/**
* @param $html
* @return array
*/
public function load($html)
{
$metaTags = ['canonical' => '', 'robots' => ''];
$this->crawler->loadHtml((string) $html);
foreach ($this->crawler->find('meta') as $meta) {
/** @var Element $meta */
$name = strtolower($meta->attr('name'));
$content = $meta->attr('content');
$metaTags[$name] = $content;
}
$links = [];
foreach ($this->crawler->find('a') as $link) {
/** @var Element $link */
$rel = $link->attr('rel');
$href = $link->attr('href');
if ('nofollow' === strtolower($rel)) {
continue;
}
$links[] = $href;
}
$this->links = array_unique($links);
$this->metaTags = $metaTags;
return ['links' => $links, 'meta' => $metaTags];
}
public function testXmlWithOptions()
{
$html = '<html><body><span></span></body></html>';
$document = new Document();
$document->loadHtml($html);
$element = $document->find('span')[0];
$prolog = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
$this->assertEquals($prolog . '<span/>', $element->xml());
$this->assertEquals($prolog . '<span></span>', $element->xml(LIBXML_NOEMPTYTAG));
}
public function testLoadHtmlException()
{
$this->setExpectedException('InvalidArgumentException');
$document = new Document();
$document->loadHtml(array('element'));
}
/**
* @expectedException InvalidArgumentException
*/
public function testLoadHtmlWithInvalidArgument()
{
$document = new Document();
$document->loadHtml(null);
}
/**
* Request callback
*
* @param callable $callback
* @throws \Exception
* @return Curl
*/
public function setCallback($callback)
{
if (!is_callable($callback)) {
throw new \Exception(sprintf('Error: %s is not a valid callable', $callback));
}
$http_response = $this->getHttpResponse();
$didom = new Document();
$dom = $didom->loadHtml($http_response);
call_user_func_array($callback, [$http_response, $dom, $this]);
return $this;
}
public function testHtmlWithOptions()
{
$html = '<html><body><span></span></body></html>';
$document = new Document();
$document->loadHtml($html);
$this->assertEquals('<html><body><span></span></body></html>', $document->html());
$this->assertEquals('<html><body><span/></body></html>', $document->html(0));
}
