From c86df75ab11643fa4649cfe6ed5c4692d4ee342b Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Mon, 15 Apr 2024 20:05:20 +0200 Subject: Adding upstream version 3.6.2. Signed-off-by: Daniel Baumann --- pre_commit/commands/__init__.py | 0 pre_commit/commands/autoupdate.py | 215 +++++++++++++++ pre_commit/commands/clean.py | 16 ++ pre_commit/commands/gc.py | 89 ++++++ pre_commit/commands/hook_impl.py | 271 +++++++++++++++++++ pre_commit/commands/init_templatedir.py | 39 +++ pre_commit/commands/install_uninstall.py | 167 ++++++++++++ pre_commit/commands/migrate_config.py | 75 ++++++ pre_commit/commands/run.py | 447 +++++++++++++++++++++++++++++++ pre_commit/commands/sample_config.py | 18 ++ pre_commit/commands/try_repo.py | 77 ++++++ pre_commit/commands/validate_config.py | 18 ++ pre_commit/commands/validate_manifest.py | 18 ++ 13 files changed, 1450 insertions(+) create mode 100644 pre_commit/commands/__init__.py create mode 100644 pre_commit/commands/autoupdate.py create mode 100644 pre_commit/commands/clean.py create mode 100644 pre_commit/commands/gc.py create mode 100644 pre_commit/commands/hook_impl.py create mode 100644 pre_commit/commands/init_templatedir.py create mode 100644 pre_commit/commands/install_uninstall.py create mode 100644 pre_commit/commands/migrate_config.py create mode 100644 pre_commit/commands/run.py create mode 100644 pre_commit/commands/sample_config.py create mode 100644 pre_commit/commands/try_repo.py create mode 100644 pre_commit/commands/validate_config.py create mode 100644 pre_commit/commands/validate_manifest.py (limited to 'pre_commit/commands') diff --git a/pre_commit/commands/__init__.py b/pre_commit/commands/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/pre_commit/commands/autoupdate.py b/pre_commit/commands/autoupdate.py new file mode 100644 index 0000000..aa0c5e2 --- /dev/null +++ b/pre_commit/commands/autoupdate.py @@ -0,0 +1,215 @@ +from __future__ import annotations + +import concurrent.futures +import os.path +import re +import tempfile +from collections.abc import Sequence +from typing import Any +from typing import NamedTuple + +import pre_commit.constants as C +from pre_commit import git +from pre_commit import output +from pre_commit import xargs +from pre_commit.clientlib import InvalidManifestError +from pre_commit.clientlib import load_config +from pre_commit.clientlib import load_manifest +from pre_commit.clientlib import LOCAL +from pre_commit.clientlib import META +from pre_commit.commands.migrate_config import migrate_config +from pre_commit.util import CalledProcessError +from pre_commit.util import cmd_output +from pre_commit.util import cmd_output_b +from pre_commit.yaml import yaml_dump +from pre_commit.yaml import yaml_load + + +class RevInfo(NamedTuple): + repo: str + rev: str + frozen: str | None = None + hook_ids: frozenset[str] = frozenset() + + @classmethod + def from_config(cls, config: dict[str, Any]) -> RevInfo: + return cls(config['repo'], config['rev']) + + def update(self, tags_only: bool, freeze: bool) -> RevInfo: + with tempfile.TemporaryDirectory() as tmp: + _git = ('git', *git.NO_FS_MONITOR, '-C', tmp) + + if tags_only: + tag_opt = '--abbrev=0' + else: + tag_opt = '--exact' + tag_cmd = (*_git, 'describe', 'FETCH_HEAD', '--tags', tag_opt) + + git.init_repo(tmp, self.repo) + cmd_output_b(*_git, 'config', 'extensions.partialClone', 'true') + cmd_output_b( + *_git, 'fetch', 'origin', 'HEAD', + '--quiet', '--filter=blob:none', '--tags', + ) + + try: + rev = cmd_output(*tag_cmd)[1].strip() + except CalledProcessError: + rev = cmd_output(*_git, 'rev-parse', 'FETCH_HEAD')[1].strip() + else: + if tags_only: + rev = git.get_best_candidate_tag(rev, tmp) + + frozen = None + if freeze: + exact = cmd_output(*_git, 'rev-parse', rev)[1].strip() + if exact != rev: + rev, frozen = exact, rev + + try: + # workaround for windows -- see #2865 + cmd_output_b(*_git, 'show', f'{rev}:{C.MANIFEST_FILE}') + cmd_output(*_git, 'checkout', rev, '--', C.MANIFEST_FILE) + except CalledProcessError: + pass # this will be caught by manifest validating code + try: + manifest = load_manifest(os.path.join(tmp, C.MANIFEST_FILE)) + except InvalidManifestError as e: + raise RepositoryCannotBeUpdatedError(f'[{self.repo}] {e}') + else: + hook_ids = frozenset(hook['id'] for hook in manifest) + + return self._replace(rev=rev, frozen=frozen, hook_ids=hook_ids) + + +class RepositoryCannotBeUpdatedError(RuntimeError): + pass + + +def _check_hooks_still_exist_at_rev( + repo_config: dict[str, Any], + info: RevInfo, +) -> None: + # See if any of our hooks were deleted with the new commits + hooks = {hook['id'] for hook in repo_config['hooks']} + hooks_missing = hooks - info.hook_ids + if hooks_missing: + raise RepositoryCannotBeUpdatedError( + f'[{info.repo}] Cannot update because the update target is ' + f'missing these hooks: {", ".join(sorted(hooks_missing))}', + ) + + +def _update_one( + i: int, + repo: dict[str, Any], + *, + tags_only: bool, + freeze: bool, +) -> tuple[int, RevInfo, RevInfo]: + old = RevInfo.from_config(repo) + new = old.update(tags_only=tags_only, freeze=freeze) + _check_hooks_still_exist_at_rev(repo, new) + return i, old, new + + +REV_LINE_RE = re.compile(r'^(\s+)rev:(\s*)([\'"]?)([^\s#]+)(.*)(\r?