Added C0403 invalid-characters-in-docstring for docstrings with invalid characters (e.g. \x00)

2 files changed, 22 insertions, 3 deletions

diff --git a/checkers/spelling.py b/checkers/spelling.py
index c716ff2..f6edd5d 100644
--- a/checkers/spelling.py
+++ b/checkers/spelling.py
@@ -61,6 +61,9 @@ class SpellingChecker(BaseTokenChecker):
                   '%s\nDid you mean: \'%s\'?',
                   'wrong-spelling-in-docstring',
                   'Used when a word in docstring is not spelled correctly.'),
+        'C0403': ('Invalid characters %r in a docstring',
+                  'invalid-characters-in-docstring',
+                  'Used when a word in docstring cannot be checked by enchant.'),
         }
     options = (('spelling-dict',
                 {'default' : '', 'type' : 'choice', 'metavar' : '<dict name>',
@@ -131,8 +134,6 @@ class SpellingChecker(BaseTokenChecker):
         line2 = re.sub("([^a-zA-Z]|^)'", " ", line2)
         # Replace punctuation signs with space e.g. and/or -> and or
         line2 = self.punctuation_regex.sub(' ', line2)
-        # Replace null bytes with space
-        line2 = line2.replace('\x00', ' ')
 
         words = []
         for word in line2.split():
@@ -170,7 +171,13 @@ class SpellingChecker(BaseTokenChecker):
                 word = word[2:]
 
             # If it is a known word, then continue.
-            if self.spelling_dict.check(word):
+            try:
+                if self.spelling_dict.check(word):
+                    continue
+            except enchant.errors.Error:
+                # this can only happen in docstrings, not comments
+                self.add_message('invalid-characters-in-docstring',
+                                 line=line_num, args=(word,))
                 continue
 
             # Store word to private dict or raise a message.
diff --git a/test/unittest_checker_spelling.py b/test/unittest_checker_spelling.py
index 8b55042..c630eb2 100644
--- a/test/unittest_checker_spelling.py
+++ b/test/unittest_checker_spelling.py
@@ -76,6 +76,18 @@ class SpellingCheckerTest(CheckerTestCase):
                           "comet' or 'comment' or 'moment' or 'foment"))):
             self.checker.visit_class(stmt)
 
+    @unittest.skipIf(spell_dict is None,
+                     "missing python-enchant package or missing "
+                     "spelling dictionaries")
+    @set_config(spelling_dict=spell_dict)
+    def test_invalid_docstring_characters(self):
+        stmt = test_utils.extract_node(
+            'def fff():\n   """test\\x00"""\n   pass')
+        with self.assertAddsMessages(
+            Message('invalid-characters-in-docstring', line=2,
+                    args=('test\x00',))):
+            self.checker.visit_function(stmt)
+
 
 if __name__ == '__main__':
     unittest.main()