task.go 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392
  1. package main
  2. import (
  3. util "app.yhyue.com/data_processing/common_utils"
  4. "app.yhyue.com/data_processing/common_utils/log"
  5. "encoding/json"
  6. "fieldproject_common/config"
  7. "fmt"
  8. es "gopkg.in/olivere/elastic.v2"
  9. "strconv"
  10. "sync"
  11. "time"
  12. )
  13. var province_map = map[string]string{
  14. "BJ": "北京", "TJ": "天津", "SH": "上海", "CQ": "重庆", "HB": "河北", "SX": "山西", "NMG": "内蒙古", "LN": "辽宁", "JL": "吉林",
  15. "HLJ": "黑龙江", "JS": "江苏", "ZJ": "浙江", "AH": "安徽", "FJ": "福建", "JX": "江西", "SD": "山东", "HEN": "河南", "HUB": "湖北",
  16. "HUN": "湖南", "GD": "广东", "GX": "广西", "HAIN": "海南", "SC": "四川", "GZ": "贵州", "YN": "云南", "XZ": "西藏", "SAX": "陕西",
  17. "GS": "甘肃", "QH": "青海", "NX": "宁夏", "XJ": "新疆",
  18. }
  19. // @Description 标讯数据
  20. // @Author J 2022/9/7 11:42
  21. func taskBiddingData1() {
  22. go SaveMethod()
  23. sess := MongoTool.GetMgoConn()
  24. defer MongoTool.DestoryMongoConn(sess)
  25. ch := make(chan bool, 10)
  26. wg := &sync.WaitGroup{}
  27. //q := map[string]interface{}{"_id": mongodb.StringTOBsonId("5a8d7f4840d2d9bbe8962002")}
  28. query := sess.DB(config.Conf.DB.Mongo.Dbname).C("bidding").Find(nil).Iter()
  29. count := 0
  30. for tmp := make(map[string]interface{}); query.Next(tmp); count++ {
  31. if count%20000 == 0 {
  32. log.Info(fmt.Sprintf("current --- %d", count))
  33. }
  34. ch <- true
  35. wg.Add(1)
  36. go func(tmp map[string]interface{}) {
  37. defer func() {
  38. <-ch
  39. wg.Done()
  40. }()
  41. if b := util.ObjToString(tmp["bid_field"]); b != "" {
  42. taskA(tmp)
  43. }
  44. }(tmp)
  45. tmp = make(map[string]interface{})
  46. }
  47. wg.Wait()
  48. log.Info(fmt.Sprintf("over --- %d", count))
  49. }
  50. // @Description 企业数据
  51. // @Author J 2022/8/23 09:08
  52. func taskCompanyData() {
  53. sess := MongoTool.GetMgoConn()
  54. defer MongoTool.DestoryMongoConn(sess)
  55. ch := make(chan bool, 10)
  56. wg := &sync.WaitGroup{}
  57. //log.Info(fmt.Sprintf("%d", MongoTool2.Count("zktest_mysql_company_info", nil)))
  58. var p1 []map[string]interface{}
  59. p1 = append(p1, map[string]interface{}{"$group": map[string]interface{}{"_id": "$company_id"}})
  60. query := sess.DB(config.Conf.DB.Mongo.Dbname).C("bidding_p_list_0907").Pipe(p1).Iter()
  61. //q := map[string]interface{}{"_id": mongodb.StringTOBsonId("61efb24b70f4a1409599badb")}
  62. //query := sess.DB(config.Conf.DB.Mongo2.Dbname).C("zktest_mysql_company_info").Find(nil).Iter()
  63. count := 0
  64. for tmp := make(map[string]interface{}); query.Next(tmp); count++ {
  65. if count%5000 == 0 {
  66. log.Info(fmt.Sprintf("current --- %d", count))
  67. }
  68. ch <- true
  69. wg.Add(1)
  70. go func(tmp map[string]interface{}) {
  71. defer func() {
  72. <-ch
  73. wg.Done()
  74. }()
  75. //taskC(tmp)
  76. taskB(util.ObjToString(tmp["_id"]))
  77. //info, _ := MongoTool.FindOne("qyxy_std", map[string]interface{}{"_id": util.ObjToString(tmp["company_id"])})
  78. //if len(*info) > 0 {
  79. // savePool <- *info
  80. //}
  81. }(tmp)
  82. tmp = make(map[string]interface{})
  83. }
  84. wg.Wait()
  85. log.Info(fmt.Sprintf("over --- %d", count))
  86. }
  87. // @Description 标讯数据企业与产品
  88. // @Author J 2022/8/24 14:37
  89. func taskA(tmp map[string]interface{}) {
  90. if tmp["entidlist"] != nil {
  91. saveMap := make(map[string]interface{})
  92. for _, eid := range tmp["entidlist"].([]interface{}) {
  93. if util.ObjToString(eid) != "" && util.ObjToString(eid) != "-" {
  94. saveMap["company_id"] = eid
