data_extract/src/jy/extract/extractcity_way.go

package extract

import (
	qu "qfw/util"
)

//最终确认确认指定地域
func ConfirmUniqueRegionInfo(regions map[string]map[string]map[string]string, area *string, city *string, district *string) bool {
	if len(regions) > 1 || len(regions) == 0 {
		return false
	}
	for k, v := range regions {
		*area = k
		if len(v) == 1 {
			for k1, v1 := range v {
				*city = k1
				if len(v1) == 1 {
					for k2, _ := range v1 {
						*district = k2
					}
				}
			}
		}
	}
	if *area != "" && *city != "" && *district != "" {
		return true
	}
	return false
}

//完整信息
func CompleteRegionInfo(area *string, city *string, district *string) {
	if *area == "北京" {
		*city = "北京市"
		if *district == "北京朝阳" { //特殊情况(北京朝阳中西医结合急诊抢救中心:5a84079740d2d9bbe88bad90)
			*district = "朝阳区"
		}
	} else if *area == "天津" {
		*city = "天津市"
	} else if *area == "上海" {
		*city = "上海市"
	} else if *area == "重庆" {
		*city = "重庆市"
	}
	if *area == "" {
		*area = "全国"
		*city = ""
		*district = ""
	}
}

//根据词获取所有的地域
func (e *ExtractTask) takeRegionsFromWords(text string) []map[string]string {
	regions := []map[string]string{}
	//全称匹配
	for pos_full, trie_full := range e.Trie_Fulls {
		if trie_full.Get(text) {
			if pos_full == 0 {
				if province := e.ProvinceMap[text]; province != "" {
					regions = append(regions, map[string]string{"area": province, "city": "", "district": ""})
				}
			} else if pos_full == 1 {
				if data := e.CityFullMap[text]; data != nil {
					if data.P.Brief != "" && data.Name != "" {
						regions = append(regions, map[string]string{"area": data.P.Brief, "city": data.Name, "district": ""})
					}
				}
			} else if pos_full == 2 {
				citys := e.DistrictCityMap[text]
				for _, c := range citys {
					if c.P.Brief != "" && c.Name != "" && text != "" {
						regions = append(regions, map[string]string{"area": c.P.Brief, "city": c.Name, "district": text})
					}
				}
			}
		}
	}
	//简称匹配
	for pos_sim, trie_sim := range e.Trie_Sims {
		if trie_sim.Get(text) {
			if pos_sim == 0 {
				if text != "" {
					regions = append(regions, map[string]string{"area": text, "city": "", "district": ""})
				}
			} else if pos_sim == 1 {
				if csMap := e.CityBriefMap[text]; csMap != nil {
					if csMap.P.Brief != "" && csMap.Name != "" {
						regions = append(regions, map[string]string{"area": csMap.P.Brief, "city": csMap.Name, "district": ""})
					}
				}
			} else if pos_sim == 2 {
				citysArr := e.DistrictSimAndAll[text]
				for _, full_citys := range citysArr {
					for district, c := range full_citys {
						if c == nil || c.P == nil || c.Name == "" {
							continue
						}
						if c.P.Brief != "" && c.Name != "" && district != "" {
							regions = append(regions, map[string]string{"area": c.P.Brief, "city": c.Name, "district": district})
						}
					}
				}
			}
		}
	}
	return regions
}

//文本取地域   from  1~jsondata文本   2~其他文本
func (e *ExtractTask) GetRegionFromText(text string, regions *map[string]map[string]map[string]string, from int) []map[string]interface{} {
	regionValues := []map[string]interface{}{}
	if text == "" {
		return regionValues
	}
	wordsArr := []string{}
	if from == 1 {
		wordsArr = e.Seg_PCD.Cut(text, true)
	} else if from == 2 {
		wordsArr = e.Seg_SV.Cut(text, true)
	}
	for _, word := range wordsArr {
		regionArr := e.takeRegionsFromWords(word)
		for _, v := range regionArr {
			area := qu.ObjToString(v["area"])
			city := qu.ObjToString(v["city"])
			district := qu.ObjToString(v["district"])
			UpdateRegionsInfo(area, city, district, regions)
			regionValues = append(regionValues, map[string]interface{}{"area": area, "city": city, "district": district})
		}
	}
	return regionValues
}