\n)$') + + +def _original_lines( + path: str, + rev_infos: list[RevInfo | None], + retry: bool = False, +) -> tuple[list[str], list[int]]: + """detect `rev:` lines or reformat the file""" + with open(path, newline='') as f: + original = f.read() + + lines = original.splitlines(True) + idxs = [i for i, line in enumerate(lines) if REV_LINE_RE.match(line)] + if len(idxs) == len(rev_infos): + return lines, idxs + elif retry: + raise AssertionError('could not find rev lines') + else: + with open(path, 'w') as f: + f.write(yaml_dump(yaml_load(original))) + return _original_lines(path, rev_infos, retry=True) + + +def _write_new_config(path: str, rev_infos: list[RevInfo | None]) -> None: + lines, idxs = _original_lines(path, rev_infos) + + for idx, rev_info in zip(idxs, rev_infos): + if rev_info is None: + continue + match = REV_LINE_RE.match(lines[idx]) + assert match is not None + new_rev_s = yaml_dump({'rev': rev_info.rev}, default_style=match[3]) + new_rev = new_rev_s.split(':', 1)[1].strip() + if rev_info.frozen is not None: + comment = f' # frozen: {rev_info.frozen}' + elif match[5].strip().startswith('# frozen:'): + comment = '' + else: + comment = match[5] + lines[idx] = f'{match[1]}rev:{match[2]}{new_rev}{comment}{match[6]}' + + with open(path, 'w', newline='') as f: + f.write(''.join(lines)) + + +def autoupdate( + config_file: str, + tags_only: bool, + freeze: bool, + repos: Sequence[str] = (), + jobs: int = 1, +) -> int: + """Auto-update the pre-commit config to the latest versions of repos.""" + migrate_config(config_file, quiet=True) + changed = False + retv = 0 + + config_repos = [ + repo for repo in load_config(config_file)['repos'] + if repo['repo'] not in {LOCAL, META} + ] + + rev_infos: list[RevInfo | None] = [None] * len(config_repos) + jobs = jobs or xargs.cpu_count() # 0 => number of cpus + jobs = min(jobs, len(repos) or len(config_repos)) # max 1-per-thread + jobs = max(jobs, 1) # at least one thread + with concurrent.futures.ThreadPoolExecutor(jobs) as exe: + futures = [ + exe.submit( + _update_one, + i, repo, tags_only=tags_only, freeze=freeze, + ) + for i, repo in enumerate(config_repos) + if not repos or repo['repo'] in repos + ] + for future in concurrent.futures.as_completed(futures): + try: + i, old, new = future.result() + except RepositoryCannotBeUpdatedError as e: + output.write_line(str(e)) + retv = 1 + else: + if new.rev != old.rev: + changed = True + if new.frozen: + new_s = f'{new.frozen} (frozen)' + else: + new_s = new.rev + msg = f'updating {old.rev} -> {new_s}' + rev_infos[i] = new + else: + msg = 'already up to date!' + + output.write_line(f'[{old.repo}] {msg}') + + if changed: + _write_new_config(config_file, rev_infos) + + return retv diff --git a/pre_commit/commands/clean.py b/pre_commit/commands/clean.py new file mode 100644 index 0000000..5119f64 --- /dev/null +++ b/pre_commit/commands/clean.py @@ -0,0 +1,16 @@ +from __future__ import annotations + +import os.path + +from pre_commit import output +from pre_commit.store import Store +from pre_commit.util import rmtree + + +def clean(store: Store) -> int: + legacy_path = os.path.expanduser('~/.pre-commit') + for directory in (store.directory, legacy_path): + if os.path.exists(directory): + rmtree(directory) + output.write_line(f'Cleaned {directory}.') + return 0 diff --git a/pre_commit/commands/gc.py b/pre_commit/commands/gc.py new file mode 100644 index 0000000..6892e09 --- /dev/null +++ b/pre_commit/commands/gc.py @@ -0,0 +1,89 @@ +from __future__ import annotations + +import os.path +from typing import Any + +import pre_commit.constants as C +from pre_commit import output +from pre_commit.clientlib import InvalidConfigError +from pre_commit.clientlib import InvalidManifestError +from pre_commit.clientlib import load_config +from pre_commit.clientlib import load_manifest +from pre_commit.clientlib import LOCAL +from pre_commit.clientlib import META +from pre_commit.store import Store + + +def _mark_used_repos( + store: Store, + all_repos: dict[tuple[str, str], str], + unused_repos: set[tuple[str, str]], + repo: dict[str, Any], +) -> None: + if repo['repo'] == META: + return + elif repo['repo'] == LOCAL: + for hook in repo['hooks']: + deps = hook.get('additional_dependencies') + unused_repos.discard(( + store.db_repo_name(repo['repo'], deps), C.LOCAL_REPO_VERSION, + )) + else: + key = (repo['repo'], repo['rev']) + path = all_repos.get(key) + # can't inspect manifest if it isn't cloned + if path is None: + return + + try: + manifest = load_manifest(os.path.join(path, C.MANIFEST_FILE)) + except InvalidManifestError: + return + else: + unused_repos.discard(key) + by_id = {hook['id']: hook for hook in manifest} + + for hook in repo['hooks']: + if hook['id'] not in by_id: + continue + + deps = hook.get( + 'additional_dependencies', + by_id[hook['id']]['additional_dependencies'], + ) + unused_repos.discard(( + store.db_repo_name(repo['repo'], deps), repo['rev'], + )) + + +def _gc_repos(store: Store) -> int: + configs = store.select_all_configs() + repos = store.select_all_repos() + + # delete config paths which do not exist + dead_configs = [p for p in configs if not os.path.exists(p)] + live_configs = [p for p in configs if os.path.exists(p)] + + all_repos = {(repo, ref): path for repo, ref, path in repos} + unused_repos = set(all_repos) + for config_path in live_configs: + try: + config = load_config(config_path) + except InvalidConfigError: + dead_configs.append(config_path) + continue + else: + for repo in config['repos']: + _mark_used_repos(store, all_repos, unused_repos, repo) + + store.delete_configs(dead_configs) + for db_repo_name, ref in unused_repos: + store.delete_repo(db_repo_name, ref, all_repos[(db_repo_name, ref)]) + return