解决:hankcs/hanlp-lucene-plugin#2 (comment)

dianhu · Jun 3, 2016 · 460bcf3 · 460bcf3
1 parent da7e84e
commit 460bcf3
Show file tree

Hide file tree

Showing 4 changed files with 4 additions and 2 deletions.
diff --git a/data/dictionary/other/CharType.dat.yes b/data/dictionary/other/CharType.dat.yes
diff --git a/src/main/java/com/hankcs/hanlp/dictionary/other/CharType.java b/src/main/java/com/hankcs/hanlp/dictionary/other/CharType.java
@@ -56,7 +56,7 @@ public class CharType
      */
     public static final byte CT_OTHER = CT_SINGLE + 12;
 
-    static byte[] type;
+    public static byte[] type;
 
     static
     {

diff --git a/src/main/java/com/hankcs/hanlp/seg/common/WordNet.java b/src/main/java/com/hankcs/hanlp/seg/common/WordNet.java
@@ -274,6 +274,7 @@ public void add(int line, List<AtomNode> atomSegment)
                     id = CoreDictionary.M_WORD_ID;
                     break;
                 case Predefine.CT_DELIMITER:
+                case Predefine.CT_OTHER:
                     nature = Nature.w;
                     break;
                 case Predefine.CT_SINGLE://12021-2129-3121

diff --git a/src/main/java/com/hankcs/hanlp/utility/TextUtility.java b/src/main/java/com/hankcs/hanlp/utility/TextUtility.java
@@ -76,7 +76,8 @@ public static int charType(String str)
             int ub2 = getUnsigned(b2);
             if (ub1 < 128)
             {
-                if (" *\"!,.?()[]{}+=/\\;:|".indexOf((char) b1) != -1)
+                if (' ' == b1) return CT_OTHER;
+                if ("*\"!,.?()[]{}+=/\\;:|".indexOf((char) b1) != -1)
                     return CT_DELIMITER;
                 if ("0123456789".indexOf((char)b1) != -1)
                     return CT_NUM;