Beispiel #1
0
string WebCrawler::crawl(string & webpage, string & stopWordsFile)
{


	
	URL startURL(webpage,"");
	
	string currentPageContents;
	
	try
	{
		currentPageContents = downloader->download(startURL); 
	}
	catch(CS240Exception &e)
	{
		string newString = e.GetMessage();
		newString += " Invalid Start URL";
		CS240Exception newE(newString);
		throw newE;
	}

	cout << "Crawling ... " << endl;
	string currentPageDescription = parser->findDescription(currentPageContents);
	
    currentPage = new Page(currentPageDescription, startURL);
				
	queue->AddPage(currentPage);
	history->AddPage(currentPage);


	while (queue->HasNext()) 
	{
		currentPage = queue->GetNext();
		URL currentURL(currentPage->GetURL());
		currentPageContents = downloader->download(currentURL);
		cout <<"  Crawling: "<<currentURL.GetResolvedURL() << endl;
		
		//Note: findLinks adds all the links in currentPageContents to the queue, if they 
		// are not already in the history. If a page is added to the queue, it is also
		// added to the history. 
		URL * newURLstoCrawl = parser->findLinks(currentPageContents,currentPage);
		int numOfURLs = parser->countLinks(currentPageContents);	

		processNewLinks(newURLstoCrawl, numOfURLs, startURL);
		
		delete [] newURLstoCrawl;		

		//Note: findWords, finds words and adds them to the index.
		parser->findWords(currentPageContents, index, currentPage);
		
		
	}
	
	//Note: the xmlPrinter will not print words that are in the stop word file.
	cout << "Generating Output File..." << endl;


	return xmlPrinter->print(index, stopWordsFile, history);


}
Beispiel #2
0
// TODO convert to use xmlui file
void KuickShow::initGUI( const KURL& startDir )
{
	KURL startURL( startDir );
	if ( !KProtocolInfo::supportsListing( startURL ) )
		startURL = KURL();

    fileWidget = new FileWidget( startURL, this, "MainWidget" );
    setFocusProxy( fileWidget );

    KActionCollection *coll = fileWidget->actionCollection();

    redirectDeleteAndTrashActions(coll);

    connect( fileWidget, SIGNAL( fileSelected( const KFileItem * ) ),
             this, SLOT( slotSelected( const KFileItem * ) ));

    connect( fileWidget, SIGNAL( fileHighlighted( const KFileItem * )),
             this, SLOT( slotHighlighted( const KFileItem * ) ));

    connect( fileWidget, SIGNAL( urlEntered( const KURL&  )),
             this, SLOT( dirSelected( const KURL& )) );


    fileWidget->setAcceptDrops(true);
    connect( fileWidget, SIGNAL( dropped( const KFileItem *, QDropEvent *, const KURL::List & )),
             this, SLOT( slotDropped( const KFileItem *, QDropEvent *, const KURL::List &)) );

    // setup actions
    KAction *open = KStdAction::open( this, SLOT( slotOpenURL() ),
                                      coll, "openURL" );

    KAction *print = KStdAction::print( this, SLOT( slotPrint() ),
                                        coll, "kuick_print" );
    print->setText( i18n("Print Image...") );

    KAction *configure = new KAction( i18n("Configure %1...").arg( KGlobal::instance()->aboutData()->programName() ), "configure",
                                      KShortcut(),
                                      this, SLOT( configuration() ),
                                      coll, "kuick_configure" );
    KAction *slide = new KAction( i18n("Start Slideshow" ), "ksslide",
                                  KShortcut( Key_F2 ),
                                  this, SLOT( startSlideShow() ),
                                  coll, "kuick_slideshow" );
    KAction *about = new KAction( i18n( "About KuickShow" ), "about",
                                  KShortcut(),
                                  this, SLOT( about() ), coll, "about" );

    oneWindowAction = new KToggleAction( i18n("Open Only One Image Window"),
                                         "window_new",
                                         KShortcut( CTRL+Key_N ), coll,
                                         "kuick_one window" );

    m_toggleBrowserAction = new KToggleAction( i18n("Show File Browser"), KShortcut( Key_Space ), coll, "toggleBrowser" );
    m_toggleBrowserAction->setCheckedState(i18n("Hide File Browser"));
    connect( m_toggleBrowserAction, SIGNAL( toggled( bool ) ),
             SLOT( toggleBrowser() ));

    KAction *showInOther = new KAction( i18n("Show Image"), KShortcut(),
                                        this, SLOT( slotShowInOtherWindow() ),
                                        coll, "kuick_showInOtherWindow" );
    KAction *showInSame = new KAction( i18n("Show Image in Active Window"),
                                       KShortcut(),
                                       this, SLOT( slotShowInSameWindow() ),
                                       coll, "kuick_showInSameWindow" );
    KAction *showFullscreen = new KAction( i18n("Show Image in Fullscreen Mode"),
					   KShortcut(), this, SLOT( slotShowFullscreen() ),
					   coll, "kuick_showFullscreen" );

    KAction *quit = KStdAction::quit( this, SLOT(slotQuit()), coll, "quit");

    // remove QString::null parameter -- ellis
    coll->readShortcutSettings( QString::null );
    m_accel = coll->accel();

    // menubar
    KMenuBar *mBar = menuBar();
    QPopupMenu *fileMenu = new QPopupMenu( mBar, "file" );
    open->plug( fileMenu );
    showInOther->plug( fileMenu );
    showInSame->plug( fileMenu );
    showFullscreen->plug( fileMenu );
    fileMenu->insertSeparator();
    slide->plug( fileMenu );
    print->plug( fileMenu );
    fileMenu->insertSeparator();
    quit->plug( fileMenu );

    QPopupMenu *editMenu = new QPopupMenu( mBar, "edit" );
    coll->action("mkdir")->plug( editMenu );
    coll->action("delete")->plug( editMenu );
    editMenu->insertSeparator();
    coll->action("properties")->plug( editMenu );


    // remove the Sorting submenu (and the separator below)
    // from the main contextmenu
    KActionMenu *sortingMenu = static_cast<KActionMenu*>( coll->action("sorting menu"));
    KActionMenu *mainActionMenu = static_cast<KActionMenu*>( coll->action("popupMenu"));
    QPopupMenu *mainPopup = mainActionMenu->popupMenu();
    int sortingIndex = mainPopup->indexOf( sortingMenu->itemId( 0 ) );
    int separatorId = mainPopup->idAt( sortingIndex + 1 );
    QMenuItem *separatorItem = mainPopup->findItem( separatorId );
    if ( separatorItem && separatorItem->isSeparator() )
        mainPopup->removeItem( separatorId );
    mainActionMenu->remove( sortingMenu );

    // add the sorting menu and a separator into the View menu
    KActionMenu *viewActionMenu = static_cast<KActionMenu*>( coll->action("view menu"));
    viewActionMenu->popupMenu()->insertSeparator( 0 );
    sortingMenu->plug( viewActionMenu->popupMenu(), 0 ); // on top of the menu


    QPopupMenu *settingsMenu = new QPopupMenu( mBar, "settings" );
    configure->plug( settingsMenu );

    mBar->insertItem( i18n("&File"), fileMenu );
    mBar->insertItem( i18n("&Edit"), editMenu );
    viewActionMenu->plug( mBar );
    mBar->insertItem( i18n("&Settings"), settingsMenu );

    // toolbar
    KToolBar *tBar = toolBar();
    tBar->setText( i18n( "Main Toolbar" ) );

    coll->action("up")->plug( tBar );
    coll->action("back")->plug( tBar );
    coll->action("forward")->plug( tBar );
    coll->action("home")->plug( tBar );
    coll->action("reload")->plug( tBar );

    tBar->insertSeparator();

    coll->action( "short view" )->plug( tBar );
    coll->action( "detailed view" )->plug( tBar );
    coll->action( "preview")->plug( tBar );

    tBar->insertSeparator();
    configure->plug( tBar );
    slide->plug( tBar );
    tBar->insertSeparator();
    oneWindowAction->plug( tBar );
    print->plug( tBar );
    tBar->insertSeparator();
    about->plug( tBar );

    QPopupMenu *help = helpMenu( QString::null, false );
    mBar->insertItem( KStdGuiItem::help().text() , help );


    KStatusBar* sBar = statusBar();
    sBar->insertItem( "           ", URL_ITEM, 10 );
    sBar->insertItem( "                          ", META_ITEM, 2 );
    sBar->setItemAlignment(URL_ITEM, QLabel::AlignVCenter | QLabel::AlignLeft);

    fileWidget->setFocus();

    KConfig *kc = KGlobal::config();
    kc->setGroup("SessionSettings");
    bool oneWindow = kc->readBoolEntry("OpenImagesInActiveWindow", true );
    oneWindowAction->setChecked( oneWindow );

    tBar->show();

    // Address box in address tool bar
    KToolBar *addressToolBar = toolBar( "address_bar" );
    const int ID_ADDRESSBAR = 1;

    cmbPath = new KURLComboBox( KURLComboBox::Directories,
                                true, addressToolBar, "address_combo_box" );
    KURLCompletion *cmpl = new KURLCompletion( KURLCompletion::DirCompletion );
    cmbPath->setCompletionObject( cmpl );
    cmbPath->setAutoDeleteCompletionObject( true );

    addressToolBar->insertWidget( ID_ADDRESSBAR, 1, cmbPath);
    addressToolBar->setItemAutoSized( ID_ADDRESSBAR );

    connect( cmbPath, SIGNAL( urlActivated( const KURL& )),
             this, SLOT( slotSetURL( const KURL& )));
    connect( cmbPath, SIGNAL( returnPressed()),
             this, SLOT( slotURLComboReturnPressed()));


    fileWidget->initActions();
    fileWidget->clearHistory();
    dirSelected( fileWidget->url() );

    setCentralWidget( fileWidget );
    setupGUI( KMainWindow::Save );

    coll->action( "reload" )->setShortcut( KStdAccel::reload() );
    coll->action( "short view" )->setShortcut(Key_F6);
    coll->action( "detailed view" )->setShortcut(Key_F7);
    coll->action( "show hidden" )->setShortcut(Key_F8);
    coll->action( "mkdir" )->setShortcut(Key_F10);
    coll->action( "preview" )->setShortcut(Key_F11);
    coll->action( "separate dirs" )->setShortcut(Key_F12);
}