task.go 26 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930
  1. package main
  2. import (
  3. "encoding/json"
  4. "field_sync/config"
  5. "field_sync/oss"
  6. "fmt"
  7. "net"
  8. "reflect"
  9. "regexp"
  10. "sort"
  11. "strconv"
  12. "strings"
  13. "time"
  14. "go.mongodb.org/mongo-driver/bson"
  15. "go.uber.org/zap"
  16. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  17. "jygit.jydev.jianyu360.cn/data_processing/common_utils/log"
  18. "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
  19. "jygit.jydev.jianyu360.cn/data_processing/common_utils/redis"
  20. "jygit.jydev.jianyu360.cn/data_processing/common_utils/udp"
  21. )
  22. var (
  23. regLetter = regexp.MustCompile("[a-z]*")
  24. cityEndReg = regexp.MustCompile("(区|县|市)$")
  25. )
  26. func biddingTask(data []byte, mapInfo map[string]interface{}) {
  27. defer util.Catch()
  28. stype := util.ObjToString(mapInfo["stype"])
  29. if stype == "bidding" {
  30. uq := bson.M{"gtid": util.ObjToString(mapInfo["gtid"]), "lteid": util.ObjToString(mapInfo["lteid"])}
  31. MgoB.Update("bidding_processing_ids", uq, bson.M{"$set": bson.M{"dataprocess": 7, "updatetime": time.Now().Unix()}}, false, true)
  32. }
  33. //领域标签处理的数据 id段
  34. if stype == "bidding_history" {
  35. MgoB.Save("field_data_record", map[string]interface{}{"gtid": mapInfo["gtid"], "lteid": mapInfo["lteid"], "status": 0})
  36. }
  37. q, _ := mapInfo["query"].(map[string]interface{})
  38. bkey, _ := mapInfo["bkey"].(string)
  39. if q == nil {
  40. q = map[string]interface{}{
  41. "_id": map[string]interface{}{
  42. "$gt": mongodb.StringTOBsonId(mapInfo["gtid"].(string)),
  43. "$lte": mongodb.StringTOBsonId(mapInfo["lteid"].(string)),
  44. },
  45. }
  46. }
  47. //extract库
  48. extractConn := MgoE.GetMgoConn()
  49. defer MgoE.DestoryMongoConn(extractConn)
  50. extractResult := extractConn.DB(MgoE.DbName).C(config.Conf.DB.MongoE.Coll).Find(q).Select(map[string]interface{}{
  51. "field_source": 0,
  52. "kvtext": 0,
  53. }).Sort("_id").Iter()
  54. eMap := map[string]map[string]interface{}{}
  55. extCount, repeatCount := 0, 0
  56. for tmp := make(map[string]interface{}); extractResult.Next(tmp); extCount++ {
  57. if util.IntAll(tmp["repeat"]) == 1 {
  58. repeatCount++
  59. }
  60. tid := mongodb.BsonIdToSId(tmp["_id"])
  61. eMap[tid] = tmp
  62. tmp = make(map[string]interface{})
  63. }
  64. log.Info("抽取表", zap.Int("数据量", extCount), zap.Int("重复数据量", repeatCount))
  65. //bidding库
  66. biddingConn := MgoB.GetMgoConn()
  67. count, _ := biddingConn.DB(MgoB.DbName).C(config.Conf.DB.MongoB.Coll).Find(&q).Count()
  68. log.Info("bidding表", zap.Int64("同步总数:", count))
  69. c := 0
  70. if count < 500000 {
  71. var res []map[string]interface{}
  72. result := biddingConn.DB(MgoB.DbName).C(config.Conf.DB.MongoB.Coll).Find(q).Select(map[string]interface{}{
  73. "contenthtml": 0,
  74. }).Iter()
  75. for tmp := make(map[string]interface{}); result.Next(tmp); {
  76. res = append(res, tmp)
  77. tmp = make(map[string]interface{})
  78. }
  79. MgoB.DestoryMongoConn(biddingConn)
  80. log.Info("查询结果", zap.Int64("bidding", count), zap.Int("抽取:", extCount))
  81. c = doIndex(res, eMap, bkey, stype)
  82. } else {
  83. log.Info("查询结果", zap.Int64("数据量太大,放弃", count))
  84. MgoB.DestoryMongoConn(biddingConn)
  85. }
  86. log.Info("bidding sync...over", zap.Int64("all", count), zap.Int("extract sync", c))
  87. NextNode(mapInfo, stype)
  88. NextNodePro(mapInfo, stype)
  89. NextNodeTidb(mapInfo, stype)
  90. if stype == "bidding_history" {
  91. NextNodeBidData(mapInfo) // bidding-data数据
  92. NextNodeTidbQyxy(mapInfo) // tidb-企业数据
  93. NextNodeHn(mapInfo)
  94. }
  95. }
  96. func biddingAllTask(data []byte, mapInfo map[string]interface{}) {
  97. defer util.Catch()
