feature(地址拆分)：地址清洗开发

1. 常规地址
2. 连号（结尾、中间）
3. 多号
4. 包含小区名称和路名

ruoyi/src/main/java/com/ruoyi/project/tool/address/AddressNodeType.java

@@ -6,12 +6,16 @@ package com.ruoyi.project.tool.address;
  * @author lihe
  */
 public enum AddressNodeType {
-    DISTRICT("区域"),
+    DISTRICT(""),
-    REGION("镇"),
+    REGION(""),
-    BLOCK("板块"),
+    BLOCK(""),
-    NONG("弄"),
+
+    NOT_FIND("没有节点"),
     ROAD("路"),
+    CNAME("小区"),
+    NONG("弄"),
     HAO("号"),
+    LIAN_HAO("连号"),
     MULTI_HAO("多号"),
     ZHUANG("幢"),
     ZUO("座"),

ruoyi/src/main/java/com/ruoyi/project/tool/address/AddressResult.java

@@ -1,8 +1,73 @@
 package com.ruoyi.project.tool.address;
 
+import com.ruoyi.project.tool.address.model.AdrNode;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * 地址结果
+ *
+ * @author lihe
+ */
 public class AddressResult {
+    private ParseContext context;
     private String district;
     private String region;
-    private Boolean multi;
+    private String block;
+    private List<String> cleanAddress;
 
+    public AddressResult(ParseContext context) {
+        this.context = context;
+    }
+
+    public String getDistrict() {
+        return context.getDistrict();
+    }
+
+    public String getRegion() {
+        return context.getRegion();
+    }
+
+    public String getBlock() {
+        return context.getBlock();
+    }
+
+    public List<String> getCleanAddress() {
+        List<AdrNode> nodeList = context.getNodeList();
+        if (0 == nodeList.size()) {
+            return null;
+        }
+
+        cleanAddress = new LinkedList<>();
+
+        for (int i = 0; i < nodeList.size(); i++) {
+
+        }
+
+        nodeList.forEach(rootNode -> {
+            StringBuilder sb = new StringBuilder();
+            loop(sb, rootNode);
+        });
+
+        return cleanAddress;
+    }
+
+    private void loop(StringBuilder sb, AdrNode rootNode) {
+        sb.append(rootNode.getContent());
+        if (null == rootNode.getChildren()) {
+            cleanAddress.add(sb.toString());
+            return;
+        }
+
+        if (rootNode.getChildren().size() > 1) {
+            for (int i = 0; i < rootNode.getChildren().size(); i++) {
+                StringBuilder newSb = new StringBuilder(sb);
+                loop(newSb, rootNode.getChildren().get(i));
+            }
+        } else {
+            loop(sb, rootNode.getChildren().get(0));
+        }
+
+    }
 }

ruoyi/src/main/java/com/ruoyi/project/tool/address/AddressUtil.java

@@ -0,0 +1,24 @@
+package com.ruoyi.project.tool.address;
+
+import com.ruoyi.project.tool.address.parse.XParse;
+
+/**
+ * 地址清洗工具类
+ *
+ * @author lihe
+ */
+public class AddressUtil {
+    /**
+     * 清洗
+     *
+     * @param text
+     * @return
+     */
+    public static AddressResult clear(String text) {
+        ParseContext context = new ParseContext(text);
+        XParse xParse = new XParse(context);
+        xParse.parse();
+        AddressResult result = new AddressResult(context);
+        return result;
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/ParseContext.java

@@ -0,0 +1,82 @@
+package com.ruoyi.project.tool.address;
+
+import com.ruoyi.project.tool.address.model.AdrNode;
+
+import java.util.ArrayList;
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * context
+ *
+ * @author lihe
+ */
+public class ParseContext {
+    private String district;
+    private String region;
+    private String block;
+    private Integer startIndex = 0;
+    private final StringBuilder sb;
+    private List<String> features = new LinkedList<>();
+    private List<AdrNode> nodeList = new ArrayList<>(20);
+
+    public ParseContext(String text) {
+        this.sb = new StringBuilder(text);
+    }
+
+    public String getDistrict() {
+        return district;
+    }
+
+    public void setDistrict(String district) {
+        this.district = district;
+        this.startIndex = this.sb.indexOf(district);
+    }
+
+    public String getRegion() {
+        return region;
+    }
+
+    public void addRegion(String region, Integer startIndex) {
+        this.region = region;
+        this.startIndex = startIndex;
+    }
+
+    public String getBlock() {
+        return block;
+    }
+
+    public void addBlock(String block, Integer startIndex) {
+        this.block = block;
+        this.startIndex = startIndex;
+    }
+
+    public void setStartIndex(Integer startIndex) {
+        this.startIndex = startIndex;
+    }
+
+    public Integer getStartIndex() {
+        return startIndex;
+    }
+
+    public void addNode(AdrNode node) {
+        this.nodeList.add(node);
+        this.startIndex = node.getEndIndex();
+    }
+
+    public StringBuilder getContent() {
+        return this.sb;
+    }
+
+    public List<AdrNode> getNodeList() {
+        return nodeList;
+    }
+
+    public void addFeature(String feature) {
+        this.features.add(feature);
+    }
+
+    public int getIndexOf(String text) {
+        return this.getContent().indexOf(text, this.getStartIndex());
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/AdrNode.java

