jyps/getEs/es_test.go

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"
	"sort"
	"strconv"
	"strings"
	"testing"
)

func TestEs(T *testing.T) {

	s := "aa"
	fmt.Println(strings.Contains(s, ","))

	//re := regexp.MustCompile(`包\d+`) // 标题只有一个包2
	//re := regexp.MustCompile(`\d+批包(\d+(?:、\d+)*)`)
	// 定义正则表达式
	//re3 := regexp.MustCompile(`(?:包)?(\d+(?:、\d+)*)包?`) //标题含有多个包；包10、12、14、17、18、19
	//re3 := 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
	////re2 := 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
	//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 := re.FindAllString(util.ObjToString(text), -1)
	//log.Println(matches)

	//title := "中电科青岛科技产业园一期设计施工总承包1标段"
	//rea := regexp.MustCompile(`总承包\d标段`) //总承包1标段
	//maches := rea.FindAllString(title, -1)
	//log.Println("rea", maches)
	//reb := regexp.MustCompile(`包\d[-~]\d`) //包1-6
	//title = "2023年度分布式光伏发电项目主要设备(包1~6)设备采购合格供应商"
	//matches := reb.FindAllString(title, -1)
	//log.Println("reb", matches)

	//re := regexp.MustCompile(`\d+批包(\d+(?:、\d+)*)`)
	re := regexp.MustCompile(`包?\d{1,2}[-~、和](包)?\d{1,2}包?`) //1-6包；01-06包；01、02包;包1、包2
	title := "中国绿发投资集团有限公司直属项目公司2023年第20批集中标段1采购非招标项目（包10、12、14、17、18、19"
	matches := re.FindAllString(title, -1)
	log.Println("rec", matches)
	//
	//re2 := regexp.MustCompile(`包?\d{1,2}`) // 标题只有一个包2
	//re3 := regexp.MustCompile(`\d{1,2}包`) // 标题只有一个包2
	re3 := regexp.MustCompile(`(标段[1-9一二三四五六七八九]|包[1-9一二三四五六七八九]?[0-9]|[1-9一二三四五六七八九]?[0-9]包|[a-kA-K]包)`) // 标题只有一个包2
	title = "济南市历下标段3区东关中心幼儿园超市包21、22、23、24"
	matches = re3.FindAllString(title, -1)
	log.Println("aaa", matches)

	//ree := regexp.MustCompile(`第\d{1,2}[批]|\d{1,2}标段|标段\d{1,2}`)
	//ree2 := regexp.MustCompile(`(第)?\d{1,2}[标段\d{1,2}]`)

	//log.Println(ree2.FindAllString(title, -1))
	// 原始数据
	data := []string{"包1", "包1-2", "包4", "包4-5", "包3", "包5", "包9", "包11", "包12", "11包", "标段1", "包13"}
	// 调用去重函数
	uniquePackages := removeDuplicates(data)
	// 输出去重后的结果
	fmt.Println("去重后的数据:", uniquePackages)

}

func removeDuplicates(data []string) []string {
	// 存储已存在的包号
	existingPackages := make(map[int]bool)

	// 存储包含包号信息的字符串
	packages := make(map[int]string)

	// 匹配包号的正则表达式
	re := regexp.MustCompile(`(包)(\d+)(?:-(\d+))?`)

	noexists := make([]string, 0)
	// 遍历数据
	for _, item := range data {
		// 提取包号信息
		matches := re.FindStringSubmatch(item)
		if len(matches) < 3 {
			noexists = append(noexists, item)
			continue
		}

		// 解析包号
		start, _ := strconv.Atoi(matches[2])
		end := start
		if len(matches[3]) > 0 {
			end, _ = strconv.Atoi(matches[3])
		}

		// 添加到已存在的包号中
		for i := start; i <= end; i++ {
			existingPackages[i] = true
		}

		// 将包含包号信息的字符串存储到 packages 中
		packages[start] = matches[0]
	}

	// 从 map 中提取去重后的包号并排序
	var uniquePackages []int
	for packageNum := range existingPackages {
		uniquePackages = append(uniquePackages, packageNum)
	}
	sort.Ints(uniquePackages)

	// 将连续的包号转换为包含范围的字符串
	var result []string
	var start, end int
	for i, num := range uniquePackages {
		if i == 0 {
			start = num
			end = num
		} else if num == end+1 {
			end = num
		} else {
			if start == end {
				result = append(result, packages[start])
			} else {
				result = append(result, fmt.Sprintf("包%d-%d", start, end))
			}
			start = num
			end = num
		}
	}
	if start == end {
		result = append(result, packages[start])
	} else {
		result = append(result, fmt.Sprintf("包%d-%d", start, end))
	}

	result = append(result, noexists...)
	return result
}

// syncEs 同步es 数据道信集群
func TestSyncEs(T *testing.T) {
	//url := "http://172.17.4.184:19805"
	url := "http://127.0.0.1:19805"
	username := "es_all"
	password := "TopJkO2E_d1x"
	index := "bidding" //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)
	}

	url2 := "http://127.0.0.1:19905"
	username2 := "jybid"
	password2 := "Top2023_JEB01i@31"
	// 创建 Elasticsearch 客户端
	client2, err := elastic.NewClient(
		elastic.SetURL(url2),
		elastic.SetBasicAuth(username2, password2),
		elastic.SetSniff(false),
	)
	if err != nil {
		log.Fatalf("创建 Elasticsearch 客户端失败：%s", err)
	}

	rangeQuery := elastic.NewRangeQuery("id").Gte("65869b436977356f55a01b0b").Lt("6586a4196977356f55a02c79")
	query := elastic.NewBoolQuery().Must(rangeQuery)

	ctx := context.Background()
	//开始滚动搜索
	scrollID := ""
	scroll := "1m"
	searchSource := elastic.NewSearchSource().
		Query(query).
		Size(10000).
		Sort("_doc", true) //升序排序
	//Sort("_doc", false) //降序排序
	searchService := client.Scroll(index).
		Size(10000).
		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"])

			client2.Index().Index(index).Id(id).BodyJson(doc).Do(ctx)
		}
		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 TestGetP(T *testing.T) {
	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("1672502400").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(1000).
		Sort("_doc", true) //升序排序
	//Sort("_doc", false) //降序排序
	searchService := client.Scroll(index).
		Size(1000).
		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~~~~~~~~~~~~~~~")

}