len(unused_repos) + + +def gc(store: Store) -> int: + with store.exclusive_lock(): + repos_removed = _gc_repos(store) + output.write_line(f'{repos_removed} repo(s) removed.') + return 0 diff --git a/pre_commit/commands/hook_impl.py b/pre_commit/commands/hook_impl.py new file mode 100644 index 0000000..49a80b7 --- /dev/null +++ b/pre_commit/commands/hook_impl.py @@ -0,0 +1,271 @@ +from __future__ import annotations + +import argparse +import os.path +import subprocess +import sys +from collections.abc import Sequence + +from pre_commit.commands.run import run +from pre_commit.envcontext import envcontext +from pre_commit.parse_shebang import normalize_cmd +from pre_commit.store import Store + +Z40 = '0' * 40 + + +def _run_legacy( + hook_type: str, + hook_dir: str, + args: Sequence[str], +) -> tuple[int, bytes]: + if os.environ.get('PRE_COMMIT_RUNNING_LEGACY'): + raise SystemExit( + f"bug: pre-commit's script is installed in migration mode\n" + f'run `pre-commit install -f --hook-type {hook_type}` to fix ' + f'this\n\n' + f'Please report this bug at ' + f'https://github.com/pre-commit/pre-commit/issues', + ) + + if hook_type == 'pre-push': + stdin = sys.stdin.buffer.read() + else: + stdin = b'' + + # not running in legacy mode + legacy_hook = os.path.join(hook_dir, f'{hook_type}.legacy') + if not os.access(legacy_hook, os.X_OK): + return 0, stdin + + with envcontext((('PRE_COMMIT_RUNNING_LEGACY', '1'),)): + cmd = normalize_cmd((legacy_hook, *args)) + return subprocess.run(cmd, input=stdin).returncode, stdin + + +def _validate_config( + retv: int, + config: str, + skip_on_missing_config: bool, +) -> None: + if not os.path.isfile(config): + if skip_on_missing_config or os.getenv('PRE_COMMIT_ALLOW_NO_CONFIG'): + print(f'`{config}` config file not found. Skipping `pre-commit`.') + raise SystemExit(retv) + else: + print( + f'No {config} file was found\n' + f'- To temporarily silence this, run ' + f'`PRE_COMMIT_ALLOW_NO_CONFIG=1 git ...`\n' + f'- To permanently silence this, install pre-commit with the ' + f'--allow-missing-config option\n' + f'- To uninstall pre-commit run `pre-commit uninstall`', + ) + raise SystemExit(1) + + +def _ns( + hook_type: str, + color: bool, + *, + all_files: bool = False, + remote_branch: str | None = None, + local_branch: str | None = None, + from_ref: str | None = None, + to_ref: str | None = None, + pre_rebase_upstream: str | None = None, + pre_rebase_branch: str | None = None, + remote_name: str | None = None, + remote_url: str | None = None, + commit_msg_filename: str | None = None, + prepare_commit_message_source: str | None = None, + commit_object_name: str | None = None, + checkout_type: str | None = None, + is_squash_merge: str | None = None, + rewrite_command: str | None = None, +) -> argparse.Namespace: + return argparse.Namespace( + color=color, + hook_stage=hook_type, + remote_branch=remote_branch, + local_branch=local_branch, + from_ref=from_ref, + to_ref=to_ref, + pre_rebase_upstream=pre_rebase_upstream, + pre_rebase_branch=pre_rebase_branch, + remote_name=remote_name, + remote_url=remote_url, + commit_msg_filename=commit_msg_filename, + prepare_commit_message_source=prepare_commit_message_source, + commit_object_name=commit_object_name, + all_files=all_files, + checkout_type=checkout_type, + is_squash_merge=is_squash_merge, + rewrite_command=rewrite_command, + files=(), + hook=None, + verbose=False, + show_diff_on_failure=False, + ) + + +def _rev_exists(rev: str) -> bool: + return not subprocess.call(('git', 'rev-list', '--quiet', rev)) + + +def _pre_push_ns( + color: bool, + args: Sequence[str], + stdin: bytes, +) -> argparse.Namespace | None: + remote_name = args[0] + remote_url = args[1] + + for line in stdin.decode().splitlines(): + parts = line.rsplit(maxsplit=3) + local_branch, local_sha, remote_branch, remote_sha = parts + if local_sha == Z40: + continue + elif remote_sha != Z40 and _rev_exists(remote_sha): + return _ns( + 'pre-push', color, + from_ref=remote_sha, to_ref=local_sha, + remote_branch=remote_branch, + local_branch=local_branch, + remote_name=remote_name, remote_url=remote_url, + ) + else: + # ancestors not found in remote + ancestors = subprocess.check_output(( + 'git', 'rev-list', local_sha, '--topo-order', '--reverse', + '--not', f'--remotes={remote_name}', + )).decode().strip() + if not ancestors: + continue + else: + first_ancestor = ancestors.splitlines()[0] + cmd = ('git', 'rev-list', '--max-parents=0', local_sha) + roots = set(subprocess.check_output(cmd).decode().splitlines()) + if first_ancestor in roots: + # pushing the whole tree including root commit + return _ns( + 'pre-push', color, + all_files=True, + remote_name=remote_name, remote_url=remote_url, + remote_branch=remote_branch, + local_branch=local_branch, + ) + else: + rev_cmd = ('git', 'rev-parse', f'{first_ancestor}^') + source = subprocess.check_output(rev_cmd).decode().strip() + return _ns( + 'pre-push', color, + from_ref=source, to_ref=local_sha, + remote_name=remote_name, remote_url=remote_url, + remote_branch=remote_branch, + local_branch=local_branch, + ) + + # nothing to push + return None + + +_EXPECTED_ARG_LENGTH_BY_HOOK = { + 'commit-msg': 1, + 'post-checkout': 3, + 'post-commit': 0, + 'pre-commit': 0, + 'pre-merge-commit': 0, + 'post-merge': 1, + 'post-rewrite': 1, + 'pre-push': 2, +} + + +def _check_args_length(hook_type: str, args: Sequence[str]) -> None: + if hook_type == 'prepare-commit-msg': + if len(args) < 1 or len(args) > 3: + raise