Files
wehub-resource-sync 498b235461
Build and test / Build and test AMD64 Ubuntu 22.04 (push) Failing after 0s
Publish Builder / amazonlinux2023 (push) Failing after 1s
Build and test / UT for Go (push) Has been skipped
Publish KRTE Images / KRTE (push) Failing after 1s
Build and test / Integration Test (push) Has been skipped
Build and test / Upload Code Coverage (push) Has been skipped
Publish Builder / rockylinux9 (push) Failing after 1s
Publish Builder / ubuntu22.04 (push) Failing after 0s
Publish Builder / ubuntu24.04 (push) Failing after 0s
Publish Gpu Builder / publish-gpu-builder (push) Failing after 1s
Publish Test Images / PyTest (push) Failing after 0s
Build and test / UT for Cpp (push) Has been cancelled
chore: import upstream snapshot with attribution
2026-07-13 12:31:17 +08:00

687 lines
22 KiB
Go

package proxy
import (
"context"
"encoding/json"
"strconv"
"strings"
"github.com/samber/lo"
"github.com/tidwall/gjson"
"go.opentelemetry.io/otel/trace"
"github.com/milvus-io/milvus-proto/go-api/v3/commonpb"
"github.com/milvus-io/milvus-proto/go-api/v3/milvuspb"
"github.com/milvus-io/milvus-proto/go-api/v3/schemapb"
"github.com/milvus-io/milvus/internal/proxy/shardclient"
"github.com/milvus-io/milvus/internal/util/function/highlight"
"github.com/milvus-io/milvus/internal/util/function/models"
"github.com/milvus-io/milvus/pkg/v3/common"
"github.com/milvus-io/milvus/pkg/v3/proto/querypb"
"github.com/milvus-io/milvus/pkg/v3/util/funcutil"
"github.com/milvus-io/milvus/pkg/v3/util/merr"
"github.com/milvus-io/milvus/pkg/v3/util/paramtable"
"github.com/milvus-io/milvus/pkg/v3/util/typeutil"
)
const (
PreTagsKey = "pre_tags"
PostTagsKey = "post_tags"
HighlightSearchTextKey = "highlight_search_text"
HighlightQueryKey = "highlight_query"
FragmentOffsetKey = "fragment_offset"
FragmentSizeKey = "fragment_size"
FragmentNumKey = "num_of_fragments"
DefaultFragmentSize = 100
DefaultFragmentNum = 5
DefaultPreTag = "<em>"
DefaultPostTag = "</em>"
)
type Highlighter interface {
AsSearchPipelineOperator(t *searchTask) (operator, error)
FieldIDs() []int64
RequiredFieldIDs() []int64
DynamicFieldNames() []string
}
// highlight task for one field
type highlightTask struct {
*querypb.HighlightTask
preTags [][]byte
postTags [][]byte
}
type highlightQuery struct {
text string
fieldName string
highlightType querypb.HighlightQueryType
}
type LexicalHighlighter struct {
tasks map[int64]*highlightTask // fieldID -> highlightTask
extraFields []int64 // extra fields id for fetch analyzer name of multi analyzers
// option for all highlight task
// TODO: support set option for each task
preTags [][]byte
postTags [][]byte
highlightSearch bool
options *querypb.HighlightOptions
queries []*highlightQuery
}
// add highlight task with search
// must used before addTaskWithQuery
func (h *LexicalHighlighter) addTaskWithSearchText(collInfo *schemaInfo, fieldID int64, fieldName string, analyzerName string, texts []string) error {
_, ok := h.tasks[fieldID]
if ok {
return merr.WrapErrParameterInvalidMsg("not support hybrid search with highlight now. fieldID: %d", fieldID)
}
task := &highlightTask{
preTags: h.preTags,
postTags: h.postTags,
HighlightTask: &querypb.HighlightTask{
FieldName: fieldName,
FieldId: fieldID,
Options: h.options,
},
}
h.tasks[fieldID] = task
task.Texts = texts
task.SearchTextNum = int64(len(texts))
// try get multi analyzer name field id
nameFieldID, err := collInfo.GetMultiAnalyzerNameFieldID(fieldID)