//更新方法
func UpdateRegionsInfo(area, city, district string, regions *map[string]map[string]map[string]string) {
	if (*regions)[area] == nil {
		city_info := map[string]map[string]string{}
		district_info := map[string]string{}
		if city != "" {
			if district != "" {
				district_info[district] = district
			}
			city_info[city] = district_info
		}
		(*regions)[area] = city_info //新增
	} else {
		city_info := (*regions)[area]
		if city != "" {
			district_info := map[string]string{}
			if city_info[city] != nil {
				district_info = city_info[city]
			}
			if district != "" {
				district_info[district] = district
			}
			city_info[city] = district_info
			(*regions)[area] = city_info
		}
	}
}

//同组合并后合理性校验
func ReasonableGroupRegionInfo(datas map[string]map[string]map[string]string) map[string]map[string]map[string]string {
	if len(datas) > 2 || len(datas) == 0 { //省份超限，无效
		return map[string]map[string]map[string]string{}
	}
	uncity, undistrict := 0, 0
	for _, v := range datas {
		uncity += len(v)
		for _, v1 := range v {
			undistrict += len(v1)
		}
	}

	if uncity > 3 {
		regions_1 := map[string]map[string]map[string]string{}
		for k, v := range datas {
			city_info := map[string]map[string]string{}
			if len(v) == 1 {
				city_info = v
			}
			regions_1[k] = city_info
		}
		//计算当前
		uncity_district := 0
		for _, v := range regions_1 {
			for _, v1 := range v {
				uncity_district += len(v1)
			}
		}
		if uncity_district > 3 {
			regions_2 := map[string]map[string]map[string]string{}
			for k, v := range regions_1 {
				city_info := map[string]map[string]string{}
				for k1, v1 := range v {
					district_info := map[string]string{}
					if len(v1) == 1 {
						district_info = v1
					}
					city_info[k1] = district_info
				}
				regions_2[k] = city_info
			}
			return regions_2
		}
		return regions_1
	}

	if undistrict > 3 {
		new_regions := map[string]map[string]map[string]string{}
		for k, v := range datas {
			city_info := map[string]map[string]string{}
			for k1, v1 := range v {
				district_info := map[string]string{}
				if len(v1) == 1 {
					district_info = v1
				}
				city_info[k1] = district_info
			}
			new_regions[k] = city_info
		}
		return new_regions
	}
	return datas
}

//两组比对~找寻补充，排除数据
func AnalysisIsUniqueInfo(regions map[string]map[string]map[string]string, all_regions *map[string]map[string]map[string]string) {
	if len(regions) == 0 {
		return
	}
	if len(*all_regions) == 0 {
		*all_regions = regions
		return
	}
	regionsArr := splitRegionsInfos(regions)          //目标数据
	all_regionsArr := splitRegionsInfos(*all_regions) //源数据
	new_all_regionsArr := []map[string]string{}       //新数据
	for _, info := range regionsArr {
		area := qu.ObjToString(info["area"])
		if (*all_regions)[area] == nil {
			continue
		}
		unmatchInfo1 := ScreenOutReasonableRegionInfo(info, &all_regionsArr, &new_all_regionsArr)
		if unmatchInfo1 != nil { //降级匹配~最多二级
			unmatchInfo2 := ScreenOutReasonableRegionInfo(unmatchInfo1, &all_regionsArr, &new_all_regionsArr)
			if unmatchInfo2 != nil { //降级匹配~最多一级
				ScreenOutReasonableRegionInfo(unmatchInfo2, &all_regionsArr, &new_all_regionsArr)
			}
		}
	}
	//根据最新有效地域组~重新构建所有信息
	reset_regions_infos := map[string]map[string]map[string]string{}
	if len(new_all_regionsArr) > 0 {
		for _, v := range new_all_regionsArr {
			area := qu.ObjToString(v["area"])
			city := qu.ObjToString(v["city"])
			district := qu.ObjToString(v["district"])
			UpdateRegionsInfo(area, city, district, &reset_regions_infos)
		}
		*all_regions = reset_regions_infos
	}
}

