bidding.go 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797
  1. package main
  2. import (
  3. "context"
  4. "encoding/json"
  5. "fmt"
  6. "github.com/olivere/elastic/v7"
  7. "github.com/xuri/excelize/v2"
  8. "io"
  9. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  10. "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
  11. "log"
  12. "strings"
  13. )
  14. func getBiddingLimitData() {
  15. //url := "http://172.17.4.184:19908"
  16. url := "http://127.0.0.1:19908"
  17. username := "jybid"
  18. password := "Top2023_JEB01i@31"
  19. index := "bidding" //索引名称
  20. // 创建 Elasticsearch 客户端
  21. client, err := elastic.NewClient(
  22. elastic.SetURL(url),
  23. elastic.SetBasicAuth(username, password),
  24. elastic.SetSniff(false),
  25. )
  26. if err != nil {
  27. log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
  28. }
  29. // 构建查询
  30. query := elastic.NewBoolQuery().
  31. Must(elastic.NewMatchQuery("detail", "开标记录"))
  32. // 执行查询
  33. searchResult, err := client.Search().Size(100).
  34. Index(index).
  35. Query(query).
  36. Do(context.Background())
  37. if err != nil {
  38. log.Fatalf("Error executing search: %s", err)
  39. }
  40. // 本地数据库
  41. MgoB = &mongodb.MongodbSim{
  42. //MongodbAddr: "172.31.31.202:27081,172.20.45.128:27080",
  43. MongodbAddr: "127.0.0.1:27083",
  44. Size: 10,
  45. DbName: "qfw",
  46. UserName: "SJZY_RWbid_ES",
  47. Password: "SJZY@B4i4D5e6S",
  48. //Direct: true,
  49. }
  50. MgoB.InitPool()
  51. for _, hit := range searchResult.Hits.Hits {
  52. var doc map[string]interface{}
  53. err := json.Unmarshal(hit.Source, &doc)
  54. if err != nil {
  55. log.Printf("解析文档失败:%s", err)
  56. continue
  57. }
  58. MgoB.SaveByOriID("wcc_bidding_test", doc)
  59. }
  60. log.Println("数据处理完毕")
  61. }
  62. // getBidding2 获取bidding数据
  63. func getBidding2() {
  64. //url := "http://172.17.4.184:19908"
  65. url := "http://127.0.0.1:19905"
  66. username := "jybid"
  67. password := "Top2023_JEB01i@31"
  68. index := "bidding_temp" //索引名
  69. // 创建 Elasticsearch 客户端
  70. client, err := elastic.NewClient(
  71. elastic.SetURL(url),
  72. elastic.SetBasicAuth(username, password),
  73. elastic.SetSniff(false),
  74. )
  75. if err != nil {
  76. log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
  77. }
  78. //85 抽取库
  79. //Mgo := &mongodb.MongodbSim{
  80. // //MongodbAddr: "127.0.0.1:27080",
  81. // MongodbAddr: "172.17.4.85:27080",
  82. // DbName: "top",
  83. // Size: 10,
  84. // //Direct: true,
  85. //}
  86. //Mgo.InitPool()
  87. //2023年01-01 2023-10-01,,1-3季度
  88. //2024-1 - 2024-4;1704038400-1711900800
  89. //2023-10-1 2024-1-1;1696089600-1704038400
  90. //城市范围
  91. //areaTermsQuery := elastic.NewTermsQuery("company_city", "北京市")
  92. //rangeQuery := elastic.NewRangeQuery("publishtime").Gte(1704038400).Lt(1735660800)
  93. //query := elastic.NewBoolQuery().
  94. // Must(areaTermsQuery).
  95. // Must(rangeQuery)
  96. //---------------------------//
  97. //query := elastic.NewBoolQuery()
  98. //query.Must(elastic.NewMatchQuery("company_area", "广东"))
  99. ////query.Must(elastic.NewTermQuery("company_type", "北京市"))
  100. // 构建查询条件
  101. //query := elastic.NewBoolQuery().
  102. // MustNot(elastic.NewTermQuery("company_type", "个体工商户")). // 排除 company_type 为 "个体工商户"
  103. // Filter(elastic.NewTermQuery("company_area", "广东")) // 过滤 company_area 为 "广东"
  104. //rangeQuery := elastic.NewRangeQuery("comeintime").Gte("1640966400").Lt("1703952000")
  105. //query := elastic.NewBoolQuery().
