Writing Python scripts to extract data from PDFs is always a nightmare - misshapen tables, arbitrary form formatting, tiny text or low-quality scanned texts. Your champion is here in the form of Natural PDF, a new Python library for wrangling data that's focused on usability and cramming in as many features as possible. Write your code just like you'd ask "real-language" questions:

- "Find the bold text that ends with a colon and get the text to the right of it"
- "Show me all pages that include diagrams"
- "Ignore everything above the thick horizontal line"

Natural PDF also comes bundled with all sorts of modern AI-based magic:

- Categorize pages or entire PDFs by text or vision models
- Use powerful layout analysis engines to recognize images, tables, headers, and paragraphs
- Multilingual OCR powered by multiple engines + LLM-correcting integration

Participants will leave with a solid understanding of how to extract data from difficult (and simple) PDFs using Python.

This session is good for those with some experience working with Python. Laptops will be provided.

Session Detail

All Sessions