Processing non-XBRL 10K/10Q

[dbojanin]

I have a group of 50 CIK's that I pulled historical 10-K/10-Q filings for and roughly 600/2700 filings support XBRL. Could the library support scraping the documents without relying on XBRL?

I could just create pyrantic objects and preprocess the html to markdown and use a reasoning model to extract to structured outputs but that creates the risk of hallucination/etc. hoping to ask around for a better way first