task.go 8.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328
  1. package main
  2. import (
  3. "fmt"
  4. "github.com/cron"
  5. "go.mongodb.org/mongo-driver/bson"
  6. "go.mongodb.org/mongo-driver/bson/primitive"
  7. "log"
  8. "mongodb"
  9. "qfw/util"
  10. "regexp"
  11. "strings"
  12. "time"
  13. )
  14. func TimeTask() {
  15. //GetPaData()
  16. c := cron.New()
  17. cronstrBd := "0 0 */" + fmt.Sprint(BdTaskTime) + " * * ?" //每TaskTime小时执行一次
  18. //cronstr := "0 0 " + fmt.Sprint(TaskTime) + " * * ?" //每天TaskTime跑一次
  19. cronstrPa := "0 0 15 ? * " + fmt.Sprint(PaTaskTime) //凭安增量数据每周二跑一次
  20. _ = c.AddFunc(cronstrBd, func() { GetBdData() })
  21. _ = c.AddFunc(cronstrPa, func() { GetPaData() })
  22. c.Start()
  23. }
  24. func GetBdData() {
  25. count := 0
  26. lastid := ""
  27. sess := MgoBd.GetMgoConn()
  28. defer MgoBd.DestoryMongoConn(sess)
  29. fields := map[string]interface{}{"data": 1, "down_time": 1}
  30. q := bson.M{"down_time": bson.M{"$gt": LastTime}}
  31. query := sess.DB(Dbname_bd).C(CollBd).Find(q).Select(fields).Iter()
  32. tmp := make(map[string]interface{})
  33. for query.Next(&tmp) {
  34. lastid = mongodb.BsonIdToSId(tmp["_id"])
  35. if count%1000 == 0 {
  36. util.Debug("baidu ----current----", count, lastid)
  37. }
  38. findEnt(tmp)
  39. count++
  40. }
  41. util.Debug("baidu 处理", count, "条数据")
  42. }
  43. func GetPaData() {
  44. count := 0
  45. lastid := ""
  46. sess := MgoMix.GetMgoConn()
  47. defer MgoMix.DestoryMongoConn(sess)
  48. fields := map[string]interface{}{"changes": 1, "company_id": 1, "company_name": 1, "company_type": 1, "establish_date": 1, "create_time": 1}
  49. query := sess.DB(Dbname_pa).C(CollPa).Find(nil).Select(fields).Iter()
  50. c := MgoMix.Count(CollPa, nil)
  51. util.Debug("ping an count ------", c)
  52. tmp := make(map[string]interface{})
  53. for query.Next(&tmp) {
  54. lastid = mongodb.BsonIdToSId(tmp["company_id"])
  55. if count%1000 == 0 {
  56. util.Debug("ping an ----current-----", count, lastid)
  57. }
  58. if strings.Contains(util.ObjToString(tmp["company_type"]), "个体") {
  59. continue
  60. }
  61. currentTime := time.Now().Unix()
  62. if tmp["changes"] != nil && len(tmp["changes"].([]interface{})) > 0 {
  63. delete(tmp, "establish_date")
  64. q := bson.M{"company_name": tmp["company_name"]}
  65. changeEnt, _ := MgoMix.FindOne(CollSave, q)
  66. if changeEnt != nil && len(*changeEnt) > 0 {
  67. tmpList := tmp["changes"].([]interface{})
  68. changeList := clearRepeat((*changeEnt)["changes"].([]interface{}))
  69. if len(tmpList) > len(changeList) {
  70. infoList := clearRepeat(tmp["changes"].([]interface{}))
  71. for _, item := range infoList {
  72. item1 := item.(map[string]interface{})
  73. setMark(item1)
  74. }
  75. tmp["changes"] = infoList
  76. tmp["updatetime"] = currentTime
  77. }
  78. }else {
  79. infoList := clearRepeat(tmp["changes"].([]interface{}))
  80. for _, item := range infoList {
  81. item1 := item.(map[string]interface{})
  82. setMark(item1)
  83. }
  84. tmp["_id"] = primitive.NewObjectID()
  85. tmp["createtime"] = currentTime
  86. tmp["updatetime"] = currentTime
  87. }
  88. update := make(map[string]interface{})
  89. tmp["datasource"] = "pingan"
  90. update["$set"] = tmp
  91. updateInfo := []map[string]interface{}{
  92. {
  93. "_id": tmp["_id"],
  94. },
  95. update,
  96. }
  97. MgoSaveCache <- updateInfo
  98. count++
  99. }else {
  100. //{
  101. // "change_code": "100000",
  102. // "change_name": "新设立公司",
  103. // "change_push": true,
  104. // "change_info": "新设立公司",
  105. // "change_keyword": ["新设立"]
  106. //},
  107. setupData := ""
  108. if tmp["establish_date"] != nil {
  109. if timeTmp, ok := tmp["establish_date"].(primitive.DateTime); ok {
  110. t := timeTmp.Time()
  111. setupData = util.FormatDate(&t, util.Date_Short_Layout)
  112. } else if timeTmp, ok := tmp["establish_date"].(string); ok && timeTmp != "" {
  113. t := timeReg.FindString(timeTmp)
  114. if t != "" {
  115. setupData = t
  116. }
  117. }
  118. }
  119. createData := ""
  120. if tmp["create_time"] != nil {
  121. if timeTmp, ok := tmp["create_time"].(primitive.DateTime); ok {
  122. t := timeTmp.Time()
  123. createData = util.FormatDate(&t, util.Date_Short_Layout)
  124. } else if timeTmp, ok := tmp["create_time"].(string); ok && timeTmp != "" {
  125. t := timeReg.FindString(timeTmp)
  126. if t != "" {
  127. createData = t
  128. }
  129. }
  130. }
  131. tm2, _ := time.Parse("2006-01-02", createData)
  132. //当前时间17天内
  133. if tm2.Unix() < (time.Now().Unix() - 17 * 60 * 60 * 24) {
  134. continue
  135. }
  136. delete(tmp, "establish_date")
  137. delete(tmp, "create_time")
  138. changeInfo := make(map[string]interface{})
  139. changeInfo["change_field"] = "新设立公司"
  140. changeInfo["change_name_new"] = "新设立公司"
  141. changeInfo["content_before"] = ""
  142. changeInfo["content_after"] = "新设立公司"
  143. changeInfo["change_date"] = setupData
  144. tmp["changes"] = []map[string]interface{}{changeInfo}
  145. tmp["_id"] = primitive.NewObjectID()
  146. tmp["createtime"] = currentTime
  147. tmp["updatetime"] = currentTime
  148. tmp["datasource"] = "pingan"
  149. update := make(map[string]interface{})
  150. update["$set"] = tmp
  151. updateInfo := []map[string]interface{}{
  152. {
  153. "_id": tmp["_id"],
  154. },
  155. update,
  156. }
  157. MgoSaveCache <- updateInfo
  158. count++
  159. }
  160. }
  161. util.Debug("pingan 处理", count, "条数据")
  162. }
  163. func findEnt(tmp map[string]interface{}) {
  164. if LastTime < util.Int64All(tmp["down_time"]) {
  165. LastTime = util.Int64All(tmp["down_time"])
  166. }
  167. data := util.ObjToMap(tmp["data"])
  168. ent := util.ObjToMap((*data)["basicData"])
  169. changeData := util.ObjToMap((*data)["changeRecordData"])
  170. infoList := (*changeData)["list"].([]interface{})
  171. currentTime := time.Now().Unix()
  172. q := bson.M{"company_name": (*ent)["entName"]}
  173. changeEnt, _ := MgoMix.FindOne(CollSave, q)
  174. update := map[string]interface{}{}
  175. if changeEnt != nil && len(*changeEnt) > 0 {
  176. //1、企业变更库有该企业信息
  177. if (*changeEnt)["changes"] != nil{
  178. (*changeEnt)["updatetime"] = currentTime
  179. if len(infoList) > len((*changeEnt)["changes"].([]interface{})) {
  180. mapArr := setChangeInfo(infoList)
  181. for _, v := range mapArr{
  182. setMark(v)
  183. }
  184. (*changeEnt)["changes"] = mapArr
  185. }
  186. update["$set"] = *changeEnt
  187. updateInfo := []map[string]interface{}{
  188. {
  189. "_id": (*changeEnt)["_id"],
  190. },
  191. update,
  192. }
  193. MgoSaveCache <- updateInfo
  194. }
  195. } else {
  196. //2、企业变更库没有该企业信息
  197. paEnt, _ := MgoMix.FindOne(CollQy, q)
  198. saveEnt := map[string]interface{}{}