  98. q, _ := mapInfo["query"].(map[string]interface{})
  99. if q == nil {
  100. q = map[string]interface{}{
  101. "_id": map[string]interface{}{
  102. "$gt": mongodb.StringTOBsonId(mapInfo["gtid"].(string)),
  103. "$lte": mongodb.StringTOBsonId(mapInfo["lteid"].(string)),
  104. },
  105. }
  106. }
  107. //extract库
  108. extractConn := MgoE.GetMgoConn()
  109. defer MgoE.DestoryMongoConn(extractConn)
  110. extractResult := extractConn.DB(MgoE.DbName).C(config.Conf.DB.MongoE.Coll).Find(q).Select(map[string]interface{}{
  111. "field_source": 0,
  112. "kvtext": 0,
  113. }).Sort("-_id").Iter()
  114. //bidding库
  115. biddingConn := MgoB.GetMgoConn()
  116. defer MgoB.DestoryMongoConn(biddingConn)
  117. count := 0
  118. var compare map[string]interface{}
  119. result := biddingConn.DB(MgoB.DbName).C(config.Conf.DB.MongoB.Coll).Find(q).Select(map[string]interface{}{
  120. "contenthtml": 0,
  121. "field_source": 0,
  122. }).Sort("-_id").Iter()
  123. for tmp := make(map[string]interface{}); result.Next(tmp); count++ {
  124. update := map[string]interface{}{}
  125. del := map[string]interface{}{} //记录extract没有值而bidding中有值的字段
  126. //对比方法----------------
  127. for {
  128. if compare == nil {
  129. compare = make(map[string]interface{})
  130. if !extractResult.Next(compare) {
  131. break
  132. }
  133. }
  134. if compare != nil {
  135. cid := mongodb.BsonIdToSId(compare["_id"])
  136. tid := mongodb.BsonIdToSId(tmp["_id"])
  137. if cid == tid {
  138. //更新bidding表;bidding表modifyinfo中的字段不更新
  139. modifyinfo := make(map[string]bool)
  140. if tmpmodifyinfo, ok := tmp["modifyinfo"].(map[string]interface{}); ok && tmpmodifyinfo != nil {
  141. for k := range tmpmodifyinfo {
  142. modifyinfo[k] = true
  143. }
  144. }
  145. for _, k := range config.Conf.Serve.FieldS {
  146. v1 := compare[k] //extract
  147. v2 := tmp[k] //bidding
  148. if v2 == nil && v1 != nil {
  149. update[k] = v1
  150. } else if v2 != nil && v1 != nil && !modifyinfo[k] {
  151. update[k] = v1
  152. } else if v2 != nil && v1 == nil && !modifyinfo[k] {
  153. if k == "s_subscopeclass" && del["subscopeclass"] == nil {
  154. continue
  155. } else if k == "s_topscopeclass" && del["topscopeclass"] == nil {
  156. continue
  157. }
  158. del[k] = 1
  159. //util.Debug("抽取结果没有值,bidding有值:field--", k, del)
  160. }
  161. }
  162. if util.IntAll(compare["repeat"]) == 1 {
  163. update["extracttype"] = -1
  164. update["dataprocess"] = 7
  165. update["repeat_id"] = compare["repeat_id"]
  166. } else {
  167. update["extracttype"] = 1
  168. update["dataprocess"] = 8
  169. }
  170. break
  171. } else {
  172. if cid < tid {
  173. compare = nil
  174. continue
  175. } else {
  176. break
  177. }
  178. }
  179. } else {
  180. break
  181. }
  182. }
  183. //------------------对比结束
  184. //处理分类
  185. if compare != nil { //extract
  186. fieldFun(compare, update)
  187. compare = nil
  188. }
  189. // entidlist
  190. extractMap := make(map[string]interface{})
  191. if update["s_winner"] != "" {
  192. cid := companyFun(update)
  193. if len(cid) > 0 {
  194. update["entidlist"] = cid
  195. extractMap["entidlist"] = cid
  196. }
  197. }
  198. if len(extractMap) > 0 {
  199. updateExtPool <- []map[string]interface{}{
  200. {"_id": tmp["_id"]},
  201. {"$set": extractMap},
  202. }
  203. }
  204. // 附件有效字段
  205. if i := validFile(tmp); i != 0 {
  206. if i == -1 {
  207. update["isValidFile"] = false
  208. } else {
  209. update["isValidFile"] = true
  210. }
  211. }
  212. if len(update) > 0 {
  213. if len(del) > 0 { //删除的字段
  214. updateBidPool <- []map[string]interface{}{{
  215. "_id": tmp["_id"],
  216. },
  217. {"$set": update, "$unset": del},
  218. }
  219. } else {
  220. updateBidPool <- []map[string]interface{}{{
