Adding upstream version 2.9.3.

Signed-off-by: Daniel Baumann <daniel@debian.org>

.github/FUNDING.yml

@@ -0,0 +1,2 @@
+github: asottile
+open_collective: pre-commit

.pre-commit-config.yaml

@@ -21,16 +21,16 @@ repos:
     hooks:
     -   id: autopep8
 -   repo: https://github.com/pre-commit/pre-commit
-    rev: v2.7.1
+    rev: v2.9.3
     hooks:
     -   id: validate_manifest
 -   repo: https://github.com/asottile/pyupgrade
-    rev: v2.7.3
+    rev: v2.7.4
     hooks:
     -   id: pyupgrade
         args: [--py36-plus]
 -   repo: https://github.com/asottile/reorder_python_imports
-    rev: v2.3.5
+    rev: v2.3.6
     hooks:
     -   id: reorder-python-imports
         args: [--py3-plus]

CHANGELOG.md

@@ -1,3 +1,65 @@
+2.9.3 - 2020-12-07
+==================
+
+### Fixes
+- Fix crash on cygwin mismatch check outside of a git directory
+    - #1721 PR by @asottile.
+    - #1720 issue by @chronoB.
+- Fix cleanup code on docker volumes for go
+    - #1725 PR by @fsouza.
+- Fix working directory detection on SUBST drives on windows
+    - #1727 PR by mrogaski.
+    - #1610 issue by @jcameron73.
+
+2.9.2 - 2020-11-25
+==================
+
+### Fixes
+- Fix default value for `types_or` so `symlink` and `directory` can be matched
+    - #1716 PR by @asottile.
+    - #1718 issue by @CodeBleu.
+
+2.9.1 - 2020-11-25
+==================
+
+### Fixes
+- Improve error message for "hook goes missing"
+    - #1709 PR by @paulhfischer.
+    - #1708 issue by @theod07.
+- Add warning for `/*` in `files` / `exclude` regexes
+    - #1707 PR by @paulhfischer.
+    - #1702 issue by @asottile.
+- Fix `healthy()` check for `language: python` on windows when the base
+  executable has non-ascii characters.
+    - #1713 PR by @asottile.
+    - #1711 issue by @Najiva.
+
+2.9.0 - 2020-11-21
+==================
+
+### Features
+- Add `types_or` which allows matching multiple disparate `types` in a hook
+    - #1677 by @MarcoGorelli.
+    - #607 by @asottile.
+- Add Github Sponsors / Open Collective links
+    - https://github.com/sponsors/asottile
+    - https://opencollective.com/pre-commit
+
+### Fixes
+- Improve cleanup for `language: dotnet`
+    - #1678 by @rkm.
+- Fix "xargs" when running windows batch files
+    - #1686 PR by @asottile.
+    - #1604 issue by @apietrzak.
+    - #1604 issue by @ufwtlsb.
+- Fix conflict with external `rbenv` and `language_version: default`
+    - #1700 PR by @asottile.
+    - #1699 issue by @abuxton.
+- Improve performance of `git status` / `git diff` commands by ignoring
+  submodules
+    - #1704 PR by @Vynce.
+    - #1701 issue by @Vynce.
+
 2.8.2 - 2020-10-30
 ==================
 

pre_commit/clientlib.py

@@ -61,6 +61,7 @@ MANIFEST_HOOK_DICT = cfgv.Map(
     cfgv.Optional('files', check_string_regex, ''),
     cfgv.Optional('exclude', check_string_regex, '^$'),
     cfgv.Optional('types', cfgv.check_array(check_type_tag), ['file']),
+    cfgv.Optional('types_or', cfgv.check_array(check_type_tag), []),
     cfgv.Optional('exclude_types', cfgv.check_array(check_type_tag), []),
 
     cfgv.Optional(
@@ -111,6 +112,18 @@ LOCAL = 'local'
 META = 'meta'
 
