PDF Extract is a Python tool built for the RaceDocs workflow. It turns technical PDFs into readable Markdown and extracts the images needed for reading or translation.
Recovering document structure
Using PyMuPDF, the tool reads text positions, fonts, and sizes to identify headings, paragraphs, and lists. It removes repeated headers and footers and rebuilds reading order across single-column, two-column, and mixed layouts.
The workflow also recovers supported tables, deduplicates embedded bitmaps, and renders clipped page compositions when masks or overlays need to be preserved. The result is a Markdown document with a corresponding set of image assets.