feature（地址清洗）重构

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

@@ -27,4 +27,12 @@ public class AddressContent {
     public String getResult() {
         return sb.toString();
     }
+
+    @Override
+    public String toString() {
+        return "AddressContent{" +
+                "addressType=" + addressType +
+                ", sb=" + sb +
+                '}';
+    }
 }

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

@@ -1,5 +1,7 @@
 package com.ruoyi.project.tool.address;
 
+import com.ruoyi.project.tool.address.model.AddressType;
+
 import java.util.ArrayList;
 import java.util.LinkedList;
 import java.util.List;
@@ -16,6 +18,7 @@ public class AddressContext {
      */
     private ArrayList<AddressNode> nodeList = new ArrayList<>();
     private StringBuilder stringBuilder;
+    private Integer startIndex = 0;
 
     public AddressContext(String text) {
         this.stringBuilder = new StringBuilder(text);
@@ -26,16 +29,30 @@ public class AddressContext {
     }
 
     public void addAddressNode(AddressNode node) {
+        startIndex = node.getEndIndex() - 1;
         this.nodeList.add(node);
     }
 
+    public Integer getStartIndex() {
+        return startIndex;
+    }
+
     public StandardAddress getResult() {
         StandardAddress standardAddress = new StandardAddress(this.stringBuilder.toString());
-//        nodeList.forEach(node->{
+        nodeList.forEach(node -> {
-////            AddressContent addressContent = new AddressContent();
+            AddressContent addressContent = new AddressContent(AddressType.CONDO);
-//
+            if (node.dataNode()) {
-////            standardAddress.addAddressContent();
+                addressContent.appendContent(node.getContent());
-//        });
+            } else {
+                if (1 == node.getContentList().size()) {
+                    addressContent.appendContent(node.getContentList().get(0).getContent());
+                } else {
+                    // 多个先组合，再拆分
+                }
+            }
+
+            standardAddress.addAddressContent(addressContent);
+        });
 
         return standardAddress;
     }

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

@@ -1,5 +1,7 @@
 package com.ruoyi.project.tool.address;
 
+import com.ruoyi.common.utils.StringUtils;
+
 import java.util.AbstractList;
 import java.util.Collection;
 import java.util.LinkedList;
@@ -55,6 +57,15 @@ public class AddressNode {
         return content;
     }
 
+    /**
+     * 数据节点
+     *
+     * @return
+     */
+    public Boolean dataNode() {
+        return StringUtils.isNotEmpty(this.content);
+    }
+
     public void setContent(String content) {
         this.content = content;
     }
@@ -62,4 +73,8 @@ public class AddressNode {
     public void addChildNode(AddressNode addressNode) {
         contentList.add(addressNode);
     }
+
+    public List<AddressNode> getContentList() {
+        return contentList;
+    }
 }

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

@@ -1,6 +1,5 @@
 package com.ruoyi.project.tool.address.service;
 
-import com.sun.xml.internal.fastinfoset.algorithm.BooleanEncodingAlgorithm;
 
 import java.util.List;
 

