57 lines
1.6 KiB
PHP
57 lines
1.6 KiB
PHP
<?php
|
|
|
|
namespace App\Services\Scrapers;
|
|
|
|
use Goutte\Client;
|
|
use Illuminate\Support\Carbon;
|
|
|
|
class VoiceScraper
|
|
{
|
|
protected $client;
|
|
protected $author = 'unknown';
|
|
protected $content = [];
|
|
protected $topics = [];
|
|
|
|
public function __construct()
|
|
{
|
|
$this->client = new Client();
|
|
}
|
|
|
|
public function extract($url)
|
|
{
|
|
$crawler = $this->client->request('GET', $url);
|
|
|
|
// Extracting title - checking for multiple class names
|
|
$title = $crawler->filter('.text-3xl')->first()->text();
|
|
|
|
|
|
// Extracting article content
|
|
$crawler->filter('.container .dv')->each(function ($node) {
|
|
$this->content[] = $node->text();
|
|
});
|
|
|
|
// Extracting topics
|
|
$this->topics = [];
|
|
$crawler->filter('.related-tags-holder a')->each(function ($node) {
|
|
$this->topics[] = [
|
|
"name" => $node->text(),
|
|
"slug" => str_replace("/", "", $node->attr('href'))
|
|
];
|
|
});
|
|
|
|
// Returning extracted data
|
|
return [
|
|
'source' => 'Voice',
|
|
'title' => $title,
|
|
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
|
|
'image' => $crawler->filter('meta[property="og:image"]')->first()->attr('content'),
|
|
'content' => $this->content,
|
|
'url' => $url,
|
|
'date' => Carbon::now(),
|
|
'guid' => str_replace("https://voice.mv/", "", $url),
|
|
'author' => $this->author,
|
|
'topics' => $this->topics
|
|
];
|
|
}
|
|
}
|