@@ -0,0 +1,64 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * node
+ *
+ * @author lihe
+ */
+public abstract class AdrNode {
+    protected AddressNodeType nodeType;
+    private Integer startIndex;
+    private Integer endIndex;
+    private final String content;
+    protected List<AdrNode> children;
+
+    public AdrNode(String content, Integer startIndex) {
+        this.content = content;
+        this.startIndex = startIndex;
+    }
+
+    public AdrNode(String content, Integer startIndex, Integer endIndex) {
+        this.content = content;
+        this.startIndex = startIndex;
+        this.endIndex = endIndex;
+    }
+
+    public AddressNodeType getNodeType() {
+        return nodeType;
+    }
+
+    public String getContent() {
+        return content;
+    }
+
+    public Integer getEndIndex() {
+        return endIndex;
+    }
+
+    public void setEndIndex(Integer endIndex) {
+        this.endIndex = endIndex;
+    }
+
+    public List<AdrNode> getChildren() {
+        return children;
+    }
+
+    /**
+     * 添加节点
+     *
+     * @param node 节点
+     */
+    public void addNode(AdrNode node) {
+        if (null == this.children) {
+            this.children = new LinkedList<>();
+        }
+        this.children.add(node);
+    }
+
+
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/CengNode.java

@@ -0,0 +1,15 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+/**
+ * 层
+ *
+ * @author lihe
+ */
+public class CengNode extends AdrNode {
+    public CengNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.CENG;
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/HaoNode.java

@@ -0,0 +1,28 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * 号
+ *
+ * @author lihe
+ */
+public class HaoNode extends AdrNode {
+    private List<String> shiList = new LinkedList<>();
+    public HaoNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.HAO;
+    }
+
+    /**
+     * 交叉路
+     *
+     * @param shi
+     */
+    public void addShi(String shi) {
+        shiList.add(shi);
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/LianHaoNode.java

@@ -0,0 +1,24 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * 号
+ *
+ * @author lihe
+ */
+public class LianHaoNode extends AdrNode {
+    private List<String> shiList = new LinkedList<>();
+
+    public LianHaoNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.LIAN_HAO;
+    }
+
+    public void addShi(String shi) {
+        shiList.add(shi);
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/NongNode.java

@@ -0,0 +1,30 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * 弄
+ *
+ * @author lihe
+ */
+public class NongNode extends AdrNode {
+    private List<String> haoList = new LinkedList<>();
+
+    public NongNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.NONG;
+    }
+
+    /**
+     * 交叉路
+     *
+     * @param road
+     */
+    public void addHao(String road) {
+        haoList.add(road);
+    }
+
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/RoadNode.java

@@ -0,0 +1,34 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+/**
+ * 路名或者小区名称
+ *
+ * @author lihe
+ */
+public class RoadNode extends AdrNode {
+    private List<String> roadList = new LinkedList<>();
+
+    public RoadNode(String content, Integer startIndex) {
+        super(content, startIndex);
+        this.nodeType = AddressNodeType.ROAD;
+    }
+
+    public RoadNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.ROAD;
+    }
+
+    /**
+     * 交叉路
+     *
+     * @param road
+     */
+    public void addRoad(String road) {
+        roadList.add(road);
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/model/ShiNode.java

@@ -0,0 +1,15 @@
+package com.ruoyi.project.tool.address.model;
+
+import com.ruoyi.project.tool.address.AddressNodeType;
+
+/**
+ * 室
+ *
+ * @author lihe
+ */
+public class ShiNode extends AdrNode {
+    public ShiNode(String content, Integer startIndex, Integer endIndex) {
+        super(content, startIndex, endIndex);
+        this.nodeType = AddressNodeType.SHI;
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/parse/XParse.java

