Merge pull request #131 from pre-commit/non_utf8_diff

Treat diffs as maybe-not-utf8.

pre_commit/prefixed_command_runner.py

@@ -56,24 +56,24 @@ class PrefixedCommandRunner(object):
         if not os.path.exists(self.prefix_dir):
             self.__makedirs(self.prefix_dir)
 
-    def run(self, cmd, retcode=0, stdin=None, **kwargs):
+    def run(self, cmd, retcode=0, stdin=None, encoding='UTF-8', **kwargs):
         popen_kwargs = {
             'stdin': subprocess.PIPE,
             'stdout': subprocess.PIPE,
             'stderr': subprocess.PIPE,
         }
         if stdin is not None:
-            stdin = stdin.encode('utf-8')
+            stdin = stdin.encode('UTF-8')
 
         popen_kwargs.update(kwargs)
         self._create_path_if_not_exists()
         replaced_cmd = _replace_cmd(cmd, prefix=self.prefix_dir)
         proc = self.__popen(replaced_cmd, **popen_kwargs)
         stdout, stderr = proc.communicate(stdin)
-        if isinstance(stdout, bytes):
+        if encoding is not None:
-            stdout = stdout.decode('UTF-8')
+            stdout = stdout.decode(encoding)
-        if isinstance(stderr, bytes):
+        if encoding is not None:
-            stderr = stderr.decode('UTF-8')
+            stderr = stderr.decode(encoding)
         returncode = proc.returncode
 
         if retcode is not None and retcode != returncode:

pre_commit/staged_files_only.py

@@ -20,19 +20,20 @@ def staged_files_only(cmd_runner):
         cmd_runner - PrefixedCommandRunner
     """
     # Determine if there are unstaged files
-    retcode, diff_stdout, _ = cmd_runner.run(
+    retcode, diff_stdout_binary, _ = cmd_runner.run(
         ['git', 'diff', '--ignore-submodules', '--binary', '--exit-code'],
         retcode=None,
+        encoding=None,
     )
-    if retcode and diff_stdout.strip():
+    if retcode and diff_stdout_binary.strip():
         patch_filename = cmd_runner.path('patch{0}'.format(int(time.time())))
         logger.warning('Unstaged files detected.')
         logger.info(
             'Stashing unstaged files to {0}.'.format(patch_filename),
         )
         # Save the current unstaged changes as a patch
-        with io.open(patch_filename, 'w', encoding='utf-8') as patch_file:
+        with io.open(patch_filename, 'wb') as patch_file:
-            patch_file.write(diff_stdout)
+            patch_file.write(diff_stdout_binary)
 
         # Clear the working directory of unstaged changes
         cmd_runner.run(['git', 'checkout', '--', '.'])

tests/staged_files_only_test.py

@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
@@ -34,9 +35,14 @@ def foo_staged(tmpdir_factory):
         yield auto_namedtuple(path=path, foo_filename=foo_filename)
 
 
-def _test_foo_state(path, foo_contents=FOO_CONTENTS, status='A'):
+def _test_foo_state(
+        path,
+        foo_contents=FOO_CONTENTS,
+        status='A',
+        encoding='UTF-8',
+):
     assert os.path.exists(path.foo_filename)
-    assert io.open(path.foo_filename, encoding='utf-8').read() == foo_contents
+    assert io.open(path.foo_filename, encoding=encoding).read() == foo_contents
     actual_status = get_short_git_status()['foo']
     assert status == actual_status
 
@@ -246,10 +252,22 @@ def test_diff_returns_1_no_diff_though(fake_logging_handler, foo_staged):
 
 def test_stage_utf8_changes(foo_staged, cmd_runner):
     contents = '\u2603'
-    with io.open('foo', 'w', encoding='utf-8') as foo_file:
+    with io.open('foo', 'w', encoding='UTF-8') as foo_file:
         foo_file.write(contents)
 
     _test_foo_state(foo_staged, contents, 'AM')
     with staged_files_only(cmd_runner):
         _test_foo_state(foo_staged)
     _test_foo_state(foo_staged, contents, 'AM')
+
+
+def test_stage_non_utf8_changes(foo_staged, cmd_runner):
+    contents = 'ú'
+    # Produce a latin-1 diff
+    with io.open('foo', 'w', encoding='latin-1') as foo_file:
+        foo_file.write(contents)
+
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')
+    with staged_files_only(cmd_runner):
+        _test_foo_state(foo_staged)
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')