findwithcapabilities.java

Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等

	m_Packages.add(tok.nextToken());
    }
    
    if (Utils.getFlag("generic", options)) {
      creator    = new GenericPropertiesCreator();
      creator.execute(false);
      props	 = creator.getInputProperties();
      tok        = new StringTokenizer(props.getProperty(m_Superclass), ",");
      m_Packages = new Vector();
      while (tok.hasMoreTokens())
	m_Packages.add(tok.nextToken());
    }
  }
  
  /**
   * Gets the current settings of this object.
   * 
   * @return an array of strings suitable for passing to setOptions
   */
  public String[] getOptions() {
    Vector 	result;
    String[]	options;
    int		i;

    result = new Vector();
    
    result.add("-num-instances");
    result.add("" + m_Capabilities.getMinimumNumberInstances());
    
    if (m_Capabilities.handles(Capability.NO_CLASS)) {
      result.add("-no-class");
    }
    else {
      if (m_Capabilities.handles(Capability.NOMINAL_CLASS))
	result.add("-nominal-class");
      if (m_Capabilities.handles(Capability.BINARY_CLASS))
	result.add("-binary-class");
      if (m_Capabilities.handles(Capability.NUMERIC_CLASS))
	result.add("-numeric-class");
      if (m_Capabilities.handles(Capability.STRING_CLASS))
	result.add("-string-class");
      if (m_Capabilities.handles(Capability.DATE_CLASS))
	result.add("-date-class");
      if (m_Capabilities.handles(Capability.RELATIONAL_CLASS))
	result.add("-relational-class");
    }
    
    if (m_Capabilities.handles(Capability.NOMINAL_ATTRIBUTES))
	result.add("-nominal-atts");
    if (m_Capabilities.handles(Capability.BINARY_ATTRIBUTES))
	result.add("-binary-atts");
    if (m_Capabilities.handles(Capability.NUMERIC_ATTRIBUTES))
	result.add("-numeric-atts");
    if (m_Capabilities.handles(Capability.STRING_ATTRIBUTES))
	result.add("-string-atts");
    if (m_Capabilities.handles(Capability.DATE_ATTRIBUTES))
	result.add("-date-atts");
    if (m_Capabilities.handles(Capability.RELATIONAL_ATTRIBUTES))
	result.add("-relational-atts");

    if (m_Capabilities.handles(Capability.ONLY_MULTIINSTANCE))
      result.add("-only-multiinstance");

    if (getHandler() != null) {
      result.add("-W");
      result.add(getHandler().getClass().getName());
      if (getHandler() instanceof OptionHandler) {
	result.add("--");
	options = ((OptionHandler) getHandler()).getOptions();
	for (i = 0; i < options.length; i++)
	  result.add(options[i]);
      }
    }
    else if (getFilename().length() != 0) {
      result.add("-t");
      result.add(getFilename());
      result.add("-c");
      result.add(m_ClassIndex.getSingleIndex());
    }

    if (m_Superclass.length() != 0) {
      result.add("-superclass");
      result.add(m_Superclass);
    }
    else {
      result.add("-packages");
      result.add(m_Packages.toString().replaceAll("\\[", "").replaceAll("\\]", ""));
    }
    
    return (String[]) result.toArray(new String[result.size()]);
  }
  
  /**
   * sets the Capabilities handler to generate the data for
   */
  public void setHandler(CapabilitiesHandler value) {
    m_Handler = value;
    setCapabilities(m_Handler.getCapabilities());
  }
  
  /**
   * returns the current set CapabilitiesHandler to generate the dataset
   * for, can be null
   */
  public CapabilitiesHandler getHandler() {
    return m_Handler;
  }
  
  /**
   * Sets the dataset filename to base the capabilities on. It immediately
   * loads the dataset and retrieves the capabilities from it.
   */
  public void setFilename(String value) {
    Instances		insts;
    
    m_Filename = value;

    if (m_Filename.length() != 0) {
      try {
        insts  = new Instances(new BufferedReader(new FileReader(m_Filename)));
        m_ClassIndex.setUpper(insts.numAttributes());
        insts.setClassIndex(Integer.parseInt(getClassIndex()) - 1);
        
        setCapabilities(Capabilities.forInstances(insts));
      }
      catch (Exception e) {
	e.printStackTrace();
      }
    }
  }
  
  /**
   * returns the current filename for the dataset to base the capabilities on
   */
  public String getFilename() {
    return m_Filename;
  }
  
  /**
   * sets the class index, -1 for none, first and last are also valid
   */
  public void setClassIndex(String value) {
    if (value.equals("-1"))
      m_ClassIndex = null;
    else
      m_ClassIndex = new SingleIndex(value);
  }
  
  /**
   * returns the current current class index, -1 if no class attribute
   */
  public String getClassIndex() {
    if (m_ClassIndex == null)
      return "-1";
    else
      return "" + m_ClassIndex.getIndex();
  }

