So I don't know what didn't work here, but I got a workaround for it.
Instead of trying out on and on I now use the print version of each article. There are no pictures in it, but that's not really a problem and a great advantage is that the downloaded content is much smaller and I also have to remove fewer tags from it, which simplifies the source code of the recipe a lot.
If anyone is interested:
Code:
def print_version(self, url):
artid = url.rsplit('-')[-2]
return u'http://www.golem.de/print.php?a=' +artid
I had to extract the article ID from the URL which is the last number group and just add this to the given
print url
Thanks for your patience, Kovid