123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797 |
- package main
- import (
- "context"
- "encoding/json"
- "fmt"
- "github.com/olivere/elastic/v7"
- "github.com/xuri/excelize/v2"
- "io"
- util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
- "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
- "log"
- "strings"
- )
- func getBiddingLimitData() {
- //url := "http://172.17.4.184:19908"
- url := "http://127.0.0.1:19908"
- username := "jybid"
- password := "Top2023_JEB01i@31"
- 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)
- }
- // 构建查询
- query := elastic.NewBoolQuery().
- Must(elastic.NewMatchQuery("detail", "开标记录"))
- // 执行查询
- searchResult, err := client.Search().Size(100).
- Index(index).
- Query(query).
- Do(context.Background())
- if err != nil {
- log.Fatalf("Error executing search: %s", err)
- }
- // 本地数据库
- MgoB = &mongodb.MongodbSim{
- //MongodbAddr: "172.31.31.202:27081,172.20.45.128:27080",
- MongodbAddr: "127.0.0.1:27083",
- Size: 10,
- DbName: "qfw",
- UserName: "SJZY_RWbid_ES",
- Password: "SJZY@B4i4D5e6S",
- //Direct: true,
- }
- MgoB.InitPool()
- for _, hit := range searchResult.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
- }
- MgoB.SaveByOriID("wcc_bidding_test", doc)
- }
- log.Println("数据处理完毕")
- }
- // getBidding2 获取bidding数据
- func getBidding2() {
- //url := "http://172.17.4.184:19908"
- url := "http://127.0.0.1:19905"
- username := "jybid"
- password := "Top2023_JEB01i@31"
- index := "bidding_temp" //索引名
- // 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)
- }
- //85 抽取库
- //Mgo := &mongodb.MongodbSim{
- // //MongodbAddr: "127.0.0.1:27080",
- // MongodbAddr: "172.17.4.85:27080",
- // DbName: "top",
- // Size: 10,
- // //Direct: true,
- //}
- //Mgo.InitPool()
- //2023年01-01 2023-10-01,,1-3季度
- //2024-1 - 2024-4;1704038400-1711900800
- //2023-10-1 2024-1-1;1696089600-1704038400
- //城市范围
- //areaTermsQuery := elastic.NewTermsQuery("company_city", "北京市")
- //rangeQuery := elastic.NewRangeQuery("publishtime").Gte(1704038400).Lt(1735660800)
- //query := elastic.NewBoolQuery().
- // Must(areaTermsQuery).
- // Must(rangeQuery)
- //---------------------------//
- //query := elastic.NewBoolQuery()
- //query.Must(elastic.NewMatchQuery("company_area", "广东"))
- ////query.Must(elastic.NewTermQuery("company_type", "北京市"))
- // 构建查询条件
- //query := elastic.NewBoolQuery().
- // MustNot(elastic.NewTermQuery("company_type", "个体工商户")). // 排除 company_type 为 "个体工商户"
- // Filter(elastic.NewTermQuery("company_area", "广东")) // 过滤 company_area 为 "广东"
- //rangeQuery := elastic.NewRangeQuery("comeintime").Gte("1640966400").Lt("1703952000")
- //query := elastic.NewBoolQuery().
- // Should(
- // elastic.NewMatchQuery("subtype", "合同"),
- // elastic.NewMatchQuery("toptype", "结果"),
- // elastic.NewMatchQuery("toptype", "招标"),
- // ).
- // MinimumShouldMatch("1"). // 至少满足一个Should条件 // buyer = "中国建筑股份有限公司"
- // Must(elastic.NewRangeQuery("publishtime").Gte(1704038400).Lt(1735660800))
- query := elastic.NewBoolQuery().
