JavaScript / WASM

<script type="module">
  // Self-hosted: this project's OWN GitHub Pages origin — no third-party CDN. The
  // ~MB engine wasm is fetched LAZILY by the first `await Dataset.fromString(...)`
  // below — NOT by this import line — so it never blocks the page paint.
  import { Dataset, DataFactory as DF } from "https://sparq.jeswr.org/wasm/sparq.js";

  const ds = await Dataset.fromString(
    '<http://ex/a> <http://ex/name> "Alice" .',
    "ntriples",
  );
  console.log(ds.size); // 1  (DatasetCore: size, add, delete, has, match, iterate)

  ds.add(DF.quad(
    DF.namedNode("http://ex/b"), DF.namedNode("http://ex/name"), DF.literal("Bob"),
  ));
  for (const q of ds.match(null, DF.namedNode("http://ex/name"), null)) {
    console.log(q.subject.value, "→", q.object.value);
  }

  // Drop to the full SPARQL engine when DatasetCore is not enough:
  console.log(ds.store.queryBoolean("ASK { ?s ?p ?o }")); // true
  ds.free();
</script>

sparq.js is a single self-contained ESM file published into this site at /wasm/sparq.js — the engine .wasm stays out of it, fetched lazily by the first await. The same named entry is also on an ESM CDN (the @jeswr/sparq npm package):

<script type="module">
  // Alternative: the same named entry from an ESM CDN (the published @jeswr/sparq npm package).
  import { Dataset } from "https://esm.sh/@jeswr/sparq";
</script>

Prefer Dataset (or SparqStore) in an app — the cold start is memoised, so it is paid at most once per page. The lower-level engine handle is also importable: the wasm-pack glue is itself an ESM module.

<script type="module">
  // Low-level: the wasm-pack `--target web` glue is itself a real ESM module.
  import init, { Store } from "https://sparq.jeswr.org/wasm/sparq_wasm.js";
  await init(); // lazily fetches + instantiates sparq_wasm_bg.wasm
  const store = Store.load("<a> <b> <c> .", "ntriples");
</script>