  221. "_id": tmp["_id"],
  222. },
  223. {"$set": update},
  224. }
  225. }
  226. }
  227. if count%50000 == 0 {
  228. log.Info("biddingTask", zap.Int("current", count))
  229. }
  230. tmp = make(map[string]interface{})
  231. }
  232. log.Info("biddingAll sync...over", zap.Int("all", count))
  233. }
  234. func doIndex(infos []map[string]interface{}, eMap map[string]map[string]interface{}, bkey, stype string) int {
  235. syncNo := 0 //抽取表数据同步数量
  236. //对比两张表数据,减少查询次数
  237. var compare map[string]interface{}
  238. var bidUpdate [][]map[string]interface{}
  239. var extUpdate [][]map[string]interface{}
  240. //SaveEsLock := &sync.Mutex{}
  241. log.Info("start ...")
  242. for n, tmp := range infos {
  243. tid := mongodb.BsonIdToSId(tmp["_id"])
  244. update := map[string]interface{}{} //要更新的mongo数据
  245. del := map[string]interface{}{}
  246. //对比方法----------------
  247. if eMap[tid] != nil {
  248. compare = eMap[tid]
  249. if stype == "bidding" {
  250. // 增量id段 正常数据
  251. if dg := util.IntAll(compare["dataging"]); dg == 1 { //extract中dataging=1跳过
  252. tmp = make(map[string]interface{})
  253. compare = nil
  254. continue
  255. }
  256. delete(eMap, tid)
  257. }
  258. if stype == "bidding_history" {
  259. //增量id段 历史数据
  260. if compare["history_updatetime"] == nil { //extract中history_updatetime不存在跳过
  261. tmp = make(map[string]interface{})
  262. compare = nil
  263. continue
  264. }
  265. delete(eMap, tid)
  266. }
  267. syncNo++
  268. for _, k := range config.Conf.Serve.FieldS {
  269. v1 := compare[k] //extract
  270. v2 := tmp[k] //bidding
  271. if v2 == nil && v1 != nil {
  272. update[k] = v1
  273. } else if v2 != nil && v1 != nil {
  274. update[k] = v1
  275. } else if v2 != nil && v1 == nil {
  276. if k == "s_subscopeclass" && del["subscopeclass"] == nil {
  277. continue
  278. } else if k == "s_topscopeclass" && del["topscopeclass"] == nil {
  279. continue
  280. } else if k == "city" || k == "district" {
  281. update[k] = ""
  282. } else {
  283. del[k] = 1
  284. }
  285. }
  286. }
  287. // 附件重采,数据同步时不更新判重标识
  288. if util.IntAll(compare["repeat"]) == 1 {
  289. update["extracttype"] = -1
  290. update["dataprocess"] = 7
  291. update["repeat_id"] = compare["repeat_id"]
  292. } else {
  293. update["extracttype"] = 1
  294. update["dataprocess"] = 8
  295. }
  296. } else {
  297. compare = nil
  298. if util.IntAll(tmp["dataging"]) == 1 { //修改未抽取的bidding数据的dataging
  299. update["dataging"] = 0
  300. }
  301. update["dataprocess"] = 8
  302. }
  303. //下面可以多线程跑的--->
  304. //处理分类
  305. if compare != nil { //extract
  306. fieldFun(compare, update)
  307. // publishtime 20230523
  308. if util.IntAll(tmp["publishtime"]) == -1 {
  309. if pb := methodPb(compare); pb > 0 {
  310. update["publishtime"] = pb
  311. }
  312. }
  313. compare = nil
  314. }
  315. //------------------对比结束
  316. //处理key descript
  317. if bkey == "" {
  318. DealInfo(&tmp, &update)
  319. }
  320. // entidlist
  321. extractMap := make(map[string]interface{})
  322. if update["s_winner"] != "" {
  323. cid := companyFun(update)
  324. if len(cid) > 0 {
  325. tmp["entidlist"] = cid
  326. update["entidlist"] = cid
  327. extractMap["entidlist"] = cid
  328. }
  329. }
  330. // 6.10 剑鱼发布信息分类处理, 写在这里是为了修改抽取表
  331. typeFunc(tmp, update, extractMap)
  332. if len(extractMap) > 0 {
  333. if extractMap["toptype"] != nil && extractMap["subtype"] == nil {
  334. extUpdate = append(extUpdate, []map[string]interface{}{
  335. {"_id": tmp["_id"]},
  336. {"$set": extractMap, "$unset": map[string]interface{}{"subtype": ""}},
  337. })
  338. } else {
  339. extUpdate = append(extUpdate, []map[string]interface{}{
  340. {"_id": tmp["_id"]},
  341. {"$set": extractMap},
  342. })
