task.go 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391
  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()
  21. defer Mgo.DestoryMongoConn(sess)
  22. q := bson.M{"jy_updatetime": bson.M{"$gt": jyUpdatetime}}
  23. var zid int
  24. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  25. count := 0
  26. ch := make(chan bool, 16)
  27. wg := &sync.WaitGroup{}
  28. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  29. if count%20000 == 0 {
  30. log.Println("current:", count)
  31. }
  32. if util.ObjToString(tmp["_operation_type"]) == "update" {
  33. continue
  34. }
  35. //废弃
  36. if util.Int64All(tmp["use_flag"]) > 8 {
  37. continue
  38. }
  39. zid = util.IntAll(tmp["_id"])
  40. ch <- true
  41. wg.Add(1)
  42. go func(tmp map[string]interface{}) {
  43. defer func() {
  44. <-ch
  45. wg.Done()
  46. }()
  47. mutex.Lock()
  48. defer mutex.Unlock()
  49. //
  50. currentTime := time.Now().Unix()
  51. query := bson.M{"company_id": tmp["company_id"]}
  52. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  53. //原来数据有changes 字段就更新,追加数据
  54. if b && len(*info) > 0 {
  55. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  56. update := make(map[string]interface{})
  57. item := make(map[string]interface{})
  58. item["change_field"] = tmp["change_field"]
  59. item["content_before"] = tmp["content_before"]
  60. item["content_after"] = tmp["content_after"]
  61. item["change_date"] = tmp["change_date"]
  62. setMark(item) //change_name_new
  63. //update["changes"] = changes
  64. update["update_time"] = currentTime
  65. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  66. MgoMix.Update("qyxy_change", map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  67. }
  68. } else {
  69. //没有的直接写入
  70. query := bson.M{"_id": tmp["company_id"]}
  71. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  72. if b1 && len(*qyxy) > 0 {
  73. save := make(map[string]interface{})
  74. var changes []map[string]interface{}
  75. item := make(map[string]interface{})
  76. item["change_field"] = tmp["change_field"]
  77. item["content_before"] = tmp["content_before"]
  78. item["content_after"] = tmp["content_after"]
  79. item["change_date"] = tmp["change_date"]
  80. setMark(item) //change_name_new
  81. changes = append(changes, item)
  82. //save["company_name"] = (*qyxy)["company_name"]
  83. save["company_id"] = (*qyxy)["_id"]
  84. save["changes"] = changes
  85. save["create_time"] = currentTime
  86. save["update_time"] = currentTime
  87. //saveInfo := map[string]interface{}{"$set": save}
  88. MgoMix.Save(CollSave, save)
  89. }
  90. }
  91. }(tmp)
  92. tmp = map[string]interface{}{}
  93. }
  94. wg.Wait()
  95. util.Debug("over---", count, zid)
  96. }
  97. //IncByID 通过传入id 更新数据
  98. func IncByID(mapinfo map[string]interface{}) {
  99. defer util.Catch()
  100. sess := Mgo.GetMgoConn()
  101. defer Mgo.DestoryMongoConn(sess)
  102. ch := make(chan bool, 16)
  103. wg := &sync.WaitGroup{}
  104. q := map[string]interface{}{
  105. "_id": map[string]interface{}{
  106. "$gt": util.Int64All(mapinfo["gtid"]),
  107. },
  108. }
  109. var zid int
  110. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  111. count := 0
  112. total := 0
  113. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  114. if count%20000 == 0 {
  115. log.Println("current:", count)
  116. log.Println("current _id:", tmp["_id"], "total", total)
  117. }
  118. //表更时间
  119. if util.ObjToString(tmp["change_date"]) < "2023-01-01" {
  120. continue
  121. }
  122. if util.ObjToString(tmp["_operation_type"]) == "update" {
  123. continue
  124. }
  125. //废弃
  126. if util.Int64All(tmp["use_flag"]) > 8 {
