package com.hughes.android.dictionary.parser.wiktionary;
+import com.hughes.android.dictionary.engine.Language;
+
import java.util.LinkedHashMap;
+import java.util.LinkedHashSet;
import java.util.Map;
+import java.util.Set;
import java.util.regex.Pattern;
public class WiktionaryLangs {
- public static final Map<String,String> isoCodeToWikiName = new LinkedHashMap<String,String>();
+ public static final Map<String,String> isoCodeToEnWikiName = new LinkedHashMap<String,String>();
static {
- isoCodeToWikiName.put("AF", "Afrikaans");
- isoCodeToWikiName.put("SQ", "Albanian");
- isoCodeToWikiName.put("AR", "Arabic");
- isoCodeToWikiName.put("HY", "Armenian");
- isoCodeToWikiName.put("BE", "Belarusian");
- isoCodeToWikiName.put("BN", "Bengali");
- isoCodeToWikiName.put("BS", "Bosnian");
- isoCodeToWikiName.put("BG", "Bulgarian");
- isoCodeToWikiName.put("CA", "Catalan");
- isoCodeToWikiName.put("HR", "Croatian");
- isoCodeToWikiName.put("CS", "Czech");
- isoCodeToWikiName.put("ZH", "Chinese|Mandarin|Cantonese");
- isoCodeToWikiName.put("DA", "Danish");
- isoCodeToWikiName.put("NL", "Dutch");
- isoCodeToWikiName.put("EN", "English");
- isoCodeToWikiName.put("EO", "Esperanto");
- isoCodeToWikiName.put("ET", "Estonian");
- isoCodeToWikiName.put("FI", "Finnish");
- isoCodeToWikiName.put("FR", "French");
- isoCodeToWikiName.put("DE", "German");
- isoCodeToWikiName.put("EL", "Greek");
- isoCodeToWikiName.put("haw", "Hawaiian");
- isoCodeToWikiName.put("HE", "Hebrew");
- isoCodeToWikiName.put("HI", "Hindi");
- isoCodeToWikiName.put("HU", "Hungarian");
- isoCodeToWikiName.put("IS", "Icelandic");
- isoCodeToWikiName.put("ID", "Indonesian");
- isoCodeToWikiName.put("GA", "Gaelic");
- isoCodeToWikiName.put("IT", "Italian");
- isoCodeToWikiName.put("LA", "Latin");
- isoCodeToWikiName.put("LV", "Latvian");
- isoCodeToWikiName.put("LT", "Lithuanian");
- isoCodeToWikiName.put("JA", "Japanese");
- isoCodeToWikiName.put("KO", "Korean");
- isoCodeToWikiName.put("KU", "Kurdish");
- isoCodeToWikiName.put("MS", "Malay");
- isoCodeToWikiName.put("MI", "Maori");
- isoCodeToWikiName.put("MN", "Mongolian");
- isoCodeToWikiName.put("NE", "Nepali");
- isoCodeToWikiName.put("NO", "Norwegian");
- isoCodeToWikiName.put("FA", "Persian");
- isoCodeToWikiName.put("PL", "Polish");
- isoCodeToWikiName.put("PT", "Portuguese");
- isoCodeToWikiName.put("PA", "Punjabi");
- isoCodeToWikiName.put("RO", "Romanian");
- isoCodeToWikiName.put("RU", "Russian");
- isoCodeToWikiName.put("SA", "Sanskrit");
- isoCodeToWikiName.put("SR", "Serbian");
- isoCodeToWikiName.put("SK", "Slovak");
- isoCodeToWikiName.put("SO", "Somali");
- isoCodeToWikiName.put("ES", "Spanish");
- isoCodeToWikiName.put("SW", "Swahili");
- isoCodeToWikiName.put("SV", "Swedish");
- isoCodeToWikiName.put("TL", "Tagalog");
- isoCodeToWikiName.put("TG", "Tajik");
- isoCodeToWikiName.put("TH", "Thai");
- isoCodeToWikiName.put("BO", "Tibetan");
- isoCodeToWikiName.put("TR", "Turkish");
- isoCodeToWikiName.put("UK", "Ukrainian");
- isoCodeToWikiName.put("UR", "Urdu");
- isoCodeToWikiName.put("VI", "Vietnamese");
- isoCodeToWikiName.put("CI", "Welsh");
- isoCodeToWikiName.put("YI", "Yiddish");
- isoCodeToWikiName.put("ZU", "Zulu");
