PDFMind

An agent for extracting structured content from PDFs using LangGraph and OpenAI.

Features

• Extract and format text content from PDFs
• Convert tables to markdown format
• Extract images with AI-generated descriptions
• Use LangGraph for agent-based orchestration

Setup

# Install Poetry if you don't have it
curl -sSL https://install.python-poetry.org | python3 -

# Install dependencies
poetry install
# Or install from pypi
pip install pdf_mind

# Install other dependencies
brew install ghostscript
brew install poppler
# apt install ghostscript poppler

N.B.: if you're on OSX, the Ghostscript module may not be found. You can fix that by doing:

mkdir -p ~/lib
ln -s "$(brew --prefix gs)/lib/libgs.dylib" ~/lib

See the Camelot docs for more details on installing the dependency. It'll work without Ghostscript.

Usage

from pdf_mind import PDFExtractionAgent

agent = PDFExtractionAgent()
result = agent.process("path/to/document.pdf")
print(result)

Alternatively, look at example.py for an example that will output metadata on extracted items and token usage:

Development

# Run tests
poetry run pytest

# Lint code
poetry run ruff check .
poetry run black .