Distribution publique GHCR (miroir, aucun compte requis) + avertissement légal dans le README et le label OCI

Deux défauts enchaînés : la requête Kitsu cumulait les genres en ET
(quatre catégories → 1 seul titre, précisément le téléchargement à
l'origine des genres), et l'exclusion échouait car « Titre - Saison 1 -
E3 » ne normalisait jamais vers « Titre » canonique.

- Une requête Kitsu par genre, fusion entrelacée (chaque genre contribue)
- normalize_title retire saison/épisode en boucle avec frontières de mots
  (« - Saison 1 - E3 », « S1 E1 » → titre canonique ; « Série 9 » intact)
- test de non-régession : le possédé avec marqueurs n'est plus recommandé

Dockerfile

@@ -18,6 +18,8 @@ RUN uv sync --frozen --no-dev --no-cache
 # ---------------------------------------------------------------------------
 FROM python:3.13-slim-bookworm
 
+LABEL org.opencontainers.image.source="https://git.lanro.eu/Roman/ohm_streaming" \
+      org.opencontainers.image.description="Ohm Stream Downloader — outil technique d'automatisation, ne fournit ni n'héberge aucun contenu. L'usage relève de la seule responsabilité de l'utilisateur (droit d'auteur)."
 # ffmpeg (téléchargements HLS), ca-certificates (scraping HTTPS),
 # gosu (bascule utilisateur non-root dans l'entrypoint)
 RUN apt-get update \

README.md

@@ -3,10 +3,34 @@
 Application web **auto-hébergée** (homelab) : centre de contrôle unique pour découvrir,
 regarder et télécharger des animes et séries VOSTFR/VF.
 
+## ⚖️ Avertissement légal
+
+Ohm Stream Downloader est un **outil technique d'automatisation** : il ne
+fournit, n'héberge ni ne distribue **aucun contenu**. Il automatise uniquement
+l'accès à des flux déjà publiquement accessibles sur des sites tiers, dont
+l'auteur n'est pas responsable et qu'il n'exploite pas.
+
+L'utilisation de ce logiciel relève de la **seule responsabilité de
+l'utilisateur**, tenu de respecter la législation de son pays — notamment le
+droit d'auteur (en France : art. L.122-4 du Code de la propriété
+intellectuelle). Le téléchargement ou la diffusion d'œuvres protégées sans
+autorisation est interdit par la loi.
+
+Projet publié à des fins **éducatives et de recherche**. L'auteur décline
+toute responsabilité en cas d'usage illégal et n'encourage en aucune façon le
+piratage.
+
+> **Disclaimer (EN)**: this project is a technical automation tool. It does
+> not provide, host or distribute any content — it merely automates access to
+> streams already publicly available on third-party websites that the author
+> neither controls nor operates. Users are solely responsible for complying
+> with their local laws, including copyright. The author assumes no liability
+> for illegal use and does not endorse piracy.
+
 ## Déploiement (Docker) — recommandé
 
 Le déploiement officiel passe par Docker Compose : l'image (ffmpeg inclus) est
-hébergée sur le **registre privé du Gitea** — rien n'est publié publiquement.
+hébergée publiquement sur **GHCR** (`ghcr.io/aulyriusohm/ohm-streaming`) — aucun compte requis.
 
 ### Installation guidée (recommandée)
 
@@ -18,14 +42,13 @@ git clone https://git.lanro.eu/Roman/ohm_streaming.git && cd ohm_streaming
 Le script vérifie Docker, demande la **destination des épisodes** (dossier dédié
 recommandé — voir « Bibliothèque Plex/Sonarr » ci-dessous), le port, génère les
 secrets, branche le montage et démarre. Non interactif aussi :
-`./scripts/install.sh --dir /srv/animes --port 8777 --skip-login`.
+`./scripts/install.sh --dir /srv/animes --port 8777`.
 
 ### À la main
 
 ```bash
-# 1. Récupérer le projet puis se connecter au registre privé (compte Gitea avec accès lecture)
+# 1. Récupérer le projet
 git clone https://git.lanro.eu/Roman/ohm_streaming.git && cd ohm_streaming
-docker login git.lanro.eu
 
 # 2. Configuration locale
 cp .env.example .env
@@ -36,7 +59,33 @@ docker compose up -d
 ```
 
