]> gitweb.fperrin.net Git - DictionaryPC.git/blobdiff - src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java
Bug-fixes to WikiTokenizer (handle weird line-feed), update to newest
[DictionaryPC.git] / src / com / hughes / android / dictionary / engine / DictionaryBuilderMain.java
index 6e640d6da2a480c99dbeaaf203677be2da415a78..d99b59c0fe5ad9014df2b5f048f846d61bcfb89d 100644 (file)
@@ -19,7 +19,7 @@ import java.util.Map;
 
 import junit.framework.TestCase;
 
-import com.hughes.android.dictionary.parser.enwiktionary.EnWiktionaryLangs;
+import com.hughes.android.dictionary.parser.wiktionary.WiktionaryLangs;
 
 public class DictionaryBuilderMain extends TestCase {
   
@@ -31,7 +31,7 @@ public class DictionaryBuilderMain extends TestCase {
     
     // Builds all the dictionaries it can, outputs list to a text file.
     
-    final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(EnWiktionaryLangs.isoCodeToWikiName);
+    final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(WiktionaryLangs.isoCodeToWikiName);
     isoToWikiName.remove("EN");
     isoToWikiName.remove("DE");
 
@@ -65,7 +65,7 @@ public class DictionaryBuilderMain extends TestCase {
     //isoToWikiName.clear();
     boolean go = false;
     for (final String foreignIso : isoToWikiName.keySet()) {
-      if (foreignIso.equals("BO")) {
+      if (foreignIso.equals("GD")) {
         go = true;
       }
       if (!go) {
@@ -93,16 +93,18 @@ public class DictionaryBuilderMain extends TestCase {
             String.format("--lang2Stoplist=%s", STOPLISTS + isoToStoplist.get(foreignIso)),
             String.format("--dictInfo=(EN)Wikitionary-based EN-%s dictionary.\n\n%s", foreignIso, isoToDedication.get(foreignIso)),
 
-            "--input2=" + INPUTS + "enWikiSplit/" + foreignIso + ".data",
+            "--input2=" + INPUTS + "wikiSplit/en/" + foreignIso + ".data",
             "--input2Name=enwiktionary." + foreignIso,
             "--input2Format=enwiktionary",
+            "--input2WiktionaryType=EnForeign",
             "--input2LangPattern=" + isoToRegex.get(foreignIso),
             "--input2LangCodePattern=" + foreignIso.toLowerCase(),
             "--input2EnIndex=1",
 
-            "--input3=" + INPUTS + "enWikiSplit/EN.data",
+            "--input3=" + INPUTS + "wikiSplit/en/EN.data",
             "--input3Name=enwiktionary.english",
             "--input3Format=enwiktionary",
+            "--input3WiktionaryType=EnToTranslation",
             "--input3LangPattern=" + isoToRegex.get(foreignIso),
             "--input3LangCodePattern=" + foreignIso.toLowerCase(),
             "--input3EnIndex=1",
@@ -125,16 +127,18 @@ public class DictionaryBuilderMain extends TestCase {
         "--input4Charset=UTF8",
         "--input4Format=chemnitz",
         
-        "--input2=" + INPUTS + "enWikiSplit/DE.data",
+        "--input2=" + INPUTS + "wikiSplit/en/DE.data",
         "--input2Name=enwiktionary.DE",
         "--input2Format=enwiktionary",
+        "--input2WiktionaryType=EnForeign",
         "--input2LangPattern=German",
         "--input2LangCodePattern=de",
         "--input2EnIndex=2",
 
-        "--input3=" + INPUTS + "enWikiSplit/EN.data",
+        "--input3=" + INPUTS + "wikiSplit/en/EN.data",
         "--input3Name=enwiktionary.english",
         "--input3Format=enwiktionary",
+        "--input3WiktionaryType=EnToTranslation",
         "--input3LangPattern=German",
         "--input3LangCodePattern=de",
         "--input3EnIndex=2",