task.go 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384
  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{"$gt": jyUpdatetime}}
  23. var zid int
  24. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  25. count := 0
  26. ch := make(chan bool, 16)
  27. wg := &sync.WaitGroup{}
  28. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  29. if count%2000 == 0 {
  30. log.Println("current:", count)
  31. }
  32. zid = util.IntAll(tmp["_id"])
  33. ch <- true
  34. wg.Add(1)
  35. go func(tmp map[string]interface{}) {
  36. defer func() {
  37. <-ch
  38. wg.Done()
  39. }()
  40. mutex.Lock()
  41. defer mutex.Unlock()
  42. //
  43. currentTime := time.Now().Unix()
  44. query := bson.M{"company_id": tmp["company_id"]}
  45. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  46. //原来数据有changes 字段就更新,追加数据
  47. if b && len(*info) > 0 {
  48. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  49. update := make(map[string]interface{})
  50. item := make(map[string]interface{})
  51. item["change_field"] = tmp["change_field"]
  52. item["content_before"] = tmp["content_before"]
  53. item["content_after"] = tmp["content_after"]
  54. item["change_date"] = tmp["change_date"]
  55. setMark(item) //change_name_new
  56. //update["changes"] = changes
  57. update["update_time"] = currentTime
  58. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  59. MgoMix.Update("qyxy_change", map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  60. }
  61. } else {
  62. //没有的直接写入
  63. query := bson.M{"_id": tmp["company_id"]}
  64. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  65. if b1 && len(*qyxy) > 0 {
  66. save := make(map[string]interface{})
  67. var changes []map[string]interface{}
  68. item := make(map[string]interface{})
  69. item["change_field"] = tmp["change_field"]
  70. item["content_before"] = tmp["content_before"]
  71. item["content_after"] = tmp["content_after"]
  72. item["change_date"] = tmp["change_date"]
  73. setMark(item) //change_name_new
  74. changes = append(changes, item)
  75. //save["company_name"] = (*qyxy)["company_name"]
  76. save["company_id"] = (*qyxy)["_id"]
  77. save["changes"] = changes
  78. save["create_time"] = currentTime
  79. save["update_time"] = currentTime
  80. //saveInfo := map[string]interface{}{"$set": save}
  81. MgoMix.Save(CollSave, save)
  82. }
  83. }
  84. }(tmp)
  85. tmp = map[string]interface{}{}
  86. }
  87. wg.Wait()
  88. util.Debug("over---", count, zid)
  89. }
  90. //IncByID 通过传入id 更新数据
  91. func IncByID(mapinfo map[string]interface{}) {
  92. defer util.Catch()
  93. sess := Mgo.GetMgoConn()
  94. defer Mgo.DestoryMongoConn(sess)
  95. ch := make(chan bool, 16)
  96. wg := &sync.WaitGroup{}
  97. q := map[string]interface{}{
  98. "_id": map[string]interface{}{
  99. "$gt": util.Int64All(mapinfo["gtid"]),
  100. },
  101. }
  102. var zid int
  103. it := sess.DB("mixdata").C("company_change").Find(q).Select(nil).Iter()
  104. count := 0
  105. total := 0
  106. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  107. if count%20000 == 0 {
  108. log.Println("current:", count)
  109. log.Println("current _id:", tmp["_id"], "total", total)
  110. }
  111. //表更时间
  112. if util.ObjToString(tmp["change_date"]) < "2023-01-01" {
  113. continue
  114. }
  115. if util.ObjToString(tmp["_operation_type"]) == "update" {
  116. continue
  117. }
  118. //废弃
  119. if util.Int64All(tmp["use_flag"]) > 8 {
  120. continue
  121. }
  122. zid = util.IntAll(tmp["_id"])
  123. ch <- true
  124. wg.Add(1)
  125. go func(tmp map[string]interface{}) {
  126. defer func() {
  127. <-ch
  128. wg.Done()