  127. continue
  128. }
  129. zid = util.IntAll(tmp["_id"])
  130. ch <- true
  131. wg.Add(1)
  132. go func(tmp map[string]interface{}) {
  133. defer func() {
  134. <-ch
  135. wg.Done()
  136. }()
  137. mutex.Lock()
  138. defer mutex.Unlock()
  139. currentTime := time.Now().Unix()
  140. query := bson.M{"company_id": tmp["company_id"]}
  141. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  142. //原来数据有changes 字段就更新,追加数据
  143. if b && len(*info) > 0 {
  144. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  145. update := make(map[string]interface{})
  146. item := make(map[string]interface{})
  147. item["change_field"] = tmp["change_field"]
  148. item["content_before"] = tmp["content_before"]
  149. item["content_after"] = tmp["content_after"]
  150. item["change_date"] = tmp["change_date"]
  151. setMark(item) //change_name_new
  152. //update["changes"] = changes
  153. update["update_time"] = currentTime
  154. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  155. MgoMix.Update("qyxy_change", map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  156. total++
  157. }
  158. } else {
  159. total++
  160. //没有的直接写入
  161. query := bson.M{"_id": tmp["company_id"]}
  162. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  163. if b1 && len(*qyxy) > 0 {
  164. save := make(map[string]interface{})
  165. var changes []map[string]interface{}
  166. item := make(map[string]interface{})
  167. item["change_field"] = tmp["change_field"]
  168. item["content_before"] = tmp["content_before"]
  169. item["content_after"] = tmp["content_after"]
  170. item["change_date"] = tmp["change_date"]
  171. setMark(item) //change_name_new
  172. changes = append(changes, item)
  173. //save["company_name"] = (*qyxy)["company_name"]
  174. save["company_id"] = (*qyxy)["_id"]
  175. save["changes"] = changes
  176. save["create_time"] = currentTime
  177. save["update_time"] = currentTime
  178. //saveInfo := map[string]interface{}{"$set": save}
  179. MgoMix.Save(CollSave, save)
  180. }
  181. }
  182. }(tmp)
  183. tmp = map[string]interface{}{}
  184. }
  185. wg.Wait()
  186. util.Debug("over---", count, zid)
  187. }
  188. func setMark(tmp map[string]interface{}) {
  189. for _, v := range ChangeMap {
  190. str := util.ObjToString(tmp["change_field"])
  191. regArr := v["change_key_reg"].([]string)
  192. for _, v1 := range regArr {
  193. matched, _ := regexp.MatchString(v1, str)
  194. if matched {
  195. tmp["change_name_new"] = v["change_name"]
  196. return
  197. }
  198. }
  199. }
  200. }
  201. func clearRepeat(list []interface{}) []interface{} {
  202. var tmp []interface{}
  203. if len(list) > 1 {
  204. for k, v := range list {
  205. if k < len(list)-1 {
  206. if fmt.Sprint(list[k]) != fmt.Sprint(list[k+1]) {
  207. tmp = append(tmp, v)
  208. }
  209. } else {
  210. tmp = append(tmp, v)
  211. }
  212. }
  213. return tmp
  214. } else {
  215. return list
  216. }
  217. }
  218. // TaskAll 存量数据
  219. func TaskAll() {
  220. defer util.Catch()
  221. sess := MgoMix.GetMgoConn()
  222. defer MgoMix.DestoryMongoConn(sess)
  223. pool := make(chan bool, 10)
  224. wg := &sync.WaitGroup{}
  225. field := bson.M{"company_name": 1}
  226. it := sess.DB("mixdata").C(CollQy).Find(nil).Select(field).Iter()
  227. count := 0
  228. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  229. if count%20000 == 0 {
  230. log.Println("current:", count, tmp["_id"])
  231. }
  232. if strings.Contains(util.ObjToString(tmp["company_type"]), "个体") {
  233. continue
  234. }
  235. pool <- true
  236. wg.Add(1)
  237. go func(tmp map[string]interface{}) {
  238. defer func() {
  239. <-pool
