// Pega pagina
$crawler = $this->goutte->request('GET', $url);
// Pega informaçoes da pagina
$data = array();
foreach ($website['scan'] as $key => $filter) {
$str = strip_tags($crawler->filterXPath($filter)->text());
$str = preg_replace($this->stop_words, "", $str);
$data[$key] = $str;
}
// coloca na lista de resultados
$this->result[$website['name']][] = $data;
}
}
/**
* Salva resultados em json
*
* @param null $dir
*/
public function save($dir = null)
{
if (is_null($dir)) {
$dir = realpath(dirname(__FILE__)) . DIRECTORY_SEPARATOR . "data" . DIRECTORY_SEPARATOR . "json";
}
foreach ($this->result as $entry => $data) {
file_put_contents($dir . DIRECTORY_SEPARATOR . $entry . ".json", json_encode($data));
}
}
}
$sp = new Spider();
$sp->run();
$sp->save();
