client = new Client; } public function extract($url, $date) { try { $crawler = $this->client->request('GET', $url); $title = $crawler->filter('h1')->first()->text(); if ($crawler->filter('figure img')->count() > 0) { $this->image = $crawler->filter('figure img')->first()->attr('src'); } if ($crawler->filter('.post_content p')->count() == 0) { return; } $crawler->filter('.post_content p')->each(function ($node) { $this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text()); }); $crawler->filter('div[class*="border-t border-grey-light border-dotted mt-7 py-3"] a')->each(function ($node) { //Removing the show more tags button if ($node->text() == "+") { return; } $this->topics[] = [ "name" => $node->text(), "slug" => str_replace("/", "", $node->attr('href')) ]; }); if ($crawler->filter('div[class*="font-waheed text-grey ml-3 pl-3 text-lg border-l border-grey border-dotted"] a')->count() == 1) { $this->author = $crawler->filter('div[class*="font-waheed text-grey ml-3 pl-3 text-lg border-l border-grey border-dotted"] a')->first()->text(); } //Remove all the alphabets from string //preg_replace("/[a-zA-Z]/", "",$string); return [ 'source' => 'Avas', 'title' => $title, 'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'), 'image' => $this->image, 'content' => $this->content, 'url' => $url, 'date' => $date, 'guid' => str_replace("https://avas.mv/", "", $url), 'author' => $this->author, 'topics' => $this->topics ]; } catch (\Exception $e) { } } }