load_data.go 5.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236
  1. package main
  2. import (
  3. "encoding/json"
  4. "log"
  5. "qfw/util"
  6. "qfw/util/redis"
  7. "strings"
  8. "sync"
  9. "time"
  10. )
  11. // 初始加载数据,默认加载最近6个月的数据
  12. func (p *ProjectTask) loadData(starttime int64) {
  13. log.Println("load project start..", starttime)
  14. p.findLock.Lock()
  15. defer p.findLock.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(field).Iter()
  24. n, count := 0, 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.AllIdsMapLock.Lock()
  74. p.AllIdsMap[id] = &ID{Id: id, P: tmp}
  75. p.AllIdsMapLock.Unlock()
  76. }
  77. case <-loadOver:
  78. return
  79. }
  80. }
  81. }()
  82. ch := make(chan bool, 3)
  83. wg := &sync.WaitGroup{}
  84. for tmp := make(map[string]interface{}); it.Next(tmp); count++ {
  85. //if count%20000 == 0 {
  86. // log.Println(fmt.Sprintf("iter --- %d", count))
  87. //}
  88. ch <- true
  89. wg.Add(1)
  90. go func(tmp map[string]interface{}) {
  91. defer func() {
  92. <-ch
  93. wg.Done()
  94. }()
  95. bys, _ := json.Marshal(tmp)
  96. var pc *ProjectCache
  97. _ = json.Unmarshal(bys, &pc)
  98. saveFiled(p, tmp, pc)
  99. redis.PutCKV("project", pc.Id.Hex(), tmp)
  100. pool <- pc
  101. }(tmp)
  102. tmp = make(map[string]interface{})
  103. }
  104. wg.Wait()
  105. time.Sleep(2 * time.Second)
  106. loadOver <- true
  107. log.Println("load project over..", n)
  108. }
  109. func (p *ProjectTask) loadSite() {
  110. log.Println("load site start..")
  111. p.findLock.Lock()
  112. defer p.findLock.Unlock()
  113. p.mapSiteLock.Lock()
  114. defer p.mapSiteLock.Unlock()
  115. sess := MongoTool.GetMgoConn()
  116. defer MongoTool.DestoryMongoConn(sess)
  117. q := map[string]interface{}{}
  118. it := sess.DB(MongoTool.DbName).C(SiteColl).Find(&q).Iter()
  119. n := 0
  120. pool := make(chan *Site, 100)
  121. over := make(chan bool)
  122. go func() {
  123. for {
  124. select {
  125. case tmp := <-pool:
  126. n++
  127. //站点有效标记state
  128. if tmp != nil && tmp.Status == 5 {
  129. p.mapSite[tmp.Site] = tmp
  130. }
  131. case <-over:
  132. return
  133. }
  134. }
  135. }()
  136. for {
  137. result := make(map[string]interface{})
  138. if it.Next(&result) {
  139. go func(res map[string]interface{}) {
  140. bys, _ := json.Marshal(result)
  141. var tmp *Site
  142. _ = json.Unmarshal(bys, &tmp)
  143. pool <- tmp
  144. }(result)
  145. } else {
  146. break
  147. }
  148. }
  149. time.Sleep(2 * time.Second)
  150. over <- true
  151. log.Println("load site over..", n)
  152. }
  153. func saveFiled(p *ProjectTask, res map[string]interface{}, tmp *ProjectCache) {
  154. proHref := util.ObjToString(res["projecthref"])
  155. if proHref != "" {
  156. p.mapHrefLock.Lock()
  157. p.mapHref[proHref] = tmp.Id.Hex()
  158. p.mapHrefLock.Unlock()
  159. }
  160. if res["entidlist"] != nil {
  161. elist := util.ObjArrToStringArr(res["entidlist"].([]interface{}))
  162. wlist := strings.Split(util.ObjToString(res["s_winner"]), ",")
  163. buyer := util.ObjToString(res["buyer"])
  164. if len(elist) == len(wlist) && buyer != "" {
  165. for i, eid := range elist {
  166. if eid != "-" {
  167. text := buyer + "," + wlist[i]
  168. ex, _ := redis.Exists(RedisCode, text)
  169. if !ex {
  170. redis.PutCKV(RedisCode, text, tmp.Id.Hex())
  171. }
  172. }
  173. }
  174. }
  175. }
  176. //tmpMap := make(map[string]InfoField)
  177. //infoMap := res["infofield"].(map[string]interface{})
  178. //for _, v := range infoMap {
  179. // var field InfoField
  180. // b, _ := json.Marshal(v)
  181. // _ = json.Unmarshal(b, &field)
  182. // tmpMap[tmp.Id.Hex()] = field
  183. //}
  184. //tmp.InfoFiled = tmpMap
  185. }
  186. // 加载spidercode数据,isflow字段
  187. func (p *ProjectTask) loadSpiderCode() {
  188. log.Println("load spider code start..")
  189. p.findLock.Lock()
  190. defer p.findLock.Unlock()
  191. p.mapSpiderLock.Lock()
  192. defer p.mapSpiderLock.Unlock()
  193. sess := MgoSpider.GetMgoConn()
  194. defer MgoSpider.DestoryMongoConn(sess)
  195. q := map[string]interface{}{}
  196. field := map[string]interface{}{"code": 1, "isflow": 1}
  197. it := sess.DB(MgoSpider.DbName).C("luaconfig").Find(&q).Select(field).Iter()
  198. n := 0
  199. pool := make(chan map[string]interface{}, 100)
  200. over := make(chan bool)
  201. go func() {
  202. for {
  203. select {
  204. case tmp := <-pool:
  205. n++
  206. code := util.ObjToString(tmp["code"])
  207. p.mapSpider[code] = util.IntAll(tmp["isflow"])
  208. case <-over:
  209. return
  210. }
  211. }
  212. }()
  213. for {
  214. result := make(map[string]interface{})
  215. if it.Next(&result) {
  216. go func(res map[string]interface{}) {
  217. pool <- result
  218. }(result)
  219. } else {
  220. break
  221. }
  222. }
  223. time.Sleep(2 * time.Second)
  224. over <- true
  225. log.Println("load spider over..", n)
  226. }