  106. // Should(
  107. // elastic.NewMatchQuery("subtype", "合同"),
  108. // elastic.NewMatchQuery("toptype", "结果"),
  109. // elastic.NewMatchQuery("toptype", "招标"),
  110. // ).
  111. // MinimumShouldMatch("1"). // 至少满足一个Should条件 // buyer = "中国建筑股份有限公司"
  112. // Must(elastic.NewRangeQuery("publishtime").Gte(1704038400).Lt(1735660800))
  113. query := elastic.NewBoolQuery().
  114. Must(elastic.NewRangeQuery("id").Gt("6882d8d0d5d8e4081fd3ada6").Lte("6882d8e4d5d8e4081fd3addf"))
  115. ctx := context.Background()
  116. //开始滚动搜索
  117. scrollID := ""
  118. scroll := "10m"
  119. searchSource := elastic.NewSearchSource().
  120. Query(query).
  121. Size(10000).
  122. Sort("_doc", true) //升序排序
  123. //Sort("_doc", false) //降序排序
  124. searchService := client.Scroll(index).
  125. Size(10000).
  126. Scroll(scroll).
  127. SearchSource(searchSource)
  128. res, err := searchService.Do(ctx)
  129. if err != nil {
  130. if err == io.EOF {
  131. fmt.Println("没有数据")
  132. } else {
  133. panic(err)
  134. }
  135. }
  136. //defer client.ClearScroll().ScrollId(scrollID).Do(ctx) // 在退出时清理资源
  137. fmt.Println("总数是:", res.TotalHits())
  138. total := 0
  139. for len(res.Hits.Hits) > 0 {
  140. for _, hit := range res.Hits.Hits {
  141. var doc map[string]interface{}
  142. err := json.Unmarshal(hit.Source, &doc)
  143. if err != nil {
  144. log.Printf("解析文档失败:%s", err)
  145. continue
  146. }
  147. err = MgoB.InsertOrUpdate("qfw", "wcc_95_bidding_temp", doc)
  148. if err != nil {
  149. log.Println("error", doc["id"])
  150. }
  151. }
  152. total = total + len(res.Hits.Hits)
  153. scrollID = res.ScrollId
  154. res, err = client.Scroll().ScrollId(scrollID).Scroll(scroll).Do(ctx)
  155. log.Println("current count:", total)
  156. if err != nil {
  157. if err == io.EOF {
  158. // 滚动到最后一批数据,退出循环
  159. break
  160. }
  161. log.Println("滚动搜索失败:", err, res)
  162. break // 处理错误时退出循环
  163. }
  164. }
  165. // 在循环外调用 ClearScroll
  166. _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
  167. if err != nil {
  168. log.Printf("清理滚动搜索失败:%s", err)
  169. }
  170. fmt.Println("结束~~~~~~~~~~~~~~~")
  171. }
  172. func getBidding() {
  173. url := "http://172.17.4.184:19908"
  174. //url := "http://127.0.0.1:19908"
  175. username := "jybid"
  176. password := "Top2023_JEB01i@31"
  177. index := "bidding" //索引名称
  178. // 创建 Elasticsearch 客户端
  179. client, err := elastic.NewClient(
  180. elastic.SetURL(url),
  181. elastic.SetBasicAuth(username, password),
  182. elastic.SetSniff(false),
  183. )
  184. if err != nil {
  185. log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
  186. }
  187. //85 抽取库
  188. //Mgo := &mongodb.MongodbSim{
  189. // //MongodbAddr: "127.0.0.1:27080",
  190. // MongodbAddr: "172.17.4.85:27080",
  191. // DbName: "top",
  192. // Size: 10,
  193. // //Direct: true,
  194. //}
  195. //Mgo.InitPool()
  196. MgoB := &mongodb.MongodbSim{
  197. MongodbAddr: "172.31.31.202:27081,172.20.45.128:27080",
  198. //MongodbAddr: "127.0.0.1:27083",
  199. Size: 10,
  200. DbName: "qfw",
  201. UserName: "SJZY_RWbid_ES",
  202. Password: "SJZY@B4i4D5e6S",
  203. //Direct: true,
  204. }
  205. MgoB.InitPool()
  206. //2023年01-01 2023-10-01,,1-3季度
  207. //2024-1 - 2024-4;1704038400-1711900800
  208. //2023-10-1 2024-1-1;1696089600-1704038400
  209. //城市范围
  210. //areaTermsQuery := elastic.NewTermsQuery("company_city", "北京市")
  211. //rangeQuery := elastic.NewRangeQuery("establish_date").Gte(1704038400)
  212. //query := elastic.NewBoolQuery().
