]> gitweb.fperrin.net Git - DictionaryPC.git/blobdiff - src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java
Examples now parsed with dispatch. Better {{l}} and {{term}} handling.
[DictionaryPC.git] / src / com / hughes / android / dictionary / engine / DictionaryBuilderMain.java
index 20bf4ba4437909286baf5b6593199c19ee8a6b8d..13ab9726d44100922f0a5b0c229c8d16f2975f15 100644 (file)
 package com.hughes.android.dictionary.engine;
 
 import java.io.File;
+import java.io.FileNotFoundException;
+import java.io.IOException;
 import java.io.PrintWriter;
 import java.io.RandomAccessFile;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.Collections;
 import java.util.LinkedHashMap;
 import java.util.List;
@@ -27,9 +30,9 @@ import junit.framework.TestCase;
 
 public class DictionaryBuilderMain extends TestCase {
   
-  static final String INPUTS = "../DictionaryData/inputs/";
-  static final String STOPLISTS = "../DictionaryData/inputs/stoplists/";
-  static final String OUTPUTS = "../DictionaryData/outputs/";
+  static final String INPUTS = "data/inputs/";
+  static final String STOPLISTS = "data/inputs/stoplists/";
+  static final String OUTPUTS = "data/outputs/";
     
   public static void main(final String[] args) throws Exception {
     
@@ -44,7 +47,6 @@ public class DictionaryBuilderMain extends TestCase {
     // German handled in file.
     isoToDedication.put("EL", "Greek dictionary dedicated to Noah Egge.");
     isoToDedication.put("IT", "Italian dictionary dedicated to Carolina Tropini, my favorite stardust in the whole universe!  Ti amo!");
-    isoToDedication.put("JA", "Japanese dictionary dedicated to Akane Watanabe.");
     isoToDedication.put("KO", "Korean dictionary dedicated to Ande Elwood--fall fashion und Fernsehturms!");
     isoToDedication.put("PT", "Portuguese dictionary dedicated to Carlos Melo, one Tough Mudder.");
     isoToDedication.put("RO", "Romanian dictionary dedicated to Radu Teodorescu.");
@@ -61,12 +63,14 @@ public class DictionaryBuilderMain extends TestCase {
     isoToStoplist.put("FR", "fr.txt");
 
     final Map<String,String>  isoToRegex = new LinkedHashMap<String, String>();
-    isoToRegex.put("ZH", ".*Chinese.*|.*Mandarin.*|.*Cantonese.*");
+    // HACK: The missing "e" prevents a full match, causing "Cantonese" to be appended to the entries.
+    isoToRegex.put("ZH", "Chinese|Mandarin|Cantones");
     
-    boolean go = false; 
-    isoToWikiName.clear();
+//    isoToWikiName.keySet().retainAll(Arrays.asList("UK", "HR", "FI"));
+    //isoToWikiName.clear();
+    boolean go = false;
     for (final String foreignIso : isoToWikiName.keySet()) {
-      if (foreignIso.equals("GA")) {
+      if (foreignIso.equals("JA")) {
         go = true;
       }
       if (!go) {
@@ -83,7 +87,7 @@ public class DictionaryBuilderMain extends TestCase {
           isoToDedication.put(foreignIso, "");
         }
         if (!isoToRegex.containsKey(foreignIso)) {
-          isoToRegex.put(foreignIso, ".*" + isoToWikiName.get(foreignIso) + ".*");
+          isoToRegex.put(foreignIso, isoToWikiName.get(foreignIso));
         }
   
         DictionaryBuilder.main(new String[] {
@@ -111,48 +115,53 @@ public class DictionaryBuilderMain extends TestCase {
         });
         
         // Print the entries for diffing.
-        final RandomAccessFile raf = new RandomAccessFile(new File(dictFile), "r");
-        final Dictionary dict = new Dictionary(raf);
-        final PrintWriter textOut = new PrintWriter(new File(dictFile + ".text"));
-        final List<PairEntry> sorted = new ArrayList<PairEntry>(dict.pairEntries);
-        Collections.sort(sorted);
-        for (final PairEntry pairEntry : sorted) {
-          textOut.println(pairEntry.getRawText(false));
-        }
-        textOut.close();
-        raf.close();
+        printToText(dictFile);
 
     }  // foreignIso
 
+    final String dictFile = OUTPUTS + "DE-EN_chemnitz_enwiktionary.quickdic"; 
     DictionaryBuilder.main(new String[] {
-        "--dictOut=" + OUTPUTS + "DE-EN_chemnitz_enwiktionary.quickdic",
+        "--dictOut=" + dictFile,
         "--lang1=DE",
         "--lang2=EN",
-        String.format("--lang1Stoplist=%s", STOPLISTS + isoToStoplist.get("DE")),
-        String.format("--lang2Stoplist=%s", STOPLISTS + isoToStoplist.get("EN")),
+        String.format("--lang1Stoplist=%s", STOPLISTS + "de.txt"),
+        String.format("--lang2Stoplist=%s", STOPLISTS + "en.txt"),
         "--dictInfo=@" + INPUTS + "de-en_chemnitz_enwiktionary.info",
 
-        "--input1=" + INPUTS + "de-en_chemnitz.txt",
-        "--input1Name=chemnitz",
-        "--input1Charset=UTF8",
-        "--input1Format=chemnitz",
+        "--input4=" + INPUTS + "de-en_chemnitz.txt",
+        "--input4Name=chemnitz",
+        "--input4Charset=UTF8",
+        "--input4Format=chemnitz",
         
         "--input2=" + INPUTS + "enWikiSplit/DE.data",
         "--input2Name=enwiktionary.DE",
         "--input2Format=enwiktionary",
-        "--input2LangPattern=" + isoToRegex.get("DE"),
+        "--input2LangPattern=German",
         "--input2LangCodePattern=de",
         "--input2EnIndex=2",
 
         "--input3=" + INPUTS + "enWikiSplit/EN.data",
         "--input3Name=enwiktionary.english",
         "--input3Format=enwiktionary",
-        "--input3LangPattern=" + isoToRegex.get("DE"),
+        "--input3LangPattern=German",
         "--input3LangCodePattern=de",
         "--input3EnIndex=2",
-        
     });
-
+    printToText(dictFile);
+    
+  }
+  
+  static void printToText(final String dictFile) throws IOException {
+    final RandomAccessFile raf = new RandomAccessFile(new File(dictFile), "r");
+    final Dictionary dict = new Dictionary(raf);
+    final PrintWriter textOut = new PrintWriter(new File(dictFile + ".text"));
+    final List<PairEntry> sorted = new ArrayList<PairEntry>(dict.pairEntries);
+    Collections.sort(sorted);
+    for (final PairEntry pairEntry : sorted) {
+      textOut.println(pairEntry.getRawText(false));
+    }
+    textOut.close();
+    raf.close();
   }
   
 }