + // The `Serializer` implementation on the previous page serialized chars as + // single-character strings so handle that representation here. + fn deserialize_char(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // Refer to the "Understanding deserializer lifetimes" page for information + // about the three deserialization flavors of strings in Serde. + fn deserialize_str(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + fn deserialize_string(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // The `Serializer` implementation on the previous page serialized byte + // arrays as JSON arrays of bytes. Handle that representation here. + fn deserialize_bytes(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + fn deserialize_byte_buf(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // An absent optional is represented as the JSON `null` and a present + // optional is represented as just the contained value. + // + // As commented in `Serializer` implementation, this is a lossy + // representation. For example the values `Some(())` and `None` both + // serialize as just `null`. Unfortunately this is typically what people + // expect when working with JSON. Other formats are encouraged to behave + // more intelligently if possible. + fn deserialize_option(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // In Serde, unit means an anonymous value containing no data. + fn deserialize_unit(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + if self.input.starts_with(&[0]) { + self.input = &self.input[1..]; + visitor.visit_unit() + } else { + Err(Error::ExpectedNull) + } + } + + // Unit struct means a named value containing no data. + fn deserialize_unit_struct(self, _name: &'static str, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + // As is done here, serializers are encouraged to treat newtype structs as + // insignificant wrappers around the data they contain. That means not + // parsing anything other than the contained value. + fn deserialize_newtype_struct(self, _name: &'static str, visitor: V) -> Result + where + V: Visitor<'de>, + { + visitor.visit_newtype_struct(self) + } + + // Deserialization of compound types like sequences and maps happens by + // passing the visitor an "Access" object that gives it the ability to + // iterate through the data contained in the sequence. + fn deserialize_seq(self, _visitor: V) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // Tuples look just like sequences in JSON. Some formats may be able to + // represent tuples more efficiently. + // + // As indicated by the length parameter, the `Deserialize` implementation + // for a tuple in the Serde data model is required to know the length of the + // tuple before even looking at the input data. + fn deserialize_tuple(self, _len: usize, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + // Tuple structs look just like sequences in JSON. + fn deserialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + // Much like `deserialize_seq` but calls the visitors `visit_map` method + // with a `MapAccess` implementation, rather than the visitor's `visit_seq` + // method with a `SeqAccess` implementation. + fn deserialize_map(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + // Parse the opening brace of the map. + if self.next_byte()? == 2 { + // Give the visitor access to each entry of the map. + let value = visitor.visit_seq(StructFieldsVisitor::new(self))?; + // Parse the closing brace of the map. + if self.next_byte()? == 3 { + Ok(value) + } else { + Err(Error::ExpectedMapEnd) + } + } else { + Err(Error::ExpectedMap) + } + } + + // Structs look just like maps in JSON. + // + // Notice the `fields` parameter - a "struct" in the Serde data model means + // that the `Deserialize` implementation is required to know what the fields + // are before even looking at the input data. Any key-value pairing in which + // the fields cannot be known ahead of time is probably a map. + fn deserialize_struct( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + self.deserialize_map(visitor) + } + + fn deserialize_enum( + self, + _name: &'static str, + _variants: &'static [&'static str], + _visitor: V, + ) -> Result + where + V: Visitor<'de>, + { + unimplemented!