High-performance Python libraries for PDF processing, data extraction, and LLM document pipelines.

This organisation maintains Python libraries for working with PDF and other document formats — from low-level manipulation to LLM-ready data extraction.

The foundation of everything here. PyMuPDF wraps the MuPDF C engine and exposes a full Python API for reading, rendering, editing, and converting PDF, XPS, EPUB, MOBI, CBZ, and image files.

pip install pymupdf

import pymupdf

doc = pymupdf.open("report.pdf")
page = doc[0]
print(page.get_text())           # extract text
pix = page.get_pixmap(dpi=150)   # render to image
pix.save("page.png")

Key capabilities: text and image extraction · page rendering at any DPI · annotation create/edit/delete · redaction · PDF creation and merging · encryption · OCR via Tesseract · form fields · 10+ output formats

→ Documentation · PyPI · Changelog

Turn any document into clean, structured data for RAG pipelines, vector stores, and LLM ingestion — in one line. No GPU, no cloud, no tokens required.

pip install pymupdf4llm

import pymupdf4llm

md   = pymupdf4llm.to_markdown("paper.pdf")       # Markdown
data = pymupdf4llm.to_json("paper.pdf")           # JSON with bboxes
text = pymupdf4llm.to_text("paper.pdf")           # plain text

Key capabilities: layout-aware extraction · multi-column reading order · table detection → Markdown · smart hybrid OCR (only where needed) · page chunking with metadata · LlamaIndex and LangChain integrations · 10–250× cheaper than vision-LLM approaches

→ Documentation · PyPI · Live demo

A drop-in LangChain document loader and parser backed by pymupdf4llm. Extracts PDF content as Markdown and feeds it directly into any LangChain retrieval chain.

pip install langchain-pymupdf4llm

from langchain_pymupdf4llm import PyMuPDF4LLMLoader

loader = PyMuPDF4LLMLoader("document.pdf", mode="single")
docs = loader.load()

→ PyPI

An MCP (Model Context Protocol) server exposing pymupdf4llm as a tool. Gives any MCP-compatible AI client (Claude, Cursor, Windsurf) direct access to PDF-to-Markdown extraction.

uvx pymupdf4llm-mcp@latest stdio

→ PyPI

A collection of working example scripts, Jupyter notebooks, and GUI demos built on PyMuPDF. Covers image handling, annotation workflows, data extraction patterns, and more — useful as a recipe book alongside the official documentation.

→ Browse examples

An optional font package extending the fonts available for text output in PyMuPDF. Includes additional Unicode-compatible typefaces beyond the 14 standard PDF fonts.

pip install pymupdf-fonts

All repositories in this organisation are available under the GNU AGPL v3 for open-source use, and under a commercial licence for proprietary applications. Commercial licences are available from Artifex Software — the creators and maintainers of MuPDF, PyMuPDF, and this organisation.

_{Maintained by Artifex Software, Inc. · pymupdf.io}