+ isoCodeToEnWikiName.put("AF", "Afrikaans");
+ isoCodeToEnWikiName.put("SQ", "Albanian");
+ isoCodeToEnWikiName.put("AR", "Arabic");
+ isoCodeToEnWikiName.put("HY", "Armenian");
+ isoCodeToEnWikiName.put("BE", "Belarusian");
+ isoCodeToEnWikiName.put("BN", "Bengali");
+ isoCodeToEnWikiName.put("BG", "Bulgarian");
+ isoCodeToEnWikiName.put("CA", "Catalan");
+ isoCodeToEnWikiName.put("SH", "Serbo-Croatian");
+ isoCodeToEnWikiName.put("CS", "Czech");
+ isoCodeToEnWikiName.put("ZH", "Chinese");
+ isoCodeToEnWikiName.put("cmn", "Mandarin");
+ isoCodeToEnWikiName.put("yue", "Cantonese");
+ isoCodeToEnWikiName.put("DA", "Danish");
+ isoCodeToEnWikiName.put("NL", "Dutch");
+ isoCodeToEnWikiName.put("EN", "English");
+ isoCodeToEnWikiName.put("EO", "Esperanto");
+ isoCodeToEnWikiName.put("ET", "Estonian");
+ isoCodeToEnWikiName.put("FI", "Finnish");
+ isoCodeToEnWikiName.put("FR", "French");
+ isoCodeToEnWikiName.put("DE", "German");
+ isoCodeToEnWikiName.put("grc", "Ancient Greek");
+ isoCodeToEnWikiName.put("EL", "Greek");
+ isoCodeToEnWikiName.put("haw", "Hawaiian");
+ isoCodeToEnWikiName.put("HE", "Hebrew");
+ isoCodeToEnWikiName.put("HI", "Hindi");
+ isoCodeToEnWikiName.put("HU", "Hungarian");
+ isoCodeToEnWikiName.put("IS", "Icelandic");
+ isoCodeToEnWikiName.put("ID", "Indonesian");
+ isoCodeToEnWikiName.put("GA", "Irish");
+ isoCodeToEnWikiName.put("GD", "Gaelic");
+ isoCodeToEnWikiName.put("GV", "Manx");
+ isoCodeToEnWikiName.put("IT", "Italian");
+ isoCodeToEnWikiName.put("LA", "Latin");
+ isoCodeToEnWikiName.put("LV", "Latvian");
+ isoCodeToEnWikiName.put("LT", "Lithuanian");
+ isoCodeToEnWikiName.put("JA", "Japanese");
+ isoCodeToEnWikiName.put("KO", "Korean");
+ isoCodeToEnWikiName.put("KU", "Kurdish");
+ isoCodeToEnWikiName.put("LO", "Lao");
+ isoCodeToEnWikiName.put("ML", "Malayalam");
+ isoCodeToEnWikiName.put("MS", "Malay");
+ isoCodeToEnWikiName.put("MI", "Maori");
+ isoCodeToEnWikiName.put("MN", "Mongolian");
+ isoCodeToEnWikiName.put("NE", "Nepali");
+ isoCodeToEnWikiName.put("NO", "Norwegian");
+ isoCodeToEnWikiName.put("FA", "Persian");
+ isoCodeToEnWikiName.put("PL", "Polish");
+ isoCodeToEnWikiName.put("PT", "Portuguese");
+ isoCodeToEnWikiName.put("PA", "Punjabi");
+ isoCodeToEnWikiName.put("RO", "Romanian");
+ isoCodeToEnWikiName.put("RU", "Russian");
+ isoCodeToEnWikiName.put("SA", "Sanskrit");
+ isoCodeToEnWikiName.put("SK", "Slovak");
+ isoCodeToEnWikiName.put("SL", "Slovene|Slovenian");
+ isoCodeToEnWikiName.put("SO", "Somali");
+ isoCodeToEnWikiName.put("ES", "Spanish");
+ isoCodeToEnWikiName.put("SW", "Swahili");
+ isoCodeToEnWikiName.put("SV", "Swedish");
+ isoCodeToEnWikiName.put("TL", "Tagalog");
+ isoCodeToEnWikiName.put("TG", "Tajik");
+ isoCodeToEnWikiName.put("TA", "Tamil");
+ isoCodeToEnWikiName.put("TH", "Thai");
+ isoCodeToEnWikiName.put("BO", "Tibetan");
+ isoCodeToEnWikiName.put("TR", "Turkish");
+ isoCodeToEnWikiName.put("UK", "Ukrainian");
+ isoCodeToEnWikiName.put("UR", "Urdu");
+ isoCodeToEnWikiName.put("VI", "Vietnamese");
+ isoCodeToEnWikiName.put("CI", "Welsh");
+ isoCodeToEnWikiName.put("YI", "Yiddish");
+ isoCodeToEnWikiName.put("ZU", "Zulu");
+ isoCodeToEnWikiName.put("AZ", "Azeri");