if err != nil {
return err
}
// set analyzer name and extra field id for multi analyzer
if nameFieldID > 0 {
// if multi analyzer name field id is found, set analyzer name to default
if analyzerName == "" {
analyzerName = "default"
}
task.AnalyzerNames = []string{}
for i := 0; i < len(texts); i++ {
task.AnalyzerNames = append(task.AnalyzerNames, analyzerName)
}
h.extraFields = append(h.extraFields, nameFieldID)
}
return nil
}
func (h *LexicalHighlighter) addTaskWithQuery(collInfo *schemaInfo, fieldID int64, query *highlightQuery, analyzerName string) error {
task, ok := h.tasks[fieldID]
if !ok {
task = &highlightTask{
HighlightTask: &querypb.HighlightTask{
Texts: []string{},
FieldId: fieldID,
FieldName: query.fieldName,
Options: h.options,
},
preTags: h.preTags,
postTags: h.postTags,
}
h.tasks[fieldID] = task
}
task.Texts = append(task.Texts, query.text)
task.Queries = append(task.Queries, &querypb.HighlightQuery{
Type: query.highlightType,
})
nameFieldID, err := collInfo.GetMultiAnalyzerNameFieldID(fieldID)
if err != nil {
return err
}
if nameFieldID > 0 {
if analyzerName == "" {
analyzerName = "default"
}
task.AnalyzerNames = append(task.AnalyzerNames, analyzerName)
if !lo.Contains(h.extraFields, nameFieldID) {
h.extraFields = append(h.extraFields, nameFieldID)
}
}
return nil
}
func (h *LexicalHighlighter) initHighlightQueries(t *searchTask, analyzerName string) error {
// add query to highlight tasks
for _, query := range h.queries {
fieldID, ok := t.schema.MapFieldID(query.fieldName)
if !ok {
return merr.WrapErrParameterInvalidMsg("highlight field not found in schema: %s", query.fieldName)
}
if err := h.addTaskWithQuery(t.schema, fieldID, query, analyzerName); err != nil {
return err
}
}
return nil
}
func (h *LexicalHighlighter) AsSearchPipelineOperator(t *searchTask) (operator, error) {
return newLexicalHighlightOperator(t, lo.Values(h.tasks))
}
func (h *LexicalHighlighter) FieldIDs() []int64 {
return append(lo.Keys(h.tasks), h.extraFields...)
}
func (h *LexicalHighlighter) RequiredFieldIDs() []int64 {
return h.FieldIDs()
}
func (h *LexicalHighlighter) DynamicFieldNames() []string {
return []string{}
}
func NewLexicalHighlighter(highlighter *commonpb.Highlighter) (*LexicalHighlighter, error) {
params := funcutil.KeyValuePair2Map(highlighter.GetParams())
h := &LexicalHighlighter{
tasks: make(map[int64]*highlightTask),
options: &querypb.HighlightOptions{},
extraFields: make([]int64, 0),
}
// set pre_tags and post_tags
if value, ok := params[PreTagsKey]; ok {
tags := []string{}
if err := json.Unmarshal([]byte(value), &tags); err != nil {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal pre_tags as string array failed: %v", err)
}
if len(tags) == 0 {
return nil, merr.WrapErrParameterInvalidMsg("pre_tags cannot be empty list")
}
h.preTags = make([][]byte, len(tags))
for i, tag := range tags {
h.preTags[i] = []byte(tag)
}
} else {
h.preTags = [][]byte{[]byte(DefaultPreTag)}
}
if value, ok := params[PostTagsKey]; ok {
tags := []string{}
if err := json.Unmarshal([]byte(value), &tags); err != nil {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal post_tags as string list failed: %v", err)
}
if len(tags) == 0 {
return nil, merr.WrapErrParameterInvalidMsg("post_tags cannot be empty list")
}
h.postTags = make([][]byte, len(tags))
for i, tag := range tags {
h.postTags[i] = []byte(tag)
}
} else {
h.postTags = [][]byte{[]byte(DefaultPostTag)}
}
