c_region.go 2.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104
  1. package clean
  2. import (
  3. "data_ai/ul"
  4. "fmt"
  5. "strings"
  6. )
  7. // 清洗地域
  8. func CleanRegion(area string, city string) (string, string) {
  9. area = fieldReg1.ReplaceAllString(area, "")
  10. city = fieldReg1.ReplaceAllString(city, "")
  11. if area == "无" {
  12. area = ""
  13. }
  14. if city == "无" {
  15. city = ""
  16. }
  17. new_area, new_city := "", ""
  18. //确定省份
  19. if area != "" {
  20. if ul.S_ProvinceDict[area] != nil {
  21. new_area = area
  22. } else {
  23. for k, _ := range ul.S_ProvinceDict {
  24. if strings.Contains(area, k) {
  25. new_area = k
  26. break
  27. }
  28. }
  29. }
  30. }
  31. //确认城市-从下往上补
  32. if city != "" {
  33. if ul.S_CityDict[city] != nil {
  34. new_city = city
  35. new_area = ul.S_CityDict[city][0].P_Name
  36. } else { //是否存在包含关系
  37. for k, v := range ul.S_CityDict {
  38. if strings.Contains(k, city) || strings.Contains(city, k) {
  39. for _, v1 := range v {
  40. new_area = v1.P_Name
  41. new_city = k
  42. }
  43. }
  44. }
  45. }
  46. if new_area != "" && new_city != "" {
  47. return new_area, new_city
  48. }
  49. //可能抽取区县信息
  50. if ul.S_DistrictDict[city] != nil {
  51. districts := ul.S_DistrictDict[city]
  52. for _, v := range districts {
  53. if new_area == "" {
  54. new_area = v.P_Name
  55. new_city = v.C_Name
  56. return new_area, new_city
  57. } else {
  58. if new_area == v.P_Name {
  59. new_city = v.C_Name
  60. return new_area, new_city
  61. }
  62. }
  63. }
  64. } else {
  65. //判定为简称区县,补充在校验...
  66. arr := []string{}
  67. arr = append(arr, fmt.Sprintf("%s区", city))
  68. arr = append(arr, fmt.Sprintf("%s县", city))
  69. arr = append(arr, fmt.Sprintf("%s市", city))
  70. for _, v := range arr {
  71. if ul.S_DistrictDict[v] != nil {
  72. districts := ul.S_DistrictDict[v]
  73. for _, v1 := range districts {
  74. if new_area == "" {
  75. new_area = v1.P_Name
  76. new_city = v1.C_Name
  77. break
  78. } else {
  79. if new_area == v1.P_Name {
  80. new_city = v1.C_Name
  81. break
  82. }
  83. }
  84. }
  85. }
  86. }
  87. }
  88. }
  89. if new_area == "北京" && new_city == "" {
  90. new_city = "北京市"
  91. } else if new_area == "天津" && new_city == "" {
  92. new_city = "天津市"
  93. } else if new_area == "上海" && new_city == "" {
  94. new_city = "上海市"
  95. } else if new_area == "重庆" && new_city == "" {
  96. new_city = "重庆市"
  97. } else {
  98. }
  99. return new_area, new_city
  100. }