knowledgeBase/rpc/knowledge/util/elasticsearch_dsl.go

package util

import (
	"fmt"
	. "knowledgeBase/rpc/knowledge/init"
	"log"
	"strconv"
	"strings"
)

var (
	Analyze string
	Segment string
)

/*项目中所用到的几类查询dsl语句构建工具类*/

// DSL4SearchByKwsAndTags 管理后台列表查询
func DSL4SearchByKwsAndTags(kws string, tags ...string) string {
	var (
		sql        = `{"query": {"bool": {"must": [%s,%s]}}}`
		queryMatch = `{"match":{"knowledgeKeyWords":{"query":"%s","minimum_should_match":"%s"}}}`
		tagsTerms  = `{"terms":{"tags.code":[%s]}}`
		tag        = `"%s"`
	)
	if kws != "" {

	}
	queryMatch = fmt.Sprintf(queryMatch, kws, "20%")
	tmp := ""
	for i, val := range tags {
		if i < len(tags)-1 {
			tmp += fmt.Sprintf(tag, val) + ","
		} else {
			tmp += fmt.Sprintf(tag, val)
		}
	}
	tagsTerms = fmt.Sprintf(tagsTerms, tmp)
	sql = fmt.Sprintf(sql, queryMatch, tagsTerms)
	return sql
}

func DSL4SmartResponse(question string, tenantId int64, msgType int) string {
	var (
		totalQuery = `{"post_filter":{%s},"query":{%s},"_source":[%s],"size":%d}`
		postFilter = `"script":{"script":"def sk=_source.must_keywords;def n=0;for(item in sk){ n++;if(que.indexOf(item)>-1){return true}};if(n==0){ return true}","params":{"que":"%s"}}`
		query      = `"bool":{"must_not":[%s],"must":[{"match":{"%s":{"query":"%s","minimum_should_match":"%s"}}},{"terms":{"tenantId":"%s"}}]}`
	)
	var typeStr string
	/*1.首先将问题使用hanlp分词*/
	//hanlpCutWords := HanlpGetNormalWords(question, "http://39.106.145.77:8080/api/segment")
	hanlpCutWords := HanlpGetNormalWords(question, C.Segment)
	question = strings.Join(hanlpCutWords, "")
	lenQuestion := len([]rune(question))
	if lenQuestion >= 2 {
		queryPercent := "40%"
		if lenQuestion < 5 {
			queryPercent = "85%"
		} else if lenQuestion < 9 {
			queryPercent = "60%"
		} else if lenQuestion < 12 {
			queryPercent = "55%"
		}
		log.Println("消息类型：", msgType)
		if msgType == 1 {
			typeStr = "keywords"
		} else if msgType == 2 { //百度语音过来的
			typeStr = "keywords.key_pinyin"
		}
		/*2使用sik分词将问题分词以获取更多查询词语*/
		//mustque := ElasticSmartIK(question, "http://39.106.145.77:9201/smart/_analyze")
		mustque := ElasticSmartIK(question, C.Es.Addr+"/"+C.Es.Index+"/_analyze")
		if mustque != "" {
			postFilter = fmt.Sprintf(postFilter, mustque)
		}
		query = fmt.Sprintf(query, "", typeStr, question, queryPercent, strconv.Itoa(int(tenantId)))
		queryDSL := fmt.Sprintf(totalQuery, postFilter, query, `"answer","question"`, 1)
		log.Println("queryDSL:", queryDSL)
		return queryDSL
	}
	return ""
}

