Well, the bullet check is easily feasible. "Too much text" and "difficult words" too.

Check https://github.com/johnwhitington/camlpdf, which does a really good job to parse PDF files properly.