@@ -0,0 +1,170 @@
+package com.ruoyi.project.tool.address.parse;
+
+import com.ruoyi.common.utils.LoadUtil;
+import com.ruoyi.project.tool.address.ParseContext;
+import com.ruoyi.project.tool.address.model.RoadNode;
+
+import java.util.List;
+import java.util.Map;
+
+/**
+ * X解析
+ *
+ * @author lihe
+ */
+public class XParse {
+    private static final int DISTRICT_LENGTH = 8;
+    private static final String SHANGHAI_SHI = "上海市";
+    private static final String SHANGHAI = "上海";
+    private static final String SHANG_SHANG = "上上";
+    private static final String SHANG = "上";
+    private Map<String, String> districtMap = LoadUtil.loadDict("address-dict/district.dict");
+    private List<String> regionList = LoadUtil.loadList("address-dict/region.dict");
+    private List<String> blockList = LoadUtil.loadList("address-dict/block.dict");
+    private List<String> roadList = LoadUtil.loadList("address-dict/road.dict");
+    private List<String> communityNameList = LoadUtil.loadList("address-dict/community.dict");
+    private ParseContext context;
+
+    public XParse(ParseContext context) {
+        this.context = context;
+    }
+
+    /**
+     * 区域、板块、
+     * 路、小区
+     */
+    public void parse() {
+        parseDistrict();
+        parseRegion();
+        parseBlock();
+        parseRoad();
+        parseCommunityName();
+    }
+
+    /**
+     * 小区名称
+     */
+    private void parseCommunityName() {
+        RoadNode communityNameNode = null;
+        for (int i = 0; i < communityNameList.size(); i++) {
+            int index = this.context.getContent().indexOf(communityNameList.get(i), this.context.getStartIndex());
+            if (-1 != index) {
+                communityNameNode = new RoadNode(communityNameList.get(i), index,
+                        index + communityNameList.get(i).length());
+                break;
+            }
+        }
+
+        if (null != communityNameNode) {
+            this.context.addNode(communityNameNode);
+
+            YParse yParse = new YParse(communityNameNode, context);
+            yParse.parse();
+            return;
+        }
+    }
+
+    /**
+     * 路名
+     */
+    private void parseRoad() {
+        int startIndex = this.context.getStartIndex();
+        int endIndex = 0;
+        RoadNode firstNode = null;
+        for (int i = 0; i < roadList.size(); i++) {
+            int index = this.context.getContent().indexOf(roadList.get(i), startIndex);
+            if (-1 != index) {
+                RoadNode roadNode = new RoadNode(roadList.get(i), index);
+                if (null == firstNode) {
+                    firstNode = roadNode;
+                } else {
+                    // 连路；
+                    if (endIndex + 1 == index) {
+                        firstNode.addRoad(firstNode.getContent());
+                        firstNode.addRoad(roadList.get(i));
+                    } else {
+                        // 新旧地址
+                    }
+                }
+                startIndex = index;
+                endIndex = index + roadList.get(i).length();
+            }
+        }
+
+        // 单点
+        if (null != firstNode) {
+            firstNode.setEndIndex(endIndex);
+            this.context.addNode(firstNode);
+
+            YParse yParse = new YParse(firstNode, context);
+            yParse.parse();
+            return;
+        }
+
+
+        for (int i = 0; i < roadList.size(); i++) {
+            // 砍掉“路”
+            // XX街加上“路”
+            String noRoad = roadList.get(i);
+            if (noRoad.endsWith("路")) {
+                noRoad = noRoad.substring(0, noRoad.length() - 1);
+            }
+
+            int index = this.context.getContent().indexOf(noRoad, startIndex);
+            if (-1 != index) {
+                RoadNode roadNode = new RoadNode(noRoad, index);
+                if (null == firstNode) {
+                    firstNode = roadNode;
+                } else {
+                    // 连路；
+                    if (endIndex + 1 == index) {
+                        firstNode.addRoad(firstNode.getContent());
+                        firstNode.addRoad(roadList.get(i));
+                    } else {
+                        // 新旧地址
+                    }
+                }
+                startIndex = index;
+                endIndex = index + roadList.get(i).length();
+            }
+        }
+    }
+
+
+    private void parseDistrict() {
+        String shanghaiAndDistrict = null;
+        if (this.context.getContent().length() >= DISTRICT_LENGTH) {
+            shanghaiAndDistrict = (SHANG + this.context.getContent().substring(0, 7)).replace(SHANG_SHANG, SHANG);
+        } else {
+            shanghaiAndDistrict = (SHANG + this.context.getContent()).replace(SHANG_SHANG, SHANG);
+        }
+
+        if (!shanghaiAndDistrict.startsWith(SHANGHAI_SHI) && !shanghaiAndDistrict.startsWith(SHANGHAI) && shanghaiAndDistrict.startsWith(SHANG)) {
+            shanghaiAndDistrict = shanghaiAndDistrict.substring(1, shanghaiAndDistrict.length() - 1);
+        }
+
+        for (Map.Entry<String, String> district : districtMap.entrySet()) {
+            if (shanghaiAndDistrict.startsWith(district.getKey())) {
+                this.context.setDistrict(district.getValue());
+            }
+        }
+    }
+
+    private void parseRegion() {
+        for (int i = 0; i < regionList.size(); i++) {
+            int index = this.context.getContent().indexOf(regionList.get(i), this.context.getStartIndex());
+            if (-1 != index) {
+                this.context.addRegion(regionList.get(i), index);
+            }
+        }
+    }
+
+    private void parseBlock() {
+        for (int i = 0; i < blockList.size(); i++) {
+            int index = this.context.getContent().indexOf(blockList.get(i), this.context.getStartIndex());
+            if (-1 != 1) {
+                this.context.addBlock(blockList.get(i), index);
+            }
+        }
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/parse/YParse.java

