Fix t.co urls followed by a comma

lsemel · lsemel · commit 52c61013ce1f · 2013-03-25T10:27:04.000-04:00
diff --git a/ttp/tests.py b/ttp/tests.py
@@ -231,6 +231,11 @@ def test_url_followed_comma(self):
         self.assertEqual(result.html, u'text <a href="http://example.com">http://example.com</a>,')
         self.assertEqual(result.urls, [u'http://example.com'])
 
+    def test_url_with_path_followed_comma(self):
+        result = self.parser.parse(u'text http://example.com/abcde, more')
+        self.assertEqual(result.html, u'text <a href="http://example.com/abcde">http://example.com/abcde</a>, more')
+        self.assertEqual(result.urls, [u'http://example.com/abcde'])
+
     def test_url_followed_brace(self):
         result = self.parser.parse(u'text http://example.com)')
         self.assertEqual(result.html, u'text <a href="http://example.com">http://example.com</a>)')
diff --git a/ttp/ttp.py b/ttp/ttp.py
@@ -59,7 +59,7 @@
 PATH_ENDING_CHARS = r'[%s\)=#/]' % UTF_CHARS
 QUERY_ENDING_CHARS = '[a-z0-9_&=#]'
 
-URL_REGEX = re.compile('((%s)((https?://|www\\.)(%s)(\/%s*%s?)?(\?%s*%s)?))'
+URL_REGEX = re.compile('((%s)((https?://|www\\.)(%s)(\/(%s*%s)?)?(\?%s*%s)?))'
                        % (PRE_CHARS, DOMAIN_CHARS, PATH_CHARS,
                           PATH_ENDING_CHARS, QUERY_CHARS, QUERY_ENDING_CHARS),
                           re.IGNORECASE)