+ isoCodeToEnWikiName.put("EU", "Basque");
+ isoCodeToEnWikiName.put("BR", "Breton");
+ isoCodeToEnWikiName.put("MR", "Marathi");
+ isoCodeToEnWikiName.put("FO", "Faroese");
+ isoCodeToEnWikiName.put("GL", "Galician");
+ isoCodeToEnWikiName.put("KA", "Georgian");
+ isoCodeToEnWikiName.put("HT", "Haitian Creole");
+ isoCodeToEnWikiName.put("LB", "Luxembourgish");
+ isoCodeToEnWikiName.put("MK", "Macedonian");
+ isoCodeToEnWikiName.put("GV", "Manx");
- isoCodeToWikiName.put("AZ", "Azeri");
- isoCodeToWikiName.put("EU", "Basque");
- isoCodeToWikiName.put("BR", "Breton");
- isoCodeToWikiName.put("MR", "Burmese");
- isoCodeToWikiName.put("FO", "Faroese");
- isoCodeToWikiName.put("GL", "Galician");
- isoCodeToWikiName.put("KA", "Georgian");
- isoCodeToWikiName.put("HT", "Haitian Creole");
- isoCodeToWikiName.put("LB", "Luxembourgish");
- isoCodeToWikiName.put("MK", "Macedonian");
+ // No longer exists in EN:
+ // isoCodeToEnWikiName.put("BS", "Bosnian");
+ // isoCodeToEnWikiName.put("SR", "Serbian");
+ // isoCodeToEnWikiName.put("HR", "Croatian");
+ // Font doesn't work:
+ //isoCodeToEnWikiName.put("MY", "Burmese");
+
+
+ {
+ //Set<String> missing = new LinkedHashSet<String>(isoCodeToEnWikiName.keySet());
+ //missing.removeAll(Language.isoCodeToResources.keySet());
+ //System.out.println(missing);
+ }
+ //assert Language.isoCodeToResources.keySet().containsAll(isoCodeToEnWikiName.keySet());
}
public static final Map<String,Map<String,String>> wikiCodeToIsoCodeToWikiName = new LinkedHashMap<String, Map<String,String>>();
static {
// en
- wikiCodeToIsoCodeToWikiName.put("en", isoCodeToWikiName);
+ wikiCodeToIsoCodeToWikiName.put("en", isoCodeToEnWikiName);
Map<String,String> isoCodeToWikiName;
isoCodeToWikiName.put("FR", "Französisch");
isoCodeToWikiName.put("EO", "Esperanto");
isoCodeToWikiName.put("CA", "Katalanisch");
- isoCodeToWikiName.put("LA", "Lateinisch");
+ isoCodeToWikiName.put("LA", "Latein");
isoCodeToWikiName.put("CS", "Tschechisch");
isoCodeToWikiName.put("HU", "Ungarisch");
isoCodeToWikiName.put("SV", "Schwedisch");
isoCodeToWikiName.put("ES", "Spanisch");
- // egrep -o '\{\{=[a-zA-Z]+=\}\}' frwiktionary-pages-articles.xml | sort | uniq -c | sort -nr
+ // egrep -o '== *\{\{langue\|[a-zA-Z]+\}\} *==' frwiktionary-pages-articles.xml | sort | uniq -c | sort -nr
isoCodeToWikiName = new LinkedHashMap<String, String>();
wikiCodeToIsoCodeToWikiName.put("fr", isoCodeToWikiName);
- isoCodeToWikiName.put("FR", Pattern.quote("{{=fr=}}"));
- isoCodeToWikiName.put("RU", Pattern.quote("{{=ru=}}"));
- isoCodeToWikiName.put("BG", Pattern.quote("{{=bg=}}")); // Bulgarian
- isoCodeToWikiName.put("EN", Pattern.quote("{{=en=}}"));
- //isoCodeToWikiName.put("", Pattern.quote("{{=sl=}}"));
- isoCodeToWikiName.put("LA", Pattern.quote("{{=la=}}"));
- isoCodeToWikiName.put("IT", Pattern.quote("{{=it=}}"));
- isoCodeToWikiName.put("EO", Pattern.quote("{{=eo=}}"));
- isoCodeToWikiName.put("CS", Pattern.quote("{{=cs=}}")); // Czech
- isoCodeToWikiName.put("NL", Pattern.quote("{{=nl=}}")); // Dutch
- //isoCodeToWikiName.put("", Pattern.quote("{{=mg=}}"));
- //isoCodeToWikiName.put("", Pattern.quote("{{=hsb=}}"));