  343. }
  344. if len(extUpdate) >= MgoBulkSize {
  345. tmps := extUpdate
  346. MgoE.UpdateBulk(config.Conf.DB.MongoE.Coll, tmps...)
  347. extUpdate = [][]map[string]interface{}{}
  348. }
  349. }
  350. // 附件有效字段
  351. if i := validFile(tmp); i != 0 {
  352. if i == -1 {
  353. tmp["isValidFile"] = false
  354. update["isValidFile"] = false
  355. } else {
  356. tmp["isValidFile"] = true
  357. update["isValidFile"] = true
  358. }
  359. }
  360. // 2024-02-21 徐志恒 情报标签字段
  361. if tmp["tag_topinformation"] != nil {
  362. update["tag_set"] = getTagSet(tmp, compare)
  363. }
  364. if len(update) > 0 {
  365. if len(del) > 0 {
  366. bidUpdate = append(bidUpdate, []map[string]interface{}{{
  367. "_id": tmp["_id"],
  368. },
  369. {"$set": update, "$unset": del},
  370. })
  371. } else {
  372. bidUpdate = append(bidUpdate, []map[string]interface{}{{
  373. "_id": tmp["_id"],
  374. },
  375. {"$set": update},
  376. })
  377. }
  378. if len(bidUpdate) >= MgoBulkSize {
  379. tmps := bidUpdate
  380. MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, tmps...)
  381. bidUpdate = [][]map[string]interface{}{}
  382. }
  383. }
  384. if n%500 == 0 {
  385. log.Info("biddingTask", zap.Int("current", n))
  386. }
  387. tmp = make(map[string]interface{})
  388. }
  389. //SaveEsLock.Lock()
  390. if len(bidUpdate) > 0 {
  391. tmps := bidUpdate
  392. MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, tmps...)
  393. bidUpdate = [][]map[string]interface{}{}
  394. }
  395. if len(extUpdate) > 0 {
  396. tmps := extUpdate
  397. MgoE.UpdateBulk(config.Conf.DB.MongoE.Coll, tmps...)
  398. extUpdate = [][]map[string]interface{}{}
  399. }
  400. //SaveEsLock.Unlock()
  401. return syncNo
  402. }
  403. // @Description subscopeclass、topscopeclass、package
  404. // 20230523 多包处理 subpackage = 1
  405. // @Author J 2022/6/7 5:54 PM
  406. func fieldFun(compare, update map[string]interface{}) {
  407. subscopeclass, _ := compare["subscopeclass"].([]interface{}) //subscopeclass
  408. if subscopeclass != nil {
  409. m1 := map[string]bool{}
  410. newclass := []string{}
  411. for _, sc := range subscopeclass {
  412. sclass, _ := sc.(string)
  413. if !m1[sclass] {
  414. m1[sclass] = true
  415. newclass = append(newclass, sclass)
  416. }
  417. }
  418. update["s_subscopeclass"] = strings.Join(newclass, ",")
  419. update["subscopeclass"] = newclass
  420. }
  421. topscopeclass, _ := compare["topscopeclass"].([]interface{}) //topscopeclass
  422. if topscopeclass != nil {
  423. m2 := map[string]bool{}
  424. newclass := []string{}
  425. for _, tc := range topscopeclass {
  426. tclass, _ := tc.(string)
  427. tclass = regLetter.ReplaceAllString(tclass, "") // 去除字母
  428. if !m2[tclass] {
  429. m2[tclass] = true
  430. newclass = append(newclass, tclass)
  431. }
  432. }
  433. update["topscopeclass"] = topscopeclass
  434. update["s_topscopeclass"] = strings.Join(newclass, ",")
  435. }
  436. if package1 := compare["package"]; package1 != nil {
  437. packageM, _ := package1.(map[string]interface{})
  438. update["package"] = packageM
  439. for _, p := range packageM {
  440. pm, _ := p.(map[string]interface{})
  441. if util.ObjToString(pm["winner"]) != "" || util.Float64All(pm["budget"]) > 0 ||
  442. util.Float64All(pm["bidamount"]) > 0 {
  443. update["multipackage"] = 1
  444. break
  445. }
  446. }
  447. } else {
  448. update["multipackage"] = 0
  449. }
  450. // subpackage