@@ -0,0 +1,148 @@
+package com.ruoyi.project.tool.address.parse;
+
+import com.ruoyi.project.tool.address.ParseContext;
+import com.ruoyi.project.tool.address.model.*;
+
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * Y解析
+ * ABC弄(DEF支弄)
+ * ABC号DEF号(室)
+ * ABC、DEF、GHI号；ABC号、DEF号、GHI号
+ *
+ * @author lihe
+ */
+public class YParse {
+    private static final String HAO = "号";
+
+    private static final String DEFAULT_NONG_PATTERN = "[\\d一二三四五六七八九十]+弄([\\d一二三四五六七八九十]+支弄)?";
+    private static final String DEFAULT_HAO_PATTERN = "([\\dA-Za-z]+)号";
+    // 连号结束？号=室
+    private static final String LIAN_HAO_PATTERN = "\\d+号\\d+号";
+    // 多号
+    private static final String MULTI_HAO_PATTERN = "(\\d+(号)?[、\\.\\-\\_]?)+号";
+    private static final String NUMBER_PATTERN = "\\d+";
+    private ParseContext context;
+    private AdrNode rootNode;
+
+
+    public YParse(AdrNode adrNode, ParseContext context) {
+        this.context = context;
+        this.rootNode = adrNode;
+    }
+
+    public void parse() {
+        parseNong();
+        parseHao();
+        parseZuo();
+    }
+
+    private void parseZuo() {
+    }
+
+    /**
+     * 连号
+     *
+     * @param text
+     * @param matchText
+     */
+    private void parseLianHao(String text, String matchText) {
+        if (text.endsWith(matchText)) {
+            // ${1}号${2}室，修改最后一个字符
+            this.context.getContent().replace(this.context.getContent().length() - 1,
+                    this.context.getContent().length() - 1, "室");
+
+            // 找到${1}号
+            Matcher haoMatch = Pattern.compile(DEFAULT_HAO_PATTERN).matcher(matchText);
+            if (haoMatch.find()) {
+                String hao = haoMatch.group();
+                int index = this.context.getIndexOf(hao);
+                HaoNode haoNode = new HaoNode(hao, index, index + hao.length());
+                this.rootNode.addNode(haoNode);
+                this.context.setStartIndex(haoNode.getEndIndex());
+
+                ZParse zParse = new ZParse(haoNode, this.context, haoNode.getEndIndex());
+                zParse.parse();
+            }
+        } else {
+            // 中间连号
+            int index = this.context.getIndexOf(matchText);
+            LianHaoNode haoNode = new LianHaoNode(matchText, index, index + matchText.length());
+            this.rootNode.addNode(haoNode);
+            this.context.setStartIndex(haoNode.getEndIndex());
+
+            ZParse zParse = new ZParse(haoNode, this.context, haoNode.getEndIndex());
+            zParse.parse();
+        }
+    }
+
+    /**
+     * 并列号
+     */
+    private void parseMultiHao(String matchText) {
+
+        Pattern numberPattern = Pattern.compile(NUMBER_PATTERN);
+        Matcher matcher = numberPattern.matcher(matchText);
+
+        while (matcher.find()) {
+            int index = this.context.getIndexOf(matcher.group());
+            if (-1 == index) {
+                continue;
+            }
+
+            LianHaoNode haoNode = new LianHaoNode(matcher.group() + "号", index, index + matcher.group().length());
+            this.rootNode.addNode(haoNode);
+            this.context.setStartIndex(haoNode.getEndIndex());
+
+            ZParse zParse = new ZParse(haoNode, this.context, haoNode.getEndIndex());
+            zParse.parse();
+        }
+    }
+
+
+    private void parseHao() {
+        // 连号
+        String text = this.context.getContent().substring(this.context.getStartIndex());
+        Pattern lianHaoPattern = Pattern.compile(LIAN_HAO_PATTERN);
+        Matcher matcher = lianHaoPattern.matcher(text);
+        if (matcher.find()) {
+            parseLianHao(text, matcher.group());
+            return;
+        }
+
+        // 并列号
+        Pattern multiHaoPattern = Pattern.compile(MULTI_HAO_PATTERN);
+        matcher = multiHaoPattern.matcher(text);
+        if (matcher.find()) {
+            parseMultiHao(matcher.group());
+            return;
+        }
+
+        // 默认的号
+        Pattern haoPattern = Pattern.compile(DEFAULT_HAO_PATTERN);
+        matcher = haoPattern.matcher(text);
+        if (matcher.find()) {
+            int index = this.context.getIndexOf(matcher.group());
+            HaoNode haoNode = new HaoNode(matcher.group(), index, index + matcher.group(0).length());
+            this.rootNode.addNode(haoNode);
+            this.context.setStartIndex(haoNode.getEndIndex());
+            ZParse zParse = new ZParse(haoNode, this.context, haoNode.getEndIndex());
+            zParse.parse();
+            return;
+        }
+    }
+
+    private void parseNong() {
+        Pattern nongPattern = Pattern.compile(DEFAULT_NONG_PATTERN);
+        Matcher matcher = nongPattern.matcher(this.context.getContent());
+        if (matcher.find()) {
+            int index = this.context.getIndexOf(matcher.group(0));
+            NongNode nongNode = new NongNode(matcher.group(0), index, index + matcher.group(0).length());
+            this.rootNode.addNode(nongNode);
+            this.rootNode = nongNode;
+            this.context.setStartIndex(nongNode.getEndIndex());
+        }
+    }
+}

