Fix a problem introduced in 320724f964 where is extracted the ID from the url with self._match_id but the problem is that ID is not always present in the url passed so the title should be extracted as proposed by the fix (and like is done in _real_extract (see line 337))