SystemExit( + f'hook-impl for {hook_type} expected 1, 2, or 3 arguments ' + f'but got {len(args)}: {args}', + ) + elif hook_type == 'pre-rebase': + if len(args) < 1 or len(args) > 2: + raise SystemExit( + f'hook-impl for {hook_type} expected 1 or 2 arguments ' + f'but got {len(args)}: {args}', + ) + elif hook_type in _EXPECTED_ARG_LENGTH_BY_HOOK: + expected = _EXPECTED_ARG_LENGTH_BY_HOOK[hook_type] + if len(args) != expected: + arguments_s = 'argument' if expected == 1 else 'arguments' + raise SystemExit( + f'hook-impl for {hook_type} expected {expected} {arguments_s} ' + f'but got {len(args)}: {args}', + ) + else: + raise AssertionError(f'unexpected hook type: {hook_type}') + + +def _run_ns( + hook_type: str, + color: bool, + args: Sequence[str], + stdin: bytes, +) -> argparse.Namespace | None: + _check_args_length(hook_type, args) + if hook_type == 'pre-push': + return _pre_push_ns(color, args, stdin) + elif hook_type in 'commit-msg': + return _ns(hook_type, color, commit_msg_filename=args[0]) + elif hook_type == 'prepare-commit-msg' and len(args) == 1: + return _ns(hook_type, color, commit_msg_filename=args[0]) + elif hook_type == 'prepare-commit-msg' and len(args) == 2: + return _ns( + hook_type, color, commit_msg_filename=args[0], + prepare_commit_message_source=args[1], + ) + elif hook_type == 'prepare-commit-msg' and len(args) == 3: + return _ns( + hook_type, color, commit_msg_filename=args[0], + prepare_commit_message_source=args[1], commit_object_name=args[2], + ) + elif hook_type in {'post-commit', 'pre-merge-commit', 'pre-commit'}: + return _ns(hook_type, color) + elif hook_type == 'post-checkout': + return _ns( + hook_type, color, + from_ref=args[0], to_ref=args[1], checkout_type=args[2], + ) + elif hook_type == 'post-merge': + return _ns(hook_type, color, is_squash_merge=args[0]) + elif hook_type == 'post-rewrite': + return _ns(hook_type, color, rewrite_command=args[0]) + elif hook_type == 'pre-rebase' and len(args) == 1: + return _ns(hook_type, color, pre_rebase_upstream=args[0]) + elif hook_type == 'pre-rebase' and len(args) == 2: + return _ns( + hook_type, color, pre_rebase_upstream=args[0], + pre_rebase_branch=args[1], + ) + else: + raise AssertionError(f'unexpected hook type: {hook_type}') + + +def hook_impl( + store: Store, + *, + config: str, + color: bool, + hook_type: str, + hook_dir: str, + skip_on_missing_config: bool, + args: Sequence[str], +) -> int: + retv, stdin = _run_legacy(hook_type, hook_dir, args) + _validate_config(retv, config, skip_on_missing_config) + ns = _run_ns(hook_type, color, args, stdin) + if ns is None: + return retv + else: + return retv | run(config, store, ns) diff --git a/pre_commit/commands/init_templatedir.py b/pre_commit/commands/init_templatedir.py new file mode 100644 index 0000000..08af656 --- /dev/null +++ b/pre_commit/commands/init_templatedir.py @@ -0,0 +1,39 @@ +from __future__ import annotations + +import logging +import os.path + +from pre_commit.commands.install_uninstall import install +from pre_commit.store import Store +from pre_commit.util import CalledProcessError +from pre_commit.util import cmd_output + +logger = logging.getLogger('pre_commit') + + +def init_templatedir( + config_file: str, + store: Store, + directory: str, + hook_types: list[str] | None, + skip_on_missing_config: bool = True, +) -> int: + install( + config_file, + store, + hook_types=hook_types, + overwrite=True, + skip_on_missing_config=skip_on_missing_config, + git_dir=directory, + ) + try: + _, out, _ = cmd_output('git', 'config', 'init.templateDir') + except CalledProcessError: + configured_path = None + else: + configured_path = os.path.realpath(os.path.expanduser(out.strip())) + dest = os.path.realpath(directory) + if configured_path != dest: + logger.warning('`init.templateDir` not set to the target directory') + logger.warning(f'maybe `git config --global init.templateDir {dest}`?') + return 0 diff --git a/pre_commit/commands/install_uninstall.py b/pre_commit/commands/install_uninstall.py new file mode 100644 index 0000000..d19e0d4 --- /dev/null +++ b/pre_commit/commands/install_uninstall.py @@ -0,0 +1,167 @@ +from __future__ import annotations + +import logging +import os.path +import shlex +import shutil +import sys + +from pre_commit import git +from pre_commit import output +from pre_commit.clientlib import InvalidConfigError +from pre_commit.clientlib import load_config +from pre_commit.repository import all_hooks +from pre_commit.repository import install_hook_envs +from pre_commit.store import Store +from pre_commit.util import make_executable +from pre_commit.util import resource_text + + +logger = logging.getLogger(__name__) + +# This is used to identify the hook file we install +PRIOR_HASHES = ( + b'4d9958c90bc262f47553e2c073f14cfe', + b'd8ee923c46731b42cd95cc869add4062', + b'49fd668cb42069aa1b6048464be5d395', + b'79f09a650522a87b0da915d0d983b2de', + b'e358c9dae00eac5d06b38dfdb1e33a8c', +) +CURRENT_HASH = b'138fd403232d2ddd5efb44317e38bf03' +TEMPLATE_START = '# start templated\n' +TEMPLATE_END = '# end templated\n' + + +def _hook_types(cfg_filename: str, hook_types: list[str] | None) -> list[str]: + if hook_types is not None: + return hook_types + else: + try: + cfg = load_config(cfg_filename) + except InvalidConfigError: + return ['pre-commit'] + else: + return cfg['default_install_hook_types'] + + +def _hook_paths( + hook_type: str, + git_dir: str | None = None, +) -> tuple[str, str]: + git_dir = git_dir if git_dir is not None else