Fix for cases where commands output UTF8 in Windows.

pre_commit/util.py

@@ -93,6 +93,15 @@ def cmd_output_b(
 ) -> tuple[int, bytes, bytes | None]:
     _setdefault_kwargs(kwargs)
 
+    if sys.platform == 'win32':
+        # In windows, pipes use CP1252 by default, and you'll get an
+        # exception if the command being run outputs unicode. So we
+        # set the pipe encoding to utf-8 instead.
+        #
+        # See https://stackoverflow.com/a/74607949/149506
+        import os
+        os.environ['PYTHONIOENCODING'] = 'utf-8'
+
     try:
         cmd = parse_shebang.normalize_cmd(cmd, env=kwargs.get('env'))
     except parse_shebang.ExecutableNotFoundError as e:

tests/util_test.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 import os.path
 import stat
 import subprocess
+import sys
 
 import pytest
 
@@ -106,3 +107,12 @@ def test_rmtree_read_only_directories(tmpdir):
     tmpdir.join('x/y/z').chmod(mode_no_w)
     tmpdir.join('x/y/z').chmod(mode_no_w)
     rmtree(str(tmpdir.join('x')))
+
+
+@pytest.mark.parametrize('fn', (cmd_output_b, cmd_output_p))
+def test_cmd_output_utf8(fn):
+    """Makes sure `cmd_output_*` works if the command being
+    run outputs UTF8 characters."""
+    ret, out, _ = fn(f'{sys.executable}', '-c', 'print("❤")')
+    assert ret == 0
+    assert out.strip().decode() == '❤'