I just discovered https://hister.org/ (I think from a link here on hacker news) ... it solves the full text history search really really well.
- Local go server to index everything into sqlite or postgres
- browser extension to send the DOM for indexing.
- This happens after each page is finished rendering, so client side rendering doesn't break, and it doesn't require fetching the url a second time on the back-end.
- Filter rules to control what gets indexed (basic url regex patterns, nothing too fancy)
- Text-based query interface for terminal dwellers like me.
- Local file indexing that actually works well so that you can search your history and your documents in one place.
- Custom extractors for understanding specific sites and file formats.