  95. if tmp["purchasinglist"] != nil {
  96. for _, p1 := range tmp["purchasinglist"].([]interface{}) {
  97. p2 := p1.(map[string]interface{})
  98. if p2["itemname"] != nil {
  99. saveMap["itemname"] = p2["itemname"]
  100. if p2["brandname"] != nil {
  101. saveMap["brand"] = p2["brandname"]
  102. }
  103. if p2["model"] != nil {
  104. saveMap["model"] = p2["model"]
  105. }
  106. MongoTool.Save("bidding_p_list_0907", saveMap)
  107. //savePool <- saveMap
  108. }
  109. }
  110. }
  111. }
  112. }
  113. }
  114. }
  115. var company_field = []string{"company_id", "company_name", "company_code", "credit_no", "org_code", "tax_code", "establish_date", "legal_person", "legal_person_caption",
  116. "company_status", "company_type", "authority", "issue_date", "operation_startdate", "operation_enddate", "capital", "company_address", "business_scope", "cancel_date",
  117. "cancel_reason", "revoke_date", "revoke_reason", "legal_person_type", "real_capital", "en_name", "list_code", "area_code", "employee_no", "company_phone", "company_email",
  118. "website", "sourcetype"}
  119. // @Description 医疗企业
  120. // @Author J 2022/8/24 14:37
  121. func taskB(tid string) {
  122. saveM := make(map[string]interface{})
  123. info, _ := MongoTool1.FindOneByField("company_base", map[string]interface{}{"company_id": tid}, "")
  124. if len(*info) == 0 {
  125. info, _ = MongoTool1.FindOneByField("special_enterprise", map[string]interface{}{"company_id": tid}, "")
  126. }
  127. if len(*info) > 0 {
  128. for _, v := range company_field {
  129. if (*info)[v] != nil {
  130. saveM[v] = (*info)[v]
  131. }
  132. if v == "area_code" {
  133. std, _ := MongoTool.FindOneByField("qyxy_std", map[string]interface{}{"_id": tid}, map[string]interface{}{"company_area": 1, "company_city": 1, "company_district": 1, "website_url": 1, "capital": 1})
  134. if len(*std) > 0 {
  135. m := make(map[string]interface{})
  136. for k, v := range map[string]string{"company_district": "district", "company_city": "city", "company_area": "area"} {
  137. if v1 := util.ObjToString((*std)[k]); v1 != "" {
  138. m = map[string]interface{}{v: v1}
  139. info := MysqlB.FindOne("code_area", m, "", "")
  140. if info != nil && len(*info) > 0 {
  141. saveM["area_code"] = (*info)["code"]
  142. break
  143. }
  144. }
  145. }
  146. if saveM["area_code"] == nil {
  147. saveM["area_code"] = "000000"
  148. }
  149. if (*std)["website_url"] != nil && len(util.ObjToString((*std)["website_url"])) <= 255 {
  150. saveM["website"] = (*std)["website_url"]
  151. }
  152. if (*std)["company_phone"] != nil {
  153. saveM["company_phone"] = (*std)["company_phone"]
  154. }
  155. if (*std)["company_email"] != nil {
  156. saveM["company_email"] = (*std)["company_email"]
  157. }
  158. if (*std)["capital"] != nil {
  159. saveM["capital"] = util.ObjToString((*std)["capital"])
  160. }
  161. }
  162. } else if v == "employee_no" {
  163. rep, _ := MongoTool1.Find("annual_report_base", map[string]interface{}{"company_id": tid}, map[string]interface{}{"_id": -1}, "", false, -1, -1)
  164. if len(*rep) > 0 && util.ObjToString((*rep)[0]["employee_no"]) != "" {
  165. i, err := strconv.Atoi(util.ObjToString((*rep)[0]["employee_no"]))
  166. if err != nil {
  167. saveM[v] = i
  168. }
  169. }
  170. }
  171. //else if v == "website" {
  172. // info, _ := MongoTool.Find("annual_report_website", map[string]interface{}{"company_id": tid}, map[string]interface{}{"_id": -1}, "", false, -1, -1)
  173. // if len(*info) > 0 && util.ObjToString((*info)[0]["website_url"]) != "" {
