feat(stats): query r2 data catalog

packages/stats/core/src/domain/inference.test.ts

@@ -1,5 +1,5 @@
 import { describe, expect, test } from "bun:test"
-import { toGeoAggregate, toModelAggregate, toProviderAggregate } from "./inference"
+import { buildStatsQueries, toGeoAggregate, toModelAggregate, toProviderAggregate } from "./inference"
 import { modelAuthor, normalizeInferenceModel, statModel, statProvider } from "./model-normalization"
 
 describe("inference stat normalization", () => {
@@ -82,6 +82,27 @@ describe("inference stat normalization", () => {
       }),
     ).toMatchObject([{ period_key: "2026-W20" }])
   })
+
+  test("builds bounded R2 SQL queries for each day and week", () => {
+    const queries = buildStatsQueries(new Date("2026-08-10T00:00:00.000Z"), new Date("2026-08-12T12:00:00.000Z"), {
+      namespace: "inference",
+      table: "generation",
+      dataset: "zen",
+    })
+
+    expect(queries).toHaveLength(8)
+    expect(queries[0]).toContain("'week' AS grain")
+    expect(queries[0]).toContain("'2026-W33' AS period_key")
+    expect(queries[2]).toContain("'2026-08-10' AS period_key")
+    expect(queries[6]).toContain("'2026-08-12' AS period_key")
+    expect(queries[0]).toContain('FROM "inference"."generation"')
+    expect(queries[0]).toContain("event_type = 'generation.completed'")
+    expect(queries[0]).toContain("product = 'go'")
+    expect(queries[0]).toContain("LIMIT 10000")
+    expect(queries[0]).toContain("approx_distinct(session) AS sessions")
+    expect(queries[1]).toContain("'geo_model' ELSE 'geo'")
+    expect(queries[1]).toContain("0 AS sessions")
+  })
 })
 
 function aggregate(model: string, provider: string) {