pre-commit-ex 4.5.1__py2.py3-none-any.whl

pre_commit/error_handler.py

@@ -0,0 +1,81 @@
+from __future__ import annotations
+
+import contextlib
+import functools
+import os.path
+import sys
+import traceback
+from collections.abc import Generator
+from typing import IO
+
+import pre_commit.constants as C
+from pre_commit import output
+from pre_commit.errors import FatalError
+from pre_commit.store import Store
+from pre_commit.util import cmd_output_b
+from pre_commit.util import force_bytes
+
+
+def _log_and_exit(
+    msg: str,
+    ret_code: int,
+    exc: BaseException,
+    formatted: str,
+) -> None:
+    error_msg = f'{msg}: {type(exc).__name__}: '.encode() + force_bytes(exc)
+    output.write_line_b(error_msg)
+
+    _, git_version_b, _ = cmd_output_b('git', '--version', check=False)
+    git_version = git_version_b.decode(errors='backslashreplace').rstrip()
+
+    storedir = Store().directory
+    log_path = os.path.join(storedir, 'pre-commit.log')
+    with contextlib.ExitStack() as ctx:
+        if os.access(storedir, os.W_OK):
+            output.write_line(f'Check the log at {log_path}')
+            log: IO[bytes] = ctx.enter_context(open(log_path, 'wb'))
+        else:  # pragma: win32 no cover
+            output.write_line(f'Failed to write to log at {log_path}')
+            log = sys.stdout.buffer
+
+        _log_line = functools.partial(output.write_line, stream=log)
+        _log_line_b = functools.partial(output.write_line_b, stream=log)
+
+        _log_line('### version information')
+        _log_line()
+        _log_line('```')
+        _log_line(f'pre-commit version: {C.VERSION}')
+        _log_line(f'git --version: {git_version}')
+        _log_line('sys.version:')
+        for line in sys.version.splitlines():
+            _log_line(f'    {line}')
+        _log_line(f'sys.executable: {sys.executable}')
+        _log_line(f'os.name: {os.name}')
+        _log_line(f'sys.platform: {sys.platform}')
+        _log_line('```')
+        _log_line()
+
+        _log_line('### error information')
+        _log_line()
+        _log_line('```')
+        _log_line_b(error_msg)
+        _log_line('```')
+        _log_line()
+        _log_line('```')
+        _log_line(formatted.rstrip())
+        _log_line('```')
+    raise SystemExit(ret_code)
+
+
+@contextlib.contextmanager
+def error_handler() -> Generator[None]:
+    try:
+        yield
+    except (Exception, KeyboardInterrupt) as e:
+        if isinstance(e, FatalError):
+            msg, ret_code = 'An error has occurred', 1
+        elif isinstance(e, KeyboardInterrupt):
+            msg, ret_code = 'Interrupted (^C)', 130
+        else:
+            msg, ret_code = 'An unexpected error has occurred', 3
+        _log_and_exit(msg, ret_code, e, traceback.format_exc())

pre_commit/errors.py

@@ -0,0 +1,5 @@
+from __future__ import annotations
+
+
+class FatalError(RuntimeError):
+    pass

pre_commit/file_lock.py

@@ -0,0 +1,75 @@
+from __future__ import annotations
+
+import contextlib
+import errno
+import sys
+from collections.abc import Callable
+from collections.abc import Generator
+
+
+if sys.platform == 'win32':  # pragma: no cover (windows)
+    import msvcrt
+
+    # https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/locking
+
+    # on windows we lock "regions" of files, we don't care about the actual
+    # byte region so we'll just pick *some* number here.
+    _region = 0xffff
+
+    @contextlib.contextmanager
+    def _locked(
+            fileno: int,
+            blocked_cb: Callable[[], None],
+    ) -> Generator[None]:
+        try:
+            msvcrt.locking(fileno, msvcrt.LK_NBLCK, _region)
+        except OSError:
+            blocked_cb()
+            while True:
+                try:
+                    msvcrt.locking(fileno, msvcrt.LK_LOCK, _region)
+                except OSError as e:
+                    # Locking violation. Returned when the _LK_LOCK or _LK_RLCK
+                    # flag is specified and the file cannot be locked after 10
+                    # attempts.
+                    if e.errno != errno.EDEADLOCK:
+                        raise
+                else:
+                    break
+
+        try:
+            yield
+        finally:
+            # From cursory testing, it seems to get unlocked when the file is
+            # closed so this may not be necessary.
+            # The documentation however states:
+            # "Regions should be locked only briefly and should be unlocked
+            # before closing a file or exiting the program."
+            msvcrt.locking(fileno, msvcrt.LK_UNLCK, _region)
+else:  # pragma: win32 no cover
+    import fcntl
+
+    @contextlib.contextmanager
+    def _locked(
+            fileno: int,
+            blocked_cb: Callable[[], None],
+    ) -> Generator[None]:
+        try:
+            fcntl.flock(fileno, fcntl.LOCK_EX | fcntl.LOCK_NB)
+        except OSError:  # pragma: no cover (tests are single-threaded)
+            blocked_cb()
+            fcntl.flock(fileno, fcntl.LOCK_EX)
+        try:
+            yield
+        finally:
+            fcntl.flock(fileno, fcntl.LOCK_UN)
+
+
+@contextlib.contextmanager
+def lock(
+        path: str,
+        blocked_cb: Callable[[], None],
+) -> Generator[None]:
+    with open(path, 'a+') as f:
+        with _locked(f.fileno(), blocked_cb):
+            yield

