jy-data-extract/standardata/src/extractcity.go

package main

import (
	qu "qfw/util"

	"go.mongodb.org/mongo-driver/bson/primitive"

	"github.com/go-ego/gse"
)

//省
type Province struct {
	Name    string
	Brief   string
	Cap     string
	Captial *City
}

//市
type City struct {
	Name  string
	Brief string
	P     *Province
}

//区或县
type District struct {
	Name string
	C    *City
}

//街道
type Street struct {
	Name string
	D    *District
}

//村、社区、居委会
type Community struct {
	Name string
	S    *Street
}

//区或县简称对应的全称和市信息
type DistrictSimFull struct {
	SimName  string
	FullName string
	C        *City
}

//打分信息
type ScoreInfo struct {
	FullAreaScore     map[string]float64
	FullCityScore     map[string]float64
	FullDistrictScore map[string]float64
	SimAreaScore      map[string]float64
	SimCityScore      map[string]float64
	SimDistrictScore  map[string]float64
}

var (
	ProvinceMap          map[string]string             //省全称简称(key:浙江省 val:浙江)
	ProvinceBriefMap     map[string]*Province          //省简称对应的省信息(key:浙江 val:&Province{})
	CityMap              map[string]string             //市全称简称(key:杭州市 val:杭州)
	CityBriefMap         map[string]*City              //市简称对应的市信息(key:杭州 val:&City{})
	CityFullMap          map[string]*City              //市全称对应的市信息(key:杭州市 val:&City{})
	DistrictCityMap      map[string]*City              //
	NewDistrictCityMap   map[string][]*City            //区或县全称对应的city(全国有相同名称的区或县,这里对应的city用slice)
	DistrictSimAndAll    map[string]string             //区或县(key:简称 val:全称)
	NewDistrictSimAndAll map[string][]map[string]*City //区或县(key:简称 val:  相同简称的区全称:所在市)
	StreetDistrictMap    map[string]*District          //街道对应的区或县
	NewStreetDistrictMap map[string][]*District        //街道全称对应的区或县
	CommunityDistrictMap map[string][]*District        //村、居委会对应的区或县
	Trie_Full_Province   *Trie                         //省全称 省、直辖市、自治区
	Trie_Full_City       *Trie                         //市全称 地级市
	Trie_Full_District   *Trie                         //县全称 市辖区、县（旗）、县级市、自治县（自治旗）、特区、林区
	Trie_Full_Street     *Trie                         //街道、乡镇全称  镇、乡、民族乡、县辖区、街道
	Trie_Full_Community  *Trie                         //村/委员会全称  村、居委会
	Trie_Sim_Province    *Trie                         //省简称
	Trie_Sim_City        *Trie                         //市简称
	Trie_Sim_District    *Trie                         //县简称
	Trie_Fulls           []*Trie                       //所有全称
	Trie_Sims            []*Trie                       //所有简称
	Seg_SV               *gse.Segmenter                //分词
	// Seg_PCD              *gse.Segmenter                //分词
)

func InitInfo() {
	InitVar()  //初始化变量
	InitSeg()  //初始化分词
	InitCity() //初始化城市信息
}

//初始化变量
func InitVar() {
	defer qu.Catch()
	//初始化Trie
	//全称
	Trie_Full_Province = &Trie{}
	Trie_Full_City = &Trie{}
	Trie_Full_District = &Trie{}
	Trie_Full_Street = &Trie{}
	Trie_Full_Community = &Trie{}
	//简称
	Trie_Sim_Province = &Trie{}
	Trie_Sim_City = &Trie{}
	Trie_Sim_District = &Trie{}

	//初始化分词
	// Seg_PCD = &gse.Segmenter{}
	Seg_SV = &gse.Segmenter{}

	//初始化map
	ProvinceMap = make(map[string]string)
	CityMap = make(map[string]string)
	DistrictSimAndAll = make(map[string]string)
	NewDistrictSimAndAll = make(map[string][]map[string]*City)

	CityBriefMap = make(map[string]*City)
	CityFullMap = make(map[string]*City)
	ProvinceBriefMap = make(map[string]*Province)
	NewDistrictCityMap = make(map[string][]*City)

	NewStreetDistrictMap = make(map[string][]*District)
	CommunityDistrictMap = make(map[string][]*District)
}

//初始化分词
func InitSeg() {
	defer qu.Catch()
	// Seg_PCD.LoadDict("res/pcd.txt")
	Seg_SV.LoadDict("res/sv.txt")
}

