ocr-and-documents

Use pymupdf to open the PDF and call get_text() on each page, or pymupdf4llm to convert directly to Markdown. It installs in about 25MB and handles text-based PDFs, tables, images, and metadata without downloading models.

pymupdf is lightweight (~25MB) and instant, handling text-based PDFs, basic tables, and embedded images. marker-pdf (~3-5GB with PyTorch) adds OCR for scanned documents in 90+ languages, equations, forms, and reading-order detection.

Use marker-pdf, which performs OCR on scanned PDFs in over 90 languages and outputs Markdown. Run the extract_marker.py script on the file, but verify you have about 5GB free disk first since it downloads PyTorch and models.

No, pymupdf cannot perform OCR on scanned images or convert images to text. For scanned documents, equations, or complex layouts, use marker-pdf instead, or provide a URL so web_extract can handle the conversion remotely.

Use pymupdf's insert_pdf method to copy page ranges into a new document for splitting, or combine multiple source files into one for merging. No extra dependencies are needed beyond pymupdf itself.

marker-pdf requires roughly 5GB of disk space for PyTorch plus about 2.5GB of models downloaded to the Hugging Face cache on first use. Run the script's --check flag to verify free space, or fall back to pymupdf for text-based PDFs.