123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167 |
- package main
- import (
- "context"
- "encoding/json"
- "fmt"
- "github.com/olivere/elastic/v7"
- "io"
- util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
- "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
- "log"
- "regexp"
- "strings"
- )
- // getProject 获取项目数据
- func getProject() {
- MgoB := &mongodb.MongodbSim{
- MongodbAddr: "172.17.189.140:27080",
- //MongodbAddr: "127.0.0.1:27083",
- Size: 10,
- DbName: "qfw",
- UserName: "SJZY_RWbid_ES",
- Password: "SJZY@B4i4D5e6S",
- //Direct: true,
- }
- MgoB.InitPool()
- url := "http://172.17.4.184:19805"
- //url := "http://127.0.0.1:19805"
- username := "es_all"
- password := "TopJkO2E_d1x"
- index := "projectset" //索引名称
- // 创建 Elasticsearch 客户端
- client, err := elastic.NewClient(
- elastic.SetURL(url),
- elastic.SetBasicAuth(username, password),
- elastic.SetSniff(false),
- )
- if err != nil {
- log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
- }
- rangeQuery := elastic.NewRangeQuery("pici").Gt("1685548800").Lte("1704038400")
- //termQ := elastic.NewTermQuery("multipackage", 0)
- //rangeQuery := elastic.NewRangeQuery("id").Gt("657b08556977356f5578cb25").Lte("657b08556977356f5578cb26")
- query := elastic.NewBoolQuery().Must(rangeQuery)
- ctx := context.Background()
- //开始滚动搜索
- scrollID := ""
- scroll := "1m"
- searchSource := elastic.NewSearchSource().
- Query(query).
- Size(100).
- Sort("_doc", true) //升序排序
- //Sort("_doc", false) //降序排序
- searchService := client.Scroll(index).
- Size(100).
- Scroll(scroll).
- SearchSource(searchSource)
- res, err := searchService.Do(ctx)
- if err != nil {
- if err == io.EOF {
- fmt.Println("没有数据")
- } else {
- panic(err)
- }
- }
- defer client.ClearScroll().ScrollId(scrollID).Do(ctx) // 在退出时清理资源
- fmt.Println("总数是:", res.TotalHits())
- total := 0
- for len(res.Hits.Hits) > 0 {
- for _, hit := range res.Hits.Hits {
- var doc map[string]interface{}
- err := json.Unmarshal(hit.Source, &doc)
- if err != nil {
- log.Printf("解析文档失败:%s", err)
- continue
- }
- //id := util.ObjToString(doc["id"])
- //log.Println(id)
- var matchWords = make([]string, 0)
- var matchList = make([]interface{}, 0)
- if list, ok := doc["list"].([]interface{}); ok {
- for _, v := range list {
- if da, ok := v.(map[string]interface{}); ok {
- if util.ObjToString(da["toptype"]) != "招标" {
- continue
- }
- title := util.ObjToString(da["title"])
- // 使用正则表达式进行匹配
- matches := GetMatches(title)
- if len(matches) > 0 {
- matchList = append(matchList, da)
- }
- matchWords = append(matchWords, matches...)
- }
- }
- }
- insert := make(map[string]interface{})
- insert["project_id"] = doc["id"]
- insert["_id"] = doc["id"]
- insert["multipackage"] = doc["multipackage"]
- insert["list"] = doc["list"]
- insert["projectname"] = doc["projectname"]
- insert["sourceinfourl"] = doc["sourceinfourl"]
- if len(matchWords) > 0 {
- insert["matchList"] = matchList
- insert["package_name"] = util.ObjToString(doc["projectname"]) + "-" + strings.Join(matchWords, "、")
- insert["type"] = 1
- }
- MgoB.SaveByOriID("wcc_project_20240304", insert)
- }
- total = total + len(res.Hits.Hits)
- scrollID = res.ScrollId
- res, err = client.Scroll().ScrollId(scrollID).Scroll(scroll).Do(ctx)
- log.Println("current count:", total)
- if err != nil {
- if err == io.EOF {
- // 滚动到最后一批数据,退出循环
- break
- }
- log.Printf("滚动搜索失败:%s", err)
- break // 处理错误时退出循环
- }
- }
- fmt.Println("结束~~~~~~~~~~~~~~~")
- }
- func GetMatches(title string) (res []string) {
- // 编译正则表达式
- re := regexp.MustCompile(`包\d+`) // 标题只有一个包2
- // 定义正则表达式
- re2 := regexp.MustCompile(`包(\d+(?:、\d+)*)`) //标题含有多个包;包10、12、14、17、18、19
- re3 := regexp.MustCompile(`\d+包`) //冀中股份2023年12月标准件计划框架协议采购2包12.7
- re4 := regexp.MustCompile(`标包\d+`) //2024一季度水火电一般工序类中小金工件(标包1)-招标公告
- //re4 := regexp.MustCompile(`\d+标段`)
- //text := "中国绿发投资集团有限公司直属项目公司2023年第20批集中采购非招标项目(包10、12、14、17、18、19"
- //matches := re3.FindAllString(util.ObjToString(text), -1)
- matches := re2.FindAllString(util.ObjToString(title), -1)
- if len(matches) > 0 {
- return matches
- }
- matches = re4.FindAllString(util.ObjToString(title), -1)
- if len(matches) > 0 {
- return matches
- }
- matches = re3.FindAllString(util.ObjToString(title), -1)
- if len(matches) > 0 {
- return matches
- }
- matches = re.FindAllString(util.ObjToString(title), -1)
- return matches
- }
|