Hi, - I checked some ways doing this, and starting over with a new thing will give you a lot of headaches - all XSLT processors have one or another problem - success depends very much on how you where using wikipedia (plugins?) and you will have to expect a lot of poking around with details and still not beeing happy with the solutions available - there are many really crazy approaches out there to generate pdf of mediawiki "markup" - many tried, not many succeeded, most of them stop at "good enough for me"-level. So it might be a good idea, not running too far away from what they are doing at http://code.pediapress.com - you will spend much less time with installing ubuntu in a virtualbox.
However there is one quite impressive tool, that does pdf conversion via css and is good for getting the job done quick and not too dirty: http://www.princexml.com/samples/ - scroll down to the mediawiki examples - they offer a free license for non-commercial projects. Good luck! Have a nice day, Snaky -- http://mail.python.org/mailman/listinfo/python-list