From: Reimar Döffinger Date: Sat, 11 Feb 2017 16:32:56 +0000 (+0100) Subject: Support pt and es wiktionary in splitter. X-Git-Url: http://gitweb.fperrin.net/?p=DictionaryPC.git;a=commitdiff_plain;h=761f748ec034b093581535f7a4befa8e6e3c6873 Support pt and es wiktionary in splitter. The ES format seems to have changed so we can now actually use it. --- diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java index 3efc676..8049257 100644 --- a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java +++ b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java @@ -193,10 +193,20 @@ 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 + // 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("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);