task.go 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754
  1. package main
  2. import (
  3. "encoding/json"
  4. "field_sync/config"
  5. "field_sync/oss"
  6. "fmt"
  7. "go.mongodb.org/mongo-driver/bson"
  8. "go.uber.org/zap"
  9. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  10. "jygit.jydev.jianyu360.cn/data_processing/common_utils/log"
  11. "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
  12. "jygit.jydev.jianyu360.cn/data_processing/common_utils/redis"
  13. "jygit.jydev.jianyu360.cn/data_processing/common_utils/udp"
  14. "net"
  15. "reflect"
  16. "regexp"
  17. "sort"
  18. "strings"
  19. "time"
  20. )
  21. var (
  22. regLetter = regexp.MustCompile("[a-z]*")
  23. cityEndReg = regexp.MustCompile("(区|县|市)$")
  24. )
  25. func biddingTask(data []byte, mapInfo map[string]interface{}) {
  26. defer util.Catch()
  27. stype := util.ObjToString(mapInfo["stype"])
  28. if stype == "bidding" {
  29. uq := bson.M{"gtid": bson.M{"$gte": util.ObjToString(mapInfo["gtid"])},
  30. "lteid": bson.M{"$lte": util.ObjToString(mapInfo["lteid"])}}
  31. MgoB.Update("bidding_processing_ids", uq, bson.M{"$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 k == "project_startdate" || k == "project_completedate" {
  272. util.Debug(v1, v2)
  273. }
  274. if v2 == nil && v1 != nil {
  275. update[k] = v1
  276. } else if v2 != nil && v1 != nil {
  277. update[k] = v1
  278. } else if v2 != nil && v1 == nil {
  279. if k == "s_subscopeclass" && del["subscopeclass"] == nil {
  280. continue
  281. } else if k == "s_topscopeclass" && del["topscopeclass"] == nil {
  282. continue
  283. } else if k == "city" || k == "district" {
  284. update[k] = ""
  285. } else {
  286. if k == "project_startdate" || k == "project_completedate" {
  287. util.Debug(v1, v2)
  288. }
  289. del[k] = 1
  290. }
  291. }
  292. }
  293. if util.IntAll(compare["repeat"]) == 1 {
  294. update["extracttype"] = -1
  295. update["dataprocess"] = 7
  296. update["repeat_id"] = compare["repeat_id"]
  297. } else {
  298. update["extracttype"] = 1
  299. update["dataprocess"] = 8
  300. }
  301. } else {
  302. compare = nil
  303. if util.IntAll(tmp["dataging"]) == 1 { //修改未抽取的bidding数据的dataging
  304. update["dataging"] = 0
  305. }
  306. update["dataprocess"] = 8
  307. }
  308. //下面可以多线程跑的--->
  309. //处理分类
  310. if compare != nil { //extract
  311. fieldFun(compare, update)
  312. // publishtime 20230523
  313. if util.IntAll(tmp["publishtime"]) == -1 {
  314. if pb := methodPb(compare); pb > 0 {
  315. update["publishtime"] = pb
  316. }
  317. }
  318. compare = nil
  319. }
  320. //------------------对比结束
  321. //处理key descript
  322. if bkey == "" {
  323. DealInfo(&tmp, &update)
  324. }
  325. // entidlist
  326. extractMap := make(map[string]interface{})
  327. if update["s_winner"] != "" {
  328. cid := companyFun(update)
  329. if len(cid) > 0 {
  330. tmp["entidlist"] = cid
  331. update["entidlist"] = cid
  332. extractMap["entidlist"] = cid
  333. }
  334. }
  335. // 6.10 剑鱼发布信息分类处理, 写在这里是为了修改抽取表
  336. typeFunc(tmp, update, extractMap)
  337. if len(extractMap) > 0 {
  338. if extractMap["toptype"] != nil && extractMap["subtype"] == nil {
  339. extUpdate = append(extUpdate, []map[string]interface{}{
  340. {"_id": tmp["_id"]},
  341. {"$set": extractMap, "$unset": map[string]interface{}{"subtype": ""}},
  342. })
  343. } else {
  344. extUpdate = append(extUpdate, []map[string]interface{}{
  345. {"_id": tmp["_id"]},
  346. {"$set": extractMap},
  347. })
