task.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419
  1. package main
  2. import (
  3. "fmt"
  4. "github.com/cron"
  5. "go.mongodb.org/mongo-driver/bson"
  6. "log"
  7. "qfw/util"
  8. "strconv"
  9. "strings"
  10. "sync"
  11. "time"
  12. )
  13. var (
  14. partner = []string{"identify_no", "stock_type", "stock_name", "identify_type", "stock_capital", "stock_realcapital"}
  15. employee = []string{"employee_name", "position"}
  16. TypeMap = map[string]string{
  17. "采购单位": "1",
  18. "投标企业": "2",
  19. "代理机构": "3",
  20. "厂商": "4",
  21. }
  22. TypeMap1 = map[string]string{
  23. "固定电话": "1",
  24. "手机号": "2",
  25. "邮箱": "3",
  26. "不存在": "4",
  27. }
  28. )
  29. //定时任务
  30. func TimeTask() {
  31. c := cron.New()
  32. cronstr := "0 0 15 ? * Tue" //每周二15点执行
  33. //cronstr := "0 */" + fmt.Sprint(TaskTime) + " * * * ?" //每TaskTime小时执行一次
  34. err := c.AddFunc(cronstr, func() { StdAdd() })
  35. if err != nil {
  36. util.Debug(err)
  37. return
  38. }
  39. c.Start()
  40. }
  41. // StdAdd 增量数据
  42. func StdAdd() {
  43. defer util.Catch()
  44. sess := Mgo.GetMgoConn()
  45. defer Mgo.DestoryMongoConn(sess)
  46. pool := make(chan bool, 5)
  47. wg := &sync.WaitGroup{}
  48. //q := bson.M{"_id": "affe29f8d061f3faa4170cafba41f316"}
  49. q := bson.M{"updatetime": bson.M{"$gt": Updatetime}}
  50. util.Debug(q)
  51. it := sess.DB(Dbname).C(Dbcoll).Find(q).Iter()
  52. count := 0
  53. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  54. if count%5000 == 0 {
  55. log.Println("current:", count)
  56. }
  57. if util.IntAll(tmp["use_flag"]) > 5 {
  58. continue
  59. }
  60. pool <- true
  61. wg.Add(1)
  62. go func(tmp map[string]interface{}) {
  63. defer func() {
  64. <-pool
  65. wg.Done()
  66. }()
  67. esMap := map[string]interface{}{}
  68. //生索引字段处理
  69. for _, field := range EsFields {
  70. if tmp[field] == nil {
  71. continue
  72. }
  73. if field == "company_name" {
  74. esMap[field] = tmp["company_name"]
  75. esMap["name"] = tmp["company_name"]
  76. } else if field == "history_name" {
  77. var nameArr []string
  78. names := util.ObjToString(tmp["history_name"])
  79. if strings.Contains(names, ",") {
  80. nameArr = append(nameArr, strings.Split(names, ",")...)
  81. }
  82. if len(nameArr) > 0 {
  83. esMap["history_name"] = nameArr
  84. }
  85. } else if field == "establish_date" {
  86. // 成立日期修改成时间戳
  87. location, err := time.ParseInLocation(util.Date_Short_Layout, util.ObjToString(tmp["establish_date"]), time.Local)
  88. if err != nil {
  89. util.Debug(err)
  90. } else {
  91. esMap["establish_date"] = location.Unix()
  92. }
  93. } else if field == "lastupdatetime" {
  94. esMap["lastupdatetime"] = tmp["update_time_msql"]
  95. } else if field == "bid_projectname" {
  96. if pname, ok := tmp["bid_projectname"].([]interface{}); ok {
  97. p1 := util.ObjArrToStringArr(pname)
  98. esMap["bid_projectname"] = strings.Join(p1, ",")
  99. }
  100. } else if field == "bid_purchasing" {
  101. if pur, ok := tmp["bid_purchasing"].([]interface{}); ok {
  102. p1 := util.ObjArrToStringArr(pur)
  103. esMap["bid_purchasing"] = strings.Join(p1, ",")
  104. }
  105. } else if field == "bid_area" {
  106. if areas, ok := tmp["bid_area"].([]interface{}); ok {
  107. p1 := util.ObjArrToStringArr(areas)
  108. esMap["bid_area"] = strings.Join(p1, ",")
  109. }
  110. } else if field == "partners" {
  111. if ps, ok := tmp["partners"].([]interface{}); ok {
  112. var parr []map[string]interface{}
  113. for _, v := range ps {
  114. p := make(map[string]interface{})
  115. v1 := v.(map[string]interface{})
  116. for _, field := range partner {
  117. if v1[field] == nil {
  118. continue
  119. }
  120. if field == "stock_capital" || field == "stock_realcapital" {
  121. if v, err := strconv.ParseFloat(util.ObjToString(v1[field]), 64); err == nil {
  122. v, _ = strconv.ParseFloat(fmt.Sprintf("%.2f", v), 64) //保留小数点两位
  123. p[field] = v
  124. }
  125. } else {
  126. p[field] = v1[field]
  127. }
  128. }
  129. if len(p) > 0 {
  130. parr = append(parr, p)
  131. }
  132. }
  133. if len(parr) > 0 {
  134. esMap[field] = parr
  135. }
  136. }
