Docling
Docling is a document processing tool that parses diverse formats including PDF, DOCX, and more, with advanced PDF understanding. It provides a unified representation and exports to Markdown, JSON, and other formats, and integrates with AI frameworks. It is for developers building document processing pipelines or AI applications.
✨ Key features
- Parses PDF, DOCX, PPTX, XLSX, HTML, EPUB, and more.
- Advanced PDF layout, reading order, table structure understanding.
- Unified DoclingDocument representation with lossless JSON export.
- Local execution for sensitive data and air-gapped environments.
- Integrations with LangChain, LlamaIndex, Crew AI, Haystack.
- OCR support for scanned PDFs and images.
🎯 Use cases
- Convert documents to Markdown for LLM ingestion.
- Extract tables and structure from PDFs for data analysis.
- Build RAG pipelines with LangChain or LlamaIndex.
- Process audio and video files with ASR transcripts.
- Parse financial reports in XBRL format.
📦 Installation
🧰 Requirements: Python 3.10 or higher, works on macOS, Linux, Windows (x86_64 and arm64). No API keys required.
pip install docling
🚀 Usage
from docling.document_converter import DocumentConverter
source = "https://arxiv.org/pdf/2408.09869" # a document via a local path or URL
converter = DocumentConverter()
result = converter.convert(source)
print(result.document.export_to_markdown()) # output: "## Docling Technical Report[...]"
⚠️ Good to know
Python 3.9 support dropped in version 2.70.0; metadata extraction and complex chemistry understanding are coming soon.
❓ FAQ
What Python versions are supported?
Docling requires Python 3.10 or higher; Python 3.9 support was dropped in version 2.70.0.
Can I run Docling locally without cloud services?
Yes, Docling supports local execution for sensitive data and air-gapped environments.
What output formats are available?
Docling can export to Markdown, HTML, WebVTT, DocLang, DocTags, and lossless JSON.
Does Docling support OCR for scanned documents?
Yes, Docling provides extensive OCR support for scanned PDFs and images.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.