Function buildCorpus

cli/workspace/memory/ranking_test.go:83–115

buildCorpus wires a FactIndex + VectorIndex over a temp dir, adds the corpus, backfills vectors, and returns a content→id map so tests can label samples.

(t *testing.T)

// buildCorpus wires a FactIndex + VectorIndex over a temp dir, adds the corpus,
// backfills vectors, and returns a content→id map so tests can label samples.
func buildCorpus(t *testing.T) (*FactIndex, *VectorIndex, map[string]string) {
	t.Helper()
	dir := t.TempDir()
	cfg := DefaultConfig()
	fi := NewFactIndex(dir, cfg, zap.NewNop())

	corpus := []struct{ content, category string }{
		{"OAuth login issues bearer tokens", "auth"},
		{"Postgres stores user rows in tables", "database"},
		{"Kubernetes rollout uses canary strategy", "deploy"},
		{"Zap emits structured logs to stdout", "logging"},
		{"Redis eviction relies on ttl", "cache"},
	}
	for _, c := range corpus {
		fi.AddFact(c.content, c.category, nil)
	}

	contentToID := make(map[string]string)
	items := make(map[string]string)
	for _, f := range fi.GetAll() {
		contentToID[f.Content] = f.ID
		items[f.ID] = f.Content
	}

	vi := NewVectorIndex(dir, conceptProvider{}, zap.NewNop())
	if !vi.Enabled() {
		t.Fatal("vector index should be enabled with a real provider")
	}
	if err := vi.BackfillFacts(context.Background(), items); err != nil {
		t.Fatalf("backfill: %v", err)
	}
	return fi, vi, contentToID
}

// TestBlendedBeatsKeyword is the headline A/B: on queries phrased with synonyms
// that never appear in the stored fact text, keyword retrieval is blind while