Implement pygrep language as a replacement for pcre

pre_commit/languages/all.py

@@ -5,6 +5,7 @@ from pre_commit.languages import docker_image
 from pre_commit.languages import golang
 from pre_commit.languages import node
 from pre_commit.languages import pcre
+from pre_commit.languages import pygrep
 from pre_commit.languages import python
 from pre_commit.languages import ruby
 from pre_commit.languages import script
@@ -54,6 +55,7 @@ languages = {
     'golang': golang,
     'node': node,
     'pcre': pcre,
+    'pygrep': pygrep,
     'python': python,
     'ruby': ruby,
     'script': script,

pre_commit/languages/pygrep.py

@@ -0,0 +1,59 @@
+from __future__ import absolute_import
+from __future__ import unicode_literals
+
+import argparse
+import re
+import sys
+
+from pre_commit import output
+from pre_commit.languages import helpers
+from pre_commit.xargs import xargs
+
+
+ENVIRONMENT_DIR = None
+get_default_version = helpers.basic_get_default_version
+healthy = helpers.basic_healthy
+install_environment = helpers.no_install
+
+
+def _process_filename_by_line(pattern, filename):
+    retv = 0
+    with open(filename, 'rb') as f:
+        for line_no, line in enumerate(f, start=1):
+            if pattern.search(line):
+                retv = 1
+                output.write('{}:{}:'.format(filename, line_no))
+                output.write_line(line.rstrip(b'\r\n'))
+    return retv
+
+
+def run_hook(repo_cmd_runner, hook, file_args):
+    exe = (sys.executable, '-m', __name__)
+    exe += tuple(hook['args']) + (hook['entry'],)
+    return xargs(exe, file_args)
+
+
+def main(argv=None):
+    parser = argparse.ArgumentParser(
+        description=(
+            'grep-like finder using python regexes.  Unlike grep, this tool '
+            'returns nonzero when it finds a match and zero otherwise.  The '
+            'idea here being that matches are "problems".'
+        ),
+    )
+    parser.add_argument('-i', '--ignore-case', action='store_true')
+    parser.add_argument('pattern', help='python regex pattern.')
+    parser.add_argument('filenames', nargs='*')
+    args = parser.parse_args(argv)
+
+    flags = re.IGNORECASE if args.ignore_case else 0
+    pattern = re.compile(args.pattern.encode(), flags)
+
+    retv = 0
+    for filename in args.filenames:
+        retv |= _process_filename_by_line(pattern, filename)
+    return retv
+
+
+if __name__ == '__main__':
+    exit(main())

pre_commit/repository.py

@@ -202,8 +202,8 @@ class LocalRepository(Repository):
     def _cmd_runner_from_deps(self, language_name, deps):
         """local repositories have a cmd runner per hook"""
         language = languages[language_name]
-        # pcre / script / system / docker_image do not have environments so
-        # they work out of the current directory
+        # pcre / pygrep / script / system / docker_image do not have
+        # environments so they work out of the current directory
         if language.ENVIRONMENT_DIR is None:
             return PrefixedCommandRunner(git.get_root())
         else: