extractudp.go 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254
  1. // extractudp
  2. package extract
  3. import (
  4. "encoding/json"
  5. "fmt"
  6. db "jy/mongodbutil"
  7. ju "jy/util"
  8. mu "mfw/util"
  9. "net"
  10. qu "qfw/util"
  11. "sync"
  12. "time"
  13. log "github.com/donnie4w/go-logger/logger"
  14. "gopkg.in/mgo.v2/bson"
  15. )
  16. var Udpclient mu.UdpClient //udp对象
  17. var nextNodes []map[string]interface{}
  18. //udp通知抽取
  19. func ExtractUdp() {
  20. nextNodes = qu.ObjArrToMapArr(ju.Config["nextNode"].([]interface{}))
  21. Udpclient = mu.UdpClient{Local: ":" + qu.ObjToString(ju.Config["udpport"]), BufSize: 1024}
  22. Udpclient.Listen(processUdpMsg)
  23. }
  24. func processUdpMsg(act byte, data []byte, ra *net.UDPAddr) {
  25. switch act {
  26. case mu.OP_TYPE_DATA:
  27. var rep map[string]interface{}
  28. err := json.Unmarshal(data, &rep)
  29. if err != nil {
  30. log.Debug(err)
  31. } else {
  32. stype, _ := rep["stype"].(string)
  33. if stype == "distributed" { //分布式抽取分支
  34. go Udpclient.WriteUdp([]byte("发送分布式抽取分支"+qu.ObjToString(rep["ip"])+"udpok"), mu.OP_NOOP, ra)
  35. InstanceId := qu.ObjToString(rep["InstanceId"])
  36. db.Mgo.Update("ecs", `{"InstanceId":"`+InstanceId+`"}`,
  37. map[string]interface{}{
  38. "$set": map[string]interface{}{
  39. "extstatus": "running",
  40. },
  41. }, true, false)
  42. ExtractByUdp("", "", ra, qu.ObjToString(rep["InstanceId"]), qu.ObjToString(rep["ip"]))
  43. db.Mgo.Update("ecs", `{"InstanceId":"`+InstanceId+`"}`,
  44. map[string]interface{}{
  45. "$set": map[string]interface{}{
  46. "extstatus": "ok",
  47. },
  48. }, true, false)
  49. log.Debug("分布式抽取完成,可以释放esc实例", qu.ObjToString(rep["ip"]))
  50. } else {
  51. sid, _ := rep["gtid"].(string)
  52. eid, _ := rep["lteid"].(string)
  53. if sid == "" || eid == "" {
  54. log.Debug("err", "sid=", sid, ",eid=", eid)
  55. } else {
  56. udpinfo, _ := rep["key"].(string)
  57. if udpinfo == "" {
  58. udpinfo = "udpok"
  59. }
  60. go Udpclient.WriteUdp([]byte(udpinfo), mu.OP_NOOP, ra)
  61. log.Debug("udp通知抽取id段", sid, " ", eid)
  62. ExtractByUdp(sid, eid, ra)
  63. log.Debug("udp通知抽取完成,eid=", eid)
  64. for _, m := range nextNodes {
  65. by, _ := json.Marshal(map[string]interface{}{
  66. "gtid": sid,
  67. "lteid": eid,
  68. "stype": qu.ObjToString(m["stype"]),
  69. })
  70. err := Udpclient.WriteUdp(by, mu.OP_TYPE_DATA, &net.UDPAddr{
  71. IP: net.ParseIP(m["addr"].(string)),
  72. Port: qu.IntAll(m["port"]),
  73. })
  74. if err != nil {
  75. log.Debug(err)
  76. }
  77. }
  78. }
  79. }
  80. }
  81. case mu.OP_NOOP: //下个节点回应
  82. log.Debug(string(data))
  83. }
  84. }
  85. var ext *ExtractTask
  86. //根据id区间抽取
  87. func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
  88. defer qu.Catch()
  89. if ext == nil {
  90. ext = &ExtractTask{}
  91. ext.Id = qu.ObjToString(ju.Config["udptaskid"])
  92. ext.InitTaskInfo()
  93. ext.TaskInfo.FDB = db.MgoFactory(3, 5, 600, ext.TaskInfo.FromDbAddr, ext.TaskInfo.FromDB)
  94. ext.TaskInfo.TDB = db.MgoFactory(3, 5, 600, ext.TaskInfo.ToDbAddr, ext.TaskInfo.ToDB)
  95. ext.InitSite()
  96. ext.InitRulePres()
  97. ext.InitRuleBacks(false)
  98. ext.InitRuleBacks(true)
  99. ext.InitRuleCore(false)
  100. ext.InitRuleCore(true)
  101. ext.InitBlockRule()
  102. ext.InitPkgCore()
  103. ext.InitTag(false)
  104. ext.InitTag(true)
  105. ext.InitClearFn(false)
  106. ext.InitClearFn(true)
  107. if ext.IsExtractCity { //版本上控制是否开始城市抽取
  108. //初始化城市DFA信息
  109. //ext.InitCityDFA()
  110. ext.InitCityInfo()
  111. ext.InitAreaCode()
  112. ext.InitPostCode()
  113. }
  114. //质量审核
  115. ext.InitAuditFields()
  116. ext.InitAuditRule()
  117. ext.InitAuditClass()
  118. ext.InitAuditRecogField()
  119. //品牌抽取是否开启
  120. ju.IsBrandGoods, _ = ju.Config["brandgoods"].(bool)
  121. ext.ResultSave(true)
  122. ext.BidSave(true)
  123. ext.IsRun = true
  124. } else {
  125. ext.BidTotal = 0
  126. }
  127. index := 0
  128. if len(instanceId) > 0 { //分布式抽取进度
  129. go Udpclient.WriteUdp([]byte(fmt.Sprintf("IP=%s,数据接收成功", instanceId[1])), mu.OP_NOOP, ra)
