Fixes for Malay$ and reorderings due to new ICU4J.

[DictionaryPC.git] / src / com / hughes / android / dictionary / parser / wiktionary / WiktionaryLangs.java

diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
index 3de77dde399cffa007e07e31eb732061d6dd162f..3af9a139a63f9e6e21e43af3b6ed1eda7e6e5bfa 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
@@ -17,7 +17,9 @@ package com.hughes.android.dictionary.parser.wiktionary;
 import com.hughes.android.dictionary.engine.Language;
 
 import java.util.LinkedHashMap;
+import java.util.LinkedHashSet;
 import java.util.Map;
+import java.util.Set;
 import java.util.regex.Pattern;
 
 public class WiktionaryLangs {
@@ -30,14 +32,14 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("HY", "Armenian");
     isoCodeToEnWikiName.put("BE", "Belarusian");
     isoCodeToEnWikiName.put("BN", "Bengali");
-    isoCodeToEnWikiName.put("BS", "Bosnian");
     isoCodeToEnWikiName.put("BG", "Bulgarian");
-    isoCodeToEnWikiName.put("MY", "Burmese");
-    isoCodeToEnWikiName.put("yue", "Cantonese");
     isoCodeToEnWikiName.put("CA", "Catalan");
+    isoCodeToEnWikiName.put("SH", "Serbo-Croatian");
     isoCodeToEnWikiName.put("HR", "Croatian");
     isoCodeToEnWikiName.put("CS", "Czech");
-    isoCodeToEnWikiName.put("ZH", "Chinese|Mandarin");
+    isoCodeToEnWikiName.put("ZH", "Chinese");
+    isoCodeToEnWikiName.put("cmn", "Mandarin");
+    isoCodeToEnWikiName.put("yue", "Cantonese");
     isoCodeToEnWikiName.put("DA", "Danish");
     isoCodeToEnWikiName.put("NL", "Dutch");
     isoCodeToEnWikiName.put("EN", "English");
@@ -56,6 +58,7 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("ID", "Indonesian");
     isoCodeToEnWikiName.put("GA", "Irish");
     isoCodeToEnWikiName.put("GD", "Gaelic");
+    isoCodeToEnWikiName.put("GV", "Manx");
     isoCodeToEnWikiName.put("IT", "Italian");
     isoCodeToEnWikiName.put("LA", "Latin");
     isoCodeToEnWikiName.put("LV", "Latvian");
@@ -64,7 +67,7 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("KO", "Korean");
     isoCodeToEnWikiName.put("KU", "Kurdish");
     isoCodeToEnWikiName.put("LO", "Lao");
-    isoCodeToEnWikiName.put("MS", "Malay");
+    isoCodeToEnWikiName.put("MS", "Malay$");
     isoCodeToEnWikiName.put("ML", "Malayalam");
     isoCodeToEnWikiName.put("MI", "Maori");
     isoCodeToEnWikiName.put("MN", "Mongolian");
@@ -77,7 +80,6 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("RO", "Romanian");
     isoCodeToEnWikiName.put("RU", "Russian");
     isoCodeToEnWikiName.put("SA", "Sanskrit");
-    isoCodeToEnWikiName.put("SR", "Serbian");
     isoCodeToEnWikiName.put("SK", "Slovak");
     isoCodeToEnWikiName.put("SL", "Slovene|Slovenian");
     isoCodeToEnWikiName.put("SO", "Somali");
@@ -96,19 +98,31 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("CI", "Welsh");
     isoCodeToEnWikiName.put("YI", "Yiddish");
     isoCodeToEnWikiName.put("ZU", "Zulu");
-    
     isoCodeToEnWikiName.put("AZ", "Azeri");
     isoCodeToEnWikiName.put("EU", "Basque");
     isoCodeToEnWikiName.put("BR", "Breton");
-    isoCodeToEnWikiName.put("MR", "Burmese");
+    isoCodeToEnWikiName.put("MR", "Marathi");
     isoCodeToEnWikiName.put("FO", "Faroese");
     isoCodeToEnWikiName.put("GL", "Galician");
     isoCodeToEnWikiName.put("KA", "Georgian");
     isoCodeToEnWikiName.put("HT", "Haitian Creole");
     isoCodeToEnWikiName.put("LB", "Luxembourgish");
     isoCodeToEnWikiName.put("MK", "Macedonian");
+    isoCodeToEnWikiName.put("GV", "Manx");
+    
+    // No longer exists in EN:
+    // isoCodeToEnWikiName.put("BS", "Bosnian");
+    // isoCodeToEnWikiName.put("SR", "Serbian");
+    
+    // Font doesn't work:
+    //isoCodeToEnWikiName.put("MY", "Burmese");
 
 
+    {
+        Set<String> missing = new LinkedHashSet<String>(isoCodeToEnWikiName.keySet());
+        missing.removeAll(Language.isoCodeToResources.keySet());
+        //System.out.println(missing);
+    }
     assert Language.isoCodeToResources.keySet().containsAll(isoCodeToEnWikiName.keySet());
   }
 
@@ -151,6 +165,8 @@ public class WiktionaryLangs {
     //isoCodeToWikiName.put("", Pattern.quote("{{langue|mg}}"));
     //isoCodeToWikiName.put("", Pattern.quote("{{langue|hsb}}"));
     isoCodeToWikiName.put("ZH", Pattern.quote("{{langue|zh}}"));
+    isoCodeToWikiName.put("cmn", Pattern.quote("{{langue|cmn}}"));
+    isoCodeToWikiName.put("yue", Pattern.quote("{{langue|yue}}"));
     isoCodeToWikiName.put("JA", Pattern.quote("{{langue|ja}}"));
     isoCodeToWikiName.put("DE", Pattern.quote("{{langue|de}}"));
     isoCodeToWikiName.put("IS", Pattern.quote("{{langue|is}}"));  // Icelandic
@@ -184,7 +200,10 @@ public class WiktionaryLangs {
           return null;
       }
       if (name.indexOf('|') != -1) {
-          return name.substring(name.indexOf('|'));
+          return name.substring(0, name.indexOf('|'));
+      }
+      if (name.indexOf('$') != -1) {
+          return name.substring(0, name.indexOf('$'));
       }
       return name;  // can be null.
   }