1 // Copyright 2011 Google Inc. All Rights Reserved.
\r
3 // Licensed under the Apache License, Version 2.0 (the "License");
\r
4 // you may not use this file except in compliance with the License.
\r
5 // You may obtain a copy of the License at
\r
7 // http://www.apache.org/licenses/LICENSE-2.0
\r
9 // Unless required by applicable law or agreed to in writing, software
\r
10 // distributed under the License is distributed on an "AS IS" BASIS,
\r
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
\r
12 // See the License for the specific language governing permissions and
\r
13 // limitations under the License.
\r
15 package com.hughes.android.dictionary.engine;
\r
17 import com.hughes.android.dictionary.R;
\r
18 import com.ibm.icu.text.Collator;
\r
20 import java.util.ArrayList;
\r
21 import java.util.LinkedHashMap;
\r
22 import java.util.Locale;
\r
23 import java.util.Map;
\r
24 import java.util.regex.Pattern;
\r
26 public class Language {
\r
28 public static final class LanguageResources {
\r
29 public final String englishName;
\r
30 public final int nameId;
\r
31 public final int flagId;
\r
33 private LanguageResources(final String englishName, int nameId, int flagId) {
\r
34 this.englishName = englishName;
\r
35 this.nameId = nameId;
\r
36 this.flagId = flagId;
\r
39 private LanguageResources(final String englishName, int nameId) {
\r
40 this(englishName, nameId, 0);
\r
45 // http://www.loc.gov/standards/iso639-2/php/code_list.php
\r
46 public static final Map<String,LanguageResources> isoCodeToResources = new LinkedHashMap<String,LanguageResources>();
\r
48 isoCodeToResources.put("AF", new LanguageResources("Afrikaans", R.string.AF, R.drawable.flag_of_south_africa));
\r
49 isoCodeToResources.put("SQ", new LanguageResources("Albanian", R.string.SQ, R.drawable.flag_of_albania));
\r
50 isoCodeToResources.put("AR", new LanguageResources("Arabic", R.string.AR));
\r
51 isoCodeToResources.put("HY", new LanguageResources("Armenian", R.string.HY, R.drawable.flag_of_armenia));
\r
52 isoCodeToResources.put("BE", new LanguageResources("Belarusian", R.string.BE, R.drawable.flag_of_belarus));
\r
53 isoCodeToResources.put("BN", new LanguageResources("Bengali", R.string.BN));
\r
54 isoCodeToResources.put("BS", new LanguageResources("Bosnian", R.string.BS, R.drawable.flag_of_bosnia_and_herzegovina));
\r
55 isoCodeToResources.put("BG", new LanguageResources("Bulgarian", R.string.BG, R.drawable.flag_of_bulgaria));
\r
56 isoCodeToResources.put("MY", new LanguageResources("Burmese", R.string.MY, R.drawable.flag_of_myanmar));
\r
57 isoCodeToResources.put("ZH", new LanguageResources("Chinese", R.string.ZH, R.drawable.flag_of_the_peoples_republic_of_china));
\r
58 isoCodeToResources.put("cmn", new LanguageResources("Mandarin", R.string.cmn, R.drawable.flag_of_the_peoples_republic_of_china));
\r
59 isoCodeToResources.put("yue", new LanguageResources("Cantonese", R.string.yue, R.drawable.flag_of_the_peoples_republic_of_china));
\r
60 isoCodeToResources.put("CA", new LanguageResources("Catalan", R.string.CA));
\r
61 isoCodeToResources.put("HR", new LanguageResources("Croatian", R.string.HR, R.drawable.flag_of_croatia));
\r
62 isoCodeToResources.put("CS", new LanguageResources("Czech", R.string.CS, R.drawable.flag_of_the_czech_republic));
\r