  174. // saveM[v] = util.ObjToString((*info)[0]["website_url"])
  175. // }
  176. //}
  177. }
  178. saveM["comeintime"] = time.Now()
  179. saveM["updatetime"] = time.Now()
  180. saveM["sourcetype"] = 3
  181. MysqlB.Insert("company_baseinfo", saveM)
  182. //savePool <- saveM
  183. } else {
  184. util.Debug("company_id err", tid)
  185. }
  186. }
  187. func taskB_1(tid string) {
  188. saveM := make(map[string]interface{})
  189. info, _ := MongoTool1.FindOneByField("company_base", map[string]interface{}{"company_id": tid}, "")
  190. if len(*info) == 0 {
  191. info, _ = MongoTool1.FindOneByField("special_enterprise", map[string]interface{}{"company_id": tid}, "")
  192. }
  193. if len(*info) > 0 {
  194. for _, v := range company_field {
  195. if (*info)[v] != nil {
  196. saveM[v] = (*info)[v]
  197. }
  198. if v == "area_code" {
  199. if a := util.ObjToString((*info)["province_short"]); a != "" {
  200. m := map[string]interface{}{"area": province_map[a]}
  201. info := MysqlB.FindOne("code_area", m, "", "")
  202. if info != nil && len(*info) > 0 {
  203. saveM["area_code"] = (*info)["code"]
  204. }
  205. }
  206. } else if (*info)["capital"] != nil {
  207. text := util.ObjToString((*info)["capital"])
  208. capital := ObjToMoney(text)
  209. capital = capital / 10000
  210. if capital != 0 {
  211. saveM["capital"] = fmt.Sprint(capital)
  212. }
  213. }
  214. }
  215. saveM["comeintime"] = time.Now()
  216. saveM["updatetime"] = time.Now()
  217. saveM["sourcetype"] = 1
  218. MysqlB.Insert("company_baseinfo", saveM)
  219. //savePool <- saveM
  220. } else {
  221. util.Debug("company_id err", tid)
  222. }
  223. }
  224. func taskC(tmp map[string]interface{}) {
  225. tid := util.ObjToString(tmp["company_id"])
  226. saveM := make(map[string]interface{})
  227. for _, v := range company_field {
  228. if tmp[v] != nil {
  229. saveM[v] = tmp[v]
  230. }
  231. if v == "area_code" {
  232. std, _ := MongoTool.FindOneByField("qyxy_std", map[string]interface{}{"_id": tid}, map[string]interface{}{"company_area": 1, "company_city": 1, "company_district": 1, "website_url": 1, "company_phone": 1, "company_email": 1, "capital": 1})
  233. if len(*std) > 0 {
  234. m := make(map[string]interface{})
  235. for k, v := range map[string]string{"company_district": "district", "company_city": "city", "company_area": "area"} {
  236. if v1 := util.ObjToString((*std)[k]); v1 != "" {
  237. m = map[string]interface{}{v: v1}
  238. info := MysqlB.FindOne("code_area", m, "", "")
  239. if info != nil && len(*info) > 0 {
  240. saveM["area_code"] = (*info)["code"]
  241. break
  242. }
  243. }
  244. }
  245. if saveM["area_code"] == nil {
  246. saveM["area_code"] = "000000"
  247. }
  248. if (*std)["website_url"] != nil && len(util.ObjToString((*std)["website_url"])) <= 255 {
  249. saveM["website"] = (*std)["website_url"]
  250. }
  251. if (*std)["company_phone"] != nil {
  252. saveM["company_phone"] = (*std)["company_phone"]
  253. }
  254. if (*std)["company_email"] != nil {
  255. saveM["company_email"] = (*std)["company_email"]
  256. }
  257. if (*std)["capital"] != nil {
  258. saveM["capital"] = fmt.Sprint((*std)["capital"])
  259. }
  260. }
  261. } else if v == "employee_no" {
  262. rep, _ := MongoTool1.Find("annual_report_base", map[string]interface{}{"company_id": tid}, map[string]interface{}{"_id": -1}, "", false, -1, -1)
  263. if len(*rep) > 0 && util.ObjToString((*rep)[0]["employee_no"]) != "" {
  264. i, err := strconv.Atoi(util.ObjToString((*rep)[0]["employee_no"]))
