task.go 4.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152
  1. package main
  2. import (
  3. "context"
  4. "fmt"
  5. //"github.com/cron"
  6. "github.com/wangbin/jiebago"
  7. "go.mongodb.org/mongo-driver/bson"
  8. "go.mongodb.org/mongo-driver/mongo/options"
  9. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  10. "log"
  11. "math/rand"
  12. "regexp"
  13. "strconv"
  14. "strings"
  15. "time"
  16. )
  17. var (
  18. AreaFiled = []string{"credit_no", "company_code"}
  19. WordsArr = []string{"研发", "研制", "开发", "生产", "制造", "制作", "加工", "种植"}
  20. seg jiebago.Segmenter
  21. regPre, _ = regexp.Compile(`^(.+[省|市|区|县|州])?(.+)`)
  22. seed = 188
  23. )
  24. // 企业基本信息
  25. var company_base = []string{"company_name", "company_code", "credit_no", "org_code", "legal_person", "company_status",
  26. "authority", "establish_date", "issue_date", "operation_startdate", "operation_enddate", "capital", "company_type",
  27. "company_status", "company_address", "business_scope", "cancel_date", "cancel_reason", "revoke_date", "revoke_reason",
  28. "legal_person_type", "real_capital", "en_name", "list_code", "tax_code", "use_flag",
  29. }
  30. var province_map = map[string]string{
  31. "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",
  32. "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",
  33. "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",
  34. "GS": "甘肃", "QH": "青海", "NX": "宁夏", "XJ": "新疆",
  35. }
  36. // company_shortname 企业简称
  37. func getStName(name string) string {
  38. regnames := regPre.FindStringSubmatch(name)
  39. lenth := len(regnames)
  40. if lenth < 1 {
  41. return ""
  42. }
  43. newstr := regnames[lenth-1]
  44. ch := seg.Cut(newstr, false)
  45. val := []string{}
  46. for word := range ch {
  47. val = append(val, word)
  48. }
  49. name2 := ""
  50. for _, v := range val {
  51. name2 = name2 + v
  52. if len([]rune(name2)) >= 4 {
  53. break
  54. }
  55. }
  56. return name2
  57. }
  58. func DealMemberNo(memberNo string) (no int) {
  59. memberNo = strings.Trim(memberNo, " ")
  60. if memberNo == "" {
  61. return no
  62. }
  63. if memberNo == "企业选择不公示" {
  64. return no
  65. }
  66. if strings.HasSuffix(memberNo, "人") {
  67. cou := strings.Split(memberNo, "人")[0]
  68. count, _ := strconv.Atoi(cou)
  69. return count
  70. }
  71. no, _ = strconv.Atoi(memberNo)
  72. return no
  73. }
  74. //updateStd 根据qyxy_std 表,更新 nseo_id
  75. func updateStd() {
  76. id := int64(GF.Env.Autoid) //起始 autoid
  77. //生成seoid
  78. //31395235740
  79. startSeoId := int64(GF.Env.Seoid) //起始 nseo_id
  80. rand.Seed(time.Now().UnixNano())
  81. count := int64(0)
  82. //var wg sync.WaitGroup
  83. sess := MongoTool.GetMgoConn()
  84. defer MongoTool.DestoryMongoConn(sess)
  85. ctx, _ := context.WithTimeout(context.Background(), 99999*time.Hour)
  86. coll := sess.M.C.Database("mixdata").Collection("qyxy_std")
  87. find := options.Find().SetBatchSize(200).SetSort(bson.D{bson.E{"autoid", 1}}).SetProjection(bson.M{"_id": 1, "autoid": 1, "establish_date": 1, "create_time_msql": 1})
  88. cur, err := coll.Find(ctx, bson.M{"autoid": bson.M{"$gt": id}}, find)
  89. if err != nil {
  90. log.Println("mgo find err", err.Error())
  91. return
  92. }
  93. for tmp := make(map[string]interface{}); cur.Next(ctx); {
  94. count++
  95. if cur != nil {
  96. cur.Decode(&tmp)
  97. id = util.Int64All(tmp["autoid"])
  98. ed, _ := tmp["establish_date"].(string)
  99. pre := ""
  100. if len(ed) == 10 {
  101. pre = ed[2:4] + ed[5:7] + ed[8:10]
  102. } else {
  103. cd, _ := tmp["create_time_msql"].(string)
  104. if len(cd) > 9 {
  105. pre = cd[2:4] + cd[5:7] + cd[8:10]
  106. }
  107. }
  108. startSeoId += int64(rand.Intn(seed) + 2)
  109. nseo_id := fmt.Sprintf("%s%d", pre, startSeoId)
  110. update := make(map[string]interface{})
  111. update["$set"] = bson.M{
  112. "nseo_id": nseo_id,
  113. "autoid": id,
  114. }
  115. where := map[string]interface{}{
  116. "_id": tmp["_id"],
  117. }
  118. go MongoTool.Update("qyxy_std", where, update, true, false)
  119. if count%10000 == 0 {
  120. log.Println("current,id", count, id, startSeoId, nseo_id)
  121. }
  122. tmp = make(map[string]interface{})
  123. } else {
  124. cur.Close(ctx)
  125. break
  126. }
  127. }
  128. log.Println("over ----;autoid: ", id, "seo_id:", startSeoId)
  129. time.Sleep(20 * time.Minute)
  130. }