From 0ed646ed0912f709e555a1e90b5457cb1d251dd7 Mon Sep 17 00:00:00 2001
From: Anthony Sottile <asottile@umich.edu>
Date: Tue, 15 Jun 2021 08:32:44 -0700
Subject: [PATCH] read legacy hooks in an encoding-agnostic way

---
 pre_commit/commands/install_uninstall.py | 14 +++++++-------
 tests/commands/install_uninstall_test.py | 17 +++++++++++++++--
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/pre_commit/commands/install_uninstall.py b/pre_commit/commands/install_uninstall.py
index 684b5980..73c8d605 100644
--- a/pre_commit/commands/install_uninstall.py
+++ b/pre_commit/commands/install_uninstall.py
@@ -21,13 +21,13 @@ logger = logging.getLogger(__name__)
 
 # This is used to identify the hook file we install
 PRIOR_HASHES = (
-    '4d9958c90bc262f47553e2c073f14cfe',
-    'd8ee923c46731b42cd95cc869add4062',
-    '49fd668cb42069aa1b6048464be5d395',
-    '79f09a650522a87b0da915d0d983b2de',
-    'e358c9dae00eac5d06b38dfdb1e33a8c',
+    b'4d9958c90bc262f47553e2c073f14cfe',
+    b'd8ee923c46731b42cd95cc869add4062',
+    b'49fd668cb42069aa1b6048464be5d395',
+    b'79f09a650522a87b0da915d0d983b2de',
+    b'e358c9dae00eac5d06b38dfdb1e33a8c',
 )
-CURRENT_HASH = '138fd403232d2ddd5efb44317e38bf03'
+CURRENT_HASH = b'138fd403232d2ddd5efb44317e38bf03'
 TEMPLATE_START = '# start templated\n'
 TEMPLATE_END = '# end templated\n'
 # Homebrew/homebrew-core#35825: be more timid about appropriate `PATH`
@@ -48,7 +48,7 @@ def _hook_paths(
 def is_our_script(filename: str) -> bool:
     if not os.path.exists(filename):  # pragma: win32 no cover (symlink)
         return False
-    with open(filename) as f:
+    with open(filename, 'rb') as f:
         contents = f.read()
     return any(h in contents for h in (CURRENT_HASH,) + PRIOR_HASHES)
 
diff --git a/tests/commands/install_uninstall_test.py b/tests/commands/install_uninstall_test.py
index bd28654f..314b8b96 100644
--- a/tests/commands/install_uninstall_test.py
+++ b/tests/commands/install_uninstall_test.py
@@ -39,7 +39,7 @@ def test_is_script():
 
 def test_is_previous_pre_commit(tmpdir):
     f = tmpdir.join('foo')
-    f.write(f'{PRIOR_HASHES[0]}\n')
+    f.write(f'{PRIOR_HASHES[0].decode()}\n')
     assert is_our_script(f.strpath)
 
 
@@ -390,6 +390,19 @@ def test_install_existing_hook_no_overwrite_idempotent(tempdir_factory, store):
         NORMAL_PRE_COMMIT_RUN.assert_matches(output[len('legacy hook\n'):])
 
 
+def test_install_with_existing_non_utf8_script(tmpdir, store):
+    cmd_output('git', 'init', str(tmpdir))
+    tmpdir.join('.git/hooks').ensure_dir()
+    tmpdir.join('.git/hooks/pre-commit').write_binary(
+        b'#!/usr/bin/env bash\n'
+        b'# garbage: \xa0\xef\x12\xf2\n'
+        b'echo legacy hook\n',
+    )
+
+    with tmpdir.as_cwd():
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+
 FAIL_OLD_HOOK = re_assert.Matches(
     r'fail!\n'
     r'\[INFO\] Initializing environment for .+\.\n'
@@ -460,7 +473,7 @@ def test_replace_old_commit_script(tempdir_factory, store):
         # Install a script that looks like our old script
         pre_commit_contents = resource_text('hook-tmpl')
         new_contents = pre_commit_contents.replace(
-            CURRENT_HASH, PRIOR_HASHES[-1],
+            CURRENT_HASH.decode(), PRIOR_HASHES[-1].decode(),
         )
 
         os.makedirs(os.path.join(path, '.git/hooks'), exist_ok=True)