91 lines
2.4 KiB
PHP
91 lines
2.4 KiB
PHP
<?php
|
|
|
|
namespace App\Services\Scrapers;
|
|
|
|
use Goutte\Client;
|
|
use Illuminate\Support\Carbon;
|
|
use Illuminate\Support\Str;
|
|
|
|
class ThiladhunScraper
|
|
{
|
|
protected $client;
|
|
|
|
protected $title;
|
|
protected $content;
|
|
protected $image;
|
|
protected $author;
|
|
protected $topics = [];
|
|
|
|
/**
|
|
* __construct.
|
|
*
|
|
* @return void
|
|
*/
|
|
public function __construct()
|
|
{
|
|
$this->client = new Client();
|
|
}
|
|
|
|
/**
|
|
* extract.
|
|
*
|
|
* @param mixed $url
|
|
* @param mixed $date
|
|
* @param mixed $guid
|
|
*
|
|
* @return array
|
|
*/
|
|
public function extract($url, $date)
|
|
{
|
|
|
|
$crawler = $this->client->request('GET', $url);
|
|
|
|
$crawler->filter('h1')->each(function ($node) {
|
|
$this->title = $node->text();
|
|
});
|
|
|
|
|
|
$crawler->filter('div.single-body.entry-content.typography-copy p')->each(function ($node) {
|
|
$this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());;
|
|
});
|
|
|
|
$crawler->filter('div[class*="entry-thumb single-entry-thumb"] img')->each(function ($node) {
|
|
$this->image = $node->attr('src');
|
|
});
|
|
|
|
$crawler->filter('.entry-tags a')->each(function ($node) {
|
|
if(!preg_match('/[^A-Za-z0-9-]/', basename($node->attr('href'))))
|
|
{
|
|
$this->topics[] = [
|
|
"name" => $node->text(),
|
|
"slug" => basename($node->attr('href'))
|
|
];
|
|
}
|
|
|
|
});
|
|
|
|
$crawler->filter('a[class*="entry-author__name"]')->each(function ($node) {
|
|
$this->author = $node->text();
|
|
});
|
|
|
|
|
|
return [
|
|
'source' => 'Thiladhun News',
|
|
'title' => $this->title,
|
|
'og_title' => str_replace(" | Thiladhun", "", $crawler->filter('title')->first()->text('content')),
|
|
'image' => $this->image,
|
|
'content' => $this->content,
|
|
'date' => Carbon::parse($date)->format("Y-m-d H:i:s"),
|
|
'url' => $url,
|
|
'author' => $this->author,
|
|
'guid' => (integer) Str::before(basename($url), '?'),
|
|
'topics' => $this->topics ? : [
|
|
[
|
|
"name" => "ވަކި މަޢުލޫއެއް ނޭންގެ",
|
|
"slug" => "no-specific-topic"
|
|
]
|
|
]
|
|
];
|
|
}
|
|
}
|