[FE-CVS] r3418 - in FormEncode/trunk: docs formencode

[<sub...@co...>]

Author: ianb
Date: 2008-05-13 16:40:04 -0600 (Tue, 13 May 2008)
New Revision: 3418

Modified:
   FormEncode/trunk/docs/news.txt
   FormEncode/trunk/formencode/validators.py
Log:
Adjustments to the URL validator (SF bug 1940971) from Jonathan Vanesco

Modified: FormEncode/trunk/docs/news.txt
===================================================================
--- FormEncode/trunk/docs/news.txt	2008-05-13 21:47:44 UTC (rev 3417)
+++ FormEncode/trunk/docs/news.txt	2008-05-13 22:40:04 UTC (rev 3418)
@@ -21,6 +21,8 @@
 * The validators ``Int`` and ``Number`` both take min/max arguments
   (from Shannon Behrens).
 
+* Some adjustments to the URL validator.
+
 1.0.1
 -----
 

Modified: FormEncode/trunk/formencode/validators.py
===================================================================
--- FormEncode/trunk/formencode/validators.py	2008-05-13 21:47:44 UTC (rev 3417)
+++ FormEncode/trunk/formencode/validators.py	2008-05-13 22:40:04 UTC (rev 3418)
@@ -1379,10 +1379,26 @@
     check_exists = False
     add_http = True
 
-    url_re = re.compile(r'^(http|https)://'
-                        r'(?:[a-z0-9\-]+|[a-z0-9][a-z0-9\-\.\_]*\.[a-z]+)'
-                        r'(?::[0-9]+)?'
-                        r'(?:/.*)?$', re.I) 
+    url_re = re.compile(r'''
+        ^(http|https)://
+        (?:[%:\w]*@)?                   # authenticator
+        (?:[a-z0-9][a-z0-9\-]{1,62}\.)* # (sub)domain - alpha followed by 62max chars (63 total)
+        [a-z]+                          # TLD
+        (?:[0-9]+)?                     # port
+
+        # files/delims/etc
+        (?:/[
+            # rfc3986 valid chars
+            # unreserved
+            a-z0-9\-\._~
+            # delims - general
+            :/\?#\[\]@
+            # delims - sub
+            !\$&\'\(\)\*\+,;=
+        ]*)?
+        $
+    ''', re.I | re.VERBOSE)
+
     scheme_re = re.compile(r'^[a-zA-Z]+:')
 
     messages = {