shihaam/karudhaas
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix voice

Browse Source
This commit is contained in:
Mohamed jinas
2024-01-07 21:18:43 +05:00
parent 46ead16ef8
commit c8329962b7
4 changed files with 37 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
app/Services/Scrapers/VoiceScraper.php
View File

@@ -8,57 +8,49 @@ use Illuminate\Support\Carbon;
class VoiceScraper
{
protected $client;
protected $author;
protected $content;
protected $author = 'unknown';
protected $content = [];
protected $topics = [];
public function __construct()
{
$this->client = new Client;
$this->client = new Client();
}
public function extract($url)
{
$crawler = $this->client->request('GET', $url);
$title = $crawler->filter('.content h1')->first()->text();
// Extracting title - checking for multiple class names
$title = $crawler->filter('.text-3xl')->first()->text();
$image = $crawler->filter('.image img')->first()->attr('src');
$crawler->filter('article p')->each(function ($node) {
$this->content[] = preg_replace("/[a-zA-Z]/", "", $node->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("/", "", $node->attr('href'))
];
});
if ($crawler->filter(".authorname a")->count() == 1) {
$this->author = str_replace("- ", "", $crawler->filter('.authorname a')->first()->text());
}
//Remove all the alphabets from string
//preg_replace("/[a-zA-Z]/", "",$string);
// Returning extracted data
return [
'source' => 'Voice',
'title' => $title,
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
'image' => $image,
'content' => $this->content,
'url' => $url,
'date' => Carbon::parse(str_replace("- ", "", $crawler->filter('.authordatecomment .date')->first()->text()))->format("Y-m-d H:i:s"),
'guid' => str_replace("https://voice.mv/", "", $url),
'author' => $this->author,
'topics' => $this->topics
'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
];
}
}
}