  265. if err != nil {
  266. saveM[v] = i
  267. }
  268. }
  269. }
  270. }
  271. saveM["comeintime"] = time.Now()
  272. saveM["updatetime"] = time.Now()
  273. saveM["sourcetype"] = 2
  274. //MysqlB.Insert("company_business_model", map[string]interface{}{"company_id": tmp["company_id"],
  275. // "business_model": util.IntAll(tmp["business_type"]), "company_field_code": "0101", "comeintime": time.Now()})
  276. MysqlB.Insert("company_baseinfo", saveM)
  277. }
  278. func taskBiddingData() {
  279. client := Es.GetEsConn()
  280. defer Es.DestoryEsConn(client)
  281. wg := &sync.WaitGroup{}
  282. //lock := &sync.Mutex{}
  283. esquery := `{"query":{"bool":{"must":[],"must_not":[{"constant_score":{"filter":{"missing":{"field":"bid_field"}}}}],"should":[{"match_all":{}}]}},"from":0,"size":10,"sort":[],"facets":{}}`
  284. q := es.NewRawStringQuery(esquery)
  285. util.Debug(Es.Count("bidding", "bidding", q))
  286. countDocs := 0
  287. res, err := client.Scroll().Index("bidding").Type("bidding").Query(q).Size(200).Do() //查询一条获取游标
  288. if err == nil {
  289. taskInfoA(res, wg, &countDocs)
  290. scrollId := res.ScrollId
  291. for {
  292. searchResult, err := client.Scroll("1m").ScrollId(scrollId).Size(200).Do() //查询
  293. if err != nil {
  294. util.Debug("Es Search Data Error:", err)
  295. break
  296. }
  297. taskInfoA(searchResult, wg, &countDocs)
  298. scrollId = searchResult.ScrollId
  299. }
  300. wg.Wait()
  301. util.Debug("over---", countDocs)
  302. _, _ = client.ClearScroll().ScrollId(scrollId).Do() //清理游标
  303. } else {
  304. util.Debug(err)
  305. }
  306. c := make(chan bool, 1)
  307. <-c
  308. }
  309. func taskInfoA(searchResult *es.SearchResult, wg *sync.WaitGroup, countDocs *int) {
  310. for _, hit := range searchResult.Hits.Hits {
  311. //开始处理数据
  312. wg.Add(1)
  313. ChEs <- true
  314. go func(tmpHit *es.SearchHit) {
  315. defer func() {
  316. <-ChEs
  317. wg.Done()
  318. }()
  319. tmp := make(map[string]interface{})
  320. if json.Unmarshal(*tmpHit.Source, &tmp) == nil {
  321. id := util.ObjToString(tmp["_id"])
  322. if tmp["entidlist"] != nil && tmp["purchasinglist"] != nil {
  323. for _, eid := range util.ObjArrToStringArr(tmp["entidlist"].([]interface{})) {
  324. if util.ObjToString(eid) != "" && util.ObjToString(eid) != "-" {
  325. saveMap := make(map[string]interface{})
  326. saveMap["infoid"] = id
  327. saveMap["company_id"] = eid
  328. for _, p1 := range tmp["purchasinglist"].([]interface{}) {
  329. p2 := p1.(map[string]interface{})
  330. if p2["itemname"] != nil {
  331. b := method3(id, util.ObjToString(p2["itemname"]))
  332. if b {
  333. saveMap["itemname"] = p2["itemname"]
  334. if p2["brandname"] != nil {
  335. saveMap["brand"] = p2["brandname"]
  336. }
  337. if p2["model"] != nil {
  338. saveMap["model"] = p2["model"]
  339. }
  340. MongoTool.Save("bidding_p_list_0907", saveMap)
  341. }
  342. }
  343. }
  344. }
  345. }
  346. }
  347. }
  348. }(hit)
  349. *countDocs += 1
  350. if *countDocs%20000 == 0 {
  351. util.Debug("Current:", *countDocs)
  352. }
  353. }
  354. }
  355. func method3(id, itemname string) bool {
  356. q := map[string]interface{}{"infoid": id, "itemname": itemname}
  357. info := MysqlB.FindOne("bid_purchasinginfo", q, "id", "")
  358. if info != nil && len(*info) > 0 {
  359. q1 := map[string]interface{}{"bid_purchasing_id": (*info)["id"]}
  360. info1 := MysqlB.FindOne("bid_purchasing_field_record", q1, "id", "")
  361. if info1 != nil && len(*info1) > 0 {
  362. return true
  363. }
  364. }
  365. return false
  366. }