69 lines
2.1 KiB
PHP
69 lines
2.1 KiB
PHP
<?php
|
|
|
|
namespace App\Services\Scrapers;
|
|
|
|
use Goutte\Client;
|
|
use Illuminate\Support\Carbon;
|
|
|
|
class DhenScraper
|
|
{
|
|
protected $client;
|
|
|
|
protected $content;
|
|
protected $author;
|
|
|
|
public function __construct()
|
|
{
|
|
$this->client = new Client;
|
|
}
|
|
|
|
public function extract($url)
|
|
{
|
|
|
|
$crawler = $this->client->request('GET', $url);
|
|
|
|
$title = $crawler->filter('h1')->first()->text();
|
|
|
|
if ($crawler->filter('article .article-visual img')->count() == 1) {
|
|
$image = $crawler->filter('article .article-visual img')->first()->attr('src');
|
|
}
|
|
|
|
if ($crawler->filter('article .article-entry p')->count() > 0) {
|
|
$crawler->filter('article .article-entry p')->each(function ($node) {
|
|
$this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());
|
|
});
|
|
} else {
|
|
$crawler->filter('.im_message_body .im_message_text')->each(function ($node) {
|
|
$this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());
|
|
});
|
|
}
|
|
|
|
|
|
|
|
|
|
if ($crawler->filter(".story-item--meta a")->count() == 1) {
|
|
$this->author = $crawler->filter('.story-item--meta a')->first()->text();
|
|
}
|
|
|
|
//Remove all the alphabets from string
|
|
//preg_replace("/[a-zA-Z]/", "",$string);
|
|
return [
|
|
'source' => 'Dhen',
|
|
'title' => $title,
|
|
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
|
|
'image' => $image ?? "images/noimg.jpg",
|
|
'content' => $this->content,
|
|
'url' => $url,
|
|
'date' => Carbon::parse(str_replace("- ", "", $crawler->filter('.story-item--meta time')->first()->text()))->format("Y-m-d H:i:s"),
|
|
'guid' => str_replace("https://dhen.mv/", "", $url),
|
|
'author' => $this->author,
|
|
'topics' => [
|
|
[
|
|
"name" => "ވަކި މަޢުލޫއެއް ނޭންގެ",
|
|
"slug" => "no-specific-topic"
|
|
]
|
|
]
|
|
];
|
|
}
|
|
}
|