Gigablast fetching results from inside pdfs. You may think “ah, but surely Google does this”, but you’re wrong. I searched for infinite dimensional spaces and Gigablast prioritized the content of the pdfs and not the title. It came up with results that made a lot more sense than what Google does. And it fetches a lot, I looked at more than 100 results for this search only (searx fetches much less) and they kept making sense up until the 100th result.
To do this, include ext:pdf or filetype:pdf depending on your search engine.