Adding upstream version 4.0.1.

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

pre_commit/all_languages.py

@@ -44,7 +44,5 @@ languages: dict[str, Language] = {
     'script': script,
     'swift': swift,
     'system': system,
-    # TODO: fully deprecate `python_venv`
-    'python_venv': python,
 }
 language_names = sorted(languages)

pre_commit/clientlib.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import functools
 import logging
+import os.path
 import re
 import shlex
 import sys
@@ -70,6 +71,43 @@ def transform_stage(stage: str) -> str:
     return _STAGES.get(stage, stage)
 
 
+MINIMAL_MANIFEST_SCHEMA = cfgv.Array(
+    cfgv.Map(
+        'Hook', 'id',
+        cfgv.Required('id', cfgv.check_string),
+        cfgv.Optional('stages', cfgv.check_array(cfgv.check_string), []),
+    ),
+)
+
+
+def warn_for_stages_on_repo_init(repo: str, directory: str) -> None:
+    try:
+        manifest = cfgv.load_from_filename(
+            os.path.join(directory, C.MANIFEST_FILE),
+            schema=MINIMAL_MANIFEST_SCHEMA,
+            load_strategy=yaml_load,
+            exc_tp=InvalidManifestError,
+        )
+    except InvalidManifestError:
+        return  # they'll get a better error message when it actually loads!
+
+    legacy_stages = {}  # sorted set
+    for hook in manifest:
+        for stage in hook.get('stages', ()):
+            if stage in _STAGES:
+                legacy_stages[stage] = True
+
+    if legacy_stages:
+        logger.warning(
+            f'repo `{repo}` uses deprecated stage names '
+            f'({", ".join(legacy_stages)}) which will be removed in a '
+            f'future version.  '
+            f'Hint: often `pre-commit autoupdate --repo {shlex.quote(repo)}` '
+            f'will fix this.  '
+            f'if it does not -- consider reporting an issue to that repo.',
+        )
+
+
 class StagesMigrationNoDefault(NamedTuple):
     key: str
     default: Sequence[str]
@@ -99,6 +137,58 @@ class StagesMigration(StagesMigrationNoDefault):
         super().apply_default(dct)
 
 
+class DeprecatedStagesWarning(NamedTuple):
+    key: str
+
+    def check(self, dct: dict[str, Any]) -> None:
+        if self.key not in dct:
+            return
+
+        val = dct[self.key]
+        cfgv.check_array(cfgv.check_any)(val)
+
+        legacy_stages = [stage for stage in val if stage in _STAGES]
+        if legacy_stages:
+            logger.warning(
+                f'hook id `{dct["id"]}` uses deprecated stage names '
+                f'({", ".join(legacy_stages)}) which will be removed in a '
+                f'future version.  '
+                f'run: `pre-commit migrate-config` to automatically fix this.',
+            )
+
+    def apply_default(self, dct: dict[str, Any]) -> None:
+        pass
+
+    def remove_default(self, dct: dict[str, Any]) -> None:
+        raise NotImplementedError
+
+
+class DeprecatedDefaultStagesWarning(NamedTuple):
+    key: str
+
+    def check(self, dct: dict[str, Any]) -> None:
+        if self.key not in dct:
+            return
+
+        val = dct[self.key]
+        cfgv.check_array(cfgv.check_any)(val)
+
+        legacy_stages = [stage for stage in val if stage in _STAGES]
+        if legacy_stages:
+            logger.warning(
+                f'top-level `default_stages` uses deprecated stage names '
+                f'({", ".join(legacy_stages)}) which will be removed in a '
+                f'future version.  '
+                f'run: `pre-commit migrate-config` to automatically fix this.',
+            )
+
+    def apply_default(self, dct: dict[str, Any]) -> None:
+        pass
+
+    def remove_default(self, dct: dict[str, Any]) -> None:
+        raise NotImplementedError
+
+
 MANIFEST_HOOK_DICT = cfgv.Map(
     'Hook', 'id',
 
