terra-route

Terra Route aims to be a fast library for routing on GeoJSON LineStrings networks, where LineStrings share identical coordinates.

Install

``
npm install terra-route
`

Docs

API Docs can be found here

Example

Here is a short example of how to use the TerraRoute class:

`typescript
import { FeatureCollection, LineString, Point, Feature } from "geojson";
import { TerraRoute } from "terra-route";

// We omit the literal values here in the README for brevity
import { network, startPoint, endPoint } from './network'

// Initialize TerraRoute instance
const router = new TerraRoute();

// network here is a FeatureCollection where the all
// identical coordinates are considered connected.
// We must build the route graph first before calling getRoute.
router.buildRouteGraph(network);

// Get shortest route where startPoint and endpoint are Feature of
// a coordinate node which is present on the network (i.e. the coordinate
// exists in one of the linestrings)
const route = router.getRoute(startPoint, endPoint);

console.log("Shortest route:", JSON.stringify(route, null, 2));
`

Terra Route Graph

Understanding and debugging routing graphs can be complicated. There is an additional helper library called Terra Route Graph (terra-route-graph) which can be used to help understand the graphs you are working with.

Benchmarks

The benchmarks make use of a series out route example route networks from OSM in a moderate sized section of East London. It runs against the GeoJSON Path Finder library and also the ngraph.graph library. You can run the benchmarks yourself using:

`
npm run benchmark
`

Here is an example output of a benchmark run for routing:

Terra Route         | █████ 151ms
GeoJSON Path Finder | ██████████████████ 571ms
ngraph.graph        | ██████████████████████████████████████████████████ 1564ms

Using default Haversine distance, Terra Route is approximately 3.75x faster than GeoJSON Path Finder with Haversine distance for A -> B path finding. If you pass in the CheapRuler distance metric (you can use the exposed createCheapRuler function), it is approximately x8 faster than GeoJSON Path Finder with Haversine distance.

For initialisation of the network, Terra Route is approximately 10x faster with Haversine than GeoJSON Path Finder. Terra Draw splits out instantiating the Class of the library from the actual graph building, which is done via buildRouteGraph`. This allows you to defer graph creation to an appropriate time.

Terra Route uses an bi-directional A algorithm for pathfinding and by default uses a four-ary heap for the underlying priority queue, although this is configurable.

Limitations

- Terra Route does not currently support weighting functions.
- Coordinates must be identical to be considered connected

Acknowledgements

Terra Route is inspired by the the prior art of geojson-path-finder and we use this library to help benchmark Terra Routes performance.

License

MIT