add index after adding data without vector schema for vector predicate

chunker/json_parser_test.go

@@ -77,8 +77,8 @@ type Person struct {
 type Product struct {
 	Uid           string `json:"uid,omitempty"`
 	Name          string `json:"name"`
-	Discription   string `json:"discription"`
-	Discription_v string `json:"discription_v"`
+	Description   string `json:"description"`
+	Description_v string `json:"description_v"`
 }
 
 func Parse(b []byte, op int) ([]*api.NQuad, error) {
@@ -1406,7 +1406,7 @@ func TestNquadsJsonEmptyStringVectorPred(t *testing.T) {
 	p := Product{
 		Uid:           "1",
 		Name:          "",
-		Discription_v: "",
+		Description_v: "",
 	}
 
 	b, err := json.Marshal([]Product{p})
@@ -1420,16 +1420,16 @@ func TestNquadsJsonEmptyStringVectorPred(t *testing.T) {
 	require.NoError(t, err)
 	require.Equal(t, 3, len(fastNQ))
 
-	// predicate Name should be empty and edge for Discription_v should not be there
+	// predicate Name should be empty and edge for Description_v should not be there
 	// we do not create edge for "" in float32vector.
 	exp := &Experiment{
 		t:   t,
 		nqs: nq,
 		schema: `name: string @index(exact) .
-		discription_v: float32vector .`,
+		description_v: float32vector .`,
 		query: `{product(func: uid(1)) {
 			name
-			discription_v
+			description_v
 		}}`,
 		expected: `{"product":[{
 			"name":""}]}`,
@@ -1443,7 +1443,7 @@ func TestNquadsJsonEmptyStringVectorPred(t *testing.T) {
 func TestNquadsJsonEmptySquareBracketVectorPred(t *testing.T) {
 	p := Product{
 		Name:          "ipad",
-		Discription_v: "[]",
+		Description_v: "[]",
 	}
 
 	b, err := json.Marshal(p)
@@ -1457,16 +1457,16 @@ func TestNquadsJsonEmptySquareBracketVectorPred(t *testing.T) {
 	require.NoError(t, err)
 	require.Equal(t, 3, len(fastNQ))
 
-	// predicate Name should have value "ipad" and edge for Discription_v should not be there
+	// predicate Name should have value "ipad" and edge for Description_v should not be there
 	// we do not create edge for [] in float32vector.
 	exp := &Experiment{
 		t:   t,
 		nqs: nq,
 		schema: `name: string @index(exact) .
-		discription_v: float32vector .`,
+		description_v: float32vector .`,
 		query: `{product(func: eq(name, "ipad")) {
 			name
-			discription_v
+			description_v
 		}}`,
 		expected: `{"product":[{
 			"name":"ipad"}]}`,
@@ -1480,7 +1480,7 @@ func TestNquadsJsonEmptySquareBracketVectorPred(t *testing.T) {
 func TestNquadsJsonValidVector(t *testing.T) {
 	p := Product{
 		Name:          "ipad",
-		Discription_v: "[1.1, 2.2, 3.3]",
+		Description_v: "[1.1, 2.2, 3.3]",
 	}
 
 	b, err := json.Marshal(p)
@@ -1498,14 +1498,14 @@ func TestNquadsJsonValidVector(t *testing.T) {
 		t:   t,
 		nqs: nq,
 		schema: `name: string @index(exact) .
-		discription_v: float32vector .`,
+		description_v: float32vector .`,
 		query: `{product(func: eq(name, "ipad")) {
 			name
-			discription_v
+			description_v
 		}}`,
 		expected: `{"product":[{
 			"name":"ipad",
-			"discription_v":[1.1, 2.2, 3.3]}]}`,
+			"description_v":[1.1, 2.2, 3.3]}]}`,
 	}
 	exp.verify()
 

dgraphapi/vector.go

@@ -90,7 +90,7 @@ func UnmarshalVectorResp(resp *api.Response) ([][]float32, error) {
 	type Data struct {
 		Vector []struct {
 			UID                 string    `json:"uid"`
-			ProjectDescriptionV []float32 `json:"project_discription_v"`
+			ProjectDescriptionV []float32 `json:"project_description_v"`
 		} `json:"vector"`
 	}
 	var data Data

