ENH: Synchronize XML (#207)

* ENH: Synchronize XML

Synchronize to and read XML to upstream

* TST: Add tests for read_xml and to_xml

* MAINT: Cap lxml

* ENH: Improve specificity of types

Remove erroneous Nones

Author: bashtage

pandas-stubs/core/frame.pyi

@@ -69,6 +69,7 @@ from pandas._typing import (
     MaskType,
     MergeHow,
     NaPosition,
+    ReadBuffer,
     Renamer,
     ReplaceMethod,
     Scalar,
@@ -81,6 +82,7 @@ from pandas._typing import (
     T as TType,
     TimestampConvention,
     WriteBuffer,
+    XMLParsers,
     np_ndarray_bool,
     np_ndarray_str,
     num,
@@ -345,6 +347,46 @@ class DataFrame(NDFrame, OpsMixin):
         render_links: _bool = ...,
         encoding: _str | None = ...,
     ) -> _str: ...
+    @overload
+    def to_xml(
+        self,
+        path_or_buffer: FilePath | WriteBuffer[bytes] | WriteBuffer[str],
+        index: bool = ...,
+        root_name: str = ...,
+        row_name: str = ...,
+        na_rep: str | None = ...,
+        attr_cols: list[HashableT] | None = ...,
+        elem_cols: list[HashableT] | None = ...,
+        namespaces: dict[str | None, str] | None = ...,
+        prefix: str | None = ...,
+        encoding: str = ...,
+        xml_declaration: bool = ...,
+        pretty_print: bool = ...,
+        parser: XMLParsers = ...,
+        stylesheet: FilePath | ReadBuffer[str] | ReadBuffer[bytes] | None = ...,
+        compression: CompressionOptions = ...,
+        storage_options: StorageOptions = ...,
+    ) -> None: ...
+    @overload
+    def to_xml(
+        self,
+        path_or_buffer: Literal[None] = ...,
+        index: bool = ...,
+        root_name: str | None = ...,
+        row_name: str | None = ...,
+        na_rep: str | None = ...,
+        attr_cols: list[HashableT] | None = ...,
+        elem_cols: list[HashableT] | None = ...,
+        namespaces: dict[str | None, str] | None = ...,
+        prefix: str | None = ...,
+        encoding: str = ...,
+        xml_declaration: bool | None = ...,
+        pretty_print: bool | None = ...,
+        parser: str | None = ...,
+        stylesheet: FilePath | ReadBuffer[str] | ReadBuffer[bytes] | None = ...,
+        compression: CompressionOptions = ...,
+        storage_options: StorageOptions = ...,
+    ) -> str: ...
     def info(
         self, verbose=..., buf=..., max_cols=..., memory_usage=..., null_counts=...
     ) -> None: ...

pandas-stubs/io/xml.pyi

@@ -25,9 +25,9 @@ def read_xml(
     parse_dates: ParseDatesArg | None = ...,
     # encoding can not be None for lxml and StringIO input
     encoding: str | None = ...,
-    parser: XMLParsers | None = ...,
+    parser: XMLParsers = ...,
     stylesheet: FilePath | ReadBuffer[bytes] | ReadBuffer[str] | None = ...,
     iterparse: dict[str, list[str]] | None = ...,
-    compression: CompressionOptions | None = ...,
-    storage_options: StorageOptions | None = ...,
+    compression: CompressionOptions = ...,
+    storage_options: StorageOptions = ...,
 ) -> DataFrame: ...

pyproject.toml

@@ -47,6 +47,7 @@ pre-commit = ">=2.19.0"
 black = ">=22.6.0"
 isort = ">=5.10.1"
 openpyxl = ">=3.0.10"
+lxml = ">=4.7.1,<4.9.0"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]

tests/test_io.py

@@ -1,8 +1,11 @@
+import io
+
 import pandas as pd
 from pandas import (
     DataFrame,
     read_clipboard,
     read_stata,
+    read_xml,
 )
 from pandas._testing import ensure_clean
 import pytest
@@ -17,6 +20,21 @@
 DF = DataFrame({"a": [1, 2, 3], "b": [0.0, 0.0, 0.0]})
 
 
+def test_xml():
+    with ensure_clean() as path:
+        check(assert_type(DF.to_xml(path), None), type(None))
+        check(assert_type(read_xml(path), DataFrame), DataFrame)
+        with open(path, "rb") as f:
+            check(assert_type(read_xml(f), DataFrame), DataFrame)
+
+
+def test_xml_str():
+    with ensure_clean() as path:
+        check(assert_type(DF.to_xml(), str), str)
+        out: str = DF.to_xml()
+        check(assert_type(read_xml(io.StringIO(out)), DataFrame), DataFrame)
+
+
 def test_read_stata_df():
     with ensure_clean() as path:
         DF.to_stata(path)