]> gitweb.fperrin.net Git - DictionaryPC.git/blob - src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
Synonyms, antonyms.
[DictionaryPC.git] / src / com / hughes / android / dictionary / parser / wiktionary / WiktionaryLangs.java
1 // Copyright 2012 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.parser.wiktionary;
16
17 import com.hughes.android.dictionary.engine.Language;
18
19 import java.util.LinkedHashMap;
20 import java.util.Map;
21 import java.util.regex.Pattern;
22
23 public class WiktionaryLangs {
24   
25   public static final Map<String,String> isoCodeToEnWikiName = new LinkedHashMap<String,String>();
26   static {
27     isoCodeToEnWikiName.put("AF", "Afrikaans");
28     isoCodeToEnWikiName.put("SQ", "Albanian");
29     isoCodeToEnWikiName.put("AR", "Arabic");
30     isoCodeToEnWikiName.put("HY", "Armenian");
31     isoCodeToEnWikiName.put("BE", "Belarusian");
32     isoCodeToEnWikiName.put("BN", "Bengali");
33     isoCodeToEnWikiName.put("BS", "Bosnian");
34     isoCodeToEnWikiName.put("BG", "Bulgarian");
35     isoCodeToEnWikiName.put("MY", "Burmese");
36     isoCodeToEnWikiName.put("yue", "Cantonese");
37     isoCodeToEnWikiName.put("CA", "Catalan");
38     isoCodeToEnWikiName.put("HR", "Croatian");
39     isoCodeToEnWikiName.put("CS", "Czech");
40     isoCodeToEnWikiName.put("ZH", "Chinese|Mandarin");
41     isoCodeToEnWikiName.put("DA", "Danish");
42     isoCodeToEnWikiName.put("NL", "Dutch");
43     isoCodeToEnWikiName.put("EN", "English");
44     isoCodeToEnWikiName.put("EO", "Esperanto");
45     isoCodeToEnWikiName.put("ET", "Estonian");
46     isoCodeToEnWikiName.put("FI", "Finnish");
47     isoCodeToEnWikiName.put("FR", "French");
48     isoCodeToEnWikiName.put("DE", "German");
49     isoCodeToEnWikiName.put("EL", "Greek");
50     isoCodeToEnWikiName.put("grc", "Ancient Greek");
51     isoCodeToEnWikiName.put("haw", "Hawaiian");
52     isoCodeToEnWikiName.put("HE", "Hebrew");
53     isoCodeToEnWikiName.put("HI", "Hindi");
54     isoCodeToEnWikiName.put("HU", "Hungarian");
55     isoCodeToEnWikiName.put("IS", "Icelandic");
56     isoCodeToEnWikiName.put("ID", "Indonesian");
57     isoCodeToEnWikiName.put("GA", "Irish");
58     isoCodeToEnWikiName.put("GD", "Gaelic");
59     isoCodeToEnWikiName.put("IT", "Italian");
60     isoCodeToEnWikiName.put("LA", "Latin");
61     isoCodeToEnWikiName.put("LV", "Latvian");
62     isoCodeToEnWikiName.put("LT", "Lithuanian");
63     isoCodeToEnWikiName.put("JA", "Japanese");
64     isoCodeToEnWikiName.put("KO", "Korean");
65     isoCodeToEnWikiName.put("KU", "Kurdish");
66     isoCodeToEnWikiName.put("LO", "Lao");
67     isoCodeToEnWikiName.put("MS", "Malay");
68     isoCodeToEnWikiName.put("ML", "Malayalam");
69     isoCodeToEnWikiName.put("MI", "Maori");
70     isoCodeToEnWikiName.put("MN", "Mongolian");
71     isoCodeToEnWikiName.put("NE", "Nepali");
72     isoCodeToEnWikiName.put("NO", "Norwegian");
73     isoCodeToEnWikiName.put("FA", "Persian");
74     isoCodeToEnWikiName.put("PL", "Polish");
75     isoCodeToEnWikiName.put("PT", "Portuguese");
76     isoCodeToEnWikiName.put("PA", "Punjabi");
77     isoCodeToEnWikiName.put("RO", "Romanian");
78     isoCodeToEnWikiName.put("RU", "Russian");
79     isoCodeToEnWikiName.put("SA", "Sanskrit");
80     isoCodeToEnWikiName.put("SR", "Serbian");
81     isoCodeToEnWikiName.put("SK", "Slovak");
82     isoCodeToEnWikiName.put("SL", "Slovene|Slovenian");
83     isoCodeToEnWikiName.put("SO", "Somali");
84     isoCodeToEnWikiName.put("ES", "Spanish");
85     isoCodeToEnWikiName.put("SW", "Swahili");
86     isoCodeToEnWikiName.put("SV", "Swedish");
87     isoCodeToEnWikiName.put("TL", "Tagalog");
88     isoCodeToEnWikiName.put("TG", "Tajik");
89     isoCodeToEnWikiName.put("TA", "Tamil");
90     isoCodeToEnWikiName.put("TH", "Thai");
91     isoCodeToEnWikiName.put("BO", "Tibetan");
92     isoCodeToEnWikiName.put("TR", "Turkish");
93     isoCodeToEnWikiName.put("UK", "Ukrainian");
94     isoCodeToEnWikiName.put("UR", "Urdu");
95     isoCodeToEnWikiName.put("VI", "Vietnamese");
96     isoCodeToEnWikiName.put("CI", "Welsh");
97     isoCodeToEnWikiName.put("YI", "Yiddish");
98     isoCodeToEnWikiName.put("ZU", "Zulu");
99     
100     isoCodeToEnWikiName.put("AZ", "Azeri");
