80 lines
2.5 KiB
PHP
80 lines
2.5 KiB
PHP
<?php
|
|
|
|
namespace App\Services\Scrapers;
|
|
|
|
use Goutte\Client;
|
|
use Illuminate\Support\Carbon;
|
|
|
|
class DhiyaresScraper
|
|
{
|
|
protected $client;
|
|
|
|
protected $title;
|
|
protected $content;
|
|
protected $topics = [];
|
|
protected $author = "unknown";
|
|
protected $image;
|
|
|
|
public function __construct()
|
|
{
|
|
$this->client = new Client;
|
|
}
|
|
|
|
public function extract($url, $date)
|
|
{
|
|
|
|
$crawler = $this->client->request('GET', $url);
|
|
|
|
$ogTitle = $crawler->filter('title')->first()->text();
|
|
|
|
if (strpos($ogTitle, 'Gallery') === 0) {
|
|
return null;
|
|
}
|
|
|
|
$crawler->filter('.content .block-text')->each(function ($node) {
|
|
$this->content[] = $node->text();
|
|
});
|
|
|
|
|
|
$crawler->filter('a[class*="bg-blue-50 border border-blue-50 font-normal text-blue-400 hover:text-white hover:bg-base p-1 px-4 mx-2 mb-4 rounded-full font-mv-bold text-base"]')->each(function ($node) {
|
|
if (!preg_match('/[^A-Za-z0-9-]/', basename($node->attr('href')))) {
|
|
$this->topics[] = [
|
|
"name" => $node->text(),
|
|
"slug" => basename($node->attr('href'))
|
|
];
|
|
}
|
|
});
|
|
|
|
if ($crawler->filter("figure img")->count() > 0) {
|
|
$this->image = $crawler->filter("figure img")->first()->attr('src');
|
|
} else {
|
|
$this->image = ''; // or a default image path
|
|
}
|
|
|
|
|
|
if ($crawler->filter('div[class*="text-base font-mv-bold"] a[class*="py-4"]')->count() == 1) {
|
|
$this->author = $crawler->filter('div[class*="text-base font-mv-bold"] a[class*="py-4"]')->first()->text();
|
|
}
|
|
//Remove all the alphabets from string
|
|
//preg_replace("/[a-zA-Z]/", "",$string);
|
|
|
|
return [
|
|
'source' => 'Dhiyares',
|
|
'title' => $crawler->filter('h1')->first()->text(),
|
|
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
|
|
'image' => $this->image,
|
|
'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 ?: [
|
|
[
|
|
"name" => "ވަކި މަޢުލޫއެއް ނޭންގެ",
|
|
"slug" => "no-specific-topic"
|
|
]
|
|
]
|
|
];
|
|
}
|
|
}
|