func DSL4SmartResponseList(question, msgType string, repositoryId []string, must_not string, size int, tags ...string) string {
	queryPercent := "20%"
	var (
		totalQuery = `{"post_filter":{%s},"query":{%s},"_source":["_id"%s],"size":%d}`
		postFilter = `"script":{"script":"def sk=_source.must_keywords;def n=0;for(item in sk){ n++;if(que.indexOf(item)>-1){return true}};if(n==0){ return true}","params":{"que":"%s"}}`
		query      = `"bool":{"must_not":[%s],"must":[{"match":{"%s":{"query":"%s","minimum_should_match":"%s"}}},{"terms":{"repositoryId":%s}},{"terms":{"tags.code":[%s]}}]}`
	)
	/*2使用sik分词将问题分词以获取更多查询词语*/
	//mustque := ElasticSmartIK(question, "http://39.106.145.77:9201/smart/_analyze")
	mustque := ElasticSmartIK(question, Analyze)
	if mustque != "" {
		postFilter = fmt.Sprintf(postFilter, mustque)
	}
	tmp := ""
	for i, val := range tags {
		if i < len(tags)-1 {
			tmp += fmt.Sprintf(`"%s"`, val) + ","
		} else {
			tmp += fmt.Sprintf(`"%s"`, val)
		}
	}
	query = fmt.Sprintf(query, must_not, msgType, question, queryPercent, repositoryId, tmp)
	queryDSL := fmt.Sprintf(totalQuery, postFilter, query, `,"answer","questions.question"`, size)
	log.Println("queryDSL:", queryDSL)
	return queryDSL
}

func DSL4SearchByKwsOrid(keyWords string, tenantId string) string {

	var (
		sql        = `{"query": {"bool": {"must": [%s%s]}}}`
		queryMatch = `{"match":{"knowledgeKeyWords":{"query":"%s","minimum_should_match":"%s"}}}`
		//idTerms    = `,{"terms":{"smart.id":[%s]}}`
		ridTerms = `,{"terms":{"tenantId":%s}}`
		//id       = `"%s"`
	)
	queryMatch = fmt.Sprintf(queryMatch, keyWords, "20%")
	//fmt.Println("queryMatch:", queryMatch)
	//tmp := ""
	ridTerms = fmt.Sprintf(ridTerms, tenantId)
	sql = fmt.Sprintf(sql, queryMatch, ridTerms)
	log.Println("sql", sql)
	return sql
}

func GetQueryOT(tags, question, keywords, repositoryId string) (qstr string) {
	var query = `{"query":{"bool":{"must":[%s%s%s%s]}}}`
	queryMatch := ``
	queryTerms := ``
	queryId := ``
	queryQues := ``
	if keywords != "" {
		queryMatch = `{"match":{"knowledgeKeyWords":{"query":"` + keywords + `","minimum_should_match":"40%"}}},`
		//query_match = `{"match":{"questions.question":{"query":"` + keywords + `","minimum_should_match":"20%"}}},`
	}
	tags = strings.Replace(tags, ` `, `","`, -1)
	if tags != "" {
		queryTerms = `{"terms":{"tags.code":["` + tags + `"]}},`
	}
	if repositoryId != "" {
		queryId = `{"terms":{"repositoryId":[` + repositoryId + `]}}`
	}
	if question != "" {
		queryQues = `,{"match": {"smart.questions.question": {"query": "` + question + `","fuzziness": "AUTO","operator": "and"}}}`
	}
	qstr = fmt.Sprintf(query, queryMatch, queryTerms, queryId, queryQues)
	return qstr
}

/*func GetFindQuery(keywords, tags, repositoryId string) string {
	var query = `{"query":{"bool":{"must":[%s%s%s]}}}`
	query_match := ``
	query_terms := ``
	query_id := ``
	if keywords != "" {
		query_match = `{"match":{"knowledgeKeyWords":{"query":"` + keywords + `","minimum_should_match":"40%"}}},`
	}
	tags = strings.Replace(tags, ` `, `","`, -1)
	if tags != "" {
		query_terms = `{"terms":{"tags.code":["` + tags + `"]}},`
	}
	if repositoryId != "" {
		query_id = `{"terms":{"repositoryId":[` + repositoryId + `]}}`
	}
	qstr := fmt.Sprintf(query, query_match, query_terms, query_id)
	return qstr
}*/