 
+class OptionalSensibleRegex(cfgv.OptionalNoDefault):
+    def check(self, dct: Dict[str, Any]) -> None:
+        super().check(dct)
+
+        if '/*' in dct.get(self.key, ''):
+            logger.warning(
+                f'The {self.key!r} field in hook {dct.get("id")!r} is a '
+                f"regex, not a glob -- matching '/*' probably isn't what you "
+                f'want here',
+            )
+
+
 class MigrateShaToRev:
     key = 'rev'
 
@@ -226,6 +239,8 @@ CONFIG_HOOK_DICT = cfgv.Map(
         for item in MANIFEST_HOOK_DICT.items
         if item.key != 'id'
     ),
+    OptionalSensibleRegex('files', cfgv.check_string),
+    OptionalSensibleRegex('exclude', cfgv.check_string),
 )
 CONFIG_REPO_DICT = cfgv.Map(
     'Repository', 'repo',

pre_commit/commands/autoupdate.py

@@ -79,14 +79,12 @@ def _check_hooks_still_exist_at_rev(
     hooks_missing = hooks - {hook['id'] for hook in manifest}
     if hooks_missing:
         raise RepositoryCannotBeUpdatedError(
-            f'Cannot update because the tip of HEAD is missing these hooks:\n'
-            f'{", ".join(sorted(hooks_missing))}',
+            f'Cannot update because the update target is missing these '
+            f'hooks:\n{", ".join(sorted(hooks_missing))}',
         )
 
 
-REV_LINE_RE = re.compile(
-    r'^(\s+)rev:(\s*)([\'"]?)([^\s#]+)(.*)(\r?\n)$', re.DOTALL,
-)
+REV_LINE_RE = re.compile(r'^(\s+)rev:(\s*)([\'"]?)([^\s#]+)(.*)(\r?\n)$')
 
 
 def _original_lines(

pre_commit/commands/migrate_config.py

@@ -1,4 +1,5 @@
 import re
+import textwrap
 
 import yaml
 
@@ -6,27 +7,22 @@ from pre_commit.clientlib import load_config
 from pre_commit.util import yaml_load
 
 
-def _indent(s: str) -> str:
-    lines = s.splitlines(True)
-    return ''.join(' ' * 4 + line if line.strip() else line for line in lines)
-
-
 def _is_header_line(line: str) -> bool:
     return line.startswith(('#', '---')) or not line.strip()
 
 
 def _migrate_map(contents: str) -> str:
-    # Find the first non-header line
-    lines = contents.splitlines(True)
-    i = 0
-    # Only loop on non empty configuration file
-    while i < len(lines) and _is_header_line(lines[i]):
-        i += 1
-
-    header = ''.join(lines[:i])
-    rest = ''.join(lines[i:])
-
     if isinstance(yaml_load(contents), list):
+        # Find the first non-header line
+        lines = contents.splitlines(True)
+        i = 0
+        # Only loop on non empty configuration file
+        while i < len(lines) and _is_header_line(lines[i]):
+            i += 1
+
+        header = ''.join(lines[:i])
+        rest = ''.join(lines[i:])
+
         # If they are using the "default" flow style of yaml, this operation
         # will yield a valid configuration
         try:
@@ -34,7 +30,7 @@ def _migrate_map(contents: str) -> str:
             yaml_load(trial_contents)
             contents = trial_contents
         except yaml.YAMLError:
-            contents = f'{header}repos:\n{_indent(rest)}'
+            contents = f'{header}repos:\n{textwrap.indent(rest, " " * 4)}'
 
     return contents
 

pre_commit/commands/run.py

@@ -83,20 +83,32 @@ class Classifier:
             self,
             names: Sequence[str],
             types: Collection[str],
+            types_or: Collection[str],
             exclude_types: Collection[str],
     ) -> List[str]:
-        types, exclude_types = frozenset(types), frozenset(exclude_types)
+        types = frozenset(types)
+        types_or = frozenset(types_or)
+        exclude_types = frozenset(exclude_types)
         ret = []
         for filename in names:
             tags = self._types_for_file(filename)
-            if tags >= types and not tags & exclude_types:
+            if (
+                    tags >= types and
+                    (not types_or or tags & types_or) and
+                    not tags & exclude_types
+            ):
                 ret.append(filename)
         return ret
 
     def filenames_for_hook(self, hook: Hook) -> Tuple[str, ...]:
         names = self.filenames
         names = filter_by_include_exclude(names, hook.files, hook.exclude)
-        names = self.by_types(names, hook.types, hook.exclude_types)
+        names = self.by_types(
+            names,
+            hook.types,
+            hook.types_or,
+            hook.exclude_types,
+        )
         return tuple(names)
 
