org.apache.lucene.analysis.ru.RussianAnalyzer Java Examples

@Inject
public RussianAnalyzerProvider(Index index, IndexSettingsService indexSettingsService, Environment env, @Assisted String name, @Assisted Settings settings) {
    super(index, indexSettingsService.getSettings(), name, settings);
    analyzer = new RussianAnalyzer(Analysis.parseStopWords(env, settings, RussianAnalyzer.getDefaultStopSet()),
                                   Analysis.parseStemExclusion(settings, CharArraySet.EMPTY_SET));
    analyzer.setVersion(version);
}
 

RussianAnalyzerProvider(IndexSettings indexSettings, Environment env, String name, Settings settings) {
    super(indexSettings, name, settings);
    analyzer = new RussianAnalyzer(
        Analysis.parseStopWords(env, settings, RussianAnalyzer.getDefaultStopSet()),
        Analysis.parseStemExclusion(settings, CharArraySet.EMPTY_SET)
    );
    analyzer.setVersion(version);
}
 

@Override
public RussianAnalyzer get() {
    return this.analyzer;
}
 

private static void initAnalysis() {
    LANGUAGE_SPECIFIC.put("ru", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new RussianAnalyzer();
        }
    });
    LANGUAGE_SPECIFIC.put("uk", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new RussianAnalyzer();
        }
    });
    LANGUAGE_SPECIFIC.put("de", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new GermanAnalyzer();
        }
    });
    LANGUAGE_SPECIFIC.put("fr", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new FrenchAnalyzer();
        }
    });
    LANGUAGE_SPECIFIC.put("it", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new ItalianAnalyzer();
        }
    });
    LANGUAGE_SPECIFIC.put("en", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new StandardAnalyzer(EnglishAnalyzer.ENGLISH_STOP_WORDS_SET);
        }
    });
    LANGUAGE_SPECIFIC.put("default", new Analysis() {
        @Override
        protected Analyzer initialValue() {
            return new StandardAnalyzer(EnglishAnalyzer.ENGLISH_STOP_WORDS_SET);
        }
    });
}
 

/**
 * The third and final phase of the life-cyle used for identifying
 * the AnalyzerPair.
 */
private AnalyzerPair construct() throws Exception {
	if (className == null) {
		return null;
	}
	if (pattern != null) {
		return new PatternAnalyzer(this, pattern);
	}
	if (softHyphens != null) {
		return new AnalyzerPair(
				languageRange,
				new TermCompletionAnalyzer(
						wordBoundary, 
						subWordBoundary, 
						softHyphens, 
						alwaysRemoveSoftHyphens));
	}
	if (wordBoundary != null) {
		return new AnalyzerPair(
				languageRange,
				new TermCompletionAnalyzer(
						wordBoundary, 
						subWordBoundary));
	}
	final Class<? extends Analyzer> cls = getAnalyzerClass();
          
          if (hasConstructor(cls, CharArraySet.class)) {
  			return new AnalyzerPair(languageRange,
  					cls.getConstructor(CharArraySet.class).newInstance(new CharArraySet(getStopWords(), false)),
  					cls.getConstructor(CharArraySet.class).newInstance(new CharArraySet(Collections.emptySet(), false)));
          }
          
          if (hasConstructor(cls, Version.class, Set.class)) {

          	// RussianAnalyzer is missing any way to access stop words.
          	if (RussianAnalyzer.class.equals(cls)) {
          		if (useDefaultStopWords()) {
          		    return new AnalyzerPair(languageRange, new RussianAnalyzer(), new RussianAnalyzer(CharArraySet.EMPTY_SET));
          		}
          		if (doNotUseStopWords()) {
          		    return new AnalyzerPair(languageRange,  new RussianAnalyzer(CharArraySet.EMPTY_SET));	
          		}
          	}
          	return new VersionSetAnalyzerPair(this, cls);
          }
          
          if (stopwords != null && !stopwords.equals(AnalyzerOptions.STOPWORDS_VALUE_NONE)) {
          	throw new RuntimeException("Bad option: language range: " + languageRange + " stopwords are not supported by " + className);
          }
          if (hasConstructor(cls, Version.class)) {
          	return new VersionAnalyzerPair(languageRange, cls);
          }
          
          if (hasConstructor(cls)) {
          	return new AnalyzerPair(languageRange, cls.newInstance());
          }
          throw new RuntimeException("Bad option: cannot find constructor for class " + className + " for language range " + languageRange);
}
 