  348. }
  349. if len(extUpdate) >= MgoBulkSize {
  350. tmps := extUpdate
  351. MgoE.UpdateBulk(config.Conf.DB.MongoE.Coll, tmps...)
  352. extUpdate = [][]map[string]interface{}{}
  353. }
  354. }
  355. // 附件有效字段
  356. if i := validFile(tmp); i != 0 {
  357. if i == -1 {
  358. tmp["isValidFile"] = false
  359. update["isValidFile"] = false
  360. } else {
  361. tmp["isValidFile"] = true
  362. update["isValidFile"] = true
  363. }
  364. }
  365. if len(update) > 0 {
  366. util.Debug(update)
  367. util.Debug(del)
  368. if len(del) > 0 {
  369. bidUpdate = append(bidUpdate, []map[string]interface{}{{
  370. "_id": tmp["_id"],
  371. },
  372. {"$set": update, "$unset": del},
  373. })
  374. } else {
  375. bidUpdate = append(bidUpdate, []map[string]interface{}{{
  376. "_id": tmp["_id"],
  377. },
  378. {"$set": update},
  379. })
  380. }
  381. if len(bidUpdate) >= MgoBulkSize {
  382. tmps := bidUpdate
  383. MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, tmps...)
  384. bidUpdate = [][]map[string]interface{}{}
  385. }
  386. }
  387. if n%500 == 0 {
  388. log.Info("biddingTask", zap.Int("current", n))
  389. }
  390. tmp = make(map[string]interface{})
  391. }
  392. //SaveEsLock.Lock()
  393. if len(bidUpdate) > 0 {
  394. tmps := bidUpdate
  395. MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, tmps...)
  396. bidUpdate = [][]map[string]interface{}{}
  397. }
  398. if len(extUpdate) > 0 {
  399. tmps := extUpdate
  400. MgoE.UpdateBulk(config.Conf.DB.MongoE.Coll, tmps...)
  401. extUpdate = [][]map[string]interface{}{}
  402. }
  403. //SaveEsLock.Unlock()
  404. return syncNo
  405. }
  406. // @Description subscopeclass、topscopeclass、package
  407. // 20230523 多包处理 subpackage = 1
  408. // @Author J 2022/6/7 5:54 PM
  409. func fieldFun(compare, update map[string]interface{}) {
  410. subscopeclass, _ := compare["subscopeclass"].([]interface{}) //subscopeclass
  411. if subscopeclass != nil {
  412. m1 := map[string]bool{}
  413. newclass := []string{}
  414. for _, sc := range subscopeclass {
  415. sclass, _ := sc.(string)
  416. if !m1[sclass] {
  417. m1[sclass] = true
  418. newclass = append(newclass, sclass)
  419. }
  420. }
  421. update["s_subscopeclass"] = strings.Join(newclass, ",")
  422. update["subscopeclass"] = newclass
  423. }
  424. topscopeclass, _ := compare["topscopeclass"].([]interface{}) //topscopeclass
  425. if topscopeclass != nil {
  426. m2 := map[string]bool{}
  427. newclass := []string{}
  428. for _, tc := range topscopeclass {
  429. tclass, _ := tc.(string)
  430. tclass = regLetter.ReplaceAllString(tclass, "") // 去除字母
  431. if !m2[tclass] {
  432. m2[tclass] = true
  433. newclass = append(newclass, tclass)
  434. }
  435. }
  436. update["topscopeclass"] = topscopeclass
  437. update["s_topscopeclass"] = strings.Join(newclass, ",")
  438. }
  439. if package1 := compare["package"]; package1 != nil {
  440. packageM, _ := package1.(map[string]interface{})
  441. update["package"] = packageM