pre_commit/git.py

@@ -0,0 +1,245 @@
+from __future__ import annotations
+
+import logging
+import os.path
+import sys
+from collections.abc import Mapping
+
+from pre_commit.errors import FatalError
+from pre_commit.util import CalledProcessError
+from pre_commit.util import cmd_output
+from pre_commit.util import cmd_output_b
+
+logger = logging.getLogger(__name__)
+
+# see #2046
+NO_FS_MONITOR = ('-c', 'core.useBuiltinFSMonitor=false')
+
+
+def zsplit(s: str) -> list[str]:
+    s = s.strip('\0')
+    if s:
+        return s.split('\0')
+    else:
+        return []
+
+
+def no_git_env(_env: Mapping[str, str] | None = None) -> dict[str, str]:
+    # Too many bugs dealing with environment variables and GIT:
+    # https://github.com/pre-commit/pre-commit/issues/300
+    # In git 2.6.3 (maybe others), git exports GIT_WORK_TREE while running
+    # pre-commit hooks
+    # In git 1.9.1 (maybe others), git exports GIT_DIR and GIT_INDEX_FILE
+    # while running pre-commit hooks in submodules.
+    # GIT_DIR: Causes git clone to clone wrong thing
+    # GIT_INDEX_FILE: Causes 'error invalid object ...' during commit
+    _env = _env if _env is not None else os.environ
+    return {
+        k: v for k, v in _env.items()
+        if not k.startswith('GIT_') or
+        k.startswith(('GIT_CONFIG_KEY_', 'GIT_CONFIG_VALUE_')) or
+        k in {
+            'GIT_EXEC_PATH', 'GIT_SSH', 'GIT_SSH_COMMAND', 'GIT_SSL_CAINFO',
+            'GIT_SSL_NO_VERIFY', 'GIT_CONFIG_COUNT',
+            'GIT_HTTP_PROXY_AUTHMETHOD',
+            'GIT_ALLOW_PROTOCOL',
+            'GIT_ASKPASS',
+        }
+    }
+
+
+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 = os.path.abspath(
+            cmd_output('git', 'rev-parse', '--show-cdup')[1].strip(),
+        )
+        inside_git_dir = cmd_output(
+            'git', 'rev-parse', '--is-inside-git-dir',
+        )[1].strip()
+    except CalledProcessError:
+        raise FatalError(
+            'git failed. Is it installed, and are you in a Git repository '
+            'directory?',
+        )
+    if inside_git_dir != 'false':
+        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:
+    opt = '--git-dir'
+    _, out, _ = cmd_output('git', 'rev-parse', opt, cwd=git_root)
+    git_dir = out.strip()
+    if git_dir != opt:
+        return os.path.normpath(os.path.join(git_root, git_dir))
+    else:
+        raise AssertionError('unreachable: no git dir')
+
+
+def get_git_common_dir(git_root: str = '.') -> str:
+    opt = '--git-common-dir'
+    _, out, _ = cmd_output('git', 'rev-parse', opt, cwd=git_root)
+    git_common_dir = out.strip()
+    if git_common_dir != opt:
+        return os.path.normpath(os.path.join(git_root, git_common_dir))
+    else:  # pragma: no cover (git < 2.5)
+        return get_git_dir(git_root)
+
+
+def is_in_merge_conflict() -> bool:
+    git_dir = get_git_dir('.')
+    return (
+        os.path.exists(os.path.join(git_dir, 'MERGE_MSG')) and
+        os.path.exists(os.path.join(git_dir, 'MERGE_HEAD'))
+    )
+
+
+def parse_merge_msg_for_conflicts(merge_msg: bytes) -> list[str]:
+    # Conflicted files start with tabs
+    return [
+        line.lstrip(b'#').strip().decode()
+        for line in merge_msg.splitlines()
+        # '#\t' for git 2.4.1
+        if line.startswith((b'\t', b'#\t'))
+    ]
+
+
+def get_conflicted_files() -> set[str]:
+    logger.info('Checking merge-conflict files only.')
+    # Need to get the conflicted files from the MERGE_MSG because they could
+    # have resolved the conflict by choosing one side or the other
+    with open(os.path.join(get_git_dir('.'), 'MERGE_MSG'), 'rb') as f:
+        merge_msg = f.read()
+    merge_conflict_filenames = parse_merge_msg_for_conflicts(merge_msg)
+
+    # This will get the rest of the changes made after the merge.
+    # If they resolved the merge conflict by choosing a mesh of both sides
+    # this will also include the conflicted files
+    tree_hash = cmd_output('git', 'write-tree')[1].strip()
