docling

What it is and what it does

Docling is a document processing SDK that converts PDFs, Word documents, spreadsheets, presentations, email, images, video, audio, and other formats into a unified, structured representation. It specializes in advanced PDF understanding—extracting page layout, reading order, table structure, code blocks, formulas, and charts—and can export to Markdown, HTML, JSON, and domain-specific schemas (DocLang, USPTO patents, JATS articles, XBRL financial reports). The library runs locally, supports OCR for scanned documents, integrates with Visual Language Models and ASR systems, and plugs into AI frameworks like LangChain, LlamaIndex, and Haystack.

It is designed for developers building document-aware AI applications, knowledge extraction pipelines, and content processing workflows. The package offers both a Python API and a command-line interface, with options to run as a service via an API server or as an MCP server for agent integration.

Use it for:

• Convert research papers or technical PDFs to structured Markdown for ingestion into RAG or LLM pipelines.
• Extract tables, charts, and text from financial reports (XBRL) or patent documents (USPTO) for automated analysis.
• Process scanned or image-based documents using OCR and Visual Language Models to recover structured content.
• Build document-aware chatbots or Q&A systems by parsing diverse input formats into a unified representation.
• Batch-convert email archives (EML, MSG) and office documents (DOCX, XLSX, PPTX) to Markdown or JSON for downstream workflows.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Docling parses diverse document formats—PDF, DOCX, PPTX, XLSX, HTML, EPUB, email, images, video, audio, and more—into a unified representation, with advanced PDF layout understanding and export to Markdown, HTML, JSON, and other formats.

Yes. Docling is actively maintained, production-stable, permissively licensed (MIT), and solves a real problem—unified parsing of many document formats with strong PDF understanding. Low install friction, no known vulnerabilities, and broad Python version support make it a solid choice for document-heavy AI and data extraction projects. Install if you need to parse or convert diverse document types at scale.

Install

docling on PyPI

pip install docling

uv add docling

poetry add docling

Installing docling

Before you install

Active maintenance with a recent release (0 days since last update) and high repository engagement (64769 stars). Single runtime dependency (docling-slim) keeps install friction low. Supports current Python versions (3.10–3.14).

License in practice

MIT license is permissive; you can use, modify, and distribute Docling freely in commercial and private projects with minimal restrictions.

Quickstart

pip install docling

from docling.document_converter import DocumentConverter

converter = DocumentConverter()
result = converter.convert("https://arxiv.org/pdf/2408.09869")
print(result.document.export_to_markdown())

Requires Python 3.10 or higher (3.9 support was dropped in version 2.70.0).

Verify before relying

• Performance characteristics and memory footprint for large documents or batch processing.
• Whether all advertised export formats (DocTags, DocLang, JATS, XBRL) are production-ready or still experimental.
• Accuracy and reliability of OCR and Visual Language Model integrations in real-world deployments.

Package facts

Evidence: docling-2.120.1-py3-none-any.whl

Keywords: convert, docling, document, docx, html, layout model, markdown, pdf, segmentation, table former, table structure

Tags

More Artificial Intelligence packages