ruoyi/src/main/java/com/ruoyi/project/tool/address/parse/ZParse.java

@@ -0,0 +1,150 @@
+package com.ruoyi.project.tool.address.parse;
+
+import com.ruoyi.project.tool.address.ParseContext;
+import com.ruoyi.project.tool.address.model.AdrNode;
+import com.ruoyi.project.tool.address.model.CengNode;
+import com.ruoyi.project.tool.address.model.HaoNode;
+import com.ruoyi.project.tool.address.model.ShiNode;
+
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * Z解析
+ * <p>
+ * 102.3室
+ * 102、103室
+ * 1-3层、1_3层——
+ *
+ * @author lihe
+ */
+public class ZParse {
+    private static final String FU_SHI = "复式";
+    private static final String NUMBER_PATTERN = "\\d+";
+    private static final String DEFAULT_SHI_PATTERN1 = "([\\dA-Za-z\\_\\-\\.\\——\\、，]+)(室?)$";
+    private static final String DEFAULT_SHI_PATTERN = "([\\dA-Za-z\\-\\_\\.，]+)室";
+    private static final String CHINESE_FLOOR_PATTERN = "([一二三四五六七八九十]+)层";
+    private static final String DEFAULT_FLOOR_PATTERN = "地下?([\\d一二三四五六七八九十])层";
+    private static final String TAIL_FLOOR_PATTERN = "(\\d[\\-\\_\\——\\—]?\\d)(层)?$";
+    private static final String DEFAULT_ZHUANG_PATTERN = "(\\d+)幢";
+    private ParseContext context;
+    private AdrNode rootNode;
+    private String floor;
+    private Integer startIndex;
+
+    public ZParse(AdrNode adrNode, ParseContext context, Integer index) {
+        this.context = context;
+        this.rootNode = adrNode;
+        this.startIndex = index;
+    }
+
+    public void parse() {
+        parseCeng();
+        parseZhuang();
+        parseShi();
+        tag();
+    }
+
+    private void parseZhuang() {
+        Pattern haoPattern = Pattern.compile(DEFAULT_ZHUANG_PATTERN);
+        Matcher matcher = haoPattern.matcher(this.context.getContent());
+        if (matcher.find()) {
+            int index = this.context.getContent().indexOf(matcher.group(0), this.startIndex);
+            HaoNode haoNode = new HaoNode(matcher.group(0), index, index + matcher.group(0).length());
+            this.rootNode.addNode(haoNode);
+            startIndex = haoNode.getEndIndex();
+        }
+    }
+
+    private void parseShi() {
+        // 复式
+        if (-1 != this.context.getContent().indexOf(FU_SHI)) {
+            String content = this.context.getContent().substring(this.startIndex);
+            Pattern shiPattern = Pattern.compile(NUMBER_PATTERN);
+            Matcher matcher = shiPattern.matcher(content);
+            if (matcher.find()) {
+                int index = this.context.getContent().indexOf(matcher.group(0), this.startIndex);
+                ShiNode shiNode = new ShiNode(matcher.group(0) + "室", index, index + matcher.group(0).length());
+                this.rootNode.addNode(shiNode);
+                startIndex = shiNode.getEndIndex();
+            }
+        } else {
+            Pattern shiPattern = Pattern.compile(DEFAULT_SHI_PATTERN);
+            Matcher matcher = shiPattern.matcher(this.context.getContent());
+            if (matcher.find()) {
+                int index = this.context.getContent().indexOf(matcher.group(0), this.startIndex);
+                ShiNode shiNode = new ShiNode(matcher.group(0), index, index + matcher.group(0).length());
+                this.rootNode.addNode(shiNode);
+                startIndex = shiNode.getEndIndex();
+            }
+        }
+    }
+
+    private void tag() {
+        // 车位、车库、地块
+        if (-1 != this.context.getContent().indexOf("车位") || -1 != this.context.getContent().indexOf("车库")
+                || -1 != this.context.getContent().indexOf("停车")) {
+            this.context.addFeature("车位");
+        }
+
+        if (-1 != this.context.getContent().indexOf("地块")) {
+            this.context.addFeature("地块");
+        }
+    }
+
+    private Boolean invalidCeng(int floorIndex) {
+        // 判断这个层是否可取？
+        Pattern shiPattern = Pattern.compile(DEFAULT_SHI_PATTERN);
+        Matcher shiMatcher = shiPattern.matcher(this.context.getContent());
+        if (shiMatcher.find()) {
+            int shiIndex = this.context.getContent().indexOf(shiMatcher.group(0), this.startIndex);
+            return floorIndex > shiIndex;
+        }
+
+        return true;
+    }
+
+    private void parseCeng() {
+        Pattern floorPattern = Pattern.compile(CHINESE_FLOOR_PATTERN);
+        Matcher matcher = floorPattern.matcher(this.context.getContent());
+        if (matcher.find()) {
+            int index = this.context.getContent().indexOf(matcher.group(0), this.startIndex);
+            if (invalidCeng(index)) {
+                return;
+            }
+            CengNode cengNode = new CengNode(matcher.group(0), index, index + matcher.group(0).length());
+            this.rootNode.addNode(cengNode);
+            rootNode.addNode(cengNode);
+            return;
+        }
+
+        // 层结尾
+        floorPattern = Pattern.compile(TAIL_FLOOR_PATTERN);
+        matcher = floorPattern.matcher(this.context.getContent());
+        if (matcher.find()) {
+            return;
+        }
+
+        // 地下层
+        floorPattern = Pattern.compile(DEFAULT_FLOOR_PATTERN);
+        matcher = floorPattern.matcher(this.context.getContent());
+        if (matcher.find()) {
+            String num = matcher.group(1)
+                    .replace("一", "1")
+                    .replace("二", "2")
+                    .replace("三", "3")
+                    .replace("四", "4")
+                    .replace("五", "5");
+            int index = this.context.getContent().indexOf(matcher.group(0), this.startIndex);
+            if (invalidCeng(index)) {
+                return;
+            }
+            CengNode cengNode = new CengNode("-" + num, index, index + matcher.group(0).length());
+            this.rootNode.addNode(cengNode);
+            rootNode.addNode(cengNode);
+            return;
+        }
+    }
+
+}
+

ruoyi/src/main/java/com/ruoyi/project/tool/address/service/BaseAddressBuilder.java

