place_name_search/src/main/java/com/skyversation/poiaddr/util/AddressSplitUtil.java

package com.skyversation.poiaddr.util;

import java.io.InputStream;
import java.util.*;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import java.util.stream.Collectors;

public class AddressSplitUtil {
    /**省级行政单位简称->全称映射表
     * 只读，请勿修改
     * */
    public static final Map<String,String> LEVEL_1_NAME_MAP;
    /**市级行政单位简称->全称映射表
     * 只读，请勿修改
     * */
    public static final Map<String,String> LEVEL_2_NAME_MAP;
    /**县级行政单位简称->全称映射表
     * 只读，请勿修改
     * */
    public static final Map<String,String> LEVEL_3_NAME_MAP;
    /**省市县三级行政单位简称树
     * 只读，请勿修改
     * */
    public static final Map<String,Map<String,Set<String>>> SIMPLE_NAME_TREE;
    private static final Map<String,Set<String>> All_CITY_IN_TREE;

    private static final Pattern LEVEL_1_SUFFIX_PATTERN = Pattern.compile("^(?:维吾尔|((?:(?!省|市|自治区).)*?族))?(?:省|市|自治区)");

    private static final Pattern LEVEL_2_SUFFIX_PATTERN = Pattern.compile("^(?:市|自治州|地区|盟)");

    private static final Pattern LEVEL_3_SUFFIX_PATTERN = Pattern.compile("^(?:县|自治县|市|区|旗|自治旗|林区|特区)");

    public static final Pattern ROAD_SUFFIX_PATTERN = Pattern.compile("^(?:旅游区|[东南西北中一二三四五六七八九十公大小支新老]{0,2}(?:大街|路|大道|街|菜市场|马路|村))");

    static {
        Map<String,String> level1NameMap = new HashMap<>();
        Map<String,String> level2NameMap = new HashMap<>();
        Map<String,String> level3NameMap = new HashMap<>();
        Map<String,Map<String,Set<String>>> simpleNameTree = new HashMap<>();
        String file = "全国省市县记录.xlsx";
        InputStream is = AddressSplitUtil.class.getResourceAsStream(file);
        if (is==null) is= AddressSplitUtil.class.getResourceAsStream("/"+file);
        if (is==null) throw new RuntimeException("无法找到"+file);
        try {
            List<Map<String, Object>> list = ExcelReaderUtils.readExcel(is);
            for (Map<String,Object> row : list) {
                Object level1Name = row.get("省份");
                Object level1SimpleName = row.get("省份简称");
                Object level2Name = row.get("地级市");
                Object level2SimpleName = row.get("地级市简称");
                Object level3Name = row.get("县级市");
                Object level3SimpleName = row.get("县级市简称");
                if (level1SimpleName != null && level1Name!=null) {
                    level1NameMap.put(level1SimpleName.toString(), level1Name.toString());
                }
                if (level2SimpleName != null && level2Name!=null) {
                    level2NameMap.put(level2SimpleName.toString(), level2Name.toString());
                }
                if (level3SimpleName != null && level3Name!=null) {
                    level3NameMap.put(level3SimpleName.toString(), level3Name.toString());
                }
                if (level1SimpleName!=null){
                    if (!simpleNameTree.containsKey(level1SimpleName.toString())){
                        simpleNameTree.put(level1SimpleName.toString(),new HashMap<>());
                    }
                    Map<String, Set<String>> level2Map = simpleNameTree.get(level1SimpleName.toString());
                    if (level2SimpleName!=null){
                        if(!level2Map.containsKey(level2SimpleName.toString())){
                            level2Map.put(level2SimpleName.toString(),new HashSet<>());
                        }
                        if (level3SimpleName!=null){
                            level2Map.get(level2SimpleName.toString()).add(level3SimpleName.toString());
                        }
                    }
                }
            }
            LEVEL_1_NAME_MAP = Collections.unmodifiableMap(level1NameMap);
            LEVEL_2_NAME_MAP = Collections.unmodifiableMap(level2NameMap);
            LEVEL_3_NAME_MAP = Collections.unmodifiableMap(level3NameMap);
            Map<String,Map<String,Set<String>>> simpleNameTree_= new HashMap<>();
            for (String key : simpleNameTree.keySet()){
                simpleNameTree_.put(key , Collections.unmodifiableMap(simpleNameTree.get(key)));
            }
            SIMPLE_NAME_TREE= Collections.unmodifiableMap(simpleNameTree_);
            All_CITY_IN_TREE= Collections.unmodifiableMap(SIMPLE_NAME_TREE.values().stream()
                    .flatMap(map -> map.entrySet().stream())
                    .collect(Collectors.toMap(
                            Map.Entry::getKey,
                            Map.Entry::getValue,
                            (oldValue, newValue) -> newValue
                    ))
            );