- Must(elastic.NewRangeQuery("id").Gt("6882d8d0d5d8e4081fd3ada6").Lte("6882d8e4d5d8e4081fd3addf"))
- ctx := context.Background()
- //开始滚动搜索
- scrollID := ""
- scroll := "10m"
- 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
- }
- err = MgoB.InsertOrUpdate("qfw", "wcc_95_bidding_temp", doc)
- if err != nil {
- log.Println("error", doc["id"])
- }
- }
- 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.Println("滚动搜索失败:", err, res)
- break // 处理错误时退出循环
- }
- }
- // 在循环外调用 ClearScroll
- _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
- if err != nil {
- log.Printf("清理滚动搜索失败:%s", err)
- }
- fmt.Println("结束~~~~~~~~~~~~~~~")
- }
- func getBidding() {
- url := "http://172.17.4.184:19908"
- //url := "http://127.0.0.1:19908"
- username := "jybid"
- password := "Top2023_JEB01i@31"
- 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)
- }
- //85 抽取库
- //Mgo := &mongodb.MongodbSim{
- // //MongodbAddr: "127.0.0.1:27080",
- // MongodbAddr: "172.17.4.85:27080",
- // DbName: "top",
- // Size: 10,
- // //Direct: true,
- //}
- //Mgo.InitPool()
- MgoB := &mongodb.MongodbSim{
- MongodbAddr: "172.31.31.202:27081,172.20.45.128:27080",
- //MongodbAddr: "127.0.0.1:27083",
- Size: 10,
- DbName: "qfw",
- UserName: "SJZY_RWbid_ES",
- Password: "SJZY@B4i4D5e6S",
- //Direct: true,
- }
- MgoB.InitPool()
- //2023年01-01 2023-10-01,,1-3季度
- //2024-1 - 2024-4;1704038400-1711900800
- //2023-10-1 2024-1-1;1696089600-1704038400
- //城市范围
- //areaTermsQuery := elastic.NewTermsQuery("company_city", "北京市")
- //rangeQuery := elastic.NewRangeQuery("establish_date").Gte(1704038400)
- //query := elastic.NewBoolQuery().
- // Must(areaTermsQuery).
- // Must(rangeQuery)
- //---------------------------//
- //query := elastic.NewBoolQuery()
- //query.Must(elastic.NewMatchQuery("company_area", "广东"))
- ////query.Must(elastic.NewTermQuery("company_type", "北京市"))
- // 构建查询条件
- query := elastic.NewBoolQuery().
- MustNot(elastic.NewTermQuery("company_type", "个体工商户")). // 排除 company_type 为 "个体工商户"
- Filter(elastic.NewTermQuery("company_area", "广东")) // 过滤 company_area 为 "广东"
- //rangeQuery := elastic.NewRangeQuery("comeintime").Gte("1640966400").Lt("1703952000")
- //query := elastic.NewBoolQuery().
- // //北京,天津,河北,上海,江苏,浙江,安徽
- // //Must(elastic.NewTermQuery("area", "北京市")).
- // Must(elastic.NewTermsQuery("subtype", "中标", "单一", "成交", "合同")).
- // Must(elastic.NewTermsQuery("area", "北京", "上海", "江苏", "浙江", "广东")).
- // Must(rangeQuery)
- ctx := context.Background()
- //开始滚动搜索
- scrollID := ""
- scroll := "10m"
- 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
- }
- //存入新表
- insert := map[string]interface{}{
- "companyaaaaaaaaaaaaaaaaaaaaaaaaaaa_name": doc["company_name"],
- "id": doc["id"],
- "credit_no": doc["credit_no"],
- "company_code": doc["company_code"],
- }
- if strings.Contains(util.ObjToString(doc["company_name"]), "银行") || strings.Contains(util.ObjToString(doc["company_name"]), "保险") || strings.Contains(util.ObjToString(doc["company_name"]), "证券") {
- insert["wcc_type"] = 1
- }
- err = MgoB.InsertOrUpdate("qfw", "wcc_2025_guangdong_qyxy", insert)
- if err != nil {
- log.Println("error", doc["id"])
- }
- }
- 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.Println("滚动搜索失败:", err, res)
- break // 处理错误时退出循环
- }
- }
- // 在循环外调用 ClearScroll
- _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
- if err != nil {
- log.Printf("清理滚动搜索失败:%s", err)
- }
- fmt.Println("结束~~~~~~~~~~~~~~~")
- }
- // updateHrefByEs 更新mgo by es
- func updateHrefByEs() {
- url := "http://172.17.4.184:19908"
