X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fparser%2Fwiktionary%2FWiktionaryLangs.java;h=01b31a23861e7f4c093fa4d101424503a22c8736;hb=2fc669d88306d563fc9c899d8d91b25d591692ea;hp=9dfa00a3a0932caf2b843bcccfa597017c1def55;hpb=e479ba38bbcb261951399326623c20ffacc147d4;p=DictionaryPC.git

diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
index 9dfa00a..01b31a2 100644
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
@@ -14,17 +14,13 @@
 
 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> isoCodeToEnWikiName = new LinkedHashMap<String,String>();
+    public static final Map<String,String> isoCodeToEnWikiName = new LinkedHashMap<>();
     static {
         isoCodeToEnWikiName.put("AF", "Afrikaans");
         isoCodeToEnWikiName.put("SQ", "Albanian");
@@ -46,6 +42,10 @@ public class WiktionaryLangs {
         isoCodeToEnWikiName.put("ET", "Estonian");
         isoCodeToEnWikiName.put("FI", "Finnish");
         isoCodeToEnWikiName.put("FR", "French");
+        // Note: must be before German since matcher
+        // simply takes first match instead of best.
+        isoCodeToEnWikiName.put("nds", "Low German");
+        isoCodeToEnWikiName.put("pdc", "Pennsylvania German");
         isoCodeToEnWikiName.put("DE", "German");
         isoCodeToEnWikiName.put("grc", "Ancient Greek");
         isoCodeToEnWikiName.put("EL", "Greek");
@@ -107,7 +107,9 @@ public class WiktionaryLangs {
         isoCodeToEnWikiName.put("HT", "Haitian Creole");
         isoCodeToEnWikiName.put("LB", "Luxembourgish");
         isoCodeToEnWikiName.put("MK", "Macedonian");
-        isoCodeToEnWikiName.put("GV", "Manx");
+        isoCodeToEnWikiName.put("scn", "Sicilian");
+        isoCodeToEnWikiName.put("cu", "Old Church Slavonic");
+        isoCodeToEnWikiName.put("rom", "Romani");
 
         // No longer exists in EN:
         // isoCodeToEnWikiName.put("BS", "Bosnian");
@@ -126,16 +128,17 @@ public class WiktionaryLangs {
         //assert Language.isoCodeToResources.keySet().containsAll(isoCodeToEnWikiName.keySet());
     }
 
-    public static final Map<String,Map<String,String>> wikiCodeToIsoCodeToWikiName = new LinkedHashMap<String, Map<String,String>>();
+    public static final Map<String,Map<String,String>> wikiCodeToIsoCodeToWikiName = new LinkedHashMap<>();
     static {
+        Map<String,String> isoCodeToWikiName;
+
         // en
         wikiCodeToIsoCodeToWikiName.put("en", isoCodeToEnWikiName);
 
-        Map<String,String> isoCodeToWikiName;
-
         // egrep -o '\{\{Wortart[^}]+\}\}' dewiktionary-pages-articles.xml | cut -d \| -f3 | sort | uniq -c | sort -nr
-        isoCodeToWikiName = new LinkedHashMap<String, String>();
+        isoCodeToWikiName = new LinkedHashMap<>();
         wikiCodeToIsoCodeToWikiName.put("de", isoCodeToWikiName);
+        isoCodeToWikiName.put("nds", "Niederdeutsch");
         isoCodeToWikiName.put("DE", "Deutsch");
         isoCodeToWikiName.put("EN", "Englisch");
         isoCodeToWikiName.put("IT", "Italienisch");
@@ -148,14 +151,16 @@ public class WiktionaryLangs {
         isoCodeToWikiName.put("HU", "Ungarisch");
         isoCodeToWikiName.put("SV", "Schwedisch");
         isoCodeToWikiName.put("ES", "Spanisch");
+        isoCodeToWikiName.put("RO", "Rumänisch");
 
         // egrep -o '== *\{\{langue\|[a-zA-Z]+\}\} *==' frwiktionary-pages-articles.xml | sort | uniq -c | sort -nr
-        isoCodeToWikiName = new LinkedHashMap<String, String>();
+        isoCodeToWikiName = new LinkedHashMap<>();
         wikiCodeToIsoCodeToWikiName.put("fr", isoCodeToWikiName);
         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("EL", Pattern.quote("{{langue|el}}"));
         isoCodeToWikiName.put("EN", Pattern.quote("{{langue|en}}"));
         //isoCodeToWikiName.put("", Pattern.quote("{{langue|sl}}"));
         isoCodeToWikiName.put("LA", Pattern.quote("{{langue|la}}"));
@@ -173,9 +178,10 @@ public class WiktionaryLangs {
         isoCodeToWikiName.put("IS", Pattern.quote("{{langue|is}}"));  // Icelandic
         isoCodeToWikiName.put("ES", Pattern.quote("{{langue|es}}"));
         isoCodeToWikiName.put("UK", Pattern.quote("{{langue|uk}}"));
+        isoCodeToWikiName.put("PT", Pattern.quote("{{langue|pt}}"));
 
         // egrep -o '= *\{\{-[a-z]+-\}\} *=' itwiktionary-pages-articles.xml | sort | uniq -c | sort -n
-        isoCodeToWikiName = new LinkedHashMap<String, String>();
+        isoCodeToWikiName = new LinkedHashMap<>();
         wikiCodeToIsoCodeToWikiName.put("it", isoCodeToWikiName);
         isoCodeToWikiName.put("IT", "\\{\\{-(it|scn|nap|cal|lmo)-\\}\\}");  // scn, nap, cal, lmo
         isoCodeToWikiName.put("EN", Pattern.quote("{{-en-}}"));
@@ -192,10 +198,21 @@ public class WiktionaryLangs {
         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>();
+        // egrep -o '== *\{\{lengua\|[a-zA-Z]+\}\} *==' frwiktionary-pages-articles.xml | sort | uniq -c | sort -nr
+        isoCodeToWikiName = new LinkedHashMap<>();
         wikiCodeToIsoCodeToWikiName.put("es", isoCodeToWikiName);
-        isoCodeToWikiName.put("ES", Pattern.quote("{{ES"));
+        isoCodeToWikiName.put("AR", Pattern.quote("{{lengua|ar}}"));
+        isoCodeToWikiName.put("ES", Pattern.quote("{{lengua|es}}"));
+        isoCodeToWikiName.put("EN", Pattern.quote("{{lengua|en}}"));
+        isoCodeToWikiName.put("FR", Pattern.quote("{{lengua|fr}}"));
+        isoCodeToWikiName.put("IT", Pattern.quote("{{lengua|it}}"));
+
+        // Pattern seems to match Italian one
+        isoCodeToWikiName = new LinkedHashMap<>();
+        wikiCodeToIsoCodeToWikiName.put("pt", isoCodeToWikiName);
+        isoCodeToWikiName.put("PT", Pattern.quote("{{-pt-}}"));
+        isoCodeToWikiName.put("EN", Pattern.quote("{{-en-}}"));
+        isoCodeToWikiName.put("ES", Pattern.quote("{{-es-}}"));
     }
     public static String getEnglishName(String langCode) {
         String name = isoCodeToEnWikiName.get(langCode);