  199. if saveEnt != nil && len(*paEnt) > 0 {
  200. //3、企业库有该企业信息
  201. saveEnt["datasource"] = "baidu"
  202. saveEnt["_id"] = primitive.NewObjectID()
  203. saveEnt["company_id"] = (*paEnt)["company_id"]
  204. saveEnt["company_name"] = (*ent)["entName"]
  205. saveEnt["createtime"] = currentTime
  206. saveEnt["updatetime"] = currentTime
  207. if (*paEnt)["changes"] != nil{
  208. changeArr := (*paEnt)["changes"].([]interface{})
  209. mapArr := setChangeInfo(infoList)
  210. for _, v := range util.ObjArrToMapArr(changeArr){
  211. setMark(v)
  212. mapArr = append(mapArr, v)
  213. }
  214. saveEnt["changes"] = mapArr
  215. }else {
  216. saveEnt["changes"] = setChangeInfo(infoList)
  217. }
  218. update["$set"] = saveEnt
  219. updateInfo := []map[string]interface{}{
  220. {
  221. "_id": saveEnt["_id"],
  222. },
  223. update,
  224. }
  225. MgoSaveCache <- updateInfo
  226. } else {
  227. //4、企业库没有该企业信息
  228. saveEnt["company_name"] = (*ent)["entName"]
  229. saveEnt["createtime"] = currentTime
  230. saveEnt["changes"] = setChangeInfo(infoList)
  231. MgoMix.Save(CollBack, saveEnt)
  232. }
  233. }
  234. }
  235. func setChangeInfo(list []interface{}) []map[string]interface{} {
  236. var arr []map[string]interface{}
  237. for _, item := range list {
  238. tmp := map[string]interface{}{}
  239. item1 := item.(map[string]interface{})
  240. tmp["change_date"] = item1["date"]
  241. tmp["content_before"] = item1["oldValue"]
  242. tmp["content_after"] = item1["newValue"]
  243. tmp["change_field"] = item1["fieldName"]
  244. setMark(tmp)
  245. arr = append(arr, tmp)
  246. }
  247. return arr
  248. }
  249. func setMark(tmp map[string]interface{}) {
  250. for _, v := range ChangeMap {
  251. str := util.ObjToString(tmp["change_field"])
  252. regArr := v["change_key_reg"].([]string)
  253. for _, v1 := range regArr {
  254. matched, _ := regexp.MatchString(v1, str)
  255. if matched {
  256. tmp["change_name_new"] = v["change_name"]
  257. return
  258. }
  259. }
  260. }
  261. }
  262. func clearRepeat(list []interface{}) []interface{} {
  263. var tmp []interface{}
  264. if len(list) > 1 {
  265. for k, v := range list{
  266. if k < len(list)-1 {
  267. if fmt.Sprint(list[k]) != fmt.Sprint(list[k+1]) {
  268. tmp = append(tmp, v)
  269. }
  270. }else {
  271. tmp = append(tmp, v)
  272. }
  273. }
  274. return tmp
  275. }else {
  276. return list
  277. }
  278. }
  279. var MgoSaveCache = make(chan []map[string]interface{}, 2000)
  280. var SP = make(chan bool, 5)
  281. func SaveData() {
  282. log.Println("Mgo Save...")
  283. arru := make([][]map[string]interface{}, 200)
  284. indexu := 0
  285. for {
  286. select {
  287. case v := <-MgoSaveCache:
  288. arru[indexu] = v
  289. indexu++
  290. if indexu == 200 {
  291. SP <- true
  292. go func(arru [][]map[string]interface{}) {
  293. defer func() {
  294. <-SP
  295. }()
  296. MgoMix.UpSertBulk(CollSave, arru...)
  297. }(arru)
  298. arru = make([][]map[string]interface{}, 200)
  299. indexu = 0
  300. }
  301. case <-time.After(1000 * time.Millisecond):
  302. if indexu > 0 {
  303. SP <- true
  304. go func(arru [][]map[string]interface{}) {
  305. defer func() {
  306. <-SP
  307. }()
  308. MgoMix.UpSertBulk(CollSave, arru...)
  309. }(arru[:indexu])
  310. arru = make([][]map[string]interface{}, 200)
  311. indexu = 0
  312. }
  313. }
  314. }
  315. }