posting/index.go

@@ -168,8 +168,7 @@ func (txn *Txn) addIndexMutations(ctx context.Context, info *indexMutationInfo)
 		//       Similarly, the current assumption is that we have at most one
 		//       Vector Index, but this assumption may break later.
 		if info.op != pb.DirectedEdge_DEL &&
-			len(data) > 0 && data[0].Tid == types.VFloatID &&
-			len(info.factorySpecs) > 0 {
+			len(data) > 0 && len(info.factorySpecs) > 0 {
 			// retrieve vector from inUuid save as inVec
 			inVec := types.BytesAsFloatArray(data[0].Value.([]byte))
 			tc := hnsw.NewTxnCache(NewViTxn(txn), txn.StartTs)
@@ -1387,6 +1386,37 @@ func rebuildTokIndex(ctx context.Context, rb *IndexRebuild) error {
 	builder.fn = func(uid uint64, pl *List, txn *Txn) ([]*pb.DirectedEdge, error) {
 		edge := pb.DirectedEdge{Attr: rb.Attr, Entity: uid}
 		edges := []*pb.DirectedEdge{}
+
+		if runForVectors {
+			val, err := pl.Value(txn.StartTs)
+			if err != nil {
+				return []*pb.DirectedEdge{}, err
+			}
+
+			if val.Tid != types.VFloatID {
+				sv, err := types.Convert(val, types.VFloatID)
+				if err != nil {
+					return []*pb.DirectedEdge{}, err
+				}
+				b := types.ValueForType(types.BinaryID)
+				if err = types.Marshal(sv, &b); err != nil {
+					return []*pb.DirectedEdge{}, err
+				}
+				edge.Value = b.Value.([]byte)
+				edge.ValueType = types.VFloatID.Enum()
+
+				inKey := x.DataKey(edge.Attr, uid)
+				p, err := txn.Get(inKey)
+				if err != nil {
+					return []*pb.DirectedEdge{}, err
+				}
+
+				if err := p.addMutation(ctx, txn, &edge); err != nil {
+					return []*pb.DirectedEdge{}, err
+				}
+			}
+		}
+
 		err := pl.Iterate(txn.StartTs, 0, func(p *pb.Posting) error {
 			// Add index entries based on p.
 			val := types.Val{
@@ -1419,7 +1449,7 @@ func rebuildTokIndex(ctx context.Context, rb *IndexRebuild) error {
 		}
 		return edges, err
 	}
-	if len(factorySpecs) != 0 {
+	if runForVectors {
 		return builder.RunWithoutTemp(ctx)
 	}
 	return builder.Run(ctx)

query/vector/vector_graphql_test.go

@@ -19,12 +19,15 @@
 package query
 
 import (
+	"context"
 	"encoding/json"
 	"fmt"
 	"math/rand"
 	"testing"
 
+	"github.com/dgraph-io/dgo/v240/protos/api"
 	"github.com/dgraph-io/dgraph/v24/dgraphapi"
+	"github.com/dgraph-io/dgraph/v24/x"
 	"github.com/stretchr/testify/require"
 )
 
@@ -256,3 +259,56 @@ func testVectorGraphQlMutationAndQuery(t *testing.T, hc *dgraphapi.HTTPClient) {
 		}
 	}
 }
+
+func TestVectorIndexDropPredicate(t *testing.T) {
+	gc, cleanup, err := dc.Client()
+	require.NoError(t, err)
+	defer cleanup()
+	require.NoError(t, gc.LoginIntoNamespace(context.Background(),
+		dgraphapi.DefaultUser, dgraphapi.DefaultPassword, x.GalaxyNamespace))
+
+	hc, err := dc.HTTPClient()
+	require.NoError(t, err)
+	require.NoError(t, hc.LoginIntoNamespace(dgraphapi.DefaultUser,
+		dgraphapi.DefaultPassword, x.GalaxyNamespace))
+
+	require.NoError(t, hc.UpdateGQLSchema(fmt.Sprintf(graphQLVectorSchema, "euclidean")))
+
+	var vectors [][]float32
+	numProjects := 100
+	projects := generateProjects(numProjects)
+	for _, project := range projects {
+		vectors = append(vectors, project.TitleV)
+		addProject(t, hc, project)
+	}
+
+	schemaWithoutIndex := `type Project  {
+		id: ID!
+		title: String!  @search(by: [exact])
+		title_v: [Float!] @embedding
+	} `
+
+	require.NoError(t, hc.UpdateGQLSchema(schemaWithoutIndex))
+
+	op := &api.Operation{
+		DropAttr: "title_v",
+	}
+	require.NoError(t, gc.Alter(context.Background(), op))
+
+	numProjects = 100
+	projects = generateProjects(numProjects)
+	for _, project := range projects {
+		vectors = append(vectors, project.TitleV)
+		addProject(t, hc, project)
+	}
+
+	require.NoError(t, hc.UpdateGQLSchema(fmt.Sprintf(graphQLVectorSchema, "euclidean")))
+
+	// similar to query
+	for _, project := range projects {
+		similarProjects := queryProjectsSimilarByEmbedding(t, hc, project.TitleV, numProjects)
+		for _, similarVec := range similarProjects {
+			require.Contains(t, vectors, similarVec.TitleV)
+		}
+	}
+}