@@ -267,6 +357,12 @@ class NotAllowed(cfgv.OptionalNoDefault):
             raise cfgv.ValidationError(f'{self.key!r} cannot be overridden')
 
 
+_COMMON_HOOK_WARNINGS = (
+    OptionalSensibleRegexAtHook('files', cfgv.check_string),
+    OptionalSensibleRegexAtHook('exclude', cfgv.check_string),
+    DeprecatedStagesWarning('stages'),
+)
+
 META_HOOK_DICT = cfgv.Map(
     'Hook', 'id',
     cfgv.Required('id', cfgv.check_string),
@@ -289,6 +385,7 @@ META_HOOK_DICT = cfgv.Map(
         item
         for item in MANIFEST_HOOK_DICT.items
     ),
+    *_COMMON_HOOK_WARNINGS,
 )
 CONFIG_HOOK_DICT = cfgv.Map(
     'Hook', 'id',
@@ -306,16 +403,13 @@ CONFIG_HOOK_DICT = cfgv.Map(
         if item.key != 'stages'
     ),
     StagesMigrationNoDefault('stages', []),
-    OptionalSensibleRegexAtHook('files', cfgv.check_string),
-    OptionalSensibleRegexAtHook('exclude', cfgv.check_string),
+    *_COMMON_HOOK_WARNINGS,
 )
 LOCAL_HOOK_DICT = cfgv.Map(
     'Hook', 'id',
 
     *MANIFEST_HOOK_DICT.items,
-
-    OptionalSensibleRegexAtHook('files', cfgv.check_string),
-    OptionalSensibleRegexAtHook('exclude', cfgv.check_string),
+    *_COMMON_HOOK_WARNINGS,
 )
 CONFIG_REPO_DICT = cfgv.Map(
     'Repository', 'repo',
@@ -368,6 +462,7 @@ CONFIG_SCHEMA = cfgv.Map(
         'default_language_version', DEFAULT_LANGUAGE_VERSION, {},
     ),
     StagesMigration('default_stages', STAGES),
+    DeprecatedDefaultStagesWarning('default_stages'),
     cfgv.Optional('files', check_string_regex, ''),
     cfgv.Optional('exclude', check_string_regex, '^$'),
     cfgv.Optional('fail_fast', cfgv.check_bool, False),

pre_commit/commands/migrate_config.py

@@ -1,13 +1,21 @@
 from __future__ import annotations
 
-import re
+import functools
+import itertools
 import textwrap
+from typing import Callable
 
 import cfgv
 import yaml
+from yaml.nodes import ScalarNode
 
 from pre_commit.clientlib import InvalidConfigError
+from pre_commit.yaml import yaml_compose
 from pre_commit.yaml import yaml_load
+from pre_commit.yaml_rewrite import MappingKey
+from pre_commit.yaml_rewrite import MappingValue
+from pre_commit.yaml_rewrite import match
+from pre_commit.yaml_rewrite import SequenceItem
 
 
 def _is_header_line(line: str) -> bool:
@@ -38,16 +46,69 @@ def _migrate_map(contents: str) -> str:
     return contents
 
 
-def _migrate_sha_to_rev(contents: str) -> str:
-    return re.sub(r'(\n\s+)sha:', r'\1rev:', contents)
+def _preserve_style(n: ScalarNode, *, s: str) -> str:
+    style = n.style or ''
+    return f'{style}{s}{style}'
 
