2 ms·
On https://mitta.us https://mitta.us, I snapshot the whole page into a PNG using a headless browser. That PNG can then be sent into a pipeline to have the text
by kordlessagain 3y ago
On https://mitta.us https://mitta.us, I snapshot the whole page into a PNG using a headless browser. That PNG can then be sent into a pipeline to have the text extracted by Google's Document AI. That text is sent off to OpenAI to embed it, and stored in Solr, where it can be used later to search for it. The main issue is that I can't pick up links with it, but most of the time that really doesn't matter as long as I can see the page like it was rendered and have the text searchable.