kenro-wasm

SpatiaLite-style spatial SQL for SQLite in JavaScript — PostGIS-compatible ST_ functions registered as JS-level UDFs on any wasm SQLite host: the official @sqlite.org/sqlite-wasm build, sql.js, and wa-sqlite. On hosts whose SQLite takes no UDFs at all — Cloudflare D1 and Durable Objects — the same functions run standalone, from JS.

kenro is a spatial SQL engine in pure Rust (~210 functions: the DE-9IM predicate family, overlay/repair/buffer, CRS transform, H3, GeoJSON, MVT vector tiles, spatial aggregates), golden-tested against PostGIS. This package is its wasm build plus one small adapter per host. The bundled wasm is the full feature tier (~661 KB gzipped, the EPSG registry included); smaller tiers are attached to the GitHub Releases.

Live demo: https://reearth.github.io/kenro/ — drag a GeoPackage in and query it, entirely client-side.

Usage

import sqlite3InitModule from "@sqlite.org/sqlite-wasm";
import initKenro, * as kenroWasm from "kenro-wasm";
import { registerKenro } from "kenro-wasm/sqlite-wasm";

await initKenro();
const sqlite3 = await sqlite3InitModule();
const db = new sqlite3.oo1.DB(":memory:");
registerKenro(db, kenroWasm, sqlite3); // sqlite3 namespace needed for aggregates

db.selectValue("SELECT ST_AsText(ST_GeomFromText('POINT(1 2)'))"); // POINT(1 2)

// sql.js
import { registerKenro } from "kenro-wasm/sqljs";
registerKenro(db, kenroWasm);

// wa-sqlite
import { registerKenro } from "kenro-wasm/wa-sqlite";
registerKenro(sqlite3, db, kenroWasm);

Host support matrix, per-host limitations (sql.js: no int64 → the h3_* functions error loudly; no R-tree module) and measured sizes: docs/wasm.md. The full function table with PostGIS/DuckDB comparison: docs/functions.md.

Cloudflare Workers, D1 and Durable Objects

Neither D1 nor Durable Object SQLite supports user-defined functions or an R-tree, so ST_Intersects(...) can never appear in their SQL. The split that does work: kenro derives a bounding box and tile cells at write time for plain SQL to index, then runs the exact predicate in JS on the survivors.

import wasmModule from "kenro-wasm/pkg/kenro_wasm_bg.wasm";  // Workers hand you the Module
import * as kenro from "kenro-wasm";
import { cellFilterSql } from "kenro-wasm/quadtree";

kenro.initSync({ module: wasmModule });                      // once per isolate

using win = kenro.Prepared.fromText(wkt, 4326);              // decode once per scan
const hits = rows.filter((r) => {
  using g = kenro.Prepared.fromBlob(r.geom);
  return g.stIntersects(win);
});

Four subpaths carry this, all typed:

| | | |---|---| | kenro-wasm → Prepared | a geometry decoded once, then predicates, GeoJSON/WKT output and reprojection chained off it | | kenro-wasm/quadtree | bounding box → variable-depth quadtree cell ids: the B-tree-indexable stand-in for the missing R-tree (sql.js lacks one too), with nothing to tune | | kenro-wasm/tiles | the same idea at one fixed zoom — simpler, and faster for windows near that zoom | | kenro-wasm/prepared | handle lifetimes where using isn't available |

A complete Worker on both backends, with tests that run in workerd: the Cloudflare example. API reference: docs/wasm.md.

License

MIT OR Apache-2.0, at your option.