geo

geo provides fundamental geospatial primitives (points, lines, polygons, multi-geometries) and a large library of algorithms — area, centroid, convex hull, boolean operations, buffering, simplification, distance/bearing calculations, and more — for working with 2D geometry in Rust. It builds on the geo-types crate for its core data structures and integrates with the wider georust ecosystem (WKT, GeoJSON, PROJ) for real-world GIS workflows.

What You Get

• Core geometry types re-exported from geo-types (Point, LineString, Polygon, MultiPolygon, etc.) plus algorithm traits implemented on them
• 68+ algorithm modules covering area, centroid, convex hull, buffering, simplification, boolean ops, distance/bearing/geodesic calculations, and Delaunay triangulation
• Optional integrations for PROJ (coordinate reprojection), serde (serialization), and spade (Delaunay triangulation backend)
• Multithreading support for compute-heavy operations like polygon overlay via the i_overlay dependency
• An extensive test suite including a JTS-compatibility test runner validating results against the reference Java Topology Suite

Common Use Cases

• Computing area, centroid, or bounding box of geographic features in a mapping or GIS application
• Performing polygon boolean operations (union, intersection, difference) for spatial analysis pipelines
• Simplifying or buffering geometries for rendering at different zoom levels in map-based applications
• Calculating geodesic distances and bearings between coordinates for routing or proximity queries

Under The Hood

Architecture - The geo crate (geo/src/algorithm/, 68+ modules) implements each geometric operation as a trait extending the core types defined in the sibling geo-types crate, so operations like .area(), .convex_hull(), or .simplify() are called directly on Point/LineString/Polygon values; the workspace also includes geo-traits (shared trait definitions for interop with other georust crates), geo-test-fixtures (shared test geometries), and jts-test-runner, which cross-validates geo’s algorithm outputs against Java Topology Suite reference results.

Tech Stack - Nearly pure Rust (99.96%) on the 2024 edition (rust-version 1.88), depending on geo-types for core structures, geographiclib-rs and robust for numerically-robust geodesic/predicate computations, rstar for spatial R-tree indexing, and optional integrations with spade (Delaunay triangulation), proj (coordinate reprojection via libproj bindings), and i_overlay (multithreaded polygon overlay).

Code Quality - The geo crate alone contains 129+ files with #[test] functions, and the dedicated jts-test-runner crate cross-checks algorithm correctness against the industry-standard JTS reference implementation, which is an unusually rigorous validation approach for a geometry library; the project has 134 contributors and very active recent commit history, indicating sustained peer review and maintenance.

API Design - Algorithms are exposed as traits implemented directly on the geometry types, so usage reads naturally as method calls (polygon.area(), line.haversine_distance(&other)) rather than free functions taking geometry arguments, and feature flags cleanly separate optional heavy dependencies (PROJ, spade) from the default build so users only pull in what they need.