src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java

   1 package com.hughes.android.dictionary.engine;
   2
   3 import java.io.File;
   4 import java.io.PrintWriter;
   5 import java.io.RandomAccessFile;
   6 import java.util.ArrayList;
   7 import java.util.Collections;
   8 import java.util.List;
   9
  10 import junit.framework.TestCase;
  11
  12 public class DictionaryBuilderMain extends TestCase {
  13
  14   static class Lang {
  15     final String nameRegex;
  16     final String code;
  17     public Lang(String nameRegex, String code) {
  18       this.nameRegex = nameRegex;
  19       this.code = code;
  20     }
  21   }
  22
  23
  24   public static void main(final String[] args) throws Exception {
  25
  26     Lang[] langs1 = new Lang[] {
  27         new Lang("^English$", "EN"),
  28         //new Lang("^German$", "DE"),
  29     };
  30     Lang[] langs2 = new Lang[] {
  31         new Lang("^Italian$", "IT"),
  32         new Lang("^German$", "DE"),
  33         new Lang("^Afrikaans$", "AF"),
  34         new Lang("^Armenian$", "HY"),
  35         new Lang("^Arabic$", "AR"),
  36         new Lang("^Chinese$|^Mandarin$", "ZH"),
  37         new Lang("^Croation$", "HR"),
  38         new Lang("^Czech$", "CS"),
  39         new Lang("^Dutch$", "NL"),
  40         new Lang("^English$", "EN"),
  41         new Lang("^Finnish$", "FI"),
  42         new Lang("^French$", "FR"),
  43         new Lang("^Greek$", "EL"),
  44         new Lang("^Hebrew$", "HE"),
  45         new Lang("^Hindi$", "HI"),
  46         new Lang("^Icelandic$", "IS"),
  47         new Lang("^Irish$", "GA"),
  48         new Lang("^Japanese$", "JA"),
  49         new Lang("^Korean$", "KO"),
  50         new Lang("^Kurdish$", "KU"),
  51         new Lang("^Lithuanian$", "LT"),
  52         new Lang("^Malay$", "MS"),
  53         new Lang("^Maori$", "MI"),
  54         new Lang("^Mongolian$", "MN"),
  55         new Lang("^Norwegian$", "NO"),
  56         new Lang("^Persian$", "FA"),
  57         new Lang("^Portuguese$", "PT"),
  58         new Lang("^Romanian$", "RO"),
  59         new Lang("^Russian$", "RU"),
  60         new Lang("^Sanskrit$", "SA"),
  61         new Lang("^Serbian$", "SR"),
  62         new Lang("^Somali$", "SO"),
  63         new Lang("^Spanish$", "ES"),
  64         new Lang("^Sudanese$", "SU"),
  65         new Lang("^Swedish$", "SV"),
  66         new Lang("^Tajik$", "TG"),
  67         new Lang("^Thai$", "TH"),
  68         new Lang("^Tibetan$", "BO"),
  69         new Lang("^Turkish$", "TR"),
  70         new Lang("^Ukranian$", "UK"),
  71         new Lang("^Vietnamese$", "VI"),
  72         new Lang("^Welsh$", "CY"),
  73         new Lang("^Yiddish$", "YI"),
  74         new Lang("^Zulu$", "ZU"),
  75     };
  76
  77     for (final Lang lang1 : langs1) {
  78       for (final Lang lang2 : langs2) {
  79         if (lang1.nameRegex.equals(lang2.nameRegex)) {
  80           continue;
  81         }
  82
  83         int enIndex = -1;
  84         if (lang2.code.equals("EN")) {
  85           enIndex = 2;
  86         }
  87         if (lang1.code.equals("EN")) {
  88           enIndex = 1;
  89         }
  90
  91         final String dictFile = String.format("dictOutputs/%s-%s_enwiktionary.quickdic", lang1.code, lang2.code);
  92         System.out.println("building dictFile: " + dictFile);
  93         DictionaryBuilder.main(new String[] {
  94             String.format("--dictOut=%s", dictFile),
  95             String.format("--lang1=%s", lang1.code),
  96             String.format("--lang2=%s", lang2.code),
  97             String.format("--dictInfo=(EN)Wikitionary-based %s-%s dictionary", lang1.code, lang2.code),
  98
  99             "--input1=dictInputs/enwiktionary-20110205-pages-articles.xml",
 100             "--input1Name=enwiktionary",
 101             "--input1Format=enwiktionary",
 102             String.format("--input1TranslationPattern1=%s", lang1.nameRegex),
 103             String.format("--input1TranslationPattern2=%s", lang2.nameRegex),
 104             String.format("--input1EnIndex=%d", enIndex),
 105         });
 106
 107         // Print the entries for diffing.
 108         final RandomAccessFile raf = new RandomAccessFile(new File(dictFile), "r");
 109         final Dictionary dict = new Dictionary(raf);
 110         final PrintWriter textOut = new PrintWriter(new File(dictFile + ".txt"));
 111         final List<PairEntry> sorted = new ArrayList<PairEntry>(dict.pairEntries);
 112         Collections.sort(sorted);
 113         for (final PairEntry pairEntry : sorted) {
 114           textOut.println(pairEntry.getRawText(false));
 115         }
 116         textOut.close();
 117         raf.close();
 118
 119       }  // langs2
 120     }  // langs1
 121
 122     DictionaryBuilder.main(new String[] {
 123         "--dictOut=dictOutputs/DE-EN_chemnitz.quickdic",
 124         "--lang1=DE",
 125         "--lang2=EN",
 126         "--dictInfo=@dictInputs/de-en_chemnitz.info",
 127
 128         "--input1=dictInputs/de-en_chemnitz.txt",
 129         "--input1Name=chemnitz",
 130         "--input1Charset=UTF8",
 131         "--input1Format=chemnitz",
 132     });
 133
 134     DictionaryBuilder.main(new String[] {
 135         "--dictOut=dictOutputs/de-en_all.quickdic",
 136         "--lang1=DE",
 137         "--lang2=EN",
 138         "--dictInfo=@dictInputs/de-en_all.info",
 139
 140         "--input2=dictInputs/de-en_chemnitz.txt",
 141         "--input2Name=dictcc",
 142         "--input2Charset=UTF8",
 143         "--input2Format=chemnitz",
 144
 145         "--input3=dictInputs/de-en_dictcc.txt",
 146         "--input3Name=dictcc",
 147         "--input3Charset=UTF8",
 148         "--input3Format=dictcc",
 149
 150         "--input1=dictInputs/enwiktionary-20101015-pages-articles",
 151         "--input1Name=enwiktionary",
 152         "--input1Format=enwiktionary",
 153         "--input1TranslationPattern1=^German$",
 154         "--input1TranslationPattern2=^English$",
 155         "--input1EnIndex=2",
 156
 157     });
 158
 159   }
 160
 161 }