     @classmethod
@@ -250,7 +262,9 @@ def _all_filenames(args: argparse.Namespace) -> Collection[str]:
 
 
 def _get_diff() -> bytes:
-    _, out, _ = cmd_output_b('git', 'diff', '--no-ext-diff', retcode=None)
+    _, out, _ = cmd_output_b(
+        'git', 'diff', '--no-ext-diff', '--ignore-submodules', retcode=None,
+    )
     return out
 
 

pre_commit/envcontext.py

@@ -8,11 +8,7 @@ from typing import Optional
 from typing import Tuple
 from typing import Union
 
-
-class _Unset(enum.Enum):
-    UNSET = 1
-
-
+_Unset = enum.Enum('_Unset', 'UNSET')
 UNSET = _Unset.UNSET
 
 

pre_commit/git.py

@@ -47,21 +47,26 @@ def no_git_env(
 
 
 def get_root() -> str:
+    # Git 2.25 introduced a change to "rev-parse --show-toplevel" that exposed
+    # underlying volumes for Windows drives mapped with SUBST.  We use
+    # "rev-parse --show-cdup" to get the appropriate path, but must perform
+    # an extra check to see if we are in the .git directory.
     try:
-        root = cmd_output('git', 'rev-parse', '--show-toplevel')[1].strip()
+        root = os.path.realpath(
+            cmd_output('git', 'rev-parse', '--show-cdup')[1].strip(),
+        )
+        git_dir = os.path.realpath(get_git_dir())
     except CalledProcessError:
         raise FatalError(
             'git failed. Is it installed, and are you in a Git repository '
             'directory?',
         )
-    else:
-        if root == '':  # pragma: no cover (old git)
-            raise FatalError(
-                'git toplevel unexpectedly empty! make sure you are not '
-                'inside the `.git` directory of your repository.',
-            )
-        else:
-            return root
+    if os.path.commonpath((root, git_dir)) == git_dir:
+        raise FatalError(
+            'git toplevel unexpectedly empty! make sure you are not '
+            'inside the `.git` directory of your repository.',
+        )
+    return root
 
 
 def get_git_dir(git_root: str = '.') -> str:
@@ -130,7 +135,9 @@ def get_staged_files(cwd: Optional[str] = None) -> List[str]:
 
 
 def intent_to_add_files() -> List[str]:
-    _, stdout, _ = cmd_output('git', 'status', '--porcelain', '-z')
+    _, stdout, _ = cmd_output(
+        'git', 'status', '--ignore-submodules', '--porcelain', '-z',
+    )
     parts = list(reversed(zsplit(stdout)))
     intent_to_add = []
     while parts:
@@ -199,7 +206,10 @@ def check_for_cygwin_mismatch() -> None:
     """See https://github.com/pre-commit/pre-commit/issues/354"""
     if sys.platform in ('cygwin', 'win32'):  # pragma: no cover (windows)
         is_cygwin_python = sys.platform == 'cygwin'
-        toplevel = get_root()
+        try:
+            toplevel = get_root()
+        except FatalError:  # skip the check if we're not in a git repo
+            return
         is_cygwin_git = toplevel.startswith('/')
 
         if is_cygwin_python ^ is_cygwin_git:

pre_commit/hook.py

@@ -22,6 +22,7 @@ class Hook(NamedTuple):
     files: str
     exclude: str
     types: Sequence[str]
+    types_or: Sequence[str]
     exclude_types: Sequence[str]
     additional_dependencies: Sequence[str]
     args: Sequence[str]

pre_commit/languages/dotnet.py

@@ -12,7 +12,6 @@ from pre_commit.hook import Hook
 from pre_commit.languages import helpers
 from pre_commit.prefix import Prefix
 from pre_commit.util import clean_path_on_failure
-from pre_commit.util import rmtree
 
