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