  442. for _, p := range packageM {
  443. pm, _ := p.(map[string]interface{})
  444. if util.ObjToString(pm["winner"]) != "" || util.Float64All(pm["budget"]) > 0 ||
  445. util.Float64All(pm["bidamount"]) > 0 {
  446. update["multipackage"] = 1
  447. break
  448. }
  449. }
  450. } else {
  451. update["multipackage"] = 0
  452. }
  453. // subpackage
  454. if compare["package"] != nil && compare["s_winner"] != nil && compare["bidamount"] != nil {
  455. pg := compare["package"].(map[string]interface{})
  456. if len(pg) > 1 {
  457. var bmt []float64
  458. var swn []string
  459. for _, p := range pg {
  460. p1 := p.(map[string]interface{})
  461. if p1["bidamount"] != nil {
  462. bmt = append(bmt, util.Float64All(p1["bidamount"]))
  463. }
  464. if w := util.ObjToString(p1["winner"]); w != "" {
  465. swn = append(swn)
  466. }
  467. }
  468. if len(bmt) > 1 && len(swn) > 1 {
  469. sn := strings.Split(util.ObjToString(compare["s_winner"]), ",")
  470. sort.Strings(sn)
  471. sort.Strings(swn)
  472. swn1 := util.ObjArrToStringArr(Duplicate(swn)) // 去重
  473. if strings.Join(swn1, ",") == strings.Join(sn, ",") {
  474. bidamount := 0.0
  475. for _, f := range bmt {
  476. bidamount += f
  477. }
  478. if bidamount == util.Float64All(compare["bidamount"]) {
  479. update["subpackage"] = 1
  480. }
  481. }
  482. }
  483. }
  484. }
  485. }
  486. // @Description entidlist
  487. // @Author J 2022/6/7 2:36 PM
  488. func companyFun(tmp map[string]interface{}) (cid []string) {
  489. sWinnerarr := strings.Split(util.ObjToString(tmp["s_winner"]), ",")
  490. for _, w := range sWinnerarr {
  491. if w != "" {
  492. id := redis.GetStr("qyxy_id", w)
  493. if id == "" {
  494. 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)
  495. if len(*ents) > 0 {
  496. id = util.ObjToString((*ents)[0]["_id"])
  497. redis.PutCKV("qyxy_id", w, id)
  498. } else {
  499. ent, _ := MgoP.FindOne(config.Conf.DB.MongoP.Coll, map[string]interface{}{"history_name": w})
  500. if len(*ent) > 0 {
  501. id = util.ObjToString((*ent)["company_id"])
  502. redis.PutCKV("qyxy_id", w, id)
  503. }
  504. }
  505. }
  506. if id == "" {
  507. id = "-"
  508. }
  509. cid = append(cid, id)
  510. }
  511. }
  512. return cid
  513. }
  514. // @Description update 修改bidding表,extractM修改抽取表
  515. // @Author J 2022/6/10 10:29 AM
  516. func typeFunc(tmp, update, extractM map[string]interface{}) {
  517. if jyData, ok := tmp["jyfb_data"].(map[string]interface{}); ok {
  518. if t := util.ObjToString(jyData["type"]); t != "" {
  519. switch t {
  520. //case "采购信息":
  521. case "招标公告":
  522. if util.ObjToString(tmp["toptype"]) != "招标" {
  523. update["toptype"] = "招标"
  524. extractM["toptype"] = "招标"
  525. delete(update, "subtype")
  526. }
  527. case "采购意向":
  528. if util.ObjToString(tmp["toptype"]) != "采购意向" {
  529. update["toptype"] = "采购意向"
  530. update["subtype"] = "采购意向"
  531. extractM["toptype"] = "采购意向"
  532. extractM["subtype"] = "采购意向"
  533. }
  534. case "招标预告":
