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

   1 // Copyright 2011 Google Inc. All Rights Reserved.
   2 //
   3 // Licensed under the Apache License, Version 2.0 (the "License");
   4 // you may not use this file except in compliance with the License.
   5 // You may obtain a copy of the License at
   6 //
   7 //     http://www.apache.org/licenses/LICENSE-2.0
   8 //
   9 // Unless required by applicable law or agreed to in writing, software
  10 // distributed under the License is distributed on an "AS IS" BASIS,
  11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  12 // See the License for the specific language governing permissions and
  13 // limitations under the License.
  14
  15 package com.hughes.android.dictionary.engine;
  16
  17 import java.io.File;
  18 import java.io.PrintWriter;
  19 import java.io.RandomAccessFile;
  20 import java.util.ArrayList;
  21 import java.util.Collections;
  22 import java.util.List;
  23
  24 import junit.framework.TestCase;
  25
  26 public class DictionaryBuilderMain extends TestCase {
  27
  28   static class Lang {
  29     final String nameRegex;
  30     final String code;
  31     public Lang(String nameRegex, String code) {
  32       this.nameRegex = nameRegex;
  33       this.code = code;
  34     }
  35   }
  36
  37
  38   public static void main(final String[] args) throws Exception {
  39
  40     Lang[] langs1 = new Lang[] {
  41         new Lang("^English$", "EN"),
  42         //new Lang("^German$", "DE"),
  43     };
  44     Lang[] langs2 = new Lang[] {
  45         new Lang("^.*Italian.*$", "IT"),
  46         new Lang("^.*Greek.*$", "EL"),
  47         new Lang("^.*Spanish.*$", "ES"),
  48         /*
  49         new Lang("^German$", "DE"),
  50         new Lang("^Afrikaans$", "AF"),
  51         new Lang("^Armenian$", "HY"),
  52         new Lang("^Arabic$", "AR"),
  53         new Lang("^Chinese$|^Mandarin$", "ZH"),
  54         new Lang("^Croation$", "HR"),
  55         new Lang("^Czech$", "CS"),
  56         new Lang("^Dutch$", "NL"),
  57         new Lang("^English$", "EN"),
  58         new Lang("^Finnish$", "FI"),
  59         new Lang("^French$", "FR"),
  60         new Lang("^Hebrew$", "HE"),
  61         new Lang("^Hindi$", "HI"),
  62         new Lang("^Icelandic$", "IS"),
  63         new Lang("^Irish$", "GA"),
  64         new Lang("^Japanese$", "JA"),
  65         new Lang("^Korean$", "KO"),
  66         new Lang("^Kurdish$", "KU"),
  67         new Lang("^Lithuanian$", "LT"),
  68         new Lang("^Malay$", "MS"),
  69         new Lang("^Maori$", "MI"),
  70         new Lang("^Mongolian$", "MN"),
  71         new Lang("^Norwegian$", "NO"),
  72         new Lang("^Persian$", "FA"),
  73         new Lang("^Portuguese$", "PT"),
  74         new Lang("^Romanian$", "RO"),
  75         new Lang("^Russian$", "RU"),
  76         new Lang("^Sanskrit$", "SA"),
  77         new Lang("^Serbian$", "SR"),
  78         new Lang("^Somali$", "SO"),
  79         new Lang("^Sudanese$", "SU"),
  80         new Lang("^Swedish$", "SV"),
  81         new Lang("^Tajik$", "TG"),
  82         new Lang("^Thai$", "TH"),
  83         new Lang("^Tibetan$", "BO"),
  84         new Lang("^Turkish$", "TR"),
  85         new Lang("^Ukranian$", "UK"),
  86         new Lang("^Vietnamese$", "VI"),
  87         new Lang("^Welsh$", "CY"),
  88         new Lang("^Yiddish$", "YI"),
  89         new Lang("^Zulu$", "ZU"),*/
  90     };
  91
  92     for (final Lang lang1 : langs1) {
  93       for (final Lang lang2 : langs2) {
  94         if (lang1.nameRegex.equals(lang2.nameRegex)) {
  95           continue;
  96         }
  97
  98         int enIndex = -1;
  99         Lang nonEnglish = null;
 100         if (lang2.code.equals("EN")) {
 101           enIndex = 2;
 102           nonEnglish = lang1;
 103         }
 104         if (lang1.code.equals("EN")) {
 105           enIndex = 1;
 106           nonEnglish = lang2;
 107         }
 108         assert nonEnglish != null;
 109
 110         final String dictFile = String.format("dictOutputs/%s-%s_enwiktionary.quickdic", lang1.code, lang2.code);
 111         System.out.println("building dictFile: " + dictFile);
 112         DictionaryBuilder.main(new String[] {
 113             String.format("--dictOut=%s", dictFile),
 114             String.format("--lang1=%s", lang1.code),
 115             String.format("--lang2=%s", lang2.code),
 116             String.format("--dictInfo=(EN)Wikitionary-based %s-%s dictionary", lang1.code, lang2.code),
 117
 118             "--input3=wikiSplit/english.data",
 119             "--input3Name=enwiktionary.english",
 120             "--input3Format=enwiktionary",
 121             "--input3LangPattern=" + nonEnglish.nameRegex,
 122             "--input3LangCodePattern=" + (enIndex == 1 ? lang2.code : lang1.code).toLowerCase(),
 123             "--input3EnIndex=" + enIndex,
 124
 125         });
 126
 127         // Print the entries for diffing.
 128         final RandomAccessFile raf = new RandomAccessFile(new File(dictFile), "r");
 129         final Dictionary dict = new Dictionary(raf);
 130         final PrintWriter textOut = new PrintWriter(new File(dictFile + ".txt"));
 131         final List<PairEntry> sorted = new ArrayList<PairEntry>(dict.pairEntries);
 132         Collections.sort(sorted);
 133         for (final PairEntry pairEntry : sorted) {
 134           textOut.println(pairEntry.getRawText(false));
 135         }
 136         textOut.close();
 137         raf.close();
 138
 139       }  // langs2
 140     }  // langs1
 141
 142     DictionaryBuilder.main(new String[] {
 143         "--dictOut=dictOutputs/DE-EN_chemnitz.quickdic",
 144         "--lang1=DE",
 145         "--lang2=EN",
 146         "--dictInfo=@dictInputs/de-en_chemnitz.info",
 147
 148         "--input1=dictInputs/de-en_chemnitz.txt",
 149         "--input1Name=chemnitz",
 150         "--input1Charset=UTF8",
 151         "--input1Format=chemnitz",
 152     });
 153
 154     DictionaryBuilder.main(new String[] {
 155         "--dictOut=dictOutputs/de-en_all.quickdic",
 156         "--lang1=DE",
 157         "--lang2=EN",
 158         "--dictInfo=@dictInputs/de-en_all.info",
 159
 160         "--input2=dictInputs/de-en_chemnitz.txt",
 161         "--input2Name=dictcc",
 162         "--input2Charset=UTF8",
 163         "--input2Format=chemnitz",
 164
 165         "--input3=dictInputs/de-en_dictcc.txt",
 166         "--input3Name=dictcc",
 167         "--input3Charset=UTF8",
 168         "--input3Format=dictcc",
 169
 170         "--input1=dictInputs/enwiktionary-20101015-pages-articles",
 171         "--input1Name=enwiktionary",
 172         "--input1Format=enwiktionary",
 173         "--input1TranslationPattern1=^German$",
 174         "--input1TranslationPattern2=^English$",
 175         "--input1EnIndex=2",
 176
 177     });
 178
 179   }
 180
 181 }