60 lines
1.8 KiB
PHP
60 lines
1.8 KiB
PHP
<?php
|
|
|
|
namespace App\Services\Scrapers;
|
|
|
|
use Exception;
|
|
use Goutte\Client;
|
|
use Illuminate\Support\Carbon;
|
|
|
|
class HamaScraper
|
|
{
|
|
protected $client;
|
|
|
|
protected $title;
|
|
protected $content;
|
|
protected $author = "unknown";
|
|
protected $topics;
|
|
|
|
public function __construct()
|
|
{
|
|
$this->client = new Client;
|
|
}
|
|
|
|
public function extract($url, $date)
|
|
{
|
|
$crawler = $this->client->request('GET', $url);
|
|
|
|
$crawler->filter('.body > p')->each(function ($node) {
|
|
$this->content[] = $node->text();
|
|
});
|
|
|
|
if ($crawler->filter('.author_name')->count() > 0) {
|
|
$this->author = $crawler->filter('.author_name')->first()->text();
|
|
}
|
|
|
|
$crawler->filter('.article-tags a')->each(function ($node) {
|
|
$this->topics[] = [
|
|
"name" => $node->text(),
|
|
"slug" => str_replace("https://hama.mv/", "", $node->attr('href'))
|
|
];
|
|
});
|
|
|
|
|
|
//Remove all the alphabets from string
|
|
//preg_replace("/[a-zA-Z]/", "",$string);
|
|
return [
|
|
'source' => 'Hama',
|
|
'title' => $crawler->filter('h1')->first()->text(),
|
|
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
|
|
'image' => $crawler->filter("figure > img")->first()->attr('data-src'),
|
|
'content' => $this->content,
|
|
'url' => $url,
|
|
'date' => Carbon::parse($date)->format("Y-m-d H:i:s"),
|
|
'guid' => basename($url),
|
|
'author' => $this->author,
|
|
'topics' => $this->topics
|
|
];
|
|
|
|
}
|
|
}
|