systest/vector/backup_test.go

@@ -54,7 +54,7 @@ func TestVectorIncrBackupRestore(t *testing.T) {
 	require.NoError(t, gc.SetupSchema(testSchema))
 
 	numVectors := 500
-	pred := "project_discription_v"
+	pred := "project_description_v"
 	allVectors := make([][][]float32, 0, 5)
 	allRdfs := make([]string, 0, 5)
 	for i := 1; i <= 5; i++ {
@@ -78,7 +78,7 @@ func TestVectorIncrBackupRestore(t *testing.T) {
 		require.NoError(t, hc.Restore(c, dgraphtest.DefaultBackupDir, "", incrFrom, i))
 		require.NoError(t, dgraphapi.WaitForRestore(c))
 		query := `{
-			vector(func: has(project_discription_v)) {
+			vector(func: has(project_description_v)) {
 				   count(uid)
 				}
 		}`
@@ -131,7 +131,7 @@ func TestVectorBackupRestore(t *testing.T) {
 	require.NoError(t, gc.SetupSchema(testSchema))
 
 	numVectors := 1000
-	pred := "project_discription_v"
+	pred := "project_description_v"
 	rdfs, vectors := dgraphapi.GenerateRandomVectors(0, numVectors, 10, pred)
 
 	mu := &api.Mutation{SetNquads: []byte(rdfs), CommitNow: true}
@@ -171,7 +171,7 @@ func TestVectorBackupRestoreDropIndex(t *testing.T) {
 	require.NoError(t, gc.SetupSchema(testSchema))
 	// add data to the vector predicate
 	numVectors := 3
-	pred := "project_discription_v"
+	pred := "project_description_v"
 	rdfs, vectors := dgraphapi.GenerateRandomVectors(0, numVectors, 1, pred)
 	mu := &api.Mutation{SetNquads: []byte(rdfs), CommitNow: true}
 	_, err = gc.Mutate(mu)
@@ -216,7 +216,7 @@ func TestVectorBackupRestoreDropIndex(t *testing.T) {
 	require.NoError(t, dgraphapi.WaitForRestore(c))
 
 	query := ` {
-		vectors(func: has(project_discription_v)) {
+		vectors(func: has(project_description_v)) {
 			   count(uid)
 			 }
 		}`
@@ -258,7 +258,7 @@ func TestVectorBackupRestoreReIndexing(t *testing.T) {
 	require.NoError(t, gc.SetupSchema(testSchema))
 
 	numVectors := 1000
-	pred := "project_discription_v"
+	pred := "project_description_v"
 	rdfs, vectors := dgraphapi.GenerateRandomVectors(0, numVectors, 10, pred)
 
 	mu := &api.Mutation{SetNquads: []byte(rdfs), CommitNow: true}

systest/vector/load_test.go

@@ -62,7 +62,7 @@ func testExportAndLiveLoad(t *testing.T, c *dgraphtest.LocalCluster, exportForma
 	require.NoError(t, gc.SetupSchema(testSchema))
 
 	numVectors := 100
-	pred := "project_discription_v"
+	pred := "project_description_v"
 	rdfs, vectors := dgraphapi.GenerateRandomVectors(0, numVectors, 10, pred)
 
 	mu := &api.Mutation{SetNquads: []byte(rdfs), CommitNow: true}
@@ -76,7 +76,7 @@ func testExportAndLiveLoad(t *testing.T, c *dgraphtest.LocalCluster, exportForma
 	require.NoError(t, gc.DropAll())
 
 	query := `{
-		vector(func: has(project_discription_v)) {
+		vector(func: has(project_description_v)) {
 			   count(uid)
 			}
 	}`