// set fragment config
if value, ok := params[FragmentSizeKey]; ok {
fragmentSize, err := strconv.ParseInt(value, 10, 64)
if err != nil || fragmentSize <= 0 {
return nil, merr.WrapErrParameterInvalidMsg("invalid fragment_size: %s", value)
}
h.options.FragmentSize = fragmentSize
} else {
h.options.FragmentSize = DefaultFragmentSize
}
if value, ok := params[FragmentNumKey]; ok {
fragmentNum, err := strconv.ParseInt(value, 10, 64)
if err != nil || fragmentNum < 0 {
return nil, merr.WrapErrParameterInvalidMsg("invalid num_of_fragments: %s", value)
}
h.options.NumOfFragments = fragmentNum
} else {
h.options.NumOfFragments = DefaultFragmentNum
}
if value, ok := params[FragmentOffsetKey]; ok {
fragmentOffset, err := strconv.ParseInt(value, 10, 64)
if err != nil || fragmentOffset < 0 {
return nil, merr.WrapErrParameterInvalidMsg("invalid fragment_offset: %s", value)
}
h.options.FragmentOffset = fragmentOffset
}
if value, ok := params[HighlightSearchTextKey]; ok {
enable, err := strconv.ParseBool(value)
if err != nil {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight_search_text as bool failed: %v", err)
}
h.highlightSearch = enable
}
if value, ok := params[HighlightQueryKey]; ok {
queries := []any{}
if err := json.Unmarshal([]byte(value), &queries); err != nil {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries as json array failed: %v", err)
}
for _, query := range queries {
m, ok := query.(map[string]any)
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: item in array is not json object")
}
text, ok := m["text"]
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: must set `text` in query")
}
textStr, ok := text.(string)
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: `text` must be string")
}
t, ok := m["type"]
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: must set `type` in query")
}
typeStr, ok := t.(string)
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: `type` must be string")
}
typeEnum, ok := querypb.HighlightQueryType_value[typeStr]
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: invalid highlight query type: %s", typeStr)
}
f, ok := m["field"]
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: must set `field` in query")
}
fieldStr, ok := f.(string)
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("unmarshal highlight queries failed: `field` must be string")
}
h.queries = append(h.queries, &highlightQuery{
text: textStr,
highlightType: querypb.HighlightQueryType(typeEnum),
fieldName: fieldStr,
})
}
}
return h, nil
}
type lexicalHighlightOperator struct {
tasks []*highlightTask
schema *schemaInfo
lbPolicy shardclient.LBPolicy
scheduler *taskScheduler
collectionName string
collectionID int64
dbName string
}
func newLexicalHighlightOperator(t *searchTask, tasks []*highlightTask) (operator, error) {
return &lexicalHighlightOperator{
tasks: tasks,
lbPolicy: t.lb,
scheduler: t.node.(*Proxy).sched,
schema: t.schema,
collectionName: t.request.CollectionName,
collectionID: t.CollectionID,
dbName: t.request.DbName,
}, nil
}
func hasSearchResultHits(result *schemapb.SearchResultData) bool {
return searchResultHitCount(result) > 0
}
func searchResultHitCount(result *schemapb.SearchResultData) int {
if result == nil {
return 0
}
rowNum := 0
for _, topk := range result.GetTopks() {
rowNum += int(topk)
}
return rowNum
}
func rowAlignedStringFieldData(fieldData *schemapb.FieldData, rowNum int) ([]string, error) {