git.get_git_common_dir() + pth = os.path.join(git_dir, 'hooks', hook_type) + return pth, f'{pth}.legacy' + + +def is_our_script(filename: str) -> bool: + if not os.path.exists(filename): # pragma: win32 no cover (symlink) + return False + with open(filename, 'rb') as f: + contents = f.read() + return any(h in contents for h in (CURRENT_HASH,) + PRIOR_HASHES) + + +def _install_hook_script( + config_file: str, + hook_type: str, + overwrite: bool = False, + skip_on_missing_config: bool = False, + git_dir: str | None = None, +) -> None: + hook_path, legacy_path = _hook_paths(hook_type, git_dir=git_dir) + + os.makedirs(os.path.dirname(hook_path), exist_ok=True) + + # If we have an existing hook, move it to pre-commit.legacy + if os.path.lexists(hook_path) and not is_our_script(hook_path): + shutil.move(hook_path, legacy_path) + + # If we specify overwrite, we simply delete the legacy file + if overwrite and os.path.exists(legacy_path): + os.remove(legacy_path) + elif os.path.exists(legacy_path): + output.write_line( + f'Running in migration mode with existing hooks at {legacy_path}\n' + f'Use -f to use only pre-commit.', + ) + + args = ['hook-impl', f'--config={config_file}', f'--hook-type={hook_type}'] + if skip_on_missing_config: + args.append('--skip-on-missing-config') + + with open(hook_path, 'w') as hook_file: + contents = resource_text('hook-tmpl') + before, rest = contents.split(TEMPLATE_START) + _, after = rest.split(TEMPLATE_END) + + # on windows always use `/bin/sh` since `bash` might not be on PATH + # though we use bash-specific features `sh` on windows is actually + # bash in "POSIXLY_CORRECT" mode which still supports the features we + # use: subshells / arrays + if sys.platform == 'win32': # pragma: win32 cover + hook_file.write('#!/bin/sh\n') + + hook_file.write(before + TEMPLATE_START) + hook_file.write(f'INSTALL_PYTHON={shlex.quote(sys.executable)}\n') + args_s = shlex.join(args) + hook_file.write(f'ARGS=({args_s})\n') + hook_file.write(TEMPLATE_END + after) + make_executable(hook_path) + + output.write_line(f'pre-commit installed at {hook_path}') + + +def install( + config_file: str, + store: Store, + hook_types: list[str] | None, + overwrite: bool = False, + hooks: bool = False, + skip_on_missing_config: bool = False, + git_dir: str | None = None, +) -> int: + if git_dir is None and git.has_core_hookpaths_set(): + logger.error( + 'Cowardly refusing to install hooks with `core.hooksPath` set.\n' + 'hint: `git config --unset-all core.hooksPath`', + ) + return 1 + + for hook_type in _hook_types(config_file, hook_types): + _install_hook_script( + config_file, hook_type, + overwrite=overwrite, + skip_on_missing_config=skip_on_missing_config, + git_dir=git_dir, + ) + + if hooks: + install_hooks(config_file, store) + + return 0 + + +def install_hooks(config_file: str, store: Store) -> int: + install_hook_envs(all_hooks(load_config(config_file), store), store) + return 0 + + +def _uninstall_hook_script(hook_type: str) -> None: + hook_path, legacy_path = _hook_paths(hook_type) + + # If our file doesn't exist or it isn't ours, gtfo. + if not os.path.exists(hook_path) or not is_our_script(hook_path): + return + + os.remove(hook_path) + output.write_line(f'{hook_type} uninstalled') + + if os.path.exists(legacy_path): + os.replace(legacy_path, hook_path) + output.write_line(f'Restored previous hooks to {hook_path}') + + +def uninstall(config_file: str, hook_types: list[str] | None) -> int: + for hook_type in _hook_types(config_file, hook_types): + _uninstall_hook_script(hook_type) + return 0 diff --git a/pre_commit/commands/migrate_config.py b/pre_commit/commands/migrate_config.py new file mode 100644 index 0000000..842fb3a --- /dev/null +++ b/pre_commit/commands/migrate_config.py @@ -0,0 +1,75 @@ +from __future__ import annotations + +import re +import textwrap + +import cfgv +import yaml + +from pre_commit.clientlib import InvalidConfigError +from pre_commit.yaml import yaml_load + + +def _is_header_line(line: str) -> bool: + return line.startswith(('#', '---')) or not line.strip() + + +def _migrate_map(contents: str) -> str: + 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: + trial_contents = f'{header}repos:\n{rest}' + yaml_load(trial_contents) + contents = trial_contents + except yaml.YAMLError: + contents = f'{header}repos:\n{textwrap.indent(rest, " " * 4)}' + + return contents + + +def _migrate_sha_to_rev(contents: str) -> str: + return re.sub(r'(\n\s+)sha:', r'\1rev:', contents) + + +def _migrate_python_venv(contents: str) -> str: + return re.sub( + r'(\n\s+)language: python_venv\b', + r'\1language: python', + contents, + ) + + +def migrate_config(config_file: str, quiet: bool = False) -> int: + with open(config_file) as f: + orig_contents = contents = f.read() + + with cfgv.reraise_as(InvalidConfigError): + with cfgv.validate_context(f'File {config_file}'): + try: + yaml_load(orig_contents) + except Exception as e: + raise cfgv.ValidationError(str(e)) + + contents = _migrate_map(contents) + contents = _migrate_sha_to_rev(contents) + contents = _migrate_python_venv(contents) + + if contents != orig_contents: + with open(config_file, 'w') as f: + f.write(contents) + + print('Configuration has been migrated.') + elif not quiet: + print('Configuration is already migrated.') + return 0 diff --git a/pre_commit/commands/run.py b/pre_commit/commands/run.py new file mode 100644 index 0000000..076f16d --- /dev/null +++ b/pre_commit/commands/run.py @@ -0,0 +1,447 @@ +from __future__ import annotations + +import argparse +import contextlib +import functools +import logging +import os +import re +import subprocess +import time +import unicodedata +from collections.abc import Generator +from collections.abc import Iterable +from collections.abc import MutableMapping +from collections.abc import Sequence +from typing import Any + +from identify.identify import tags_from_path + +from pre_commit import color +from pre_commit import git +from pre_commit import output +from pre_commit.all_languages import languages +from pre_commit.clientlib import load_config +from pre_commit.hook import Hook +from pre_commit.repository import all_hooks +from pre_commit.repository import install_hook_envs +from pre_commit.staged_files_only import staged_files_only +from pre_commit.store import Store +from pre_commit.util import cmd_output_b + + +logger = logging.getLogger('pre_commit') + + +def _len_cjk(msg: str) -> int: + widths = {'A': 1, 'F': 2, 'H': 1, 'N': 1, 'Na': 1, 'W': 2} + return sum(widths[unicodedata.east_asian_width(c)] for c in msg) + + +def _start_msg(*, start: str, cols: int, end_len: int) -> str: + dots = '.' * (cols - _len_cjk(start) - end_len - 1) + return f'{start}{dots}' + + +def _full_msg( + *, + start: str, + cols: int, + end_msg: str, + end_color: str, + use_color: bool, + postfix: str = '', +) -> str: + dots = '.' * (cols - _len_cjk(start) - len(postfix) - len(end_msg) - 1) + end = color.format_color(end_msg, end_color, use_color) + return f'{start}{dots}{postfix}{end}\n' + + +def filter_by_include_exclude( + names: Iterable[str], + include: str, + exclude: str, +) -> Generator[str, None, None]: + include_re, exclude_re = re.compile(include), re.compile(exclude) + return ( + filename for filename in names + if include_re.search(filename) + if not exclude_re.search(filename) + ) + + +class Classifier: + def __init__(self, filenames: Iterable[str]) -> None: + self.filenames = [f for f in filenames if os.path.lexists(f)] + + @functools.cache + def _types_for_file(self, filename: str) -> set[str]: + return tags_from_path(filename) + + def by_types( + self, + names: Iterable[str], + types: Iterable[str], + types_or: Iterable[str], + exclude_types: Iterable[str], + ) -> Generator[str, None, None]: + types = frozenset(types) + types_or = frozenset(types_or) + exclude_types = frozenset(exclude_types) + for filename in names: + tags = self._types_for_file(filename) + if ( + tags >= types and + (not types_or or tags & types_or) and + not tags & exclude_types + ): + yield filename + + def filenames_for_hook(self, hook: Hook) -> Generator[str, None, None]: + return self.by_types( + filter_by_include_exclude( + self.filenames, + hook.files, + hook.exclude, + ), + hook.types, + hook.types_or, + hook.exclude_types, + ) + + @classmethod + def from_config( + cls, + filenames: Iterable[str], + include: str, + exclude: str, + ) -> Classifier: + # on windows we normalize all filenames to use forward slashes + # this makes it easier to filter using the `files:` regex + # this also makes improperly quoted shell-based hooks work better + # see #1173 + if os.altsep == '/' and os.sep == '\\': + filenames = (f.replace(os.sep, os.altsep) for f in filenames) + filenames = filter_by_include_exclude(filenames, include, exclude) + return Classifier(filenames) + + +def _get_skips(environ: MutableMapping[str, str]) -> set[str]: + skips = environ.get('SKIP', '') + return {skip.strip() for skip in skips.split(',') if skip.strip()} + + +SKIPPED = 'Skipped' +NO_FILES = '(no files to check)' + + +def _subtle_line(s: str, use_color: bool) -> None: + output.write_line(color.format_color(s, color.SUBTLE, use_color)) + + +def _run_single_hook( + classifier: Classifier, + hook: Hook, + skips: set[str], + cols: int, + diff_before: bytes, + verbose: bool, + use_color: bool, +) -> tuple[bool, bytes]: + filenames = tuple(classifier.filenames_for_hook(hook)) + + if hook.id in skips or hook.alias in skips: + output.write( + _full_msg( + start=hook.name, + end_msg=SKIPPED, + end_color=color.YELLOW, + use_color=use_color, + cols=cols, + ), + ) + duration = None + retcode = 0 + diff_after = diff_before + files_modified = False + out = b'' + elif not filenames and not hook.always_run: + output.write( + _full_msg( + start=hook.name, + postfix=NO_FILES, + end_msg=SKIPPED, + end_color=color.TURQUOISE, + use_color=use_color, + cols=cols, + ), + ) + duration = None + retcode = 0 + diff_after = diff_before + files_modified = False + out = b'' + else: + # print hook and dots first in case the hook takes a while to run + output.write(_start_msg(start=hook.name, end_len=6, cols=cols)) + + if not hook.pass_filenames: + filenames = () + time_before = time.monotonic() + language = languages[hook.language] + with language.in_env(hook.prefix, hook.language_version): + retcode, out = language.run_hook( + hook.prefix, + hook.entry, + hook.args, + filenames, + is_local=hook.src == 'local', + require_serial=hook.require_serial, + color=use_color, + ) + duration = round(time.monotonic() - time_before, 2) or 0 + diff_after = _get_diff() + + # if the hook makes changes, fail the commit + files_modified = diff_before != diff_after + + if retcode or files_modified: + print_color = color.RED + status = 'Failed' + else: + print_color = color.GREEN + status = 'Passed' + + output.write_line(color.format_color(status, print_color, use_color)) + + if verbose or hook.verbose or retcode or files_modified: + _subtle_line(f'- hook id: {hook.id}', use_color) + + if (verbose or hook.verbose) and duration is not None: + _subtle_line(f'- duration: {duration}s', use_color) + + if retcode: + _subtle_line(f'- exit code: {retcode}', use_color) + + # Print a message if failing due to file modifications + if files_modified: + _subtle_line('- files were modified by this hook', use_color) + + if out.strip(): + output.write_line() + output.write_line_b(out.strip(), logfile_name=hook.log_file) + output.write_line() + + return files_modified or bool(retcode), diff_after + + +def _compute_cols(hooks: Sequence[Hook]) -> int: + """Compute the number of columns to display hook messages. The widest + that will be displayed is in the no files skipped case: + + Hook name...