() + } + + // An identifier in Serde is the type that identifies a field of a struct or + // the variant of an enum. In JSON, struct fields and enum variants are + // represented as strings. In other formats they may be represented as + // numeric indices. + fn deserialize_identifier(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + // Like `deserialize_any` but indicates to the `Deserializer` that it makes + // no difference which `Visitor` method is called because the data is + // ignored. + // + // Some deserializers are able to implement this more efficiently than + // `deserialize_any`, for example by rapidly skipping over matched + // delimiters without paying close attention to the data in between. + // + // Some formats are not able to implement this at all. Formats that can + // implement `deserialize_any` and `deserialize_ignored_any` are known as + // self-describing. + fn deserialize_ignored_any(self, visitor: V) -> Result + where + V: Visitor<'de>, + { + self.deserialize_any(visitor) + } +} + +struct StructFieldsVisitor<'a, 'de> { + de: &'a mut Deserializer<'de>, +} + +impl<'a, 'de> StructFieldsVisitor<'a, 'de> { + fn new(de: &'a mut Deserializer<'de>) -> Self { + StructFieldsVisitor { de } + } +} + +impl<'de, 'a> SeqAccess<'de> for StructFieldsVisitor<'a, 'de> { + type Error = Error; + + fn next_element_seed(&mut self, seed: T) -> Result> + where + T: DeserializeSeed<'de>, + { + // Check if there are no more elements. + if self.de.peek_byte()? == 3 { + return Ok(None); + } + // Deserialize an array element. + seed.deserialize(&mut *self.de).map(Some) + } +} diff --git a/samples/rust/local_data_format/error.rs b/samples/rust/local_data_format/error.rs new file mode 100644 index 00000000000000..8d913580f71363 --- /dev/null +++ b/samples/rust/local_data_format/error.rs @@ -0,0 +1,73 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use core::fmt::{self, Display}; +use serde::{de, ser}; + +pub type Result = kernel::error::Result; + +// This is a bare-bones implementation. A real library would provide additional +// information in its error type, for example the line and column at which the +// error occurred, the byte offset into the input, or the current key being +// processed. +#[derive(Debug)] +pub enum Error { + // One or more variants that can be created by data structures through the + // `ser::Error` and `de::Error` traits. For example the Serialize impl for + // Mutex might return an error because the mutex is poisoned, or the + // Deserialize impl for a struct may return an error because a required + // field is missing. + Message, + + // Zero or more variants that can be created directly by the Serializer and + // Deserializer without going through `ser::Error` and `de::Error`. These + // are specific to the format, in this case JSON. + Eof, + Syntax, + ExpectedBoolean, + InvalidBooleanValue, + ExpectedInteger, + ExpectedString, + ExpectedNull, + ExpectedArray, + ExpectedArrayComma, + ExpectedArrayEnd, + ExpectedMap, + ExpectedMapColon, + ExpectedMapComma, + ExpectedMapEnd, + ExpectedEnum, + TrailingCharacters, +} + +impl ser::Error for Error { + fn custom(_msg: T) -> Self { + Error::Message + } +} + +impl de::Error for Error { + fn custom(_msg: T) -> Self { + Error::Message + } +} + +impl Display for Error { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match self { + Error::Message => write!(f, "message"), + Error::Eof => f.write_str("unexpected end of input"), + /* and so forth */ + _ => unimplemented!(), + } + } +} + +//impl core::error::Error for Error {} diff --git a/samples/rust/local_data_format/ser.rs b/samples/rust/local_data_format/ser.rs new file mode 100644 index 00000000000000..f4f17eb6da66e6 --- /dev/null +++ b/samples/rust/local_data_format/ser.rs @@ -0,0 +1,443 @@ +// SPDX-License-Identifier: Apache-2.0 OR MIT + +// Copyright 2018 Serde Developers +// +// Licensed under the Apache License, Version 2.0 or the MIT license +// , at your +// option. This file may not be copied, modified, or distributed +// except according to those terms. + +use super::error::{Error, Result}; +use alloc::vec::Vec; +use serde::ser::{self, Serialize}; + +pub struct Serializer { + // This string starts empty and JSON is appended as values are serialized. + output: Vec, +} + +// By convention, the public API of a Serde serializer is one or more `to_abc` +// functions such as `to_string`, `to_bytes`, or `to_writer` depending on what +// Rust types the serializer is able to produce as output. +// +// This basic serializer supports only `to_string`. +pub fn to_vec(value: &T) -> Result> +where + T: Serialize, +{ + let mut serializer = Serializer { output: Vec::new() }; + value.serialize(&mut serializer)?; + Ok(serializer.output) +} + +impl<'a> ser::Serializer for &'a mut Serializer { + // The output type produced by this `Serializer` during successful + // serialization. Most serializers that produce text or binary output should + // set `Ok = ()` and serialize into an `io::Write` or buffer contained + // within the `Serializer` instance, as happens here. Serializers that build + // in-memory data structures may be simplified by using `Ok` to propagate + // the data structure around. + type Ok = (); + + // The error type when some error occurs during serialization. + type Error = Error; + + // Associated types for keeping track of additional state while serializing + // compound data structures like sequences and maps. In this case no + // additional state is required beyond what is already stored in the + // Serializer struct. + type SerializeSeq = Self; + type SerializeTuple = Self; + type SerializeTupleStruct = Self; + type SerializeTupleVariant = Self; + type SerializeMap = Self; + type SerializeStruct = Self; + type SerializeStructVariant = Self; + + // Here we go with the simple methods. The following 12 methods receive one + // of the primitive types of the data model and map it to JSON by appending + // into the output string. + fn serialize_bool(self, v: bool) -> Result<()> { + self.output.try_push(1).unwrap(); + self.output.try_push(if v { 43 } else { 42 }).unwrap(); + Ok(()) + } + + // JSON does not distinguish between different sizes of integers, so all + // signed integers will be serialized the same and all unsigned integers + // will be serialized the same. Other formats, especially compact binary + // formats, may need independent logic for the different sizes. + fn serialize_i8(self, v: i8) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + fn serialize_i16(self, v: i16) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + fn serialize_i32(self, v: i32) -> Result<()> { + self.serialize_i64(i64::from(v)) + } + + // Not particularly efficient but this is example code anyway. A more + // performant approach would be to use the `itoa` crate. + fn serialize_i64(self, _v: i64) -> Result<()> { + unimplemented!(); + } + + fn serialize_u8(self, v: u8) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u16(self, v: u16) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u32(self, v: u32) -> Result<()> { + self.serialize_u64(u64::from(v)) + } + + fn serialize_u64(self, _v: u64) -> Result<()> { + unimplemented!(); + } + + // Serialize a char as a single-character string. Other formats may + // represent this differently. + fn serialize_char(self, _v: char) -> Result<()> { + unimplemented!(); + } + + // This only works for strings that don't require escape sequences but you + // get the idea. For example it would emit invalid JSON if the input string + // contains a '"' character. + fn serialize_str(self, _v: &str) -> Result<()> { + unimplemented!(); + } + + // Serialize a byte array as an array of bytes. Could also use a base64 + // string here. Binary formats will typically represent byte arrays more + // compactly. + fn serialize_bytes(self, v: &[u8]) -> Result<()> { + use serde::ser::SerializeSeq; + let mut seq = self.serialize_seq(Some(v.len()))?; + for byte in v { + seq.serialize_element(byte)?; + } + seq.end() + } + + // An absent optional is represented as the JSON `null`. + fn serialize_none(self) -> Result<()> { + self.serialize_unit() + } + + // A present optional is represented as just the contained value. Note that + // this is a lossy representation. For example the values `Some(())` and + // `None` both serialize as just `null`. Unfortunately this is typically + // what people expect when working with JSON. Other formats are encouraged + // to behave more intelligently if possible. + fn serialize_some(self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + // In Serde, unit means an anonymous value containing no data. Map this to + // JSON as `null`. + fn serialize_unit(self) -> Result<()> { + self.output.try_push(0).unwrap(); + Ok(()) + } + + // Unit struct means a named value containing no data. Again, since there is + // no data, map this to JSON as `null`. There is no need to serialize the + // name in most formats. + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + self.serialize_unit() + } + + // When serializing a unit variant (or any other kind of variant), formats + // can choose whether to keep track of it by index or by name. Binary + // formats typically use the index of the variant and human-readable formats + // typically use the name. + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<()> { + self.serialize_str(variant) + } + + // As is done here, serializers are