        } catch (Exception e) {
            throw new RuntimeException(e);
        }
    }

    private static class SplittingAddress {
        int province = -1;
        int city = -1;
        int county = -1;
        Map<Integer,String> provinceInChoose = new HashMap<>();
        Map<Integer,String> cityInChoose = new HashMap<>();
        Map<Integer,String> countyInChoose = new HashMap<>();
        String sourceAddress ;
        SplittingAddress(String sourceAddress){
            this.sourceAddress=sourceAddress;
        }
        String[] toStringList(){
            String[] output = new String[4];
            output[0]=((province!=-1)?LEVEL_1_NAME_MAP.get(provinceInChoose.get(province)):"");
            output[1]=((city!=-1)?LEVEL_2_NAME_MAP.get(cityInChoose.get(city)):"");
            output[2]=((county!=-1)?LEVEL_3_NAME_MAP.get(countyInChoose.get(county)):"");
            return  output;
        }
        String getOtherAddress(){
            int max = Math.max(province,Math.max(county,city));
            String maxName = "";
            if(max==-1){
                return sourceAddress;
            }
            if (province==max){
                maxName=provinceInChoose.get(province);
            }
            if (city==max){
                maxName=cityInChoose.get(city);
            }
            if (county==max){
                maxName=countyInChoose.get(county);
            }
            String sub = sourceAddress.substring(max+maxName.length());
            Matcher m = LEVEL_1_SUFFIX_PATTERN.matcher(sub);
            if (m.find()){
                sub = sub.substring(m.end());
            }
            m = LEVEL_2_SUFFIX_PATTERN.matcher(sub);
            if (m.find()){
                sub = sub.substring(m.end());
            }
            m = LEVEL_3_SUFFIX_PATTERN.matcher(sub);
            if (m.find()){
                sub = sub.substring(m.end());
            }
            return sub;
        }
        void findProvince(){
            Map<Integer,String> results = contain(this.sourceAddress,SIMPLE_NAME_TREE.keySet());
            for (int index : results.keySet()){
                String name = results.get(index);
                String sub = this.sourceAddress.substring(index+name.length());
                //去除南京路，北京大道型选手
                if (ROAD_SUFFIX_PATTERN.matcher(sub).find()){
                    continue;
                }
                provinceInChoose.put(index,name);
                //匹配到后缀时直接当做第一选择
                if (LEVEL_1_SUFFIX_PATTERN.matcher(sub).find()){
                    province = index;
                }
            }
            //仅有一个选择时当成一选
            if (provinceInChoose.size()==1){
                province = (int)provinceInChoose.keySet().toArray()[0];
            }
        }
        void findCity(){
            Map<Integer,String> results =null;
            //首先尝试在一选下匹配
            if (province!=-1){
                results  = contain(this.sourceAddress,SIMPLE_NAME_TREE.get(provinceInChoose.
                        get(province)).keySet());
            }
            //一选不存在或匹配无结果，直接搜全国
            if (results==null||results.isEmpty()){
                results = contain(this.sourceAddress,LEVEL_2_NAME_MAP.keySet());
            }

            Iterator<Integer> iterator = results.keySet().iterator();
            while (iterator.hasNext()) {
                int key = iterator.next();
                String name = results.get(key);
                if (key > 0 && name.equals("南县") &&"滦辉甘桦灌苍阜屏定全沂莒汝衡南郁平宁思广洛商南".indexOf(sourceAddress.charAt(key - 1)) != -1) {
                    iterator.remove();
                }

            }
            for (int index : results.keySet()){
                String name = results.get(index);
                String sub = this.sourceAddress.substring(index+name.length());
                //去除南京路，北京大道型选手
                if (ROAD_SUFFIX_PATTERN.matcher(sub).find()){
                    continue;
                }
                cityInChoose.put(index,name);
                //匹配到后缀时直接当做第一选择
                if (LEVEL_2_SUFFIX_PATTERN.matcher(sub).find()){
                    city = index;
                }
            }