 
-def _migrate_python_venv(contents: str) -> str:
-    return re.sub(
-        r'(\n\s+)language: python_venv\b',
-        r'\1language: python',
-        contents,
+def _fix_stage(n: ScalarNode) -> str:
+    return _preserve_style(n, s=f'pre-{n.value}')
+
+
+def _migrate_composed(contents: str) -> str:
+    tree = yaml_compose(contents)
+    rewrites: list[tuple[ScalarNode, Callable[[ScalarNode], str]]] = []
+
+    # sha -> rev
+    sha_to_rev_replace = functools.partial(_preserve_style, s='rev')
+    sha_to_rev_matcher = (
+        MappingValue('repos'),
+        SequenceItem(),
+        MappingKey('sha'),
     )
+    for node in match(tree, sha_to_rev_matcher):
+        rewrites.append((node, sha_to_rev_replace))
+
+    # python_venv -> python
+    language_matcher = (
+        MappingValue('repos'),
+        SequenceItem(),
+        MappingValue('hooks'),
+        SequenceItem(),
+        MappingValue('language'),
+    )
+    python_venv_replace = functools.partial(_preserve_style, s='python')
+    for node in match(tree, language_matcher):
+        if node.value == 'python_venv':
+            rewrites.append((node, python_venv_replace))
+
+    # stages rewrites
+    default_stages_matcher = (MappingValue('default_stages'), SequenceItem())
+    default_stages_match = match(tree, default_stages_matcher)
+    hook_stages_matcher = (
+        MappingValue('repos'),
+        SequenceItem(),
+        MappingValue('hooks'),
+        SequenceItem(),
+        MappingValue('stages'),
+        SequenceItem(),
+    )
+    hook_stages_match = match(tree, hook_stages_matcher)
+    for node in itertools.chain(default_stages_match, hook_stages_match):
+        if node.value in {'commit', 'push', 'merge-commit'}:
+            rewrites.append((node, _fix_stage))
+
+    rewrites.sort(reverse=True, key=lambda nf: nf[0].start_mark.index)
+
+    src_parts = []
+    end: int | None = None
+    for node, func in rewrites:
+        src_parts.append(contents[node.end_mark.index:end])
+        src_parts.append(func(node))
+        end = node.start_mark.index
+    src_parts.append(contents[:end])
+    src_parts.reverse()
+    return ''.join(src_parts)
 
 
 def migrate_config(config_file: str, quiet: bool = False) -> int:
@@ -62,8 +123,7 @@ def migrate_config(config_file: str, quiet: bool = False) -> int:
                 raise cfgv.ValidationError(str(e))
 
     contents = _migrate_map(contents)
-    contents = _migrate_sha_to_rev(contents)
-    contents = _migrate_python_venv(contents)
+    contents = _migrate_composed(contents)
 
     if contents != orig_contents:
         with open(config_file, 'w') as f:

pre_commit/commands/run.py

@@ -61,7 +61,7 @@ def filter_by_include_exclude(
         names: Iterable[str],
         include: str,
         exclude: str,
-) -> Generator[str, None, None]:
+) -> Generator[str]:
     include_re, exclude_re = re.compile(include), re.compile(exclude)
     return (
         filename for filename in names
@@ -84,7 +84,7 @@ class Classifier:
             types: Iterable[str],
             types_or: Iterable[str],
             exclude_types: Iterable[str],
-    ) -> Generator[str, None, None]:
+    ) -> Generator[str]:
         types = frozenset(types)
         types_or = frozenset(types_or)
         exclude_types = frozenset(exclude_types)
@@ -97,7 +97,7 @@ class Classifier:
             ):
                 yield filename
 
-    def filenames_for_hook(self, hook: Hook) -> Generator[str, None, None]:
+    def filenames_for_hook(self, hook: Hook) -> Generator[str]:
         return self.by_types(
             filter_by_include_exclude(
                 self.filenames,

pre_commit/envcontext.py

@@ -33,7 +33,7 @@ def format_env(parts: SubstitutionT, env: MutableMapping[str, str]) -> str:
 def envcontext(
         patch: PatchesT,
         _env: MutableMapping[str, str] | None = None,
-) -> Generator[None, None, None]:
+) -> Generator[None]:
     """In this context, `os.environ` is modified according to `patch`.
 