63 isoCodeToResources.put("DA", new LanguageResources("Danish", R.string.DA, R.drawable.flag_of_denmark));
\r
64 isoCodeToResources.put("NL", new LanguageResources("Dutch", R.string.NL, R.drawable.flag_of_the_netherlands));
\r
65 isoCodeToResources.put("EN", new LanguageResources("English", R.string.EN, R.drawable.flag_of_the_united_kingdom));
\r
66 isoCodeToResources.put("EO", new LanguageResources("Esperanto", R.string.EO));
\r
67 isoCodeToResources.put("ET", new LanguageResources("Estonian", R.string.ET, R.drawable.flag_of_estonia));
\r
68 isoCodeToResources.put("FI", new LanguageResources("Finnish", R.string.FI, R.drawable.flag_of_finland));
\r
69 isoCodeToResources.put("FR", new LanguageResources("French", R.string.FR, R.drawable.flag_of_france));
\r
70 isoCodeToResources.put("DE", new LanguageResources("German", R.string.DE, R.drawable.flag_of_germany));
\r
71 isoCodeToResources.put("EL", new LanguageResources("Greek", R.string.EL, R.drawable.flag_of_greece));
\r
72 isoCodeToResources.put("grc", new LanguageResources("Ancient Greek", R.string.grc));
\r
73 isoCodeToResources.put("haw", new LanguageResources("Hawaiian", R.string.haw));
\r
74 isoCodeToResources.put("HE", new LanguageResources("Hebrew", R.string.HE, R.drawable.flag_of_israel));
\r
75 isoCodeToResources.put("HI", new LanguageResources("Hindi", R.string.HI));
\r
76 isoCodeToResources.put("HU", new LanguageResources("Hungarian", R.string.HU, R.drawable.flag_of_hungary));
\r
77 isoCodeToResources.put("IS", new LanguageResources("Icelandic", R.string.IS, R.drawable.flag_of_iceland));
\r
78 isoCodeToResources.put("ID", new LanguageResources("Indonesian", R.string.ID, R.drawable.flag_of_indonesia));
\r
79 isoCodeToResources.put("GA", new LanguageResources("Irish", R.string.GA, R.drawable.flag_of_ireland));
\r
80 isoCodeToResources.put("GD", new LanguageResources("Scottish Gaelic", R.string.GD));
\r
81 isoCodeToResources.put("GV", new LanguageResources("Manx", R.string.GV));
\r
82 isoCodeToResources.put("IT", new LanguageResources("Italian", R.string.IT, R.drawable.flag_of_italy));
\r
83 isoCodeToResources.put("LA", new LanguageResources("Latin", R.string.LA));
\r
84 isoCodeToResources.put("LV", new LanguageResources("Latvian", R.string.LV, R.drawable.flag_of_latvia));
\r
85 isoCodeToResources.put("LT", new LanguageResources("Lithuanian", R.string.LT, R.drawable.flag_of_lithuania));
\r
86 isoCodeToResources.put("JA", new LanguageResources("Japanese", R.string.JA, R.drawable.flag_of_japan));
\r
87 isoCodeToResources.put("KO", new LanguageResources("Korean", R.string.KO, R.drawable.flag_of_south_korea));
\r
88 isoCodeToResources.put("KU", new LanguageResources("Kurdish", R.string.KU));
\r
89 isoCodeToResources.put("MS", new LanguageResources("Malay", R.string.MS, R.drawable.flag_of_malaysia));
\r
90 isoCodeToResources.put("MI", new LanguageResources("Maori", R.string.MI));
\r
91 isoCodeToResources.put("MN", new LanguageResources("Mongolian", R.string.MN, R.drawable.flag_of_mongolia));
\r
92 isoCodeToResources.put("NE", new LanguageResources("Nepali", R.string.NE, R.drawable.flag_of_nepal));
\r
93 isoCodeToResources.put("NO", new LanguageResources("Norwegian", R.string.NO, R.drawable.flag_of_norway));
\r
94 isoCodeToResources.put("FA", new LanguageResources("Persian", R.string.FA, R.drawable.flag_of_iran));