  137. } else if field == "employees" {
  138. if ps, ok := tmp["employees"].([]interface{}); ok {
  139. var parr []map[string]interface{}
  140. for _, v := range ps {
  141. p := make(map[string]interface{})
  142. v1 := v.(map[string]interface{})
  143. for _, field := range employee {
  144. if v1[field] == nil {
  145. continue
  146. } else {
  147. p[field] = v1[field]
  148. }
  149. }
  150. if len(p) > 0 {
  151. parr = append(parr, p)
  152. }
  153. }
  154. if len(parr) > 0 {
  155. esMap[field] = parr
  156. }
  157. }
  158. } else if field == "bid_unittype" {
  159. if t2, ok := tmp["bid_unittype"].([]interface{}); ok {
  160. var arr []string
  161. for _, v := range util.ObjArrToStringArr(t2) {
  162. arr = append(arr, TypeMap[v])
  163. }
  164. esMap["bid_unittype"] = strings.Join(arr, ",")
  165. }
  166. } else if field == "bid_contracttype" {
  167. if t2, ok := tmp["bid_contracttype"].([]interface{}); ok {
  168. var arr []string
  169. for _, v := range util.ObjArrToStringArr(t2) {
  170. arr = append(arr, TypeMap1[v])
  171. }
  172. esMap["bid_contracttype"] = strings.Join(arr, ",")
  173. }
  174. } else {
  175. esMap[field] = tmp[field]
  176. }
  177. }
  178. company_type := util.ObjToString(tmp["company_type"])
  179. company_name := util.ObjToString(tmp["company_name"])
  180. if company_type == "个体工商户" {
  181. if len([]rune(company_name)) >= 5 {
  182. esMap["company_type_int"] = 31
  183. } else {
  184. esMap["company_type_int"] = 32
  185. }
  186. } else if company_type == "其他" || company_type == "" {
  187. if len([]rune(company_name)) >= 4 {
  188. esMap["company_type_int"] = 21
  189. } else {
  190. esMap["company_type_int"] = 22
  191. }
  192. } else {
  193. if company_type == "内资分公司" {
  194. esMap["company_type_int"] = 12
  195. } else if len([]rune(company_name)) >= 4 {
  196. esMap["company_type_int"] = 11
  197. } else {
  198. esMap["company_type_int"] = 13
  199. }
  200. }
  201. EsSaveCache <- esMap // 保存es
  202. }(tmp)
  203. tmp = make(map[string]interface{})
  204. }
  205. wg.Wait()
  206. log.Println("Run Over...Count:", count)
  207. }
  208. // StdAll 存量数据生es
  209. func StdAll() {
  210. defer util.Catch()
  211. sess := Mgo.GetMgoConn()
  212. defer Mgo.DestoryMongoConn(sess)
  213. pool := make(chan bool, 10)
  214. wg := &sync.WaitGroup{}
  215. //q := bson.M{"_id": "f9ad04e5529023e8af0b2ad8b49bf227"}
  216. it := sess.DB(Dbname).C(Dbcoll).Find(nil).Iter()
  217. count := 0
  218. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  219. if count%20000 == 0 {
  220. log.Println("current:", count, tmp["_id"])
  221. }
  222. pool <- true
  223. wg.Add(1)
  224. go func(tmp map[string]interface{}) {
  225. defer func() {
  226. <-pool
  227. wg.Done()
  228. }()
  229. esMap := map[string]interface{}{}
  230. //生索引字段处理
  231. for _, field := range EsFields {
  232. if tmp[field] == nil {
  233. continue
  234. }
  235. if field == "company_name" {
  236. esMap[field] = tmp["company_name"]
  237. esMap["name"] = tmp["company_name"]
  238. } else if field == "history_name" {
  239. var nameArr []string
  240. for _, v := range strings.Split(util.ObjToString(tmp["history_name"]), ";") {
  241. if v != "" {
  242. nameArr = append(nameArr, v)
  243. }
  244. }
  245. if len(nameArr) > 0 {
  246. esMap["history_name"] = nameArr
  247. }
  248. } else if field == "establish_date" {
  249. // 成立日期修改成时间戳
  250. location, err := time.ParseInLocation(util.Date_Short_Layout, util.ObjToString(tmp["establish_date"]), time.Local)
  251. if err != nil {
  252. util.Debug(err)
  253. } else {
  254. esMap["establish_date"] = location.Unix()
  255. }
  256. } else if field == "lastupdatetime" {
  257. esMap["lastupdatetime"] = tmp["update_time_msql"]
  258. } else if field == "bid_projectname" {
  259. if pname, ok := tmp["bid_projectname"].([]interface{}); ok {
  260. p1 := util.ObjArrToStringArr(pname)
  261. esMap["bid_projectname"] = strings.Join(p1, ",")
  262. }
  263. } else if field == "bid_purchasing" {
  264. if pur, ok := tmp["bid_purchasing"].([]interface{}); ok {
  265. p1 := util.ObjArrToStringArr(pur)
  266. esMap["bid_purchasing"] = strings.Join(p1, ",")