+    merge_diff_filenames = zsplit(
+        cmd_output(
+            'git', 'diff', '--name-only', '--no-ext-diff', '-z',
+            '-m', tree_hash, 'HEAD', 'MERGE_HEAD', '--',
+        )[1],
+    )
+    return set(merge_conflict_filenames) | set(merge_diff_filenames)
+
+
+def get_staged_files(cwd: str | None = None) -> list[str]:
+    return zsplit(
+        cmd_output(
+            'git', 'diff', '--staged', '--name-only', '--no-ext-diff', '-z',
+            # Everything except for D
+            '--diff-filter=ACMRTUXB',
+            cwd=cwd,
+        )[1],
+    )
+
+
+def intent_to_add_files() -> list[str]:
+    _, stdout, _ = cmd_output(
+        'git', 'diff', '--no-ext-diff', '--ignore-submodules',
+        '--diff-filter=A', '--name-only', '-z',
+    )
+    return zsplit(stdout)
+
+
+def get_all_files() -> list[str]:
+    return zsplit(cmd_output('git', 'ls-files', '-z')[1])
+
+
+def get_changed_files(old: str, new: str) -> list[str]:
+    diff_cmd = ('git', 'diff', '--name-only', '--no-ext-diff', '-z')
+    try:
+        _, out, _ = cmd_output(*diff_cmd, f'{old}...{new}')
+    except CalledProcessError:  # pragma: no cover (new git)
+        # on newer git where old and new do not have a merge base git fails
+        # so we try a full diff (this is what old git did for us!)
+        _, out, _ = cmd_output(*diff_cmd, f'{old}..{new}')
+
+    return zsplit(out)
+
+
+def head_rev(remote: str) -> str:
+    _, out, _ = cmd_output('git', 'ls-remote', '--exit-code', remote, 'HEAD')
+    return out.split()[0]
+
+
+def has_diff(*args: str, repo: str = '.') -> bool:
+    cmd = ('git', 'diff', '--quiet', '--no-ext-diff', *args)
+    return cmd_output_b(*cmd, cwd=repo, check=False)[0] == 1
+
+
+def has_core_hookpaths_set() -> bool:
+    _, out, _ = cmd_output_b('git', 'config', 'core.hooksPath', check=False)
+    return bool(out.strip())
+
+
+def init_repo(path: str, remote: str) -> None:
+    if os.path.isdir(remote):
+        remote = os.path.abspath(remote)
+
+    git = ('git', *NO_FS_MONITOR)
+    env = no_git_env()
+    # avoid the user's template so that hooks do not recurse
+    cmd_output_b(*git, 'init', '--template=', path, env=env)
+    cmd_output_b(*git, 'remote', 'add', 'origin', remote, cwd=path, env=env)
+
+
+def commit(repo: str = '.') -> None:
+    env = no_git_env()
+    name, email = 'pre-commit', 'asottile+pre-commit@umich.edu'
+    env['GIT_AUTHOR_NAME'] = env['GIT_COMMITTER_NAME'] = name
+    env['GIT_AUTHOR_EMAIL'] = env['GIT_COMMITTER_EMAIL'] = email
+    cmd = ('git', 'commit', '--no-edit', '--no-gpg-sign', '-n', '-minit')
+    cmd_output_b(*cmd, cwd=repo, env=env)
+
+
+def git_path(name: str, repo: str = '.') -> str:
+    _, out, _ = cmd_output('git', 'rev-parse', '--git-path', name, cwd=repo)
+    return os.path.join(repo, out.strip())
+
+
+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'
+        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:
+            exe_type = {True: '(cygwin)', False: '(windows)'}
+            logger.warning(
+                f'pre-commit has detected a mix of cygwin python / git\n'
+                f'This combination is not supported, it is likely you will '
+                f'receive an error later in the program.\n'
+                f'Make sure to use cygwin git+python while using cygwin\n'
+                f'These can be installed through the cygwin installer.\n'
+                f' - python {exe_type[is_cygwin_python]}\n'
+                f' - git {exe_type[is_cygwin_git]}\n',
+            )
+
+
+def get_best_candidate_tag(rev: str, git_repo: str) -> str:
+    """Get the best tag candidate.
+
+    Multiple tags can exist on a SHA. Sometimes a moving tag is attached
+    to a version tag. Try to pick the tag that looks like a version.
+    """
+    tags = cmd_output(
+        'git', *NO_FS_MONITOR, 'tag', '--points-at', rev, cwd=git_repo,
+    )[1].splitlines()
+    for tag in tags:
+        if '.' in tag:
+            return tag
+    return rev

