task.go 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383
  1. package main
  2. import (
  3. "fmt"
  4. "go.mongodb.org/mongo-driver/bson"
  5. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  6. "log"
  7. "regexp"
  8. "strings"
  9. "sync"
  10. "time"
  11. )
  12. var MgoSaveCache = make(chan map[string]interface{}, 5000)
  13. var SP = make(chan bool, 5)
  14. var updatePool = make(chan []map[string]interface{}, 5000)
  15. var updateSp = make(chan bool, 5)
  16. var mutex sync.Mutex
  17. //IncData 增量处理数据
  18. func IncData() {
  19. defer util.Catch()
  20. sess := Mgo.GetMgoConn() //181 凭安
  21. defer Mgo.DestoryMongoConn(sess)
  22. q := bson.M{"jy_updatetime": bson.M{"$gte": jyUpdatetime}}
  23. var zid int
  24. queryCount, _ := sess.DB("mixdata").C("company_change").Find(q).Count()
  25. util.Debug("queryCount", queryCount)
  26. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  27. count := 0
  28. realCount := 0
  29. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  30. if count%20000 == 0 {
  31. log.Println("current:", count)
  32. log.Println("current: realCount", realCount)
  33. }
  34. if util.ObjToString(tmp["_operation_type"]) == "update" {
  35. continue
  36. }
  37. //废弃
  38. if util.Int64All(tmp["use_flag"]) > 8 {
  39. continue
  40. }
  41. zid = util.IntAll(tmp["_id"])
  42. currentTime := time.Now().Unix()
  43. query := bson.M{"company_id": tmp["company_id"]}
  44. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  45. //原来数据有changes 字段就更新,追加数据
  46. if b && len(*info) > 0 {
  47. realCount++
  48. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  49. update := make(map[string]interface{})
  50. item := make(map[string]interface{})
  51. item["change_field"] = tmp["change_field"]
  52. item["content_before"] = tmp["content_before"]
  53. item["content_after"] = tmp["content_after"]
  54. item["change_date"] = tmp["change_date"]
  55. setMark(item) //change_name_new
  56. //update["changes"] = changes
  57. update["update_time"] = currentTime
  58. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  59. MgoMix.Update(CollSave, map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  60. }
  61. } else {
  62. //没有的直接写入
  63. query := bson.M{"_id": tmp["company_id"]}
  64. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  65. if b1 && len(*qyxy) > 0 {
  66. realCount++
  67. save := make(map[string]interface{})
  68. var changes []map[string]interface{}
  69. item := make(map[string]interface{})
  70. item["change_field"] = tmp["change_field"]
  71. item["content_before"] = tmp["content_before"]
  72. item["content_after"] = tmp["content_after"]
  73. item["change_date"] = tmp["change_date"]
  74. setMark(item) //change_name_new
  75. changes = append(changes, item)
  76. //save["company_name"] = (*qyxy)["company_name"]
  77. save["company_id"] = (*qyxy)["_id"]
  78. save["changes"] = changes
  79. save["create_time"] = currentTime
  80. save["update_time"] = currentTime
  81. //saveInfo := map[string]interface{}{"$set": save}
  82. MgoMix.Save(CollSave, save)
  83. }
  84. }
  85. }
  86. util.Debug("over---", count, zid)
  87. }
  88. //IncByID 通过传入id 更新数据
  89. func IncByID(mapinfo map[string]interface{}) {
  90. defer util.Catch()
  91. sess := Mgo.GetMgoConn()
  92. defer Mgo.DestoryMongoConn(sess)
  93. ch := make(chan bool, 16)
  94. wg := &sync.WaitGroup{}
  95. q := map[string]interface{}{
  96. "_id": map[string]interface{}{
  97. "$gt": util.Int64All(mapinfo["gtid"]),
  98. },
  99. }
  100. var zid int
  101. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  102. count := 0
  103. total := 0
  104. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  105. if count%20000 == 0 {
  106. log.Println("current:", count)
  107. log.Println("current _id:", tmp["_id"], "total", total)