//初始化城市信息
func InitCity() {
	defer qu.Catch()
	InitCityAll() //城市全称信息
	InitCitySim() //城市简称信息
	Trie_Fulls = []*Trie{Trie_Full_Province, Trie_Full_City, Trie_Full_District, Trie_Full_Street, Trie_Full_Community}
	Trie_Sims = []*Trie{Trie_Sim_Province, Trie_Sim_City, Trie_Sim_District}
}

func InitCityAll() {
	query := map[string]interface{}{
		"s_type":    "cityall",
		"delete":    false,
		"s_version": Version,
	}
	list, _ := MgoExt.Find(VersionColl, query, nil, nil)
	qu.Debug("all---", len(list))
	for _, v := range list {
		name := qu.ObjToString(v["s_name"])
		tmp := v["content"].(map[string]interface{})
		//加载省信息
		Trie_Full_Province.AddWords(name) //加入省全称Trie(k:浙江省)
		p := &Province{}
		p.Name = name                       //省全称:浙江省
		p.Brief = tmp["brief"].(string)     //省简称:浙江
		Trie_Sim_Province.AddWords(p.Brief) //加入省简称Trie(k:浙江)
		ProvinceMap[name] = p.Brief         //浙江省:浙江
		ProvinceBriefMap[p.Brief] = p       //浙江:省信息{}
		p.Cap = tmp["captial"].(string)     //省会(杭州)
		//加载市信息
		city, _ := tmp["city"].(map[string]interface{})
		for k, v := range city {
			Trie_Full_City.AddWords(k) //加入市全称Trie(k:杭州市)
			v1m, _ := v.(map[string]interface{})
			c := &City{}
			c.Name = k                      //市全称:杭州市
			c.Brief = v1m["brief"].(string) //市简称:杭州
			Trie_Sim_City.AddWords(c.Brief) //加入市简称Trie(k:杭州)
			CityMap[k] = c.Brief            //杭州市:杭州
			CityBriefMap[c.Brief] = c       //杭州:市信息{}
			CityFullMap[k] = c              //杭州市:市信息{}
			c.P = p
			if c.Name == p.Cap {
				p.Captial = c //加载province中的省会市信息{}
			}
			//区县
			districtmap, _ := v1m["area"].(map[string]interface{}) //区或县
			for district, streets := range districtmap {
				d := &District{}
				d.Name = district
				d.C = c
				//省直辖市，河南济源市没有区一级，目前区一级写的还是济源市
				//匹配时，如果匹配到区，拿区和市比对，相同则代表是省直辖市，不要区一级
				Trie_Full_District.AddWords(district) //加入区或县全称Trie
				ctmp := NewDistrictCityMap[district]
				if len(ctmp) == 0 {
					tmpcarr := []*City{c}
					NewDistrictCityMap[district] = tmpcarr
				} else {
					NewDistrictCityMap[district] = append(NewDistrictCityMap[district], c)
				}
				//街道
				streetmap, _ := streets.(map[string]interface{})
				for street, communitys := range streetmap {
					s := &Street{}
					s.Name = street
					s.D = d
					Trie_Full_Street.AddWords(street) //加入街道全称Trie
					dtmp := NewStreetDistrictMap[street]
					if len(dtmp) == 0 {
						tmpdarr := []*District{d}
						NewStreetDistrictMap[street] = tmpdarr
					} else {
						NewStreetDistrictMap[street] = append(NewStreetDistrictMap[street], d)
					}
					//村、居委会
					for _, ct_tmp := range communitys.(primitive.A) {
						ct := qu.ObjToString(ct_tmp)
						Trie_Full_Community.AddWords(ct) //加入居委会、村全称Trie
						cttmp := CommunityDistrictMap[ct]
						if len(cttmp) == 0 {
							tmpdarr := []*District{d}
							CommunityDistrictMap[ct] = tmpdarr
						} else {
							CommunityDistrictMap[ct] = append(CommunityDistrictMap[ct], d)
						}
					}
				}
			}
		}
	}
}

func InitCitySim() {
	defer qu.Catch()
	query := map[string]interface{}{
		"s_type":    "citysim",
		"delete":    false,
		"s_version": Version,
	}
	list, _ := MgoExt.Find(VersionColl, query, nil, nil)
	qu.Debug("sim---", len(list))
	for _, v := range list {
		tmp := v["content"].(map[string]interface{})
		city, _ := tmp["city"].(map[string]interface{})
		for _, v1 := range city {
			v1m, _ := v1.(map[string]interface{})
			cb := v1m["brief"].(string)                 //市简称
			arr := v1m["area"].(map[string]interface{}) //区或县简称
			for districtsim, districtall := range arr {
				dfullstr, _ := districtall.(string)
				Trie_Sim_District.AddWords(districtsim) //加入区或县简称Trie
				c := CityBriefMap[cb]
				dfullarr := NewDistrictSimAndAll[districtsim]
				dfullcity := map[string]*City{dfullstr: c}
				if len(dfullarr) == 0 {
					tmparr := []map[string]*City{dfullcity}
					NewDistrictSimAndAll[districtsim] = tmparr
				} else {
					NewDistrictSimAndAll[districtsim] = append(NewDistrictSimAndAll[districtsim], dfullcity)
				}
			}
		}
	}
}