encouraged to treat newtype structs as + // insignificant wrappers around the data they contain. + fn serialize_newtype_struct(self, _name: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + // Note that newtype variant (and all of the other variant serialization + // methods) refer exclusively to the "externally tagged" enum + // representation. + // + // Serialize this to JSON in externally tagged form as `{ NAME: VALUE }`. + fn serialize_newtype_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + // Now we get to the serialization of compound types. + // + // The start of the sequence, each value, and the end are three separate + // method calls. This one is responsible only for serializing the start, + // which in JSON is `[`. + // + // The length of the sequence may or may not be known ahead of time. This + // doesn't make a difference in JSON because the length is not represented + // explicitly in the serialized form. Some serializers may only be able to + // support sequences for which the length is known up front. + fn serialize_seq(self, _len: Option) -> Result { + unimplemented!(); + } + + // Tuples look just like sequences in JSON. Some formats may be able to + // represent tuples more efficiently by omitting the length, since tuple + // means that the corresponding `Deserialize implementation will know the + // length without needing to look at the serialized data. + fn serialize_tuple(self, len: usize) -> Result { + self.serialize_seq(Some(len)) + } + + // Tuple structs look just like sequences in JSON. + fn serialize_tuple_struct( + self, + _name: &'static str, + len: usize, + ) -> Result { + self.serialize_seq(Some(len)) + } + + // Tuple variants are represented in JSON as `{ NAME: [DATA...] }`. Again + // this method is only responsible for the externally tagged representation. + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + unimplemented!(); + } + + // Maps are represented in JSON as `{ K: V, K: V, ... }`. + fn serialize_map(self, _len: Option) -> Result { + self.output.try_push(2).unwrap(); + Ok(self) + } + + // Structs look just like maps in JSON. In particular, JSON requires that we + // serialize the field names of the struct. Other formats may be able to + // omit the field names when serializing structs because the corresponding + // Deserialize implementation is required to know what the keys are without + // looking at the serialized data. + fn serialize_struct(self, _name: &'static str, len: usize) -> Result { + self.serialize_map(Some(len)) + } + + // Struct variants are represented in JSON as `{ NAME: { K: V, ... } }`. + // This is the externally tagged representation. + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result { + unimplemented!(); + } + + fn collect_str(self, _: &T) -> Result<()> + where + T: core::fmt::Display, + { + unimplemented!() + } +} + +// The following 7 impls deal with the serialization of compound types like +// sequences and maps. Serialization of such types is begun by a Serializer +// method and followed by zero or more calls to serialize individual elements of +// the compound type and one call to end the compound type. +// +// This impl is SerializeSeq so these methods are called after `serialize_seq` +// is called on the Serializer. +impl<'a> ser::SerializeSeq for &'a mut Serializer { + // Must match the `Ok` type of the serializer. + type Ok = (); + // Must match the `Error` type of the serializer. + type Error = Error; + + // Serialize a single element of the sequence. + fn serialize_element(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + // Close the sequence. + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Same thing but for tuples. +impl<'a> ser::SerializeTuple for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_element(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Same thing but for tuple structs. +impl<'a> ser::SerializeTupleStruct for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Tuple variants are a little different. Refer back to the +// `serialize_tuple_variant` method above: +// +// self.output += "{"; +// variant.serialize(&mut *self)?; +// self.output += ":["; +// +// So the `end` method in this impl is responsible for closing both the `]` and +// the `}`. +impl<'a> ser::SerializeTupleVariant for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} + +// Some `Serialize` types are not able to hold a key and value in memory at the +// same time so `SerializeMap` implementations are required to support +// `serialize_key` and `serialize_value` individually. +// +// There