Add pcre type.

pre_commit/languages/all.py

@@ -1,6 +1,7 @@
 from __future__ import unicode_literals
 
 from pre_commit.languages import node
+from pre_commit.languages import pcre
 from pre_commit.languages import python
 from pre_commit.languages import ruby
 from pre_commit.languages import script
@@ -36,6 +37,7 @@ from pre_commit.languages import system
 
 languages = {
     'node': node,
+    'pcre': pcre,
     'python': python,
     'ruby': ruby,
     'script': script,

pre_commit/languages/helpers.py

@@ -1,10 +1,14 @@
 from __future__ import unicode_literals
 
 
+def file_args_to_stdin(file_args):
+    return '\n'.join(list(file_args) + [''])
+
+
 def run_hook(env, hook, file_args):
     return env.run(
         ' '.join(['xargs', hook['entry']] + hook['args']),
-        stdin='\n'.join(list(file_args) + ['']),
+        stdin=file_args_to_stdin(file_args),
         retcode=None,
     )
 

pre_commit/languages/pcre.py

@@ -0,0 +1,27 @@
+from __future__ import unicode_literals
+
+from pre_commit.languages.helpers import file_args_to_stdin
+from pre_commit.util import shell_escape
+
+
+ENVIRONMENT_DIR = None
+
+
+def install_environment(repo_cmd_runner, version='default'):
+    """Installation for pcre type is a noop."""
+    raise AssertionError('Cannot install pcre repo.')
+
+
+def run_hook(repo_cmd_runner, hook, file_args):
+    # For PCRE the entry is the regular expression to match
+    return repo_cmd_runner.run(
+        [
+            'xargs', 'sh', '-c',
+            # Grep usually returns 0 for matches, and nonzero for non-matches
+            # so we flip it here.
+            '! grep -H -n -P {0} $@'.format(shell_escape(hook['entry'])),
+            '--',
+        ],
+        stdin=file_args_to_stdin(file_args),
+        retcode=None,
+    )

pre_commit/languages/script.py

@@ -1,16 +1,20 @@
 from __future__ import unicode_literals
 
+from pre_commit.languages.helpers import file_args_to_stdin
+
+
 ENVIRONMENT_DIR = None
 
 
 def install_environment(repo_cmd_runner, version='default'):
     """Installation for script type is a noop."""
+    raise AssertionError('Cannot install script repo.')
 
 
 def run_hook(repo_cmd_runner, hook, file_args):
     return repo_cmd_runner.run(
         ['xargs', '{{prefix}}{0}'.format(hook['entry'])] + hook['args'],
         # TODO: this is duplicated in pre_commit/languages/helpers.py
-        stdin='\n'.join(list(file_args) + ['']),
+        stdin=file_args_to_stdin(file_args),
         retcode=None,
     )

pre_commit/languages/system.py

@@ -2,18 +2,20 @@ from __future__ import unicode_literals
 
 import shlex
 
+from pre_commit.languages.helpers import file_args_to_stdin
+
 
 ENVIRONMENT_DIR = None
 
 
 def install_environment(repo_cmd_runner, version='default'):
     """Installation for system type is a noop."""
+    raise AssertionError('Cannot install system repo.')
 
 
 def run_hook(repo_cmd_runner, hook, file_args):
     return repo_cmd_runner.run(
         ['xargs'] + shlex.split(hook['entry']) + hook['args'],
-        # TODO: this is duplicated in pre_commit/languages/helpers.py
+        stdin=file_args_to_stdin(file_args),
-        stdin='\n'.join(list(file_args) + ['']),
         retcode=None,
     )

pre_commit/util.py

@@ -49,7 +49,10 @@ def clean_path_on_failure(path):
         raise
 
 
-# TODO: asottile.contextlib this with a forward port of nested
 @contextlib.contextmanager
 def noop_context():
     yield
+
+
+def shell_escape(arg):
+    return "'" + arg.replace("'", "'\"'\"'".strip()) + "'"

testing/resources/pcre_hooks_repo/hooks.yaml

@@ -0,0 +1,10 @@
+-   id: regex-with-quotes
+    name: Regex with quotes
+    entry: "foo'bar"
+    language: pcre
+    files: ''
+-   id: other-regex
+    name: Other regex
+    entry: ^\[INFO\]
+    language: pcre
+    files: ''

tests/languages/all_test.py

@@ -1,5 +1,6 @@
 from __future__ import unicode_literals
 
+import inspect
 import pytest
 
 from pre_commit.languages.all import all_languages
@@ -7,7 +8,27 @@ from pre_commit.languages.all import languages
 
