Bug-fixes to WikiTokenizer (handle weird line-feed), update to newest

[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 f729b224a8fa7a3fd2eb4dd99a2c78632deb60b2..7b84dd644845682305cab6e6f7482f70ceb55b21 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/LanguageTest.java
+++ b/src/com/hughes/android/dictionary/engine/LanguageTest.java
@@ -17,11 +17,14 @@ package com.hughes.android.dictionary.engine;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collections;
+import java.util.LinkedHashSet;
 import java.util.List;
+import java.util.Set;
 
 import junit.framework.TestCase;
 
-import com.hughes.android.dictionary.parser.enwiktionary.EnWiktionaryLangs;
+import com.hughes.android.dictionary.parser.DictFileParser;
+import com.hughes.android.dictionary.parser.wiktionary.WiktionaryLangs;
 import com.ibm.icu.text.Transliterator;
 
 public class LanguageTest extends TestCase {
@@ -167,8 +170,25 @@ public class LanguageTest extends TestCase {
        
   }
 
+  public void testThai() {
+    final Language th = Language.lookup("TH");
+    final Transliterator transliterator = Transliterator.createFromRules("", th.getDefaultNormalizerRules(), Transliterator.FORWARD);
+    // Not sure these are right, just to know...
+    assertEquals("d", transliterator.transliterate("ด"));
+    assertEquals("di", transliterator.transliterate("ด ี"));
+    assertEquals("dii", transliterator.transliterate("ดีี"));
+    
+    assertEquals(Collections.singleton("ดีี"), DictFileParser.tokenize("ดีี", DictFileParser.NON_CHAR));
+  }
+
+  
   public void testEnWiktionaryNames() {
-    assertEquals(EnWiktionaryLangs.isoCodeToWikiName.keySet(), Language.isoCodeToResourceId.keySet());
+    final Set<String> enLangs = new LinkedHashSet<String>(WiktionaryLangs.isoCodeToWikiName.keySet());
+    for (final String code : WiktionaryLangs.isoCodeToWikiName.keySet()) {
+      enLangs.add(code.toLowerCase());
+    }
+    assertEquals(enLangs.toString(), Language.isoCodeToResources.keySet().toString());
+    assertEquals(enLangs, Language.isoCodeToResources.keySet());
   }
 
 }