Install¶

Ursa is a Python package with a compiled Rust core. The wheels on PyPI include the core, so you do not need a Rust toolchain.

Requirements¶

Python 3.10 or later.
Linux (x86_64, aarch64), macOS (x86_64, Apple silicon), or Windows (x64).

Install the package¶

The package name on PyPI is ursa-graph. The import name is ursa.

pipuv

pip install ursa-graph

uv add ursa-graph

Install ursa-graph, not ursa

The PyPI name ursa belongs to a different, abandoned project. If you install it, import ursa does not give you this library.

Optional extras¶

Ursa moves data as Apache Arrow. pyarrow is the only required dependency. The extras below add one interop path each. Ursa imports the extra only when you call the matching function.

Extra	Adds	Enables
`polars`	`polars`	`result.to_polars()` and `ur.from_polars(df, ...)`
`networkx`	`networkx`	`ur.from_networkx(G)` and `ur.nodes_from_networkx(G)`
`numpy`	`numpy`	`ur.from_numpy(array)`
`scipy`	`scipy`	`ur.from_scipy_sparse(matrix)`
`all`	all of the above	everything

pip install 'ursa-graph[polars]'
pip install 'ursa-graph[all]'

pandas is not an extra. If pandas is already installed, ur.from_pandas(df, ...) and the frame constructors accept a pandas.DataFrame.

Check the installation¶

import ursa as ur

print(ur.__version__)        # the package version
print(ur.__core_version__)   # the Rust core version; the same number

If ur.__core_version__ is None, the compiled core did not load. Make sure you installed a wheel for your platform and Python version.

Install from source¶

A source install compiles the Rust core. You need a Rust toolchain (1.88 or later) and maturin.

git clone https://github.com/cldixon/ursa
cd ursa
uv sync          # builds the extension and installs the dev tools
uv run pytest    # runs the test suite

Next step¶

Continue with the quickstart.