\r
95 isoCodeToResources.put("PL", new LanguageResources("Polish", R.string.PL, R.drawable.flag_of_poland));
\r
96 isoCodeToResources.put("PT", new LanguageResources("Portuguese", R.string.PT));
\r
97 isoCodeToResources.put("PA", new LanguageResources("Punjabi", R.string.PA));
\r
98 isoCodeToResources.put("RO", new LanguageResources("Romanian", R.string.RO, R.drawable.flag_of_romania));
\r
99 isoCodeToResources.put("RU", new LanguageResources("Russian", R.string.RU, R.drawable.flag_of_russia));
\r
100 isoCodeToResources.put("SA", new LanguageResources("Sanskrit", R.string.SA));
\r
101 isoCodeToResources.put("SR", new LanguageResources("Serbian", R.string.SR, R.drawable.flag_of_serbia));
\r
102 isoCodeToResources.put("SK", new LanguageResources("Slovak", R.string.SK, R.drawable.flag_of_slovakia));
\r
103 isoCodeToResources.put("SL", new LanguageResources("Slovenian", R.string.SL, R.drawable.flag_of_slovenia));
\r
104 isoCodeToResources.put("SO", new LanguageResources("Somali", R.string.SO, R.drawable.flag_of_somalia));
\r
105 isoCodeToResources.put("ES", new LanguageResources("Spanish", R.string.ES, R.drawable.flag_of_spain));
\r
106 isoCodeToResources.put("SW", new LanguageResources("Swahili", R.string.SW));
\r
107 isoCodeToResources.put("SV", new LanguageResources("Swedish", R.string.SV, R.drawable.flag_of_sweden));
\r
108 isoCodeToResources.put("TL", new LanguageResources("Tagalog", R.string.TL));
\r
109 isoCodeToResources.put("TG", new LanguageResources("Tajik", R.string.TG, R.drawable.flag_of_tajikistan));
\r
110 isoCodeToResources.put("TH", new LanguageResources("Thai", R.string.TH, R.drawable.flag_of_thailand));
\r
111 isoCodeToResources.put("BO", new LanguageResources("Tibetan", R.string.BO));
\r
112 isoCodeToResources.put("TR", new LanguageResources("Turkish", R.string.TR, R.drawable.flag_of_turkey));
\r
113 isoCodeToResources.put("UK", new LanguageResources("Ukrainian", R.string.UK, R.drawable.flag_of_ukraine));
\r
114 isoCodeToResources.put("UR", new LanguageResources("Urdu", R.string.UR));
\r
115 isoCodeToResources.put("VI", new LanguageResources("Vietnamese", R.string.VI, R.drawable.flag_of_vietnam));
\r
116 isoCodeToResources.put("CI", new LanguageResources("Welsh", R.string.CI));
\r
117 isoCodeToResources.put("YI", new LanguageResources("Yiddish", R.string.YI));
\r
118 isoCodeToResources.put("ZU", new LanguageResources("Zulu", R.string.ZU));
\r
119 isoCodeToResources.put("AZ", new LanguageResources("Azeri", R.string.AZ, R.drawable.flag_of_azerbaijan));
\r
120 isoCodeToResources.put("EU", new LanguageResources("Basque", R.string.EU));
\r
121 isoCodeToResources.put("BR", new LanguageResources("Breton", R.string.BR));
\r
122 isoCodeToResources.put("MR", new LanguageResources("Burmese", R.string.MR));
\r
123 isoCodeToResources.put("FO", new LanguageResources("Faroese", R.string.FO));
\r
124 isoCodeToResources.put("GL", new LanguageResources("Galician", R.string.GL));
\r
125 isoCodeToResources.put("KA", new LanguageResources("Georgian", R.string.KA, R.drawable.flag_of_georgia));
\r
126 isoCodeToResources.put("HT", new LanguageResources("Haitian Creole", R.string.HT, R.drawable.flag_of_haiti));
\r
127 isoCodeToResources.put("LB", new LanguageResources("Luxembourgish", R.string.LB, R.drawable.flag_of_luxembourg));
