Strange Results in Word Cloud

I’m getting what appears to be a lot of code in some of my word clouds. As far as I can tell they’re linked to Pages documents, but not all Pages documents do this, I think. They’re not related to anything I can actually see in the documents (certainly not text). Any ideas what it might be?

Never seen anything like this related to Pages documents and DEVONthink just uses what Spotlight’s Pages plugin returns.

In my experience it’s usually caused by logs, crash reports or email attachments without proper MIME encoding.