Docling

Overview

Docling simplifies document processing by parsing diverse formats, including advanced PDF understanding, and providing seamless integrations with the generative AI ecosystem. It offers a unified DoclingDocument representation and export formats such as Markdown, HTML, and lossless JSON.

Key Features

• Parses multiple document formats including PDF, DOCX, PPTX, XLSX, HTML, EPUB, Apple Pages, WAV, MP3, WebVTT, Box Notes, email formats, images, LaTeX, DocLang, and plain text.
• Advanced PDF understanding including page layout, reading order, table structure, code, formulas, and image classification.
• Unified DoclingDocument representation with export formats including Markdown, HTML, WebVTT, DocLang, DocTags, and lossless JSON.
• Local execution for sensitive data and air-gapped environments.
• Plug-and-play integrations with LangChain, LlamaIndex, Crew AI, and Haystack, plus MCP server and API server support.
• Extensive OCR support, Visual Language Models, and Automatic Speech Recognition for audio.

Use Cases

• Convert documents into structured Markdown or JSON for generative AI and RAG pipelines.
• Build RAG applications with integrations such as Haystack, LangChain, and LlamaIndex.
• Extract information from PDFs, office files, emails, EPUB, and images with OCR.
• Run document processing locally for sensitive or air-gapped environments.

Getting Started

• Install Docling with pip install docling.
• Use the Docling CLI to convert a document and generate a Markdown file with structured content.
• Use the Python DocumentConverter to convert a source and export the document to Markdown.
• Detailed installation and usage documentation is available.

Deployment & Requirements

• Python 3.10 or higher is required; Python 3.9 support was dropped in docling version 2.70.0.
• Works on macOS, Linux, and Windows for x86_64 and arm64 architectures.
• Local execution is available for sensitive data and air-gapped environments.
• Can run as a service with the API server (docling-serve).

Before You Adopt

• License: MIT. Review its terms before using, modifying, or distributing the project.
• Python 3.9 support was dropped in docling version 2.70.0, so deployments should use Python 3.10 or higher.
• Metadata extraction is listed as coming soon, indicating it is not currently supported.
• Local execution is available for sensitive data and air-gapped environments.