- isoCodeToWikiName.put("ZH", Pattern.quote("{{=zh=}}"));
- isoCodeToWikiName.put("JA", Pattern.quote("{{=ja=}}"));
- isoCodeToWikiName.put("DE", Pattern.quote("{{=de=}}"));
- isoCodeToWikiName.put("IS", Pattern.quote("{{=is=}}")); // Icelandic
- isoCodeToWikiName.put("ES", Pattern.quote("{{=es=}}"));
- isoCodeToWikiName.put("UK", Pattern.quote("{{=uk=}}"));
+ isoCodeToWikiName.put("FR", Pattern.quote("{{langue|fr}}"));
+ isoCodeToWikiName.put("RU", Pattern.quote("{{langue|ru}}"));
+ isoCodeToWikiName.put("AR", Pattern.quote("{{langue|ar}}")); // Arabic
+ isoCodeToWikiName.put("BG", Pattern.quote("{{langue|bg}}")); // Bulgarian
+ isoCodeToWikiName.put("EN", Pattern.quote("{{langue|en}}"));
+ //isoCodeToWikiName.put("", Pattern.quote("{{langue|sl}}"));
+ isoCodeToWikiName.put("LA", Pattern.quote("{{langue|la}}"));
+ isoCodeToWikiName.put("IT", Pattern.quote("{{langue|it}}"));
+ isoCodeToWikiName.put("EO", Pattern.quote("{{langue|eo}}"));
+ isoCodeToWikiName.put("CS", Pattern.quote("{{langue|cs}}")); // Czech
+ isoCodeToWikiName.put("NL", Pattern.quote("{{langue|nl}}")); // Dutch
+ //isoCodeToWikiName.put("", Pattern.quote("{{langue|mg}}"));
+ //isoCodeToWikiName.put("", Pattern.quote("{{langue|hsb}}"));
+ isoCodeToWikiName.put("ZH", Pattern.quote("{{langue|zh}}"));
+ isoCodeToWikiName.put("cmn", Pattern.quote("{{langue|cmn}}"));
+ isoCodeToWikiName.put("yue", Pattern.quote("{{langue|yue}}"));
+ isoCodeToWikiName.put("JA", Pattern.quote("{{langue|ja}}"));
+ isoCodeToWikiName.put("DE", Pattern.quote("{{langue|de}}"));
+ isoCodeToWikiName.put("IS", Pattern.quote("{{langue|is}}")); // Icelandic
+ isoCodeToWikiName.put("ES", Pattern.quote("{{langue|es}}"));
+ isoCodeToWikiName.put("UK", Pattern.quote("{{langue|uk}}"));
// egrep -o '= *\{\{-[a-z]+-\}\} *=' itwiktionary-pages-articles.xml | sort | uniq -c | sort -n
isoCodeToWikiName = new LinkedHashMap<String, String>();
isoCodeToWikiName.put("PL", Pattern.quote("{{-pl-}}"));
isoCodeToWikiName.put("NL", Pattern.quote("{{-nl-}}"));
isoCodeToWikiName.put("LV", Pattern.quote("{{-lv-}}"));
- isoCodeToWikiName.put("LV", Pattern.quote("{{-la-}}"));
+ isoCodeToWikiName.put("LA", Pattern.quote("{{-la-}}"));
isoCodeToWikiName.put("HU", Pattern.quote("{{-hu-}}"));
- isoCodeToWikiName.put("PL", Pattern.quote("{{-pl-}}"));
isoCodeToWikiName.put("EL", Pattern.quote("{{-grc-}}"));
isoCodeToWikiName.put("SV", Pattern.quote("{{-sv-}}"));
+ isoCodeToWikiName.put("RU", Pattern.quote("{{-ru-}}"));
+ // There seems to be no consistent pattern and few foreign language entries anyway
+ isoCodeToWikiName = new LinkedHashMap<String, String>();
+ wikiCodeToIsoCodeToWikiName.put("es", isoCodeToWikiName);
+ isoCodeToWikiName.put("ES", Pattern.quote("{{ES"));
+ }
+ public static String getEnglishName(String langCode) {
+ String name = isoCodeToEnWikiName.get(langCode);
+ if (name == null) {
+ name = isoCodeToEnWikiName.get(langCode.toUpperCase());
+ }
+ if (name == null) {
+ return null;
+ }
+ if (name.indexOf('|') != -1) {
+ return name.substring(0, name.indexOf('|'));
+ }
+ if (name.indexOf('$') != -1) {
+ return name.substring(0, name.indexOf('$'));
+ }
+ return name; // can be null.
}
}
projects / DictionaryPC.git / blobdiff