//选取规则方法
func ScreenOutReasonableRegionInfo(info map[string]string, regions_infosArr *[]map[string]string, new_regions_infosArr *[]map[string]string) map[string]string {
	area := qu.ObjToString(info["area"])
	city := qu.ObjToString(info["city"])
	district := qu.ObjToString(info["district"])
	is_Exist := false
	for _, s := range *regions_infosArr {
		s_area := qu.ObjToString(s["area"])
		s_city := qu.ObjToString(s["city"])
		s_district := qu.ObjToString(s["district"])
		if s_area == area && s_city == city && s_district == district {
			is_Exist = true
			*new_regions_infosArr = append(*new_regions_infosArr, info)
		} else {
			//判断是否为补充原则
			if area != "" && city != "" && district != "" { //3级补2级 3级补2级
				if s_area == area && s_city == city && s_district == "" {
					is_Exist = true
					*new_regions_infosArr = append(*new_regions_infosArr, info)
				} else if s_area == area && s_city == "" && s_district == "" {
					is_Exist = true
					*new_regions_infosArr = append(*new_regions_infosArr, info)
				}
			} else if area != "" && city != "" && district == "" { //2级补1级
				if s_area == area && s_city == "" {
					is_Exist = true
					*new_regions_infosArr = append(*new_regions_infosArr, info)
				}
			}
		}
	}

	//若没有补充~针对二三级 再次进行去掉末位一级，再次进行一轮比对
	if !is_Exist {
		if area != "" && city != "" && district != "" {
			return map[string]string{"area": area, "city": city, "district": ""}
		}
		if area != "" && city != "" && district == "" {
			return map[string]string{"area": area, "city": "", "district": ""}
		}
	}
	return nil
}

//拆分地域数据~目的更好的合并选取
func splitRegionsInfos(infos map[string]map[string]map[string]string) []map[string]string {
	infosArr := []map[string]string{}
	for k, v := range infos {
		if len(v) > 0 {
			for k1, v1 := range v {
				if len(v1) > 0 {
					for k2, _ := range v1 {
						infosArr = append(infosArr, map[string]string{"area": k, "city": k1, "district": k2})
					}
				} else {
					infosArr = append(infosArr, map[string]string{"area": k, "city": k1, "district": ""})
				}
			}
		} else {
			infosArr = append(infosArr, map[string]string{"area": k, "city": "", "district": ""})
		}
	}
	return infosArr
}

//日志流程记录~组级别
func LogProcessRecordingForGroupInfo(key string, valueArr []string, fieldInfos map[string]interface{}, groupInfos map[string]map[string]map[string]string, finallyInfos map[string]map[string]map[string]string, logRecordInfo *[]map[string]interface{}) {
	groupArr := splitRegionsInfos(groupInfos)
	finalluArr := splitRegionsInfos(finallyInfos)
	data := map[string]interface{}{
		key + "_value":   valueArr,
		key + "_group":   groupArr,
		"finally_region": finalluArr,
	}
	for k, v := range fieldInfos {
		data[k] = v
	}
	*logRecordInfo = append(*logRecordInfo, data)
}

//日志流程记录~初步
func LogProcessRecordingForTentative(key string, valueArr interface{}, finallyInfos map[string]map[string]map[string]string, logRecordInfo *[]map[string]interface{}) {
	finallyArr := splitRegionsInfos(finallyInfos)
	data := map[string]interface{}{
		key + "_value":   valueArr,
		"finally_region": finallyArr,
	}
	*logRecordInfo = append(*logRecordInfo, data)
}

//同组合并的地域数据
//func MergeGroupRegionInfo(datas_1, datas_2 map[string]map[string]map[string]string) map[string]map[string]map[string]string {
//	regions := map[string]map[string]map[string]string{}
//	if len(datas_1) > 0 && len(datas_2) == 0 {
//		return datas_1
//	}
//	if len(datas_2) > 0 && len(datas_1) == 0 {
//		return datas_2
//	}
//	for k, v := range datas_1 {
//		area, city, district := "", "", ""
//		area = k
//		if len(v) > 0 {
//			for k1, v1 := range v {
//				city = k1
//				if len(v1) > 0 {
//					for k2, _ := range v1 {
//						district = k2
//						UpdateRegionsInfo(area, city, district, &regions)
//					}
//				} else {
//					UpdateRegionsInfo(area, city, district, &regions)
//				}
//			}
//		} else {
//			UpdateRegionsInfo(area, city, district, &regions)
//		}
//	}
//
//	for k, v := range datas_2 {
//		area, city, district := "", "", ""
//		area = k
//		if len(v) > 0 {
//			for k1, v1 := range v {
//				city = k1
//				if len(v1) > 0 {
//					for k2, _ := range v1 {
//						district = k2
//						UpdateRegionsInfo(area, city, district, &regions)
//					}
//				} else {
//					UpdateRegionsInfo(area, city, district, &regions)
//				}
//			}
//		} else {
//			UpdateRegionsInfo(area, city, district, &regions)
//		}
//	}
//	return regions
//}