ArchiveBox can extract text from HTML (and possibly PDFs too). I think it can be configured to extract subtitles from YouTube videos as well. So it can do full text searches. Basically you could have your own, offline & curated search-engine.