  /**
   * enables the given capability
   * 
   * @param c		the capability to enable
   */
  public void enable(Capability c) {
    m_Capabilities.enable(c);
  }

  /**
   * disables the given capability
   * 
   * @param c		the capability to disable
   */
  public void disable(Capability c) {
    m_Capabilities.disable(c);
  }

  /**
   * returns true if the given capability can be handled
   * 
   * @param c		the capability to check
   * @return		true if the capability can be handled
   */
  public boolean handles(Capability c) {
    return m_Capabilities.handles(c);
  }
  
  /**
   * The capabilities to search for.
   *
   * @return            the capabilities to search for
   * @see               Capabilities
   */
  public Capabilities getCapabilities() {
    return m_Capabilities;
  }
  
  /**
   * Uses the given Capabilities for the search.
   * 
   * @param c		the capabilities to use for the search
   */
  public void setCapabilities(Capabilities c) {
    m_Capabilities = (Capabilities) c.clone();
  }

  /**
   * returns the matches from the last find call
   * 
   * @return		the matching classname from the last find run
   */
  public Vector getMatches() {
    return m_Matches;
  }

  /**
   * returns the misses from the last find call
   * 
   * @return		the classnames that didn't match from the last find run
   */
  public Vector getMisses() {
    return m_Misses;
  }
  
  /**
   * returns a list with all the classnames that fit the criteria
   * 
   * @return		contains all classnames that fit the criteria
   */
  public Vector find() {
    Vector		list;
    int			i;
    Class		cls;
    Object		obj;
    CapabilitiesHandler	handler;
    boolean		fits;
    Capabilities	caps;
    
    m_Matches = new Vector();
    m_Misses  = new Vector();

    list = ClassDiscovery.find(m_Superclass, (String[]) m_Packages.toArray(new String[m_Packages.size()]));
    for (i = 0; i < list.size(); i++) {
      try {
	cls = Class.forName((String) list.get(i));
	obj = cls.newInstance();
	
	// exclude itself
	if (cls == this.getClass())
	  continue;
	
	// really a CapabilitiesHandler?
	if (!(obj instanceof CapabilitiesHandler))
	  continue;

	// check capabilities enumeration
	handler = (CapabilitiesHandler) obj;
	caps    = handler.getCapabilities();
	fits    = true;
	for (Capability cap: Capability.values()) {
	  if (m_Capabilities.handles(cap)) {
	    if (!(caps.handles(cap))) {
	      fits = false;
	      break;
	    }
	  }
	}
	if (!fits) {
	  m_Misses.add(list.get(i));
	  continue;
	}
	
	// other stuff
	if (caps.getMinimumNumberInstances() > m_Capabilities.getMinimumNumberInstances()) {
	  m_Misses.add(list.get(i));
	  continue;
	}
	
	// matches all criteria!
	m_Matches.add(list.get(i));
      }
      catch (Exception e) {
	// ignore
      }
    }
    
    return m_Matches;
  }

  /**
   * Executes the location of classes with parameters from the commandline.
   * 
   * @param args	the commandline parameters
   */
  public static void main(String[] args) {
    FindWithCapabilities 	find;
    Vector 			list;
    String			result;
    int				i;
    boolean			printMisses;
    
    printMisses = false;
    
    try {
      find = new FindWithCapabilities();
      
      try {
	printMisses = Utils.getFlag("misses", args);
        find.setOptions(args);
        Utils.checkForRemainingOptions(args);
      } 
      catch (Exception ex) {
        result = ex.getMessage() + "\n\n" + find.getClass().getName().replaceAll(".*\\.", "") + " Options:\n\n";
        Enumeration enm = find.listOptions();
        while (enm.hasMoreElements()) {
          Option option = (Option) enm.nextElement();
          result += option.synopsis() + "\n" + option.description() + "\n";
        }
        throw new Exception(result);
      }
      
      System.out.println("\nSearching for the following Capabilities:\n\n" + find.getCapabilities());
      
      find.find();

      // matches
      list = find.getMatches();
      if (list.size() == 1)
	System.out.println("\nFound " + list.size() + " class that matched the criteria:\n");
      else
	System.out.println("\nFound " + list.size() + " classes that matched the criteria:\n");
      for (i = 0; i < list.size(); i++)
	System.out.println(list.get(i));
      
      // misses
      if (printMisses) {
        list = find.getMisses();
        if (list.size() == 1)
  	System.out.println("\nFound " + list.size() + " class that didn't match the criteria:\n");
        else
  	System.out.println("\nFound " + list.size() + " classes that didn't match the criteria:\n");
        for (i = 0; i < list.size(); i++)
  	System.out.println(list.get(i));
      }
    } 
    catch (Exception ex) {
      System.err.println(ex.getMessage());
    }
  }
}