From bf912cfebb8fdd3d16bb9335e93f7c9dada949f8 Mon Sep 17 00:00:00 2001
From: Anthony Sottile <asottile@umich.edu>
Date: Fri, 6 Jun 2014 06:53:00 -0700
Subject: [PATCH] Add checking for manifest files regex.

---
 pre_commit/clientlib/validate_base.py     |  9 ++++++
 pre_commit/clientlib/validate_config.py   |  6 ++--
 pre_commit/clientlib/validate_manifest.py | 34 ++++++++++++++++-------
 tests/clientlib/validate_manifest_test.py | 15 ++++++----
 4 files changed, 45 insertions(+), 19 deletions(-)

diff --git a/pre_commit/clientlib/validate_base.py b/pre_commit/clientlib/validate_base.py
index 1b8856aa..27f363a9 100644
--- a/pre_commit/clientlib/validate_base.py
+++ b/pre_commit/clientlib/validate_base.py
@@ -4,12 +4,21 @@ import argparse
 import jsonschema
 import jsonschema.exceptions
 import os.path
+import re
 import yaml
 
 from pre_commit.jsonschema_extensions import apply_defaults
 from pre_commit.util import entry
 
 
+def is_regex_valid(regex):
+    try:
+        re.compile(regex)
+        return True
+    except re.error:
+        return False
+
+
 def get_validator(
     json_schema,
     exception_type,
diff --git a/pre_commit/clientlib/validate_config.py b/pre_commit/clientlib/validate_config.py
index d9f97698..997c84fe 100644
--- a/pre_commit/clientlib/validate_config.py
+++ b/pre_commit/clientlib/validate_config.py
@@ -1,8 +1,8 @@
-import re
 import sys
 
 from pre_commit.clientlib.validate_base import get_run_function
 from pre_commit.clientlib.validate_base import get_validator
+from pre_commit.clientlib.validate_base import is_regex_valid
 
 
 class InvalidConfigError(ValueError):
@@ -42,9 +42,7 @@ CONFIG_JSON_SCHEMA = {
 
 
 def try_regex(repo, hook, value, field_name):
-    try:
-        re.compile(value)
-    except re.error:
+    if not is_regex_valid(value):
         raise InvalidConfigError(
             'Invalid {0} regex at {1}, {2}: {3}'.format(
                 field_name, repo, hook, value,
diff --git a/pre_commit/clientlib/validate_manifest.py b/pre_commit/clientlib/validate_manifest.py
index 72e8a53f..fed5a033 100644
--- a/pre_commit/clientlib/validate_manifest.py
+++ b/pre_commit/clientlib/validate_manifest.py
@@ -2,6 +2,7 @@ import sys
 
 from pre_commit.clientlib.validate_base import get_run_function
 from pre_commit.clientlib.validate_base import get_validator
+from pre_commit.clientlib.validate_base import is_regex_valid
 from pre_commit.languages.all import all_languages
 
 
@@ -29,18 +30,31 @@ MANIFEST_JSON_SCHEMA = {
 }
 
 
+def validate_languages(hook_config):
+    if hook_config['language'] not in all_languages:
+        raise InvalidManifestError(
+            'Expected language {0} for {1} to be one of {2!r}'.format(
+                hook_config['id'],
+                hook_config['language'],
+                all_languages,
+            )
+        )
+
+
+def validate_files(hook_config):
+    if not is_regex_valid(hook_config['files']):
+        raise InvalidManifestError(
+            'Invalid files regex at {0}: {1}'.format(
+                hook_config['id'],
+                hook_config['files'],
+            )
+        )
+
+
 def additional_manifest_check(obj):
     for hook_config in obj:
-        language = hook_config['language']
-
-        if language not in all_languages:
-            raise InvalidManifestError(
-                'Expected language {0} for {1} to be one of {2!r}'.format(
-                    hook_config['id'],
-                    hook_config['language'],
-                    all_languages,
-                )
-            )
+        validate_languages(hook_config)
+        validate_files(hook_config)
 
 
 load_manifest = get_validator(
diff --git a/tests/clientlib/validate_manifest_test.py b/tests/clientlib/validate_manifest_test.py
index 800d292b..5dba2747 100644
--- a/tests/clientlib/validate_manifest_test.py
+++ b/tests/clientlib/validate_manifest_test.py
@@ -28,20 +28,25 @@ def test_additional_manifest_check_raises_for_bad_language():
 
 
 @pytest.mark.parametrize(
-    'obj', ([{'language': 'python'}], [{'language': 'ruby'}]),
+    'obj',
+    (
+        [{'language': 'python', 'files': ''}],
+        [{'language': 'ruby', 'files': ''}]
+    ),
 )
-def test_additional_manifest_check_languages(obj):
+def test_additional_manifest_check_passing(obj):
     additional_manifest_check(obj)
 
 
 @pytest.mark.parametrize(
     'obj',
     (
-        [{'id': 'a', 'language': 'not a language'}],
-        [{'id': 'a', 'language': 'python3'}],
+        [{'id': 'a', 'language': 'not a language', 'files': ''}],
+        [{'id': 'a', 'language': 'python3', 'files': ''}],
+        [{'id': 'a', 'language': 'python', 'files': 'invalid regex('}],
     ),
 )
-def test_additional_manifest_check_languages_failing(obj):
+def test_additional_manifest_failing(obj):
     with pytest.raises(InvalidManifestError):
         additional_manifest_check(obj)