@@ -24,9 +24,10 @@ public abstract class BaseAddressBuilder implements AddressBuilder {
     private static final int DISTRICT_LENGTH = 8;
     private static final String SHANGHAI_SHI = "上海市";
     private static final String SHANGHAI = "上海";
+    private static final String SHANG_SHANG = "上上";
     private static final String SHANG = "上";
     private static final String HAO = "号";
-    private static final String SHANG_SHANG = "上上";
+
     private static final String CHINESE_FLOOR_PATTERN = "([一二三四五六七八九十]+)层";
     private static final String NUMBER_FLOOR_PATTERN = "(\\d+)层";
     private static final String DEFAULT_HAO_PATTERN = "([\\dA-Za-z]+)号";

ruoyi/src/test/java/com/ruoyi/CleanUtilTests.java

@@ -1,10 +1,10 @@
 package com.ruoyi;
 
 import com.ruoyi.common.utils.LoadUtil;
-import com.ruoyi.project.tool.address.AddressContent;
+import com.ruoyi.project.tool.address.*;
-import com.ruoyi.project.tool.address.CleanUtil;
-import com.ruoyi.project.tool.address.StandardAddress;
 import com.ruoyi.project.tool.address.model.CleanAddress;
+import com.ruoyi.project.tool.address.model.LianHaoNode;
+import com.ruoyi.project.tool.address.parse.ZParse;
 import com.ruoyi.project.tool.address.utils.DefaultAddressBuilder;
 import org.junit.Assert;
 import org.junit.Test;
@@ -15,11 +15,34 @@ import java.io.IOException;
 import java.io.UnsupportedEncodingException;
 import java.util.Base64;
 import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 public class CleanUtilTests {
 
     @Test
-    public void testSortDistrict() {
+    public void printStreet() {
+        List<String> textList = LoadUtil.loadList("jie.dict");
+        textList.forEach(item -> {
+            if (!item.contains("路") && item.contains("街") && !item.contains("街坊")) {
+                System.out.println(item);
+            }
+        });
+    }
+
+    @Test
+    public void printNoRoad() {
+        List<String> textList = LoadUtil.loadList("icbc.dict");
+        textList.forEach(item -> {
+            if (!item.contains("路")) {
+                System.out.println(item);
+            }
+        });
+    }
+
+
+    @Test
+    public void testSort() {
         List<String> textList = LoadUtil.loadList("temp.dict");
         textList.sort((x, y) -> y.length() - x.length());
         textList.forEach(item -> {
@@ -27,6 +50,91 @@ public class CleanUtilTests {
         });
     }
 
+    @Test
+    public void testClear3() {
+        List<String> textList = LoadUtil.loadList("test.dict");
+        textList.forEach(item -> {
+            AddressResult addressResult = AddressUtil.clear(item);
+            if (null == addressResult.getCleanAddress()) {
+                return;
+            }
+            addressResult.getCleanAddress().forEach(adr -> {
+                System.out.println(String.format("%s\t%s\t", addressResult.getDistrict(), adr));
+            });
+        });
+    }
+
+    /**
+     * 中间连号
+     */
+    @Test
+    public void testLianHaoInMiddle() {
+        AddressResult addressResult = AddressUtil.clear("上海市浦东新区拱北路278号22号702室");
+        if (null == addressResult.getCleanAddress()) {
+            return;
+        }
+
+        Assert.assertEquals(addressResult.getCleanAddress().get(0), "拱北路278号22号702室");
+        Assert.assertEquals(addressResult.getDistrict(), "浦东");
+    }
+
+    /**
+     * 尾部连号
+     */
+    @Test
+    public void testLianHaoInEnd() {
+        AddressResult addressResult = AddressUtil.clear("上海市浦东新区拱北路278弄22号702号");
+        if (null == addressResult.getCleanAddress()) {
+            return;
+        }
+
+        Assert.assertEquals(addressResult.getCleanAddress().get(0), "拱北路278弄22号702室");
+        Assert.assertEquals(addressResult.getDistrict(), "浦东");
+    }
+
+    /**
+     * 多个号
+     */
+    @Test
+    public void testMultiHao() {
+        AddressResult addressResult = AddressUtil.clear("上海市松江区泗泾镇德悦路375弄24、25号704室");
+        if (null == addressResult.getCleanAddress()) {
+            return;
+        }
+
+        Assert.assertEquals("德悦路375弄24号704室", addressResult.getCleanAddress().get(0));
+        Assert.assertEquals("德悦路375弄25号704室", addressResult.getCleanAddress().get(1));
+        Assert.assertEquals(addressResult.getDistrict(), "松江");
+    }
+
+    @Test
+    public void testWhileMatch() {
+        Pattern numberPattern = Pattern.compile("\\d+");
+        Matcher matcher = numberPattern.matcher("24、25号");
+
+        while (matcher.find()) {
+
+            System.out.println(matcher.group());
+
+        }
+    }
+
+
+    /**
+     * 室、地下
+     */
+    @Test
+    public void testShiAndDiXiaShi() {
+        AddressResult addressResult = AddressUtil.clear("闵行区北华路168弄35号1002室及30号地下1层车位665室");
+        if (null == addressResult.getCleanAddress()) {
+            return;
+        }
+        addressResult.getCleanAddress().forEach(adr -> {
+            System.out.println(String.format("%s\t%s\t", addressResult.getDistrict(), adr));
+        });
+    }
+
+
     @Test
     public void testClear2() {
         //

