load_data.go 4.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209
  1. package main
  2. import (
  3. "encoding/json"
  4. "log"
  5. "qfw/util"
  6. "qfw/util/redis"
  7. "time"
  8. )
  9. // 初始加载数据,默认加载最近6个月的数据
  10. func (p *ProjectTask) loadData(starttime int64) {
  11. log.Println("load project start..", starttime)
  12. p.findLock.Lock()
  13. defer p.findLock.Unlock()
  14. p.AllIdsMapLock.Lock()
  15. defer p.AllIdsMapLock.Unlock()
  16. sess := MongoTool.GetMgoConn()
  17. defer MongoTool.DestoryMongoConn(sess)
  18. loadOver := make(chan bool)
  19. q := map[string]interface{}{
  20. "lasttime": map[string]interface{}{"$gte": starttime},
  21. }
  22. //field := map[string]interface{}{"list": 0}
  23. it := sess.DB(MongoTool.DbName).C(p.coll).Find(&q).Select(nil).Iter()
  24. n := 0
  25. pool := make(chan *ProjectCache, 1000)
  26. go func() {
  27. for {
  28. select {
  29. case tmp := <-pool:
  30. n++
  31. if n%10000 == 0 {
  32. log.Println("current", n, "\n", tmp.Id, len(p.mapPn), len(p.mapPc), len(p.mapPb), len(p.mapHref)) //, tmp.ProjectName, tmp.MPN, tmp.ProjectCode, tmp.MPC, tmp.Buyer)
  33. }
  34. if tmp != nil {
  35. id := tmp.Id.Hex()
  36. for _, v := range append([]string{tmp.ProjectName}, tmp.MPN...) {
  37. if v != "" {
  38. //v = pcReplace.ReplaceAllString(v, "")
  39. if v != "" {
  40. k := p.mapPn[v]
  41. if k == nil {
  42. k = &Key{Arr: []string{id}}
  43. p.mapPn[v] = k
  44. } else {
  45. k.Arr = append(k.Arr, id)
  46. }
  47. }
  48. }
  49. }
  50. for _, v := range append([]string{tmp.ProjectCode}, tmp.MPC...) {
  51. if v != "" {
  52. //v = pcReplace.ReplaceAllString(v, "")
  53. if v != "" {
  54. k := p.mapPc[v]
  55. if k == nil {
  56. k = &Key{Arr: []string{id}}
  57. p.mapPc[v] = k
  58. } else {
  59. k.Arr = append(k.Arr, id)
  60. }
  61. }
  62. }
  63. }
  64. if tmp.Buyer != "" && len([]rune(tmp.Buyer)) > 2 {
  65. k := p.mapPb[tmp.Buyer]
  66. if k == nil {
  67. k = &Key{Arr: []string{id}}
  68. p.mapPb[tmp.Buyer] = k
  69. } else {
  70. k.Arr = append(k.Arr, id)
  71. }
  72. }
  73. p.AllIdsMap[id] = &ID{Id: id, P: tmp}
  74. }
  75. case <-loadOver:
  76. return
  77. }
  78. }
  79. }()
  80. for {
  81. result := make(map[string]interface{})
  82. if it.Next(&result) {
  83. go func(res map[string]interface{}) {
  84. bys, _ := json.Marshal(result)
  85. var tmp *ProjectCache
  86. _ = json.Unmarshal(bys, &tmp)
  87. saveFiled(p, result, tmp)
  88. redis.PutCKV("project", tmp.Id.Hex(), result)
  89. pool <- tmp
  90. }(result)
  91. } else {
  92. break
  93. }
  94. }
  95. time.Sleep(2 * time.Second)
  96. loadOver <- true
  97. log.Println("load project over..", n)
  98. }
  99. func (p *ProjectTask) loadSite() {
  100. log.Println("load site start..")
  101. p.findLock.Lock()
  102. defer p.findLock.Unlock()
  103. p.mapSiteLock.Lock()
  104. defer p.mapSiteLock.Unlock()
  105. sess := MongoTool.GetMgoConn()
  106. defer MongoTool.DestoryMongoConn(sess)
  107. q := map[string]interface{}{}
  108. it := sess.DB(MongoTool.DbName).C(SiteColl).Find(&q).Iter()
  109. n := 0
  110. pool := make(chan *Site, 100)
  111. over := make(chan bool)
  112. go func() {
  113. for {
  114. select {
  115. case tmp := <-pool:
  116. n++
  117. //站点有效标记state
  118. if tmp != nil && tmp.Status == 5 {
  119. p.mapSite[tmp.Site] = tmp
  120. }
  121. case <-over:
  122. return
  123. }
  124. }
  125. }()
  126. for {
  127. result := make(map[string]interface{})
  128. if it.Next(&result) {
  129. go func(res map[string]interface{}) {
  130. bys, _ := json.Marshal(result)
  131. var tmp *Site
  132. _ = json.Unmarshal(bys, &tmp)
  133. pool <- tmp
  134. }(result)
  135. } else {
  136. break
  137. }
  138. }
  139. time.Sleep(2 * time.Second)
  140. over <- true
  141. log.Println("load site over..", n)
  142. }
  143. func saveFiled(p *ProjectTask, res map[string]interface{}, tmp *ProjectCache) {
  144. proHref := util.ObjToString(res["projecthref"])
  145. if proHref != "" {
  146. p.mapHrefLock.Lock()
  147. p.mapHref[proHref] = tmp.Id.Hex()
  148. p.mapHrefLock.Unlock()
  149. }
  150. //tmpMap := make(map[string]InfoField)
  151. //infoMap := res["infofield"].(map[string]interface{})
  152. //for _, v := range infoMap {
  153. // var field InfoField
  154. // b, _ := json.Marshal(v)
  155. // _ = json.Unmarshal(b, &field)
  156. // tmpMap[tmp.Id.Hex()] = field
  157. //}
  158. //tmp.InfoFiled = tmpMap
  159. }
  160. // 加载spidercode数据,isflow字段
  161. func (p *ProjectTask) loadSpiderCode() {
  162. log.Println("load spider code start..")
  163. p.findLock.Lock()
  164. defer p.findLock.Unlock()
  165. p.mapSpiderLock.Lock()
  166. defer p.mapSpiderLock.Unlock()
  167. sess := MgoSpider.GetMgoConn()
  168. defer MgoSpider.DestoryMongoConn(sess)
  169. q := map[string]interface{}{}
  170. field := map[string]interface{}{"code": 1, "isflow": 1}
  171. it := sess.DB(MgoSpider.DbName).C("luaconfig").Find(&q).Select(field).Iter()
  172. n := 0
  173. pool := make(chan map[string]interface{}, 100)
  174. over := make(chan bool)
  175. go func() {
  176. for {
  177. select {
  178. case tmp := <-pool:
  179. n++
  180. code := util.ObjToString(tmp["code"])
  181. p.mapSpider[code] = util.IntAll(tmp["isflow"])
  182. case <-over:
  183. return
  184. }
  185. }
  186. }()
  187. for {
  188. result := make(map[string]interface{})
  189. if it.Next(&result) {
  190. go func(res map[string]interface{}) {
  191. pool <- result
  192. }(result)
  193. } else {
  194. break
  195. }
  196. }
  197. time.Sleep(2 * time.Second)
  198. over <- true
  199. log.Println("load spider over..", n)
  200. }