  129. }()
  130. mutex.Lock()
  131. defer mutex.Unlock()
  132. currentTime := time.Now().Unix()
  133. query := bson.M{"company_id": tmp["company_id"]}
  134. info, b := MgoMix.FindOneByField(CollSave, query, bson.M{"changes": 1})
  135. //原来数据有changes 字段就更新,追加数据
  136. if b && len(*info) > 0 {
  137. if util.ObjToString(tmp["_operation_type"]) == "insert" {
  138. update := make(map[string]interface{})
  139. item := make(map[string]interface{})
  140. item["change_field"] = tmp["change_field"]
  141. item["content_before"] = tmp["content_before"]
  142. item["content_after"] = tmp["content_after"]
  143. item["change_date"] = tmp["change_date"]
  144. setMark(item) //change_name_new
  145. //update["changes"] = changes
  146. update["update_time"] = currentTime
  147. saveInfo := map[string]interface{}{"$set": update, "$push": map[string]interface{}{"changes": item}}
  148. MgoMix.Update("qyxy_change", map[string]interface{}{"company_id": util.ObjToString(tmp["company_id"])}, saveInfo, true, false)
  149. total++
  150. }
  151. } else {
  152. total++
  153. //没有的直接写入
  154. query := bson.M{"_id": tmp["company_id"]}
  155. qyxy, b1 := MgoMix.FindOne("qyxy_std", query)
  156. if b1 && len(*qyxy) > 0 {
  157. save := make(map[string]interface{})
  158. var changes []map[string]interface{}
  159. item := make(map[string]interface{})
  160. item["change_field"] = tmp["change_field"]
  161. item["content_before"] = tmp["content_before"]
  162. item["content_after"] = tmp["content_after"]
  163. item["change_date"] = tmp["change_date"]
  164. setMark(item) //change_name_new
  165. changes = append(changes, item)
  166. //save["company_name"] = (*qyxy)["company_name"]
  167. save["company_id"] = (*qyxy)["_id"]
  168. save["changes"] = changes
  169. save["create_time"] = currentTime
  170. save["update_time"] = currentTime
  171. //saveInfo := map[string]interface{}{"$set": save}
  172. MgoMix.Save(CollSave, save)
  173. }
  174. }
  175. }(tmp)
  176. tmp = map[string]interface{}{}
  177. }
  178. wg.Wait()
  179. util.Debug("over---", count, zid)
  180. }
  181. func setMark(tmp map[string]interface{}) {
  182. for _, v := range ChangeMap {
  183. str := util.ObjToString(tmp["change_field"])
  184. regArr := v["change_key_reg"].([]string)
  185. for _, v1 := range regArr {
  186. matched, _ := regexp.MatchString(v1, str)
  187. if matched {
  188. tmp["change_name_new"] = v["change_name"]
  189. return
  190. }
  191. }
  192. }
  193. }
  194. func clearRepeat(list []interface{}) []interface{} {
  195. var tmp []interface{}
  196. if len(list) > 1 {
  197. for k, v := range list {
  198. if k < len(list)-1 {
  199. if fmt.Sprint(list[k]) != fmt.Sprint(list[k+1]) {
  200. tmp = append(tmp, v)
  201. }
  202. } else {
  203. tmp = append(tmp, v)
  204. }
  205. }
  206. return tmp
  207. } else {
  208. return list
  209. }
  210. }
  211. // TaskAll 存量数据
  212. func TaskAll() {
  213. defer util.Catch()
  214. sess := MgoMix.GetMgoConn()
  215. defer MgoMix.DestoryMongoConn(sess)
  216. pool := make(chan bool, 10)
  217. wg := &sync.WaitGroup{}
  218. field := bson.M{"company_name": 1}
  219. it := sess.DB("mixdata").C(CollQy).Find(nil).Select(field).Iter()
  220. count := 0
  221. for tmp := make(map[string]interface{}); it.Next(&tmp); count++ {
  222. if count%20000 == 0 {
  223. log.Println("current:", count, tmp["_id"])
  224. }
  225. if strings.Contains(util.ObjToString(tmp["company_type"]), "个体") {
  226. continue
  227. }
  228. pool <- true
  229. wg.Add(1)
  230. go func(tmp map[string]interface{}) {
  231. defer func() {
  232. <-pool
  233. wg.Done()
  234. }()
