Merge pull request #3585 from pre-commit/hazmat

asottile · web-flow · commit 3358a3b54077 · 2025-11-22T14:03:09.000-05:00
add pre-commit hazmat
diff --git a/pre_commit/commands/hazmat.py b/pre_commit/commands/hazmat.py
@@ -0,0 +1,95 @@
+from __future__ import annotations
+
+import argparse
+import subprocess
+from collections.abc import Sequence
+
+from pre_commit.parse_shebang import normalize_cmd
+
+
+def add_parsers(parser: argparse.ArgumentParser) -> None:
+    subparsers = parser.add_subparsers(dest='tool')
+
+    cd_parser = subparsers.add_parser(
+        'cd', help='cd to a subdir and run the command',
+    )
+    cd_parser.add_argument('subdir')
+    cd_parser.add_argument('cmd', nargs=argparse.REMAINDER)
+
+    ignore_exit_code_parser = subparsers.add_parser(
+        'ignore-exit-code', help='run the command but ignore the exit code',
+    )
+    ignore_exit_code_parser.add_argument('cmd', nargs=argparse.REMAINDER)
+
+    n1_parser = subparsers.add_parser(
+        'n1', help='run the command once per filename',
+    )
+    n1_parser.add_argument('cmd', nargs=argparse.REMAINDER)
+
+
+def _cmd_filenames(cmd: tuple[str, ...]) -> tuple[
+    tuple[str, ...],
+    tuple[str, ...],
+]:
+    for idx, val in enumerate(reversed(cmd)):
+        if val == '--':
+            split = len(cmd) - idx
+            break
+    else:
+        raise SystemExit('hazmat entry must end with `--`')
+
+    return cmd[:split - 1], cmd[split:]
+
+
+def cd(subdir: str, cmd: tuple[str, ...]) -> int:
+    cmd, filenames = _cmd_filenames(cmd)
+
+    prefix = f'{subdir}/'
+    new_filenames = []
+    for filename in filenames:
+        if not filename.startswith(prefix):
+            raise SystemExit(f'unexpected file without {prefix=}: {filename}')
+        else:
+            new_filenames.append(filename.removeprefix(prefix))
+
+    cmd = normalize_cmd(cmd)
+    return subprocess.call((*cmd, *new_filenames), cwd=subdir)
+
+
+def ignore_exit_code(cmd: tuple[str, ...]) -> int:
+    cmd = normalize_cmd(cmd)
+    subprocess.call(cmd)
+    return 0
+
+
+def n1(cmd: tuple[str, ...]) -> int:
+    cmd, filenames = _cmd_filenames(cmd)
+    cmd = normalize_cmd(cmd)
+    ret = 0
+    for filename in filenames:
+        ret |= subprocess.call((*cmd, filename))
+    return ret
+
+
+def impl(args: argparse.Namespace) -> int:
+    args.cmd = tuple(args.cmd)
+    if args.tool == 'cd':
+        return cd(args.subdir, args.cmd)
+    elif args.tool == 'ignore-exit-code':
+        return ignore_exit_code(args.cmd)
+    elif args.tool == 'n1':
+        return n1(args.cmd)
+    else:
+        raise NotImplementedError(f'unexpected tool: {args.tool}')
+
+
+def main(argv: Sequence[str] | None = None) -> int:
+    parser = argparse.ArgumentParser()
+    add_parsers(parser)
+    args = parser.parse_args(argv)
+
+    return impl(args)
+
+
+if __name__ == '__main__':
+    raise SystemExit(main())
diff --git a/pre_commit/lang_base.py b/pre_commit/lang_base.py
@@ -5,6 +5,7 @@
 import random
 import re
 import shlex
+import sys
 from collections.abc import Generator
 from collections.abc import Sequence
 from typing import Any
@@ -171,7 +172,10 @@ def run_xargs(
 
