OpenReading Core / Open source
Every document.
One clear
contract.
The open-source engine for reading, comparing, and routing documents. Run it as a CLI, a Python library, or an HTTP process on your machine.
Start on your machine with PyMuPDF.
Add OCR and hosted backends when you need them.
A page in.
Structured data out.
After installation, one command reads the bundled example with a local backend.
Walk through your first parseuv run openreading parse \
examples/schedule_a_2024.pdf \
--backend pymupdf > sa.jsonsucceededOne JSON envelope in sa.jsonFrom a first parse
to a plan of your own.
Follow the tutorial in order, or go straight to the question in front of you. Every command comes from the repository walkthrough.
Get a result
Install locally, read your first page, and understand the response.
Find the differences
Read the same document two ways. See what one backend misses.
Write your rules
Choose a backend order, escalate on quality, and inspect every decision.
Put it to work
Process a folder, add a vendor key, serve locally, and resume a run.
The code is the source.
The guide is your way in.
This field guide is built directly from the tutorial README. Diagrams share the same vector artwork across the guide and the repository.
Open the field guide