Page MenuHomePhabricator

PDF export on el.wp returns content from random other Wikipedias and not from article itself
Closed, DuplicatePublic

Description

As it was reported and I checked in el.wikipedia:

Go to article Κάνδαλος Λαρισσού Αχαΐας.
Click "Export to PDF" and open the exported pdf.
It contains content from fr:hack, The links in the french language content, link to french titles in greek wikipedia.

Go to article Καρυά Αχαΐας (a disambig page).
Click "Export to PDF" and open the exported pdf.
It contains content from an english wikipedia Talk Page. Links lead to greek wikipedia.

Event Timeline

geraki raised the priority of this task from to Unbreak Now!.
geraki updated the task description. (Show Details)
geraki subscribed.
Aklapper renamed this task from PDF export returns content from other language wikipedias and not from the article to PDF export on el.wp returns content from random other Wikipedias and not from article itself.Apr 28 2015, 12:26 PM
Aklapper lowered the priority of this task from Unbreak Now! to High.
Aklapper added a project: OCG-PDF-renderer.
Aklapper set Security to None.

Thanks for taking the time to report this! Confirming on both pages.

For https://el.wikipedia.org/w/index.php?title=Ειδικό:Συλλογή&bookcmd=render_article&arttitle=Κάνδαλος+Λαρισσού+Αχαΐας&oldid=5220096&writer=rdf2latex , the displayed French article in the PDF is an ancient revision, probably https://fr.wikipedia.org/w/index.php?title=.hack&oldid=5220078 or 5220096 (the latter revision removed lots of categories from that fr article).

The English content in the PDF for Καρυά Αχαΐας is https://en.wikipedia.org/w/index.php?title=User_talk:195.92.198.72&oldid=5219251 (also ancient).

Cannot reproduce with random other el.wp articles.
https://wikitech.wikimedia.org/wiki/OCG has no information how to debug this further client-side, so I'm afraid this requires shell access / devs. CC'ing the Parsing Team.

Hm, perhaps I need to purge the OCG cache of rendered PDFs as well.