priorityclip-geo

Turn overlapping polygons into a deterministic, priority-respecting partition.

Overlapping administrative areas, work zones, or classification polygons cannot always coexist in a final layer. priorityclip assigns every shared square metre to exactly one source feature, preserves the total union, and writes an audit table showing what was removed.

Quick start

python -m pip install priorityclip-geo
priorityclip partition areas.gpkg \
  --layer candidates \
  --priority priority \
  --output partitioned.gpkg \
  --report areas.csv

Lower numeric values win. Equal values retain stable input order. Output rows are written in the order in which they were processed.

Reproducible example

The bundled example creates three synthetic polygons in EPSG:3857. No external data is required.

python examples/make_example.py
priorityclip partition examples/overlap.geojson \
  --priority priority \
  --output examples/partitioned.gpkg \
  --report examples/areas.csv

Expected area audit:

The result has zero pairwise overlap and the same 200 m² union as the input.

Python API

from priorityclip import PartitionResult, partition

result: PartitionResult = partition(frame, "priority")
result.geometries.to_file("partitioned.gpkg", layer="partitioned")
result.report.to_csv("areas.csv", index=False)

partition(frame, priority, *, tolerance=None) accepts a GeoDataFrame and the name of a numeric priority column. PartitionResult exposes:

• geometries: attributes plus the non-overlapping output geometry;
• report: source index, priority, source/result/removed area, part count, and whether the result became empty;
• source and result union areas;
• maximum coverage error and pairwise overlap.

Guarantees

• Lower numeric priority owns shared area.
• Equal priorities preserve source order.
• Source attributes and one output row per input row are retained.
• The output union matches the input union within a scale-aware tolerance.
• Pairwise result overlap remains below the same tolerance.
• CLI outputs are replaced atomically after successful processing.

Limits

• Input geometries must be valid Polygon or MultiPolygon features.
• The CRS must be projected with horizontal units in metres.
• Invalid, missing, or empty input geometries are rejected rather than repaired.
• Repeated pairwise overlap validation is intended for small and medium datasets; very large layers may need spatial indexing in a future release.

Run priorityclip --help for all CLI options. Exit code 0 means success and 2 means invalid input or a processing failure.

Development

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

Copyright 2026 Alena Nikitina. Licensed under the Apache License 2.0.