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

114 lines
3 KiB
PHP
Raw Permalink 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\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\Http\HttpFetcher;
use App\Services\Factories\ArticleParserFactory;
use App\Services\Factories\HomepageParserFactory;
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
{
2025-07-07 00:51:32 +02:00
/**
* @return Collection<int, Article>
*/
2025-07-05 18:26:04 +02:00
public static 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 self::getArticlesFromRssFeed($feed);
} elseif ($feed->type === 'website') {
return self::getArticlesFromWebsiteFeed($feed);
}
LogSaver::warning("Unsupported feed type", null, [
'feed_id' => $feed->id,
'feed_type' => $feed->type
]);
return collect();
}
2025-07-07 00:51:32 +02:00
/**
* @return Collection<int, Article>
*/
2025-07-05 18:26:04 +02:00
private static function getArticlesFromRssFeed(Feed $feed): Collection
{
// TODO: Implement RSS feed parsing
// For now, return empty collection
return collect();
}
2025-06-29 21:20:45 +02:00
2025-07-07 00:51:32 +02:00
/**
* @return Collection<int, Article>
*/
2025-07-05 18:26:04 +02:00
private static function getArticlesFromWebsiteFeed(Feed $feed): Collection
{
try {
// Try to get parser for this feed
$parser = HomepageParserFactory::getParserForFeed($feed);
2025-06-29 21:20:45 +02:00
2025-07-05 18:26:04 +02:00
if (! $parser) {
LogSaver::warning("No parser available for feed URL", null, [
'feed_id' => $feed->id,
'feed_url' => $feed->url
]);
2025-06-29 21:20:45 +02:00
2025-07-05 18:26:04 +02:00
return collect();
2025-06-29 21:33:18 +02:00
}
2025-06-29 21:20:45 +02:00
2025-07-05 18:26:04 +02:00
$html = HttpFetcher::fetchHtml($feed->url);
$urls = $parser->extractArticleUrls($html);
return collect($urls)
->map(fn (string $url) => self::saveArticle($url, $feed->id));
2025-06-29 21:20:45 +02:00
} catch (Exception $e) {
2025-07-05 18:26:04 +02:00
LogSaver::error("Failed to fetch articles from website feed", null, [
'feed_id' => $feed->id,
'feed_url' => $feed->url,
'error' => $e->getMessage()
]);
2025-07-03 21:34:39 +02:00
2025-07-05 18:26:04 +02:00
return collect();
2025-06-29 21:20:45 +02:00
}
}
2025-07-07 00:51:32 +02:00
/**
* @return array<string, mixed>
*/
2025-06-29 21:33:18 +02:00
public static 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) {
2025-07-05 18:26:04 +02:00
LogSaver::error('Exception while fetching article data', null, [
2025-06-29 21:33:18 +02:00
'url' => $article->url,
'error' => $e->getMessage()
]);
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
}
}
2025-07-05 18:26:04 +02:00
private static function saveArticle(string $url, ?int $feedId = null): Article
2025-06-29 09:37:49 +02:00
{
2025-07-03 21:16:22 +02:00
$existingArticle = Article::where('url', $url)->first();
2025-07-03 21:34:39 +02:00
2025-07-03 21:16:22 +02:00
if ($existingArticle) {
return $existingArticle;
}
2025-07-03 21:34:39 +02:00
2025-07-05 18:26:04 +02:00
return Article::create([
'url' => $url,
'feed_id' => $feedId
]);
2025-06-29 09:37:49 +02:00
}
}