Java实现根据地址智能识别省市区县

import lombok.extern.slf4j.slf4j;
import org.apache.commons.lang3.stringutils;
 
import java.util.*;
import java.util.regex.matcher;
import java.util.regex.pattern;
 
/**
 * 身份证信息分析工具类util
 */
@slf4j
public class analysisidcardutil {
 
    /**
     * 获取详细地址
     *
     * @param fulladdress 身份证完整地址
     */
    public static string getdetailedaddress(string fulladdress) {
        string[] addressarrays = splicedetailedaddress(fulladdress);
        return addressarrays[addressarrays.length - 1];
    }
 
    /**
     * 获取省市区地址，如：安徽省合肥市包河区
     *
     * @param fulladdress 身份证完整地址
     */
    public static string getpcaaddress(string fulladdress) {
        string[] addressarrays = splicedetailedaddress(fulladdress);
        stringbuilder areabuffer = new stringbuilder();
        for (int i = 0; i < addressarrays.length - 1; i++) {
            if (stringutils.isnotempty(addressarrays[i])) {
                areabuffer.append(addressarrays[i]);
            }
        }
        return areabuffer.tostring();
    }
 
    /**
     * 身份证完成地址拆分为[省、市、区、详细地址]数组，如：[安徽省, 合肥市, 包河区, 幸福大街001号]
     *
     * @param fulladdress 身份证完整地址
     */
    public static string[] splicedetailedaddress(string fulladdress) {
        string[] arr = new string[4];
        try {
            string tempstr = fulladdress;
            // 省
            string province = null;
            int provinceidx = processprovince(tempstr);
            if (provinceidx > -1) {
                province = tempstr.substring(0, provinceidx + 1);
                tempstr = tempstr.substring(provinceidx + 1);
            }
            // 市
            string city = null;
            int cityidx = processcity(tempstr);
            if (cityidx > -1) {
                city = tempstr.substring(0, cityidx + 1);
                tempstr = tempstr.substring(cityidx + 1);
            }
            // 区
            string county = null;
            int countyidx = processcounty(tempstr);
            if (countyidx > -1) {
                county = tempstr.substring(0, countyidx + 1);
                tempstr = tempstr.substring(countyidx + 1);
            }
            string street = tempstr;
            arr[0] = province;
            arr[1] = city;
            arr[2] = county;
            arr[3] = street;
        } catch (exception e) {
            log.error("身份证详细地址转义失败：{}", e.getmessage());
        }
        return arr;
    }
 
    /**
     * 拆分身份证完整地址中的省份信息
     *
     * @param address 地址
     */
    private static int processprovince(string address) {
        int[] idx = new int[3];
        int provinceidx;
        if ((provinceidx = address.indexof("省")) > -1) {
            idx[0] = provinceidx;
        }
        if ((provinceidx = address.indexof("市")) > -1) {
            idx[1] = provinceidx;
        }
        if ((provinceidx = address.indexof("区")) > -1) {
            idx[2] = provinceidx;
        }
        arrays.sort(idx);
        for (int j : idx) {
            if (j > 0) {
                return j;
            }
        }
        return provinceidx;
    }
 
    /**
     * 拆分身份证完整地址中的市区/县/自治州信息
     *
     * @param address 地址
     */
    private static int processcity(string address) {
        int[] idx = new int[7];
        int cityidx = -1;
        if ((cityidx = address.indexof("县")) > -1) {
            idx[0] = cityidx;
        }
        if ((cityidx = address.indexof("自治州")) > -1) {
            idx[1] = cityidx + 2;
        }
        if ((cityidx = address.indexof("市辖区")) > -1) {
            idx[2] = cityidx + 2;
        }
        if ((cityidx = address.indexof("市")) > -1) {
            idx[3] = cityidx;
        }
        if ((cityidx = address.indexof("区")) > -1) {
            idx[4] = cityidx;
        }
        if ((cityidx = address.indexof("地区")) > -1) {
            idx[5] = cityidx + 1;
        }
        if ((cityidx = address.indexof("盟")) > -1) {
            idx[6] = cityidx;
        }
        arrays.sort(idx);
        for (int j : idx) {
            if (j > 0) {
                return j;
            }
        }
        return cityidx;
    }
 
    /**
     * 拆分身份证完整地址中的县/旗/岛信息
     *
     * @param address 地址
     */
    private static int processcounty(string address) {
        int[] idx = new int[6];
        int countyidx;
        if ((countyidx = address.indexof("县")) > -1) {
            idx[0] = countyidx;
        }
        if ((countyidx = address.indexof("旗")) > -1) {
            idx[1] = countyidx;
        }
        if ((countyidx = address.indexof("海域")) > -1) {
            idx[2] = countyidx + 1;
        }
        if ((countyidx = address.indexof("市")) > -1) {
            idx[3] = countyidx;
        }
        if ((countyidx = address.indexof("区")) > -1) {
            idx[4] = countyidx;
        }
        if ((countyidx = address.indexof("岛")) > -1) {
            idx[5] = countyidx;
        }
        arrays.sort(idx);
        for (int j : idx) {
            if (j > 0) {
                return j;
            }
        }
        return countyidx;
    }
 
    /**
     * 身份证地址提取省市区
     *
     * @param fulladdress 身份证完整地址
     */
    public static map<string, string> addressresolution(string fulladdress) {
        // 定义正则
        string regex = "(?<province>[^省]+自治区|.*?省|.*?行政区|.*?市)(?<city>[^市]+自治州|.*?地区|.*?行政单位|.+盟|市辖区|.*?市|.*?县)(?<area>[^县]+县|.+区|.+市|.+旗|.+海域|.+岛)?(?<town>[^区]+区|.+镇)?(?<detail>.*)";
 
        matcher matcher = pattern.compile(regex).matcher(fulladdress);
        string province, city, area, town, detail;
        map<string, string> map = new linkedhashmap<>();
 
        while (matcher.find()) {
            province = matcher.group("province");
            map.put("province", stringutils.isempty(province) ? "" : province.trim());
 
            city = matcher.group("city");
            map.put("city", stringutils.isempty(city) ? "" : city.trim());
 
            area = matcher.group("area");
            map.put("area", stringutils.isempty(area) ? "" : area.trim());
 
            town = matcher.group("town");
            map.put("town", stringutils.isempty(town) ? "" : town.trim());
 
            detail = matcher.group("detail");
            map.put("detail", stringutils.isempty(detail) ? "" : detail.trim());
        }
        return map;
    }
 
    public static void main(string[] args) {
 
        string address1 = "上海市上海市浦东新区世纪大道xx号上海中心大厦xx楼a座";
        string address2 = "内蒙古自治区呼伦贝尔市鄂温克族自治旗额尔古纳市阿尔山北路xxx号蒙古民族文化博物馆x楼xx展厅";
        string address3 = "广东省广州市番禺区沙湾镇大巷涌路xxx号";
 
        system.out.println("详细地址1：" + getdetailedaddress(address1));
        system.out.println("详细地址2：" + getdetailedaddress(address2));
        system.out.println("省市区地址拼接1：" + getpcaaddress(address1));
        system.out.println("省市区地址拼接2：" + getpcaaddress(address2));
        system.out.println("地址数组1：" + arrays.tostring(splicedetailedaddress(address1)));
        system.out.println("地址数组2：" + arrays.tostring(splicedetailedaddress(address2)));
 
        system.out.println("地址提取省市区：" + addressresolution(address2));
        system.out.println("地址提取省市区：" + addressresolution(address3));
    }
}