 
 def hook_cmd(entry: str, args: Sequence[str]) -> tuple[str, ...]:
-    return (*shlex.split(entry), *args)
+    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(
diff --git a/pre_commit/main.py b/pre_commit/main.py
@@ -10,6 +10,7 @@
 from pre_commit import clientlib
 from pre_commit import git
 from pre_commit.color import add_color_option
+from pre_commit.commands import hazmat
 from pre_commit.commands.autoupdate import autoupdate
 from pre_commit.commands.clean import clean
 from pre_commit.commands.gc import gc
@@ -41,7 +42,7 @@
 os.environ.pop('PYTHONEXECUTABLE', None)
 
 COMMANDS_NO_GIT = {
-    'clean', 'gc', 'init-templatedir', 'sample-config',
+    'clean', 'gc', 'hazmat', 'init-templatedir', 'sample-config',
     'validate-config', 'validate-manifest',
 }
 
@@ -245,6 +246,11 @@ def _add_cmd(name: str, *, help: str) -> argparse.ArgumentParser:
 
     _add_cmd('gc', help='Clean unused cached repos.')
 
+    hazmat_parser = _add_cmd(
+        'hazmat', help='Composable tools for rare use in hook `entry`.',
+    )
+    hazmat.add_parsers(hazmat_parser)
+
     init_templatedir_parser = _add_cmd(
         'init-templatedir',
         help=(
@@ -389,6 +395,8 @@ def _add_cmd(name: str, *, help: str) -> argparse.ArgumentParser:
             return clean(store)
         elif args.command == 'gc':
             return gc(store)
+        elif args.command == 'hazmat':
+            return hazmat.impl(args)
         elif args.command == 'hook-impl':
             return hook_impl(
                 store,
diff --git a/tests/commands/hazmat_test.py b/tests/commands/hazmat_test.py
@@ -0,0 +1,99 @@
+from __future__ import annotations
+
+import sys
+
+import pytest
+
+from pre_commit.commands.hazmat import _cmd_filenames
+from pre_commit.commands.hazmat import main
+from testing.util import cwd
+
+
+def test_cmd_filenames_no_dash_dash():
+    with pytest.raises(SystemExit) as excinfo:
+        _cmd_filenames(('no', 'dashdash', 'here'))
+    msg, = excinfo.value.args
+    assert msg == 'hazmat entry must end with `--`'
+
+
+def test_cmd_filenames_no_filenames():
+    cmd, filenames = _cmd_filenames(('hello', 'world', '--'))
+    assert cmd == ('hello', 'world')
+    assert filenames == ()
+
+
+def test_cmd_filenames_some_filenames():
+    cmd, filenames = _cmd_filenames(('hello', 'world', '--', 'f1', 'f2'))
+    assert cmd == ('hello', 'world')
+    assert filenames == ('f1', 'f2')
+
+
+def test_cmd_filenames_multiple_dashdash():
+    cmd, filenames = _cmd_filenames(('hello', '--', 'arg', '--', 'f1', 'f2'))
+    assert cmd == ('hello', '--', 'arg')
+    assert filenames == ('f1', 'f2')
+
+
+def test_cd_unexpected_filename():
+    with pytest.raises(SystemExit) as excinfo:
+        main(('cd', 'subdir', 'cmd', '--', 'subdir/1', 'not-subdir/2'))
+    msg, = excinfo.value.args
+    assert msg == "unexpected file without prefix='subdir/': not-subdir/2"
+
+
+def _norm(out):
+    return out.replace('\r\n', '\n')
+
+
+def test_cd(tmp_path, capfd):
+    subdir = tmp_path.joinpath('subdir')
+    subdir.mkdir()
+    subdir.joinpath('a').write_text('a')
+    subdir.joinpath('b').write_text('b')
+
+    with cwd(tmp_path):
+        ret = main((
+            'cd', 'subdir',
+            sys.executable, '-c',
+            'import os; print(os.getcwd());'
+            'import sys; [print(open(f).read()) for f in sys.argv[1:]]',
+            '--',
+            'subdir/a', 'subdir/b',
+        ))
+
+    assert ret == 0
+    out, err = capfd.readouterr()
+    assert _norm(out) == f'{subdir}\na\nb\n'
+    assert err == ''
+
+
+def test_ignore_exit_code(capfd):
+    ret = main((
+        'ignore-exit-code', sys.executable, '-c', 'raise SystemExit("bye")',
+    ))
+    assert ret == 0
+    out, err = capfd.readouterr()
+    assert out == ''
+    assert _norm(err) == 'bye\n'
+
+
+def test_n1(capfd):
+    ret = main((
+        'n1', sys.executable, '-c', 'import sys; print(sys.argv[1:])',
+        '--',
+        'foo', 'bar', 'baz',
+    ))
+    assert ret == 0
+    out, err = capfd.readouterr()
+    assert _norm(out) == "['foo']\n['bar']\n['baz']\n"
+    assert err == ''
+
+
+def test_n1_some_error_code():
+    ret = main((
+        'n1', sys.executable, '-c',
+        'import sys; raise SystemExit(sys.argv[1] == "error")',
+        '--',
+        'ok', 'error', 'ok',
+    ))
+    assert ret == 1
diff --git a/tests/lang_base_test.py b/tests/lang_base_test.py
@@ -164,3 +164,15 @@ def test_basic_run_hook(tmp_path):
     assert ret == 0
     out = out.replace(b'\r\n', b'\n')
     assert out == b'hi hello file file file\n'
+
+
+def test_hook_cmd():
+    assert lang_base.hook_cmd('echo hi', ()) == ('echo', 'hi')
+
+
+def test_hook_cmd_hazmat():
+    ret = lang_base.hook_cmd('pre-commit hazmat cd a echo -- b', ())
+    assert ret == (
+        sys.executable, '-m', 'pre_commit.commands.hazmat',
+        'cd', 'a', 'echo', '--', 'b',
+    )
diff --git a/tests/main_test.py b/tests/main_test.py
@@ -9,6 +9,7 @@
 