- //url := "http://127.0.0.1:19908"
- //url := "http://127.0.0.1:19905"
- username := "jybid"
- password := "Top2023_JEB01i@31"
- //index := "bidding" //索引名称
- index := "biddingall" //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)
- }
- //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()
- defer util.Catch()
- sess := MgoB.GetMgoConn()
- defer MgoB.DestoryMongoConn(sess)
- where := map[string]interface{}{
- "itype": 0,
- }
- it := sess.DB("qfw").C("bidding_es_update_id").Find(where).Select(nil).Iter()
- log.Println("taskRun 开始")
- count := 0
- for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
- if count%100 == 0 {
- log.Println("current:", count)
- }
- id := mongodb.BsonIdToSId(tmp["_id"])
- res, _ := GetByID(client, index, id)
- if res == nil {
- //没有找到
- update := map[string]interface{}{
- "itype": 0,
- }
- MgoB.UpdateById("bidding_es_update_id", id, map[string]interface{}{"$set": update})
- } else {
- // 找到对应数据了
- update := map[string]interface{}{
- "href": res["href"],
- "itype": 1,
- }
- update2 := map[string]interface{}{
- "href": res["href"],
- }
- MgoB.UpdateById("bidding_es_update_id", id, map[string]interface{}{"$set": update})
- MgoB.UpdateById("bidding", id, map[string]interface{}{"$set": update2})
- }
- }
- log.Println("over", count)
- }
- // GetByID 根据索引和 ID 获取数据
- func GetByID(ec *elastic.Client, index string, id string) (map[string]interface{}, error) {
- ctx := context.Background()
- // 执行 Get 请求
- res, err := ec.Get().
- Index(index).
- Id(id).
- Do(ctx)
- if err != nil {
- if elastic.IsNotFound(err) {
- return nil, fmt.Errorf("document not found for index '%s' and id '%s'", index, id)
- }
- return nil, fmt.Errorf("failed to get document: %w", err)
- }
- // 解析文档源
- if !res.Found {
- return nil, fmt.Errorf("document not found for index '%s' and id '%s'", index, id)
- }
- // 解析 JSON 数据到 map[string]interface{}
- var source map[string]interface{}
- if err := json.Unmarshal(res.Source, &source); err != nil {
- return nil, fmt.Errorf("failed to unmarshal document source: %w", err)
- }
- return source, nil
- }
- // updateXlsxDa 更新中标金额
- func updateXlsxDa() {
- f, err := excelize.OpenFile("202502项目2.xlsx")
- if err != nil {
- log.Fatal("❌ 无法打开 Excel 文件:", err)
- }
- defer func() {
- f.Save()
- if err := f.Close(); err != nil {
- fmt.Println(err)
- }
- }()
- rows, err := f.GetRows("Sheet1") // 替换为正确的 Sheet 名称
- if err != nil {
- log.Fatal("❌ 无法读取 Excel:", err)
- }
- //
- for i := 1; i < len(rows); i++ {
- row := rows[i]
- // 确保 E 列(索引 4)不为空
- if len(row) < 5 || row[4] == "" {
- continue
- }
- //
- log.Println("项目名称-------------", row[3])
- // 从 F 列(索引 5)开始,每 4 列为一组
- for colIndex := 5; colIndex+1 < len(row); colIndex += 4 {
- // 提取中标金额(第 1 列)和标题(第 2 列)
- //bidAmount := row[colIndex] // F、J、N...列
- title := row[colIndex+1] // G、K、O...列
- if title == "" {
- continue
- }
- where := map[string]interface{}{
- "title": title,
- }
- bidd, _ := MgoB.FindOne("wcc_biddind_20250318", where)
- if len(*bidd) > 0 {
- bidamount := (*bidd)["bidamount"]
- sss := getColumnLetter(colIndex + 1)
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", sss, i+1), bidamount) // 中标金额
- }
- // 只打印非空数据
- //if bidAmount != "" && title != "" {
- // fmt.Printf("中标金额: %s, 标题: %s\n", bidAmount, title)
- //}
- }
- }
- }
- // getDataFromFile getDataFromFile
- func getDataFromFile() {
- url := "http://172.17.4.184:19908"
- //url := "http://127.0.0.1:19908"
- username := "jybid"
- password := "Top2023_JEB01i@31"
- //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)
- }
- // 2. 读取 Excel
