X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fengine%2FWiktionarySplitter.java;h=628d3567e7a1ad030a5e828599e8458088501db0;hb=1aa4de25c859304d21acfadd18cb546d1c21415b;hp=f5c85cf0f4f1e8cb41d500892f2bf52f64781c07;hpb=7819736ae570bf597936f0dc640f60644da15fc8;p=DictionaryPC.git diff --git a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java index f5c85cf..628d356 100644 --- a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java +++ b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java @@ -20,8 +20,8 @@ import java.io.File; import java.io.FileOutputStream; import java.io.IOException; import java.util.ArrayList; -import java.util.Arrays; import java.util.List; +import java.util.Map; import java.util.regex.Matcher; import java.util.regex.Pattern; @@ -32,8 +32,12 @@ import javax.xml.parsers.SAXParserFactory; import org.xml.sax.Attributes; import org.xml.sax.SAXException; +import com.hughes.android.dictionary.parser.enwiktionary.EnWiktionaryLangs; + public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler { + private static final String FILE_TO_SPLIT = "data/inputs/enwiktionary-pages-articles.xml"; + static class Section implements java.io.Serializable { private static final long serialVersionUID = -7676549898325856822L; @@ -70,7 +74,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler { final WiktionarySplitter wiktionarySplitter = new WiktionarySplitter(); // Configure things. - final File file = new File(args[0]); + final List selectors = wiktionarySplitter.selectors; for (int i = 1; i < args.length; i += 2) { final Selector selector = new Selector(args[i], args[i+1]); @@ -78,58 +82,13 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler { } if (selectors.isEmpty()) { - selectors.addAll(Arrays.asList( -// new Selector("../DictionaryData/inputs/enWikiSplit/AF.data", ".*[Aa]frikaans.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/AR.data", ".*[Aa]rabic.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/HY.data", ".*[Aa]rmenian.*"), - new Selector("../DictionaryData/inputs/enWikiSplit/HR.data", ".*[Cc]roatian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/CS.data", ".*[Cc]zech.*"), - new Selector("../DictionaryData/inputs/enWikiSplit/ZH.data", ".*[Cc]hinese.*|.*[Mm]andarin.*|.*Cantonese.*") -// new Selector("../DictionaryData/inputs/enWikiSplit/DA.data", ".*[Dd]anish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/NL.data", ".*[Dd]utch.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/EN.data", ".*[Ee]nglish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/FI.data", ".*[Ff]innish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/FR.data", ".*[Ff]rench.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/DE.data", ".*[Gg]erman.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/EL.data", ".*[Gg]reek.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/haw.data", ".*[Hh]awaiian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/HE.data", ".*[Hh]ebrew.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/HI.data", ".*[Hh]indi.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/IS.data", ".*[Ii]celandic.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/GA.data", ".*[Ii]rish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/IT.data", ".*[Ii]talian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/LT.data", ".*[Ll]ithuanian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/JA.data", ".*[Jj]apanese.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/KO.data", ".*[Kk]orean.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/KU.data", ".*[Kk]urdish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/MS.data", ".*[Mm]alay.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/MI.data", ".*[Mm]aori.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/MN.data", ".*[Mm]ongolian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/NO.data", ".*[Nn]orwegian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/FA.data", ".*[Pp]ersian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/PT.data", ".*[Pp]ortuguese.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/PL.data", ".*[Pp]olish.*") -// new Selector("../DictionaryData/inputs/enWikiSplit/RO.data", ".*[Rr]omanian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/RU.data", ".*[Rr]ussian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/SA.data", ".*[Ss]anskrit.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/SR.data", ".*[Ss]erbian.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/SO.data", ".*[Ss]omali.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/ES.data", ".*[Ss]panish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/SV.data", ".*[Ss]wedish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/TG.data", ".*[Tt]ajik.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/TH.data", ".*[Tt]hai.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/BO.data", ".*[Tt]ibetan.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/TR.data", ".*[Tt]urkish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/UK.data", ".*[Uu]krainian.*") -// new Selector("../DictionaryData/inputs/enWikiSplit/VI.data", ".*[Vv]ietnamese.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/CI.data", ".*[Ww]elsh.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/YI.data", ".*[Yy]iddish.*"), -// new Selector("../DictionaryData/inputs/enWikiSplit/ZU.data", ".*[Zz]ulu.*") - )); + for (final Map.Entry entry : EnWiktionaryLangs.isoCodeToWikiName.entrySet()) { + selectors.add(new Selector(String.format("data/inputs/enWikiSplit/%s.data", entry.getKey()), entry.getValue())); + } } // Do it. - parser.parse(file, wiktionarySplitter); + parser.parse(new File(FILE_TO_SPLIT), wiktionarySplitter); // Shutdown. for (final Selector selector : selectors) {