src/com/hughes/android/dictionary/engine/LanguageTest.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.util.ArrayList;
  18 import java.util.Arrays;
  19 import java.util.Collections;
  20 import java.util.List;
  21
  22 import junit.framework.TestCase;
  23
  24 import com.ibm.icu.text.Transliterator;
  25
  26 public class LanguageTest extends TestCase {
  27
  28   public void testGermanSort() {
  29     final Transliterator normalizer = Transliterator.createFromRules("", Language.de.getDefaultNormalizerRules(), Transliterator.FORWARD);
  30     assertEquals("aüääss", normalizer.transform("aueAeAEß"));
  31     final List<String> words = Arrays.asList(
  32         "er-ben",
  33         "erben",
  34         "Erben",
  35         "Erbse",
  36         "Erbsen",
  37         "essen",
  38         "Essen",
  39         "Grosformat",
  40         "Grosformats",
  41         "Grossformat",
  42         "Großformat",
  43         "Grossformats",
  44         "Großformats",
  45         "Großpoo",
  46         "Großpoos",
  47         "Hörvermögen",
  48         "Hörweite",
  49         "hos",
  50         "Höschen",
  51         "Hostel",
  52         "hulle",
  53         "Hulle",
  54         "huelle",
  55         "Huelle",
  56         "hülle",
  57         "Hülle",
  58         "Huellen",
  59         "Hüllen",
  60         "Hum"
  61         );
  62     final NormalizeComparator comparator = new NormalizeComparator(normalizer, Language.de.getCollator());
  63     assertEquals(1, comparator.compare("hülle", "huelle"));
  64     assertEquals(-1, comparator.compare("huelle", "hülle"));
  65
  66     assertEquals(-1, comparator.compare("hülle", "Hülle"));
  67
  68     assertEquals("hülle", normalizer.transform("Hülle"));
  69     assertEquals("hulle", normalizer.transform("Hulle"));
  70
  71
  72     final List<String> sorted = new ArrayList<String>(words);
  73 //    Collections.shuffle(shuffled, new Random(0));
  74     Collections.sort(sorted, comparator);
  75     System.out.println(sorted.toString());
  76     for (int i = 0; i < words.size(); ++i) {
  77       System.out.println(words.get(i) + "\t" + sorted.get(i));
  78       assertEquals(words.get(i), sorted.get(i));
  79     }
  80   }
  81
  82   public void testEnglishSort() {
  83     final Transliterator normalizer = Transliterator.createFromRules("", Language.en.getDefaultNormalizerRules(), Transliterator.FORWARD);
  84
  85     final List<String> words = Arrays.asList(
  86         "pre-print",
  87         "preppie",
  88         "preppy",
  89         "preprocess");
  90
  91     final List<String> sorted = new ArrayList<String>(words);
  92     final NormalizeComparator comparator = new NormalizeComparator(normalizer, Language.en.getCollator());
  93     Collections.sort(sorted, comparator);
  94     for (int i = 0; i < words.size(); ++i) {
  95       if (i > 0) {
  96         assertTrue(comparator.compare(words.get(i-1), words.get(i)) < 0);
  97       }
  98       System.out.println(words.get(i) + "\t" + sorted.get(i));
  99       assertEquals(words.get(i), sorted.get(i));
 100     }
 101
 102     assertTrue(comparator.compare("pre-print", "preppy") < 0);
 103
 104   }
 105
 106   public void testLanguage() {
 107     assertEquals(Language.de, Language.lookup("de"));
 108     assertEquals(Language.en, Language.lookup("en"));
 109     assertEquals("es", Language.lookup("es").getIsoCode());
 110   }
 111
 112   public void testTextNorm() {
 113     //final Transliterator transliterator = Transliterator.getInstance("Any-Latin; Upper; Lower; 'oe' > 'o'; NFD; [:Nonspacing Mark:] Remove; NFC", Transliterator.FORWARD);
 114     final Transliterator transliterator = Transliterator.createFromRules("", ":: Any-Latin; :: Upper; :: Lower; 'oe' > 'o'; :: NFD; :: [:Nonspacing Mark:] Remove; :: NFC ;", Transliterator.FORWARD);
 115     assertEquals("hoschen", transliterator.transliterate("Höschen"));
 116     assertEquals("hoschen", transliterator.transliterate("Hoeschen"));
 117     assertEquals("grosspoo", transliterator.transliterate("Großpoo"));
 118
 119     assertEquals("kyanpasu", transliterator.transliterate("キャンパス"));
 120     assertEquals("alphabetikos katalogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
 121     assertEquals("biologiceskom", transliterator.transliterate("биологическом"));
 122   }
 123   public void testHalfTextNorm() {
 124     final Transliterator transliterator = Transliterator.createFromRules("", ":: Any-Latin; ' ' > ; :: Lower; ", Transliterator.FORWARD);
 125     assertEquals("kyanpasu", transliterator.transliterate("キャンパス"));
 126     assertEquals("alphabētikóskatálogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
 127     assertEquals("biologičeskom", transliterator.transliterate("биологическом"));
 128
 129     assertEquals("xièxiè", transliterator.transliterate("謝謝"));
 130     assertEquals("xièxiè", transliterator.transliterate("谢谢"));
 131
 132     assertEquals("diànnǎo", transliterator.transliterate("電腦"));
 133     assertEquals("diànnǎo", transliterator.transliterate("电脑"));
 134     assertEquals("jìsuànjī", transliterator.transliterate("計算機"));
 135     assertEquals("jìsuànjī", transliterator.transliterate("计算机"));
 136   }
 137
 138
 139   public void testChinese() {
 140     final Language zh = Language.lookup("zh");
 141     final Transliterator transliterator = Transliterator.createFromRules("", zh.getDefaultNormalizerRules(), Transliterator.FORWARD);
 142
 143     assertEquals("xiexie", transliterator.transliterate("謝謝"));
 144     assertEquals("xiexie", transliterator.transliterate("谢谢"));
 145
 146     assertEquals("diannao", transliterator.transliterate("電腦"));
 147     assertEquals("diannao", transliterator.transliterate("电脑"));
 148     assertEquals("jisuanji", transliterator.transliterate("計算機"));
 149     assertEquals("jisuanji", transliterator.transliterate("计算机"));
 150
 151     assertEquals("chengjiu", transliterator.transliterate("成就"));
 152
 153   }
 154
 155   public void testArabic() {
 156     final Language ar = Language.lookup("ar");
 157     final Transliterator transliterator = Transliterator.createFromRules("", ar.getDefaultNormalizerRules(), Transliterator.FORWARD);
 158     // These don't seem quite right....
 159     assertEquals("haswb", transliterator.transliterate("حاسوب"));
 160     assertEquals("kmbywtr", transliterator.transliterate("كمبيوتر"));
 161
 162     assertEquals("{\u200eكمبيوتر\u200e}", Language.fixBidiText("{كمبيوتر}"));
 163   }
 164
 165 }