fedi-feed-router/app/Services/Article/ArticleFetcher.php

62 lines
1.6 KiB
PHP
Raw Normal View History

2025-06-29 09:37:49 +02:00
<?php
2025-06-29 17:46:06 +02:00
namespace App\Services\Article;
2025-06-29 09:37:49 +02:00
use App\Actions\FetchRssArticlesAction;
use App\Actions\FetchWebsiteArticlesAction;
2025-06-29 09:37:49 +02:00
use App\Models\Article;
2025-07-05 18:26:04 +02:00
use App\Models\Feed;
2025-06-29 21:33:18 +02:00
use App\Services\Factories\ArticleParserFactory;
use App\Services\Http\HttpFetcher;
2025-07-05 18:26:04 +02:00
use App\Services\Log\LogSaver;
2025-06-29 09:37:49 +02:00
use Exception;
2025-06-29 17:13:18 +02:00
use Illuminate\Support\Collection;
2025-06-29 09:37:49 +02:00
class ArticleFetcher
{
2025-08-15 02:50:42 +02:00
public function __construct(
private LogSaver $logSaver,
private FetchRssArticlesAction $fetchRssArticles,
private FetchWebsiteArticlesAction $fetchWebsiteArticles,
2025-08-15 02:50:42 +02:00
) {}
2025-07-07 00:51:32 +02:00
/**
* @return Collection<int, Article>
*/
2025-08-15 02:50:42 +02:00
public function getArticlesFromFeed(Feed $feed): Collection
2025-06-29 21:20:45 +02:00
{
2025-07-05 18:26:04 +02:00
if ($feed->type === 'rss') {
return $this->fetchRssArticles->execute($feed);
2025-07-05 18:26:04 +02:00
} elseif ($feed->type === 'website') {
return $this->fetchWebsiteArticles->execute($feed);
2025-07-05 18:26:04 +02:00
}
$this->logSaver->warning('Unsupported feed type', null, [
2025-07-05 18:26:04 +02:00
'feed_id' => $feed->id,
'feed_type' => $feed->type,
2025-07-05 18:26:04 +02:00
]);
return collect();
}
2025-07-07 00:51:32 +02:00
/**
* @return array<string, mixed>
*/
2025-08-15 02:50:42 +02:00
public function fetchArticleData(Article $article): array
2025-06-29 09:37:49 +02:00
{
try {
2025-06-29 21:33:18 +02:00
$html = HttpFetcher::fetchHtml($article->url);
$parser = ArticleParserFactory::getParser($article->url);
2025-06-29 09:37:49 +02:00
2025-06-29 21:33:18 +02:00
return $parser->extractData($html);
2025-06-29 09:37:49 +02:00
} catch (Exception $e) {
2025-08-15 02:50:42 +02:00
$this->logSaver->error('Exception while fetching article data', null, [
2025-06-29 21:33:18 +02:00
'url' => $article->url,
'error' => $e->getMessage(),
2025-06-29 21:33:18 +02:00
]);
2025-07-03 21:34:39 +02:00
2025-06-29 21:33:18 +02:00
return [];
2025-06-29 09:37:49 +02:00
}
}
}