data_ai/clean/c_region.go

package clean

import (
	"data_ai/ul"
	"fmt"
	"strings"
)

// 清洗地域
func CleanRegion(area string, city string) (string, string) {
	area = fieldReg1.ReplaceAllString(area, "")
	city = fieldReg1.ReplaceAllString(city, "")
	if area == "无" {
		area = ""
	}
	if city == "无" {
		city = ""
	}
	new_area, new_city := "", ""
	//确定省份
	if area != "" {
		if ul.S_ProvinceDict[area] != nil {
			new_area = area
		} else {
			for k, _ := range ul.S_ProvinceDict {
				if strings.Contains(area, k) {
					new_area = k
					break
				}
			}
		}
	}
	//确认城市
	if city != "" {
		//校验城市
		if ul.S_CityDict[city] != nil {
			new_city = city
			if new_area == "" {
				new_area = ul.S_CityDict[city][0].P_Name
			}
		} else {
			for k, v := range ul.S_CityDict {
				if strings.Contains(k, city) || strings.Contains(city, k) {
					for _, v1 := range v {
						if new_area == "" {
							new_area = v1.P_Name
							new_city = k
							break
						} else {
							if new_area == v1.P_Name {
								new_city = k
								break
							}
						}
					}
				}
			}
		}
		//校验区县
		if ul.S_DistrictDict[city] != nil {
			districts := ul.S_DistrictDict[city]
			for _, v := range districts {
				if new_area == "" {
					new_area = v.P_Name
					new_city = v.C_Name
					break
				} else {
					if new_area == v.P_Name {
						new_city = v.C_Name
						break
					}
				}
			}
		} else {
			//判定为简称区县，补充在校验...
			arr := []string{}
			arr = append(arr, fmt.Sprintf("%s区", city))
			arr = append(arr, fmt.Sprintf("%s县", city))
			arr = append(arr, fmt.Sprintf("%s市", city))
			for _, v := range arr {
				if ul.S_DistrictDict[v] != nil {
					districts := ul.S_DistrictDict[v]
					for _, v1 := range districts {
						if new_area == "" {
							new_area = v1.P_Name
							new_city = v1.C_Name
							break
						} else {
							if new_area == v1.P_Name {
								new_city = v1.C_Name
								break
							}
						}
					}
				}
			}
		}
	}

	if new_area == "北京" && new_city == "" {
		new_city = "北京市"
	} else if new_area == "天津" && new_city == "" {
		new_city = "天津市"
	} else if new_area == "上海" && new_city == "" {
		new_city = "上海市"
	} else if new_area == "重庆" && new_city == "" {
		new_city = "重庆市"
	} else {

	}
	return new_area, new_city
}