 Puis ouvrir http://localhost:8777 — le **premier compte créé est administrateur**.
-Les données (`data/`, `downloads/`) sont montées en volumes : elles survivent aux
+Les données (`data/`, `downloads/`) sont montées en volumes : elles survivent aux mises à jour.
+
+### Variante Podman
+
+Même déploiement via `docker-compose.podman.yml` (socket Podman pour Watchtower,
+au lieu du socket Docker) :
+
+```bash
+# 0. Socket Podman (une seule fois) — rootless :
+systemctl --user enable --now podman.socket
+sudo loginctl enable-linger $USER      # persister après déconnexion
+export PODMAN_SOCKET=/run/user/$(id -u)/podman/podman.sock
+#    rootful : sudo systemctl enable --now podman.socket  (aucun export nécessaire)
+
+# 1. Récupérer le projet
+git clone https://git.lanro.eu/Roman/ohm_streaming.git && cd ohm_streaming
+
+# 2. Configuration locale (identique à Docker)
+cp .env.example .env   # → OHM_SECRET_KEY et WATCHTOWER_TOKEN obligatoires
+
+# 3. Démarrage
+podman compose -f docker-compose.podman.yml up -d
+#    (« podman compose » délègue à docker-compose ; sinon : podman-compose)
+```
+
+La mise à jour depuis la page Admin (bouton Watchtower) fonctionne à l'identique ;
+à la main : `podman compose -f docker-compose.podman.yml pull && podman compose -f docker-compose.podman.yml up -d`.
 
 
 ### Bibliothèque Plex / Sonarr existante
@@ -111,24 +160,30 @@ Variables d'environnement (préfixe `OHM_`, voir `.env.example`) :
 
 ## Fonctionnalités
 
-- **Recherche unifiée** sur plusieurs sources (Vostfree, French-Manga) — chaque source
+- **Recherche unifiée** sur plusieurs sources (Vostfree, French-Manga, VoirAnime pour les
+  animes ; French-Stream pour les séries et films VF/VOSTFR) — chaque source
   est un module interchangeable activable/désactivable à chaud (page Admin), dont l'URL
   est modifiable à la volée (utile si un site change de domaine).
 - **Extraction en 2 niveaux** : page d'épisode → lecteurs embarqués → URL directe
-  (Sibnet, SendVid, VidMoly, Uqload, Vidzy, Luluvdo).
+  (Sibnet, SendVid, VidMoly, Uqload, Vidzy, Luluvdo ; VoirAnime résout via son
+  endpoint « prepare » : MP4 direct ou HLS relayé par le proxy du site).
+- **Préférence de contenus par compte** : sélecteur topbar ⛨ Animés / 📺 Séries /
+  ✨ Les deux — la recherche et les nouveautés filtrent par type de média, et les
+  sources hors périmètre ne sont même pas interrogées.
 - **Proxy vidéo intégré** (`/api/proxy`) : contourne les protections (tokens liés à l'IP, Referer/UA obligatoires), réécrit les playlists HLS.
-- **Streaming HLS** via hls.js ; **téléchargement HLS** via ffmpeg (remux mp4, progression temps réel).
-- **Métadonnées enrichies** via Kitsu (synopsis, genres, note, images) avec cache 72 h.
 - **Téléchargements** : file parallèle, pause/reprise (HTTP Range), retry, anti-doublons,
   progression en temps réel (SSE), persistance au redémarrage.
 - **Bibliothèque locale** : streaming avec range requests, reprise de lecture,
   navigation épisode suivant/précédent.
 - **Comptes** : JWT court + refresh token (rotation), rôles admin/utilisateur,
   administration des comptes.
-- **Découverte** (`/discover`) : 🆕 nouveautés fusionnées de toutes les sources, triées
-  par date de sortie réelle (enrichissement Kitsu, badge « en cours de diffusion »),
-  🔥 incontournables (top popularité Kitsu) et ✨ recommandations par genres déduites
-  des téléchargements et favoris, titres déjà possédés exclus (cache mémoire).
+- **Découverte** (`/discover`) : 🎭 exploration par genre (animés via Kitsu, séries/films
+  via French-Stream — état dans l'URL, partageable), 🆕 nouveautés en rails séparés
+  (animés triés par date de sortie via Kitsu ; séries & films de French-Stream), 🔥
+  incontournables animés (top popularité Kitsu) et ✨ recommandations par genres déduites
+  des téléchargements et favoris, titres déjà possédés exclus (cache mémoire). Sans
+  Sonarr, les genres des séries/films téléchargés sont lus sur leur fiche source ;
+  sans aucun historique, une carte d'amorçage invite à télécharger un premier titre.
 
