支持98年人民日报的复合词语料格式，如"[中央/n 人民/n 广播/vn 电台/n]nt"

Author: hankcs

src/main/java/com/hankcs/hanlp/corpus/document/sentence/Sentence.java

@@ -49,7 +49,7 @@ public String toString()
 
     public static Sentence create(String param)
     {
-        Pattern pattern = Pattern.compile("(\\[(([^\\s]+/[0-9a-zA-Z]+)\\s+)+?([^\\s]+/[0-9a-zA-Z]+)]/[0-9a-zA-Z]+)|([^\\s]+/[0-9a-zA-Z]+)");
+        Pattern pattern = Pattern.compile("(\\[(([^\\s]+/[0-9a-zA-Z]+)\\s+)+?([^\\s]+/[0-9a-zA-Z]+)]/?[0-9a-zA-Z]+)|([^\\s]+/[0-9a-zA-Z]+)");
         Matcher matcher = pattern.matcher(param);
         List<IWord> wordList = new LinkedList<IWord>();
         while (matcher.find())

src/main/java/com/hankcs/hanlp/corpus/document/sentence/word/CompoundWord.java

@@ -96,11 +96,11 @@ public CompoundWord(List<Word> innerList, String label)
     public static CompoundWord create(String param)
     {
         if (param == null) return null;
-        int cutIndex = param.lastIndexOf('/');
+        int cutIndex = param.lastIndexOf(']');
         if (cutIndex <= 2 || cutIndex == param.length() - 1) return null;
-        String wordParam  = param.substring(1, cutIndex - 1);
+        String wordParam  = param.substring(1, cutIndex);
         List<Word> wordList = new LinkedList<Word>();
-        for (String single : wordParam.split(" "))
+        for (String single : wordParam.split("\\s+"))
         {
             if (single.length() == 0) continue;
             Word word = Word.create(single);
@@ -112,6 +112,10 @@ public static CompoundWord create(String param)
             wordList.add(word);
         }
         String labelParam = param.substring(cutIndex + 1);
+        if (labelParam.startsWith("/"))
+        {
+            labelParam = labelParam.substring(1);
+        }
         return new CompoundWord(wordList, labelParam);
     }
 }

src/test/java/com/hankcs/test/corpus/TestWord.java

@@ -24,6 +24,7 @@ public void testCreate() throws Exception
     {
         assertEquals("人民网/nz", Word.create("人民网/nz").toString());
         assertEquals("[纽约/nsf 时报/n]/nz", CompoundWord.create("[纽约/nsf 时报/n]/nz").toString());
+        assertEquals("[中央/n 人民/n 广播/vn 电台/n]/nt", CompoundWord.create("[中央/n 人民/n 广播/vn 电台/n]nt").toString());
     }
 
     public void testSpace() throws Exception