Revision: 8642 Author: xqt Date: 2010-10-12 15:05:50 +0000 (Tue, 12 Oct 2010)
Log Message: ----------- add sites rejecting bots for weblinkchecker (patch bug #3085935 submitted by masti, thanks!)
Modified Paths: -------------- trunk/pywikipedia/weblinkchecker.py
Modified: trunk/pywikipedia/weblinkchecker.py =================================================================== --- trunk/pywikipedia/weblinkchecker.py 2010-10-11 05:17:59 UTC (rev 8641) +++ trunk/pywikipedia/weblinkchecker.py 2010-10-12 15:05:50 UTC (rev 8642) @@ -207,6 +207,9 @@ 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 + re.compile('.*[./@]ssd.jpl.nasa.gov(/.*)?'), # bot rejected on the site + re.compile('.*[./@]itis.gov(/.*)?'), # bot rejected on the site + re.compile('.*[./@]cev.lu(/.*)?'), # bot rejected on the site ]
def weblinksIn(text, withoutBracketed = False, onlyBracketed = False):