ruoyi/src/test/resources/jie.dict

@@ -0,0 +1,677 @@
+梅园三村,
+钱仓路
+其昌栈大街
+钱仓路
+其昌栈大街
+花木西街
+龙王庙镇西街
+东三街
+东三街
+东三街
+东三街
+塘南路
+塘南路
+塘南路
+东三街
+北蔡大街
+北蔡大街
+北蔡大街
+后长街
+后长街
+灵岩南路
+杨思南街
+杨思路
+振兴中路
+杨思路
+思浦路
+思浦路
+思浦路
+思浦路
+灵岩南路
+杨思南街
+杨思路
+灵岩南路
+玉泉街
+玉泉街
+玉泉街
+玉泉街
+玉泉街
+玉泉街
+振兴东路
+后长街
+振兴东路
+后长街
+振兴东路
+后长街
+金桥横街
+万安街
+万安路
+万安街
+万安街
+慈善街
+大同路
+大同路
+石家街
+石家街
+石家街
+石家街
+石家街
+石家街
+学前街
+学前街
+学前街
+王桥街
+塘东街
+塘东街
+塘东街
+南市街
+南市街
+南市街
+南市街
+南市街
+三灶浜路
+操场街
+西市街
+西市街
+益民路
+益民路
+益民路
+西市街
+西市街
+北市街
+迎春街
+普庆路
+景园街
+普园路
+川六公路
+南八灶街
+南八灶街
+西大街
+西大街
+西大街
+西大街
+西大街
+西大街
+西大街
+西大街
+中大街
+中大街
+关岳西路
+南八灶街
+南八灶街
+顾家宅
+东大街
+宫宵公寓,
+祝桥镇祝振街
+南塘街
+南塘街
+下沙镇下沙新街
+下沙镇下沙新街银信楼
+东后老街
+东后老街
+北门大街
+北门大街
+北门大街
+东门大街
+东门大街
+南门大街
+西门大街
+沿河泾南路
+东门大街
+南门大街
+南门大街
+南门大街
+卫星东路
+三角街
+三角街
+三角街
+三角街
+东门大街
+十字街大楼
+文师街
+文友街
+文益街
+西门大街
+书院镇船山街
+书院镇石潭街
+书院镇石潭街
+书院镇首善街
+书院镇首善街
+申港街道竹柏路
+申港街道竹柏路
+申港街道竹柏路
+申港街道竹柏路
+申港街道雪绒花路
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉新村,
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊泉街
+菊盛路
+菊泉街
+菊太路
+菊泉街
+菊太路
+布长街
+东西巷街
+市二路
+市二路
+塘东街
+塘东街
+塘东街
+塘东街
+市河街
+市河街
+市河街
+市河街
+塘西街
+塘西街
+陈行街
+宝钢五村,
+宝钢五村西门街,
+宝钢五村后河沿,
+团结路
+堡镇堡镇南路
+堡镇中路
+正大街
+正大街
+向阳路
+城桥镇中街山路
+川心街
+川心街
+川心街
+川心街
+新崇南路
+人民路
+新崇中路
+八一路
+富民街
+花园弄
+兴贤街
+中街山路
+西引路
+东街供销四村
+东街新村
+东街新村二村
+奉城镇南街
+洪庙镇唐城街
+鹏贸南街
+平安镇堂前街
+四团镇四新街
+四新街
+四新街
+四新南街
+塘外镇文化街
+青村镇西街
+新街二村
+新街三村
+新街一村
+蟠龙街
+池沟路
+池沟路
+公安街
+公安街
+保宁路
+文迁路
+公安街
+虹镇老街
+虹镇老街
+虹镇老街
+虹镇老街
+虹镇老街
+光启路
+光启路
+光启路
+光启路
+光启路
+县左街
+县左街
+县左街
+学院路
+柳江街
+柳江街
+南仓街
+南仓街
+先棉祠街
+先棉祠街
+先棉祠街
+先棉祠街
+学前街
+学前街
+学院路
+东街
+巡道街
+永泰街
+悦来街
+中华路
+中华路
+老太平街
+庄家街
+大林路
+江阴街
+江阴街
+江阴街
+江阴街
+沙家街
+大林路
+制造局后街
+黄渡镇劳动街
+黄渡镇劳动街
+黄渡镇劳动街
+北大街
+北大街
+北大街
+北大街
+北大街
+城中街
+东大街
+东大街
+东大街
+东大街
+嘉定镇北大街
+嘉定镇北大街
+嘉定镇东大街
+金沙路
+嘉定镇南大街
+嘉定镇南大街
+嘉定镇南大街
+嘉定镇南大街
+嘉定镇西大街
+嘉定镇张马路
+嘉定镇中下塘街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆街
+马陆塘街
+马陆塘街
+双单路
+双单路
+马陆塘街
+马陆塘街
+南大街
+南大街
+南大街
+南大街
+南大街
+南大街
+清河路
+梅园路
+人民街
+人民街
+人民街
+沙霞路
+沙霞路
+南大街
+塔城路
+塔城路
+南大街
+南大街
+南大街
+温宿路
+北大街
+北大街
+项泾西街
+新成路街道仓场路
+新成路街道仓场路
+新成路街道迎园中路
+育英街
+育英南街
+中下塘街
+中下塘街
+中下塘街
+和平街
+和平街
+解放街
+南翔镇解放街
+南翔镇解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+解放街
+劳动街
+民主东街
+民主东街
+民主东街
+民主东街
+民主东街
+民主东街
+民主东街
+民主东街
+民主东街
+民主街
+民主街
+民主街
+民主街
+民主街
+民主街
+民主街
+民主街
+解放街
+解放街
+解放街
