Sun Scraper and service fix
This commit is contained in:
@@ -11,7 +11,7 @@ class SunScraper
|
||||
protected $title;
|
||||
protected $content;
|
||||
protected $topics = [];
|
||||
protected $author;
|
||||
protected $author = "unknown";
|
||||
|
||||
public function __construct()
|
||||
{
|
||||
@@ -23,25 +23,29 @@ class SunScraper
|
||||
|
||||
$crawler = $this->client->request('GET', $url);
|
||||
|
||||
$crawler->filter('h1')->each(function ($node) {
|
||||
$crawler->filter('.component-article-title h1')->each(function ($node) {
|
||||
$this->title = $node->text();
|
||||
});
|
||||
|
||||
$crawler->filter('.component-article-content p')->each(function ($node) {
|
||||
$this->content[] = preg_replace("/[a-zA-Z]/","",$node->text());
|
||||
$this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());
|
||||
});
|
||||
|
||||
$crawler->filter('.component-article-tag li a')->each(function ($node) {
|
||||
|
||||
|
||||
$this->topics[] = [
|
||||
"name" => $node->first()->text(),
|
||||
"slug" => str_replace("https://sun.mv/", "", $node->first()->attr('href'))
|
||||
];
|
||||
});
|
||||
|
||||
if($crawler->filter(".author .name")->count() == 1)
|
||||
{
|
||||
$this->author = $crawler->filter(".author .name")->first()->text();
|
||||
}
|
||||
//Remove all the alphabets from string
|
||||
//preg_replace("/[a-zA-Z]/", "",$string);
|
||||
return [
|
||||
return [
|
||||
'source' => 'Sun',
|
||||
'title' => $this->title,
|
||||
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
|
||||
@@ -49,8 +53,8 @@ class SunScraper
|
||||
'content' => $this->content,
|
||||
'url' => $url,
|
||||
'date' => $crawler->filter(".author .time")->first()->text(),
|
||||
'guid' => str_replace("https://sun.mv/","",$url),
|
||||
'author' => $crawler->filter(".author .name")->first()->text(),
|
||||
'guid' => str_replace("https://sun.mv/", "", $url),
|
||||
'author' => $this->author,
|
||||
'topics' => collect($this->topics)->unique()->values()->toArray()
|
||||
];
|
||||
}
|
||||
|
||||
@@ -0,0 +1,27 @@
|
||||
<?php
|
||||
namespace App\Services;
|
||||
|
||||
use App\Services\Feeds\SunFeed;
|
||||
use App\Services\Scrapers\SunScraper;
|
||||
|
||||
class SunService
|
||||
{
|
||||
/**
|
||||
* Scrap all the rss articles from Sun
|
||||
*
|
||||
* @return array
|
||||
*/
|
||||
public function scrape(): array
|
||||
{
|
||||
//Return only the rss that contains "news" keyboard in its url
|
||||
$articles = (new SunFeed)->get();
|
||||
|
||||
$articlesitems = [];
|
||||
//Looping through the articles and scraping and while scraping it creates a new instance of the scraper.
|
||||
foreach ($articles as $article) {
|
||||
$articlesitems[] = (new SunScraper)->extract("https://sun.mv/".$article["id"]);
|
||||
}
|
||||
|
||||
return $articlesitems;
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user