lorax-arg 0.1

Lorax single-port app (bundled UI + backend)

lorax-arg

lorax-arg is a pip-installable distribution that runs Lorax as a single-port app. It ships the backend package in the same wheel so the install does not depend on a separate lorax PyPI release. CLI entrypoint: lorax (with lorax-arg as an alias).

• React UI served at /
• Backend API served at /api/*
• Socket.IO served at /api/socket.io/

Development (monorepo)

This package can run against the backend Python package from packages/backend/ during development.

Build + sync UI assets (for wheel builds)

Build the website with VITE_API_BASE=/api, then copy the dist/ output into the Python package:

npm ci
VITE_API_BASE=/api npm --workspace packages/website run build
python packages/app/scripts/sync_ui_assets.py

Run from source without copying assets

export LORAX_APP_STATIC_DIR=packages/website/dist
python -m pip install -e .
lorax

Clean (remove build artifacts)

rm -rf build dist *.egg-info
find packages/app -type d -name __pycache__ -prune -exec rm -rf {} +

Production build (bundled UI)

Build the UI and bundle it into the Python package, then build the wheel/sdist:

npm ci
VITE_API_BASE=/api npm --workspace packages/website run build
python packages/app/scripts/sync_ui_assets.py

python -m pip install -U build
python -m build .

You can also run:

lorax build

Production install

Install the built wheel locally (recommended for production verification):

python -m pip install --force-reinstall dist/lorax_arg-*.whl
lorax

Publish online (PyPI)

1. Update version in the repo-root pyproject.toml.
2. Build fresh artifacts:

rm -rf build dist *.egg-info
npm ci
VITE_API_BASE=/api npm --workspace packages/website run build
python packages/app/scripts/sync_ui_assets.py
python -m pip install -U build twine
python -m build .

3. Upload to TestPyPI (recommended), then PyPI:

python -m twine upload --repository testpypi dist/*
# verify: python -m pip install -i https://test.pypi.org/simple lorax-arg
python -m twine upload dist/*

Notes:

• You need PyPI credentials (or a token) configured in ~/.pypirc or via TWINE_USERNAME/TWINE_PASSWORD.
• Keep README.md and metadata in sync with what you want on PyPI.