/**
 * Returns the default stopwords set used by Lucene language analyzer for the specified language.
 *
 * @param language The language for which the stopwords are. The supported languages are English, French, Spanish,
 *                 Portuguese, Italian, Romanian, German, Dutch, Swedish, Norwegian, Danish, Russian, Finnish,
 *                 Irish, Hungarian, Turkish, Armenian, Basque and Catalan.
 * @return The default stopwords set used by Lucene language analyzers.
 */
private static CharArraySet getDefaultStopwords(String language) {
    switch (language) {
        case "English":
            return EnglishAnalyzer.getDefaultStopSet();
        case "French":
            return FrenchAnalyzer.getDefaultStopSet();
        case "Spanish":
            return SpanishAnalyzer.getDefaultStopSet();
        case "Portuguese":
            return PortugueseAnalyzer.getDefaultStopSet();
        case "Italian":
            return ItalianAnalyzer.getDefaultStopSet();
        case "Romanian":
            return RomanianAnalyzer.getDefaultStopSet();
        case "German":
            return GermanAnalyzer.getDefaultStopSet();
        case "Dutch":
            return DutchAnalyzer.getDefaultStopSet();
        case "Swedish":
            return SwedishAnalyzer.getDefaultStopSet();
        case "Norwegian":
            return NorwegianAnalyzer.getDefaultStopSet();
        case "Danish":
            return DanishAnalyzer.getDefaultStopSet();
        case "Russian":
            return RussianAnalyzer.getDefaultStopSet();
        case "Finnish":
            return FinnishAnalyzer.getDefaultStopSet();
        case "Irish":
            return IrishAnalyzer.getDefaultStopSet();
        case "Hungarian":
            return HungarianAnalyzer.getDefaultStopSet();
        case "Turkish":
            return SpanishAnalyzer.getDefaultStopSet();
        case "Armenian":
            return SpanishAnalyzer.getDefaultStopSet();
        case "Basque":
            return BasqueAnalyzer.getDefaultStopSet();
        case "Catalan":
            return CatalanAnalyzer.getDefaultStopSet();
        default:
            return CharArraySet.EMPTY_SET;
    }
}
 

@Override
public List<PreBuiltAnalyzerProviderFactory> getPreBuiltAnalyzerProviderFactories() {
    List<PreBuiltAnalyzerProviderFactory> analyzers = new ArrayList<>();
    analyzers.add(new PreBuiltAnalyzerProviderFactory("standard_html_strip", CachingStrategy.ELASTICSEARCH,
        () -> new StandardHtmlStripAnalyzer(CharArraySet.EMPTY_SET)));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("pattern", CachingStrategy.ELASTICSEARCH,
        () -> new PatternAnalyzer(Regex.compile("\\W+" /*PatternAnalyzer.NON_WORD_PATTERN*/, null), true,
        CharArraySet.EMPTY_SET)));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("snowball", CachingStrategy.LUCENE,
        () -> new SnowballAnalyzer("English", EnglishAnalyzer.ENGLISH_STOP_WORDS_SET)));

    // Language analyzers:
    analyzers.add(new PreBuiltAnalyzerProviderFactory("arabic", CachingStrategy.LUCENE, ArabicAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("armenian", CachingStrategy.LUCENE, ArmenianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("basque", CachingStrategy.LUCENE, BasqueAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("bengali", CachingStrategy.LUCENE, BengaliAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("brazilian", CachingStrategy.LUCENE, BrazilianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("bulgarian", CachingStrategy.LUCENE, BulgarianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("catalan", CachingStrategy.LUCENE, CatalanAnalyzer::new));
    // chinese analyzer: only for old indices, best effort
    analyzers.add(new PreBuiltAnalyzerProviderFactory("chinese", CachingStrategy.ONE, StandardAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("cjk", CachingStrategy.LUCENE, CJKAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("czech", CachingStrategy.LUCENE, CzechAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("danish", CachingStrategy.LUCENE, DanishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("dutch", CachingStrategy.LUCENE, DutchAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("english", CachingStrategy.LUCENE, EnglishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("finnish", CachingStrategy.LUCENE, FinnishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("french", CachingStrategy.LUCENE, FrenchAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("galician", CachingStrategy.LUCENE, GalicianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("german", CachingStrategy.LUCENE, GermanAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("greek", CachingStrategy.LUCENE, GreekAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("hindi", CachingStrategy.LUCENE, HindiAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("hungarian", CachingStrategy.LUCENE, HungarianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("indonesian", CachingStrategy.LUCENE, IndonesianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("irish", CachingStrategy.LUCENE, IrishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("italian", CachingStrategy.LUCENE, ItalianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("latvian", CachingStrategy.LUCENE, LatvianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("lithuanian", CachingStrategy.LUCENE, LithuanianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("norwegian", CachingStrategy.LUCENE, NorwegianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("persian", CachingStrategy.LUCENE, PersianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("portuguese", CachingStrategy.LUCENE, PortugueseAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("romanian", CachingStrategy.LUCENE, RomanianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("russian", CachingStrategy.LUCENE, RussianAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("sorani", CachingStrategy.LUCENE, SoraniAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("spanish", CachingStrategy.LUCENE, SpanishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("swedish", CachingStrategy.LUCENE, SwedishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("turkish", CachingStrategy.LUCENE, TurkishAnalyzer::new));
    analyzers.add(new PreBuiltAnalyzerProviderFactory("thai", CachingStrategy.LUCENE, ThaiAnalyzer::new));
    return analyzers;
}
 

@Override
public RussianAnalyzer get() {
    return this.analyzer;
}