  451. if compare["package"] != nil && compare["s_winner"] != nil && compare["bidamount"] != nil {
  452. pg := compare["package"].(map[string]interface{})
  453. if len(pg) > 1 {
  454. var bmt []float64
  455. var swn []string
  456. for _, p := range pg {
  457. p1 := p.(map[string]interface{})
  458. if p1["bidamount"] != nil {
  459. bmt = append(bmt, util.Float64All(p1["bidamount"]))
  460. }
  461. if w := util.ObjToString(p1["winner"]); w != "" {
  462. swn = append(swn)
  463. }
  464. }
  465. if len(bmt) > 1 && len(swn) > 1 {
  466. sn := strings.Split(util.ObjToString(compare["s_winner"]), ",")
  467. sort.Strings(sn)
  468. sort.Strings(swn)
  469. swn1 := util.ObjArrToStringArr(Duplicate(swn)) // 去重
  470. if strings.Join(swn1, ",") == strings.Join(sn, ",") {
  471. bidamount := 0.0
  472. for _, f := range bmt {
  473. bidamount += f
  474. }
  475. if bidamount == util.Float64All(compare["bidamount"]) {
  476. update["subpackage"] = 1
  477. }
  478. }
  479. }
  480. }
  481. }
  482. }
  483. // @Description entidlist
  484. // @Author J 2022/6/7 2:36 PM
  485. func companyFun(tmp map[string]interface{}) (cid []string) {
  486. sWinnerarr := strings.Split(util.ObjToString(tmp["s_winner"]), ",")
  487. for _, w := range sWinnerarr {
  488. if w != "" {
  489. id := redis.GetStr("qyxy_id", w)
  490. if id == "" {
  491. ents, _ := MgoQ.Find(config.Conf.DB.MongoQ.Coll, map[string]interface{}{"company_name": w}, map[string]interface{}{"updatetime": -1}, map[string]interface{}{"company_name": 1}, false, -1, -1)
  492. if len(*ents) > 0 {
  493. id = util.ObjToString((*ents)[0]["_id"])
  494. redis.PutCKV("qyxy_id", w, id)
  495. } else {
  496. ent, _ := MgoP.FindOne(config.Conf.DB.MongoP.Coll, map[string]interface{}{"history_name": w})
  497. if len(*ent) > 0 {
  498. id = util.ObjToString((*ent)["company_id"])
  499. redis.PutCKV("qyxy_id", w, id)
  500. }
  501. }
  502. }
  503. if id == "" {
  504. id = "-"
  505. }
  506. cid = append(cid, id)
  507. }
  508. }
  509. return cid
  510. }
  511. // @Description update 修改bidding表,extractM修改抽取表
  512. // @Author J 2022/6/10 10:29 AM
  513. func typeFunc(tmp, update, extractM map[string]interface{}) {
  514. if jyData, ok := tmp["jyfb_data"].(map[string]interface{}); ok {
  515. if t := util.ObjToString(jyData["type"]); t != "" {
  516. switch t {
  517. //case "采购信息":
  518. case "招标公告":
  519. if util.ObjToString(tmp["toptype"]) != "招标" {
  520. update["toptype"] = "招标"
  521. extractM["toptype"] = "招标"
  522. delete(update, "subtype")
  523. }
  524. case "采购意向":
  525. if util.ObjToString(tmp["toptype"]) != "采购意向" {
  526. update["toptype"] = "采购意向"
  527. update["subtype"] = "采购意向"
  528. extractM["toptype"] = "采购意向"
  529. extractM["subtype"] = "采购意向"
  530. }
  531. case "招标预告":
  532. if util.ObjToString(tmp["toptype"]) != "预告" {
  533. update["toptype"] = "预告"
  534. extractM["toptype"] = "预告"
  535. delete(update, "subtype")
  536. }
  537. case "招标结果":
  538. if util.ObjToString(tmp["toptype"]) != "结果" {
  539. update["toptype"] = "结果"
  540. extractM["toptype"] = "结果"
  541. delete(update, "subtype")
  542. }
  543. }
  544. }
  545. }
  546. }
  547. // @Description 附件有效字段(isValidFile)
  548. // @Author J 2022/7/8 14:41
  549. func validFile(tmp map[string]interface{}) int {
  550. isContinue := false
  551. if pinfo, o := tmp["projectinfo"].(map[string]interface{}); o {
  552. if atts, o1 := pinfo["attachments"].(map[string]interface{}); o1 {
  553. for _, att := range atts {
