milvus/tests/integration/getvector/get_vector_test.go
yihao.dai 014387fd94
Forbid to get quantized vector from ChunkManager (#24334)
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2023-05-24 23:03:27 +08:00

371 lines
11 KiB
Go

// Licensed to the LF AI & Data foundation under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package getvector
import (
"context"
"fmt"
"strconv"
"testing"
"github.com/golang/protobuf/proto"
"github.com/stretchr/testify/suite"
"github.com/milvus-io/milvus-proto/go-api/commonpb"
"github.com/milvus-io/milvus-proto/go-api/milvuspb"
"github.com/milvus-io/milvus-proto/go-api/schemapb"
"github.com/milvus-io/milvus/pkg/common"
"github.com/milvus-io/milvus/pkg/util/distance"
"github.com/milvus-io/milvus/pkg/util/funcutil"
"github.com/milvus-io/milvus/pkg/util/typeutil"
"github.com/milvus-io/milvus/tests/integration"
)
type TestGetVectorSuite struct {
integration.MiniClusterSuite
// test params
nq int
topK int
indexType string
metricType string
pkType schemapb.DataType
vecType schemapb.DataType
// expected
searchFailed bool
}
func (s *TestGetVectorSuite) run() {
ctx, cancel := context.WithCancel(s.Cluster.GetContext())
defer cancel()
collection := fmt.Sprintf("TestGetVector_%d_%d_%s_%s_%s",
s.nq, s.topK, s.indexType, s.metricType, funcutil.GenRandomStr())
const (
NB = 10000
dim = 128
)
pkFieldName := "pkField"
vecFieldName := "vecField"
pk := &schemapb.FieldSchema{
FieldID: 100,
Name: pkFieldName,
IsPrimaryKey: true,
Description: "",
DataType: s.pkType,
TypeParams: []*commonpb.KeyValuePair{
{
Key: common.MaxLengthKey,
Value: "100",
},
},
IndexParams: nil,
AutoID: false,
}
fVec := &schemapb.FieldSchema{
FieldID: 101,
Name: vecFieldName,
IsPrimaryKey: false,
Description: "",
DataType: s.vecType,
TypeParams: []*commonpb.KeyValuePair{
{
Key: common.DimKey,
Value: fmt.Sprintf("%d", dim),
},
},
IndexParams: nil,
}
schema := integration.ConstructSchema(collection, dim, false, pk, fVec)
marshaledSchema, err := proto.Marshal(schema)
s.Require().NoError(err)
createCollectionStatus, err := s.Cluster.Proxy.CreateCollection(ctx, &milvuspb.CreateCollectionRequest{
CollectionName: collection,
Schema: marshaledSchema,
ShardsNum: 2,
})
s.Require().NoError(err)
s.Require().Equal(createCollectionStatus.GetErrorCode(), commonpb.ErrorCode_Success)
fieldsData := make([]*schemapb.FieldData, 0)
if s.pkType == schemapb.DataType_Int64 {
fieldsData = append(fieldsData, integration.NewInt64FieldData(pkFieldName, NB))
} else {
fieldsData = append(fieldsData, integration.NewStringFieldData(pkFieldName, NB))
}
var vecFieldData *schemapb.FieldData
if s.vecType == schemapb.DataType_FloatVector {
vecFieldData = integration.NewFloatVectorFieldData(vecFieldName, NB, dim)
} else {
vecFieldData = integration.NewBinaryVectorFieldData(vecFieldName, NB, dim)
}
fieldsData = append(fieldsData, vecFieldData)
hashKeys := integration.GenerateHashKeys(NB)
_, err = s.Cluster.Proxy.Insert(ctx, &milvuspb.InsertRequest{
CollectionName: collection,
FieldsData: fieldsData,
HashKeys: hashKeys,
NumRows: uint32(NB),
})
s.Require().NoError(err)
s.Require().Equal(createCollectionStatus.GetErrorCode(), commonpb.ErrorCode_Success)
// flush
flushResp, err := s.Cluster.Proxy.Flush(ctx, &milvuspb.FlushRequest{
CollectionNames: []string{collection},
})
s.Require().NoError(err)
segmentIDs, has := flushResp.GetCollSegIDs()[collection]
ids := segmentIDs.GetData()
s.Require().NotEmpty(segmentIDs)
s.Require().True(has)
segments, err := s.Cluster.MetaWatcher.ShowSegments()
s.Require().NoError(err)
s.Require().NotEmpty(segments)
s.WaitForFlush(ctx, ids)
// create index
_, err = s.Cluster.Proxy.CreateIndex(ctx, &milvuspb.CreateIndexRequest{
CollectionName: collection,
FieldName: vecFieldName,
IndexName: "_default",
ExtraParams: integration.ConstructIndexParam(dim, s.indexType, s.metricType),
})
s.Require().NoError(err)
s.Require().Equal(createCollectionStatus.GetErrorCode(), commonpb.ErrorCode_Success)
s.WaitForIndexBuilt(ctx, collection, vecFieldName)
// load
_, err = s.Cluster.Proxy.LoadCollection(ctx, &milvuspb.LoadCollectionRequest{
CollectionName: collection,
})
s.Require().NoError(err)
s.Require().Equal(createCollectionStatus.GetErrorCode(), commonpb.ErrorCode_Success)
s.WaitForLoad(ctx, collection)
// search
nq := s.nq
topk := s.topK
outputFields := []string{vecFieldName}
params := integration.GetSearchParams(s.indexType, s.metricType)
searchReq := integration.ConstructSearchRequest("", collection, "",
vecFieldName, s.vecType, outputFields, s.metricType, params, nq, dim, topk, -1)
searchResp, err := s.Cluster.Proxy.Search(ctx, searchReq)
s.Require().NoError(err)
if s.searchFailed {