+南翔镇民主东街
+南翔镇民主东街
+南翔镇民主街
+南翔镇民主街
+南翔镇民主街
+南翔镇红翔新村
+漕泾镇中心街
+中心街
+枫泾镇南大街
+枫泾镇新街
+海汇街
+海汇街
+金山卫镇东街
+金山卫镇钱圩东街
+龙临街
+山阳镇海汇街
+山阳镇海汇街
+干巷镇干溪街
+廊下镇廊下大街
+廊下镇廊下大街
+廊下镇廊下大街
+廊下镇下大街
+吕巷镇璜溪东街
+璜溪东街
+吕巷镇璜溪东街
+璜溪东街
+吕巷镇建新大街
+吕巷镇新东街
+吕巷镇新东街
+亭林镇开乐大街
+亭林镇松隐大街
+亭林镇松隐大街
+亭林镇新建北街
+亭林镇新建西街
+朱行镇开乐大街
+朱行镇开乐大街
+朱行镇开乐大街
+朱行镇开乐大街
+朱行镇兴乐街
+珠港街
+城中路
+新安街
+临仓街
+临源街
+临源街
+临源街
+秀州街
+秀州街
+秀州街
+秀州街
+众安街
+朱泾镇东林东街
+朱泾镇东林街
+朱泾镇东林街
+朱泾镇东林街
+朱泾镇东林街
+朱泾镇金南街
+朱泾镇金南街
+朱泾镇金南街
+朱泾镇临仓街
+朱泾镇临仓街
+朱泾镇临仓街
+朱泾镇临源街
+朱泾镇万安街
+朱泾镇万安街
+朱泾镇万安街
+朱泾镇万安街
+朱泾镇万安街
+朱泾镇西林街
+朱泾镇新汇街
+朱泾镇秀州街
+朱泾镇秀州街
+朱泾镇秀州街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+朱泾镇众安街
+万春街
+万春街
+万春街
+万春街
+万春街
+万春街
+万春街
+万春街
+升平街
+升平街
+汝南街
+汝南街
+汝南街
+汝南街
+汝南街
+汝南街
+汝南街
+南华街
+南华街
+中心街
+南北大街
+南北大街
+新西街
+东新街
+富岩路
+东街
+马桥新村,
+中街
+新街
+马桥镇东新街
+马桥镇西街
+马桥镇中街
+中街
+中街
+中街
+中街
+南中街
+西新路
+马桥西街
+马桥西街
+马桥西街
+富强街
+富强街
+七莘路
+富强街
+富强街
+富强街
+宝隆新村
+南街
+南街
+莘浜路
+莘庄镇西街
+莘浜路
+小西街
+颛桥镇后东街
+南大街
+南大街
+郁家弄,
+南大街
+南大街
+兰溪路
+南大街
+白鹤镇南街
+白鹤镇南街
+华新镇华富街
+华新镇华富街
+华新镇华富街
+华新镇华富街
+华新镇华强街
+华新镇华强街
+华新镇华强街
+华新镇华强街
+华新镇华强街
+华新镇华强街
+华新镇华新街
+华新镇华新街
+华新镇华新街
+重固镇大街
+重固镇大街
+重固镇大街
+广场街
+香花桥街道崧雅路
+赵巷镇崧雅路
+盈浦街道盈清路
+盈清路
+赵屯镇久远路
+赵屯镇久远路
+白鹤镇友爱东路
+赵屯镇久远路
+赵屯镇久远路
+香花桥街道香大路
+久业路
+练塘镇前进街
+九亭大街
+九亭大街
+九亭大街
+九亭大街
+涞亭南路
+九亭大街
+九亭大街
+九亭大街
+九亭大街
+九亭大街
+九亭大街
+九亭大街
+九亭镇九亭大街
+九亭镇九亭大街
+九亭镇九亭大街
+九亭镇九亭大街
+虬泾路
+九亭大街
+淡家浜街
+淡家浜街
+长水街
+平原街
+平原街
+平原街
+平原街
+平原街
+平原街
+平原街
+秦安街
+秦安街
+平原街
+东外街
+沪松路
+北门街
+北门街
+阔街
+李塔街
+通波新村,
+乐都路
+乐都路
+荣乐中路
+下塘街
+小塔前
+长桥街
+中山二路
+潭东街
+谷阳北路
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新北街
+新南街
+新南街
+新南街
+新南街
+新南街
+新南街
+新南街
+新桥镇新南街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新镇街
+新中街
+新中街
+新中街
+百花街
+丁香园
+桂林东街
+桂林东街
+桂林东街
+桂林东街
+桂林东街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+桂林西街
+钦州南路
+钦州南路
+钦州南路
+桂林东街
+小闸镇街
+怀安街
+紫阳路
+紫阳路
+汇站街
+南丹路汇站街
+汇站街
+纪念路
+学府路
+学府街
+北大街
+蒲松北路
+和顺街
+菊泉街
+环城东路
+城中南路
+学前街
+中街小区,
+中新路
+中新路
+中新路
+菊泉街

ruoyi/src/test/resources/test.dict

@@ -0,0 +1,12 @@
+上海市普陀区长风一村40号601-2室
+上海市松江区泗泾镇佘月路18弄28号9层901室
+上海市浦东新区晨晖路825弄49号103（复式）室
+宝山区美艾路177弄167号1_3层
+奉贤区南桥镇江海新村(B)89幢578号402室
+上海市嘉定区百安公路2699弄505号1层1-3层室
+上海市松江区泗泾镇泗宝路298弄7号10层1004室
+奉贤区南桥镇江海新村（C）6幢218号502室
+上海市闸北区洛川东路310号503，504室
+上海市松江区泗泾镇德悦路375弄24、25号704室
+闵行区北华路168弄35号1002室及30号地下1层车位665室
+上海市浦东新区拱北路278号22号702室