if fieldData == nil {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, field data is nil")
}
data := fieldData.GetScalars().GetStringData().GetData()
validData := fieldData.GetValidData()
if len(validData) == 0 {
if len(data) != rowNum {
return nil, merr.WrapErrServiceInternalMsg(
"get highlight failed, string field %s:%d has %d rows, expected %d",
fieldData.GetFieldName(), fieldData.GetFieldId(), len(data), rowNum)
}
return append([]string(nil), data...), nil
}
if len(validData) != rowNum {
return nil, merr.WrapErrServiceInternalMsg(
"get highlight failed, string field %s:%d has ValidData length %d, expected %d",
fieldData.GetFieldName(), fieldData.GetFieldId(), len(validData), rowNum)
}
validCount := lo.CountBy(validData, func(valid bool) bool { return valid })
if len(data) != rowNum && len(data) != validCount {
return nil, merr.WrapErrServiceInternalMsg(
"get highlight failed, string field %s:%d has %d compact rows, expected %d valid rows",
fieldData.GetFieldName(), fieldData.GetFieldId(), len(data), validCount)
}
iterator := typeutil.GetDataIterator(fieldData)
texts := make([]string, rowNum)
for i := range texts {
if value := iterator(i); value != nil {
text, ok := value.(string)
if !ok {
return nil, merr.WrapErrServiceInternalMsg(
"get highlight failed, string field %s:%d has non-string data at row %d",
fieldData.GetFieldName(), fieldData.GetFieldId(), i)
}
texts[i] = text
}
}
return texts, nil
}
func (op *lexicalHighlightOperator) run(ctx context.Context, span trace.Span, inputs ...any) ([]any, error) {
result := inputs[0].(*milvuspb.SearchResults)
resultData := result.GetResults()
// skip highlight if result has no hits. FieldsData may contain empty field templates
// even when Topks/Ids/Scores show zero matched rows.
if !hasSearchResultHits(resultData) {
return []any{result}, nil
}
datas := resultData.GetFieldsData()
if len(datas) == 0 {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, field data is empty for non-empty search result")
}
rowNum := searchResultHitCount(resultData)
req := &querypb.GetHighlightRequest{
Topks: result.GetResults().GetTopks(),
Tasks: lo.Map(op.tasks, func(task *highlightTask, _ int) *querypb.HighlightTask { return task.HighlightTask }),
}
for _, task := range req.GetTasks() {
textFieldDatas, ok := lo.Find(datas, func(data *schemapb.FieldData) bool { return data.FieldId == task.GetFieldId() })
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("get highlight failed, text field not in output field %s: %d", task.GetFieldName(), task.GetFieldId())
}
texts, err := rowAlignedStringFieldData(textFieldDatas, rowNum)
if err != nil {
return nil, err
}
task.Texts = append(task.Texts, texts...)
task.CorpusTextNum = int64(len(texts))
nameFieldID, err := op.schema.GetMultiAnalyzerNameFieldID(task.GetFieldId())
if err != nil {
return nil, err
}
// if use multi analyzer
// get analyzer field data
if nameFieldID > 0 {
analyzerFieldDatas, ok := lo.Find(datas, func(data *schemapb.FieldData) bool { return data.FieldId == nameFieldID })
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("get highlight failed, analyzer name field: %d for multi analyzer not in output field", nameFieldID)
}
analyzerNames, err := rowAlignedStringFieldData(analyzerFieldDatas, rowNum)
if err != nil {
return nil, err
}
for i, name := range analyzerNames {
if name == "" {
analyzerNames[i] = "default"
}
}
task.AnalyzerNames = append(task.AnalyzerNames, analyzerNames...)
