Merge pull request #1614 from iBug/patch-4

Catch shortened URL starting with www
Charcoal-SE · Feb 17, 2018 · 0f9c3b1 · 0f9c3b1
2 parents 39d097a + e00340f
commit 0f9c3b1
Showing 1 changed file with 7 additions and 7 deletions.
diff --git a/findspam.py b/findspam.py
@@ -991,16 +991,16 @@ class FindSpam:
          'sites': [], 'reason': 'link at end of {}', 'title': False, 'body': True, 'username': False,
          'stripcodeblocks': False, 'body_summary': False, 'questions': False, 'max_rep': 1, 'max_score': 0},
         # Shortened URL near the end of question
-        {'regex': r"(?is)://(goo\.gl|bit\.ly|bit\.do|tinyurl\.com|fb\.me|cl\.ly|t\.co|is\.gd|j\.mp|tr\.im|ow\.ly|"
+        {'regex': r"(?is)://(?:w+\.)?(goo\.gl|bit\.ly|bit\.do|tinyurl\.com|fb\.me|cl\.ly|t\.co|is\.gd|j\.mp|tr\.im|"
                   r"wp\.me|alturl\.com|tiny\.cc|9nl\.me|post\.ly|dyo\.gs|bfy\.tw|amzn\.to|adf\.ly|adfoc\.us|"
-                  r"surl\.cn\.com|clkmein\.com|bluenik\.com|rurl\.us|adyou\.co|buff\.ly)/.{0,200}$", 'all': True,
-         'sites': ["superuser.com", "askubuntu.com"], 'reason': "shortened URL in {}", 'title': False, 'body': True,
-         'username': False, 'stripcodeblocks': True, 'body_summary': False, 'answers': False, 'max_rep': 1,
-         'max_score': 0},
+                  r"surl\.cn\.com|clkmein\.com|bluenik\.com|rurl\.us|adyou\.co|buff\.ly|ow\.ly)/.{0,200}$",
+         'all': True, 'sites': ["superuser.com", "askubuntu.com"], 'reason': "shortened URL in {}", 'title': False,
+         'body': True, 'username': False, 'stripcodeblocks': True, 'body_summary': False, 'answers': False,
+         'max_rep': 1, 'max_score': 0},
         # Shortened URL in an answer
-        {'regex': r"(?is)://(goo\.gl|bit\.ly|bit\.do|tinyurl\.com|fb\.me|cl\.ly|t\.co|is\.gd|j\.mp|tr\.im|ow\.ly|"
+        {'regex': r"(?is)://(?:w+\.)?(goo\.gl|bit\.ly|bit\.do|tinyurl\.com|fb\.me|cl\.ly|t\.co|is\.gd|j\.mp|tr\.im|"
                   r"wp\.me|alturl\.com|tiny\.cc|9nl\.me|post\.ly|dyo\.gs|bfy\.tw|amzn\.to|adf\.ly|adfoc\.us|"
-                  r"surl\.cn\.com|clkmein\.com|bluenik\.com|rurl\.us|adyou\.co|buff\.ly)/",
+                  r"surl\.cn\.com|clkmein\.com|bluenik\.com|rurl\.us|adyou\.co|buff\.ly|ow\.ly)/",
          'all': True, 'sites': ["codegolf.stackexchange.com"], 'reason': "shortened URL in {}", 'title': False,
          'body': True, 'username': False, 'stripcodeblocks': True, 'body_summary': False, 'questions': False,
          'max_rep': 1, 'max_score': 0},