WasmFS: Support reading from stdin (#19699)

This moves the stdin getChar() function into a shared location and then
calls it from the old and new FSes.

The moved code is unchanged except for a closure suppression which
somehow became necessary after the movement (which is odd).

Author: kripken

src/library_fs_shared.js

@@ -102,6 +102,73 @@ mergeInto(LibraryManager.library, {
     if (canWrite) mode |= {{{ cDefs.S_IWUGO }}};
     return mode;
   },
+
+  $FS_stdin_getChar_buffer: [],
+
+  // getChar has 3 particular return values:
+  // a.) the next character represented as an integer
+  // b.) undefined to signal that no data is currently available
+  // c.) null to signal an EOF
+  $FS_stdin_getChar__deps: [
+    '$FS_stdin_getChar_buffer',
+    '$intArrayFromString',
+  ],
+  $FS_stdin_getChar: () => {
+    if (!FS_stdin_getChar_buffer.length) {
+      var result = null;
+#if ENVIRONMENT_MAY_BE_NODE
+      if (ENVIRONMENT_IS_NODE) {
+        // we will read data by chunks of BUFSIZE
+        var BUFSIZE = 256;
+        var buf = Buffer.alloc(BUFSIZE);
+        var bytesRead = 0;
+
+        // For some reason we must suppress a closure warning here, even though
+        // fd definitely exists on process.stdin, and is even the proper way to
+        // get the fd of stdin,
+        // https://github.com/nodejs/help/issues/2136#issuecomment-523649904
+        // This started to happen after moving this logic out of library_tty.js,
+        // so it is related to the surrounding code in some unclear manner.
+        /** @suppress {missingProperties} */
+        var fd = process.stdin.fd;
+
+        try {
+          bytesRead = fs.readSync(fd, buf, 0, BUFSIZE, -1);
+        } catch(e) {
+          // Cross-platform differences: on Windows, reading EOF throws an exception, but on other OSes,
+          // reading EOF returns 0. Uniformize behavior by treating the EOF exception to return 0.
+          if (e.toString().includes('EOF')) bytesRead = 0;
+          else throw e;
+        }
+
+        if (bytesRead > 0) {
+          result = buf.slice(0, bytesRead).toString('utf-8');
+        } else {
+          result = null;
+        }
+      } else
+#endif
+      if (typeof window != 'undefined' &&
+        typeof window.prompt == 'function') {
+        // Browser.
+        result = window.prompt('Input: ');  // returns null on cancel
+        if (result !== null) {
+          result += '\n';
+        }
+      } else if (typeof readline == 'function') {
+        // Command line.
+        result = readline();
+        if (result !== null) {
+          result += '\n';
+        }
+      }
+      if (!result) {
+        return null;
+      }
+      FS_stdin_getChar_buffer = intArrayFromString(result, true);
+    }
+    return FS_stdin_getChar_buffer.shift();
+  },
 });
 
 // Normally only the FS things that the compiler sees are needed are included.

src/library_sigs.js

@@ -414,6 +414,7 @@ sigs = {
   _wasmfs_opfs_set_size_access__sig: 'vpijp',
   _wasmfs_opfs_set_size_file__sig: 'vpijp',
   _wasmfs_opfs_write_access__sig: 'iipii',
+  _wasmfs_stdin_get_char__sig: 'i',
   _wasmfs_thread_utils_heartbeat__sig: 'vp',
   abort__sig: 'v',
   alBuffer3f__sig: 'viifff',

src/library_tty.js

@@ -5,7 +5,11 @@
  */
 
 mergeInto(LibraryManager.library, {
-  $TTY__deps: ['$FS', '$intArrayFromString', '$UTF8ArrayToString'],
+  $TTY__deps: [
+    '$FS',
+    '$UTF8ArrayToString',
+    '$FS_stdin_getChar'
+  ],
 #if !MINIMAL_RUNTIME
   $TTY__postset: function() {
     addAtInit('TTY.init();');
@@ -97,56 +101,8 @@ mergeInto(LibraryManager.library, {
       }
     },
     default_tty_ops: {
-      // get_char has 3 particular return values:
-      // a.) the next character represented as an integer
-      // b.) undefined to signal that no data is currently available
-      // c.) null to signal an EOF
       get_char: function(tty) {
-        if (!tty.input.length) {
-          var result = null;
-#if ENVIRONMENT_MAY_BE_NODE
-          if (ENVIRONMENT_IS_NODE) {
-            // we will read data by chunks of BUFSIZE
-            var BUFSIZE = 256;
-            var buf = Buffer.alloc(BUFSIZE);
-            var bytesRead = 0;
-
-            try {
-              bytesRead = fs.readSync(process.stdin.fd, buf, 0, BUFSIZE, -1);
-            } catch(e) {
-              // Cross-platform differences: on Windows, reading EOF throws an exception, but on other OSes,
-              // reading EOF returns 0. Uniformize behavior by treating the EOF exception to return 0.
-              if (e.toString().includes('EOF')) bytesRead = 0;
-              else throw e;
-            }
-
-            if (bytesRead > 0) {
-              result = buf.slice(0, bytesRead).toString('utf-8');
-            } else {
-              result = null;
-            }
-          } else
-#endif
-          if (typeof window != 'undefined' &&
-            typeof window.prompt == 'function') {
-            // Browser.
-            result = window.prompt('Input: ');  // returns null on cancel
-            if (result !== null) {
-              result += '\n';
-            }
-          } else if (typeof readline == 'function') {
-            // Command line.
-            result = readline();
-            if (result !== null) {
-              result += '\n';
-            }
-          }
-          if (!result) {
-            return null;
-          }
-          tty.input = intArrayFromString(result, true);
-        }
-        return tty.input.shift();
+        return FS_stdin_getChar();
       },
       put_char: function(tty, val) {
         if (val === null || val === {{{ charCode('\n') }}}) {

