/**
   * Analyzes the given value using the given Analyzer.
   *
   * @param value Value to analyze
   * @param context The {@link AnalysisContext analysis context}.
   * @return NamedList containing the tokens produced by analyzing the given value
   */
  protected NamedList<? extends Object> analyzeValue(String value, AnalysisContext context) {

    Analyzer analyzer = context.getAnalyzer();

    if (!TokenizerChain.class.isInstance(analyzer)) {

      TokenStream tokenStream = null;
      try {
        tokenStream = analyzer.reusableTokenStream(context.getFieldName(), new StringReader(value));
        tokenStream.reset();
      } catch (IOException e) {
        throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, e);
      }
      NamedList<List<NamedList>> namedList = new NamedList<List<NamedList>>();
      namedList.add(
          tokenStream.getClass().getName(),
          convertTokensToNamedLists(analyzeTokenStream(tokenStream), context));
      return namedList;
    }

    TokenizerChain tokenizerChain = (TokenizerChain) analyzer;
    CharFilterFactory[] cfiltfacs = tokenizerChain.getCharFilterFactories();
    TokenizerFactory tfac = tokenizerChain.getTokenizerFactory();
    TokenFilterFactory[] filtfacs = tokenizerChain.getTokenFilterFactories();

    NamedList<Object> namedList = new NamedList<Object>();

    if (cfiltfacs != null) {
      String source = value;
      for (CharFilterFactory cfiltfac : cfiltfacs) {
        CharStream reader = CharReader.get(new StringReader(source));
        reader = cfiltfac.create(reader);
        source = writeCharStream(namedList, reader);
      }
    }

    TokenStream tokenStream = tfac.create(tokenizerChain.charStream(new StringReader(value)));
    List<AttributeSource> tokens = analyzeTokenStream(tokenStream);

    namedList.add(tokenStream.getClass().getName(), convertTokensToNamedLists(tokens, context));

    ListBasedTokenStream listBasedTokenStream = new ListBasedTokenStream(tokens);

    for (TokenFilterFactory tokenFilterFactory : filtfacs) {
      tokenStream = tokenFilterFactory.create(listBasedTokenStream);
      List<AttributeSource> tokenList = analyzeTokenStream(tokenStream);
      namedList.add(
          tokenStream.getClass().getName(), convertTokensToNamedLists(tokenList, context));
      listBasedTokenStream = new ListBasedTokenStream(tokenList);
    }

    return namedList;
  }
Пример #2
0
  private static SimpleOrderedMap<Object> getAnalyzerInfo(Analyzer analyzer) {
    SimpleOrderedMap<Object> aninfo = new SimpleOrderedMap<Object>();
    aninfo.add("className", analyzer.getClass().getName());
    if (analyzer instanceof TokenizerChain) {

      TokenizerChain tchain = (TokenizerChain) analyzer;

      CharFilterFactory[] cfiltfacs = tchain.getCharFilterFactories();
      SimpleOrderedMap<Map<String, Object>> cfilters = new SimpleOrderedMap<Map<String, Object>>();
      for (CharFilterFactory cfiltfac : cfiltfacs) {
        Map<String, Object> tok = new HashMap<String, Object>();
        String className = cfiltfac.getClass().getName();
        tok.put("className", className);
        tok.put("args", cfiltfac.getArgs());
        cfilters.add(className.substring(className.lastIndexOf('.') + 1), tok);
      }
      if (cfilters.size() > 0) {
        aninfo.add("charFilters", cfilters);
      }

      SimpleOrderedMap<Object> tokenizer = new SimpleOrderedMap<Object>();
      TokenizerFactory tfac = tchain.getTokenizerFactory();
      tokenizer.add("className", tfac.getClass().getName());
      tokenizer.add("args", tfac.getArgs());
      aninfo.add("tokenizer", tokenizer);

      TokenFilterFactory[] filtfacs = tchain.getTokenFilterFactories();
      SimpleOrderedMap<Map<String, Object>> filters = new SimpleOrderedMap<Map<String, Object>>();
      for (TokenFilterFactory filtfac : filtfacs) {
        Map<String, Object> tok = new HashMap<String, Object>();
        String className = filtfac.getClass().getName();
        tok.put("className", className);
        tok.put("args", filtfac.getArgs());
        filters.add(className.substring(className.lastIndexOf('.') + 1), tok);
      }
      if (filters.size() > 0) {
        aninfo.add("filters", filters);
      }
    }
    return aninfo;
  }