}
}
task := &HighlightTask{
ctx: ctx,
lb: op.lbPolicy,
Condition: NewTaskCondition(ctx),
GetHighlightRequest: req,
collectionName: op.collectionName,
collectionID: op.collectionID,
dbName: op.dbName,
}
if err := op.scheduler.dqQueue.Enqueue(task); err != nil {
return nil, err
}
if err := task.WaitToFinish(); err != nil {
return nil, err
}
HighlightResults := []*commonpb.HighlightResult{}
if rowNum != 0 {
if len(task.result.GetResults()) != len(req.GetTasks())*rowNum {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, result count %d does not match task count %d and row count %d", len(task.result.GetResults()), len(req.GetTasks()), rowNum)
}
rowDatas := lo.Map(task.result.Results, func(result *querypb.HighlightResult, i int) *commonpb.HighlightData {
return buildStringFragments(op.tasks[i/rowNum], i%rowNum, result.GetFragments())
})
for i, task := range req.GetTasks() {
HighlightResults = append(HighlightResults, &commonpb.HighlightResult{
FieldName: task.GetFieldName(),
Datas: rowDatas[i*rowNum : (i+1)*rowNum],
})
}
}
result.Results.HighlightResults = HighlightResults
return []any{result}, nil
}
func buildStringFragments(task *highlightTask, idx int, frags []*querypb.HighlightFragment) *commonpb.HighlightData {
startOffset := int(task.GetSearchTextNum()) + len(task.Queries)
text := []rune(task.Texts[startOffset+idx])
preTagsNum := len(task.preTags)
postTagsNum := len(task.postTags)
result := &commonpb.HighlightData{Fragments: make([]string, 0)}
for _, frag := range frags {
var fragString strings.Builder
cursor := int(frag.GetStartOffset())
for i := 0; i < len(frag.GetOffsets())/2; i++ {
startOffset := int(frag.Offsets[i<<1])
endOffset := int(frag.Offsets[(i<<1)+1])
if cursor < startOffset {
fragString.WriteString(string(text[cursor:startOffset]))
}
fragString.WriteString(string(task.preTags[i%preTagsNum]))
fragString.WriteString(string(text[startOffset:endOffset]))
fragString.WriteString(string(task.postTags[i%postTagsNum]))
cursor = endOffset
}
if cursor < int(frag.GetEndOffset()) {
fragString.WriteString(string(text[cursor:frag.GetEndOffset()]))
}
result.Fragments = append(result.Fragments, fragString.String())
}
return result
}
type SemanticHighlighter struct {
highlight *highlight.SemanticHighlight
}
func newSemanticHighlighter(t *searchTask, extraInfo *models.ModelExtraInfo) (*SemanticHighlighter, error) {
conf := paramtable.Get().FunctionCfg.ZillizProviders.GetValue()
highlight, err := highlight.NewSemanticHighlight(t.schema.CollectionSchema, t.request.GetHighlighter().GetParams(), conf, extraInfo)
if err != nil {
return nil, err
}
return &SemanticHighlighter{highlight: highlight}, nil
}
func (h *SemanticHighlighter) FieldIDs() []int64 {
return h.highlight.FieldIDs()
}
func (h *SemanticHighlighter) RequiredFieldIDs() []int64 {
return h.highlight.RequiredFieldIDs()
}
func (h *SemanticHighlighter) DynamicFieldNames() []string {
return h.highlight.DynamicFieldNames()
}
func (h *SemanticHighlighter) AsSearchPipelineOperator(t *searchTask) (operator, error) {
return &semanticHighlightOperator{highlight: h.highlight}, nil
}
type semanticHighlightOperator struct {
highlight *highlight.SemanticHighlight
}
func (op *semanticHighlightOperator) run(ctx context.Context, span trace.Span, inputs ...any) ([]any, error) {
result := inputs[0].(*milvuspb.SearchResults)
resultData := result.GetResults()
if !hasSearchResultHits(resultData) {
return []any{result}, nil
}
datas := resultData.GetFieldsData()
if len(datas) == 0 {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, field data is empty for non-empty search result")
}
highlightResults := []*commonpb.HighlightResult{}
topks := result.Results.GetTopks()
// Process schema fields
for _, fieldID := range op.highlight.FieldIDs() {