  108. }
  109. //表更时间
  110. if util.ObjToString(tmp["change_date"]) < "2023-01-01" {
  111. continue
  112. }
  113. if util.ObjToString(tmp["_operation_type"]) == "update" {
  114. continue
  115. }
  116. //废弃
  117. if util.Int64All(tmp["use_flag"]) > 8 {
  118. continue
  119. }
  120. zid = util.IntAll(tmp["_id"])
  121. ch <- true
  122. wg.Add(1)
  123. go func(tmp map[string]interface{}) {
  124. defer func() {
  125. <-ch
  126. wg.Done()
  127. }()
  128. mutex.Lock()
  129. defer mutex.Unlock()
  130. currentTime := time.Now().Unix()
  131. query := bson.M{"company_id": tmp["company_id"]}
  132. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  133. //原来数据有changes 字段就更新,追加数据
  134. if b && len(*info) > 0 {
  135. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  136. update := make(map[string]interface{})
  137. item := make(map[string]interface{})
  138. item["change_field"] = tmp["change_field"]
  139. item["content_before"] = tmp["content_before"]
  140. item["content_after"] = tmp["content_after"]
  141. item["change_date"] = tmp["change_date"]
  142. setMark(item) //change_name_new
  143. //update["changes"] = changes
  144. update["update_time"] = currentTime
  145. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  146. MgoMix.Update("qyxy_change", map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  147. total++
  148. }
  149. } else {
  150. total++
  151. //没有的直接写入
  152. query := bson.M{"_id": tmp["company_id"]}
  153. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  154. if b1 && len(*qyxy) > 0 {
  155. save := make(map[string]interface{})
  156. var changes []map[string]interface{}
  157. item := make(map[string]interface{})
  158. item["change_field"] = tmp["change_field"]
  159. item["content_before"] = tmp["content_before"]
  160. item["content_after"] = tmp["content_after"]
  161. item["change_date"] = tmp["change_date"]
  162. setMark(item) //change_name_new
  163. changes = append(changes, item)
  164. //save["company_name"] = (*qyxy)["company_name"]
  165. save["company_id"] = (*qyxy)["_id"]
  166. save["changes"] = changes
  167. save["create_time"] = currentTime
  168. save["update_time"] = currentTime
  169. //saveInfo := map[string]interface{}{"$set": save}
  170. MgoMix.Save(CollSave, save)
  171. }
  172. }
  173. }(tmp)
  174. tmp = map[string]interface{}{}
  175. }
  176. wg.Wait()
  177. util.Debug("over---", count, zid)
  178. }
  179. func setMark(tmp map[string]interface{}) {
  180. for _, v := range ChangeMap {
  181. str := util.ObjToString(tmp["change_field"])
  182. regArr := v["change_key_reg"].([]string)
  183. for _, v1 := range regArr {
  184. matched, _ := regexp.MatchString(v1, str)
  185. if matched {
  186. tmp["change_name_new"] = v["change_name"]
  187. return
  188. }
  189. }
  190. }
  191. }
  192. func clearRepeat(list []interface{}) []interface{} {
  193. var tmp []interface{}
  194. if len(list) > 1 {
  195. for k, v := range list {
  196. if k < len(list)-1 {
  197. if fmt.Sprint(list[k]) != fmt.Sprint(list[k+1]) {
  198. tmp = append(tmp, v)
  199. }
  200. } else {
  201. tmp = append(tmp, v)
  202. }
  203. }
  204. return tmp
  205. } else {
  206. return list
  207. }
  208. }
  209. // TaskAll 存量数据
  210. func TaskAll() {
  211. defer util.Catch()
  212. sess := MgoMix.GetMgoConn()
  213. defer MgoMix.DestoryMongoConn(sess)
  214. pool := make(chan bool, 10)
  215. wg := &sync.WaitGroup{}
  216. field := bson.M{"company_name": 1}
  217. it := sess.DB("mixdata").C(CollQy).Find(nil).Select(field).Iter()
  218. count := 0
  219. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  220. if count%20000 == 0 {
  221. log.Println("current:", count, tmp["_id"])
  222. }
  223. if strings.Contains(util.ObjToString(tmp["company_type"]), "个体") {
  224. continue
