public function save(array $publications)
{
if (0 === count($publications)) {
return;
}
$repository = $this->_em->getRepository('ECENetagoraBundle:KnownLink');
$references = $this->getExistingReferences($publications);
foreach ($publications as $publication) {
if (!in_array($publication->getReference(), $references)) {
// Scrap the publication and get the latest uri
try {
$response = $this->browser->get($publication->getLinkUrl());
} catch (\Exception $e) {
// The Http request failed...
continue;
}
$urls = explode("\n", $response->getHeader('Location'));
$url = array_pop($urls);
$knownLink = $this->_em->getRepository('ECENetagoraBundle:KnownLink')->findOneByUrl($url);
// We found a corresponding KnownLink that we can affect to the publication
if ($knownLink) {
$publication->setKnownLink($knownLink);
$this->_em->persist($publication);
continue;
}
// Otherwise, we need to create a new KnownLink and guess the category
$crawler = new Crawler();
$crawler->addContent($response->getContent());
$guesser = new CategoryGuesser($url, $response, $crawler);
$guesser->guess();
$category = $this->_em->getRepository('ECENetagoraBundle:Category')->findOneByType($guesser->getCategory());
/*if (null === $category) {
var_dump($guesser->getScores());die;
echo $guesser->getCategory();die;
}*/
$knownLink = new KnownLink();
$knownLink->setCategory($category);
$knownLink->setUrl($url);
$knownLink->fromArray($guesser->getMetadata());
$publication->setKnownLink($knownLink);
$this->_em->persist($knownLink);
$this->_em->persist($publication);
}
}
$this->_em->flush();
}
public function getCategory()
{
if ($this->knownLink) {
return $this->knownLink->getCategory();
}
}
