Dump now contain a `bytes` attribute which breaks parsing.

[d2a-raudenaerde]

For example: <text bytes="28961" xml:space="preserve"> This makes the cleaner return no content.

[tballison]

I think there was a bytes before...at least according to the source code. The problem is that there's a regex that expects the xml:space... before the bytes element.

This just hit me, and it is a complete showstopper for wikiclean. I've added preprocessing hackery for now. :D

[lintool]

Unfortunately, I'm tied up with EMNLP deadlines until next week... but PR welcome?!