is a third optional method on the `SerializeMap` trait. The +// `serialize_entry` method allows serializers to optimize for the case where +// key and value are both available simultaneously. In JSON it doesn't make a +// difference so the default behavior for `serialize_entry` is fine. +impl<'a> ser::SerializeMap for &'a mut Serializer { + type Ok = (); + type Error = Error; + + // The Serde data model allows map keys to be any serializable type. JSON + // only allows string keys so the implementation below will produce invalid + // JSON if the key serializes as something other than a string. + // + // A real JSON serializer would need to validate that map keys are strings. + // This can be done by using a different Serializer to serialize the key + // (instead of `&mut **self`) and having that other serializer only + // implement `serialize_str` and return an error on any other data type. + fn serialize_key(&mut self, _key: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!() + } + + // It doesn't make a difference whether the colon is printed at the end of + // `serialize_key` or at the beginning of `serialize_value`. In this case + // the code is a bit simpler having it here. + fn serialize_value(&mut self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!() + } + + fn end(self) -> Result<()> { + self.output.try_push(3).unwrap(); + Ok(()) + } +} + +// Structs are like maps in which the keys are constrained to be compile-time +// constant strings. +impl<'a> ser::SerializeStruct for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(&mut **self) + } + + fn end(self) -> Result<()> { + self.output.try_push(3).unwrap(); + Ok(()) + } +} + +// Similar to `SerializeTupleVariant`, here the `end` method is responsible for +// closing both of the curly braces opened by `serialize_struct_variant`. +impl<'a> ser::SerializeStructVariant for &'a mut Serializer { + type Ok = (); + type Error = Error; + + fn serialize_field(&mut self, _key: &'static str, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + unimplemented!(); + } + + fn end(self) -> Result<()> { + unimplemented!(); + } +} diff --git a/samples/rust/rust_serde.rs b/samples/rust/rust_serde.rs new file mode 100644 index 00000000000000..3d4957dd082209 --- /dev/null +++ b/samples/rust/rust_serde.rs @@ -0,0 +1,73 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Rust `serde` sample. +//! +//! It uses a data format from the `kernel` crate, as well as defining +//! one here ("local"). Then it uses both on a type that uses `serve_derive`. + +use kernel::prelude::*; +use serde_derive::{Deserialize, Serialize}; + +module! { + type: RustSerde, + name: "rust_serde", + author: "Rust for Linux Contributors", + description: "Rust `serde` sample", + license: "GPL", +} + +struct RustSerde; + +pub mod local_data_format { + #![allow(missing_docs)] + + mod de; + mod error; + mod ser; + + pub use de::{from_bytes, Deserializer}; + pub use error::{Error, Result}; + pub use ser::{to_vec, Serializer}; +} + +#[derive(Serialize, Deserialize, Debug)] +struct S { + a: (), + b: bool, + c: bool, + d: (), +} + +impl kernel::Module for RustSerde { + fn init(_module: &'static ThisModule) -> Result { + pr_info!("Rust serde sample (init)\n"); + + let original = S { + a: (), + b: false, + c: true, + d: (), + }; + crate::pr_info!(" original = {:?}", original); + + let serialized = kernel::test_serde::to_vec(&original).unwrap(); + crate::pr_info!(" serialized = {:?}", serialized); + + let deserialized: S = kernel::test_serde::from_bytes(&serialized).unwrap(); + crate::pr_info!(" deserialized = {:?}", deserialized); + + let serialized = local_data_format::to_vec(&deserialized).unwrap(); + crate::pr_info!(" serialized (local) = {:?}", serialized); + + let deserialized: S = local_data_format::from_bytes(&serialized).unwrap(); + crate::pr_info!("deserialized (local) = {:?}", deserialized); + + Ok(RustSerde) + } +} + +impl Drop for RustSerde { + fn drop(&mut self) { + pr_info!("Rust serde sample (exit)\n"); + } +} diff --git a/scripts/Makefile.build b/scripts/Makefile.build index 9f94fc83f08652..c121637c2ea145 100644 --- a/scripts/Makefile.build +++ b/scripts/Makefile.build @@ -284,7 +284,7 @@ rust_common_cmd = \ -Zallow-features=$(rust_allowed_features) \ -Zcrate-attr=no_std \ -Zcrate-attr='feature($(rust_allowed_features))' \ - --extern alloc --extern kernel \ + --extern alloc --extern kernel --extern serde --extern serde_derive \ --crate-type rlib -L $(objtree)/rust/ \ --crate-name $(basename $(notdir $@)) \ --emit=dep-info=$(depfile)