Improve PDF parsing results for a broader range of documents

[erichare]

Feature Request

Users have reported less-than-ideal results with PDF parsing in the File Upload component. Under the hood, we essentially use pypdf and join together the elements returned, but do no further processing and don't allow any configuration of the parameters.

Motivation

This issue is meant to track improvements that could be made, which could range from exploring other libraries, exposing more configuration options, or implementing some custom routines that produce a better set of outputs when parsing PDFs.

Your Contribution

I can track some potential improvements and ultimately work on a PR