feat(scraping): implémenter le health check de tous les scrapers

- Commande CheckAllScrapersHealth + handler avec ports dédiés - Value Object ContentSourceHealthCheckData - Resource API Platform et State Processor - Adapters InMemory et tests unitaires + fonctionnels
2026-03-16 00:08:57 +01:00
parent 795cbeccc3
commit 01474c264b
11 changed files with 453 additions and 0 deletions
--- a/tests/Domain/Scraping/Adapter/InMemoryContentSourceForHealthCheckRepository.php
+++ b/tests/Domain/Scraping/Adapter/InMemoryContentSourceForHealthCheckRepository.php
@@ -0,0 +1,27 @@
+<?php
+
+namespace App\Tests\Domain\Scraping\Adapter;
+
+use App\Domain\Scraping\Domain\Contract\Repository\ContentSourceForHealthCheckInterface;
+use App\Domain\Scraping\Domain\Model\ValueObject\ContentSourceHealthCheckData;
+
+class InMemoryContentSourceForHealthCheckRepository implements ContentSourceForHealthCheckInterface
+{
+    /** @var ContentSourceHealthCheckData[] */
+    private array $sources = [];
+
+    public function add(ContentSourceHealthCheckData $data): void
+    {
+        $this->sources[] = $data;
+    }
+
+    public function getAll(): array
+    {
+        return $this->sources;
+    }
+
+    public function clear(): void
+    {
+        $this->sources = [];
+    }
+}
--- a/tests/Domain/Scraping/Adapter/InMemoryContentSourceHealthRepository.php
+++ b/tests/Domain/Scraping/Adapter/InMemoryContentSourceHealthRepository.php
@@ -0,0 +1,41 @@
+<?php
+
+namespace App\Tests\Domain\Scraping\Adapter;
+
+use App\Domain\Scraping\Domain\Contract\Repository\ContentSourceHealthRepositoryInterface;
+
+class InMemoryContentSourceHealthRepository implements ContentSourceHealthRepositoryInterface
+{
+    /** @var array<int, array{status: string, testedAt: ?\DateTimeImmutable, error: ?string}> */
+    private array $statuses = [];
+
+    public function markAsTesting(int $sourceId): void
+    {
+        $this->statuses[$sourceId] = ['status' => 'testing', 'testedAt' => null, 'error' => null];
+    }
+
+    public function markAsHealthy(int $sourceId, \DateTimeImmutable $testedAt): void
+    {
+        $this->statuses[$sourceId] = ['status' => 'ok', 'testedAt' => $testedAt, 'error' => null];
+    }
+
+    public function markAsUnhealthy(int $sourceId, \DateTimeImmutable $testedAt, string $error): void
+    {
+        $this->statuses[$sourceId] = ['status' => 'ko', 'testedAt' => $testedAt, 'error' => $error];
+    }
+
+    public function getStatus(int $sourceId): ?string
+    {
+        return $this->statuses[$sourceId]['status'] ?? null;
+    }
+
+    public function getError(int $sourceId): ?string
+    {
+        return $this->statuses[$sourceId]['error'] ?? null;
+    }
+
+    public function clear(): void
+    {
+        $this->statuses = [];
+    }
+}
--- a/tests/Domain/Scraping/Application/CommandHandler/CheckAllScrapersHealthHandlerTest.php
+++ b/tests/Domain/Scraping/Application/CommandHandler/CheckAllScrapersHealthHandlerTest.php
@@ -0,0 +1,154 @@
+<?php
+
+namespace App\Tests\Domain\Scraping\Application\CommandHandler;
+
+use App\Domain\Scraping\Application\Command\CheckAllScrapersHealth;
+use App\Domain\Scraping\Application\CommandHandler\CheckAllScrapersHealthHandler;
+use App\Domain\Scraping\Application\CommandHandler\TestScraperConfigurationHandler;
+use App\Domain\Scraping\Domain\Model\ValueObject\ContentSourceHealthCheckData;
+use App\Tests\Domain\Scraping\Adapter\InMemoryContentSourceForHealthCheckRepository;
+use App\Tests\Domain\Scraping\Adapter\InMemoryContentSourceHealthRepository;
+use App\Tests\Domain\Scraping\Adapter\InMemoryScraperAdapter;
+use App\Tests\Domain\Scraping\Adapter\InMemoryScraperFactory;
+use PHPUnit\Framework\TestCase;
+use Psr\Log\NullLogger;
+
+class CheckAllScrapersHealthHandlerTest extends TestCase
+{
+    private InMemoryContentSourceForHealthCheckRepository $sourceRepo;
+    private InMemoryContentSourceHealthRepository $healthRepo;
+    private InMemoryScraperFactory $scraperFactory;
+    private CheckAllScrapersHealthHandler $handler;
+
+    protected function setUp(): void
+    {
+        $this->sourceRepo = new InMemoryContentSourceForHealthCheckRepository();
+        $this->healthRepo = new InMemoryContentSourceHealthRepository();
+        $this->scraperFactory = new InMemoryScraperFactory();
+        $this->scraperFactory->addScraper('html', new InMemoryScraperAdapter());
+
+        $testScraperHandler = new TestScraperConfigurationHandler($this->scraperFactory);
+
+        $this->handler = new CheckAllScrapersHealthHandler(
+            $this->sourceRepo,
+            $this->healthRepo,
+            $testScraperHandler,
+            new NullLogger(),
+        );
+    }
+
+    public function testSourceWithoutTestSlugIsSkipped(): void
+    {
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 1,
+            baseUrl: 'https://example.com',
+            chapterUrlFormat: 'https://example.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: null,
+            testChapterNumber: null,
+        ));
+
+        $this->handler->handle(new CheckAllScrapersHealth());
+
+        $this->assertNull($this->healthRepo->getStatus(1));
+    }
+
+    public function testSourceWithTestSlugIsMarkedAsHealthyOnSuccess(): void
+    {
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 2,