            //仅有一个选择时当成一选
            if (cityInChoose.size()==1){
                city = (int)cityInChoose.keySet().toArray()[0];
            }
        }
        void findCounty(){
            Map<Integer,String> results = null;
            //尝试一选
            if (city!=-1){
                results=contain(sourceAddress,All_CITY_IN_TREE.get(cityInChoose.get(city)));
            }
            //一选不存在或匹配无结果，先搜全省
            if ((results == null || results.isEmpty()) && province != -1) {
                results = contain(sourceAddress, SIMPLE_NAME_TREE.get(provinceInChoose.get(province)).values().stream().
                        flatMap(Set::stream).collect(Collectors.toSet()));
            }
            //最后全国
            if (results == null || results.isEmpty()) {
                results = contain(sourceAddress, LEVEL_3_NAME_MAP.keySet());
            }
            for (int index : results.keySet()) {
                String name = results.get(index);
                String sub = this.sourceAddress.substring(index + name.length());
                //去除南京路，北京大道型选手
                if (ROAD_SUFFIX_PATTERN.matcher(sub).find()) {
                    continue;
                }
                countyInChoose.put(index, name);
                //匹配到后缀时直接当做第一选择
                if (LEVEL_3_SUFFIX_PATTERN.matcher(sub).find()) {
                    county = index;
                }
            }
            //仅有一个选择时当成一选
            if (countyInChoose.size()==1){
                county = (int)countyInChoose.keySet().toArray()[0];
            }
        }
    }
    /**
     * 检查字符串含有哪些字符,输出这些匹配字符的位置和字符的map
     * @param s 被检查字符串
     * @param nameList 检查范围
     */
    private static Map<Integer,String> contain(String s,Iterable<String> nameList){
        Map<Integer,String> output = new HashMap<Integer,String>();
        for (String name:nameList){
            if (name.isEmpty())continue;
            int index = -1;
            while ((index = s.indexOf(name, index + 1)) != -1){
                output.put(index,name);
            }
        }
        return output;
    }

    /**
     * 分离地址字符串,请优先使用shanghaiAddressSplitUtil，此类只分词到县<br/>
     * 注意，当输入的地址错误时不会自动修正，未找到的级会被空置<br/>
     * 例如输入"北京青浦区盈港路515号1061室" ，输出[北京市，北京市，青浦区，盈港路515号1061室]<br/>
     * 输入"安徽怀宁县黄墩镇老埂村双闸组" ，输出[安徽省，怀宁县，黄墩镇，盈港路515号1061室]<br/>
     * @return 结果为[省级，城级，县级，余下的部分]，分离失败则返回null<br/>
     * @see ShanghaiAddressSplitUtil
     */
    public static String[] splitAddress(String address){
        SplittingAddress a = new SplittingAddress(address.replaceAll("\\s+",""));
        a.findProvince();
        a.findCity();
        a.findCounty();
        String[] output = a.toStringList();
        output[3]=(a.getOtherAddress());
        return output;

    }
    //测试用
    public static void main(String[] args) {
        System.out.println(Arrays.toString(splitAddress("安徽省安徽省颍上县垂岗乡陶嘴村东道场31号")));
        System.out.println(Arrays.toString(splitAddress("荣乐西路1058弄32号501室")));
        System.out.println(Arrays.toString(splitAddress("泗泾镇新家园路30弄21号402室")));
        System.out.println(Arrays.toString(splitAddress("山东省山东省单县莱河镇宋楼行政村霍井村041号")));
        System.out.println(Arrays.toString(splitAddress("安徽省五河县安徽省五河县朱顶乡胡庄村447号")));
        System.out.println(Arrays.toString(splitAddress("九亭镇九亭大街506弄22号101室")));
        System.out.println(Arrays.toString(splitAddress("陕西省宝鸡市凤翔区陕西省凤翔区尹家务乡槐中村5组024号")));
        System.out.println(Arrays.toString(splitAddress("江苏省海门市江苏省海门市正余镇王灶河村十三组36号")));
        System.out.println(Arrays.toString(splitAddress("泗泾镇古楼公路519弄1号1102室")));
        System.out.println(Arrays.toString(splitAddress("奉贤县奉城镇奉粮路115号")));
        System.out.println(Arrays.toString(splitAddress("上海市奉贤区南桥镇沪杭支路24号14幢165室")));
        System.out.println(Arrays.toString(splitAddress("浦东新区周浦镇年家浜路10、12号1层")));
    }
}