ruoyi/src/main/java/com/ruoyi/project/tool/address/utils/AddressNodeParse.java

@@ -75,7 +75,7 @@ public class AddressNodeParse {
         }
 
         for (Map.Entry<String, String> district : districtMap.entrySet()) {
-            int index = shanghaiAndDistrict.indexOf(district.getKey());
+            int index = shanghaiAndDistrict.indexOf(district.getKey(), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.DISTRICT, district.getValue());
                 addressNode.setStartIndex(index);
@@ -92,7 +92,7 @@ public class AddressNodeParse {
      */
     public void parseRegion() {
         for (int i = 0; i < regionList.size(); i++) {
-            int index = this.context.getContent().indexOf(regionList.get(i));
+            int index = this.context.getContent().indexOf(regionList.get(i), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.REGION, regionList.get(i));
                 addressNode.setStartIndex(index);
@@ -124,9 +124,8 @@ public class AddressNodeParse {
      */
     public void parseRoad() {
         AddressNode addressNode = new AddressNode(AddressNodeType.ROAD);
-
         for (int i = 0; i < roadList.size(); i++) {
-            int index = this.context.getContent().indexOf(roadList.get(i));
+            int index = this.context.getContent().indexOf(roadList.get(i), this.context.getStartIndex());
             if (-1 != index) {
                 if (null == addressNode.getStartIndex()) {
                     addressNode.setStartIndex(index);
@@ -171,12 +170,13 @@ public class AddressNodeParse {
         Pattern nongPattern = Pattern.compile(DEFAULT_NONG_PATTERN);
         Matcher matcher = nongPattern.matcher(this.context.getContent());
         if (matcher.find()) {
-            int index = this.context.getContent().indexOf(matcher.group(0));
+            int index = this.context.getContent().indexOf(matcher.group(0), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.NONG, matcher.group(0));
                 addressNode.setStartIndex(index);
                 addressNode.setEndIndex(index + matcher.group(0).length() + 1);
                 addressNode.addChildNode(addressNode);
+                this.context.addAddressNode(addressNode);
             }
         }
     }
@@ -188,12 +188,13 @@ public class AddressNodeParse {
         Pattern haoPattern = Pattern.compile(DEFAULT_HAO_PATTERN);
         Matcher matcher = haoPattern.matcher(this.context.getContent());
         if (matcher.find()) {
-            int index = this.context.getContent().indexOf(matcher.group(0));
+            int index = this.context.getContent().indexOf(matcher.group(0), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.HAO, matcher.group(0));
                 addressNode.setStartIndex(index);
                 addressNode.setEndIndex(index + matcher.group(0).length() + 1);
                 addressNode.addChildNode(addressNode);
+                this.context.addAddressNode(addressNode);
             }
         }
     }
@@ -205,12 +206,13 @@ public class AddressNodeParse {
         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));
+            int index = this.context.getContent().indexOf(matcher.group(0), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.ZHUANG, matcher.group(0));
                 addressNode.setStartIndex(index);
                 addressNode.setEndIndex(index + matcher.group(0).length() + 1);
                 addressNode.addChildNode(addressNode);
+                this.context.addAddressNode(addressNode);
             }
         }
     }
@@ -222,12 +224,13 @@ public class AddressNodeParse {
         Pattern haoPattern = Pattern.compile(DEFAULT_FLOOR_PATTERN);
         Matcher matcher = haoPattern.matcher(this.context.getContent());
         if (matcher.find()) {
-            int index = this.context.getContent().indexOf(matcher.group(0));
+            int index = this.context.getContent().indexOf(matcher.group(0), this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.CENG, matcher.group(0));
                 addressNode.setStartIndex(index);
                 addressNode.setEndIndex(index + matcher.group(0).length() + 1);
                 addressNode.addChildNode(addressNode);
+                this.context.addAddressNode(addressNode);
             }
         }
     }
@@ -239,12 +242,13 @@ public class AddressNodeParse {
         Pattern haoPattern = Pattern.compile(DEFAULT_SHI_PATTERN);
         Matcher matcher = haoPattern.matcher(this.context.getContent());
         if (matcher.find()) {
-            int index = this.context.getContent().indexOf(matcher.group(0));
+            int index = this.context.getContent().indexOf(matcher.group(0),this.context.getStartIndex());
             if (-1 != index) {
                 AddressNode addressNode = new AddressNode(AddressNodeType.SHI, matcher.group(0));
                 addressNode.setStartIndex(index);
                 addressNode.setEndIndex(index + matcher.group(0).length() + 1);
                 addressNode.addChildNode(addressNode);
+                this.context.addAddressNode(addressNode);
             }
         }
     }

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

@@ -1,8 +1,11 @@
 package com.ruoyi;
 
 import com.ruoyi.common.utils.LoadUtil;
+import com.ruoyi.project.tool.address.AddressContent;
 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.utils.DefaultAddressBuilder;
 import org.junit.Assert;
 import org.junit.Test;
 
@@ -24,6 +27,24 @@ public class CleanUtilTests {
         });
     }
 
+    @Test
+    public void testClear2() {
+        //
+        List<String> textList = LoadUtil.loadList("icbc.dict");
+
+        textList.forEach(item -> {
+            DefaultAddressBuilder defaultAddressBuilder = new DefaultAddressBuilder();
+            StandardAddress standardAddress = defaultAddressBuilder.clear(item);
+            List<AddressContent> list = standardAddress.getResult();
+            list.forEach(x -> {
+                System.out.println(x);
+            });
+//
+//            CleanAddress cleanAddress = CleanUtil.clear(item);
+//            System.out.println(cleanAddress);
+        });
+    }
+
     @Test
     public void testClear() {
         List<String> textList = LoadUtil.loadList("icbc.dict");