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

<?php

namespace App\Services\Article;

use App\Models\Article;
use App\Models\Feed;
use App\Services\Factories\ArticleParserFactory;
use App\Services\Factories\HomepageParserFactory;
use App\Services\Http\HttpFetcher;
use App\Services\Log\LogSaver;
use Exception;
use Illuminate\Support\Collection;

class ArticleFetcher
{
    public function __construct(
        private LogSaver $logSaver
    ) {}

    /**
     * @return Collection<int, Article>
     */
    public function getArticlesFromFeed(Feed $feed): Collection
    {
        if ($feed->type === 'rss') {
            return $this->getArticlesFromRssFeed($feed);
        } elseif ($feed->type === 'website') {
            return $this->getArticlesFromWebsiteFeed($feed);
        }

        $this->logSaver->warning('Unsupported feed type', null, [
            'feed_id' => $feed->id,
            'feed_type' => $feed->type,
        ]);

        return collect();
    }

    /**
     * @return Collection<int, Article>
     */
    private function getArticlesFromRssFeed(Feed $feed): Collection
    {
        try {
            $xml = HttpFetcher::fetchHtml($feed->url);

            $previousUseErrors = libxml_use_internal_errors(true);

            try {
                $rss = simplexml_load_string($xml);
            } finally {
                libxml_clear_errors();
                libxml_use_internal_errors($previousUseErrors);
            }

            if ($rss === false || ! isset($rss->channel->item)) {
                $this->logSaver->warning('Failed to parse RSS feed XML', null, [
                    'feed_id' => $feed->id,
                    'feed_url' => $feed->url,
                ]);

                return collect();
            }

            $articles = collect();
            foreach ($rss->channel->item as $item) {
                $link = (string) $item->link;
                if ($link !== '') {
                    $articles->push($this->saveArticle($link, $feed->id));
                }
            }

            return $articles;
        } catch (Exception $e) {
            $this->logSaver->error('Failed to fetch articles from RSS feed', null, [
                'feed_id' => $feed->id,
                'feed_url' => $feed->url,
                'error' => $e->getMessage(),
            ]);

            return collect();
        }
    }

    /**
     * @return Collection<int, Article>
     */
    private function getArticlesFromWebsiteFeed(Feed $feed): Collection
    {
        try {
            // Try to get parser for this feed
            $parser = HomepageParserFactory::getParserForFeed($feed);

            if (! $parser) {
                $this->logSaver->warning('No parser available for feed URL', null, [
                    'feed_id' => $feed->id,
                    'feed_url' => $feed->url,
                ]);

                return collect();
            }

            $html = HttpFetcher::fetchHtml($feed->url);
            $urls = $parser->extractArticleUrls($html);

            return collect($urls)
                ->map(fn (string $url) => $this->saveArticle($url, $feed->id));

        } catch (Exception $e) {
            $this->logSaver->error('Failed to fetch articles from website feed', null, [
                'feed_id' => $feed->id,
                'feed_url' => $feed->url,
                'error' => $e->getMessage(),
            ]);

            return collect();
        }
    }

    /**
     * @return array<string, mixed>
     */
    public function fetchArticleData(Article $article): array
    {
        try {
            $html = HttpFetcher::fetchHtml($article->url);
            $parser = ArticleParserFactory::getParser($article->url);

            return $parser->extractData($html);
        } catch (Exception $e) {
            $this->logSaver->error('Exception while fetching article data', null, [
                'url' => $article->url,
                'error' => $e->getMessage(),
            ]);

            return [];
        }
    }

    private function saveArticle(string $url, ?int $feedId = null): Article
    {
        $fallbackTitle = $this->generateFallbackTitle($url);

        try {
            $article = Article::firstOrCreate(
                ['url' => $url],
                [
                    'feed_id' => $feedId,
                    'title' => $fallbackTitle,
                ]
            );

            if ($article->wasRecentlyCreated) {
                $article->dispatchFetchedEvent();
            }

            return $article;
        } catch (Exception $e) {
            $this->logSaver->error('Failed to create article', null, [
                'url' => $url,
                'feed_id' => $feedId,
                'error' => $e->getMessage(),
            ]);
            throw $e;
        }
    }

    private function generateFallbackTitle(string $url): string
    {
        // Extract filename from URL as a basic fallback title
        $path = parse_url($url, PHP_URL_PATH);
        $filename = basename($path ?: $url);

        // Remove file extension and convert to readable format
        $title = preg_replace('/\.[^.]*$/', '', $filename);
        $title = str_replace(['-', '_'], ' ', $title);
        $title = ucwords($title);

        return $title ?: 'Untitled Article';
    }
}