  554. if att == nil {
  555. continue
  556. }
  557. if reflect.TypeOf(att).String() == "string" {
  558. continue
  559. }
  560. att1 := att.(map[string]interface{})
  561. if fid := util.ObjToString(att1["fid"]); fid != "" {
  562. isContinue = true
  563. break
  564. }
  565. }
  566. if isContinue {
  567. if attachTxt, o := tmp["attach_text"].(map[string]interface{}); o {
  568. if len(attachTxt) > 0 {
  569. for _, at := range attachTxt {
  570. at1 := at.(map[string]interface{})
  571. if len(at1) > 0 {
  572. for k, _ := range at1 {
  573. if reflect.TypeOf(at1[k]).String() == "string" {
  574. continue
  575. }
  576. at2 := at1[k].(map[string]interface{})
  577. s := strings.ToLower(util.ObjToString(at2["file_name"]))
  578. if !strings.Contains(s, "jpg") || !strings.Contains(s, "jpeg") != strings.Contains(s, "png") ||
  579. strings.Contains(s, "pdf") {
  580. if strings.Contains(s, "swf") || strings.Contains(s, "html") {
  581. return -1
  582. } else if AnalysisFile(oss.OssGetObject(util.ObjToString(at2["attach_url"]))) {
  583. return 1
  584. }
  585. }
  586. }
  587. break
  588. } else {
  589. break
  590. }
  591. }
  592. }
  593. }
  594. flag := false
  595. for _, att := range atts {
  596. if att == nil {
  597. continue
  598. }
  599. if reflect.TypeOf(att).String() == "string" {
  600. continue
  601. }
  602. att1 := att.(map[string]interface{})
  603. if fid := util.ObjToString(att1["fid"]); fid != "" {
  604. ftype := strings.ToLower(util.ObjToString(tmp["ftype"]))
  605. if ftype != "swf" && ftype != "html" && oss.OssObjExists("jy-datafile", fid) {
  606. return 1
  607. } else {
  608. flag = true
  609. }
  610. }
  611. }
  612. if flag {
  613. return -1
  614. }
  615. }
  616. }
  617. }
  618. return 0
  619. }
  620. // @Description id不变,内容变化 重新索引数据
  621. // @Author J 2022/8/10 13:29
  622. func taskinfo(id string) {
  623. tmp, _ := MgoB.FindById("bidding", id, nil)
  624. if tmp == nil || len(*tmp) == 0 {
  625. log.Info(fmt.Sprintf("taskinfo bidding id=%s 未查询到数据", id))
  626. return
  627. }
  628. extractM, _ := MgoE.FindById(config.Conf.DB.MongoE.Coll, id, nil)
  629. if extractM == nil || len(*extractM) == 0 {
  630. extractM, _ = MgoE.FindById(config.Conf.DB.MongoE.Coll1, id, nil)
  631. if extractM == nil || len(*extractM) == 0 {
  632. log.Info(fmt.Sprintf("taskinfo extract id=%s 未查询到数据", id))
  633. return
  634. }
  635. }
  636. update := map[string]interface{}{} //要更新的mongo数据
  637. //更新bidding表字段
  638. for _, k := range config.Conf.Serve.FieldS {
  639. v1 := (*extractM)[k] //extract
  640. v2 := (*tmp)[k] //bidding
  641. if v2 == nil && v1 != nil {
  642. update[k] = v1
  643. } else if v2 != nil && v1 != nil {
  644. update[k] = v1
  645. } else if v2 != nil && v1 == nil {
  646. if k == "city" || k == "district" {
  647. update[k] = ""
  648. }
  649. }
  650. }
  651. if util.IntAll((*extractM)["repeat"]) == 1 {
  652. update["extracttype"] = -1
  653. update["dataprocess"] = 7
  654. update["repeat_id"] = (*extractM)["repeat_id"]
  655. } else {
  656. update["extracttype"] = 1
  657. update["dataprocess"] = 8
  658. }
  659. //处理分类
  660. fieldFun(*extractM, update)
  661. extractMap := make(map[string]interface{})
  662. if util.ObjToString((*tmp)["s_winner"]) != "" {
  663. cid := companyFun(*tmp)
  664. if len(cid) > 0 {
  665. update["entidlist"] = cid
  666. extractMap["entidlist"] = cid
  667. }
  668. MgoE.UpdateById(config.Conf.DB.MongoE.Coll, id, map[string]interface{}{"$set": extractMap})