pre_commit/hook.py

@@ -0,0 +1,60 @@
+from __future__ import annotations
+
+import logging
+from collections.abc import Sequence
+from typing import Any
+from typing import NamedTuple
+
+from pre_commit.prefix import Prefix
+
+logger = logging.getLogger('pre_commit')
+
+
+class Hook(NamedTuple):
+    src: str
+    prefix: Prefix
+    id: str
+    name: str
+    entry: str
+    language: str
+    alias: str
+    files: str
+    exclude: str
+    types: Sequence[str]
+    types_or: Sequence[str]
+    exclude_types: Sequence[str]
+    additional_dependencies: Sequence[str]
+    args: Sequence[str]
+    always_run: bool
+    fail_fast: bool
+    pass_filenames: bool
+    description: str
+    language_version: str
+    log_file: str
+    minimum_pre_commit_version: str
+    require_serial: bool
+    stages: Sequence[str]
+    verbose: bool
+
+    @property
+    def install_key(self) -> tuple[Prefix, str, str, tuple[str, ...]]:
+        return (
+            self.prefix,
+            self.language,
+            self.language_version,
+            tuple(self.additional_dependencies),
+        )
+
+    @classmethod
+    def create(cls, src: str, prefix: Prefix, dct: dict[str, Any]) -> Hook:
+        # TODO: have cfgv do this (?)
+        extra_keys = set(dct) - _KEYS
+        if extra_keys:
+            logger.warning(
+                f'Unexpected key(s) present on {src} => {dct["id"]}: '
+                f'{", ".join(sorted(extra_keys))}',
+            )
+        return cls(src=src, prefix=prefix, **{k: dct[k] for k in _KEYS})
+
+
+_KEYS = frozenset(set(Hook._fields) - {'src', 'prefix'})