s.Require().NotEqual(searchResp.GetStatus().GetErrorCode(), commonpb.ErrorCode_Success)
s.T().Logf("reason:%s", searchResp.GetStatus().GetReason())
return
}
s.Require().Equal(searchResp.GetStatus().GetErrorCode(), commonpb.ErrorCode_Success)
result := searchResp.GetResults()
if s.pkType == schemapb.DataType_Int64 {
s.Require().Len(result.GetIds().GetIntId().GetData(), nq*topk)
} else {
s.Require().Len(result.GetIds().GetStrId().GetData(), nq*topk)
}
s.Require().Len(result.GetScores(), nq*topk)
s.Require().GreaterOrEqual(len(result.GetFieldsData()), 1)
var vecFieldIndex = -1
for i, fieldData := range result.GetFieldsData() {
if typeutil.IsVectorType(fieldData.GetType()) {
vecFieldIndex = i
break
}
}
s.Require().EqualValues(nq, result.GetNumQueries())
s.Require().EqualValues(topk, result.GetTopK())
// check output vectors
if s.vecType == schemapb.DataType_FloatVector {
s.Require().Len(result.GetFieldsData()[vecFieldIndex].GetVectors().GetFloatVector().GetData(), nq*topk*dim)
rawData := vecFieldData.GetVectors().GetFloatVector().GetData()
resData := result.GetFieldsData()[vecFieldIndex].GetVectors().GetFloatVector().GetData()
if s.pkType == schemapb.DataType_Int64 {
for i, id := range result.GetIds().GetIntId().GetData() {
expect := rawData[int(id)*dim : (int(id)+1)*dim]
actual := resData[i*dim : (i+1)*dim]
s.Require().ElementsMatch(expect, actual)
}
} else {
for i, idStr := range result.GetIds().GetStrId().GetData() {
id, err := strconv.Atoi(idStr)
s.Require().NoError(err)
expect := rawData[id*dim : (id+1)*dim]
actual := resData[i*dim : (i+1)*dim]
s.Require().ElementsMatch(expect, actual)
}
}
} else {
s.Require().Len(result.GetFieldsData()[vecFieldIndex].GetVectors().GetBinaryVector(), nq*topk*dim/8)
rawData := vecFieldData.GetVectors().GetBinaryVector()
resData := result.GetFieldsData()[vecFieldIndex].GetVectors().GetBinaryVector()
if s.pkType == schemapb.DataType_Int64 {
for i, id := range result.GetIds().GetIntId().GetData() {
dataBytes := dim / 8
for j := 0; j < dataBytes; j++ {
expect := rawData[int(id)*dataBytes+j]
actual := resData[i*dataBytes+j]
s.Require().Equal(expect, actual)
}
}
} else {
for i, idStr := range result.GetIds().GetStrId().GetData() {
dataBytes := dim / 8
id, err := strconv.Atoi(idStr)
s.Require().NoError(err)
for j := 0; j < dataBytes; j++ {
expect := rawData[id*dataBytes+j]
actual := resData[i*dataBytes+j]
s.Require().Equal(expect, actual)
}
}
}
}
status, err := s.Cluster.Proxy.DropCollection(ctx, &milvuspb.DropCollectionRequest{
CollectionName: collection,
})
s.Require().NoError(err)
s.Require().Equal(status.GetErrorCode(), commonpb.ErrorCode_Success)
}
func (s *TestGetVectorSuite) TestGetVector_FLAT() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexFaissIDMap
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_IVF_FLAT() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexFaissIvfFlat
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_IVF_PQ() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexFaissIvfPQ
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = true
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_IVF_SQ8() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexFaissIvfSQ8
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = true
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_HNSW() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexHNSW
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_IP() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexHNSW
s.metricType = distance.IP
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_StringPK() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexHNSW
s.metricType = distance.L2
s.pkType = schemapb.DataType_VarChar
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_BinaryVector() {
s.nq = 10
s.topK = 10
s.indexType = integration.IndexFaissBinIvfFlat
s.metricType = distance.JACCARD
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_BinaryVector
s.searchFailed = false
s.run()
}
func (s *TestGetVectorSuite) TestGetVector_Big_NQ_TOPK() {
s.T().Skip("skip big NQ Top due to timeout")
s.nq = 10000
s.topK = 200
s.indexType = integration.IndexHNSW
s.metricType = distance.L2
s.pkType = schemapb.DataType_Int64
s.vecType = schemapb.DataType_FloatVector
s.searchFailed = false
s.run()
}
//func (s *TestGetVectorSuite) TestGetVector_DISKANN() {
// s.nq = 10
// s.topK = 10
// s.indexType = integration.IndexDISKANN
// s.metricType = distance.L2
// s.pkType = schemapb.DataType_Int64
// s.vecType = schemapb.DataType_FloatVector
// s.searchFailed = false
// s.run()
//}
func TestGetVector(t *testing.T) {
suite.Run(t, new(TestGetVectorSuite))
}