     `patch` is an iterable of 2-tuples (key, value):

pre_commit/error_handler.py

@@ -68,7 +68,7 @@ def _log_and_exit(
 
 
 @contextlib.contextmanager
-def error_handler() -> Generator[None, None, None]:
+def error_handler() -> Generator[None]:
     try:
         yield
     except (Exception, KeyboardInterrupt) as e:

pre_commit/file_lock.py

@@ -20,7 +20,7 @@ if sys.platform == 'win32':  # pragma: no cover (windows)
     def _locked(
             fileno: int,
             blocked_cb: Callable[[], None],
-    ) -> Generator[None, None, None]:
+    ) -> Generator[None]:
         try:
             msvcrt.locking(fileno, msvcrt.LK_NBLCK, _region)
         except OSError:
@@ -53,7 +53,7 @@ else:  # pragma: win32 no cover
     def _locked(
             fileno: int,
             blocked_cb: Callable[[], None],
-    ) -> Generator[None, None, None]:
+    ) -> Generator[None]:
         try:
             fcntl.flock(fileno, fcntl.LOCK_EX | fcntl.LOCK_NB)
         except OSError:  # pragma: no cover (tests are single-threaded)
@@ -69,7 +69,7 @@ else:  # pragma: win32 no cover
 def lock(
         path: str,
         blocked_cb: Callable[[], None],
-) -> Generator[None, None, None]:
+) -> Generator[None]:
     with open(path, 'a+') as f:
         with _locked(f.fileno(), blocked_cb):
             yield

pre_commit/lang_base.py

@@ -127,7 +127,7 @@ def no_install(
 
 
 @contextlib.contextmanager
-def no_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def no_env(prefix: Prefix, version: str) -> Generator[None]:
     yield
 
 

pre_commit/languages/conda.py

@@ -41,7 +41,7 @@ def get_env_patch(env: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/coursier.py

@@ -70,7 +70,7 @@ def get_env_patch(target_dir: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/dart.py

@@ -29,7 +29,7 @@ def get_env_patch(venv: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/dotnet.py

@@ -30,14 +30,14 @@ def get_env_patch(venv: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield
 
 
 @contextlib.contextmanager
-def _nuget_config_no_sources() -> Generator[str, None, None]:
+def _nuget_config_no_sources() -> Generator[str]:
     with tempfile.TemporaryDirectory() as tmpdir:
         nuget_config = os.path.join(tmpdir, 'nuget.config')
         with open(nuget_config, 'w') as f:

pre_commit/languages/golang.py

@@ -121,7 +121,7 @@ def _install_go(version: str, dest: str) -> None:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir, version)):
         yield

pre_commit/languages/haskell.py

@@ -24,7 +24,7 @@ def get_env_patch(target_dir: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/lua.py

@@ -44,7 +44,7 @@ def get_env_patch(d: str) -> PatchesT:  # pragma: win32 no cover
 
 
 @contextlib.contextmanager  # pragma: win32 no cover
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/node.py

@@ -59,7 +59,7 @@ def get_env_patch(venv: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/perl.py

@@ -33,7 +33,7 @@ def get_env_patch(venv: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/python.py

@@ -152,7 +152,7 @@ def norm_version(version: str) -> str | None:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/r.py

@@ -85,7 +85,7 @@ def health_check(prefix: Prefix, version: str) -> str | None:
 
 
 @contextlib.contextmanager
-def _r_code_in_tempfile(code: str) -> Generator[str, None, None]:
+def _r_code_in_tempfile(code: str) -> Generator[str]:
     """
     To avoid quoting and escaping issues, avoid `Rscript [options] -e {expr}`
     but use `Rscript [options] path/to/file_with_expr.R`
@@ -105,7 +105,7 @@ def get_env_patch(venv: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/languages/ruby.py