  130. for {
  131. tsk, b := db.Mgo.FindOne("esctask", `{"state":{"$lt":1}}`)
  132. if tsk != nil && !b {
  133. break
  134. }
  135. db.Mgo.UpdateById("esctask", (*tsk)["_id"], map[string]interface{}{
  136. "$set": map[string]interface{}{
  137. "InstanceId": instanceId[0],
  138. "state": 1,
  139. "runtime": time.Now().Format(qu.Date_Full_Layout),
  140. },
  141. })
  142. query := bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(qu.ObjToString((*tsk)["sid"])), "$lte": bson.ObjectIdHex(qu.ObjToString((*tsk)["eid"]))}}
  143. count1 := ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl, query)
  144. count2 := ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl+"_back", query)
  145. log.Debug("timestr", (*tsk)["timestr"], "count", count1+count2)
  146. list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, -1, -1)
  147. for _, v := range *list {
  148. if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
  149. continue
  150. }
  151. var j, jf *ju.Job
  152. var isSite bool
  153. if ext.IsFileField && v["projectinfo"] != nil {
  154. v["isextFile"] = true
  155. j, jf,isSite = ext.PreInfo(v)
  156. } else {
  157. j, _,isSite = ext.PreInfo(v)
  158. }
  159. go ext.ExtractProcess(j, jf,isSite)
  160. index++
  161. ext.TaskInfo.ProcessPool <- true
  162. }
  163. list2, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl+"_back", query, nil, Fields, false, -1, -1)
  164. for _, v := range *list2 {
  165. if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
  166. continue
  167. }
  168. var j, jf *ju.Job
  169. var isSite bool
  170. if ext.IsFileField && v["projectinfo"] != nil {
  171. v["isextFile"] = true
  172. j, jf,isSite = ext.PreInfo(v)
  173. } else {
  174. j, _,isSite = ext.PreInfo(v)
  175. }
  176. go ext.ExtractProcess(j, jf,isSite)
  177. index++
  178. ext.TaskInfo.ProcessPool <- true
  179. }
  180. db.Mgo.UpdateById("esctask", (*tsk)["_id"], map[string]interface{}{
  181. "$set": map[string]interface{}{
  182. "InstanceId": instanceId[0],
  183. "oktime": time.Now().Format(qu.Date_Full_Layout),
  184. "state": 1,
  185. },
  186. })
  187. db.Mgo.Update("ecs", `{"InstanceId":"`+instanceId[0]+`"}`,
  188. map[string]interface{}{
  189. "$inc": map[string]interface{}{
  190. "totalnum": count1 + count2,
  191. "step": 1,
  192. },
  193. }, true, false)
  194. }
  195. log.Debug("抽取完成", "index:", index, "bidtotal:", ext.BidTotal)
  196. } else { //普通抽取
  197. query := bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(sid), "$lte": bson.ObjectIdHex(eid)}}
  198. count := ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl, query)
  199. log.Debug("查询条件为:", query, "查询条数:", count)
  200. pageNum := (count + PageSize - 1) / PageSize
  201. limit := PageSize
  202. if count < PageSize {
  203. limit = count
  204. }
  205. wg := sync.WaitGroup{}
  206. for i := 0; i < pageNum; i++ {
  207. query = bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(sid)}}
  208. fmt.Printf("page=%d,query=%v\n", i+1, query)
  209. list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
  210. for _, v := range *list {
  211. if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
  212. log.Debug(index, qu.BsonIdToSId(v["_id"]), "//去除含敏感词数据")
  213. continue
  214. }
  215. _id := qu.BsonIdToSId(v["_id"])
  216. var j, jf *ju.Job
  217. var isSite bool
  218. if ext.IsFileField && v["projectinfo"] != nil {
  219. v["isextFile"] = true
  220. j, jf,isSite = ext.PreInfo(v)
  221. } else {
  222. j, _,isSite = ext.PreInfo(v)
  223. }
  224. ext.TaskInfo.ProcessPool <- true
  225. wg.Add(1)
  226. go func(wg *sync.WaitGroup, j, jf *ju.Job) {
  227. defer wg.Done()
  228. //log.Debug(index,j.SourceMid,)
  229. ext.ExtractProcess(j, jf,isSite)
  230. }(&wg, j, jf)
  231. index++
  232. if index%1000 == 0 {
  233. log.Debug("index:", index, ",页码:", i+1, ",_id:", _id)
  234. }
  235. sid = _id
  236. if sid >= eid {
  237. break
  238. }
  239. }
  240. }
  241. wg.Wait()
  242. ext.BidSave(false)
  243. log.Debug("抽取完成,", "count:", count, ",index:", index, ",bidtotal:", ext.BidTotal, ",eid:", eid)
  244. }
  245. }