2012/3/28 Bináris wikiposta@gmail.com:
For me they return still encoded titles. :-(
...they do for me, too. Strangely enough. Considering it did work back in august (which you can check by fetching r9429), someone broke it. Bisecting indicates this revision: ------------------------------------------------------------------------ r9489 | xqt | 2011-09-03 12:07:18 +0200 (Sat, 03 Sep 2011) | 1 line
revert r5856 due to bug #2989218 ------------------------------------------------------------------------
https://www.mediawiki.org/w/index.php?title=Special:Code/pywikipedia/9489 http://sourceforge.net/tracker/?func=detail&group_id=93107&atid=6031...
Oh, and I think the fix is already in cosmetic_changes.py, too. Check def cleanUpLinks (line 314).
It seems I was wrong about that. It indeed does not fix this, even though it /does/ fix %xx urls.
Another issue: á is encoded as .C3.A1. However, a literal .C3.A1 in section title will also appear the same. Is there any way to decide if .C3.A1 stands for á or for .C3.A1?
Check the mediawiki source, as it's mediawiki that does this transformation.
Best, Merlijn