Make URL matching code less matchy

Author: Richard Jones

CHANGES.txt

@@ -4,6 +4,7 @@ are given with the most recent entry first.
 2008-09-01 1.4.6
 Fixed:
 - Bug introduced in 1.4.5 in RDBMS full-text indexing
+- Make URL matching code less matchy
 
 
 2008-08-19 1.4.5

roundup/cgi/templating.py

@@ -1247,11 +1247,18 @@ def is_view_ok(self):
 class StringHTMLProperty(HTMLProperty):
     hyper_re = re.compile(r'''(
         (?P<url>
-          ((ht|f)tp(s?)://|www\.)?          # prefix
+         (
+          (ht|f)tp(s?)://                   # protocol
           ([\w]+:\w+@)?                     # username/password
-          (([\w\-]+\.)+([\w]{2,5}))         # hostname
-          (:[\d]{1,5})?                     # port
-          (/[\w\-$.+!*(),;:@&=?/~\\#%]*)?   # path etc.
+          ([\w\-]+\.)+                      # hostname
+          [\w]{2,5}                         # TLD
+         |                                  # ... or ...
+          www\.                             # "www."
+          ([\w\-]+\.)+                      # hostname
+          [\w]{2,5}                         # TLD
+         )
+         (:[\d]{1,5})?                     # port
+         (/[\w\-$.+!*(),;:@&=?/~\\#%]*)?   # path etc.
         )|
         (?P<email>[-+=%/\w\.]+@[\w\.\-]+)|
         (?P<item>(?P<class>[A-Za-z_]+)(\s*)(?P<id>\d+))

test/test_templating.py

@@ -91,9 +91,13 @@ def lookup(key) :
     def test_url_match(self):
         '''Test the URL regular expression in StringHTMLProperty.
         '''
-        def t(s, **groups):
+        def t(s, nothing=False, **groups):
             m = StringHTMLProperty.hyper_re.search(s)
-            self.assertNotEquals(m, None, '%r did not match'%s)
+            if nothing:
+                self.assertEquals(m, None, '%r matched'%s)
+                return
+            else:
+                self.assertNotEquals(m, None, '%r did not match'%s)
             d = m.groupdict()
             for g in groups:
                 self.assertEquals(d[g], groups[g], '%s %r != %r in %r'%(g, d[g],
@@ -103,10 +107,12 @@ def t(s, **groups):
         t('http://roundup.net/', url='http://roundup.net/')
         t('<HTTP://roundup.net/>', url='HTTP://roundup.net/')
         t('www.a.ex', url='www.a.ex')
+        t('foo.a.ex', nothing=True)
+        t('StDevValidTimeSeries.GetObservation', nothing=True)
         t('http://a.ex', url='http://a.ex')
         t('http://a.ex/?foo&bar=baz\\.@!$%()qwerty',
             url='http://a.ex/?foo&bar=baz\\.@!$%()qwerty')
-        t('www.net', url='www.net')
+        t('www.foo.net', url='www.foo.net')
         t('[email protected]', email='[email protected]')
         t('[email protected]', email='[email protected]')
         t('i1', **{'class':'i', 'id':'1'})