 ENVIRONMENT_DIR = 'dotnetenv'
 BIN_DIR = 'bin'
@@ -76,9 +75,9 @@ def install_environment(
             ),
         )
 
-        # Cleanup build output
-        for d in ('bin', 'obj', build_dir):
-            rmtree(prefix.path(d))
+        # Clean the git dir, ignoring the environment dir
+        clean_cmd = ('git', 'clean', '-ffxd', '-e', f'{ENVIRONMENT_DIR}-*')
+        helpers.run_setup_cmd(prefix, clean_cmd)
 
 
 def run_hook(

pre_commit/languages/python.py

@@ -36,7 +36,7 @@ def _version_info(exe: str) -> str:
 
 def _read_pyvenv_cfg(filename: str) -> Dict[str, str]:
     ret = {}
-    with open(filename) as f:
+    with open(filename, encoding='UTF-8') as f:
         for line in f:
             try:
                 k, v = line.split('=')

pre_commit/languages/ruby.py

@@ -52,7 +52,6 @@ def get_env_patch(
     else:  # pragma: win32 no cover
         patches += (
             ('RBENV_ROOT', venv),
-            ('RBENV_VERSION', language_version),
             (
                 'PATH', (
                     os.path.join(venv, 'gems', 'bin'), os.pathsep,
@@ -61,6 +60,9 @@ def get_env_patch(
                 ),
             ),
         )
+    if language_version not in {'system', 'default'}:  # pragma: win32 no cover
+        patches += (('RBENV_VERSION', language_version),)
+
     return patches
 
 

pre_commit/meta_hooks/check_useless_excludes.py

@@ -47,8 +47,10 @@ def check_useless_excludes(config_file: str) -> int:
             # the defaults applied during runtime
             hook = apply_defaults(hook, MANIFEST_HOOK_DICT)
             names = classifier.filenames
-            types, exclude_types = hook['types'], hook['exclude_types']
-            names = classifier.by_types(names, types, exclude_types)
+            types = hook['types']
+            types_or = hook['types_or']
+            exclude_types = hook['exclude_types']
+            names = classifier.by_types(names, types, types_or, exclude_types)
             include, exclude = hook['files'], hook['exclude']
             if not exclude_matches_any(names, include, exclude):
                 print(

pre_commit/util.py

@@ -255,7 +255,7 @@ def rmtree(path: str) -> None:
         excvalue = exc[1]
         if (
                 func in (os.rmdir, os.remove, os.unlink) and
-                excvalue.errno == errno.EACCES
+                excvalue.errno in {errno.EACCES, errno.EPERM}
         ):
             for p in (path, os.path.dirname(path)):
                 os.chmod(p, os.stat(p).st_mode | stat.S_IWUSR)

pre_commit/xargs.py

@@ -137,6 +137,16 @@ def xargs(
     except parse_shebang.ExecutableNotFoundError as e:
         return e.to_output()[:2]
 
+    # on windows, batch files have a separate length limit than windows itself
+    if (
+            sys.platform == 'win32' and
+            cmd[0].lower().endswith(('.bat', '.cmd'))
+    ):  # pragma: win32 cover
+        # this is implementation details but the command gets translated into
+        # full/path/to/cmd.exe /c *cmd
+        cmd_exe = parse_shebang.find_executable('cmd.exe')
+        _max_length = 8192 - len(cmd_exe) - len(' /c ')
+
     partitions = partition(cmd, varargs, target_concurrency, _max_length)
 
     def run_cmd_partition(

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = pre_commit
-version = 2.8.2
+version = 2.9.3
 description = A framework for managing and maintaining multi-language pre-commit hooks.
 long_description = file: README.md
 long_description_content_type = text/markdown