  669. }
  670. // 附件有效字段
  671. if i := validFile(*tmp); i != 0 {
  672. if i == -1 {
  673. update["isValidFile"] = false
  674. } else {
  675. update["isValidFile"] = true
  676. }
  677. }
  678. if len(update) > 0 {
  679. MgoB.UpdateById(config.Conf.DB.MongoB.Coll, id, map[string]interface{}{"$set": update})
  680. }
  681. mapinfo := map[string]interface{}{
  682. "infoid": id,
  683. "stype": "index-by-id",
  684. }
  685. datas, _ := json.Marshal(mapinfo)
  686. var next = &net.UDPAddr{
  687. IP: net.ParseIP(config.Conf.Udp.Next.Addr),
  688. Port: util.IntAll(config.Conf.Udp.Next.Port),
  689. }
  690. log.Info("nsq data over", zap.Any("es", next), zap.String("mapinfo", string(datas)))
  691. _ = UdpClient.WriteUdp(datas, udp.OP_TYPE_DATA, next)
  692. }
  693. var DateTimeSelect = []string{"bidopentime", "bidendtime", "signaturedate", "comeintime"}
  694. // @Description 发布时间处理
  695. // @Author J 2023/5/23 14:32
  696. func methodPb(tmp map[string]interface{}) int64 {
  697. if tmp["ext_publishtime"] != nil {
  698. if newPb := util.Int64All(tmp["ext_publishtime"]); newPb < time.Now().Unix() && newPb > 1420041600 {
  699. return newPb
  700. }
  701. }
  702. for _, d := range DateTimeSelect {
  703. if tmp[d] != nil && util.Int64All(tmp[d]) < time.Now().Unix() {
  704. return util.Int64All(tmp[d])
  705. }
  706. }
  707. return 0
  708. }
  709. // Duplicate
  710. // @Description 去重
  711. // @Author J 2023/5/24 09:53
  712. func Duplicate(a interface{}) (ret []interface{}) {
  713. va := reflect.ValueOf(a)
  714. for i := 0; i < va.Len(); i++ {
  715. if i > 0 && reflect.DeepEqual(va.Index(i-1).Interface(), va.Index(i).Interface()) {
  716. continue
  717. }
  718. ret = append(ret, va.Index(i).Interface())
  719. }
  720. return ret
  721. }
  722. // @Description 获取情报标签
  723. // @Author 徐志恒 2024/2/21 09:53
  724. func getTagSet(tmp, compare map[string]interface{}) map[string]map[string]interface{} {
  725. tagSet := map[string]map[string]interface{}{}
  726. wuye := map[string]interface{}{}
  727. buyer := util.ObjToString(compare["buyer"])
  728. publishtime := util.Int64All(tmp["publishtime"])
  729. bidamount := util.Float64All(compare["bidamount"])
  730. wuye["isfirsthand"] = 62
  731. if buyer != "" {
  732. sql := `{
  733. "query": {
  734. "bool": {
  735. "must": [
  736. {
  737. "term": {
  738. "buyer": "` + buyer + `"
  739. }
  740. },
  741. {
  742. "term": {
  743. "tag_topinformation": "情报_物业"
  744. }
  745. },
  746. {
  747. "term": {
  748. "subtype": "合同"
  749. }
  750. },
  751. {
  752. "range": {
  753. "publishtime": {
  754. "lte": ` + fmt.Sprint(publishtime) + `
  755. }
  756. }
  757. }
  758. ]
  759. }
  760. },
  761. "sort": {
  762. "publishtime": "asc"
  763. },
  764. "_source": [
  765. "s_winner"
  766. ],
  767. "size": 10000
  768. }`
  769. data := Es.Get("bidding", "bidding", sql)
  770. if data != nil && len(*data) > 0 {
  771. count := 0
  772. first := util.ObjToString((*data)[0]["s_winner"])
  773. for k, v := range *data {
  774. winner := util.ObjToString(v["s_winner"])
  775. if k > 0 && first != winner {
  776. first = winner
  777. count++
  778. }
  779. }
  780. changehand := fmt.Sprintf("%.2f", float64(count)/float64(len(*data)))
  781. changehands, _ := strconv.ParseFloat(changehand, 64)
  782. wuye["changehand"] = changehands
  783. log.Println("changehands", count, len(*data), changehands)
  784. if changehands > 0.3 {
