X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fengine%2FDictionaryBuilder.java;h=10c52a5c9a0824a23e24126c691de284357a3e79;hb=7d5ada9329d101b59b55691dd2f63ce3e3860011;hp=502da0b5a5c13ea3189230852b43c34d8933873a;hpb=1aa4de25c859304d21acfadd18cb546d1c21415b;p=DictionaryPC.git diff --git a/src/com/hughes/android/dictionary/engine/DictionaryBuilder.java b/src/com/hughes/android/dictionary/engine/DictionaryBuilder.java index 502da0b..10c52a5 100644 --- a/src/com/hughes/android/dictionary/engine/DictionaryBuilder.java +++ b/src/com/hughes/android/dictionary/engine/DictionaryBuilder.java @@ -20,7 +20,6 @@ import java.io.PrintStream; import java.io.RandomAccessFile; import java.nio.charset.Charset; import java.util.ArrayList; -import java.util.Arrays; import java.util.LinkedHashSet; import java.util.List; import java.util.Map; @@ -32,18 +31,19 @@ import javax.xml.parsers.ParserConfigurationException; import org.xml.sax.SAXException; import com.hughes.android.dictionary.parser.DictFileParser; -import com.hughes.android.dictionary.parser.enwiktionary.EnWiktionaryXmlParser; +import com.hughes.android.dictionary.parser.Parser; +import com.hughes.android.dictionary.parser.wiktionary.EnForeignParser; +import com.hughes.android.dictionary.parser.wiktionary.EnToTranslationParser; import com.hughes.util.Args; import com.hughes.util.FileUtil; - public class DictionaryBuilder { public final Dictionary dictionary; public final List indexBuilders = new ArrayList(); - public DictionaryBuilder(final String dictInfo, final Language lang0, final Language lang1, final String normalizerRules1, final String normalizerRules2, final Set lang1Stoplist, final Set lang2Stoplist) { - dictionary = new Dictionary(dictInfo); + public DictionaryBuilder(final String dictInfoString, final Language lang0, final Language lang1, final String normalizerRules1, final String normalizerRules2, final Set lang1Stoplist, final Set lang2Stoplist) { + dictionary = new Dictionary(dictInfoString); indexBuilders.add(new IndexBuilder(this, lang0.getIsoCode(), lang0.getIsoCode() + "->" + lang1.getIsoCode(), lang0, normalizerRules1, lang1Stoplist, false)); indexBuilders.add(new IndexBuilder(this, lang1.getIsoCode(), lang1.getIsoCode() + "->" + lang0.getIsoCode(), lang1, normalizerRules2, lang2Stoplist, true)); } @@ -127,31 +127,43 @@ public class DictionaryBuilder { if (inputName == null) { fatalError("Must specify human readable name for: " + prefix + "Name"); } + String pageLimitString = keyValueArgs.remove(prefix + "PageLimit"); + if (pageLimitString == null) { + pageLimitString = "-1"; + } + final int pageLimit = Integer.parseInt(pageLimitString); - final EntrySource entrySource = new EntrySource(dictionaryBuilder.dictionary.sources.size(), inputName, dictionaryBuilder.dictionary.pairEntries.size()); + final EntrySource entrySource = new EntrySource(dictionaryBuilder.dictionary.sources.size(), inputName, 0); System.out.println(""); String inputFormat = keyValueArgs.remove(prefix + "Format"); if ("tab_separated".equals(inputFormat)) { final boolean flipColumns = "true".equals(keyValueArgs.remove(prefix + "FlipColumns")); - new DictFileParser(charset, flipColumns, DictFileParser.TAB, null, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file); + new DictFileParser(charset, flipColumns, DictFileParser.TAB, null, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parse(file, entrySource, pageLimit); } else if ("chemnitz".equals(inputFormat)) { final boolean flipColumns = "true".equals(keyValueArgs.remove(prefix + "FlipColumns")); - new DictFileParser(charset, flipColumns, DictFileParser.DOUBLE_COLON, DictFileParser.PIPE, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file); + new DictFileParser(charset, flipColumns, DictFileParser.DOUBLE_COLON, DictFileParser.PIPE, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parse(file, entrySource, pageLimit); } else if ("enwiktionary".equals(inputFormat)) { + final String type = keyValueArgs.remove(prefix + "WiktionaryType"); final Pattern langPattern = Pattern.compile(keyValueArgs.remove(prefix + "LangPattern"), Pattern.CASE_INSENSITIVE); final Pattern langCodePattern = Pattern.compile(keyValueArgs.remove(prefix + "LangCodePattern")); final int enIndex = Integer.parseInt(keyValueArgs.remove(prefix + "EnIndex")) - 1; - String pageLimit = keyValueArgs.remove(prefix + "PageLimit"); - if (pageLimit == null) { - pageLimit = "-1"; - } if (enIndex < 0 || enIndex >= 2) { fatalError("Must be 1 or 2: " + prefix + "EnIndex"); } - new EnWiktionaryXmlParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(1-enIndex), - langPattern, langCodePattern, enIndex != 0).parse(file, Integer.parseInt(pageLimit)); + final Parser parser; + if ("EnToTranslation".equals(type)) { + parser = new EnToTranslationParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(1-enIndex), + langPattern, langCodePattern, enIndex != 0); + } else if ("EnForeign".equals(type)) { + parser = new EnForeignParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(1-enIndex), + langPattern, langCodePattern, enIndex != 0); + } else { + fatalError("Invalid WiktionaryType (use EnToTranslation or EnForeign): " + type); + return; + } + parser.parse(file, entrySource, pageLimit); } else { fatalError("Invalid or missing input format: " + inputFormat); }