- f, err := excelize.OpenFile("投资入库项目_202504.xlsx")
- if err != nil {
- log.Fatal("❌ 无法打开 Excel 文件:", err)
- }
- defer func() {
- f.Save()
- if err := f.Close(); err != nil {
- fmt.Println(err)
- }
- }()
- // 3. 读取 B 列(项目名称)
- rows, err := f.GetRows("Sheet1") // 替换为正确的 Sheet 名称
- if err != nil {
- log.Fatal("❌ 无法读取 Excel:", err)
- }
- for i := 1; i < len(rows); i++ {
- row := rows[i]
- if i == 0 || len(row) < 4 { // 跳过表头或无效数据
- continue
- }
- //已经有数据,跳过
- if len(row) > 6 {
- continue
- }
- projectName := row[1]
- if projectName == "" {
- continue
- }
- buyer2 := row[2] //单位明细
- if i%100 == 0 {
- log.Println("projectName", projectName)
- }
- // 查询 ES,获取最相似的前 6 个完整文档
- results, err := searchES(client, projectName, buyer2)
- if err != nil {
- log.Printf("⚠ 查询失败: %s, 错误: %v\n", projectName, err)
- continue
- }
- //
- if len(results) > 0 {
- startCol := 8 // C列对应索引3 := 3 // C 列开始
- for _, result := range results {
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["bidamount"]) // 中标金额
- //f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["budget"]) // 预算金额
- //f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["total_investment"]) // 预算金额
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+1), i+1), result["title"]) // 标题
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+2), i+1), result["jyhref"]) // 标讯链接
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+3), i+1), result["score"]) // 相似度
- startCol += 4 // **向右移动 4 列,防止数据覆盖**
- }
- f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", "G", i+1), "结果") // 相似度
- }
- }
- log.Println("over over")
- }
- func searchES(client *elastic.Client, projectName, buyer2 string) ([]map[string]interface{}, error) {
- query := elastic.NewBoolQuery().
- Must(
- //elastic.NewMatchQuery("projectname.pname", projectName), // 模糊匹配 projectname
- //elastic.NewMatchQuery("title", projectName), // 模糊匹配 projectname
- elastic.NewMatchQuery("detail", projectName), // 模糊匹配 projectname
- //elastic.NewTermQuery("area", "安徽"), // 过滤区域
- elastic.NewTermsQuery("subtype", "中标", "成交", "合同", "单一"), // 过滤 subtype
- //elastic.NewTermsQuery("toptype", "招标", "预告", "采购意向"), // 过滤 subtype
- //elastic.NewTermsQuery("toptype", "拟建"), // 过滤 subtype
- )
- searchResult, err := client.Search().
- Index("bidding").
- Query(query).
- Size(70). // 先取 12 条,确保足够数据
- Do(context.Background())
- if err != nil {
- return nil, err
- }
- // 结果集
- var results []map[string]interface{}
- seenProjectNames := make(map[string]bool) // 用于去重
- seenProjectCode := make(map[string]bool) // 用于去重
- bidamountMap := make(map[float64]bool)
- for _, hit := range searchResult.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
- }
- // 获取 `projectname`,防止 key 不存在时的错误
- projectNameValue, ok := doc["projectname"].(string)
- bidamount := util.Float64All(doc["bidamount"])
- if !ok {
- log.Printf("⚠️ 缺少 projectname 字段,跳过:%v", doc["projectname"])
- continue
- }
- projectCodeValue := util.ObjToString(doc["projectcode"])
- if seenProjectCode[projectCodeValue] {
- continue
- }
- if projectCodeValue != "" {
- seenProjectCode[projectCodeValue] = true
- }
- // **处理额外字段**
- id := util.ObjToString(doc["id"])
- bidData, _ := MgoB.FindById("bidding", id, nil)
- if util.Float64All((*bidData)["total_investment"]) > 0 {
- doc["total_investment"] = (*bidData)["total_investment"]
- }
- doc["jyhref"] = GetJyURLByID(id)
- score := *hit.Score
- site := util.ObjToString(doc["site"])
- if site == "中华人民共和国自然资源部" {
- doc["title"] = "土地出让" + "-" + util.ObjToString(doc["title"])
- }
- doc["score"] = score //相似度