  213. // Must(areaTermsQuery).
  214. // Must(rangeQuery)
  215. //---------------------------//
  216. //query := elastic.NewBoolQuery()
  217. //query.Must(elastic.NewMatchQuery("company_area", "广东"))
  218. ////query.Must(elastic.NewTermQuery("company_type", "北京市"))
  219. // 构建查询条件
  220. query := elastic.NewBoolQuery().
  221. MustNot(elastic.NewTermQuery("company_type", "个体工商户")). // 排除 company_type 为 "个体工商户"
  222. Filter(elastic.NewTermQuery("company_area", "广东")) // 过滤 company_area 为 "广东"
  223. //rangeQuery := elastic.NewRangeQuery("comeintime").Gte("1640966400").Lt("1703952000")
  224. //query := elastic.NewBoolQuery().
  225. // //北京,天津,河北,上海,江苏,浙江,安徽
  226. // //Must(elastic.NewTermQuery("area", "北京市")).
  227. // Must(elastic.NewTermsQuery("subtype", "中标", "单一", "成交", "合同")).
  228. // Must(elastic.NewTermsQuery("area", "北京", "上海", "江苏", "浙江", "广东")).
  229. // Must(rangeQuery)
  230. ctx := context.Background()
  231. //开始滚动搜索
  232. scrollID := ""
  233. scroll := "10m"
  234. searchSource := elastic.NewSearchSource().
  235. Query(query).
  236. Size(10000).
  237. Sort("_doc", true) //升序排序
  238. //Sort("_doc", false) //降序排序
  239. searchService := client.Scroll(index).
  240. Size(10000).
  241. Scroll(scroll).
  242. SearchSource(searchSource)
  243. res, err := searchService.Do(ctx)
  244. if err != nil {
  245. if err == io.EOF {
  246. fmt.Println("没有数据")
  247. } else {
  248. panic(err)
  249. }
  250. }
  251. //defer client.ClearScroll().ScrollId(scrollID).Do(ctx) // 在退出时清理资源
  252. fmt.Println("总数是:", res.TotalHits())
  253. total := 0
  254. for len(res.Hits.Hits) > 0 {
  255. for _, hit := range res.Hits.Hits {
  256. var doc map[string]interface{}
  257. err := json.Unmarshal(hit.Source, &doc)
  258. if err != nil {
  259. log.Printf("解析文档失败:%s", err)
  260. continue
  261. }
  262. //存入新表
  263. insert := map[string]interface{}{
  264. "companyaaaaaaaaaaaaaaaaaaaaaaaaaaa_name": doc["company_name"],
  265. "id": doc["id"],
  266. "credit_no": doc["credit_no"],
  267. "company_code": doc["company_code"],
  268. }
  269. 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") {
  270. insert["wcc_type"] = 1
  271. }
  272. err = MgoB.InsertOrUpdate("qfw", "wcc_2025_guangdong_qyxy", insert)
  273. if err != nil {
  274. log.Println("error", doc["id"])
  275. }
  276. }
  277. total = total + len(res.Hits.Hits)
  278. scrollID = res.ScrollId
  279. res, err = client.Scroll().ScrollId(scrollID).Scroll(scroll).Do(ctx)
  280. log.Println("current count:", total)
  281. if err != nil {
  282. if err == io.EOF {
  283. // 滚动到最后一批数据,退出循环
