Development Guide

Setup

Setup Python env

python -m venv .venv
. ./.venv/bin/activate

Install dependencies:

poetry install

Install CUDA 12.1 version of torch and torchvision (Linux/Windows, GPU support):

poetry run pip install torch==2.5.1+cu121 torchvision==0.20.1+cu121 --index-url https://download.pytorch.org/whl/cu121

Development Workflow

Run Tests

poetry run python test.py

Run Lint

Check code quality with pylint:

python lint.py

Or directly:

poetry run pylint doc_page_extractor

macOS Development Setup

For macOS developers, PyTorch CUDA version is not compatible. Use the following steps:

# Install only main dependencies first (skip dev group to avoid CUDA installation)
poetry install

# Install PyTorch CPU version
poetry run pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu

Build Package

Clean old builds and create distribution files:

python build.py

Before Submitting PR

Make sure all checks pass:

poetry run python test.py
python lint.py