Update polyglot so that it cleans out bad utf-8 chars, instead of dying

Question

I\'m trying to tweak the polyglot function for detecting the language. Basically, I\'m downloading a websites HTML, stripping it of any HTML using html2tx