# geometry-rtree
Part of the [boost_geometry](https://crates.io/crates/boost_geometry) workspace — a Rust port of [Boost.Geometry](https://www.boost.org/doc/libs/release/libs/geometry/). Most users should depend on the facade crate, which re-exports this one; depend on this crate directly only for a slimmer build.
An R-tree spatial index over the geometry kernel.
Mirrors `boost/geometry/index/rtree.hpp` and the support headers
under `boost/geometry/index/detail/`. Stores any [`Indexable`] value
(a value with an axis-aligned bounding box) and answers Boost's box
predicates plus logical/satisfies queries and k-nearest-neighbour
search. It also provides insertion, condensing removal, count,
iteration, clear, bulk packing, and opt-in serde persistence.
The split strategy is a type parameter of [`Rtree`]. The default is
[`AsymmetricRStarSplit`] with six-child branches and 12-value
leaves for insertion, and four-child branches/four-value leaves for
bulk packing; symmetric [`RStarSplit`], [`Quadratic`], and [`Linear`]
configurations remain available. Bulk loading via [`FromIterator`] uses
Sort-Tile-Recursive packing for a balanced tree in one pass.
See [`split`] for parameter semantics, validity constraints, tuning
guidance, and the benchmark evidence behind the default.
Cartesian, 2D, `f64` for v1.
### Index polygons and query a point
Implement [`Indexable`] for an application type by returning its
axis-aligned bounds, then build an [`Rtree`] from an iterator. This example
uses rectangular polygons, so a bounds intersection with a point is also an
exact polygon intersection. For other polygon shapes, treat the result as a
candidate set and apply an exact point-in-polygon test afterward.
```rust
use geometry_rtree::{Bounds, Indexable, Predicate, Rtree};
#[derive(Debug)]
struct Parcel {
name: &'static str,
boundary: [[f64; 2]; 5],
bounds: Bounds,
}
impl Parcel {
fn rectangle(name: &'static str, min: [f64; 2], max: [f64; 2]) -> Self {
Self {
name,
boundary: [
min,
[min[0], max[1]],
max,
[max[0], min[1]],
min,
],
bounds: Bounds::new(min, max),
}
}
}
impl Indexable for Parcel {
fn bounds(&self) -> Bounds {
self.bounds
}
}
let parcels = [
Parcel::rectangle("park", [0.0, 0.0], [4.0, 3.0]),
Parcel::rectangle("school", [5.0, 0.0], [8.0, 2.0]),
Parcel::rectangle("lake", [1.0, 5.0], [3.0, 7.0]),
];
let tree: Rtree<Parcel> = parcels.into_iter().collect();
let point = Bounds::point([2.0, 1.0]);
let hits = tree.query(Predicate::Intersects(point));
assert_eq!(hits.len(), 1);
assert_eq!(hits[0].name, "park");
assert_eq!(hits[0].boundary[0], [0.0, 0.0]);
```
Module layout:
* [`bounds`] — the axis-aligned box arithmetic (area, enlargement,
union, distance) the tree keys on.
* [`indexable`] — the [`Indexable`] trait.
* [`node`] — the leaf / branch [`Node`](node::Node) enum.
* [`split`] — the [`SplitParameters`] strategies.
* [`predicate`] — built-in and composable query predicates.
* [`rtree`](mod@rtree) — the [`Rtree`] and its mutation / query /
nearest / bulk-load operations.
* [`query_iter`] — [`QueryIter`] and [`QueryWithIter`], the lazy
spatial-query walks.
* [`nearest_iter`] — [`NearestIter`], the
unbounded nearest-first stream.
* [`values`] — iteration over every stored value.
* `serialization` (feature-gated) — serde value-sequence persistence.
* `search_frontier` / `nearest_bound` (crate-internal) — the nearest
search's stack-first frontier and k-th-best rank buffer.
[`Indexable`]: indexable::Indexable
## License
BSL-1.0 — see [LICENSE](https://github.com/pentatonick/boost_geometry/blob/main/LICENSE).