/*****************************************************************************
* function name : SearchFieldByString
* params : query field
* return :
*
* description : 在指定字段检索相应的query，用于正常检索
*
******************************************************************************/
func (this *IndexSet) SearchFieldByString(query string, field string) ([]utils.DocIdInfo, bool) {
var terms []string
if this.FieldInfo[field].SType == 9 {
terms = this.IvtIndex[field].GetCustomInterface().SegmentFunc(query, true)
} else {
terms = this.Segmenter.SegmentByType(query, this.FieldInfo[field].SType, false)
}
//fmt.Printf("terms : %v \n",terms)
/*
var terms []string
switch this.FieldInfo[field].SType {
case 1: //正常切词
terms = utils.RemoveDuplicatesAndEmpty(this.Segmenter.Segment(query, false))
case 2: //按单个字符进行切词
terms = utils.RemoveDuplicatesAndEmpty(strings.Split(query, ""))
case 3: //按规定的分隔符进行切词
terms = utils.RemoveDuplicatesAndEmpty(strings.Split(query, ","))
}
*/
//按照最大切分进行切词
//terms := utils.RemoveDuplicatesAndEmpty(this.Segmenter.Segment(query, false))
//this.Logger.Info("TERMS :: %v ", terms)
//首先按照字段检索
//交集结果
var res_list []utils.DocIdInfo
_, ok := this.IvtIndex[field]
if !ok {
return nil, false
}
isFound := true
for index, term := range terms {
l, ok := this.IvtIndex[field].Find(term)
if !ok {
isFound = false
break
}
//this.Logger.Info("[Term : %v ] [Field: %v ] DocIDs : %v", term, field, l)
//求交集
if index == 0 {
res_list = l
} else {
res_list, ok = utils.Interaction(l, res_list)
if !ok {
isFound = false
break
}
}
}
if len(res_list) > 0 && isFound == true {
return res_list, true
}
return nil, false
}
func (this *invert) query(key interface{}) ([]utils.DocIdNode, bool) {
//this.Logger.Info("[DEBUG] invert Query %v", key)
keystr, ok := key.(string)
if !ok {
return nil, false
}
//全词匹配模式
if this.fieldType == utils.IDX_TYPE_STRING || this.fieldType == utils.GATHER_TYPE {
return this.queryTerm(keystr)
}
var queryterms []string
switch this.fieldType {
case utils.IDX_TYPE_STRING_LIST: //分号切割模式
queryterms = strings.Split(keystr, ";")
case utils.IDX_TYPE_STRING_SINGLE: //单字模式
queryterms = utils.GSegmenter.SegmentSingle(keystr)
case utils.IDX_TYPE_STRING_SEG: //分词模式
queryterms = utils.GSegmenter.Segment(keystr, false)
default:
return this.queryTerm(keystr)
}
if len(queryterms) == 1 {
return this.queryTerm(queryterms[0])
}
var fDocids []utils.DocIdNode
// var sDocids []utils.DocIdNode
var hasRes bool
var match bool
fDocids, match = this.queryTerm(queryterms[0])
//fDocids=append(fDocids,sDocids...)
if match {
for _, term := range queryterms[1:] {
subDocids, ok := this.queryTerm(term)
if !ok {
return nil, false
}
fDocids, hasRes = utils.Interaction(fDocids, subDocids)
if !hasRes {
return nil, false
}
}
}
if len(fDocids) == 0 {
return nil, false
}
return fDocids, true
}
func (this *IndexSet) SearchByRules(rules /*map[string]interface{}*/ []SearchRule) ([]utils.DocIdInfo, bool) {
functime := utils.InitTime()
fmt.Printf("SearchByRules: %v \n", functime("Start"))
var res []utils.DocIdInfo
for index, rule := range rules {
var sub_res []utils.DocIdInfo
var ok bool
if rule.Field == "query" {
sub_res, ok = this.Search(rule.Query)
} else {
//this.Logger.Info(" Field : %v Query : %v",rule.Field, rule.Query)
fmt.Printf("SearchByRules: %v \n", functime("Start SearchField"))
sub_res, ok = this.SearchField(rule.Query, rule.Field)
fmt.Printf("SearchByRules: %v \n", functime("End SearchField"))
}
if !ok {
return nil, false
}
if index == 0 {
res = sub_res
} else {
fmt.Printf("SearchByRules: %v \n", functime("Start Interaction"))
res, ok = utils.Interaction(res, sub_res)
fmt.Printf("SearchByRules: %v \n", functime("End Interaction"))
if !ok {
return nil, false
}
}
//this.Logger.Info(" RES :: %v ", res)
}
//BitMap过滤失效的doc_id
//this.Logger.Info(" %v ",res)
fmt.Printf("SearchByRules: %v \n", functime("Start Bitmap"))
r := make([]utils.DocIdInfo, len(res))
r_index := 0
for i, _ := range res {
//this.Logger.Info(" %v ",res[i].DocId)
if this.BitMap.GetBit(uint64(res[i].DocId)) == 0 {
r[r_index] = res[i]
r_index++
