@yang-29/vctrs-wasm

A fast vector database that runs in the browser via WebAssembly. ~220KB, zero dependencies.

Same Rust core as the Python and Node.js bindings.

Install

npm install @yang-29/vctrs-wasm

Usage

import init, { VctrsDatabase } from "@yang-29/vctrs-wasm";

await init();

const db = new VctrsDatabase(384, "cosine");

db.add("doc1", new Float32Array(embedding), { title: "hello" });
db.add("doc2", new Float32Array(embedding2), { title: "world" });

const results = db.search(new Float32Array(query), 5);
for (const r of results) {
  console.log(r.id, r.distance, r.metadata);
  r.free();
}

db.free();

CDN (no bundler)

<script type="module">
import init, { VctrsDatabase } from "https://unpkg.com/@yang-29/vctrs-wasm/vctrs_wasm.js";
await init();
const db = new VctrsDatabase(4, "cosine");
</script>

API

| Method | Description | |--------|-------------| | new VctrsDatabase(dim, metric) | Create database. Metrics: "cosine", "euclidean", "dot" | | db.add(id, vector, metadata) | Add vector (Float32Array) with optional metadata | | db.upsert(id, vector, metadata) | Insert or update | | db.search(vector, k) | Find k nearest neighbors → SearchResult[] | | db.delete(id) | Delete by ID → boolean | | db.contains(id) | Check existence → boolean | | db.length | Vector count | | db.dim | Dimensionality |

SearchResult has .id, .distance, .metadata properties. Call .free() when done.

Notes

• Runs entirely in-memory — no persistence across page reloads
• Call .free() on results and the database to avoid memory leaks
• Single-threaded (no SIMD/BLAS) — suitable for datasets under ~10k vectors