+            baseUrl: 'https://example.com',
+            chapterUrlFormat: 'https://example.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: 'one-piece',
+            testChapterNumber: 1.0,
+        ));
+
+        $this->handler->handle(new CheckAllScrapersHealth());
+
+        $this->assertSame('ok', $this->healthRepo->getStatus(2));
+        $this->assertNull($this->healthRepo->getError(2));
+    }
+
+    public function testSourceIsMarkedAsUnhealthyWhenScraperThrows(): void
+    {
+        $failingScraper = new InMemoryScraperAdapter();
+        $failingScraper->simulateError(new \RuntimeException('Connexion refusée'));
+        $this->scraperFactory->addScraper('html', $failingScraper);
+
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 3,
+            baseUrl: 'https://example.com',
+            chapterUrlFormat: 'https://example.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: 'one-piece',
+            testChapterNumber: 1.0,
+        ));
+
+        $this->handler->handle(new CheckAllScrapersHealth());
+
+        $this->assertSame('ko', $this->healthRepo->getStatus(3));
+        $this->assertNotNull($this->healthRepo->getError(3));
+    }
+
+    public function testMultipleSourcesAreAllProcessed(): void
+    {
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 10,
+            baseUrl: 'https://siteA.com',
+            chapterUrlFormat: 'https://siteA.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: 'manga-a',
+            testChapterNumber: 1.0,
+        ));
+
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 11,
+            baseUrl: 'https://siteB.com',
+            chapterUrlFormat: 'https://siteB.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: null,
+            testChapterNumber: null,
+        ));
+
+        $this->sourceRepo->add(new ContentSourceHealthCheckData(
+            id: 12,
+            baseUrl: 'https://siteC.com',
+            chapterUrlFormat: 'https://siteC.com/{slug}/{chapterNumber}',
+            scrapingType: 'html',
+            imageSelector: 'img',
+            nextPageSelector: null,
+            chapterSelector: null,
+            testSlug: 'manga-c',
+            testChapterNumber: 3.0,
+        ));
+
+        $this->handler->handle(new CheckAllScrapersHealth());
+
+        $this->assertSame('ok', $this->healthRepo->getStatus(10));
+        $this->assertNull($this->healthRepo->getStatus(11)); // skippée
+        $this->assertSame('ok', $this->healthRepo->getStatus(12));
+    }
+
+    protected function tearDown(): void
+    {
+        $this->sourceRepo->clear();
+        $this->healthRepo->clear();
+        $this->scraperFactory->clear();
+    }
+}
--- a/tests/Feature/Scraping/CheckAllScrapersHealthTest.php
+++ b/tests/Feature/Scraping/CheckAllScrapersHealthTest.php
@@ -0,0 +1,72 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Tests\Feature\Scraping;
+
+use App\Entity\ContentSource;
+use App\Tests\Feature\AbstractApiTestCase;
+use Symfony\Component\HttpFoundation\Response;
+use Zenstruck\Foundry\Test\ResetDatabase;
+
+final class CheckAllScrapersHealthTest extends AbstractApiTestCase
+{
+    use ResetDatabase;
+
+    private function post(): void
+    {
+        static::createClient()->request('POST', '/api/scraping/check-all-health', [
+            'json' => new \stdClass(),
+        ]);
+    }
+
+    public function testItReturns202WithNoSources(): void
+    {
+        $this->post();
+
+        $this->assertResponseStatusCodeSame(Response::HTTP_ACCEPTED);
+    }
+
+    public function testItReturns202WithSourcesHavingNoTestConfig(): void
+    {
+        $source = new ContentSource();
+        $source->setBaseUrl('https://example.com')
+               ->setChapterUrlFormat('https://example.com/{slug}/{chapterNumber}')
+               ->setScrapingType('html');
+
+        $this->entityManager->persist($source);
+        $this->entityManager->flush();
+
+        $this->post();
+
+        $this->assertResponseStatusCodeSame(Response::HTTP_ACCEPTED);
+
+        // La source sans testSlug ne doit pas avoir son statut modifié
+        $this->entityManager->clear();
+        $reloaded = $this->entityManager->find(ContentSource::class, $source->getId());
+        $this->assertSame('unknown', $reloaded->getHealthStatus());
+    }
+
+    public function testHealthStatusIsUpdatedForSourcesWithTestConfig(): void
+    {
+        $source = new ContentSource();
+        $source->setBaseUrl('https://example.com')
+               ->setChapterUrlFormat('https://example.com/{slug}/{chapterNumber}')
+               ->setScrapingType('html')
+               ->setTestSlug('one-piece')
+               ->setTestChapterNumber(1.0);
+
+        $this->entityManager->persist($source);
+        $this->entityManager->flush();
+
+        $this->post();
+
+        $this->assertResponseStatusCodeSame(Response::HTTP_ACCEPTED);
+
+        // Le statut ne doit plus être 'unknown' après le test
+        $this->entityManager->clear();
+        $reloaded = $this->entityManager->find(ContentSource::class, $source->getId());
+        $this->assertNotSame('unknown', $reloaded->getHealthStatus());
+        $this->assertNotSame('testing', $reloaded->getHealthStatus()); // doit être terminé
+    }
+}