63 lines
2.0 KiB
PHP
63 lines
2.0 KiB
PHP
<?php
|
|
|
|
namespace App\Domain\Scraping\Infrastructure\Service\Scraper;
|
|
|
|
use App\Domain\Scraping\Domain\Model\ScrapingJob;
|
|
use App\Domain\Scraping\Domain\Model\ValueObject\ImageUrl;
|
|
use App\Domain\Scraping\Domain\Model\ValueObject\PageNumber;
|
|
use Symfony\Component\DomCrawler\Crawler;
|
|
|
|
class HtmlScraper extends AbstractScraper
|
|
{
|
|
protected function scrapePages(ScrapingJob $job): array
|
|
{
|
|
$url = $this->buildUrl($job);
|
|
$response = $this->httpClient->request('GET', $url);
|
|
|
|
$crawler = new Crawler($response->getContent());
|
|
$images = $crawler->filter('img.manga-page'); // Adapter selon le site
|
|
|
|
$pages = [];
|
|
$images->each(function (Crawler $image) use (&$pages) {
|
|
$pages[] = [
|
|
'url' => $image->attr('src'),
|
|
'number' => count($pages) + 1
|
|
];
|
|
});
|
|
|
|
return $pages;
|
|
}
|
|
|
|
protected function downloadPage(ScrapingJob $job, array $page, string $tempDir): void
|
|
{
|
|
$imageUrl = new ImageUrl($page['url']);
|
|
$pageNumber = new PageNumber($page['number']);
|
|
|
|
$fileName = sprintf('%s/%03d.%s',
|
|
$tempDir,
|
|
$pageNumber->getValue(),
|
|
$imageUrl->getExtension()
|
|
);
|
|
|
|
$response = $this->httpClient->request('GET', $imageUrl->getValue());
|
|
file_put_contents($fileName, $response->getContent());
|
|
|
|
$job->addPage($pageNumber, $imageUrl);
|
|
$this->dispatchProgressEvent($job, $page['number'], count($pages));
|
|
}
|
|
|
|
public function supports(string $sourceType): bool
|
|
{
|
|
return $sourceType === 'html';
|
|
}
|
|
|
|
private function buildUrl(ScrapingJob $job): string
|
|
{
|
|
// À implémenter selon votre logique de construction d'URL
|
|
// Vous aurez probablement besoin d'injecter un service pour récupérer les informations du chapitre
|
|
return sprintf('https://example.com/manga/%s/chapter/%s',
|
|
$job->getMangaId(),
|
|
$job->getChapterId()
|
|
);
|
|
}
|
|
} |