knowledgeBase/rpc/knowledge/util/elasticsearch_dsl.go

package util

import (
	"encoding/json"
	"fmt"
	"github.com/zeromicro/go-zero/core/logx"
	"strings"
)

/*项目中所用到的几类查询dsl语句构建工具类*/

func DSL4SmartResponse(question string, entId string, msgType int, addr, index, segment string) 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"}}},{"term":{"entId":"%s"}}]}`
	)
	var typeStr string
	/*1.首先将问题使用hanlp分词*/
	hanlpCutWords := HanlpGetNormalWords(question, segment)
	if len(hanlpCutWords) == 0 {
		hanlpCutWords = append(hanlpCutWords, question)
	}
	question = strings.Join(hanlpCutWords, " ")
	lenQuestion := len([]rune(question))
	//logx.Info("lenQuestion", lenQuestion)
	if lenQuestion >= 2 {
		queryPercent := "40%"
		if lenQuestion < 5 {
			queryPercent = "85%"
		} else if lenQuestion < 9 {
			queryPercent = "60%"
		} else if lenQuestion < 12 {
			queryPercent = "55%"
		}
		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, addr+"/"+index+"/_analyze")
		if mustque != "" {
			postFilter = fmt.Sprintf(postFilter, mustque)
		}
		query = fmt.Sprintf(query, "", typeStr, question, queryPercent, entId)
		queryDSL := fmt.Sprintf(totalQuery, postFilter, query, `"answer","question"`, 1)
		//logx.Info("queryDSL:", queryDSL)
		return queryDSL
	}
	return ""
}

func DSL4SearchByKwsOrid(keyWords string, entId string, mark int) string {

	var (
		sql        = `{"query": {"bool": {"must": [%s%s]}}}`
		queryMatch = `{"match":{"keywords":{"query":"%s","minimum_should_match":"%s"}}}`
		ridTerms   = `,{"term":{"entId":%s}}`
	)
	if mark == 1 {
		queryMatch = fmt.Sprintf(queryMatch, keyWords, "30%")
	} else {
		queryMatch = fmt.Sprintf(queryMatch, keyWords, "20%")
	}
	ridTerms = fmt.Sprintf(ridTerms, entId)
	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
}

var queryStr = `{"_source": ["question","intention","answer"],"size": %d, "min_score":%v,"query": {"bool": {"must": [{"script_score": {"query": {"term":{"entId":%v}},"script": {"source": "cosine(params.queryVector,doc['questionVector'])+1", "params": {"queryVector": %v}}}}]}}}`

func GetAnswerQueryStr(question string, entId string, size int, minScore float64) string {
	qv, _ := EncodeVector(question)
	bs, err := json.Marshal(qv)
	if err != nil {
		logx.Info("向量序列化失败：", err)
	}
	return fmt.Sprintf(queryStr, size, minScore, entId, string(bs))
}