101     isoCodeToEnWikiName.put("EU", "Basque");
102     isoCodeToEnWikiName.put("BR", "Breton");
103     isoCodeToEnWikiName.put("MR", "Burmese");
104     isoCodeToEnWikiName.put("FO", "Faroese");
105     isoCodeToEnWikiName.put("GL", "Galician");
106     isoCodeToEnWikiName.put("KA", "Georgian");
107     isoCodeToEnWikiName.put("HT", "Haitian Creole");
108     isoCodeToEnWikiName.put("LB", "Luxembourgish");
109     isoCodeToEnWikiName.put("MK", "Macedonian");
110     
111     assert isoCodeToEnWikiName.keySet().equals(Language.isoCodeToResources.keySet());
112   }
113
114   public static final Map<String,Map<String,String>> wikiCodeToIsoCodeToWikiName = new LinkedHashMap<String, Map<String,String>>();
115   static {
116     // en
117     wikiCodeToIsoCodeToWikiName.put("en", isoCodeToEnWikiName);
118     
119     Map<String,String> isoCodeToWikiName;
120     
121     // egrep -o '\{\{Wortart[^}]+\}\}' dewiktionary-pages-articles.xml | cut -d \| -f3 | sort | uniq -c | sort -nr
122     isoCodeToWikiName = new LinkedHashMap<String, String>();
123     wikiCodeToIsoCodeToWikiName.put("de", isoCodeToWikiName);
124     isoCodeToWikiName.put("DE", "Deutsch");
125     isoCodeToWikiName.put("EN", "Englisch");
126     isoCodeToWikiName.put("IT", "Italienisch");
127     isoCodeToWikiName.put("PL", "Polnisch");
128     isoCodeToWikiName.put("FR", "Französisch");
129     isoCodeToWikiName.put("EO", "Esperanto");
130     isoCodeToWikiName.put("CA", "Katalanisch");
131     isoCodeToWikiName.put("LA", "Lateinisch");
132     isoCodeToWikiName.put("CS", "Tschechisch");
133     isoCodeToWikiName.put("HU", "Ungarisch");
134     isoCodeToWikiName.put("SV", "Schwedisch");
135     isoCodeToWikiName.put("ES", "Spanisch");
136
137     // egrep -o '\{\{=[a-zA-Z]+=\}\}' frwiktionary-pages-articles.xml | sort | uniq -c | sort -nr
138     isoCodeToWikiName = new LinkedHashMap<String, String>();
139     wikiCodeToIsoCodeToWikiName.put("fr", isoCodeToWikiName);
140     isoCodeToWikiName.put("FR", Pattern.quote("{{=fr=}}"));
141     isoCodeToWikiName.put("RU", Pattern.quote("{{=ru=}}"));
142     isoCodeToWikiName.put("BG", Pattern.quote("{{=bg=}}"));  // Bulgarian
143     isoCodeToWikiName.put("EN", Pattern.quote("{{=en=}}"));
144     //isoCodeToWikiName.put("", Pattern.quote("{{=sl=}}"));
145     isoCodeToWikiName.put("LA", Pattern.quote("{{=la=}}"));
146     isoCodeToWikiName.put("IT", Pattern.quote("{{=it=}}"));
147     isoCodeToWikiName.put("EO", Pattern.quote("{{=eo=}}"));
148     isoCodeToWikiName.put("CS", Pattern.quote("{{=cs=}}"));  // Czech
149     isoCodeToWikiName.put("NL", Pattern.quote("{{=nl=}}"));  // Dutch
150     //isoCodeToWikiName.put("", Pattern.quote("{{=mg=}}"));
151     //isoCodeToWikiName.put("", Pattern.quote("{{=hsb=}}"));
152     isoCodeToWikiName.put("ZH", Pattern.quote("{{=zh=}}"));
153     isoCodeToWikiName.put("JA", Pattern.quote("{{=ja=}}"));
154     isoCodeToWikiName.put("DE", Pattern.quote("{{=de=}}"));
155     isoCodeToWikiName.put("IS", Pattern.quote("{{=is=}}"));  // Icelandic
156     isoCodeToWikiName.put("ES", Pattern.quote("{{=es=}}"));
157     isoCodeToWikiName.put("UK", Pattern.quote("{{=uk=}}"));
158
159     // egrep -o '= *\{\{-[a-z]+-\}\} *=' itwiktionary-pages-articles.xml | sort | uniq -c | sort -n
160     isoCodeToWikiName = new LinkedHashMap<String, String>();
161     wikiCodeToIsoCodeToWikiName.put("it", isoCodeToWikiName);
162     isoCodeToWikiName.put("IT", "\\{\\{-(it|scn|nap|cal|lmo)-\\}\\}");  // scn, nap, cal, lmo
163     isoCodeToWikiName.put("EN", Pattern.quote("{{-en-}}"));
164     isoCodeToWikiName.put("FR", Pattern.quote("{{-fr-}}"));
165     isoCodeToWikiName.put("DE", Pattern.quote("{{-de-}}"));
166     isoCodeToWikiName.put("ES", Pattern.quote("{{-es-}}"));
167     isoCodeToWikiName.put("JA", Pattern.quote("{{-ja-}}"));
168     isoCodeToWikiName.put("PL", Pattern.quote("{{-pl-}}"));
169     isoCodeToWikiName.put("NL", Pattern.quote("{{-nl-}}"));
170     isoCodeToWikiName.put("LV", Pattern.quote("{{-lv-}}"));
171     isoCodeToWikiName.put("LA", Pattern.quote("{{-la-}}"));
172     isoCodeToWikiName.put("HU", Pattern.quote("{{-hu-}}"));
173     isoCodeToWikiName.put("EL", Pattern.quote("{{-grc-}}"));
174     isoCodeToWikiName.put("SV", Pattern.quote("{{-sv-}}"));
175
176   }
177   
178 }