  225. }
  226. pool <- true
  227. wg.Add(1)
  228. go func(tmp map[string]interface{}) {
  229. defer func() {
  230. <-pool
  231. wg.Done()
  232. }()
  233. tmp["company_id"] = tmp["_id"]
  234. delete(tmp, "_id")
  235. query := bson.M{"company_id": tmp["company_id"]}
  236. info, b := Mgo.Find("company_change", query, nil, nil, false, -1, -1)
  237. if b && len(*info) > 0 {
  238. disposeFuc(*info, tmp)
  239. }
  240. }(tmp)
  241. tmp = map[string]interface{}{}
  242. }
  243. wg.Wait()
  244. }
  245. func taskInfo(id string) {
  246. update := make(map[string]interface{})
  247. q := map[string]interface{}{"company_id": id}
  248. info, _ := Mgo.Find("company_change", q, nil, nil, false, -1, -1)
  249. if len(*info) > 0 {
  250. var changes []map[string]interface{}
  251. currentTime := time.Now().Unix()
  252. for _, v := range *info {
  253. item := make(map[string]interface{})
  254. item["change_field"] = v["change_field"]
  255. item["content_before"] = v["content_before"]
  256. item["content_after"] = v["content_after"]
  257. item["change_date"] = v["change_date"]
  258. setMark(item) //change_name_new
  259. changes = append(changes, item)
  260. }
  261. update["changes"] = changes
  262. update["update_time"] = currentTime
  263. }
  264. util.Debug(update)
  265. MgoMix.Update("qyxy_change", q, map[string]interface{}{"$set": update}, false, false)
  266. }
  267. func disposeFuc(maps []map[string]interface{}, tmp map[string]interface{}) {
  268. var changes []map[string]interface{}
  269. currentTime := time.Now().Unix()
  270. for _, v := range maps {
  271. item := make(map[string]interface{})
  272. item["change_field"] = v["change_field"]
  273. item["content_before"] = v["content_before"]
  274. item["content_after"] = v["content_after"]
  275. item["change_date"] = v["change_date"]
  276. setMark(item) //change_name_new
  277. changes = append(changes, item)
  278. }
  279. tmp["changes"] = changes
  280. tmp["create_time"] = currentTime
  281. tmp["update_time"] = currentTime
  282. MgoSaveCache <- tmp
  283. }
  284. //SaveData 存量保存
  285. func SaveData() {
  286. log.Println("Mgo Save...")
  287. arru := make([]map[string]interface{}, 200)
  288. indexu := 0
  289. for {
  290. select {
  291. case v := <-MgoSaveCache:
  292. arru[indexu] = v
  293. indexu++
  294. if indexu == 200 {
  295. SP <- true
  296. go func(arru []map[string]interface{}) {
  297. defer func() {
  298. <-SP
  299. }()
  300. MgoMix.SaveBulk(CollSave, arru...)
  301. }(arru)
  302. arru = make([]map[string]interface{}, 200)
  303. indexu = 0
  304. }
  305. case <-time.After(1000 * time.Millisecond):
  306. if indexu > 0 {
  307. SP <- true
  308. go func(arru []map[string]interface{}) {
  309. defer func() {
  310. <-SP
  311. }()
  312. MgoMix.SaveBulk(CollSave, arru...)
  313. }(arru[:indexu])
  314. arru = make([]map[string]interface{}, 200)
  315. indexu = 0
  316. }
  317. }
  318. }
  319. }
  320. func updateMethod() {
  321. arru := make([][]map[string]interface{}, 200)
  322. indexu := 0
  323. for {
  324. select {
  325. case v := <-updatePool:
  326. arru[indexu] = v
  327. indexu++
  328. if indexu == 200 {
  329. updateSp <- true
  330. go func(arru [][]map[string]interface{}) {
  331. defer func() {
  332. <-updateSp
  333. }()
  334. MgoMix.UpSertBulk(CollSave, arru...)
  335. }(arru)
  336. arru = make([][]map[string]interface{}, 200)
  337. indexu = 0
  338. }
  339. case <-time.After(1000 * time.Millisecond):
  340. if indexu > 0 {
  341. updateSp <- true
  342. go func(arru [][]map[string]interface{}) {
  343. defer func() {
  344. <-updateSp
  345. }()
  346. MgoMix.UpSertBulk(CollSave, arru...)
  347. }(arru[:indexu])
  348. arru = make([][]map[string]interface{}, 200)
  349. indexu = 0
  350. }
  351. }
  352. }
  353. }