修复可变DAT的entrySet方法 fix #1038

hankcs · Dec 8, 2018 · 43f0ea8 · 43f0ea8
1 parent 54a230b
commit 43f0ea8
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 0 deletions.
diff --git a/src/main/java/com/hankcs/hanlp/collection/trie/datrie/IntArrayList.java b/src/main/java/com/hankcs/hanlp/collection/trie/datrie/IntArrayList.java
@@ -4,6 +4,7 @@
 import com.hankcs.hanlp.corpus.io.ICacheAble;
 
 import java.io.*;
+import java.util.ArrayList;
 
 /**
  * 动态数组
@@ -205,4 +206,15 @@ private void readObject(ObjectInputStream in) throws IOException, ClassNotFoundE
         exponentialExpanding = in.readBoolean();
         exponentialExpandFactor = in.readDouble();
     }
+
+    @Override
+    public String toString()
+    {
+        ArrayList<Integer> head = new ArrayList<Integer>(20);
+        for (int i = 0; i < Math.min(size, 20); ++i)
+        {
+            head.add(data[i]);
+        }
+        return head.toString();
+    }
 }
diff --git a/src/main/java/com/hankcs/hanlp/collection/trie/datrie/MutableDoubleArrayTrieInteger.java b/src/main/java/com/hankcs/hanlp/collection/trie/datrie/MutableDoubleArrayTrieInteger.java
@@ -1245,6 +1245,7 @@ public KeyValuePair()
                             from = b + i;
                             path.append(from);
                             b = base.get(from);
+                            i = 0;
                             if (getCheck(b + UNUSED_CHAR_VALUE) == from)
                             {
                                 value = getLeafValue(getBase(b + UNUSED_CHAR_VALUE));

diff --git a/src/test/java/com/hankcs/hanlp/model/perceptron/feature/ImmutableFeatureMDatMapTest.java b/src/test/java/com/hankcs/hanlp/model/perceptron/feature/ImmutableFeatureMDatMapTest.java
@@ -1,14 +1,36 @@
 package com.hankcs.hanlp.model.perceptron.feature;
 
 import com.hankcs.hanlp.HanLP;
+import com.hankcs.hanlp.collection.trie.datrie.MutableDoubleArrayTrieInteger;
 import com.hankcs.hanlp.model.perceptron.model.LinearModel;
 import junit.framework.TestCase;
 
+import java.util.Map;
+import java.util.TreeMap;
+
 public class ImmutableFeatureMDatMapTest extends TestCase
 {
     public void testCompress() throws Exception
     {
         LinearModel model = new LinearModel(HanLP.Config.PerceptronCWSModelPath);
         model.compress(0.1);
     }
+
+    public void testFeatureMap() throws Exception
+    {
+        LinearModel model = new LinearModel(HanLP.Config.PerceptronCWSModelPath);
+        ImmutableFeatureMDatMap featureMap = (ImmutableFeatureMDatMap) model.featureMap;
+        MutableDoubleArrayTrieInteger dat = featureMap.dat;
+        System.out.println(featureMap.size());
+        System.out.println(featureMap.entrySet().size());
+        System.out.println(featureMap.idOf("\u0001/\u00014"));
+        TreeMap<String, Integer> map = new TreeMap<String, Integer>();
+        for (Map.Entry<String, Integer> entry : dat.entrySet())
+        {
+            map.put(entry.getKey(), entry.getValue());
+            assertEquals(entry.getValue().intValue(), dat.get(entry.getKey()));
+        }
+        System.out.println(map.size());
+        assertEquals(dat.size(), map.size());
+    }
 }