Mangarr/src/Domain/Scraping/Application/CommandHandler/ScrapeChapterHandler.php

<?php

namespace App\Domain\Scraping\Application\CommandHandler;

use App\Domain\Scraping\Application\Command\ScrapeChapter;
use App\Domain\Scraping\Domain\Contract\Repository\ChapterRepositoryInterface;
use App\Domain\Scraping\Domain\Contract\Repository\MangaRepositoryInterface;
use App\Domain\Scraping\Domain\Contract\Repository\SourceRepositoryInterface;
use App\Domain\Shared\Domain\Contract\ImageStorageInterface;
use App\Domain\Scraping\Domain\Contract\Service\ImageDownloaderInterface;
use App\Domain\Scraping\Domain\Contract\Service\ScraperFactoryInterface;
use App\Domain\Shared\Domain\Event\ChapterScraped;
use App\Domain\Scraping\Domain\Event\ChapterScrapingFailed;
use App\Domain\Scraping\Domain\Event\ChapterScrapingStarted;
use App\Domain\Scraping\Domain\Model\Chapter;
use App\Domain\Scraping\Domain\Model\Source;
use App\Domain\Scraping\Domain\Model\ValueObject\ScrapingRequest;
use App\Domain\Scraping\Domain\Model\ValueObject\TempDirectory;
use App\Domain\Shared\Domain\Contract\JobRepositoryInterface;
use Symfony\Component\Messenger\MessageBusInterface;

readonly class ScrapeChapterHandler
{
    public function __construct(
        private ScraperFactoryInterface $scraperFactory,
        private ImageDownloaderInterface $imageDownloader,
        private ImageStorageInterface $imageStorage,
        private JobRepositoryInterface $jobRepository,
        private ChapterRepositoryInterface $chapterRepository,
        private MangaRepositoryInterface $mangaRepository,
        private SourceRepositoryInterface $sourceRepository,
        private MessageBusInterface $eventBus,
    ) {
    }

    public function handle(ScrapeChapter $command): void
    {
        /** @var Chapter $chapter */
        $chapter = $this->chapterRepository->getById($command->chapterId);
        $manga = $this->mangaRepository->getById($chapter->mangaId);

        $job = $this->jobRepository->get($command->jobId);
        $job->context['chapterId'] = $command->chapterId;
        $job->context['mangaTitle'] = $manga->getTitle();
        $job->start();
        $this->jobRepository->save($job);

        $this->eventBus->dispatch(new ChapterScrapingStarted($job->id, $manga->getTitle(), $chapter->chapterNumber));

        $sources = $this->getSourcesToTry($manga);
        $slugsToTry = array_merge([$manga->getSlug()], $manga->getAlternativeSlugs());
        $success = false;
        $lastException = null;

        foreach ($sources as $source) {
            foreach ($slugsToTry as $slug) {
                try {
                    $job->context['sourceId'] = $source->getId()->getValue();
                    $job->context['slug'] = $slug;
                    $this->jobRepository->save($job);

                    $scrapingParameters = $source->getScrappingParameters();
                    $scrapingParameters['chapterNumber'] = $chapter->chapterNumber;
                    $scrapingType = $scrapingParameters['scrapingType'] ?? 'html';

                    $scrapingRequest = new ScrapingRequest(
                        $scrapingType,
                        $source->buildChapterUrl($slug, $chapter->chapterNumber),
                        $scrapingParameters
                    );

                    $scraper = $this->scraperFactory->getScraperWithFallback($scrapingType);
                    $scrapingResult = $scraper->scrape($scrapingRequest);

                    $tempDir = new TempDirectory();
                    $downloadResults = $this->imageDownloader->downloadBatch(
                        $scrapingResult->getImageUrls(),
                        $tempDir,
                        $job->id
                    );

                    $localPaths = array_map(fn ($r) => $r->getLocalPath(), $downloadResults);
                    $pagesDirectory = $this->imageStorage->storeChapterImages($command->chapterId, $localPaths);
                    $pageCount = count($downloadResults);

                    $job->complete();
                    $this->jobRepository->save($job);

                    $this->eventBus->dispatch(new ChapterScraped($job->id, $command->chapterId, $pagesDirectory, $pageCount));
                    $tempDir->cleanup();

                    $success = true;
                    break;

                } catch (\Exception $e) {
                    $lastException = $e;
                }
            }

            if ($success) {
                break;
            }
        }

        if (!$success) {
            $errorMessage = $lastException?->getMessage() ?? 'Failed to scrape chapter from all available sources';
            $job->fail($errorMessage);
            $this->jobRepository->save($job);
            $this->eventBus->dispatch(new ChapterScrapingFailed($job->id, $chapter->mangaId, $chapter->chapterNumber, $errorMessage));
        }
    }

    /**
     * @param \App\Domain\Scraping\Domain\Model\Manga $manga
     * @return Source[]
     */
    private function getSourcesToTry(\App\Domain\Scraping\Domain\Model\Manga $manga): array
    {
        if ($manga->hasPreferredSources()) {
            $preferredSources = [];
            foreach ($manga->getPreferredSources() as $sourceId) {
                $source = $this->sourceRepository->getById($sourceId);
                if ($source) {
                    $preferredSources[] = $source;
                }

                if (count($preferredSources) >= 3) {
                    break;
                }
            }

            if (!empty($preferredSources)) {
                return $preferredSources;
            }
        }

        return $this->sourceRepository->getAll();
    }
}