Fixed combining marks on Unicode regexes.

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

diff --git a/src/com/hughes/android/dictionary/engine/DictionaryBuilderTest.java b/src/com/hughes/android/dictionary/engine/DictionaryBuilderTest.java
index 7cdb3130b0c9a82495db5f0dac4b61216f7294e7..e3a8acc401c75e1d0b6b032777c0bafd16816ff7 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/DictionaryBuilderTest.java
+++ b/src/com/hughes/android/dictionary/engine/DictionaryBuilderTest.java
@@ -27,11 +27,32 @@ import junit.framework.TestCase;
 public class DictionaryBuilderTest extends TestCase {
   
   public static final String TEST_INPUTS = "testdata/inputs/";
-  public static final String WIKISPLIT = "data/inputs/enWikiSplit/";
+  public static final String WIKISPLIT = "data/inputs/wikiSplit/en/";
   public static final String STOPLISTS = "data/inputs/stoplists/";
   public static final String GOLDENS = "testdata/goldens/";
 
   public static final String TEST_OUTPUTS = "testdata/outputs/";
+  
+  public void testWiktionary_IT_EN() throws Exception {
+    wiktionaryTestWithLangToEn("wiktionary.it_en.quickdic", "IT", "it.txt",
+        "EN.data", "enwiktionary.english", "Italian", "it");
+  }
+
+  public void testWiktionary_ZH_EN() throws Exception {
+    wiktionaryTestWithLangToEn("wiktionary.zh_en.quickdic", "ZH", "empty.txt",
+        // These missing "e" prevents a complete match, forcing the name to be printed
+        "EN.data", "enwiktionary.english", "Chinese|Mandarin|Cantones", "zh");
+  }
+
+  public void testWiktionary_DE_EN() throws Exception {
+    wiktionaryTestWithLangToEn("wiktionary.de_en.quickdic", "DE", "de.txt",
+        "EN.data", "enwiktionary.english", "German", "it");
+  }
+
+  public void testWiktionary_IT_IT() throws Exception {
+    wiktionaryTestWithLangToEn("wiktionary.it_it.quickdic", "IT", "it.txt",
+        "IT.data", "enwiktionary.italian", "Italian", "it");
+  }
 
   // French
   public void testWiktionary_FR_FR() throws Exception {
@@ -53,32 +74,17 @@ public class DictionaryBuilderTest extends TestCase {
         "ZH.data", "enwiktionary.chinese", "Chinese|Mandarin|Cantones", "zh");
   }
 
-  public void testWiktionary_ZH_EN() throws Exception {
-    wiktionaryTestWithLangToEn("wiktionary.zh_en.quickdic", "ZH", "empty.txt",
-        // These missing "e" prevents a complete match, forcing the name to be printed
-        "EN.data", "enwiktionary.english", "Chinese|Mandarin|Cantones", "zh");
-  }
-  
   // German
   public void testWiktionary_DE_DE() throws Exception {
     wiktionaryTestWithLangToEn("wiktionary.de_de.quickdic", "DE", "de.txt",
         "DE.data", "enwiktionary.german", "German", "it");
   }
 
-  public void testWiktionary_DE_EN() throws Exception {
-    wiktionaryTestWithLangToEn("wiktionary.de_en.quickdic", "DE", "de.txt",
-        "EN.data", "enwiktionary.english", "German", "it");
-  }
-
-  // Italian
-  public void testWiktionary_IT_IT() throws Exception {
-    wiktionaryTestWithLangToEn("wiktionary.it_it.quickdic", "IT", "it.txt",
-        "IT.data", "enwiktionary.italian", "Italian", "it");
-  }
-
-  public void testWiktionary_IT_EN() throws Exception {
-    wiktionaryTestWithLangToEn("wiktionary.it_en.quickdic", "IT", "it.txt",
-        "EN.data", "enwiktionary.english", "Italian", "it");
+  // Thai
+  public void testWiktionary_TH_TH() throws Exception {
+    wiktionaryTestWithLangToEn("wiktionary.th_th.quickdic", "TH", "empty.txt",
+        // These missing "e" prevents a complete match, forcing the name to be printed.
+        "TH.data", "enwiktionary.thai", "Thai", "th");
   }
 
   public void wiktionaryTestWithLangToEn(final String name, final String lang1,
@@ -86,6 +92,7 @@ public class DictionaryBuilderTest extends TestCase {
       final String langPattern, final String langCode) throws Exception {
     final File result = new File(TEST_OUTPUTS + name);
     System.out.println("Writing to: " + result);
+    final String type = data.equals("EN.data") ? "EnToTranslation" : "EnForeign";
     DictionaryBuilder.main(new String[] {
         "--dictOut=" + result.getAbsolutePath(),
         "--lang1=" + lang1,
@@ -97,6 +104,7 @@ public class DictionaryBuilderTest extends TestCase {
         "--input4=" + WIKISPLIT + data,
         "--input4Name=" + dictName,
         "--input4Format=enwiktionary",
+        "--input4WiktionaryType=" + type,
         "--input4LangPattern=" + langPattern,
         "--input4LangCodePattern=" + langCode,
         "--input4EnIndex=2",
@@ -126,7 +134,7 @@ public class DictionaryBuilderTest extends TestCase {
         "--input2=" + TEST_INPUTS + "de-en_dictcc_simulated",
         "--input2Name=dictcc",
         "--input2Charset=UTF8",
-        "--input2Format=dictcc",
+        "--input2Format=tab_separated",
 
         "--print=" + result.getPath() + ".text",
     });