Originally Posted by ldolse
It looks like in this case there is probably a bug where some type of junk in the document is being interpreted as a URL. You'll probably need to post the pdf file to the bug tracker.
Alternatively you can try to use search and replace to find and remove whatever string is causing that failure - I'm pretty sure it's failing after search and replace would be done, I don't think it's forced into xhtml until after preprocessing.
Where is the bug tracker?