opendataloader-project/opendataloader-pdf

PDF Parser for AI-ready data. Automate PDF accessibility. Open-source.

OpenDataLoader PDF – AI‑ready PDF parsing & accessibility

What it does – Turns PDFs (digital, scanned, or already‑tagged) into structured, AI‑friendly formats:

  • Markdown, JSON (with per‑element bounding boxes), HTML – ready for chunking and retrieval‑augmented generation.
  • Tagged PDF – automatically adds PDF/UA‑compatible tags so screen‑readers can read the document.
  • Enterprise add‑on – PDF/UA‑1/2 export and a visual “accessibility studio”.

Why it matters – PDF content is notoriously hard for LLMs: reading order is lost, tables break, and there are no coordinates for citations. OpenDataLoader PDF provides a deterministic local parser plus an optional “hybrid” AI mode that routes difficult pages to a remote model, achieving the highest benchmark scores among open‑source PDF parsers (overall 0.907, table 0.928).

Key technical points

  • Local engine (Java 11+, called from Python/Node/Java SDKs) gives fast, deterministic output with XY‑Cut++ reading‑order detection.
  • Hybrid mode – a lightweight server (opendataloader-pdf-hybrid) forwards complex pages to an AI backend for OCR, table reconstruction, formula extraction (LaTeX), and image‑description generation (uses a 256 M SmolVLM model). This yields >90 % table accuracy on border‑less tables.
  • Bounding boxes for every element (text, heading, list, image, table, formula) enable precise source‑citation in RAG pipelines.
  • Accessibility – free auto‑tagging follows the PDF Association’s Well‑Tagged PDF spec and is validated with veraPDF. Full PDF/UA compliance is offered as a paid enterprise feature.
  • Multi‑language OCR – 80+ languages via EasyOCR (default) or alternative engines (RapidOCR, Tesseract, Apple Vision, NVIDIA Nemotron OCR).

How to use (quick‑start)

pip install -U opendataloader-pdf   # Python SDK (requires Java 11+)

import opendataloader_pdf
opendataloader_pdf.convert(
    input_path=["file1.pdf", "folder/"],
    output_dir="out/",
    format="markdown,json"
)

Node and Java SDKs are available with analogous commands.

When to enable hybrid mode – For PDFs that contain:

  • Complex or border‑less tables
  • Scanned images (OCR needed)
  • Mathematical formulas
  • Charts that need AI‑generated alt‑text Install the extra dependency and run the backend server, e.g.:
pip install "opendataloader-pdf[hybrid]"
opendataloader-pdf-hybrid --port 5002   # start AI server
opendataloader-pdf --hybrid docling-fast file.pdf   # client call

Performance – Local mode processes a page in ~0.015 s; hybrid mode adds ~0.45 s/page for AI‑enhanced pages. No GPU is required for the core parser; AI enrichments run on CPU (or optional GPU‑enabled OCR backends).

Ecosystem – The library ships on PyPI, npm, and Maven Central, making it easy to embed in Python, JavaScript, or Java applications. A LangChain integration is provided for seamless RAG pipelines.

License & pricing – Core extraction and auto‑tagging are Apache‑2.0 (free). PDF/UA export, the visual accessibility studio, and enterprise‑grade support are commercial add‑ons.

In short, OpenDataLoader PDF is a genuine open‑source tool that bridges PDF parsing and accessibility with the needs of modern LLM‑driven workflows.

Related

  • Project
    PaddlePaddle/PaddleOCR

    PaddleOCR is an open‑source OCR and document‑AI toolkit that converts images or PDFs into structured, LLM‑ready text, tables and markdown/JSON. It offers multilingual scene‑text models (PP‑OCRv6) and lightweight vision‑language models (PaddleOCR‑VL, PP‑StructureV3) with state‑of‑the‑art accuracy, fast CPU/GPU inference, and deployment options ranging from Docker to browser SDKs. Integrated with RAG platforms like Dify and RAGFlow, it serves as a data‑engine for building AI agents and retrieval‑augmented generation pipelines.

  • Project
    pymupdf/PyMuPDF

    PyMuPDF is a high-performance Python library for PDF and other document processing, offering fast text extraction, layout analysis, conversion, rendering, and manipulation. It provides LLM-ready outputs like Markdown and JSON, runs locally, and supports OCR, tables, forms, and redaction.

  • Project
    docling-project/docling-parse

    Docling Parse is a fast, configurable PDF‑parsing library (Python API with a C++ core) that extracts text, words, lines, vector shapes and bitmap images together with their coordinates. It supports both sequential and multi‑threaded parsing, lets you choose what to compute or materialize per page, and includes rendering utilities. Ideal for AI workflows that need layout‑aware document data.

  • Project
    magicrew/doc7

    A document conversion tool that turns PDFs, Office files, and images into AI-ready Markdown using multimodal vision models, supporting local inference for privacy and cost-efficiency.

  • Project
    docling-project/docling

    Docling is an open‑source Python library/CLI that parses many document types (PDF, Word, slides, spreadsheets, HTML, e‑books, audio, video, email, etc.) into a unified, structured representation. It extracts layout, tables, code, formulas, images, and speech‑to‑text, then exports to Markdown, HTML, JSON, DocLang, DocTags, etc. Designed for generative‑AI workflows, it offers plug‑and‑play integrations with LangChain, LlamaIndex, Crew AI, Haystack, and supports local execution for privacy‑sensitive data.