  535. if util.ObjToString(tmp["toptype"]) != "预告" {
  536. update["toptype"] = "预告"
  537. extractM["toptype"] = "预告"
  538. delete(update, "subtype")
  539. }
  540. case "招标结果":
  541. if util.ObjToString(tmp["toptype"]) != "结果" {
  542. update["toptype"] = "结果"
  543. extractM["toptype"] = "结果"
  544. delete(update, "subtype")
  545. }
  546. }
  547. }
  548. }
  549. }
  550. // @Description 附件有效字段(isValidFile)
  551. // @Author J 2022/7/8 14:41
  552. func validFile(tmp map[string]interface{}) int {
  553. isContinue := false
  554. if pinfo, o := tmp["projectinfo"].(map[string]interface{}); o {
  555. if atts, o1 := pinfo["attachments"].(map[string]interface{}); o1 {
  556. for _, att := range atts {
  557. if att == nil {
  558. continue
  559. }
  560. if reflect.TypeOf(att).String() == "string" {
  561. continue
  562. }
  563. att1 := att.(map[string]interface{})
  564. if fid := util.ObjToString(att1["fid"]); fid != "" {
  565. isContinue = true
  566. break
  567. }
  568. }
  569. if isContinue {
  570. if attachTxt, o := tmp["attach_text"].(map[string]interface{}); o {
  571. if len(attachTxt) > 0 {
  572. for _, at := range attachTxt {
  573. at1 := at.(map[string]interface{})
  574. if len(at1) > 0 {
  575. for k, _ := range at1 {
  576. if reflect.TypeOf(at1[k]).String() == "string" {
  577. continue
  578. }
  579. at2 := at1[k].(map[string]interface{})
  580. s := strings.ToLower(util.ObjToString(at2["file_name"]))
  581. if !strings.Contains(s, "jpg") || !strings.Contains(s, "jpeg") != strings.Contains(s, "png") ||
  582. strings.Contains(s, "pdf") {
  583. if strings.Contains(s, "swf") || strings.Contains(s, "html") {
  584. return -1
  585. } else if AnalysisFile(oss.OssGetObject(util.ObjToString(at2["attach_url"]))) {
  586. return 1
  587. }
  588. }
  589. }
  590. break
  591. } else {
  592. break
  593. }
  594. }
  595. }
  596. }
  597. flag := false
  598. for _, att := range atts {
  599. if att == nil {
  600. continue
  601. }
  602. if reflect.TypeOf(att).String() == "string" {
  603. continue
  604. }
  605. att1 := att.(map[string]interface{})
  606. if fid := util.ObjToString(att1["fid"]); fid != "" {
  607. ftype := strings.ToLower(util.ObjToString(tmp["ftype"]))
  608. if ftype != "swf" && ftype != "html" && oss.OssObjExists("jy-datafile", fid) {
  609. return 1
  610. } else {
  611. flag = true
  612. }
  613. }
  614. }
  615. if flag {
  616. return -1
  617. }
  618. }
  619. }
  620. }
  621. return 0
  622. }
  623. // @Description id不变,内容变化 重新索引数据
  624. // @Author J 2022/8/10 13:29
  625. func taskinfo(id string) {
  626. tmp, _ := MgoB.FindById("bidding", id, nil)
  627. if tmp == nil || len(*tmp) == 0 {
  628. log.Info(fmt.Sprintf("taskinfo bidding id=%s 未查询到数据", id))
  629. return
  630. }
  631. extractM, _ := MgoE.FindById(config.Conf.DB.MongoE.Coll, id, nil)
  632. if extractM == nil || len(*extractM) == 0 {
  633. extractM, _ = MgoE.FindById(config.Conf.DB.MongoE.Coll1, id, nil)
  634. if extractM == nil || len(*extractM) == 0 {