  284. break
  285. }
  286. log.Println("滚动搜索失败:", err, res)
  287. break // 处理错误时退出循环
  288. }
  289. }
  290. // 在循环外调用 ClearScroll
  291. _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
  292. if err != nil {
  293. log.Printf("清理滚动搜索失败:%s", err)
  294. }
  295. fmt.Println("结束~~~~~~~~~~~~~~~")
  296. }
  297. // updateHrefByEs 更新mgo by es
  298. func updateHrefByEs() {
  299. url := "http://172.17.4.184:19908"
  300. //url := "http://127.0.0.1:19908"
  301. //url := "http://127.0.0.1:19905"
  302. username := "jybid"
  303. password := "Top2023_JEB01i@31"
  304. //index := "bidding" //索引名称
  305. index := "biddingall" //索引名称
  306. // 创建 Elasticsearch 客户端
  307. client, err := elastic.NewClient(
  308. elastic.SetURL(url),
  309. elastic.SetBasicAuth(username, password),
  310. elastic.SetSniff(false),
  311. )
  312. if err != nil {
  313. log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
  314. }
  315. //MgoB = &mongodb.MongodbSim{
  316. // //MongodbAddr: "172.17.189.140:27080",
  317. // MongodbAddr: "127.0.0.1:27083",
  318. // Size: 10,
  319. // DbName: "qfw",
  320. // UserName: "SJZY_RWbid_ES",
  321. // Password: "SJZY@B4i4D5e6S",
  322. // Direct: true,
  323. //}
  324. //MgoB.InitPool()
  325. defer util.Catch()
  326. sess := MgoB.GetMgoConn()
  327. defer MgoB.DestoryMongoConn(sess)
  328. where := map[string]interface{}{
  329. "itype": 0,
  330. }
  331. it := sess.DB("qfw").C("bidding_es_update_id").Find(where).Select(nil).Iter()
  332. log.Println("taskRun 开始")
  333. count := 0
  334. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  335. if count%100 == 0 {
  336. log.Println("current:", count)
  337. }
  338. id := mongodb.BsonIdToSId(tmp["_id"])
  339. res, _ := GetByID(client, index, id)
  340. if res == nil {
  341. //没有找到
  342. update := map[string]interface{}{
  343. "itype": 0,
  344. }
  345. MgoB.UpdateById("bidding_es_update_id", id, map[string]interface{}{"$set": update})
  346. } else {
  347. // 找到对应数据了
  348. update := map[string]interface{}{
  349. "href": res["href"],
  350. "itype": 1,
  351. }
  352. update2 := map[string]interface{}{
  353. "href": res["href"],
  354. }
  355. MgoB.UpdateById("bidding_es_update_id", id, map[string]interface{}{"$set": update})
  356. MgoB.UpdateById("bidding", id, map[string]interface{}{"$set": update2})
  357. }
  358. }
  359. log.Println("over", count)
  360. }
  361. // GetByID 根据索引和 ID 获取数据
  362. func GetByID(ec *elastic.Client, index string, id string) (map[string]interface{}, error) {
  363. ctx := context.Background()
  364. // 执行 Get 请求
  365. res, err := ec.Get().
