[Pywikipedia-l] SVN: [4792] trunk/pywikipedia/weblinkchecker.py

[wikipedian at svn.wikimedia.org]

Revision: 4792
Author:   wikipedian
Date:     2008-01-02 20:18:54 +0000 (Wed, 02 Jan 2008)

Log Message:
-----------
fixed bug: [ 1862171 ] How to hide Links from weblinkchcker.py ?

Modified Paths:
--------------
    trunk/pywikipedia/weblinkchecker.py

Modified: trunk/pywikipedia/weblinkchecker.py
===================================================================
--- trunk/pywikipedia/weblinkchecker.py	2008-01-02 16:19:31 UTC (rev 4791)
+++ trunk/pywikipedia/weblinkchecker.py	2008-01-02 20:18:54 UTC (rev 4792)
@@ -145,12 +145,22 @@
 }
 
 ignorelist = [
-    re.compile('.*[\./@]example.com(/.*)?'), # reserved for documentation
-    re.compile('.*[\./@]example.net(/.*)?'), # reserved for documentation
-    re.compile('.*[\./@]example.org(/.*)?'), # reserved for documentation
-    re.compile('.*[\./@]gso.gbv.de(/.*)?'),  # bot somehow can't handle their redirects 
-    re.compile('.*[\./@]berlinonline.de(/.*)?'), # a de: user wants to fix them by hand and doesn't want them to be deleted, see [[de:Benutzer:BLueFiSH.as/BZ]].
-    re.compile('.*[\./@]bodo.kommune.no(/.*)?'), # bot can't handle their redirects
+    # Officialy reserved for testing, documentation, etc. in
+    # http://tools.ietf.org/html/rfc2606#page-2
+    # top-level domains:
+    re.compile('.*[\./@]test(/.*)?'),
+    re.compile('.*[\./@]example(/.*)?'),
+    re.compile('.*[\./@]invalid(/.*)?'),
+    re.compile('.*[\./@]localhost(/.*)?'),
+    # second-level domains:
+    re.compile('.*[\./@]example\.com(/.*)?'),
+    re.compile('.*[\./@]example\.net(/.*)?'),
+    re.compile('.*[\./@]example\.org(/.*)?'),
+
+    # Other special cases
+    re.compile('.*[\./@]gso\.gbv\.de(/.*)?'),  # bot somehow can't handle their redirects 
+    re.compile('.*[\./@]berlinonline\.de(/.*)?'), # a de: user wants to fix them by hand and doesn't want them to be deleted, see [[de:Benutzer:BLueFiSH.as/BZ]].
+    re.compile('.*[\./@]bodo\.kommune\.no(/.*)?'), # bot can't handle their redirects
 ]
 
 def compileLinkR(withoutBracketed = False, onlyBracketed = False):