Fix word count issue for Chinese and Japanese

javatests/org/chromium/distiller/DocumentTitleGetterTest.java

 // Copyright 2014 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 package org.chromium.distiller;
 
 import com.google.gwt.dom.client.Element;
 
 public class DocumentTitleGetterTest extends DomDistillerJsTestCase {
     public void testNonStringWithoutRoot() {
@@ skipping 41 matching lines @@
                 "before dash - after dash", null);
         assertEquals("before dash - after dash", title);
     }
 
     public void test1Dash2LongParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "part with 6 words before dash - part with 6 words after dash", null);
         assertEquals("part with 6 words before dash", title);
     }
 
+    public void test1Dash2LongPartsChinese() {
+        String title = DocumentTitleGetter.getDocumentTitle(
+                "比較長一點的句子 - 這是不要的部分", null);
+        assertEquals("比較長一點的句子", title);
+    }
+
     public void test1DashLongAndShortParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "part with 6 words before dash - after dash", null);
         assertEquals("part with 6 words before dash", title);
     }
 
     public void test1DashShortAndLongParts() {
-        // TODO(kuan): if using RegExp.split, this fails with "ant test.prod".
         String title = DocumentTitleGetter.getDocumentTitle(
                 "before dash - part with 6 words after dash", null);
         assertEquals("part with 6 words after dash", title);
     }
 
+    public void test1DashShortAndLongPartsChinese() {
+        String title = DocumentTitleGetter.getDocumentTitle(
+                "短語 - 比較長一點的句子", null);
+        assertEquals("比較長一點的句子", title);
+    }
+
     public void test2DashesShortParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "before dash - between dash0 and dash1 - after dash1", null);
         assertEquals("before dash - between dash0 and dash1", title);
     }
 
     public void test2DashesShortAndLongParts() {
         // TODO(kuan): if using RegExp.split, this fails with "ant test.prod".
         String title = DocumentTitleGetter.getDocumentTitle(
                 "before - - part with 6 words after dash", null);
         assertEquals("- part with 6 words after dash", title);
     }
 
     public void test1Bar2ShortParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "before bar | after bar", null);
         assertEquals("before bar | after bar", title);
     }
 
     public void test2ColonsShortParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "start : midway : end", null);
         assertEquals("start : midway : end", title);
     }
 
+    public void test2ColonsShortPartsChinese() {
+        String title = DocumentTitleGetter.getDocumentTitle(
+                "開始 : 中間 : 最後", null);
+        assertEquals("開始 : 中間 : 最後", title);
+    }
+
     public void test2ColonsShortAndLongParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "start : midway : part with 6 words at end", null);
         assertEquals("part with 6 words at end", title);
     }
 
+    public void test2ColonsShortAndLongPartsChinese() {
+        String title = DocumentTitleGetter.getDocumentTitle(
+                "開始 : 中間 : 最後比較長的部分", null);
+        assertEquals("最後比較長的部分", title);
+    }
+
     public void test2ColonsShortAndLongAndShortParts() {
         String title = DocumentTitleGetter.getDocumentTitle(
                 "start : part with 6 words at midway : end", null);
         assertEquals("part with 6 words at midway : end", title);
     }
 
+    public void test2ColonsShortAndLongAndShortPartsChinese() {
+        String title = DocumentTitleGetter.getDocumentTitle(
+                "開始 : 中間要的部分 : 最後", null);
+        assertEquals("中間要的部分 : 最後", title);
+    }
+
     public void testH1WithShortText() {
         Element root = TestUtil.createDiv(0);
         Element h1 = TestUtil.createHeading(1, "short heading");
         root.appendChild(h1);
 
         String title = DocumentTitleGetter.getDocumentTitle("short title", root);
         assertEquals("short title", title);
     }
 
     public void testH1WithLongText() {
@@ skipping 40 matching lines @@
         String title = DocumentTitleGetter.getDocumentTitle("short title", null);
         assertEquals("short title", title);
     }
 
     public void testShortTitleWithoutH1() {
         Element root = TestUtil.createDiv(0);
         String title = DocumentTitleGetter.getDocumentTitle("short title", root);
         assertEquals("short title", title);
     }
 }