Repolode
Submit
docling-project

docling-project/docling

Open-source document parser that converts PDF, Office, HTML and other files into structured data for generative AI pipelines.

Open-source alternative to:Adobe Acrobat

View on GitHubWebsite
Stars
69,115
Forks
5,107
Open issues
874
Last commit
3 hours ago

What is it?

Docling is a Python library and CLI for parsing documents and converting them into a unified representation called DoclingDocument. It accepts a wide range of inputs, including PDF, DOCX, PPTX, XLSX, HTML, EPUB, email files, images and audio, and exports the result as Markdown, HTML, lossless JSON or other formats.

Its PDF pipeline handles page layout, reading order, table structure, formulas, code blocks and OCR for scanned pages. Vision language models such as GraniteDocling can be used as an alternative pipeline, and speech recognition models can transcribe audio and video.

Docling can run fully locally, which suits sensitive data and air-gapped environments. It also ships with integrations for LangChain, LlamaIndex, CrewAI and Haystack, an MCP server, and a docling-serve API server.

Key features

  • 01Parses PDF, Office files, HTML, EPUB, email, images and audio into one document model
  • 02Advanced PDF understanding covering layout, reading order, tables and formulas
  • 03OCR for scanned PDFs and images
  • 04Exports to Markdown, HTML and lossless JSON
  • 05Runs locally for sensitive data and air-gapped setups
  • 06Integrations with LangChain, LlamaIndex, CrewAI and Haystack

Use cases

  • Turning PDF reports or papers into Markdown for RAG and LLM ingestion
  • Extracting tables from financial or research documents into structured data
  • Building a document-processing service with the docling-serve API
  • Feeding office documents into agent frameworks through LangChain or LlamaIndex

Quick start

  1. 1

    Install the package

    pip install docling
  2. 2

    Convert a document from the command line

    docling https://arxiv.org/pdf/2206.01062
  3. 3

    Convert a document in Python

Things to know

  • Requires Python 3.10 or higher, as Python 3.9 support was dropped in version 2.70.0.
  • Some advanced features such as VLM and ASR models rely on separate model downloads, and their own licenses apply.

Topics

#document-parsing#pdf#ocr#llm#rag#markdown#python#data-extraction#ai#convert

FAQ

What is docling?
Open-source document parser that converts PDF, Office, HTML and other files into structured data for generative AI pipelines. Docling is a Python library and CLI for parsing documents and converting them into a unified representation called DoclingDocument. It accepts a wide range of inputs, including PDF, DOCX, PPTX, XLSX, HTML, EPUB, email files, images and audio, and exports the result as Markdown, HTML, lossless JSON or other formats.
Is docling free to use? What license does it have?
docling is open source under the MIT license. Check the license terms before using it in commercial products.
Is docling still maintained?
Actively maintained. The last commit to docling was on Oct 19, 2026.
What can I use docling for?
Turning PDF reports or papers into Markdown for RAG and LLM ingestion Extracting tables from financial or research documents into structured data Building a document-processing service with the docling-serve API Feeding office documents into agent frameworks through LangChain or LlamaIndex
How do I install docling?
Install the package: pip install docling Convert a document from the command line: docling https://arxiv.org/pdf/2206.01062 Convert a document in Python
Where can I get docling?
The source code and installation instructions are on GitHub: https://github.com/docling-project/docling

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects