webgraph.java

一个用java语言编写的网络爬虫程序

                n.fixed = true;
                placeNodeOnGraph (n, 0, 0);
            }
            else {
                // root node of an additional tree -- drop it randomly
                // within window
                
                Dimension d = size ();
                
                int x = (int)(Math.random() * d.width);
                int y = (int)(Math.random() * d.height);
                placeNodeOnScreen (n, x, y);
            }
        }
        else {
            // check if parent's filter allows this link to be displayed
            if (!shouldDisplay (parent.filter, link.getStatus ()))
                return;

            // hang the node off its parent in a random direction
            double len = getRestLength();
            double angle = Math.random () * 2 * Math.PI;
            double x = parent.x + len*Math.cos(angle);
            double y = parent.y + len*Math.sin(angle);

            update (n);
            addNode (n);
            placeNodeOnGraph (n, x, y);

            WebEdge e = new WebEdge (link, parent, n);
            links.put (link, e);
            update (e);
            addEdge (e);
        }
    }

    void update (WebEdge e) {
        e.color = Colors.parseColor (e.link.getLabel ("Workbench.color"));
        e.thick = e.link.hasLabel ("Workbench.thick");
    }

    void update (WebNode n) {
        Page page = n.link.getPage ();
        int status = n.link.getStatus ();

        if (page == null) {
            // not downloaded yet
            switch (n.rendering) {
            case ICON:
                n.name = null;
                n.icon = getIcon (LinkEvent.eventName[status]);
                break;
            case TITLE:
            case ABSOLUTE_URL:
                n.name = n.link.getURL().toString();
                n.icon = null;
                break;
            case RELATIVE_URL: {
                Link origin = n.link.getSource().getOrigin();
                if (origin != null)
                    n.name = Link.relativeTo (origin.getURL(), n.link.getURL());
                else
                    n.name = n.link.getURL().toString();
                n.icon = null;
                break;
            }
            }
        }
        else {
            switch (n.rendering) {
            case ICON:
                n.name = null;
                n.icon = getIcon (page.getLabel ("Workbench.icon", 
                                                 LinkEvent.eventName[status]));
                break;
            case TITLE:
                n.name = page.getTitle ();
                if (n.name == null)
                    n.name = "[" + n.link.getURL().toString() + "]";
                n.icon = null;
                break;
            case ABSOLUTE_URL:
                n.name = n.link.getURL().toString();
                n.icon = null;
                break;
            case RELATIVE_URL: {
                Link origin = n.link.getSource().getOrigin();
                if (origin != null)
                    n.name = Link.relativeTo (origin.getURL(), n.link.getURL());
                else
                    n.name = n.link.getURL().toString();
                n.icon = null;
                break;
            }
            }
            n.color = Colors.parseColor (page.getLabel ("Workbench.color"));
            n.scale = page.getNumericLabel ("Workbench.size", 
                            new Integer (1)).doubleValue ();
        }

        if (n.icon == null) {
            FontMetrics fm = getFontMetrics ();
            n.width = fm.stringWidth (n.name) + 10;
            n.height = fm.getHeight () + 4;
        }
        else {
            n.width = (int)(n.icon.getWidth(this) * n.scale);
            n.height = (int)(n.icon.getHeight(this) * n.scale);
        }
    }

    WebEdge findEdge (Link l) {
        if (l == null)
                return null;
            Object obj = links.get (l);
            if (obj instanceof WebEdge)
                return (WebEdge)obj;
            else
                return null;
    }        

    WebNode findNode (Link l) {
        if (l == null)
            return null;
        Object obj = links.get (l);
        if (obj instanceof WebEdge)
            return (WebNode)((WebEdge)obj).to;
        else
            return (WebNode)obj;
    }
    
    WebNode findParent (Link l) {
        if (l == null)
            return null;
        Page source = l.getSource ();
        Link origin	= source.getOrigin ();
        return findNode (origin);
    }

    /*
     * LinkView listeners
     */

    private Vector listeners = new Vector ();

    /**
     * Add a listener for LinkViewEvents.  A LinkViewEvent is sent every time a
     * node or edge in the graph is double-clicked.
     * @param listener Object that wants to receive LinkViewEvents 
     */
    public void addLinkViewListener (LinkViewListener listener) {
        if (!listeners.contains (listener))
            listeners.addElement (listener);
    }

    /**
     * Removes a listener from the set of LinkViewEvent listeners.  If it is not found in the set,
     * does nothing.
     *
     * @param listen a listener
     */
    public void removeLinkViewListener (CrawlListener listener) {
        listeners.removeElement (listener);
    }

    void fireEvent (Link link) {
        LinkViewEvent event = new LinkViewEvent (this, link);

        for (int j=0, len=listeners.size(); j<len; ++j) {
            LinkViewListener listen = (LinkViewListener)listeners.elementAt(j);
            listen.viewLink (event);
        }
    }

    void doubleClick (int x, int y) {
        Object over = pick (x, y);
        if (over != null) {
            Link link;
            if (over instanceof WebNode)
                link = ((WebNode)over).link;
            else
                link = ((WebEdge)over).link;
            fireEvent (link);
        }
    }
        
    public boolean handleEvent (Event event) {
        if (event.id == Event.MOUSE_DOWN && event.clickCount == 2) {
            doubleClick (event.x, event.y);
        }
        else
            return super.handleEvent (event);
        return true;
    }
    
    public Link getSelectedLink () {
        WebNode n = (WebNode)getSelectedNode();
        if (n != null)
            return n.link;
            
        WebEdge e = (WebEdge)getSelectedEdge();
        if (e != null)
            return e.link;
            
        return null;
    }
    
    /**
     * Create a new Frame containing a WebGraph connected to a crawler.
     */ 
    public static Frame monitor (Crawler crawler) {
        Frame win = new ClosableFrame ("Graph: " + crawler.getName ());

        WebGraph g = new WebGraph ();
        crawler.addCrawlListener (g);
        crawler.addLinkListener (g);
        g.setNodeCharge (1000);
        g.setRestLength (50);

        win.add ("Center", g);
        win.pack ();
        win.show ();

        return win;
    }

    static String[] getTip (Link link) {
        Vector result = new Vector ();

        String exception = link.getLabel ("exception");
        if (exception != null && exception.length() > 0)
            result.addElement ("*** " + exception);

        String anchor = link.toText ();
        if (anchor != null && anchor.length() > 0)
            result.addElement (anchor);

        String url = link.getURL ().toString ();
        if (url != null && url.length() > 0)
            result.addElement (url);

        String labels = link.getObjectLabels ();
        if (labels != null && labels.length() > 0)
            result.addElement (labels);

        String[] tip = new String[result.size ()];
        result.copyInto (tip);
        return tip;
    }

    static String[] getTip (Page page) {
        Vector result = new Vector ();

        String title = page.getTitle ();
        if (title != null && title.length() > 0)
            result.addElement (title);

        String url = page.getURL ().toString();
        if (url != null && url.length() > 0)
            result.addElement (url);

        String labels = page.getObjectLabels ();
        if (labels != null && labels.length() > 0)
            result.addElement (labels);

        String[] tip = new String[result.size ()];
        result.copyInto (tip);
        return tip;
    }


    Hashtable icons = new Hashtable ();
       // maps String (CrawlEvent name or user-defined icon name) to Image

    Image pageIcon;
    Image linkIcon;
    Image retrievingIcon;