(no files to check) Skipped + """ + if hooks: + name_len = max(_len_cjk(hook.name) for hook in hooks) + else: + name_len = 0 + + cols = name_len + 3 + len(NO_FILES) + 1 + len(SKIPPED) + return max(cols, 80) + + +def _all_filenames(args: argparse.Namespace) -> Iterable[str]: + # these hooks do not operate on files + if args.hook_stage in { + 'post-checkout', 'post-commit', 'post-merge', 'post-rewrite', + 'pre-rebase', + }: + return () + elif args.hook_stage in {'prepare-commit-msg', 'commit-msg'}: + return (args.commit_msg_filename,) + elif args.from_ref and args.to_ref: + return git.get_changed_files(args.from_ref, args.to_ref) + elif args.files: + return args.files + elif args.all_files: + return git.get_all_files() + elif git.is_in_merge_conflict(): + return git.get_conflicted_files() + else: + return git.get_staged_files() + + +def _get_diff() -> bytes: + _, out, _ = cmd_output_b( + 'git', 'diff', '--no-ext-diff', '--no-textconv', '--ignore-submodules', + check=False, + ) + return out + + +def _run_hooks( + config: dict[str, Any], + hooks: Sequence[Hook], + skips: set[str], + args: argparse.Namespace, +) -> int: + """Actually run the hooks.""" + cols = _compute_cols(hooks) + classifier = Classifier.from_config( + _all_filenames(args), config['files'], config['exclude'], + ) + retval = 0 + prior_diff = _get_diff() + for hook in hooks: + current_retval, prior_diff = _run_single_hook( + classifier, hook, skips, cols, prior_diff, + verbose=args.verbose, use_color=args.color, + ) + retval |= current_retval + if retval and (config['fail_fast'] or hook.fail_fast): + break + if retval and args.show_diff_on_failure and prior_diff: + if args.all_files: + output.write_line( + 'pre-commit hook(s) made changes.\n' + 'If you are seeing this message in CI, ' + 'reproduce locally with: `pre-commit run --all-files`.\n' + 'To run `pre-commit` as part of git workflow, use ' + '`pre-commit install`.', + ) + output.write_line('All changes made by hooks:') + # args.color is a boolean. + # See user_color function in color.py + git_color_opt = 'always' if args.color else 'never' + subprocess.call(( + 'git', '--no-pager', 'diff', '--no-ext-diff', + f'--color={git_color_opt}', + )) + + return retval + + +def _has_unmerged_paths() -> bool: + _, stdout, _ = cmd_output_b('git', 'ls-files', '--unmerged') + return bool(stdout.strip()) + + +def _has_unstaged_config(config_file: str) -> bool: + retcode, _, _ = cmd_output_b( + 'git', 'diff', '--quiet', '--no-ext-diff', config_file, check=False, + ) + # be explicit, other git errors don't mean it has an unstaged config. + return retcode == 1 + + +def run( + config_file: str, + store: Store, + args: argparse.Namespace, + environ: MutableMapping[str, str] = os.environ, +) -> int: + stash = not args.all_files and not args.files + + # Check if we have unresolved merge conflict files and fail fast. + if stash and _has_unmerged_paths(): + logger.error('Unmerged files. Resolve before committing.') + return 1 + if bool(args.from_ref) != bool(args.to_ref): + logger.error('Specify both --from-ref and --to-ref.') + return 1 + if stash and _has_unstaged_config(config_file): + logger.error( + f'Your pre-commit configuration is unstaged.\n' + f'`git add {config_file}` to fix this.', + ) + return 1 + if ( + args.hook_stage in {'prepare-commit-msg', 'commit-msg'} and + not args.commit_msg_filename + ): + logger.error( + f'`--commit-msg-filename` is required for ' + f'`--hook-stage {args.hook_stage}`', + ) + return 1 + # prevent recursive post-checkout hooks (#1418) + if ( + args.hook_stage == 'post-checkout' and + environ.get('_PRE_COMMIT_SKIP_POST_CHECKOUT') + ): + return 0 + + # Expose prepare_commit_message_source / commit_object_name + # as environment variables for the hooks + if args.prepare_commit_message_source: + environ['PRE_COMMIT_COMMIT_MSG_SOURCE'] = ( + args.prepare_commit_message_source + ) + + if args.commit_object_name: + environ['PRE_COMMIT_COMMIT_OBJECT_NAME'] = args.commit_object_name + + # Expose from-ref / to-ref as environment variables for hooks to consume + if args.from_ref and args.to_ref: + # legacy names + environ['PRE_COMMIT_ORIGIN'] = args.from_ref + environ['PRE_COMMIT_SOURCE'] = args.to_ref + # new names + environ['PRE_COMMIT_FROM_REF'] = args.from_ref + environ['PRE_COMMIT_TO_REF'] = args.to_ref + + if args.pre_rebase_upstream and args.pre_rebase_branch: + environ['PRE_COMMIT_PRE_REBASE_UPSTREAM'] = args.pre_rebase_upstream + environ['PRE_COMMIT_PRE_REBASE_BRANCH'] = args.pre_rebase_branch + + if ( + args.remote_name and args.remote_url and + args.remote_branch and args.local_branch + ): + environ['PRE_COMMIT_LOCAL_BRANCH'] = args.local_branch + environ['PRE_COMMIT_REMOTE_BRANCH'] = args.remote_branch + environ['PRE_COMMIT_REMOTE_NAME'] = args.remote_name + environ['PRE_COMMIT_REMOTE_URL'] = args.remote_url + + if args.checkout_type: + environ['PRE_COMMIT_CHECKOUT_TYPE'] = args.checkout_type + + if args.is_squash_merge: + environ['PRE_COMMIT_IS_SQUASH_MERGE'] = args.is_squash_merge + + if args.rewrite_command: + environ['PRE_COMMIT_REWRITE_COMMAND'] = args.rewrite_command + + # Set pre_commit flag + environ['PRE_COMMIT'] = '1' + + with contextlib.ExitStack() as exit_stack: + if stash: + exit_stack.enter_context(staged_files_only(store.directory)) + + config = load_config(config_file) + hooks = [ + hook + for hook in all_hooks(config, store) + if not args.hook or hook.id == args.hook or hook.alias == args.hook + if args.hook_stage in hook.stages + ] + + if args.hook and not hooks: + output.write_line( + f'No hook with id `{args.hook}` in stage `{args.hook_stage}`', + ) + return 1 + + skips = _get_skips(environ) + to_install = [ + hook + for hook in hooks + if hook.id not in skips and hook.alias not in skips + ] + install_hook_envs(to_install, store) + + return _run_hooks(config, hooks, skips, args) + + # https://github.com/python/mypy/issues/7726 + raise AssertionError('unreachable') diff --git a/pre_commit/commands/sample_config.py b/pre_commit/commands/sample_config.py new file mode 100644 index 0000000..ce22f65 --- /dev/null +++ b/pre_commit/commands/sample_config.py @@ -0,0 +1,18 @@ +from __future__ import annotations +SAMPLE_CONFIG = '''\ +# See https://pre-commit.com for more information +# See https://pre-commit.com/hooks.html for more hooks +repos: +- repo: https://github.com/pre-commit/pre-commit-hooks + rev: v3.2.0 + hooks: + - id: trailing-whitespace + - id: end-of-file-fixer + - id: check-yaml + - id: check-added-large-files +''' + + +def sample_config() -> int: + print(SAMPLE_CONFIG, end='') + return 0 diff --git a/pre_commit/commands/try_repo.py b/pre_commit/commands/try_repo.py new file mode 100644 index 0000000..539ed3c --- /dev/null +++ b/pre_commit/commands/try_repo.py @@ -0,0 +1,77 @@ +from __future__ import annotations + +import argparse +import logging +import os.path +import tempfile + +import pre_commit.constants as C +from pre_commit import git +from pre_commit import output +from pre_commit.clientlib import load_manifest +from pre_commit.commands.run import run +from pre_commit.store import Store +from pre_commit.util import cmd_output_b +from pre_commit.xargs import xargs +from pre_commit.yaml import yaml_dump + +logger = logging.getLogger(__name__) + + +def _repo_ref(tmpdir: str, repo: str, ref: str | None) -> tuple[str, str]: + # if `ref` is explicitly passed, use it + if ref is not None: + return repo, ref + + ref = git.head_rev(repo) + # if it exists on disk, we'll try and clone it with the local changes + if os.path.exists(repo) and git.has_diff('HEAD', repo=repo): + logger.warning('Creating temporary repo with uncommitted changes...') + + shadow = os.path.join(tmpdir, 'shadow-repo') + cmd_output_b('git', 'clone', repo, shadow) + cmd_output_b('git', 'checkout', ref, '-b', '_pc_tmp', cwd=shadow) + + idx = git.git_path('index', repo=shadow) + objs = git.git_path('objects', repo=shadow) + env = dict(os.environ, GIT_INDEX_FILE=idx, GIT_OBJECT_DIRECTORY=objs) + + staged_files = git.get_staged_files(cwd=repo) + if staged_files: + xargs(('git', 'add', '--'), staged_files, cwd=repo, env=env) + + cmd_output_b('git', 'add', '-u', cwd=repo, env=env) + git.commit(repo=shadow) + + return shadow, git.head_rev(shadow) + else: + return repo, ref + + +def try_repo(args: argparse.Namespace) -> int: + with tempfile.TemporaryDirectory() as tempdir: + repo, ref = _repo_ref(tempdir, args.repo, args.ref) + + store = Store(tempdir) + if args.hook: + hooks = [{'id': args.hook}] + else: + repo_path = store.clone(repo, ref) + manifest = load_manifest(os.path.join(repo_path, C.MANIFEST_FILE)) + manifest = sorted(manifest, key=lambda hook: hook['id']) + hooks = [{'id': hook['id']} for hook in manifest] + + config = {'repos': [{'repo': repo, 'rev': ref, 'hooks': hooks}]} + config_s = yaml_dump(config) + + config_filename = os.path.join(tempdir, C.CONFIG_FILE) + with open(config_filename, 'w') as cfg: + cfg.write(config_s) + + output.write_line('=' * 79) + output.write_line('Using config:') + output.write_line('=' * 79) + output.write(config_s) + output.write_line('=' * 79) + + return run(config_filename, store, args) diff --git a/pre_commit/commands/validate_config.py b/pre_commit/commands/validate_config.py new file mode 100644 index 0000000..b3de635 --- /dev/null +++ b/pre_commit/commands/validate_config.py @@ -0,0 +1,18 @@ +from __future__ import annotations + +from collections.abc import Sequence + +from pre_commit import clientlib + + +def validate_config(filenames: Sequence[str]) -> int: + ret = 0 + + for filename in filenames: + try: + clientlib.load_config(filename) + except clientlib.InvalidConfigError as e: + print(e) + ret = 1 + + return ret diff --git a/pre_commit/commands/validate_manifest.py b/pre_commit/commands/validate_manifest.py new file mode 100644 index 0000000..8493c6e --- /dev/null +++ b/pre_commit/commands/validate_manifest.py @@ -0,0 +1,18 @@ +from __future__ import annotations + +from collections.abc import Sequence + +from pre_commit import clientlib + + +def validate_manifest(filenames: Sequence[str]) -> int: + ret = 0 + + for filename in filenames: + try: + clientlib.load_manifest(filename) + except clientlib.InvalidManifestError as e: + print(e) + ret = 1 + + return ret -- cgit v1.2.3