Comment by 3abiton

This is an interesting approach! Why not offload PDF extraction to other frameorks that apply OCR pdf -> .md