load_data.go 4.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205
  1. package main
  2. import (
  3. "encoding/json"
  4. "log"
  5. "qfw/util"
  6. "time"
  7. )
  8. // 初始加载数据,默认加载最近6个月的数据
  9. func (p *ProjectTask) loadData(starttime int64) {
  10. log.Println("load project start..", starttime)
  11. p.findLock.Lock()
  12. defer p.findLock.Unlock()
  13. p.AllIdsMapLock.Lock()
  14. defer p.AllIdsMapLock.Unlock()
  15. sess := MongoTool.GetMgoConn()
  16. defer MongoTool.DestoryMongoConn(sess)
  17. q := map[string]interface{}{
  18. "lasttime": map[string]interface{}{"$gte": starttime},
  19. }
  20. it := sess.DB(MongoTool.DbName).C(p.coll).Find(&q).Iter()
  21. n := 0
  22. pool := make(chan *ProjectInfo, 100)
  23. over := make(chan bool)
  24. go func() {
  25. for {
  26. select {
  27. case tmp := <-pool:
  28. n++
  29. if n%10000 == 0 {
  30. 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)
  31. }
  32. if tmp != nil {
  33. id := tmp.Id.Hex()
  34. for _, v := range append([]string{tmp.ProjectName}, tmp.MPN...) {
  35. if v != "" {
  36. v = pcReplace.ReplaceAllString(v, "")
  37. if v != "" {
  38. k := p.mapPn[v]
  39. if k == nil {
  40. k = &Key{Arr: []string{id}}
  41. p.mapPn[v] = k
  42. } else {
  43. k.Arr = append(k.Arr, id)
  44. }
  45. }
  46. }
  47. }
  48. for _, v := range append([]string{tmp.ProjectCode}, tmp.MPC...) {
  49. if v != "" {
  50. v = pcReplace.ReplaceAllString(v, "")
  51. if v != "" {
  52. k := p.mapPc[v]
  53. if k == nil {
  54. k = &Key{Arr: []string{id}}
  55. p.mapPc[v] = k
  56. } else {
  57. k.Arr = append(k.Arr, id)
  58. }
  59. }
  60. }
  61. }
  62. if tmp.Buyer != "" && len([]rune(tmp.Buyer)) > 2 {
  63. k := p.mapPb[tmp.Buyer]
  64. if k == nil {
  65. k = &Key{Arr: []string{id}}
  66. p.mapPb[tmp.Buyer] = k
  67. } else {
  68. k.Arr = append(k.Arr, id)
  69. }
  70. }
  71. p.AllIdsMap[id] = &ID{Id: id, P: tmp}
  72. }
  73. case <-over:
  74. return
  75. }
  76. }
  77. }()
  78. for {
  79. result := make(map[string]interface{})
  80. if it.Next(&result) {
  81. go func(res map[string]interface{}) {
  82. bys, _ := json.Marshal(result)
  83. var tmp *ProjectInfo
  84. _ = json.Unmarshal(bys, &tmp)
  85. saveFiled(p, result, tmp)
  86. pool <- tmp
  87. }(result)
  88. } else {
  89. break
  90. }
  91. }
  92. time.Sleep(2 * time.Second)
  93. over <- true
  94. log.Println("load project over..", n)
  95. }
  96. func (p *ProjectTask) loadSite() {
  97. log.Println("load site start..")
  98. p.findLock.Lock()
  99. defer p.findLock.Unlock()
  100. p.mapSiteLock.Lock()
  101. defer p.mapSiteLock.Unlock()
  102. sess := MongoTool.GetMgoConn()
  103. defer MongoTool.DestoryMongoConn(sess)
  104. q := map[string]interface{}{}
  105. it := sess.DB(MongoTool.DbName).C(SiteColl).Find(&q).Iter()
  106. n := 0
  107. pool := make(chan *Site, 100)
  108. over := make(chan bool)
  109. go func() {
  110. for {
  111. select {
  112. case tmp := <-pool:
  113. n++
  114. //站点有效标记state
  115. if tmp != nil && tmp.Status == 5 {
  116. p.mapSite[tmp.Site] = tmp
  117. }
  118. case <-over:
  119. return
  120. }
  121. }
  122. }()
  123. for {
  124. result := make(map[string]interface{})
  125. if it.Next(&result) {
  126. go func(res map[string]interface{}) {
  127. bys, _ := json.Marshal(result)
  128. var tmp *Site
  129. _ = json.Unmarshal(bys, &tmp)
  130. pool <- tmp
  131. }(result)
  132. } else {
  133. break
  134. }
  135. }
  136. time.Sleep(2 * time.Second)
  137. over <- true
  138. log.Println("load site over..", n)
  139. }
  140. func saveFiled(p *ProjectTask, res map[string]interface{}, tmp *ProjectInfo) {
  141. proHref := util.ObjToString(res["projecthref"])
  142. if proHref != "" {
  143. p.mapHrefLock.Lock()
  144. p.mapHref[proHref] = tmp.Id.Hex()
  145. p.mapHrefLock.Unlock()
  146. }
  147. tmpMap := make(map[string]InfoField)
  148. infoMap := res["infofield"].(map[string]interface{})
  149. for _, v := range infoMap {
  150. var field InfoField
  151. b, _ := json.Marshal(v)
  152. _ = json.Unmarshal(b, &field)
  153. tmpMap[tmp.Id.Hex()] = field
  154. }
  155. tmp.InfoFiled = tmpMap
  156. }
  157. // 加载spidercode数据,isflow字段
  158. func (p *ProjectTask) loadSpiderCode() {
  159. log.Println("load spider code start..")
  160. p.findLock.Lock()
  161. defer p.findLock.Unlock()
  162. p.mapSpiderLock.Lock()
  163. defer p.mapSpiderLock.Unlock()
  164. sess := MgoSpider.GetMgoConn()
  165. defer MgoSpider.DestoryMongoConn(sess)
  166. q := map[string]interface{}{}
  167. field := map[string]interface{}{"code": 1, "isflow": 1}
  168. it := sess.DB(MgoSpider.DbName).C("luaconfig").Find(&q).Select(field).Iter()
  169. n := 0
  170. pool := make(chan map[string]interface{}, 100)
  171. over := make(chan bool)
  172. go func() {
  173. for {
  174. select {
  175. case tmp := <-pool:
  176. n++
  177. code := util.ObjToString(tmp["code"])
  178. p.mapSpider[code] = util.IntAll(tmp["isflow"])
  179. case <-over:
  180. return
  181. }
  182. }
  183. }()
  184. for {
  185. result := make(map[string]interface{})
  186. if it.Next(&result) {
  187. go func(res map[string]interface{}) {
  188. pool <- result
  189. }(result)
  190. } else {
  191. break
  192. }
  193. }
  194. time.Sleep(2 * time.Second)
  195. over <- true
  196. log.Println("load spider over..", n)
  197. }