@@ -73,7 +73,7 @@ def get_env_patch(
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir, version)):
         yield

pre_commit/languages/rust.py

@@ -61,7 +61,7 @@ def get_env_patch(target_dir: str, version: str) -> PatchesT:
 
 
 @contextlib.contextmanager
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir, version)):
         yield

pre_commit/languages/swift.py

@@ -27,7 +27,7 @@ def get_env_patch(venv: str) -> PatchesT:  # pragma: win32 no cover
 
 
 @contextlib.contextmanager  # pragma: win32 no cover
-def in_env(prefix: Prefix, version: str) -> Generator[None, None, None]:
+def in_env(prefix: Prefix, version: str) -> Generator[None]:
     envdir = lang_base.environment_dir(prefix, ENVIRONMENT_DIR, version)
     with envcontext(get_env_patch(envdir)):
         yield

pre_commit/logging_handler.py

@@ -32,7 +32,7 @@ class LoggingHandler(logging.Handler):
 
 
 @contextlib.contextmanager
-def logging_handler(use_color: bool) -> Generator[None, None, None]:
+def logging_handler(use_color: bool) -> Generator[None]:
     handler = LoggingHandler(use_color)
     logger.addHandler(handler)
     logger.setLevel(logging.INFO)

pre_commit/repository.py

@@ -3,7 +3,6 @@ from __future__ import annotations
 import json
 import logging
 import os
-import shlex
 from collections.abc import Sequence
 from typing import Any
 
@@ -68,14 +67,6 @@ def _hook_install(hook: Hook) -> None:
     logger.info('Once installed this environment will be reused.')
     logger.info('This may take a few minutes...')
 
-    if hook.language == 'python_venv':
-        logger.warning(
-            f'`repo: {hook.src}` uses deprecated `language: python_venv`.  '
-            f'This is an alias for `language: python`.  '
-            f'Often `pre-commit autoupdate --repo {shlex.quote(hook.src)}` '
-            f'will fix this.',
-        )
-
     lang = languages[hook.language]
     assert lang.ENVIRONMENT_DIR is not None
 

pre_commit/staged_files_only.py

@@ -33,7 +33,7 @@ def _git_apply(patch: str) -> None:
 
 
 @contextlib.contextmanager
-def _intent_to_add_cleared() -> Generator[None, None, None]:
+def _intent_to_add_cleared() -> Generator[None]:
     intent_to_add = git.intent_to_add_files()
     if intent_to_add:
         logger.warning('Unstaged intent-to-add files detected.')
@@ -48,7 +48,7 @@ def _intent_to_add_cleared() -> Generator[None, None, None]:
 
 
 @contextlib.contextmanager
-def _unstaged_changes_cleared(patch_dir: str) -> Generator[None, None, None]:
+def _unstaged_changes_cleared(patch_dir: str) -> Generator[None]:
     tree = cmd_output('git', 'write-tree')[1].strip()
     diff_cmd = (
         'git', 'diff-index', '--ignore-submodules', '--binary',
@@ -105,7 +105,7 @@ def _unstaged_changes_cleared(patch_dir: str) -> Generator[None, None, None]:
 
 
 @contextlib.contextmanager
-def staged_files_only(patch_dir: str) -> Generator[None, None, None]:
+def staged_files_only(patch_dir: str) -> Generator[None]:
     """Clear any unstaged changes from the git working directory inside this
     context.
     """

pre_commit/store.py

@@ -10,6 +10,7 @@ from collections.abc import Sequence
 from typing import Callable
 
 import pre_commit.constants as C
+from pre_commit import clientlib
 from pre_commit import file_lock
 from pre_commit import git
 from pre_commit.util import CalledProcessError
@@ -101,7 +102,7 @@ class Store:
             os.replace(tmpfile, self.db_path)
 
     @contextlib.contextmanager
-    def exclusive_lock(self) -> Generator[None, None, None]:
+    def exclusive_lock(self) -> Generator[None]:
         def blocked_cb() -> None:  # pragma: no cover (tests are in-process)
             logger.info('Locking pre-commit directory')
 
