Right now it is not possible to index PDF files in the form of media items. I believe this would be very valuable for sites that have lots of pdf's uploaded. This way Vragen.ai can have more context of these files.
My proposed solution is to detect whenever a media item with mime type 'application/pdf' is being indexed. We can then construct a URL to it and pass it to Vragen.ai. Then on the side of Vragen.ai it can crawl the provided URL and render it to text + index it. This way we are not passing the entire PDF in an HTTP request and also do not have to render the PDF to HTML on the Drupal side.
None I can think of.
When a media item is detected it will send a new document request without content. The API will soon be equipped to handle this.
Active
2.0
Code
Thanks!
Left a comment, also phpcs is not passing :)
Ty and merged. Nobnew release though, do younneed one?
I think a new release is a good idea, especially as there is also a bug with reading settings on beta3.
I'd also rather use a beta version then the dev branch.
Thanks!