fileparser.java

一个决策树的Applet（转载

   * <p>
   * This method exists to keep the file format opaque to parent objects.
   *
   * @throws InvalidMetaFileException If a syntax error is found.
   */
  public String extractString()
    throws InvalidMetaFileException
  {
    return extractString( '\"', '\"', true );
  }

  /**
   * Extracts a sample from the data file.  The method
   * expects that the data file has one sample per line,
   * which consists of a comma-delimited list of
   * attributes.  A line may or not be terminated by an
   * additional comma.
   *
   * @return A Vector of Strings from the comma
   *         delimited sample, in the order they appear
   *         on the line in the file, or null if the
   *         end of the file has been reached.
   *
   * @throws InvalidDataFileException If a syntax error
   *         is found.
   *
   * @throws IOException If an IO error occurs while
   *         reading a line from the file.
   */
  public Vector extractDataSample()
    throws InvalidDataFileException,
           IOException
  {
    do {
      m_fileText = m_lineReader.readLine();

      if( m_fileText == null ) return null;
    }
    while( isEmptyLine( m_fileText ) );

    int pos;
    Vector attribVals = new Vector();

    // Parse the line.
    while( (pos = m_fileText.indexOf( ',' )) > -1 ) {
      if( pos == 0 )
        throw new
          InvalidDataFileException( "Syntax error " +
            "in data file (line " + m_lineReader.getLineNumber() + "): " +
            "Null attribute before comma." );

      attribVals.add( m_fileText.substring( 0, pos ).trim() );

      // Strip off the value we just grabbed, and
      // the trailing comma.
      if( (pos + 1) == m_fileText.length() )
        m_fileText = new String();
      else
        m_fileText = m_fileText.substring( pos + 1 );
    }

    // There may or may not be something left on the
    // line after parsing the last comma - if there
    // is, add it to our values vector.
    if( !isEmptyLine( m_fileText ) )
      attribVals.add( m_fileText.trim() );

    return attribVals;
  }

  /**
   * Returns the line number for the line that was
   * last read from the current file.
   *
   * <p>
   * <b>Note:</b> This method is only useful for
   * data files.
   *
   * @return the current line number, or -1 if no file
   *         is open.
   */
  public int getCurrentLineNum()
  {
    if( m_lineReader == null ) return -1;
    return m_lineReader.getLineNumber();
  }

  // Private methods

  /**
   * Extracts and returns a string from the current
   * file text, using the supplied characters as delimiters
   * (starting from the beginning of the text).  The string
   * and the delimiters are removed from the file text.
   *
   * @param startDelim Delimiter character that marks the
   *        start of the string.
   *
   * @param stopDelim  Delimiter character that marks the
   *        end of the string.
   *
   * @param trim Determines if whitespace is trimmed from
   *        the beginning and the end of the extracted string.
   *        Set to true to remove whitespace, or false
   *        otherwise.
   *
   * @throws InvalidMetaFileException If there
   *         is a problem extracting a String contained
   *         within the specified delimiters.
   */
  private
    String extractString( char startDelim, char stopDelim, boolean trim )
      throws InvalidMetaFileException
  {
    int d1pos, d2pos;      // delimiter positions

    d1pos = m_fileText.indexOf( startDelim );

    if( d1pos < 0 )
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: Missing opening " +
          "delimiter character " + startDelim + "." );
    else if( m_fileText.length() == (d1pos+1) )
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: Empty after first " +
          "delimiter character." );

    d2pos = m_fileText.indexOf( stopDelim, d1pos + 1 );

    if( d2pos < 0 )
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: Missing closing " +
          "delimiter character." );

    if( (d1pos+1) == d2pos )
      // The string is empty.
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: String between delimiters is empty." );

    // Extract the attribute name from the file text.
    String extractedString = m_fileText.substring( d1pos+1, d2pos );

    if( (d2pos+1) < m_fileText.length() )
      m_fileText = m_fileText.substring( d2pos+1 );
    else
      // There was nothing after the last delimiter.
      m_fileText = new String( "" );

    if( trim ) return extractedString.trim();

    return extractedString;
  }

  /**
   * Strips the specified string, and anything before
   * it, out of the file text.  The method also searches
   * for and removes the '=' sign immediately following
   * the string.
   *
   * @throws InvalidMetaFileException If the string
   *         is not located in the file text, or there
   *         is no trailing '=' sign.
   */
  private void stripString( String stringToStrip )
    throws InvalidMetaFileException
  {
    // Determine where in the file buffer the string
    //  is located, and strip it, and anything before it, out.
    int pos;

    if( (pos = m_fileText.indexOf( stringToStrip )) == -1 )
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: Missing '" +
          stringToStrip + "' keyword(s)." );

    try {
      m_fileText = m_fileText.substring( pos + stringToStrip.length() );
    }
    catch( StringIndexOutOfBoundsException e ) {
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: Empty after '" +
          stringToStrip + "' keyword(s)." );
    }

    if( !stripCharacter( '=' ) )
      throw new
        InvalidMetaFileException( "Syntax error in " +
          "configuration file: No '=' sign after '" +
          stringToStrip + "' keyword(s)." );
  }

  /**
   * Starts at the beginning of the current file text
   * and finds the first non-whitespace character.  If
   * the character matches, the method removes
   * the sign and any whitespace before and after it.
   * Otherwise, the method leaves the file text untouched.
   *
   * @param  charToStrip Character to remove.
   *
   * @return true if the character was found and chopped,
   *         false otherwise.
   */
  private boolean stripCharacter( char charToStrip )
  {
    int pos = 0;
    boolean foundChar = false;

    while( pos < m_fileText.length() ) {
      char currChar = m_fileText.charAt( pos );

      if( currChar == ' '  || currChar == '\r' ||
          currChar == '\t' || currChar == '\n' )
        pos++;
      else if( currChar == charToStrip ) {
        foundChar = true;
        break;
      }
      else
        break;
    }

    if( foundChar ) {
      pos++;

      while( pos < m_fileText.length() ) {
        char currChar = m_fileText.charAt( pos );

        if( currChar == ' '  || currChar == '\r' ||
            currChar == '\t' || currChar == '\n' )
          pos++;
        else
          break;
      }

      // Chop the file text.
      if( pos < m_fileText.length() )
        m_fileText = m_fileText.substring( pos );
      else
        // The only thing after the character
        // was whitespace.
        m_fileText = new String( "" );
    }

    return foundChar;
  }

  /**
   * Returns true if the supplied string only
   * contains whitespace (tab and space) characters,
   * or is empty.
   */
  private boolean isEmptyLine( String line )
  {
    if( line.length() == 0 ) return true;

    int pos = 0;
    boolean isEmpty = true;

    while( pos < line.length() ) {
      char currChar = m_fileText.charAt( pos );

      if( currChar == ' '  || currChar == '\r' ||
          currChar == '\t' || currChar == '\n' )
        pos++;
      else {
        isEmpty = false;
        break;
      }
    }

    return isEmpty;
  }
}