pre_commit/lang_base.py

@@ -0,0 +1,196 @@
+from __future__ import annotations
+
+import contextlib
+import os
+import random
+import re
+import shlex
+import sys
+from collections.abc import Generator
+from collections.abc import Sequence
+from typing import Any
+from typing import ContextManager
+from typing import NoReturn
+from typing import Protocol
+
+import pre_commit.constants as C
+from pre_commit import parse_shebang
+from pre_commit import xargs
+from pre_commit.prefix import Prefix
+from pre_commit.util import cmd_output_b
+
+FIXED_RANDOM_SEED = 1542676187
+
+SHIMS_RE = re.compile(r'[/\\]shims[/\\]')
+
+
+class Language(Protocol):
+    # Use `None` for no installation / environment
+    @property
+    def ENVIRONMENT_DIR(self) -> str | None: ...
+    # return a value to replace `'default` for `language_version`
+    def get_default_version(self) -> str: ...
+    # return whether the environment is healthy (or should be rebuilt)
+    def health_check(self, prefix: Prefix, version: str) -> str | None: ...
+
+    # install a repository for the given language and language_version
+    def install_environment(
+            self,
+            prefix: Prefix,
+            version: str,
+            additional_dependencies: Sequence[str],
+    ) -> None:
+        ...
+
+    # modify the environment for hook execution
+    def in_env(self, prefix: Prefix, version: str) -> ContextManager[None]: ...
+
+    # execute a hook and return the exit code and output
+    def run_hook(
+            self,
+            prefix: Prefix,
+            entry: str,
+            args: Sequence[str],
+            file_args: Sequence[str],
+            *,
+            is_local: bool,
+            require_serial: bool,
+            color: bool,
+    ) -> tuple[int, bytes]:
+        ...
+
+
+def exe_exists(exe: str) -> bool:
+    found = parse_shebang.find_executable(exe)
+    if found is None:  # exe exists
+        return False
+
+    homedir = os.path.expanduser('~')
+    try:
+        common: str | None = os.path.commonpath((found, homedir))
+    except ValueError:  # on windows, different drives raises ValueError
+        common = None
+
+    return (
+        # it is not in a /shims/ directory
+        not SHIMS_RE.search(found) and
+        (
+            # the homedir is / (docker, service user, etc.)
+            os.path.dirname(homedir) == homedir or
+            # the exe is not contained in the home directory
+            common != homedir
+        )
+    )
+
+
+def setup_cmd(prefix: Prefix, cmd: tuple[str, ...], **kwargs: Any) -> None:
+    cmd_output_b(*cmd, cwd=prefix.prefix_dir, **kwargs)
+
+
+def environment_dir(prefix: Prefix, d: str, language_version: str) -> str:
+    return prefix.path(f'{d}-{language_version}')
+
+
+def assert_version_default(binary: str, version: str) -> None:
+    if version != C.DEFAULT:
+        raise AssertionError(
+            f'for now, pre-commit requires system-installed {binary} -- '
+            f'you selected `language_version: {version}`',
+        )
+
+
+def assert_no_additional_deps(
+        lang: str,
+        additional_deps: Sequence[str],
+) -> None:
+    if additional_deps:
+        raise AssertionError(
+            f'for now, pre-commit does not support '
+            f'additional_dependencies for {lang} -- '
+            f'you selected `additional_dependencies: {additional_deps}`',
+        )
+
+
+def basic_get_default_version() -> str:
+    return C.DEFAULT
+
+
+def basic_health_check(prefix: Prefix, language_version: str) -> str | None:
+    return None
+
+
+def no_install(
+        prefix: Prefix,
+        version: str,
+        additional_dependencies: Sequence[str],
+) -> NoReturn:
+    raise AssertionError('This language is not installable')
+
+
+@contextlib.contextmanager
+def no_env(prefix: Prefix, version: str) -> Generator[None]:
+    yield
+
+
+def target_concurrency() -> int:
+    if 'PRE_COMMIT_NO_CONCURRENCY' in os.environ:
+        return 1
+    else:
+        # Travis appears to have a bunch of CPUs, but we can't use them all.
+        if 'TRAVIS' in os.environ:
+            return 2
+        else:
+            return xargs.cpu_count()
+
+
+def _shuffled(seq: Sequence[str]) -> list[str]:
+    """Deterministically shuffle"""
+    fixed_random = random.Random()
+    fixed_random.seed(FIXED_RANDOM_SEED, version=1)
+
+    seq = list(seq)
+    fixed_random.shuffle(seq)
+    return seq
+
+
+def run_xargs(
+        cmd: tuple[str, ...],
+        file_args: Sequence[str],
+        *,
+        require_serial: bool,
+        color: bool,
+) -> tuple[int, bytes]:
+    if require_serial:
+        jobs = 1
+    else:
+        # Shuffle the files so that they more evenly fill out the xargs
+        # partitions, but do it deterministically in case a hook cares about
+        # ordering.
+        file_args = _shuffled(file_args)
+        jobs = target_concurrency()
+    return xargs.xargs(cmd, file_args, target_concurrency=jobs, color=color)
+
+
+def hook_cmd(entry: str, args: Sequence[str]) -> tuple[str, ...]:
+    cmd = shlex.split(entry)
+    if cmd[:2] == ['pre-commit', 'hazmat']:
+        cmd = [sys.executable, '-m', 'pre_commit.commands.hazmat', *cmd[2:]]
+    return (*cmd, *args)
+
+
+def basic_run_hook(
+        prefix: Prefix,
+        entry: str,
+        args: Sequence[str],
+        file_args: Sequence[str],
+        *,
+        is_local: bool,
+        require_serial: bool,
+        color: bool,
+) -> tuple[int, bytes]:
+    return run_xargs(
+        hook_cmd(entry, args),
+        file_args,
+        require_serial=require_serial,
+        color=color,
+    )