\r
128 isoCodeToResources.put("MK", new LanguageResources("Macedonian", R.string.MK));
\r
129 isoCodeToResources.put("LO", new LanguageResources("Lao", R.string.LO));
\r
130 isoCodeToResources.put("ML", new LanguageResources("Malayalam", R.string.ML));
\r
131 isoCodeToResources.put("SL", new LanguageResources("Slovenian", R.string.SL, R.drawable.flag_of_slovenia));
\r
132 isoCodeToResources.put("TA", new LanguageResources("Tamil", R.string.TA));
\r
133 isoCodeToResources.put("SH", new LanguageResources("Serbo-Croatian", R.string.SH));
\r
135 // Hack to allow lower-case ISO codes to work:
\r
136 for (final String isoCode : new ArrayList<String>(isoCodeToResources.keySet())) {
\r
137 isoCodeToResources.put(isoCode.toLowerCase(), isoCodeToResources.get(isoCode));
\r
143 private static final Map<String, Language> registry = new LinkedHashMap<String, Language>();
\r
145 final String isoCode;
\r
146 final Locale locale;
\r
148 private Collator collator;
\r
150 private Language(final Locale locale, final String isoCode) {
\r
151 this.locale = locale;
\r
152 this.isoCode = isoCode;
\r
154 registry.put(isoCode.toLowerCase(), this);
\r
158 public String toString() {
\r
159 return locale.toString();
\r
162 public String getIsoCode() {
\r
166 public synchronized Collator getCollator() {
\r
167 // Don't think this is thread-safe...
\r
168 // if (collator == null) {
\r
169 this.collator = Collator.getInstance(locale);
\r
170 this.collator.setStrength(Collator.IDENTICAL);
\r
175 public String getDefaultNormalizerRules() {
\r
176 return ":: Any-Latin; ' ' > ; :: Lower; :: NFD; :: [:Nonspacing Mark:] Remove; :: NFC ;";
\r
180 * A practical pattern to identify strong RTL characters. This pattern is not
\r
181 * completely correct according to the Unicode standard. It is simplified for
\r
182 * performance and small code size.
\r
184 private static final String rtlChars =
\r
185 "\u0591-\u07FF\uFB1D-\uFDFD\uFE70-\uFEFC";
\r
187 private static final String puncChars =
\r
188 "\\[\\]\\(\\)\\{\\}\\=";
\r
190 private static final Pattern RTL_LEFT_BOUNDARY = Pattern.compile("(["+ puncChars +"])([" + rtlChars + "])");
\r
191 private static final Pattern RTL_RIGHT_BOUNDARY = Pattern.compile("([" + rtlChars + "])(["+ puncChars +"])");
\r
193 public static String fixBidiText(String text) {
\r
194 // text = RTL_LEFT_BOUNDARY.matcher(text).replaceAll("$1\u200e $2");
\r
195 // text = RTL_RIGHT_BOUNDARY.matcher(text).replaceAll("$1 \u200e$2");
\r
199 // ----------------------------------------------------------------
\r
201 public static final Language en = new Language(Locale.ENGLISH, "EN");
\r
202 public static final Language fr = new Language(Locale.FRENCH, "FR");
\r
203 public static final Language it = new Language(Locale.ITALIAN, "IT");
\r
205 public static final Language de = new Language(Locale.GERMAN, "DE") {
\r
207 public String getDefaultNormalizerRules() {
\r
208 return ":: Lower; 'ae' > 'ä'; 'oe' > 'ö'; 'ue' > 'ü'; 'ß' > 'ss'; ";
\r
212 // ----------------------------------------------------------------
\r
214 public static synchronized Language lookup(final String isoCode) {
\r
215 Language lang = registry.get(isoCode.toLowerCase());
\r
216 if (lang == null) {
\r
217 lang = new Language(new Locale(isoCode), isoCode);
\r