  635. log.Info(fmt.Sprintf("taskinfo extract id=%s 未查询到数据", id))
  636. return
  637. }
  638. }
  639. update := map[string]interface{}{} //要更新的mongo数据
  640. //更新bidding表字段
  641. for _, k := range config.Conf.Serve.FieldS {
  642. v1 := (*extractM)[k] //extract
  643. v2 := (*tmp)[k] //bidding
  644. if v2 == nil && v1 != nil {
  645. update[k] = v1
  646. } else if v2 != nil && v1 != nil {
  647. update[k] = v1
  648. } else if v2 != nil && v1 == nil {
  649. if k == "city" || k == "district" {
  650. update[k] = ""
  651. }
  652. }
  653. }
  654. if util.IntAll((*extractM)["repeat"]) == 1 {
  655. update["extracttype"] = -1
  656. update["dataprocess"] = 7
  657. update["repeat_id"] = (*extractM)["repeat_id"]
  658. } else {
  659. update["extracttype"] = 1
  660. update["dataprocess"] = 8
  661. }
  662. //处理分类
  663. fieldFun(*extractM, update)
  664. extractMap := make(map[string]interface{})
  665. if util.ObjToString((*tmp)["s_winner"]) != "" {
  666. cid := companyFun(*tmp)
  667. if len(cid) > 0 {
  668. update["entidlist"] = cid
  669. extractMap["entidlist"] = cid
  670. }
  671. MgoE.UpdateById(config.Conf.DB.MongoE.Coll, id, map[string]interface{}{"$set": extractMap})
  672. }
  673. // 附件有效字段
  674. if i := validFile(*tmp); i != 0 {
  675. if i == -1 {
  676. update["isValidFile"] = false
  677. } else {
  678. update["isValidFile"] = true
  679. }
  680. }
  681. if len(update) > 0 {
  682. MgoB.UpdateById(config.Conf.DB.MongoB.Coll, id, map[string]interface{}{"$set": update})
  683. }
  684. mapinfo := map[string]interface{}{
  685. "infoid": id,
  686. "stype": "index-by-id",
  687. }
  688. datas, _ := json.Marshal(mapinfo)
  689. var next = &net.UDPAddr{
  690. IP: net.ParseIP(config.Conf.Udp.Next.Addr),
  691. Port: util.IntAll(config.Conf.Udp.Next.Port),
  692. }
  693. log.Info("nsq data over", zap.Any("es", next), zap.String("mapinfo", string(datas)))
  694. _ = UdpClient.WriteUdp(datas, udp.OP_TYPE_DATA, next)
  695. }
  696. var DateTimeSelect = []string{"bidopentime", "bidendtime", "signaturedate", "comeintime"}
  697. // @Description 发布时间处理
  698. // @Author J 2023/5/23 14:32
  699. func methodPb(tmp map[string]interface{}) int64 {
  700. if tmp["ext_publishtime"] != nil {
  701. if newPb := util.Int64All(tmp["ext_publishtime"]); newPb < time.Now().Unix() && newPb > 1420041600 {
  702. return newPb
  703. }
  704. }
  705. for _, d := range DateTimeSelect {
  706. if tmp[d] != nil && util.Int64All(tmp[d]) < time.Now().Unix() {
  707. return util.Int64All(tmp[d])
  708. }
  709. }
  710. return 0
  711. }
  712. // Duplicate
  713. // @Description 去重
  714. // @Author J 2023/5/24 09:53
  715. func Duplicate(a interface{}) (ret []interface{}) {
  716. va := reflect.ValueOf(a)
  717. for i := 0; i < va.Len(); i++ {
  718. if i > 0 && reflect.DeepEqual(va.Index(i-1).Interface(), va.Index(i).Interface()) {
  719. continue
  720. }
  721. ret = append(ret, va.Index(i).Interface())
  722. }
  723. return ret
  724. }