testing/resources/exclude_types_repo/bin/hook.sh

@@ -1,3 +1,3 @@
 #!/usr/bin/env bash
-echo $@
+echo "$@"
 exit 1

testing/resources/failing_hook_repo/bin/hook.sh

@@ -1,4 +1,4 @@
 #!/usr/bin/env bash
 echo 'Fail'
-echo $@
+echo "$@"
 exit 1

testing/resources/modified_file_returns_zero_repo/bin/hook2.sh

@@ -1,2 +1,2 @@
 #!/usr/bin/env bash
-echo $@
+echo "$@"

testing/resources/script_hooks_repo/bin/hook.sh

@@ -1,4 +1,4 @@
 #!/usr/bin/env bash
 
-echo $@
+echo "$@"
 echo 'Hello World'

testing/resources/types_or_repo/.pre-commit-hooks.yaml

@@ -0,0 +1,6 @@
+-   id: python-cython-files
+    name: Python and Cython files
+    entry: bin/hook.sh
+    language: script
+    types: [file]
+    types_or: [python, cython]

testing/resources/types_or_repo/bin/hook.sh

@@ -0,0 +1,3 @@
+#!/usr/bin/env bash
+echo "$@"
+exit 1

testing/resources/types_repo/bin/hook.sh

@@ -1,3 +1,3 @@
 #!/usr/bin/env bash
-echo $@
+echo "$@"
 exit 1

testing/zipapp/make

@@ -99,6 +99,9 @@ def main() -> int:
         shebang = '/usr/bin/env python3'
         zipapp.create_archive(tmpdir, filename, interpreter=shebang)
 
+    with open(f'{filename}.sha256sum', 'w') as f:
+        subprocess.check_call(('sha256sum', filename), stdout=f)
+
     return 0
 
 

tests/clientlib_test.py

@@ -166,6 +166,23 @@ def test_validate_warn_on_unknown_keys_at_top_level(tmpdir, caplog):
     ]
 
 
+def test_validate_optional_sensible_regex(caplog):
+    config_obj = {
+        'id': 'flake8',
+        'files': 'dir/*.py',
+    }
+    cfgv.validate(config_obj, CONFIG_HOOK_DICT)
+
+    assert caplog.record_tuples == [
+        (
+            'pre_commit',
+            logging.WARNING,
+            "The 'files' field in hook 'flake8' is a regex, not a glob -- "
+            "matching '/*' probably isn't what you want here",
+        ),
+    ]
+
+
 @pytest.mark.parametrize('fn', (validate_config_main, validate_manifest_main))
 def test_mains_not_ok(tmpdir, fn):
     not_yaml = tmpdir.join('f.notyaml')

tests/commands/autoupdate_test.py

@@ -1,9 +1,12 @@
 import shlex
+from unittest import mock
 
 import pytest
+import yaml
 
 import pre_commit.constants as C
 from pre_commit import git
+from pre_commit import util
 from pre_commit.commands.autoupdate import _check_hooks_still_exist_at_rev
 from pre_commit.commands.autoupdate import autoupdate
 from pre_commit.commands.autoupdate import RepositoryCannotBeUpdatedError
@@ -173,6 +176,11 @@ def test_autoupdate_out_of_date_repo(out_of_date, tmpdir, store):
     assert cfg.read() == fmt.format(out_of_date.path, out_of_date.head_rev)
 
 
+def test_autoupdate_pure_yaml(out_of_date, tmpdir, store):
+    with mock.patch.object(util, 'Dumper', yaml.SafeDumper):
+        test_autoupdate_out_of_date_repo(out_of_date, tmpdir, store)
+
+
 def test_autoupdate_only_one_to_update(up_to_date, out_of_date, tmpdir, store):
     fmt = (
         'repos:\n'

tests/commands/migrate_config_test.py

@@ -2,24 +2,9 @@ import pytest
 
 import pre_commit.constants as C
 from pre_commit.clientlib import InvalidConfigError
-from pre_commit.commands.migrate_config import _indent
 from pre_commit.commands.migrate_config import migrate_config
 
