]> gitweb.fperrin.net Git - DictionaryPC.git/blob - src/com/hughes/android/dictionary/engine/LanguageTest.java
7037172d88e8bee2327440f5288445c0b018da48
[DictionaryPC.git] / 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.LinkedHashSet;
21 import java.util.List;
22 import java.util.Set;
23
24 import junit.framework.TestCase;
25
26 import com.hughes.android.dictionary.parser.wiktionary.WiktionaryLangs;
27 import com.ibm.icu.text.Transliterator;
28
29 public class LanguageTest extends TestCase {
30   
31   public void testGermanSort() {
32     final Transliterator normalizer = Transliterator.createFromRules("", Language.de.getDefaultNormalizerRules(), Transliterator.FORWARD);
33     assertEquals("aüääss", normalizer.transform("aueAeAEß"));
34     final List<String> words = Arrays.asList(
35         "er-ben",
36         "erben",
37         "Erben",
38         "Erbse",
39         "Erbsen",
40         "essen",
41         "Essen",
42         "Grosformat",
43         "Grosformats",
44         "Grossformat",
45         "Großformat",
46         "Grossformats",
47         "Großformats",
48         "Großpoo",
49         "Großpoos",
50         "Hörvermögen",
51         "Hörweite",
52         "hos",
53         "Höschen",
54         "Hostel",
55         "hulle",
56         "Hulle",
57         "huelle",
58         "Huelle",
59         "hülle",
60         "Hülle",
61         "Huellen",
62         "Hüllen",
63         "Hum"
64         );
65     final NormalizeComparator comparator = new NormalizeComparator(normalizer, Language.de.getCollator());
66     assertEquals(1, comparator.compare("hülle", "huelle"));
67     assertEquals(-1, comparator.compare("huelle", "hülle"));
68     
69     assertEquals(-1, comparator.compare("hülle", "Hülle"));
70     
71     assertEquals("hülle", normalizer.transform("Hülle"));
72     assertEquals("hulle", normalizer.transform("Hulle"));
73
74     
75     final List<String> sorted = new ArrayList<String>(words);
76 //    Collections.shuffle(shuffled, new Random(0));
77     Collections.sort(sorted, comparator);
78     System.out.println(sorted.toString());
79     for (int i = 0; i < words.size(); ++i) {
80       System.out.println(words.get(i) + "\t" + sorted.get(i));
81       assertEquals(words.get(i), sorted.get(i));
82     }
83   }
84
85   public void testEnglishSort() {
86     final Transliterator normalizer = Transliterator.createFromRules("", Language.en.getDefaultNormalizerRules(), Transliterator.FORWARD);
87
88     final List<String> words = Arrays.asList(
89         "pre-print", 
90         "preppie", 
91         "preppy",
92         "preprocess");
93     
94     final List<String> sorted = new ArrayList<String>(words);
95     final NormalizeComparator comparator = new NormalizeComparator(normalizer, Language.en.getCollator());
96     Collections.sort(sorted, comparator);
97     for (int i = 0; i < words.size(); ++i) {
98       if (i > 0) {
99         assertTrue(comparator.compare(words.get(i-1), words.get(i)) < 0);
100       }
101       System.out.println(words.get(i) + "\t" + sorted.get(i));
102       assertEquals(words.get(i), sorted.get(i));
103     }
104     
105     assertTrue(comparator.compare("pre-print", "preppy") < 0);
106
107   }
108   
109   public void testLanguage() {
110     assertEquals(Language.de, Language.lookup("de"));
111     assertEquals(Language.en, Language.lookup("en"));
112     assertEquals("es", Language.lookup("es").getIsoCode());
113   }
114
115   public void testTextNorm() {
116     //final Transliterator transliterator = Transliterator.getInstance("Any-Latin; Upper; Lower; 'oe' > 'o'; NFD; [:Nonspacing Mark:] Remove; NFC", Transliterator.FORWARD);