 import pre_commit.constants as C
 from pre_commit import main
+from pre_commit.commands import hazmat
 from pre_commit.errors import FatalError
 from pre_commit.util import cmd_output
 from testing.auto_namedtuple import auto_namedtuple
@@ -157,6 +158,17 @@ def test_all_cmds(command, mock_commands, mock_store_dir):
     assert_only_one_mock_called(mock_commands)
 
 
+def test_hazmat(mock_store_dir):
+    with mock.patch.object(hazmat, 'impl') as mck:
+        main.main(('hazmat', 'cd', 'subdir', '--', 'cmd', '--', 'f1', 'f2'))
+    assert mck.call_count == 1
+    (arg,), dct = mck.call_args
+    assert dct == {}
+    assert arg.tool == 'cd'
+    assert arg.subdir == 'subdir'
+    assert arg.cmd == ['cmd', '--', 'f1', 'f2']
+
+
 def test_try_repo(mock_store_dir):
     with mock.patch.object(main, 'try_repo') as patch:
         main.main(('try-repo', '.'))
diff --git a/tests/repository_test.py b/tests/repository_test.py
@@ -506,3 +506,14 @@ def test_args_with_spaces_and_quotes(tmp_path):
 
     expected = b"['i have spaces', 'and\"\\'quotes', '$and !this']\n"
     assert ret == (0, expected)
+
+
+def test_hazmat(tmp_path):
+    ret = run_language(
+        tmp_path, unsupported,
+        f'pre-commit hazmat ignore-exit-code {shlex.quote(sys.executable)} '
+        f"-c 'import sys; raise SystemExit(sys.argv[1:])'",
+        ('f1', 'f2'),
+    )
+    expected = b"['f1', 'f2']\n"
+    assert ret == (0, expected)