  785. wuye["changehandindex"] = 61
  786. }
  787. if len(*data) > 1 {
  788. wuye["isfirsthand"] = 0
  789. }
  790. }
  791. }
  792. if tmp["projectinfo"] != nil {
  793. projectInfo := util.ObjToMap(tmp["projectinfo"])
  794. if projectInfo != nil && len(*projectInfo) > 0 {
  795. if (*projectInfo)["attachments"] != nil {
  796. wuye["isfile"] = 63
  797. }
  798. }
  799. }
  800. wuye["period"] = getperiod(compare)
  801. wuye["scale"] = getBidamountRange(bidamount)
  802. if tmp["property_form"] != nil {
  803. property_form := util.ObjArrToStringArr(tmp["property_form"].([]interface{}))
  804. wuye["property_form"] = getpropertyform(property_form)
  805. }
  806. tagSet["wuye"] = wuye
  807. return tagSet
  808. }
  809. func getBidamountRange(value float64) int {
  810. switch {
  811. case value < 500000:
  812. return 1
  813. case value >= 500000 && value < 1000000:
  814. return 2
  815. case value >= 1000000 && value < 2000000:
  816. return 3
  817. case value >= 2000000 && value < 5000000:
  818. return 4
  819. default:
  820. return 5
  821. }
  822. }
  823. func getpropertyform(value []string) string {
  824. arr := []string{}
  825. categories := map[string]string{
  826. "住宅": "21",
  827. "政府办公楼": "22",
  828. "学校": "23",
  829. "医院": "24",
  830. "产业园区": "25",
  831. "旅游景区": "26",
  832. "交通运输": "27",
  833. "商务办公楼": "28",
  834. "酒店": "29",
  835. }
  836. for _, v := range value {
  837. if categories[v] != "" {
  838. arr = append(arr, categories[v])
  839. }
  840. }
  841. return strings.Join(arr, ",")
  842. }
  843. func getperiod(data map[string]interface{}) int {
  844. res := 16
  845. signaturedate := util.Int64All(data["signaturedate"]) //合同签订日期
  846. expiredate := util.Int64All(data["expiredate"]) //合同截止日期
  847. // contractperiod := util.ObjToString(data["contractperiod"]) //合同期限
  848. project_duration := util.IntAll(data["project_duration"]) //工期时长
  849. project_timeunit := util.ObjToString(data["project_timeunit"]) //工期单位
  850. result := float64(0)
  851. if expiredate > 0 && signaturedate > 0 {
  852. result = calculateYearDifference(signaturedate, expiredate)
  853. } else if project_duration > 0 && project_timeunit != "" {
  854. if strings.Contains(project_timeunit, "年") {
  855. if project_duration == 1 {
  856. res = 12
  857. } else if project_duration == 2 {
  858. res = 13
  859. } else if project_duration == 3 {
  860. res = 14
  861. } else if project_duration == 5 {
  862. res = 15
  863. }
  864. return res
  865. } else if strings.Contains(project_timeunit, "月") {
  866. result = float64(project_duration) / 12
  867. } else if strings.Contains(project_timeunit, "周") {
  868. result = float64(project_duration) * 7 / 365
  869. } else if strings.Contains(project_timeunit, "日") || strings.Contains(project_timeunit, "天") {
  870. result = float64(project_duration) / 365
  871. }
  872. }
  873. if result == 0 {
  874. res = 16
  875. } else if result < 1 {
  876. res = 11
  877. } else if result >= 1 && result < 2 {
  878. res = 12
  879. } else if result >= 2 && result < 3 {
  880. res = 13
  881. } else if result >= 3 && result < 4 {
  882. res = 14
  883. } else if result >= 5 {
  884. res = 15
  885. }
  886. return res
  887. }
  888. func calculateYearDifference(startTime int64, endTime int64) float64 {
  889. start := time.Unix(startTime, 0)
  890. end := time.Unix(endTime, 0)
  891. duration := end.Sub(start)
  892. years := duration.Hours() / 24 / 365
  893. return years
  894. }