Live pushdown to DuckDB, query by query

<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/@toclocoinc/lattice-grid@1.28.0/lattice-grid.min.css">

<div id="grid" style="height: 560px"></div>
<pre id="sql" style="margin-top: 12px"></pre>

<script type="module">
  import * as LatticeGrid from 'https://cdn.jsdelivr.net/npm/@toclocoinc/lattice-grid@1.28.0/modules/htmx.esm.js';

  // The grid ships no query engine. The page starts DuckDB-Wasm itself and hands
  // the adapter a live connection, so reaching for pushdown never grows the grid
  // you load.
  import * as duckdb from 'https://cdn.jsdelivr.net/npm/@duckdb/duckdb-wasm@1.29.0/+esm';
  const bundle = await duckdb.selectBundle(duckdb.getJsDelivrBundles());
  const worker = await duckdb.createWorker(bundle.mainWorker);
  const db = new duckdb.AsyncDuckDB(new duckdb.ConsoleLogger(duckdb.LogLevel.ERROR), worker);
  await db.instantiate(bundle.mainModule, bundle.pthreadWorker);
  const connection = await db.connect();

  // from is any SQL table expression. read_parquet streams the file, so a host
  // that honours range requests lets DuckDB read only the bytes a query touches.
  const adapter = LatticeGrid.duckdbAdapter({
    connection,
    from: "read_parquet('/demo-data/readings.parquet')",
  });
  const source = LatticeGrid.createPushdownSource({
    adapter,
    compute: LatticeGrid,   // the grid finishes anything SQL could not express
    pageSize: 200,          // only the visible window comes back, per interaction
  });

  // Wrap execute to show the statement DuckDB ran and how long it took.
  const sql = document.getElementById('sql');
  const inner = adapter.execute.bind(adapter);
  adapter.execute = async (query, request) => {
    const started = performance.now();
    const result = await inner(query, request);
    const ms = Math.round(performance.now() - started);
    const f = adapter.sqlFor(query);
    sql.textContent = f.sql + '\n-- ' + ms + ' ms, ' + result.rows.length + ' of ' + (result.total ?? '?') + ' rows';
    return result;
  };

  LatticeGrid.createGrid(document.getElementById('grid'), {
    rowKey: 'reading_id',
    toolPanel: { side: 'left', panels: ['filters', 'columns'] },
    columns: [
      { field: 'reading_id', title: '#', type: 'number', layout: { width: 110 } },
      { field: 'plant', title: 'Plant', filter: { type: 'set' } },
      { field: 'line', title: 'Line', filter: { type: 'set' } },
      { field: 'product', title: 'Product' },
      { field: 'bore_mm', title: 'Bore', type: 'millimetres',
        spec: { lower: 9.95, upper: 10.05, target: 10 } },
      { field: 'cycle_seconds', title: 'Cycle', type: 'seconds' },
      { field: 'in_spec', title: 'In spec', type: 'boolean' },
      { field: 'taken_at', title: 'Taken', type: 'dateString' },
    ],
    source,   // a hundred thousand rows; DuckDB answers each filter and sort in SQL
  });
</script>