fieldDatas, ok := lo.Find(datas, func(data *schemapb.FieldData) bool { return data.FieldId == fieldID })
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("get highlight failed, text field not in output field %d", fieldID)
}
texts := fieldDatas.GetScalars().GetStringData().GetData()
fieldName := op.highlight.GetFieldName(fieldID)
highlightResult, err := op.processFieldHighlight(ctx, fieldName, texts, topks)
if err != nil {
return nil, err
}
highlightResults = append(highlightResults, highlightResult)
}
// Process dynamic fields
if op.highlight.HasDynamicFields() {
// Find $meta field data
dynamicFieldID := op.highlight.DynamicFieldID()
metaFieldData, ok := lo.Find(datas, func(data *schemapb.FieldData) bool {
return data.FieldId == dynamicFieldID || data.FieldName == common.MetaFieldName
})
if !ok {
return nil, merr.WrapErrParameterInvalidMsg("get highlight failed, dynamic field ($meta) not in output field")
}
// Safely get JSON data with nil checks
scalars := metaFieldData.GetScalars()
if scalars == nil {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, dynamic field ($meta) has no scalar data")
}
jsonData := scalars.GetJsonData()
if jsonData == nil {
return nil, merr.WrapErrServiceInternalMsg("get highlight failed, dynamic field ($meta) has no JSON data")
}
jsonDataBytes := jsonData.GetData()
dynFieldNames := op.highlight.DynamicFieldNames()
// Extract all dynamic fields in one pass to avoid parsing JSON multiple times
allDynFieldTexts, err := extractMultipleDynamicFieldTexts(jsonDataBytes, dynFieldNames)
if err != nil {
return nil, err
}
for _, dynFieldName := range dynFieldNames {
texts := allDynFieldTexts[dynFieldName]
highlightResult, err := op.processFieldHighlight(ctx, dynFieldName, texts, topks)
if err != nil {
return nil, err
}
highlightResults = append(highlightResults, highlightResult)
}
}
result.Results.HighlightResults = highlightResults
return []any{result}, nil
}
// processFieldHighlight processes highlight for a single field and returns the result
func (op *semanticHighlightOperator) processFieldHighlight(ctx context.Context, fieldName string, texts []string, topks []int64) (*commonpb.HighlightResult, error) {
highlights, scores, err := op.highlight.Process(ctx, topks, texts)
if err != nil {
return nil, err
}
if len(highlights) != len(scores) {
return nil, merr.WrapErrServiceInternalMsg("Highlights size must equal to scores size, but got highlights size [%d], scores size [%d]", len(highlights), len(scores))
}
result := &commonpb.HighlightResult{
FieldName: fieldName,
Datas: make([]*commonpb.HighlightData, len(highlights)),
}
for i := range highlights {
result.Datas[i] = &commonpb.HighlightData{Fragments: highlights[i], Scores: scores[i]}
}
return result, nil
}
// extractMultipleDynamicFieldTexts extracts text values from JSON data for multiple field names in one pass
// This avoids parsing JSON multiple times when there are multiple dynamic fields
func extractMultipleDynamicFieldTexts(jsonDataBytes [][]byte, fieldNames []string) (map[string][]string, error) {
result := make(map[string][]string, len(fieldNames))
for _, fieldName := range fieldNames {
result[fieldName] = make([]string, len(jsonDataBytes))
}
for i, jsonBytes := range jsonDataBytes {
if len(jsonBytes) == 0 {
// JSON is empty, all fields use empty string (graceful degradation)
continue
}
if !gjson.ValidBytes(jsonBytes) {
return nil, merr.WrapErrDataIntegrityMsg("failed to unmarshal JSON at index %d: invalid JSON", i)
}
for _, fieldName := range fieldNames {
r := gjson.GetBytes(jsonBytes, fieldName)
if !r.Exists() {
// field missing, use empty string (graceful degradation)
continue
}
if r.Type != gjson.String {
return nil, merr.WrapErrParameterInvalidMsg("dynamic field %s is not a string type, got %s", fieldName, r.Type.String())
}
result[fieldName][i] = r.String()
}
}
return result, nil
}