Merging upstream version 2.9.3.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

13 files changed, 92 insertions, 49 deletions

diff --git a/pre_commit/clientlib.py b/pre_commit/clientlib.py
index 87679bf..20d4492 100644
--- a/pre_commit/clientlib.py
+++ b/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',
diff --git a/pre_commit/commands/autoupdate.py b/pre_commit/commands/autoupdate.py
index 87f6d53..33a3473 100644
--- a/pre_commit/commands/autoupdate.py
+++ b/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(
diff --git a/pre_commit/commands/migrate_config.py b/pre_commit/commands/migrate_config.py
index d580ff1..a155f6b 100644
--- a/pre_commit/commands/migrate_config.py
+++ b/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
+    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:])
+        header = ''.join(lines[:i])
+        rest = ''.join(lines[i:])
 
-    if isinstance(yaml_load(contents), list):
         # 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
 
diff --git a/pre_commit/commands/run.py b/pre_commit/commands/run.py
index 0d335e2..1e8fad2 100644
--- a/pre_commit/commands/run.py
+++ b/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
 
 
diff --git a/pre_commit/envcontext.py b/pre_commit/envcontext.py
index 4ab0d8c..92d975d 100644
--- a/pre_commit/envcontext.py
+++ b/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
 
 
diff --git a/pre_commit/git.py b/pre_commit/git.py
index 13ba664..5096274 100644
--- a/pre_commit/git.py
+++ b/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:
diff --git a/pre_commit/hook.py b/pre_commit/hook.py
index b65ac42..ea77394 100644
--- a/pre_commit/hook.py
+++ b/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]
diff --git a/pre_commit/languages/dotnet.py b/pre_commit/languages/dotnet.py
index a8abc86..094d2f1 100644
--- a/pre_commit/languages/dotnet.py
+++ b/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(
diff --git a/pre_commit/languages/python.py b/pre_commit/languages/python.py
index 65f521c..43b7280 100644
--- a/pre_commit/languages/python.py
+++ b/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('=')
diff --git a/pre_commit/languages/ruby.py b/pre_commit/languages/ruby.py
index 1a0f0c7..81bc954 100644
--- a/pre_commit/languages/ruby.py
+++ b/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
 
 
diff --git a/pre_commit/meta_hooks/check_useless_excludes.py b/pre_commit/meta_hooks/check_useless_excludes.py
index db6865c..12be03f 100644
--- a/pre_commit/meta_hooks/check_useless_excludes.py
+++ b/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(
diff --git a/pre_commit/util.py b/pre_commit/util.py
index f4cf704..b5f40ad 100644
--- a/pre_commit/util.py
+++ b/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)
diff --git a/pre_commit/xargs.py b/pre_commit/xargs.py
index 7538b54..60a057c 100644
--- a/pre_commit/xargs.py
+++ b/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(