Esempio n. 1
0
  @Override
  public boolean incrementToken() throws IOException {

    while (input.incrementToken()) {
      char text[] = termAtt.termBuffer();
      int termLength = termAtt.termLength();
      if (!stopTable.contains(text, 0, termLength)) {
        return true;
      }
    }
    return false;
  }
  public final boolean incrementToken() throws IOException {
    int increment = 0;
    while (input.incrementToken()) {
      if (!stopWords.contains(termAttr.termBuffer(), 0, termAttr.termLength())) {
        posIncrAttr.setPositionIncrement(posIncrAttr.getPositionIncrement() + increment);
        return true;
      }

      increment += posIncrAttr.getPositionIncrement();
    }

    return false;
  }
  @Override
  public boolean incrementToken() throws IOException {
    if (!input.incrementToken()) return false;

    char[] termBuffer = termAtt.termBuffer();
    int termBufferLength = termAtt.termLength();
    char[] backup = null;
    if (factory.maxWordCount < CapitalizationFilterFactory.DEFAULT_MAX_WORD_COUNT) {
      // make a backup in case we exceed the word count
      backup = new char[termBufferLength];
      System.arraycopy(termBuffer, 0, backup, 0, termBufferLength);
    }
    if (termBufferLength < factory.maxTokenLength) {
      int wordCount = 0;

      int lastWordStart = 0;
      for (int i = 0; i < termBufferLength; i++) {
        char c = termBuffer[i];
        if (c <= ' ' || c == '.') {
          int len = i - lastWordStart;
          if (len > 0) {
            factory.processWord(termBuffer, lastWordStart, len, wordCount++);
            lastWordStart = i + 1;
            i++;
          }
        }
      }

      // process the last word
      if (lastWordStart < termBufferLength) {
        factory.processWord(
            termBuffer, lastWordStart, termBufferLength - lastWordStart, wordCount++);
      }

      if (wordCount > factory.maxWordCount) {
        termAtt.setTermBuffer(backup, 0, termBufferLength);
      }
    }

    return true;
  }