Example splitting fixes, tokenizer newline handling, Chinese

[DictionaryPC.git] / src / com / hughes / android / dictionary / engine / LanguageTest.java

diff --git a/src/com/hughes/android/dictionary/engine/LanguageTest.java b/src/com/hughes/android/dictionary/engine/LanguageTest.java
index 692bc41428d1de9e1fc6d4974bc5d1aba8be57b8..2d9b6a0afd052fdaa439738a87724788821b8f07 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/LanguageTest.java
+++ b/src/com/hughes/android/dictionary/engine/LanguageTest.java
@@ -122,6 +122,45 @@ public class LanguageTest extends TestCase {
     assertEquals("alphabetikos katalogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
     assertEquals("biologiceskom", transliterator.transliterate("биологическом"));
   }
+  public void testHalfTextNorm() {
+    final Transliterator transliterator = Transliterator.createFromRules("", ":: Any-Latin; ' ' > ; :: Lower; ", Transliterator.FORWARD);
+    assertEquals("kyanpasu", transliterator.transliterate("キャンパス"));
+    assertEquals("alphabētikóskatálogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
+    assertEquals("biologičeskom", transliterator.transliterate("биологическом"));
+
+    assertEquals("xièxiè", transliterator.transliterate("謝謝"));
+    assertEquals("xièxiè", transliterator.transliterate("谢谢"));
+
+    assertEquals("diànnǎo", transliterator.transliterate("電腦"));
+    assertEquals("diànnǎo", transliterator.transliterate("电脑"));
+    assertEquals("jìsuànjī", transliterator.transliterate("計算機"));
+    assertEquals("jìsuànjī", transliterator.transliterate("计算机"));
+  }
+
+  
+  public void testChinese() {
+    final Language zh = Language.lookup("zh");
+    final Transliterator transliterator = Transliterator.createFromRules("", zh.getDefaultNormalizerRules(), Transliterator.FORWARD);
+    
+    assertEquals("xiexie", transliterator.transliterate("謝謝"));
+    assertEquals("xiexie", transliterator.transliterate("谢谢"));
+
+    assertEquals("diannao", transliterator.transliterate("電腦"));
+    assertEquals("diannao", transliterator.transliterate("电脑"));
+    assertEquals("jisuanji", transliterator.transliterate("計算機"));
+    assertEquals("jisuanji", transliterator.transliterate("计算机"));
+    
+    assertEquals("chengjiu", transliterator.transliterate("成就"));
+    
+  }
+  
+  public void testArabic() {
+    final Language ar = Language.lookup("ar");
+    final Transliterator transliterator = Transliterator.createFromRules("", ar.getDefaultNormalizerRules(), Transliterator.FORWARD);
+    // These don't seem quite right....
+    assertEquals("haswb", transliterator.transliterate("حاسوب"));
+    assertEquals("kmbywtr", transliterator.transliterate("كمبيوتر"));
+  }