/**
* Adds all the images from the page to the web
* @param zibo\library\spider\Web $web The spider web
* @param zibo\library\spider\WebNode $prey The current prey in the web
* @param string $baseUrl Base URL of the crawl
* @param string $preyBaseUrl Base URL of the prey
* @param zibo\library\xml\dom\Document $dom The DOM document of the current prey
* @return null
*/
protected function biteDocument(Web $web, WebNode $prey, $baseUrl, $preyBaseUrl, Document $dom)
{
$images = $dom->getElementsByTagName('img');
foreach ($images as $image) {
$url = $image->getAttribute('src');
if (!$url) {
continue;
}
$url = $this->getAbsoluteUrl($url, $baseUrl, $preyBaseUrl);
$link = $web->getNode($url);
$link->addType(WebNode::TYPE_IMAGE);
$link->addReference($prey);
$prey->addLink($link);
}
}
/**
* Adds all the used javascripts to the web
* @param zibo\library\spider\Web $web The spider web
* @param zibo\library\spider\WebNode $prey The current prey in the web
* @param string $baseUrl Base URL of the crawl
* @param string $preyBaseUrl Base URL of the prey
* @param zibo\library\xml\dom\Document $dom The DOM document of the current prey
* @return null
*/
protected function biteDocument(Web $web, WebNode $prey, $baseUrl, $preyBaseUrl, Document $dom)
{
$scripts = $dom->getElementsByTagName('script');
foreach ($scripts as $script) {
$type = $script->getAttribute('type');
$url = $script->getAttribute('src');
if ($type != 'text/javascript' || !$url) {
continue;
}
$url = $this->getAbsoluteUrl($url, $baseUrl, $preyBaseUrl);
$link = $web->getNode($url);
$link->addType(WebNode::TYPE_JS);
$link->addReference($prey);
$prey->addLink($link);
}
}
/**
* Adds the URL's from the anchors in the page to the web
* @param zibo\library\spider\Web $web The spider web
* @param zibo\library\spider\WebNode $prey The current prey in the web
* @param string $baseUrl Base URL of the crawl
* @param string $preyBaseUrl Base URL of the prey
* @param zibo\library\xml\dom\Document $dom The DOM document of the current prey
* @return null
*/
protected function biteDocument(Web $web, WebNode $prey, $baseUrl, $preyBaseUrl, Document $dom)
{
$anchors = $dom->getElementsByTagName('a');
foreach ($anchors as $anchor) {
$url = $anchor->getAttribute('href');
if (!$url || String::startsWith($url, '#')) {
continue;
}
if (!String::startsWith($url, 'mailto:')) {
$url = $this->getAbsoluteUrl($url, $baseUrl, $preyBaseUrl);
}
$node = $web->getNode($url);
$node->addReference($prey);
$prey->addLink($node);
}
}
/**
* Adds the used style sheets to the web
* @param zibo\library\spider\Web $web The spider web
* @param zibo\library\spider\WebNode $prey The current prey in the web
* @param string $baseUrl Base URL of the crawl
* @param string $preyBaseUrl Base URL of the prey
* @param zibo\library\xml\dom\Document $dom The DOM document of the current prey
* @return null
*/
protected function biteDocument(Web $web, WebNode $prey, $baseUrl, $preyBaseUrl, Document $dom)
{
$links = $dom->getElementsByTagName('link');
foreach ($links as $link) {
$type = $link->getAttribute('type');
$rel = $link->getAttribute('rel');
$url = $link->getAttribute('href');
if ($type != 'text/css' || $rel != 'stylesheet' || !$url) {
continue;
}
$url = $this->getAbsoluteUrl($url, $baseUrl, $preyBaseUrl);
$link = $web->getNode($url);
$link->addType(WebNode::TYPE_CSS);
$link->addReference($prey);
$prey->addLink($link);
}
}
/**
* Reads the dependencies from the provided file and adds them to the
* provided container
* @param zibo\core\di\DependencyContainer $container
* @param zibo\library\filesystem\File $file
* @return null
*/
private function readDependencies(DependencyContainer $container, File $file)
{
$dom = new Document();
$dom->load($file);
$dependencyElements = $dom->getElementsByTagName(self::TAG_DEPENDENCY);
foreach ($dependencyElements as $dependencyElement) {
$interface = $dependencyElement->getAttribute(self::ATTRIBUTE_INTERFACE);
$className = $dependencyElement->getAttribute(self::ATTRIBUTE_CLASS);
$id = $dependencyElement->getAttribute(self::ATTRIBUTE_ID);
if (!$id) {
$id = null;
}
$dependency = new Dependency($className, $id);
$this->readCalls($dependency, $dependencyElement);
$container->addDependency($interface, $dependency);
}
}
