public function handle(Spizer_Document $doc)
{
// Silently skip all non-HTML documents
if (!$doc instanceof Spizer_Document_Html) {
return;
}
// Add document URL to the list of visited pages
$baseUrl = $doc->getUrl();
if (!in_array($baseUrl, $this->targets)) {
$this->targets[] = $baseUrl;
}
$pagerize = new Diggin_Scraper_Helper_Simplexml_Pagerize(simplexml_import_dom($doc->getDomDocument()), array('baseUrl' => $this->toUrl($doc->getUrl())));
if ($nextLink = $pagerize->getNextLink()) {
$max_follow = $this->_config['max_follow'];
if (!$max_follow or $this->page_count <= $max_follow) {
$this->addToQueue($nextLink, $baseUrl);
++$this->page_count;
}
}
}
