tpt-gis

A pure Rust, planetary-scale Geographic Information Systems engine — a zero-FFI replacement for the GDAL/PROJ/GEOS stack, with no_std support for edge/embedded targets and cloud-native streaming for planetary-scale raster/vector data.

License: Dual MIT / Apache-2.0, at your option. Status: Active development — see todo.md for the full roadmap.

Why

GDAL and PROJ are the backbone of global mapping, autonomous vehicles, and climate science — and a 30-year-old C++ monolith with a build system that is a rite of passage of pain. Integrating them into modern web backends, mobile apps, or edge devices means wrestling with FFI and a dependency chain most build systems were never designed for.

tpt-gis rewrites the core mathematical projections, geometry engines, and data formats from scratch in safe, no_std-capable Rust: zero-FFI coordinate transformations, high-performance spatial indexing, and modern cloud-native raster/vector parsing.

Architecture

tpt-gis/
├── tpt-gis-core      # CRS definitions, datum transformations, geodesic math (no_std)
├── tpt-gis-geom      # Vector geometry (points, lines, polygons), topological ops (no_std)
├── tpt-gis-raster    # Grid data, map algebra, Cloud Optimized GeoTIFF (COG) parsing
├── tpt-gis-io        # Parsers/writers for GeoJSON, WKB/WKT, Shapefile, GeoPackage
├── tpt-gis-index     # R-Trees, Quadtrees, H3/S2 hexagonal grid integration
└── tpt-gis           # Facade crate re-exporting the five crates behind feature flags

See spec.txt for the full design document, todo.md for the phased task breakdown, and docs/ for usage guides.

Quickstart

Add the facade crate to your Cargo.toml:

[dependencies]
tpt-gis = "0.1"

Then use any of the engine crates behind feature flags. Most users only need the core (geodesy) and geom (vector topology) features:

use tpt_gis::core::{geodesic, Ellipsoid, GeoPoint};
use tpt_gis::geom::{Point, Polygon};

// Geodesic distance between two places on the WGS84 ellipsoid (Vincenty).
let london = GeoPoint::new(51.5074, -0.1278);
let paris = GeoPoint::new(48.8566, 2.3522);
let km = geodesic::inverse(&Ellipsoid::WGS84, london, paris).unwrap().distance_m / 1000.0;
assert!((km - 343.6).abs() < 1.0); // ≈ 343.6 km

// Point-in-polygon test (zero-allocation, borrowed rings).
let ring = [
    Point::new(0.0, 0.0),
    Point::new(0.0, 1.0),
    Point::new(1.0, 1.0),
    Point::new(1.0, 0.0),
    Point::new(0.0, 0.0),
];
let zone = Polygon::from_exterior(&ring);
assert!(zone.contains_point(Point::new(0.5, 0.5)));

For cloud-native raster, enable raster and http to stream a remote Cloud Optimized GeoTIFF without downloading the whole file (see tpt-gis-raster).

Python bindings

tpt-gis ships first-class Python bindings — tpt-gis-py, built with pyo3 + maturin — so you can use the geometry, format, and geodesy engines from Python without leaving the Rust safety model.

Install from PyPI (wheels for Linux/macOS/Windows, Python 3.8+):

pip install tpt-gis-py

Or build from source:

cd bindings/tpt-gis-py
pip install maturin
maturin develop

import tpt_gis_py as gis

poly = gis.Geometry.from_wkt("POLYGON ((0 0, 10 0, 10 10, 0 10, 0 0))")
assert poly.contains_point(5.0, 5.0)

london = gis.GeoPoint(51.5074, -0.1278)
paris = gis.GeoPoint(48.8566, 2.3522)
gis.geodesic_distance(london, paris)  # ≈ 343_500 m on WGS84

See Python bindings for the full API (WKT/WKB/GeoJSON round-trip, shapefile reading, geodesic inverse/direct) and the current v0.1 scope (raster/COG, the index crate, GeoPackage, projections, and numpy interop are deferred).

Building

cargo build --workspace
cargo test --workspace

tpt-gis-core and tpt-gis-geom are no_std by default (enable the std feature for host-native math intrinsics) and are verified to cross-compile for embedded (thumbv7em-none-eabihf, riscv32imc-unknown-none-elf = ESP32-C3) and WebAssembly (wasm32-unknown-unknown, wasm32-wasip1) targets.

Examples

• examples/spatial-join-cli — MVP 1: a planetary-scale points-in-polygons spatial join engine CLI (reproject, join, GeoJSON/CSV output, throughput benchmark).
• examples/drone-geofence-demo — MVP 2: a real-time no_std drone geofence / navigation engine (breach detection + escape-vector).
• examples/wasm-geofence-demo — a browser/WASM demo that reuses the drone geofence check_position logic in a wasm-bindgen + canvas harness.

Crates

The workspace splits the engine into five crates plus a convenience facade crate tpt-gis that re-exports them behind feature flags — most users should depend on tpt-gis rather than the individual crates. The facade also exposes a prelude module for one-line imports of the common types.

Contributing

This project is currently issues-only — bug reports and design discussion via GitHub Issues are welcome, but code pull requests are not accepted at this time. See CONTRIBUTING.md. Contributions must use permissive (MIT/Apache-2.0-compatible) licensed dependencies — no GPL/LGPL.

License

Licensed under either of

• Apache License, Version 2.0
• MIT license

at your option. Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this project shall be dual licensed as above, without any additional terms or conditions.