 
 @pytest.mark.parametrize('language', all_languages)
-def test_all_languages_support_interface(language):
+def test_install_environment_argspec(language):
-    assert hasattr(languages[language], 'install_environment')
+    expected_argspec = inspect.ArgSpec(
-    assert hasattr(languages[language], 'run_hook')
+        args=['repo_cmd_runner', 'version'],
+        varargs=None,
+        keywords=None,
+        defaults=('default',),
+    )
+    argspec = inspect.getargspec(languages[language].install_environment)
+    assert argspec == expected_argspec
+
+
+@pytest.mark.parametrize('language', all_languages)
+def test_ENVIRONMENT_DIR(language):
     assert hasattr(languages[language], 'ENVIRONMENT_DIR')
+
+
+@pytest.mark.parametrize('language', all_languages)
+def test_run_hook_argpsec(language):
+    expected_argspec = inspect.ArgSpec(
+        args=['repo_cmd_runner', 'hook', 'file_args'],
+        varargs=None, keywords=None, defaults=None,
+    )
+    argspec = inspect.getargspec(languages[language].run_hook)
+    assert argspec == expected_argspec

tests/languages/helpers_test.py

@@ -0,0 +1,16 @@
+from __future__ import absolute_import
+from __future__ import unicode_literals
+
+from pre_commit.languages.helpers import file_args_to_stdin
+
+
+def test_file_args_to_stdin_empty():
+    assert file_args_to_stdin([]) == ''
+
+
+def test_file_args_to_stdin_some():
+    assert file_args_to_stdin(['foo', 'bar']) == 'foo\nbar\n'
+
+
+def test_file_args_to_stdin_tuple():
+    assert file_args_to_stdin(('foo', 'bar')) == 'foo\nbar\n'

tests/repository_test.py

@@ -1,6 +1,7 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
+import io
 import mock
 import os.path
 import pytest
@@ -25,12 +26,20 @@ def test_install_python_repo_in_env(tmpdir_factory, store):
     assert os.path.exists(os.path.join(store.directory, repo.sha, 'py_env'))
 
 
-def _test_hook_repo(tmpdir_factory, store, repo_path, hook_id, args, expected):
+def _test_hook_repo(
+        tmpdir_factory,
+        store,
+        repo_path,
+        hook_id,
+        args,
+        expected,
+        expected_return_code=0,
+):
     path = make_repo(tmpdir_factory, repo_path)
     config = make_config_from_repo(path)
     repo = Repository.create(config, store)
     ret = repo.run_hook(hook_id, args)
-    assert ret[0] == 0
+    assert ret[0] == expected_return_code
     assert ret[1] == expected
 
 
@@ -102,6 +111,69 @@ def test_run_a_script_hook(tmpdir_factory, store):
     )
 
 
+@pytest.mark.integration
+def test_pcre_hook_no_match(tmpdir_factory, store):
+    path = git_dir(tmpdir_factory)
+    with local.cwd(path):
+        with io.open('herp', 'w') as herp:
+            herp.write('foo')
+
+        with io.open('derp', 'w') as derp:
+            derp.write('bar')
+
+        _test_hook_repo(
+            tmpdir_factory, store, 'pcre_hooks_repo',
+            'regex-with-quotes', ['herp', 'derp'], '',
+        )
+
+        _test_hook_repo(
+            tmpdir_factory, store, 'pcre_hooks_repo',
+            'other-regex', ['herp', 'derp'], '',
+        )
+
+
+@pytest.mark.integration
+def test_pcre_hook_matching(tmpdir_factory, store):
+    path = git_dir(tmpdir_factory)
+    with local.cwd(path):
+        with io.open('herp', 'w') as herp:
+            herp.write("\nherpfoo'bard\n")
+
+        with io.open('derp', 'w') as derp:
+            derp.write('[INFO] information yo\n')
+
+        _test_hook_repo(
+            tmpdir_factory, store, 'pcre_hooks_repo',
+            'regex-with-quotes', ['herp', 'derp'], "herp:2:herpfoo'bard\n",
+            expected_return_code=123,
+        )
+
+        _test_hook_repo(
+            tmpdir_factory, store, 'pcre_hooks_repo',
+            'other-regex', ['herp', 'derp'], 'derp:1:[INFO] information yo\n',
+            expected_return_code=123,
+        )
+
+
+@pytest.mark.integration
+def test_pcre_many_files(tmpdir_factory, store):
+    # This is intended to simulate lots of passing files and one failing file
+    # to make sure it still fails.  This is not the case when naively using
+    # a system hook with `grep -H -n '...'` and expected_return_code=123.
+    path = git_dir(tmpdir_factory)
+    with local.cwd(path):
+        with io.open('herp', 'w') as herp:
+            herp.write('[INFO] info\n')
+
+        _test_hook_repo(
+            tmpdir_factory, store, 'pcre_hooks_repo',
+            'other-regex',
+            ['/dev/null'] * 15000 + ['herp'],
+            'herp:1:[INFO] info\n',
+            expected_return_code=123,
+        )
+
+
 @pytest.mark.integration
 def test_cwd_of_hook(tmpdir_factory, store):
     # Note: this doubles as a test for `system` hooks

tests/util_test.py

@@ -11,6 +11,7 @@ from plumbum import local
 from pre_commit.util import clean_path_on_failure
 from pre_commit.util import entry
 from pre_commit.util import memoize_by_cwd
+from pre_commit.util import shell_escape
 
 
 @pytest.fixture
@@ -99,3 +100,15 @@ def test_clean_path_on_failure_cleans_for_system_exit(in_tmpdir):
             raise MySystemExit
 
     assert not os.path.exists('foo')
+
+
+@pytest.mark.parametrize(
+    ('input_str', 'expected'),
+    (
+        ('', "''"),
+        ('foo"bar', "'foo\"bar'"),
+        ("foo'bar", "'foo'\"'\"'bar'")
+    ),
+)
+def test_shell_escape(input_str, expected):
+    assert shell_escape(input_str) == expected