- Layout: h-screen overflow-hidden, <main> flex-col avec mt-16
- Pages avec toolbar: toolbar hors du conteneur scrollable (flex-col + overflow-y-auto flex-1)
- Pages sans toolbar: wrapper overflow-y-auto h-full
- app.scss: scrollbar-width/color limité à Firefox via @supports (-moz-appearance: none) pour éviter le conflit avec les pseudo-éléments webkit sur Chrome 121+
- Suppression des flèches de scrollbar via ::-webkit-scrollbar-button
- html/body overflow:hidden pour éviter la double scrollbar
`fillVolumeGaps` incorrectly left chapters null when surrounded by two
different non-null volumes (e.g. Vol10 → null → Vol11). Simplify the
condition to always prefer the previous volume, covering all cases.
Also fix `InMemoryMangaRepository::findExistingChaptersByNumbers` to
return an array keyed by chapter number, matching the Doctrine contract.
Add 5 unit tests for MangadxChapterSynchronizationService covering
volume transitions, start-of-series gaps, explicit volumes, FR/EN
priority, and deduplication of existing chapters.
- Title: cascade en → fr → ja-ro → ko-ro → zh-ro → first available to avoid silently dropping mangas without English title (e.g. One Piece stored as ja-ro)
- Image: use uploads.mangadex.org CDN with .512.jpg thumbnail suffix instead of mangadex.org/covers which fails in prod
- Correction de l'affichage du texte dans le toast (suppression de w-0/truncate)
- Déplacement des toasts en bas à gauche avec animation slide depuis la gauche
- Inversion de l'ordre des éléments : bouton fermeture > texte > icône > bande couleur
- Fix timing : ChapterScrapingStarted synchrone pour notif "démarrage" avant le scraping
- Ajout make notify-test pour tester les 4 types de notifications
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add mangaTitle to ScrapingJob context at creation time
- Fix job.js constructor to map failureReason, attempts, maxAttempts, context from API
- JobItem: show readable type label, manga name, chapter number, source and attempts counter
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Docker resolves bind mounts at container start time, not dynamically when
the Deployer symlink changes. Running cache:clear before restart means
docker exec sees the old release code, causing errors on changed config.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Symfony's YAML route loader does not support the priority key (only PHP config does).
Relying on vue_app being defined first in routes.yaml to ensure it is matched
before legacy controller routes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move vue_app before controllers in routes.yaml AND keep priority:1.
Using both guarantees Symfony matches the Vue SPA catch-all first
regardless of how the router compiles equal-priority routes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Without priority:1, Symfony matched legacy controllers (e.g. app_activity at /activity)
before the vue_app catch-all on hard reload. Now vue_app matches first for all paths
except /api/* and /legacy* which still route to Symfony controllers.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The var/ directory is a persistent Docker volume. Without explicit cache:clear,
docker restart keeps serving old cached routes (e.g. / → MangaController).
New code is already visible via bind mount before restart, so docker exec works.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Route `/` now serves the Vue SPA directly (catch-all `/{req}`)
- Legacy Twig interface moved to `/legacy`
- Vue Router base changed from `/vue/` to `/`
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace the per-page API call (base64 payload) with static image URLs
served directly by Caddy from public/images/pages/{chapterId}/.
- LocalImageStorage now stores to public/images/ (was MANGA_DATA_PATH)
- LegacyChapterRepository returns /images/pages/{id}/{file} URLs,
uses getimagesize() instead of loading file content into memory
- Delete GetChapterPage query/handler/response, ChapterPageResource,
ChapterPageProvider, PageContent model
- Remove getPageContent() from ChapterRepositoryInterface
- Frontend: loadChapter() fetches chapter + all pages in parallel,
ReaderPage uses URL instead of base64 data URI, InfiniteReader drops
lazy-load observer side effect, readerStore drops loadedPages/preload
- GetChapterPagesTest: extract fixture images from CBZ at runtime,
ignore tests/Fixtures/pages/ in .gitignore
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Le domaine Scraping ne génère plus d'archives CBZ ni ne modifie les
entités du domaine Manga directement. Il scrape, stocke les images
individuellement, et émet un événement partagé.
- Suppression : CbzGeneratorInterface, CbzGenerator, CbzGenerationRequest,
CbzPath, CbzGenerationException
- Suppression : save() de ChapterRepositoryInterface (Scraping)
- Suppression : cbzPath du modèle Chapter (Scraping)
- Ajout : ImageStorageInterface + LocalImageStorage
(stockage dans {MANGA_DATA_PATH}/pages/{chapterId}/)
- ScrapeChapterHandler utilise ImageStorage au lieu du générateur CBZ
- ChapterScraped déplacé dans Domain/Shared/Domain/Event/
avec jobId, chapterId, pagesDirectory, pageCount
- Routing Messenger ajouté
- Ajout : ChapterScrapedEventListener + ChapterScrapedMessageHandler
pour mettre à jour Chapter.pagesDirectory via le Repository Manga
- LegacyChapterRepository en dual-mode :
pagesDirectory en priorité, fallback cbzPath (backward compat)
- Requêtes prev/next : filtrent pagesDirectory IS NOT NULL OR cbzPath IS NOT NULL
- ChapterContext expose pagesDirectory
- phparkitect.php : App\Domain\Shared\Domain\Event autorisé dans
les couches Application (correction violations pré-existantes
ChapterImported/VolumeImported + nouvelle ChapterScraped)
- 218/218 tests passent (+3 nouveaux)
- InMemoryImageStorage créé pour les tests unitaires
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