  366. Index(index).
  367. Id(id).
  368. Do(ctx)
  369. if err != nil {
  370. if elastic.IsNotFound(err) {
  371. return nil, fmt.Errorf("document not found for index '%s' and id '%s'", index, id)
  372. }
  373. return nil, fmt.Errorf("failed to get document: %w", err)
  374. }
  375. // 解析文档源
  376. if !res.Found {
  377. return nil, fmt.Errorf("document not found for index '%s' and id '%s'", index, id)
  378. }
  379. // 解析 JSON 数据到 map[string]interface{}
  380. var source map[string]interface{}
  381. if err := json.Unmarshal(res.Source, &source); err != nil {
  382. return nil, fmt.Errorf("failed to unmarshal document source: %w", err)
  383. }
  384. return source, nil
  385. }
  386. // updateXlsxDa 更新中标金额
  387. func updateXlsxDa() {
  388. f, err := excelize.OpenFile("202502项目2.xlsx")
  389. if err != nil {
  390. log.Fatal("❌ 无法打开 Excel 文件:", err)
  391. }
  392. defer func() {
  393. f.Save()
  394. if err := f.Close(); err != nil {
  395. fmt.Println(err)
  396. }
  397. }()
  398. rows, err := f.GetRows("Sheet1") // 替换为正确的 Sheet 名称
  399. if err != nil {
  400. log.Fatal("❌ 无法读取 Excel:", err)
  401. }
  402. //
  403. for i := 1; i < len(rows); i++ {
  404. row := rows[i]
  405. // 确保 E 列(索引 4)不为空
  406. if len(row) < 5 || row[4] == "" {
  407. continue
  408. }
  409. //
  410. log.Println("项目名称-------------", row[3])
  411. // 从 F 列(索引 5)开始,每 4 列为一组
  412. for colIndex := 5; colIndex+1 < len(row); colIndex += 4 {
  413. // 提取中标金额(第 1 列)和标题(第 2 列)
  414. //bidAmount := row[colIndex] // F、J、N...列
  415. title := row[colIndex+1] // G、K、O...列
  416. if title == "" {
  417. continue
  418. }
  419. where := map[string]interface{}{
  420. "title": title,
  421. }
  422. bidd, _ := MgoB.FindOne("wcc_biddind_20250318", where)
  423. if len(*bidd) > 0 {
  424. bidamount := (*bidd)["bidamount"]
  425. sss := getColumnLetter(colIndex + 1)
  426. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", sss, i+1), bidamount) // 中标金额
  427. }
  428. // 只打印非空数据
  429. //if bidAmount != "" && title != "" {
  430. // fmt.Printf("中标金额: %s, 标题: %s\n", bidAmount, title)
  431. //}
  432. }
  433. }
  434. }
  435. // getDataFromFile getDataFromFile
  436. func getDataFromFile() {
  437. url := "http://172.17.4.184:19908"
  438. //url := "http://127.0.0.1:19908"
  439. username := "jybid"
  440. password := "Top2023_JEB01i@31"
  441. //index := "bidding" //索引名称
  442. // 创建 Elasticsearch 客户端
  443. client, err := elastic.NewClient(
  444. elastic.SetURL(url),
  445. elastic.SetBasicAuth(username, password),
  446. elastic.SetSniff(false),
  447. )
  448. if err != nil {
  449. log.Fatalf("创建 Elasticsearch 客户端失败:%s", err)
  450. }
  451. // 2. 读取 Excel
  452. f, err := excelize.OpenFile("投资入库项目_202504.xlsx")
  453. if err != nil {
  454. log.Fatal("❌ 无法打开 Excel 文件:", err)
  455. }
  456. defer func() {
  457. f.Save()
  458. if err := f.Close(); err != nil {
  459. fmt.Println(err)
  460. }
  461. }()
  462. // 3. 读取 B 列(项目名称)
  463. rows, err := f.GetRows("Sheet1") // 替换为正确的 Sheet 名称
  464. if err != nil {
  465. log.Fatal("❌ 无法读取 Excel:", err)
  466. }
  467. for i := 1; i < len(rows); i++ {
  468. row := rows[i]
  469. if i == 0 || len(row) < 4 { // 跳过表头或无效数据
  470. continue
  471. }
  472. //已经有数据,跳过
  473. if len(row) > 6 {
  474. continue
  475. }
  476. projectName := row[1]
  477. if projectName == "" {
  478. continue
  479. }
  480. buyer2 := row[2] //单位明细
  481. if i%100 == 0 {
  482. log.Println("projectName", projectName)
  483. }
  484. // 查询 ES,获取最相似的前 6 个完整文档
  485. results, err := searchES(client, projectName, buyer2)
  486. if err != nil {
  487. log.Printf("⚠ 查询失败: %s, 错误: %v\n", projectName, err)
  488. continue
  489. }
  490. //
  491. if len(results) > 0 {
  492. startCol := 8 // C列对应索引3 := 3 // C 列开始
  493. for _, result := range results {
  494. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["bidamount"]) // 中标金额
  495. //f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["budget"]) // 预算金额
  496. //f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol), i+1), result["total_investment"]) // 预算金额
  497. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+1), i+1), result["title"]) // 标题
  498. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+2), i+1), result["jyhref"]) // 标讯链接
  499. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", getColumnLetter(startCol+3), i+1), result["score"]) // 相似度
  500. startCol += 4 // **向右移动 4 列,防止数据覆盖**
  501. }
  502. f.SetCellValue("Sheet1", fmt.Sprintf("%s%d", "G", i+1), "结果") // 相似度
  503. }
  504. }
  505. log.Println("over over")
  506. }
  507. func searchES(client *elastic.Client, projectName, buyer2 string) ([]map[string]interface{}, error) {
  508. query := elastic.NewBoolQuery().
