Stream an export

Build an ordered event query using the shared setup, then consume its async iterator.

Query

export function exportEvents(db: AnalyticsDB, tenantId: number) {
  return db
    .selectFrom("events")
    .prewhere("tenant_id", "=", tenantId)
    .select("user_id", "event_type", "created_at")
    .orderBy("created_at")
    .orderBy("user_id")
    .orderBy("event_type");
}

const query = exportEvents(db, 1);
for await (const row of query.stream({ abortSignal: requestSignal })) {
  await writeRow(row);
}

requestSignal comes from your request or job, and writeRow is your destination's async write function. Await each write so a slow destination controls consumption. With Node writable streams, handle the writable's backpressure when implementing that function. Avoid collecting rows into an array in the actual export.

The iterator yields { user_id: string; event_type: string; created_at: string }. The client receives chunks, so this does not promise one network read per row or zero buffering. Keep the shared client open until consumption ends.

SQL

SELECT user_id, event_type, created_at
FROM events
PREWHERE tenant_id = {p0:Int64}
ORDER BY created_at ASC, user_id ASC, event_type ASC

Parameters: p0 = 1.

Rows yielded

Shown together for readability; the export consumes them individually:

[
  { "user_id": "1001", "event_type": "signup", "created_at": "2026-08-01 10:00:00" },
  { "user_id": "1002", "event_type": "signup", "created_at": "2026-08-02 10:00:00" },
  { "user_id": "1001", "event_type": "purchase", "created_at": "2026-08-03 10:00:00" },
  { "user_id": "1001", "event_type": "purchase", "created_at": "2026-08-04 10:00:00" }
]

Errors may arrive after some rows have been written. Choose how your export reports partial output. Streaming does not support WITH TOTALS; use executeWithTotals() for that result shape.

Streaming is available in 0.9.1; the abortSignal option requires 0.10.0 or newer. See release guidance and the execution reference.