Default to UTF-8 encoding in open() calls

pre_commit/commands/install_uninstall.py

@@ -1,11 +1,11 @@
 from __future__ import print_function
 from __future__ import unicode_literals
 
-import io
 import os.path
 import sys
 
 from pre_commit import output
+from pre_commit import util
 from pre_commit.util import make_executable
 from pre_commit.util import mkdirp
 from pre_commit.util import resource_filename
@@ -25,7 +25,7 @@ CURRENT_HASH = '138fd403232d2ddd5efb44317e38bf03'
 def is_our_script(filename):
     if not os.path.exists(filename):
         return False
-    contents = io.open(filename).read()
+    contents = util.open(filename).read()
     return any(h in contents for h in (CURRENT_HASH,) + PRIOR_HASHES)
 
 
@@ -54,15 +54,15 @@ def install(
             )
         )
 
-    with io.open(hook_path, 'w') as pre_commit_file_obj:
+    with util.open(hook_path, 'w') as pre_commit_file_obj:
         if hook_type == 'pre-push':
-            with io.open(resource_filename('pre-push-tmpl')) as fp:
+            with util.open(resource_filename('pre-push-tmpl')) as fp:
                 pre_push_contents = fp.read()
         else:
             pre_push_contents = ''
 
         skip_on_missing_conf = 'true' if skip_on_missing_conf else 'false'
-        contents = io.open(resource_filename('hook-tmpl')).read().format(
+        contents = util.open(resource_filename('hook-tmpl')).read().format(
             sys_executable=sys.executable,
             hook_type=hook_type,
             pre_push=pre_push_contents,

pre_commit/languages/ruby.py

@@ -1,11 +1,11 @@
 from __future__ import unicode_literals
 
 import contextlib
-import io
 import os.path
 import shutil
 import tarfile
 
+from pre_commit import util
 from pre_commit.envcontext import envcontext
 from pre_commit.envcontext import Var
 from pre_commit.languages import helpers
@@ -65,7 +65,7 @@ def _install_rbenv(
             tf.extractall(repo_cmd_runner.path(directory, 'plugins'))
 
     activate_path = repo_cmd_runner.path(directory, 'bin', 'activate')
-    with io.open(activate_path, 'w') as activate_file:
+    with util.open(activate_path, 'w') as activate_file:
         # This is similar to how you would install rbenv to your home directory
         # However we do a couple things to make the executables exposed and
         # configure it to work in our directory.

pre_commit/parse_shebang.py

@@ -1,7 +1,6 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
-import io
 import os.path
 import shlex
 import string
@@ -41,7 +40,7 @@ def parse_filename(filename):
     if not os.path.exists(filename) or not os.access(filename, os.X_OK):
         return ()
 
-    with io.open(filename, 'rb') as f:
+    with open(filename, 'rb') as f:
         return parse_bytesio(f)
 
 

pre_commit/repository.py

@@ -1,6 +1,5 @@
 from __future__ import unicode_literals
 
-import io
 import json
 import logging
 import os
@@ -13,6 +12,7 @@ from cached_property import cached_property
 import pre_commit.constants as C
 from pre_commit import five
 from pre_commit import git
+from pre_commit import util
 from pre_commit.clientlib import is_local_repo
 from pre_commit.clientlib import MANIFEST_HOOK_DICT
 from pre_commit.languages.all import languages
@@ -41,13 +41,13 @@ def _read_installed_state(cmd_runner, venv):
     if not os.path.exists(filename):
         return None
     else:
-        return json.loads(io.open(filename).read())
+        return json.loads(util.open(filename).read())
 
 
 def _write_installed_state(cmd_runner, venv, state):
     state_filename = _state_filename(cmd_runner, venv)
     staging = state_filename + 'staging'
-    with io.open(staging, 'w') as state_file:
+    with util.open(staging, 'w') as state_file:
         state_file.write(five.to_text(json.dumps(state)))
     # Move the file into place atomically to indicate we've installed
     os.rename(staging, state_filename)

pre_commit/schema.py

@@ -3,13 +3,14 @@ from __future__ import unicode_literals
 
 import collections
 import contextlib
-import io
 import os.path
 import re
 import sys
 
 import six
 
+from pre_commit import util
+
 
 class ValidationError(ValueError):
     def __init__(self, error_msg, ctx=None):
@@ -266,7 +267,7 @@ def load_from_filename(filename, schema, load_strategy, exc_tp):
         if not os.path.exists(filename):
             raise ValidationError('{} does not exist'.format(filename))
 
-        with io.open(filename) as f:
+        with util.open(filename) as f:
             contents = f.read()
 
         with validate_context('File {}'.format(filename)):

pre_commit/staged_files_only.py

@@ -1,7 +1,6 @@
 from __future__ import unicode_literals
 
 import contextlib
-import io
 import logging
 import time
 
@@ -35,7 +34,7 @@ def staged_files_only(cmd_runner):
             'Stashing unstaged files to {}.'.format(patch_filename),
         )
         # Save the current unstaged changes as a patch
-        with io.open(patch_filename, 'wb') as patch_file:
+        with open(patch_filename, 'wb') as patch_file:
             patch_file.write(diff_stdout_binary)
 
         # Clear the working directory of unstaged changes

pre_commit/store.py

@@ -1,7 +1,6 @@
 from __future__ import unicode_literals
 
 import contextlib
-import io
 import logging
 import os.path
 import sqlite3
@@ -10,6 +9,7 @@ import tempfile
 from cached_property import cached_property
 
 import pre_commit.constants as C
+from pre_commit import util
 from pre_commit.prefixed_command_runner import PrefixedCommandRunner
 from pre_commit.util import clean_path_on_failure
 from pre_commit.util import cmd_output
@@ -46,7 +46,7 @@ class Store(object):
         self.__created = False
 
     def _write_readme(self):
-        with io.open(os.path.join(self.directory, 'README'), 'w') as readme:
+        with util.open(os.path.join(self.directory, 'README'), 'w') as readme:
             readme.write(
                 'This directory is maintained by the pre-commit project.\n'
                 'Learn more: https://github.com/pre-commit/pre-commit\n'

pre_commit/util.py

@@ -3,6 +3,7 @@ from __future__ import unicode_literals
 import contextlib
 import errno
 import functools
+import io
 import os.path
 import shutil
 import stat
@@ -16,6 +17,14 @@ from pre_commit import five
 from pre_commit import parse_shebang
 
 
+def open(*args, **kwargs):
+    """Some strange interaction with python2, osx, and vscode makes default
+    encoding detection go wrong.  See #519
+    """
+    kwargs.setdefault('encoding', 'UTF-8')
+    return io.open(*args, **kwargs)
+
+
 @contextlib.contextmanager
 def cwd(path):
     original_cwd = os.getcwd()