added warning if globs are used instead of regex

paulhfischer · asottile · commit 610716d3d1ca · 2020-11-22T18:53:48.000-08:00
diff --git a/pre_commit/clientlib.py b/pre_commit/clientlib.py
@@ -112,6 +112,18 @@ def validate_manifest_main(argv: Optional[Sequence[str]] = None) -> int:
 META = 'meta'
 
 
+class OptionalSensibleRegex(cfgv.OptionalNoDefault):
+    def check(self, dct: Dict[str, Any]) -> None:
+        super().check(dct)
+
+        if '/*' in dct.get(self.key, ''):
+            logger.warning(
+                f'The {self.key!r} field in hook {dct.get("id")!r} is a '
+                f"regex, not a glob -- matching '/*' probably isn't what you "
+                f'want here',
+            )
+
+
 class MigrateShaToRev:
     key = 'rev'
 
@@ -227,6 +239,8 @@ def warn_unknown_keys_repo(
         for item in MANIFEST_HOOK_DICT.items
         if item.key != 'id'
     ),
+    OptionalSensibleRegex('files', cfgv.check_string),
+    OptionalSensibleRegex('exclude', cfgv.check_string),
 )
 CONFIG_REPO_DICT = cfgv.Map(
     'Repository', 'repo',
diff --git a/tests/clientlib_test.py b/tests/clientlib_test.py
@@ -166,6 +166,23 @@ def test_validate_warn_on_unknown_keys_at_top_level(tmpdir, caplog):
     ]
 
 
+def test_validate_optional_sensible_regex(caplog):
+    config_obj = {
+        'id': 'flake8',
+        'files': 'dir/*.py',
+    }
+    cfgv.validate(config_obj, CONFIG_HOOK_DICT)
+
+    assert caplog.record_tuples == [
+        (
+            'pre_commit',
+            logging.WARNING,
+            "The 'files' field in hook 'flake8' is a regex, not a glob -- "
+            "matching '/*' probably isn't what you want here",
+        ),
+    ]
+
+
 @pytest.mark.parametrize('fn', (validate_config_main, validate_manifest_main))
 def test_mains_not_ok(tmpdir, fn):
     not_yaml = tmpdir.join('f.notyaml')