//r = append(r,res[i])
}
}
fmt.Printf("SearchByRules: %v \n", functime("End Bitmap"))
//TODO 自定义过滤
fmt.Printf("SearchByRules: %v \n", functime("End SearchByRules"))
return r[:r_index], true
}
/*****************************************************************************
* function name : SearchString
* params : query
* return : doc链
*
* description : 检索函数，根据关键字进行检索，先在同一个字段中检索，如果没有结果进行跨字段检索
*
******************************************************************************/
func (this *IndexSet) SearchString(query string) ([]utils.DocIdInfo, bool) {
//按照最大切分进行切词
//terms := utils.RemoveDuplicatesAndEmpty(this.Segmenter.Segment(query,false))
//首先按照字段检索
var res_list []utils.DocIdInfo
var ok bool
for key, _ := range this.IvtIndex {
if this.IvtIndex[key].GetType() != 1 {
continue
}
res_list, ok = this.SearchFieldByString(query, key)
if ok {
return res_list, true
}
}
//如果数量不够，跨字段检索 TODO
var res_merge []utils.DocIdInfo
if len(res_list) == 0 {
terms := utils.RemoveDuplicatesAndEmpty(this.Segmenter.Segment(query, false))
//this.Logger.Info("OUT TERMS :: %v ",terms)
for index, term := range terms {
l, ok := this.SearchFieldsByTerm(term)
if !ok {
return nil, false
}
if index == 0 {
res_merge = l
} else {
res_merge, ok = utils.Interaction(res_merge, l)
//this.Logger.Info("Interaction Term:%v Docids: %v",term,res_merge)
if !ok {
return nil, false
}
}
}
}
return res_merge, true
}
func (this *IndexSet) SearchByRules(rules /*map[string]interface{}*/ []SearchRule) ([]utils.DocIdInfo, bool) {
var res []utils.DocIdInfo
for index, rule := range rules {
var sub_res []utils.DocIdInfo
var ok bool
if rule.Field == "query" {
sub_res, ok = this.Search(rule.Query)
} else {
//this.Logger.Info(" Field : %v Query : %v",rule.Field, rule.Query)
sub_res, ok = this.SearchField(rule.Query, rule.Field)
}
if !ok {
return nil, false
}
if index == 0 {
res = sub_res
} else {
res, ok = utils.Interaction(res, sub_res)
if !ok {
return nil, false
}
}
//this.Logger.Info(" RES :: %v ", res)
}
//BitMap过滤失效的doc_id
this.Logger.Info(" %v ", res)
r := make([]utils.DocIdInfo, 0)
for i, _ := range res {
//this.Logger.Info(" %v ",res[i].DocId)
if this.BitMap.GetBit(uint64(res[i].DocId)) == 0 {
r = append(r, res[i])
}
}
//TODO 自定义过滤
return r, true
}
func (this *IndexSet) SearchByRules(rules /*map[string]interface{}*/ []SearchRule) ([]utils.DocIdInfo, bool) {
//functime := utils.InitTime()
//fmt.Printf("SearchByRules: %v \n", functime("Start"))
var res []utils.DocIdInfo
for index, rule := range rules {
var sub_res []utils.DocIdInfo
var ok bool
if rule.Field == "query" {
sub_res, ok = this.Search(rule.Query)
} else {
sub_res, ok = this.SearchField(rule.Query, rule.Field)
}
if !ok {
return nil, false
}
if index == 0 {
res = sub_res
} else {
res, ok = utils.Interaction(res, sub_res)
if !ok {
return nil, false
}
}
}
//BitMap过滤失效的doc_id
r := make([]utils.DocIdInfo, len(res))
r_index := 0
for i, _ := range res {
if this.BitMap.GetBit(uint64(res[i].DocId)) == 0 {
r[r_index] = res[i]
r_index++
}
}
//TODO 自定义过滤
return r[:r_index], true
}
/*****************************************************************************
* function name : SearchFieldByString
* params : query field
* return :
*
* description : 在指定字段检索相应的query，用于正常检索
*
******************************************************************************/
func (this *IndexSet) SearchFieldByString(query string, field string) ([]utils.DocIdInfo, bool) {
//按照最大切分进行切词
terms := utils.RemoveDuplicatesAndEmpty(this.Segmenter.Segment(query, false))
//this.Logger.Info("TERMS :: %v ", terms)
//首先按照字段检索
//交集结果
var res_list []utils.DocIdInfo
_, ok := this.IvtIndex[field]
if !ok {
return nil, false
}
isFound := true
for index, term := range terms {
l, ok := this.IvtIndex[field].Find(term)
if !ok {
isFound = false
break
}
//this.Logger.Info("[Term : %v ] [Field: %v ] DocIDs : %v", term, field, l)
//求交集
if index == 0 {
res_list = l
} else {
res_list, ok = utils.Interaction(l, res_list)
if !ok {
isFound = false
break
}
}
}
if len(res_list) > 0 && isFound == true {
return res_list, true
}
return nil, false
}