  240. wg.Done()
  241. }()
  242. tmp["company_id"] = tmp["_id"]
  243. delete(tmp, "_id")
  244. query := bson.M{"company_id": tmp["company_id"]}
  245. info, b := Mgo.Find("company_change", query, nil, nil, false, -1, -1)
  246. if b && len(*info) > 0 {
  247. disposeFuc(*info, tmp)
  248. }
  249. }(tmp)
  250. tmp = map[string]interface{}{}
  251. }
  252. wg.Wait()
  253. }
  254. func taskInfo(id string) {
  255. update := make(map[string]interface{})
  256. q := map[string]interface{}{"company_id": id}
  257. info, _ := Mgo.Find("company_change", q, nil, nil, false, -1, -1)
  258. if len(*info) > 0 {
  259. var changes []map[string]interface{}
  260. currentTime := time.Now().Unix()
  261. for _, v := range *info {
  262. item := make(map[string]interface{})
  263. item["change_field"] = v["change_field"]
  264. item["content_before"] = v["content_before"]
  265. item["content_after"] = v["content_after"]
  266. item["change_date"] = v["change_date"]
  267. setMark(item) //change_name_new
  268. changes = append(changes, item)
  269. }
  270. update["changes"] = changes
  271. update["update_time"] = currentTime
  272. }
  273. util.Debug(update)
  274. MgoMix.Update("qyxy_change", q, map[string]interface{}{"$set": update}, false, false)
  275. }
  276. func disposeFuc(maps []map[string]interface{}, tmp map[string]interface{}) {
  277. var changes []map[string]interface{}
  278. currentTime := time.Now().Unix()
  279. for _, v := range maps {
  280. item := make(map[string]interface{})
  281. item["change_field"] = v["change_field"]
  282. item["content_before"] = v["content_before"]
  283. item["content_after"] = v["content_after"]
  284. item["change_date"] = v["change_date"]
  285. setMark(item) //change_name_new
  286. changes = append(changes, item)
  287. }
  288. tmp["changes"] = changes
  289. tmp["create_time"] = currentTime
  290. tmp["update_time"] = currentTime
  291. MgoSaveCache <- tmp
  292. }
  293. //SaveData 存量保存
  294. func SaveData() {
  295. log.Println("Mgo Save...")
  296. arru := make([]map[string]interface{}, 200)
  297. indexu := 0
  298. for {
  299. select {
  300. case v := <-MgoSaveCache:
  301. arru[indexu] = v
  302. indexu++
  303. if indexu == 200 {
  304. SP <- true
  305. go func(arru []map[string]interface{}) {
  306. defer func() {
  307. <-SP
  308. }()
  309. MgoMix.SaveBulk(CollSave, arru...)
  310. }(arru)
  311. arru = make([]map[string]interface{}, 200)
  312. indexu = 0
  313. }
  314. case <-time.After(1000 * time.Millisecond):
  315. if indexu > 0 {
  316. SP <- true
  317. go func(arru []map[string]interface{}) {
  318. defer func() {
  319. <-SP
  320. }()
  321. MgoMix.SaveBulk(CollSave, arru...)
  322. }(arru[:indexu])
  323. arru = make([]map[string]interface{}, 200)
  324. indexu = 0
  325. }
  326. }
  327. }
  328. }
  329. func updateMethod() {
  330. arru := make([][]map[string]interface{}, 200)
  331. indexu := 0
  332. for {
  333. select {
  334. case v := <-updatePool:
  335. arru[indexu] = v
  336. indexu++
  337. if indexu == 200 {
  338. updateSp <- true
  339. go func(arru [][]map[string]interface{}) {
  340. defer func() {
  341. <-updateSp
  342. }()
  343. MgoMix.UpSertBulk(CollSave, arru...)
  344. }(arru)
  345. arru = make([][]map[string]interface{}, 200)
  346. indexu = 0
  347. }
  348. case <-time.After(1000 * time.Millisecond):
  349. if indexu > 0 {
  350. updateSp <- true
  351. go func(arru [][]map[string]interface{}) {
  352. defer func() {
  353. <-updateSp
  354. }()
  355. MgoMix.UpSertBulk(CollSave, arru...)
  356. }(arru[:indexu])
  357. arru = make([][]map[string]interface{}, 200)
  358. indexu = 0
  359. }
  360. }
  361. }
  362. }