  267. }
  268. } else if field == "bid_area" {
  269. if areas, ok := tmp["bid_area"].([]interface{}); ok {
  270. p1 := util.ObjArrToStringArr(areas)
  271. esMap["bid_area"] = strings.Join(p1, ",")
  272. }
  273. } else if field == "partners" {
  274. if ps, ok := tmp["partners"].([]interface{}); ok {
  275. var parr []map[string]interface{}
  276. for _, v := range ps {
  277. p := make(map[string]interface{})
  278. v1 := v.(map[string]interface{})
  279. for _, field := range partner {
  280. if v1[field] == nil {
  281. continue
  282. }
  283. if field == "stock_capital" || field == "stock_realcapital" {
  284. text := util.ObjToString(v1[field])
  285. if strings.Contains(text, "万元") {
  286. text = strings.Replace(text, "万元", "", -1)
  287. }
  288. if v, err := strconv.ParseFloat(text, 64); err == nil {
  289. v, _ = strconv.ParseFloat(fmt.Sprintf("%.2f", v), 64) //保留小数点两位
  290. p[field] = v
  291. }
  292. } else {
  293. p[field] = v1[field]
  294. }
  295. }
  296. if len(p) > 0 {
  297. parr = append(parr, p)
  298. }
  299. }
  300. if len(parr) > 0 {
  301. esMap[field] = parr
  302. }
  303. }
  304. } else if field == "employees" {
  305. if ps, ok := tmp["employees"].([]interface{}); ok {
  306. var parr []map[string]interface{}
  307. for _, v := range ps {
  308. p := make(map[string]interface{})
  309. v1 := v.(map[string]interface{})
  310. for _, field := range employee {
  311. if v1[field] == nil {
  312. continue
  313. } else {
  314. p[field] = v1[field]
  315. }
  316. }
  317. if len(p) > 0 {
  318. parr = append(parr, p)
  319. }
  320. }
  321. if len(parr) > 0 {
  322. esMap[field] = parr
  323. }
  324. }
  325. } else if field == "bid_unittype" {
  326. if t2, ok := tmp["bid_unittype"].([]interface{}); ok {
  327. var arr []string
  328. for _, v := range util.ObjArrToStringArr(t2) {
  329. arr = append(arr, TypeMap[v])
  330. }
  331. esMap["bid_unittype"] = strings.Join(arr, ",")
  332. }
  333. } else if field == "bid_contracttype" {
  334. if t2, ok := tmp["bid_contracttype"].([]interface{}); ok {
  335. var arr []string
  336. for _, v := range util.ObjArrToStringArr(t2) {
  337. arr = append(arr, TypeMap1[v])
  338. }
  339. esMap["bid_contracttype"] = strings.Join(arr, ",")
  340. }
  341. } else {
  342. esMap[field] = tmp[field]
  343. }
  344. }
  345. company_type := util.ObjToString(tmp["company_type"])
  346. company_name := util.ObjToString(tmp["company_name"])
  347. if company_type == "个体工商户" {
  348. if len([]rune(company_name)) >= 5 {
  349. esMap["company_type_int"] = 31
  350. } else {
  351. esMap["company_type_int"] = 32
  352. }
  353. } else if company_type == "其他" || company_type == "" {
  354. if len([]rune(company_name)) >= 4 {
  355. esMap["company_type_int"] = 21
  356. } else {
  357. esMap["company_type_int"] = 22
  358. }
  359. } else {
  360. if company_type == "内资分公司" {
  361. esMap["company_type_int"] = 12
  362. } else if len([]rune(company_name)) >= 4 {
  363. esMap["company_type_int"] = 11
  364. } else {
  365. esMap["company_type_int"] = 13
  366. }
  367. }
  368. EsSaveCache <- esMap // 保存es
  369. }(tmp)
  370. tmp = make(map[string]interface{})
  371. }
  372. wg.Wait()
  373. log.Println("Run Over...Count:", count)
  374. }
  375. // SaveEs 过滤后数据存库
  376. func SaveEs() {
  377. log.Println("Es Save...")
  378. arru := make([]map[string]interface{}, 100)
  379. indexu := 0
  380. for {
  381. select {
  382. case v := <-EsSaveCache:
  383. arru[indexu] = v
  384. indexu++
  385. if indexu == 100 {
  386. SP <- true
  387. go func(arru []map[string]interface{}) {
  388. defer func() {
  389. <-SP
  390. }()
  391. Es.BulkSave(Index, Itype, &arru, true)
  392. }(arru)
  393. arru = make([]map[string]interface{}, 100)
  394. indexu = 0
  395. }
  396. case <-time.After(1000 * time.Millisecond):
  397. if indexu > 0 {
  398. SP <- true
  399. go func(arru []map[string]interface{}) {
  400. defer func() {
  401. <-SP
  402. }()
  403. Es.BulkSave(Index, Itype, &arru, true)
  404. }(arru[:indexu])
  405. arru = make([]map[string]interface{}, 100)
  406. indexu = 0
  407. }
  408. }
  409. }
  410. }