实现了用户自定义词性,同时支持代码动态增加和词典文件增加用户词性

src/main/java/com/hankcs/hanlp/corpus/io/ByteArray.java

@@ -105,13 +105,12 @@ public boolean hasMore()
      */
     public String nextString()
     {
-        StringBuilder sb = new StringBuilder();
-        int length = nextInt();
-        for (int i = 0; i < length; ++i)
+        char[] buffer = new char[nextInt()];
+        for (int i = 0; i < buffer.length; ++i)
         {
-            sb.append(nextChar());
+            buffer[i] = nextChar();
         }
-        return sb.toString();
+        return new String(buffer);
     }
 
     public float nextFloat()

src/main/java/com/hankcs/hanlp/corpus/io/IOUtil.java

@@ -477,4 +477,14 @@ public static void loadDictionary(BufferedReader br, TreeMap<String, CoreDiction
         }
         br.close();
     }
+
+    public static void writeCustomNature(DataOutputStream out, LinkedHashSet<Nature> customNatureCollector) throws IOException
+    {
+        if (customNatureCollector.size() == 0) return;
+        out.writeInt(-customNatureCollector.size());
+        for (Nature nature : customNatureCollector)
+        {
+            TextUtility.writeString(nature.toString(), out);
+        }
+    }
 }

src/main/java/com/hankcs/hanlp/corpus/tag/Nature.java

@@ -11,6 +11,8 @@
  */
 package com.hankcs.hanlp.corpus.tag;
 
+import com.hankcs.hanlp.corpus.util.CustomNatureUtility;
+
 /**
  * 词性
  *
@@ -810,4 +812,21 @@ public static Nature fromString(String name)
             return null;
         }
     }
-}
+
+    /**
+     * 创建自定义词性,如果已有该对应词性,则直接返回已有的词性
+     * @param name 字符串词性
+     * @return Enum词性
+     */
+    public static Nature create(String name)
+    {
+        try
+        {
+            return Nature.valueOf(name);
+        }
+        catch (Exception e)
+        {
+            return CustomNatureUtility.addNature(name);
+        }
+    }
+}

src/main/java/com/hankcs/hanlp/corpus/util/CustomNatureUtility.java

@@ -0,0 +1,72 @@
+/*
+ * <summary></summary>
+ * <author>He Han</author>
+ * <email>[email protected]</email>
+ * <create-date>2016/1/4 16:02</create-date>
+ *
+ * <copyright file="CustomNatureUtility.java" company="码农场">
+ * Copyright (c) 2008-2016, 码农场. All Right Reserved, http://www.hankcs.com/
+ * This source is subject to Hankcs. Please contact Hankcs to get more information.
+ * </copyright>
+ */
+package com.hankcs.hanlp.corpus.util;
+
+import com.hankcs.hanlp.corpus.tag.Nature;
+import com.hankcs.hanlp.dictionary.CustomDictionary;
+import com.hankcs.hanlp.recognition.nr.PersonRecognition;
+import com.hankcs.hanlp.recognition.nt.OrganizationRecognition;
+import com.hankcs.hanlp.seg.common.Vertex;
+import static com.hankcs.hanlp.utility.Predefine.logger;
+
+/**
+ * 运行时动态增加词性工具
+ *
+ * @author hankcs
+ */
+public class CustomNatureUtility
+{
+    static
+    {
+        logger.warning("已激活自定义词性功能,由于采用了反射技术,用户需对本地环境的兼容性和稳定性负责!\n" +
+                               "如果用户代码X.java中有switch(nature)语句,需要调用CustomNatureUtility.registerSwitchClass(X.class)注册X这个类");
+    }
+
+    /**
+     * 动态增加词性工具
+     */
+    private static EnumBuster<Nature> enumBuster = new EnumBuster<Nature>(Nature.class,
+                                                                          CustomDictionary.class,
+                                                                          Vertex.class,
+                                                                          PersonRecognition.class,
+                                                                          OrganizationRecognition.class);
+
+    /**
+     * 增加词性
+     * @param name 词性名称
+     * @return 词性
+     */
+    public static Nature addNature(String name)
+    {
+        Nature customNature = enumBuster.make(name);
+        enumBuster.addByValue(customNature);
+
+        return customNature;
+    }
+
+    /**
+     * 注册switch(nature)语句类
+     * @param switchUsers 任何使用了switch(nature)语句的类
+     */
+    public static void registerSwitchClass(Class... switchUsers)
+    {
+        enumBuster.registerSwitchClass(switchUsers);
+    }
+
+    /**
+     * 还原对词性的全部修改
+     */
+    public static void restore()
+    {
+        enumBuster.restore();
+    }
+}