compas-nest 0.1.1

2D irregular nesting (OpenNest) for COMPAS.

compas_nest

Documentation: https://petrasvestartas.github.io/compas_nest/

2D irregular nesting for the COMPAS framework — Python bindings for the OpenNest C++ engines, built with nanobind.

Nest polylines with holes into sheets with holes, with live terminal progress and compas_viewer visualization.

Two engines

Class	Engine	Notes
opennest_collision	physics / overlap-relaxation (np_nest)	dependency-free; iteration-budget driven; nests parts into holes
opennest	NFP + genetic algorithm (nfp_nest)	bundled Clipper2; generation/fitness driven; carries part attributes through placement

Install

pip install compas_nest

From source (editable)

One-step (uv, macOS + Windows Git Bash + Linux): creates a local .venv, installs every dependency, and builds the package:

git clone --recurse-submodules https://github.com/petrasvestartas/compas_nest.git
cd compas_nest
bash bash/install.sh

After editing C++ sources, rebuild without recreating the env: bash bash/build.sh --test.

conda:

conda env create -f environment.yml
conda activate compas_nest
pip install --no-build-isolation -ve .

plain pip:

pip install nanobind "scikit-build-core[pyproject]"
pip install --no-build-isolation -ve .

The C++ engine sources live under external/nest/ (nest_physics_cpp/ + opennest_cpp/, the latter bundling Clipper2 and a minimal Boost subset). They are self-contained — no CGAL/Boost/Eigen download is needed.

Quick start

from compas.geometry import Polyline
from compas_nest import nest_geo, nest_sheets, opennest_collision

def rect(x0, y0, w, h):
    return Polyline([[x0, y0, 0], [x0+w, y0, 0], [x0+w, y0+h, 0], [x0, y0+h, 0], [x0, y0, 0]])

geo = nest_geo()
geo.add_part(rect(0, 0, 20, 10), copies=3)
geo.add_part(rect(0, 0, 15, 15), holes=[rect(5, 5, 5, 5)], copies=2)

sheets = nest_sheets()
sheets.add_sheet(rect(0, 0, 100, 100), holes=[rect(40, 40, 10, 10)])

result = opennest_collision(iterations=2000, num_rotations=64).solve(geo, sheets)

for group in result.placed_polylines():
    print("sheet", group["sheet_id"], "->", len(group["parts"]), "parts")

# serialize placed polylines (with holes) + transformations to COMPAS JSON
result.to_json("data/output/quickstart.json")

See the examples/ folder for the viewer workflows, and the documentation for examples, API reference and credits.

License & credits

MIT — see LICENSE. Attributions for the underlying work are in CREDITS.md.