 
-@pytest.mark.parametrize(
-    ('s', 'expected'),
-    (
-        ('', ''),
-        ('a', '    a'),
-        ('foo\nbar', '    foo\n    bar'),
-        ('foo\n\nbar\n', '    foo\n\n    bar\n'),
-        ('\n\n\n', '\n\n\n'),
-    ),
-)
-def test_indent(s, expected):
-    assert _indent(s) == expected
-
-
 def test_migrate_config_normal_format(tmpdir, capsys):
     cfg = tmpdir.join(C.CONFIG_FILE)
     cfg.write(

tests/commands/run_test.py

@@ -219,6 +219,19 @@ def test_types_hook_repository(cap_out, store, tempdir_factory):
         assert b'bar.notpy' not in printed
 
 
+def test_types_or_hook_repository(cap_out, store, tempdir_factory):
+    git_path = make_consuming_repo(tempdir_factory, 'types_or_repo')
+    with cwd(git_path):
+        stage_a_file('bar.notpy')
+        stage_a_file('bar.pxd')
+        stage_a_file('bar.py')
+        ret, printed = _do_run(cap_out, store, git_path, run_opts())
+        assert ret == 1
+        assert b'bar.notpy' not in printed
+        assert b'bar.pxd' in printed
+        assert b'bar.py' in printed
+
+
 def test_exclude_types_hook_repository(cap_out, store, tempdir_factory):
     git_path = make_consuming_repo(tempdir_factory, 'exclude_types_repo')
     with cwd(git_path):
@@ -951,6 +964,27 @@ def test_classifier_does_not_normalize_backslashes_non_windows(tmpdir):
                 assert classifier.filenames == [r'a/b\c']
 
 
+def test_classifier_empty_types_or(tmpdir):
+    tmpdir.join('bar').ensure()
+    os.symlink(tmpdir.join('bar'), tmpdir.join('foo'))
+    with tmpdir.as_cwd():
+        classifier = Classifier(('foo', 'bar'))
+        for_symlink = classifier.by_types(
+            classifier.filenames,
+            types=['symlink'],
+            types_or=[],
+            exclude_types=[],
+        )
+        for_file = classifier.by_types(
+            classifier.filenames,
+            types=['file'],
+            types_or=[],
+            exclude_types=[],
+        )
+        assert for_symlink == ['foo']
+        assert for_file == ['bar']
+
+
 @pytest.fixture
 def some_filenames():
     return (

tests/languages/python_test.py

@@ -23,6 +23,13 @@ def test_read_pyvenv_cfg(tmpdir):
     assert python._read_pyvenv_cfg(pyvenv_cfg) == expected
 
 
+def test_read_pyvenv_cfg_non_utf8(tmpdir):
+    pyvenv_cfg = tmpdir.join('pyvenv_cfg')
+    pyvenv_cfg.write_binary('hello = hello john.š\n'.encode())
+    expected = {'hello': 'hello john.š'}
+    assert python._read_pyvenv_cfg(pyvenv_cfg) == expected
+
+
 def test_norm_version_expanduser():
     home = os.path.expanduser('~')
     if os.name == 'nt':  # pragma: nt cover

tests/repository_test.py

@@ -901,6 +901,7 @@ def test_manifest_hooks(tempdir_factory, store):
             'post-commit', 'manual', 'post-checkout', 'push',
         ),
         types=['file'],
+        types_or=[],
         verbose=False,
     )
 

tests/xargs_test.py

@@ -195,3 +195,12 @@ def test_xargs_color_true_makes_tty():
     )
     assert retcode == 0
     assert out == b'True\n'
+
+
+@pytest.mark.xfail(os.name == 'posix', reason='nt only')
+@pytest.mark.parametrize('filename', ('t.bat', 't.cmd', 'T.CMD'))
+def test_xargs_with_batch_files(tmpdir, filename):
+    f = tmpdir.join(filename)
+    f.write('echo it works\n')
+    retcode, out = xargs.xargs((str(f),), ('x',) * 8192)
+    assert retcode == 0, (retcode, out)