scigraphs-utils 0.1.0

Scientific graph layout utilities.

scigraphs-utils

Utilities for scientific graph layouts.

The Python import package is scigraphs_utils.

import numpy as np
from scigraphs_utils import sfdp_layout

edges = np.array([[0, 1], [1, 2], [2, 3]], dtype=np.int32)
xy = sfdp_layout(4, edges, seed=42, overlap="false", maxiter=100)

For Graphviz parameters that are not listed explicitly, pass them as keyword arguments. They are forwarded as Graphviz graph attributes:

from scigraphs_utils import graphviz_layout

xy = graphviz_layout(4, edges, engine="sfdp", start=42, K=1.2)

Development

The current MVP builds a native extension against a local Graphviz installation:

python -m pip install -e ".[test]"
pytest

To build against a private Graphviz prefix, run:

tools/build_graphviz.sh
GRAPHVIZ_PREFIX="$PWD/build/vendor/graphviz-install" python -m pip install -e ".[test]"

Wheels

The project is configured for cibuildwheel CPython 3.13 builds. The wheel build hook compiles a pinned Graphviz release into a private prefix and builds the native extension against that prefix. On Linux, auditwheel then bundles the required shared libraries into the wheel so users do not need Graphviz installed at runtime.

python -m pip install ".[wheel]"
python -m cibuildwheel --output-dir wheelhouse