Merge pull request #509 from pawamoy/patch-1

Fix parsing of git output with unusual characters

Author: asottile

pre_commit_hooks/check_executables_have_shebangs.py

@@ -12,6 +12,14 @@
 EXECUTABLE_VALUES = frozenset(('1', '3', '5', '7'))
 
 
+def zsplit(s: str) -> List[str]:
+    s = s.strip('\0')
+    if s:
+        return s.split('\0')
+    else:
+        return []
+
+
 def check_executables(paths: List[str]) -> int:
     if sys.platform == 'win32':  # pragma: win32 cover
         return _check_git_filemode(paths)
@@ -26,9 +34,9 @@ def check_executables(paths: List[str]) -> int:
 
 
 def _check_git_filemode(paths: Sequence[str]) -> int:
-    outs = cmd_output('git', 'ls-files', '--stage', '--', *paths)
+    outs = cmd_output('git', 'ls-files', '-z', '--stage', '--', *paths)
     seen: Set[str] = set()
-    for out in outs.splitlines():
+    for out in zsplit(outs):
         metadata, path = out.split('\t')
         tagmode = metadata.split(' ', 1)[0]
 

tests/check_executables_have_shebangs_test.py

@@ -73,6 +73,21 @@ def test_check_git_filemode_passing(tmpdir):
         assert check_executables_have_shebangs._check_git_filemode(files) == 0
 
 
+def test_check_git_filemode_passing_unusual_characters(tmpdir):
+    with tmpdir.as_cwd():
+        cmd_output('git', 'init', '.')
+
+        f = tmpdir.join('mañana.txt')
+        f.write('#!/usr/bin/env bash')
+        f_path = str(f)
+        cmd_output('chmod', '+x', f_path)
+        cmd_output('git', 'add', f_path)
+        cmd_output('git', 'update-index', '--chmod=+x', f_path)
+
+        files = (f_path,)
+        assert check_executables_have_shebangs._check_git_filemode(files) == 0
+
+
 def test_check_git_filemode_failing(tmpdir):
     with tmpdir.as_cwd():
         cmd_output('git', 'init', '.')
@@ -87,6 +102,16 @@ def test_check_git_filemode_failing(tmpdir):
         assert check_executables_have_shebangs._check_git_filemode(files) == 1
 
 
+@pytest.mark.parametrize('out', ('\0f1\0f2\0', '\0f1\0f2', 'f1\0f2\0'))
+def test_check_zsplits_correctly(out):
+    assert check_executables_have_shebangs.zsplit(out) == ['f1', 'f2']
+
+
+@pytest.mark.parametrize('out', ('\0\0', '\0', ''))
+def test_check_zsplit_returns_empty(out):
+    assert check_executables_have_shebangs.zsplit(out) == []
+
+
 @pytest.mark.parametrize(
     ('content', 'mode', 'expected'),
     (