src/library_wasmfs.js

@@ -427,4 +427,14 @@ FS.createPreloadedFile = FS_createPreloadedFile;
         }
       }, 50);
   },
+
+  _wasmfs_stdin_get_char__deps: ['$FS_stdin_getChar'],
+  _wasmfs_stdin_get_char: () => {
+    // Return the read character, or -1 to indicate EOF.
+    var c = FS_stdin_getChar();
+    if (typeof c === 'number') {
+      return c;
+    }
+    return -1;
+  }
 });

system/lib/standalone/standalone.c

@@ -217,6 +217,26 @@ void emscripten_out(const char* text) { wasi_writeln(1, text); }
 
 void emscripten_err(const char* text) { wasi_writeln(2, text); }
 
+__attribute__((import_module("wasi_snapshot_preview1"),
+               import_name("fd_read"))) __wasi_errno_t
+imported__wasi_fd_read(__wasi_fd_t fd,
+                        const __wasi_ciovec_t* iovs,
+                        size_t iovs_len,
+                        __wasi_size_t* nread);
+
+int _wasmfs_stdin_get_char(void) {
+  char c;
+  struct __wasi_ciovec_t iov;
+  iov.buf = (uint8_t*)&c;
+  iov.buf_len = 1;
+  __wasi_size_t nread;
+  imported__wasi_fd_read(0, &iov, 1, &nread);
+  if (nread == 0) {
+    return -1;
+  }
+  return c;
+}
+
 // In the non-standalone build we define this helper function in JS to avoid
 // signture mismatch issues.
 // See: https://github.com/emscripten-core/posixtestsuite/issues/6

system/lib/wasmfs/special_files.cpp

@@ -12,6 +12,7 @@
 #include <wasi/api.h>
 
 #include "special_files.h"
+#include "wasmfs_internal.h"
 
 namespace wasmfs::SpecialFiles {
 
@@ -45,8 +46,15 @@ class StdinFile : public DataFile {
   }
 
   ssize_t read(uint8_t* buf, size_t len, off_t offset) override {
-    // TODO: Implement reading from stdin.
-    abort();
+    for (size_t i = 0; i < len; i++) {
+      auto c = _wasmfs_stdin_get_char();
+      if (c < 0) {
+        // No more input can be read, return what we did read.
+        return i;
+      }
+      buf[i] = c;
+    }
+    return len;
   };
 
   int flush() override { return 0; }

system/lib/wasmfs/wasmfs_internal.h

@@ -18,6 +18,9 @@ void _wasmfs_get_preloaded_child_path(int index, char* childName);
 size_t _wasmfs_get_preloaded_file_size(uint32_t index);
 void _wasmfs_copy_preloaded_file_data(uint32_t index, uint8_t* data);
 
+// Returns the next character from stdin, or -1 on EOF.
+int _wasmfs_stdin_get_char(void);
+
 #ifdef __cplusplus
 }
 #endif

test/other/metadce/test_metadce_cxx_wasmfs.imports

@@ -8,6 +8,7 @@ env._wasmfs_get_preloaded_file_mode
 env._wasmfs_get_preloaded_file_size
 env._wasmfs_get_preloaded_parent_path
 env._wasmfs_get_preloaded_path_name
+env._wasmfs_stdin_get_char
 env.abort
 env.emscripten_console_error
 env.emscripten_date_now

test/other/metadce/test_metadce_cxx_wasmfs.jssize

@@ -1 +1 @@
-12355
+12922

test/other/metadce/test_metadce_cxx_wasmfs.sent

@@ -8,6 +8,7 @@ _wasmfs_get_preloaded_file_mode
 _wasmfs_get_preloaded_file_size
 _wasmfs_get_preloaded_parent_path
 _wasmfs_get_preloaded_path_name
+_wasmfs_stdin_get_char
 abort
 emscripten_console_error
 emscripten_date_now

test/other/metadce/test_metadce_cxx_wasmfs.size

@@ -1 +1 @@
-164439
+164531

test/other/metadce/test_metadce_files_wasmfs.exports

@@ -1,4 +1,4 @@
-r
 s
 t
 u
+v

test/other/metadce/test_metadce_files_wasmfs.imports

@@ -15,3 +15,4 @@ a.n
 a.o
 a.p
 a.q
+a.r

test/other/metadce/test_metadce_files_wasmfs.jssize

@@ -1 +1 @@
-6750
+7314

test/other/metadce/test_metadce_files_wasmfs.sent

@@ -15,3 +15,4 @@ n
 o
 p
 q
+r

test/other/metadce/test_metadce_files_wasmfs.size

@@ -1 +1 @@
-52397
+52463

test/test_other.py

@@ -1480,6 +1480,7 @@ def test_export_all_and_exported_functions(self):
     self.emcc('lib.c', ['-sEXPORTED_FUNCTIONS=_libfunc2', '-sEXPORT_ALL', '--pre-js', 'main.js'], output_filename='a.out.js')
     self.assertContained('libfunc\n', self.run_js('a.out.js'))
 
+  @also_with_wasmfs
   @crossplatform
   def test_stdin(self):
     def run_test():