  509. Must(
  510. //elastic.NewMatchQuery("projectname.pname", projectName), // 模糊匹配 projectname
  511. //elastic.NewMatchQuery("title", projectName), // 模糊匹配 projectname
  512. elastic.NewMatchQuery("detail", projectName), // 模糊匹配 projectname
  513. //elastic.NewTermQuery("area", "安徽"), // 过滤区域
  514. elastic.NewTermsQuery("subtype", "中标", "成交", "合同", "单一"), // 过滤 subtype
  515. //elastic.NewTermsQuery("toptype", "招标", "预告", "采购意向"), // 过滤 subtype
  516. //elastic.NewTermsQuery("toptype", "拟建"), // 过滤 subtype
  517. )
  518. searchResult, err := client.Search().
  519. Index("bidding").
  520. Query(query).
  521. Size(70). // 先取 12 条,确保足够数据
  522. Do(context.Background())
  523. if err != nil {
  524. return nil, err
  525. }
  526. // 结果集
  527. var results []map[string]interface{}
  528. seenProjectNames := make(map[string]bool) // 用于去重
  529. seenProjectCode := make(map[string]bool) // 用于去重
  530. bidamountMap := make(map[float64]bool)
  531. for _, hit := range searchResult.Hits.Hits {
  532. var doc map[string]interface{}
  533. err := json.Unmarshal(hit.Source, &doc)
  534. if err != nil {
  535. log.Printf("解析文档失败:%s", err)
  536. continue
  537. }
  538. // 获取 `projectname`,防止 key 不存在时的错误
  539. projectNameValue, ok := doc["projectname"].(string)
  540. bidamount := util.Float64All(doc["bidamount"])
  541. if !ok {
  542. log.Printf("⚠️ 缺少 projectname 字段,跳过:%v", doc["projectname"])
  543. continue
  544. }
  545. projectCodeValue := util.ObjToString(doc["projectcode"])
  546. if seenProjectCode[projectCodeValue] {
  547. continue
  548. }
  549. if projectCodeValue != "" {
  550. seenProjectCode[projectCodeValue] = true
  551. }
  552. // **处理额外字段**
  553. id := util.ObjToString(doc["id"])
  554. bidData, _ := MgoB.FindById("bidding", id, nil)
  555. if util.Float64All((*bidData)["total_investment"]) > 0 {
  556. doc["total_investment"] = (*bidData)["total_investment"]
  557. }
  558. doc["jyhref"] = GetJyURLByID(id)
  559. score := *hit.Score
  560. site := util.ObjToString(doc["site"])
  561. if site == "中华人民共和国自然资源部" {
  562. doc["title"] = "土地出让" + "-" + util.ObjToString(doc["title"])
  563. }
  564. doc["score"] = score //相似度
  565. detail := util.ObjToString(doc["detail"])
  566. if !strings.Contains(detail, projectName) {
  567. continue
  568. }
  569. //detail := util.ObjToString(doc["detail"])
  570. //if !strings.Contains(detail, buyer2) {
  571. // continue
  572. //}
  573. // **去重逻辑**:如果 `projectname` 已经出现过,则跳过
  574. if seenProjectNames[projectNameValue] {
  575. continue
  576. }
  577. if bidamountMap[bidamount] {
  578. continue
  579. }
  580. // **记录该 `projectname`,避免重复**
  581. seenProjectNames[projectNameValue] = true
  582. bidamountMap[bidamount] = true
  583. // **存入 MongoDB**
  584. //err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
  585. //if err != nil {
  586. // log.Println("MongoDB 插入失败:", doc["id"])
  587. //}
  588. // **加入结果集**
  589. results = append(results, doc)
  590. // **如果已经找到 6 条不同 `projectname`,就跳出循环**
  591. if len(results) >= 10 {
  592. break
  593. }
  594. }
  595. //2、判断正文包含采购单位
  596. for _, hit := range searchResult.Hits.Hits {
  597. var doc map[string]interface{}
  598. err := json.Unmarshal(hit.Source, &doc)
  599. if err != nil {
  600. log.Printf("解析文档失败:%s", err)
  601. continue
