diff --git a/app/Services/Scrapers/SunScraper.php b/app/Services/Scrapers/SunScraper.php new file mode 100644 index 0000000..541dfb1 --- /dev/null +++ b/app/Services/Scrapers/SunScraper.php @@ -0,0 +1,57 @@ +client = new Client; + } + + public function extract($url) + { + + $crawler = $this->client->request('GET', $url); + + $crawler->filter('h1')->each(function ($node) { + $this->title = $node->text(); + }); + + $crawler->filter('.component-article-content p')->each(function ($node) { + $this->content[] = preg_replace("/[a-zA-Z]/","",$node->text()); + }); + + $crawler->filter('.component-article-tag li a')->each(function ($node) { + + $this->topics[] = [ + "name" => $node->first()->text(), + "slug" => str_replace("https://sun.mv/", "", $node->first()->attr('href')) + ]; + }); + + //Remove all the alphabets from string + //preg_replace("/[a-zA-Z]/", "",$string); + return [ + 'source' => 'Sun', + 'title' => $this->title, + 'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'), + 'image' => $crawler->filter(".component-article-featured img")->first()->attr('src'), + 'content' => $this->content, + 'url' => $url, + 'date' => $crawler->filter(".author .time")->first()->text(), + 'guid' => str_replace("https://sun.mv/","",$url), + 'author' => $crawler->filter(".author .name")->first()->text(), + 'topics' => collect($this->topics)->unique()->values()->toArray() + ]; + } +}