@@ -112,7 +113,7 @@ class Store:
     def connect(
             self,
             db_path: str | None = None,
-    ) -> Generator[sqlite3.Connection, None, None]:
+    ) -> Generator[sqlite3.Connection]:
         db_path = db_path or self.db_path
         # sqlite doesn't close its fd with its contextmanager >.<
         # contextlib.closing fixes this.
@@ -136,6 +137,7 @@ class Store:
             deps: Sequence[str],
             make_strategy: Callable[[str], None],
     ) -> str:
+        original_repo = repo
         repo = self.db_repo_name(repo, deps)
 
         def _get_result() -> str | None:
@@ -168,6 +170,9 @@ class Store:
                     'INSERT INTO repos (repo, ref, path) VALUES (?, ?, ?)',
                     [repo, ref, directory],
                 )
+
+            clientlib.warn_for_stages_on_repo_init(original_repo, directory)
+
         return directory
 
     def _complete_clone(self, ref: str, git_cmd: Callable[..., None]) -> None:

pre_commit/util.py

@@ -25,7 +25,7 @@ def force_bytes(exc: Any) -> bytes:
 
 
 @contextlib.contextmanager
-def clean_path_on_failure(path: str) -> Generator[None, None, None]:
+def clean_path_on_failure(path: str) -> Generator[None]:
     """Cleans up the directory on an exceptional failure."""
     try:
         yield

pre_commit/xargs.py

@@ -120,7 +120,6 @@ def partition(
 @contextlib.contextmanager
 def _thread_mapper(maxsize: int) -> Generator[
     Callable[[Callable[[TArg], TRet], Iterable[TArg]], Iterable[TRet]],
-    None, None,
 ]:
     if maxsize == 1:
         yield map

pre_commit/yaml.py

@@ -6,6 +6,7 @@ from typing import Any
 import yaml
 
 Loader = getattr(yaml, 'CSafeLoader', yaml.SafeLoader)
+yaml_compose = functools.partial(yaml.compose, Loader=Loader)
 yaml_load = functools.partial(yaml.load, Loader=Loader)
 Dumper = getattr(yaml, 'CSafeDumper', yaml.SafeDumper)
 

pre_commit/yaml_rewrite.py

@@ -0,0 +1,52 @@
+from __future__ import annotations
+
+from collections.abc import Generator
+from collections.abc import Iterable
+from typing import NamedTuple
+from typing import Protocol
+
+from yaml.nodes import MappingNode
+from yaml.nodes import Node
+from yaml.nodes import ScalarNode
+from yaml.nodes import SequenceNode
+
+
+class _Matcher(Protocol):
+    def match(self, n: Node) -> Generator[Node]: ...
+
+
+class MappingKey(NamedTuple):
+    k: str
+
+    def match(self, n: Node) -> Generator[Node]:
+        if isinstance(n, MappingNode):
+            for k, _ in n.value:
+                if k.value == self.k:
+                    yield k
+
+
+class MappingValue(NamedTuple):
+    k: str
+
+    def match(self, n: Node) -> Generator[Node]:
+        if isinstance(n, MappingNode):
+            for k, v in n.value:
+                if k.value == self.k:
+                    yield v
+
+
+class SequenceItem(NamedTuple):
+    def match(self, n: Node) -> Generator[Node]:
+        if isinstance(n, SequenceNode):
+            yield from n.value
+
+
+def _match(gen: Iterable[Node], m: _Matcher) -> Iterable[Node]:
+    return (n for src in gen for n in m.match(src))
+
+
+def match(n: Node, matcher: tuple[_Matcher, ...]) -> Generator[ScalarNode]:
+    gen: Iterable[Node] = (n,)
+    for m in matcher:
+        gen = _match(gen, m)
+    return (n for n in gen if isinstance(n, ScalarNode))