  602. }
  603. // 获取 `projectname`,防止 key 不存在时的错误
  604. projectNameValue, ok := doc["projectname"].(string)
  605. bidamount := util.Float64All(doc["bidamount"])
  606. if !ok {
  607. log.Printf("⚠️ 缺少 projectname 字段,跳过:%v", doc["projectname"])
  608. continue
  609. }
  610. // **处理额外字段**
  611. id := util.ObjToString(doc["id"])
  612. doc["jyhref"] = GetJyURLByID(id)
  613. score := *hit.Score
  614. doc["score"] = score //相似度
  615. site := util.ObjToString(doc["site"])
  616. if site == "中华人民共和国自然资源部" {
  617. doc["title"] = "土地出让" + "-" + util.ObjToString(doc["title"])
  618. }
  619. //detail := util.ObjToString(doc["detail"])
  620. //if !strings.Contains(detail, projectName) {
  621. // continue
  622. //}
  623. //判断正文包含采购单位
  624. detail := util.ObjToString(doc["detail"])
  625. if !strings.Contains(detail, buyer2) {
  626. continue
  627. }
  628. // **去重逻辑**:如果 `projectname` 已经出现过,则跳过
  629. if seenProjectNames[projectNameValue] {
  630. continue
  631. }
  632. if bidamountMap[bidamount] {
  633. continue
  634. }
  635. // **记录该 `projectname`,避免重复**
  636. seenProjectNames[projectNameValue] = true
  637. bidamountMap[bidamount] = true
  638. // **存入 MongoDB**
  639. //err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
  640. //if err != nil {
  641. // log.Println("MongoDB 插入失败:", doc["id"])
  642. //}
  643. // **加入结果集**
  644. results = append(results, doc)
  645. // **如果已经找到 6 条不同 `projectname`,就跳出循环**
  646. if len(results) >= 10 {
  647. break
  648. }
  649. }
  650. //for _, vv := range results {
  651. // id := util.ObjToString(vv["id"])
  652. // bidData, _ := MgoB.FindById("bidding", id, nil)
  653. // MgoB.SaveByOriID("wcc_biddind_20250326", bidData)
  654. //}
  655. return results, nil
  656. }
  657. // searchES 查询 ES,返回前 6 个相似 `projectname`
  658. func searchES222(client *elastic.Client, projectName string) ([]map[string]interface{}, error) {
  659. //query := elastic.NewMatchQuery("projectname", projectName)
  660. query := elastic.NewBoolQuery().
  661. Must(
  662. elastic.NewMatchQuery("projectname.pname", projectName), // 匹配 projectname
  663. elastic.NewMatchQuery("area", "安徽"), // 匹配 projectname
  664. elastic.NewTermsQuery("subtype", "中标", "成交", "合同"), // 过滤 subtype
  665. )
  666. searchResult, err := client.Search().
  667. Index("bidding").
  668. Query(query).
  669. Size(12). // 取前 6 条数据
  670. Do(context.Background())
  671. if err != nil {
  672. return nil, err
  673. }
  674. var results []map[string]interface{}
  675. for _, hit := range searchResult.Hits.Hits {
  676. var doc map[string]interface{}
  677. err := json.Unmarshal(hit.Source, &doc)
  678. if err != nil {
  679. log.Printf("解析文档失败:%s", err)
  680. continue
  681. }
  682. id := util.ObjToString(doc["id"])
  683. doc["jyhref"] = GetJyURLByID(id)
  684. score := *hit.Score
  685. doc["score"] = score //相似度
  686. err = MgoB.InsertOrUpdate("qfw", "wcc_20250312_bidding", doc)
  687. //if err != nil {
  688. // log.Println("error", doc["id"])
  689. //}
  690. results = append(results, doc)
  691. }
  692. return results, nil
  693. }
  694. func getColumnLetter(index int) string {
  695. col := ""
  696. for index > 0 {
  697. index--
  698. col = string(rune('A'+index%26)) + col
  699. index /= 26
  700. }
  701. return col
  702. }