Function TestBulkLoadVectorEdgeCases

systest/vector/load_test.go:631–784

(t *testing.T)

}

func TestBulkLoadVectorEdgeCases(t *testing.T) {
	// Schema with multiple vector predicates - some will have data, some won't
	schema := `
		vec_with_data: float32vector @index(hnsw(exponent: "5", metric: "euclidean")) .
		vec_single: float32vector @index(hnsw(exponent: "5", metric: "euclidean")) .
		vec_empty: float32vector @index(hnsw(exponent: "5", metric: "euclidean")) .
		name: string @index(exact) .
	`

	vectorDim := 10

	// Step 1: Create source cluster
	sourceConf := dgraphtest.NewClusterConfig().
		WithNumAlphas(1).
		WithNumZeros(1).
		WithReplicas(1).
		WithACL(time.Hour)
	sourceCluster, err := dgraphtest.NewLocalCluster(sourceConf)
	require.NoError(t, err)
	defer func() { sourceCluster.Cleanup(t.Failed()) }()
	require.NoError(t, sourceCluster.Start())

	gc, cleanup, err := sourceCluster.Client()
	require.NoError(t, err)
	defer cleanup()
	require.NoError(t, gc.LoginIntoNamespace(context.Background(),
		dgraphapi.DefaultUser, dgraphapi.DefaultPassword, x.RootNamespace))

	hc, err := sourceCluster.HTTPClient()
	require.NoError(t, err)
	require.NoError(t, hc.LoginIntoNamespace(dgraphapi.DefaultUser,
		dgraphapi.DefaultPassword, x.RootNamespace))

	require.NoError(t, gc.SetupSchema(schema))

	// Generate data:
	// - vec_with_data: 100 vectors
	// - vec_single: 1 vector
	// - vec_empty: 0 vectors (schema only)

	var rdfBuilder strings.Builder
	var vectorsWithData [][]float32

	// Add 100 vectors to vec_with_data
	for i := 0; i < 100; i++ {
		vec := dgraphapi.GenerateRandomVector(vectorDim)
		vectorsWithData = append(vectorsWithData, vec)
		vecStr := fmt.Sprintf(`"[%s]"`, strings.Trim(strings.Join(strings.Fields(fmt.Sprint(vec)), ", "), "[]"))
		rdfBuilder.WriteString(fmt.Sprintf("<0x%x> <vec_with_data> %s .\n", i+10, vecStr))
		rdfBuilder.WriteString(fmt.Sprintf("<0x%x> <name> \"entity_%d\" .\n", i+10, i))
	}

	// Add single vector to vec_single
	singleVec := dgraphapi.GenerateRandomVector(vectorDim)
	singleVecStr := fmt.Sprintf(`"[%s]"`, strings.Trim(strings.Join(strings.Fields(fmt.Sprint(singleVec)), ", "), "[]"))
	rdfBuilder.WriteString(fmt.Sprintf("<0x%x> <vec_single> %s .\n", 1000, singleVecStr))
	rdfBuilder.WriteString(fmt.Sprintf("<0x%x> <name> \"single_entity\" .\n", 1000))