117     final Transliterator transliterator = Transliterator.createFromRules("", ":: Any-Latin; :: Upper; :: Lower; 'oe' > 'o'; :: NFD; :: [:Nonspacing Mark:] Remove; :: NFC ;", Transliterator.FORWARD);
118     assertEquals("hoschen", transliterator.transliterate("Höschen"));
119     assertEquals("hoschen", transliterator.transliterate("Hoeschen"));
120     assertEquals("grosspoo", transliterator.transliterate("Großpoo"));
121
122     assertEquals("kyanpasu", transliterator.transliterate("キャンパス"));
123     assertEquals("alphabetikos katalogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
124     assertEquals("biologiceskom", transliterator.transliterate("биологическом"));
125   }
126   public void testHalfTextNorm() {
127     final Transliterator transliterator = Transliterator.createFromRules("", ":: Any-Latin; ' ' > ; :: Lower; ", Transliterator.FORWARD);
128     assertEquals("kyanpasu", transliterator.transliterate("キャンパス"));
129     assertEquals("alphabētikóskatálogos", transliterator.transliterate("Αλφαβητικός Κατάλογος"));
130     assertEquals("biologičeskom", transliterator.transliterate("биологическом"));
131
132     assertEquals("xièxiè", transliterator.transliterate("謝謝"));
133     assertEquals("xièxiè", transliterator.transliterate("谢谢"));
134
135     assertEquals("diànnǎo", transliterator.transliterate("電腦"));
136     assertEquals("diànnǎo", transliterator.transliterate("电脑"));
137     assertEquals("jìsuànjī", transliterator.transliterate("計算機"));
138     assertEquals("jìsuànjī", transliterator.transliterate("计算机"));
139   }
140
141   
142   public void testChinese() {
143     final Language zh = Language.lookup("zh");
144     final Transliterator transliterator = Transliterator.createFromRules("", zh.getDefaultNormalizerRules(), Transliterator.FORWARD);
145     
146     assertEquals("xiexie", transliterator.transliterate("謝謝"));
147     assertEquals("xiexie", transliterator.transliterate("谢谢"));
148
149     assertEquals("diannao", transliterator.transliterate("電腦"));
150     assertEquals("diannao", transliterator.transliterate("电脑"));
151     assertEquals("jisuanji", transliterator.transliterate("計算機"));
152     assertEquals("jisuanji", transliterator.transliterate("计算机"));
153     
154     assertEquals("chengjiu", transliterator.transliterate("成就"));
155     
156   }
157   
158   public void testArabic() {
159     final Language ar = Language.lookup("ar");
160     final Transliterator transliterator = Transliterator.createFromRules("", ar.getDefaultNormalizerRules(), Transliterator.FORWARD);
161     // These don't seem quite right....
162     assertEquals("haswb", transliterator.transliterate("حاسوب"));
163     assertEquals("kmbywtr", transliterator.transliterate("كمبيوتر"));
164
165     assertEquals("{\u200e كمبيوتر \u200e}", Language.fixBidiText("{كمبيوتر}"));
166     assertEquals("{a=\u200e كمبيوتر \u200e}", Language.fixBidiText("{a=كمبيوتر}"));
167     assertEquals("(\u200e كمبيوتر \u200e)", Language.fixBidiText("(كمبيوتر)"));
168     assertEquals("أنثى أنْثَى (’únθā) {f}, إناث (’ināθ) {p}, اناثى (’anāθā) {p}", Language.fixBidiText("أنثى أنْثَى (’únθā) {f}, إناث (’ināθ) {p}, اناثى (’anāθā) {p}"));
169        
170   }
171
172   public void testEnWiktionaryNames() {
173     final Set<String> enLangs = new LinkedHashSet<String>(WiktionaryLangs.isoCodeToWikiName.keySet());
174     for (final String code : WiktionaryLangs.isoCodeToWikiName.keySet()) {
175       enLangs.add(code.toLowerCase());
176     }
177     assertEquals(enLangs.toString(), Language.isoCodeToResources.keySet().toString());
178     assertEquals(enLangs, Language.isoCodeToResources.keySet());
179   }
180
181 }