  235. tmp["company_id"] = tmp["_id"]
  236. delete(tmp, "_id")
  237. query := bson.M{"company_id": tmp["company_id"]}
  238. info, b := Mgo.Find("company_change", query, nil, nil, false, -1, -1)
  239. if b && len(*info) > 0 {
  240. disposeFuc(*info, tmp)
  241. }
  242. }(tmp)
  243. tmp = map[string]interface{}{}
  244. }
  245. wg.Wait()
  246. }
  247. func taskInfo(id string) {
  248. update := make(map[string]interface{})
  249. q := map[string]interface{}{"company_id": id}
  250. info, _ := Mgo.Find("company_change", q, nil, nil, false, -1, -1)
  251. if len(*info) > 0 {
  252. var changes []map[string]interface{}
  253. currentTime := time.Now().Unix()
  254. for _, v := range *info {
  255. item := make(map[string]interface{})
  256. item["change_field"] = v["change_field"]
  257. item["content_before"] = v["content_before"]
  258. item["content_after"] = v["content_after"]
  259. item["change_date"] = v["change_date"]
  260. setMark(item) //change_name_new
  261. changes = append(changes, item)
  262. }
  263. update["changes"] = changes
  264. update["update_time"] = currentTime
  265. }
  266. util.Debug(update)
  267. MgoMix.Update("qyxy_change", q, map[string]interface{}{"$set": update}, false, false)
  268. }
  269. func disposeFuc(maps []map[string]interface{}, tmp map[string]interface{}) {
  270. var changes []map[string]interface{}
  271. currentTime := time.Now().Unix()
  272. for _, v := range maps {
  273. item := make(map[string]interface{})
  274. item["change_field"] = v["change_field"]
  275. item["content_before"] = v["content_before"]
  276. item["content_after"] = v["content_after"]
  277. item["change_date"] = v["change_date"]
  278. setMark(item) //change_name_new
  279. changes = append(changes, item)
  280. }
  281. tmp["changes"] = changes
  282. tmp["create_time"] = currentTime
  283. tmp["update_time"] = currentTime
  284. MgoSaveCache <- tmp
  285. }
  286. //SaveData 存量保存
  287. func SaveData() {
  288. log.Println("Mgo Save...")
  289. arru := make([]map[string]interface{}, 200)
  290. indexu := 0
  291. for {
  292. select {
  293. case v := <-MgoSaveCache:
  294. arru[indexu] = v
  295. indexu++
  296. if indexu == 200 {
  297. SP <- true
  298. go func(arru []map[string]interface{}) {
  299. defer func() {
  300. <-SP
  301. }()
  302. MgoMix.SaveBulk(CollSave, arru...)
  303. }(arru)
  304. arru = make([]map[string]interface{}, 200)
  305. indexu = 0
  306. }
  307. case <-time.After(1000 * time.Millisecond):
  308. if indexu > 0 {
  309. SP <- true
  310. go func(arru []map[string]interface{}) {
  311. defer func() {
  312. <-SP
  313. }()
  314. MgoMix.SaveBulk(CollSave, arru...)
  315. }(arru[:indexu])
  316. arru = make([]map[string]interface{}, 200)
  317. indexu = 0
  318. }
  319. }
  320. }
  321. }
  322. func updateMethod() {
  323. arru := make([][]map[string]interface{}, 200)
  324. indexu := 0
  325. for {
  326. select {
  327. case v := <-updatePool:
  328. arru[indexu] = v
  329. indexu++
  330. if indexu == 200 {
  331. updateSp <- true
  332. go func(arru [][]map[string]interface{}) {
  333. defer func() {
  334. <-updateSp
  335. }()
  336. MgoMix.UpSertBulk(CollSave, arru...)
  337. }(arru)
  338. arru = make([][]map[string]interface{}, 200)
  339. indexu = 0
  340. }
  341. case <-time.After(1000 * time.Millisecond):
  342. if indexu > 0 {
  343. updateSp <- true
  344. go func(arru [][]map[string]interface{}) {
  345. defer func() {
  346. <-updateSp
  347. }()
  348. MgoMix.UpSertBulk(CollSave, arru...)
  349. }(arru[:indexu])
  350. arru = make([][]map[string]interface{}, 200)
  351. indexu = 0
  352. }
  353. }
  354. }
  355. }