- detail := util.ObjToString(doc["detail"])
- if !strings.Contains(detail, projectName) {
- continue
- }
- //detail := util.ObjToString(doc["detail"])
- //if !strings.Contains(detail, buyer2) {
- // continue
- //}
- // **去重逻辑**:如果 `projectname` 已经出现过,则跳过
- if seenProjectNames[projectNameValue] {
- continue
- }
- if bidamountMap[bidamount] {
- continue
- }
- // **记录该 `projectname`,避免重复**
- seenProjectNames[projectNameValue] = true
- bidamountMap[bidamount] = true
- // **存入 MongoDB**
- //err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
- //if err != nil {
- // log.Println("MongoDB 插入失败:", doc["id"])
- //}
- // **加入结果集**
- results = append(results, doc)
- // **如果已经找到 6 条不同 `projectname`,就跳出循环**
- if len(results) >= 10 {
- break
- }
- }
- //2、判断正文包含采购单位
- for _, hit := range searchResult.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
- }
- // 获取 `projectname`,防止 key 不存在时的错误
- projectNameValue, ok := doc["projectname"].(string)
- bidamount := util.Float64All(doc["bidamount"])
- if !ok {
- log.Printf("⚠️ 缺少 projectname 字段,跳过:%v", doc["projectname"])
- continue
- }
- // **处理额外字段**
- id := util.ObjToString(doc["id"])
- doc["jyhref"] = GetJyURLByID(id)
- score := *hit.Score
- doc["score"] = score //相似度
- site := util.ObjToString(doc["site"])
- if site == "中华人民共和国自然资源部" {
- doc["title"] = "土地出让" + "-" + util.ObjToString(doc["title"])
- }
- //detail := util.ObjToString(doc["detail"])
- //if !strings.Contains(detail, projectName) {
- // continue
- //}
- //判断正文包含采购单位
- detail := util.ObjToString(doc["detail"])
- if !strings.Contains(detail, buyer2) {
- continue
- }
- // **去重逻辑**:如果 `projectname` 已经出现过,则跳过
- if seenProjectNames[projectNameValue] {
- continue
- }
- if bidamountMap[bidamount] {
- continue
- }
- // **记录该 `projectname`,避免重复**
- seenProjectNames[projectNameValue] = true
- bidamountMap[bidamount] = true
- // **存入 MongoDB**
- //err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
- //if err != nil {
- // log.Println("MongoDB 插入失败:", doc["id"])
- //}
- // **加入结果集**
- results = append(results, doc)
- // **如果已经找到 6 条不同 `projectname`,就跳出循环**
- if len(results) >= 10 {
- break
- }
- }
- //for _, vv := range results {
- // id := util.ObjToString(vv["id"])
- // bidData, _ := MgoB.FindById("bidding", id, nil)
- // MgoB.SaveByOriID("wcc_biddind_20250326", bidData)
- //}
- return results, nil
- }
- // searchES 查询 ES,返回前 6 个相似 `projectname`
- func searchES222(client *elastic.Client, projectName string) ([]map[string]interface{}, error) {
- //query := elastic.NewMatchQuery("projectname", projectName)
- query := elastic.NewBoolQuery().
- Must(
- elastic.NewMatchQuery("projectname.pname", projectName), // 匹配 projectname
- elastic.NewMatchQuery("area", "安徽"), // 匹配 projectname
- elastic.NewTermsQuery("subtype", "中标", "成交", "合同"), // 过滤 subtype
- )
- searchResult, err := client.Search().
- Index("bidding").
- Query(query).
- Size(12). // 取前 6 条数据
- Do(context.Background())
- if err != nil {
- return nil, err
- }
- var results []map[string]interface{}
- for _, hit := range searchResult.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"])
- doc["jyhref"] = GetJyURLByID(id)
- score := *hit.Score
- doc["score"] = score //相似度
- err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
- //if err != nil {
- // log.Println("error", doc["id"])
- //}
- results = append(results, doc)
- }
- return results, nil
- }
- func getColumnLetter(index int) string {
- col := ""
- for index > 0 {
- index--
- col = string(rune('A'+index%26)) + col
- index /= 26
- }
- return col
- }
|