 
 ## Intégration Sonarr / Prowlarr (*arr)
@@ -198,7 +253,7 @@ app/
 ├── scrapers/
 │   ├── base.py          # contrats SourceScraper/HosterExtractor + registres
 │   ├── configs/         # sélecteurs YAML externalisés (réparer sans coder)
-│   ├── sources/         # vostfree, french_manga
+│   ├── sources/         # vostfree, french_manga, voiranime, french_stream
 ├── services/            # downloads, kitsu, discover, sonarr, torznab, settings
 └── templates/ + static/ # UI htmx + Alpine.js, thème sombre
 ```
@@ -209,6 +264,6 @@ app/
 ## Tests
 
 ```bash
-uv run pytest        # 86 tests
+uv run pytest        # 132 tests
 uv run ruff check .  # lint
 ```

app/auth.py

@@ -25,6 +25,10 @@ class User:
     username: str
     is_admin: bool
     is_active: bool
+    content_preference: str = "both"  # anime | serie | both
+
+
+VALID_CONTENT_PREFERENCES = ("anime", "serie", "both")
 
 
 # ---------------------------------------------------------------- mots de passe
@@ -56,7 +60,8 @@ async def create_user(username: str, password: str) -> User:
 
 async def authenticate(username: str, password: str) -> User | None:
     row = await db.fetchone(
-        "SELECT id, username, password_hash, is_admin, is_active FROM users WHERE username = ?",
+        "SELECT id, username, password_hash, is_admin, is_active, content_preference "
+        "FROM users WHERE username = ?",
         (username.strip(),),
     )
     if row is None or not verify_password(password, row["password_hash"]):
@@ -66,18 +71,35 @@ async def authenticate(username: str, password: str) -> User | None:
         logger.warning("Compte désactivé : %r", username)
         return None
     return User(
-        id=row["id"], username=row["username"], is_admin=bool(row["is_admin"]), is_active=True
+        id=row["id"],
+        username=row["username"],
+        is_admin=bool(row["is_admin"]),
+        is_active=True,
+        content_preference=row["content_preference"],
     )
 
 
 async def get_user(user_id: int) -> User | None:
     row = await db.fetchone(
-        "SELECT id, username, is_admin, is_active FROM users WHERE id = ?", (user_id,)
+        "SELECT id, username, is_admin, is_active, content_preference FROM users WHERE id = ?",
+        (user_id,),
     )
     if row is None or not row["is_active"]:
         return None
     return User(
-        id=row["id"], username=row["username"], is_admin=bool(row["is_admin"]), is_active=True
+        id=row["id"],
+        username=row["username"],
+        is_admin=bool(row["is_admin"]),
+        is_active=True,
+        content_preference=row["content_preference"],
+    )
+
+
+async def set_content_preference(user_id: int, preference: str) -> None:
+    if preference not in VALID_CONTENT_PREFERENCES:
+        raise ValueError(f"Préférence invalide : {preference!r}")
+    await db.execute(
+        "UPDATE users SET content_preference = ? WHERE id = ?", (preference, user_id)
     )
 
 

app/config.py

@@ -34,7 +34,10 @@ class Settings(BaseSettings):
     kitsu_base_url: str = "https://kitsu.io/api/edge"
     metadata_cache_ttl_hours: int = 72
 
-    # Mise à jour (déploiement Docker — Watchtower compagnon)
+    # Mise à jour (dépôt public — lecture anonyme de l'API Gitea)
+    gitea_url: str = "https://git.lanro.eu"
+    gitea_repo: str = "Roman/ohm_streaming"
+    # Déploiement Docker — Watchtower compagnon
     watchtower_url: str = ""
     watchtower_token: str = ""
 

app/db.py

@@ -11,6 +11,8 @@ CREATE TABLE IF NOT EXISTS users (
     password_hash TEXT NOT NULL,
     is_admin INTEGER NOT NULL DEFAULT 0,
     is_active INTEGER NOT NULL DEFAULT 1,
+    content_preference TEXT NOT NULL DEFAULT 'both'
+        CHECK (content_preference IN ('anime','serie','both')),
     created_at TEXT NOT NULL DEFAULT (datetime('now'))
 );
 
@@ -89,8 +91,18 @@ class Database:
         await self._conn.execute("PRAGMA journal_mode=WAL")
         await self._conn.execute("PRAGMA foreign_keys=ON")
         await self._conn.executescript(SCHEMA)
+        await self._migrate()
         await self._conn.commit()
 
+    async def _migrate(self) -> None:
+        """Migrations légères : colonnes ajoutées après coup pour les bases existantes."""
+        cursor = await self._conn.execute("PRAGMA table_info(users)")
+        columns = {row["name"] for row in await cursor.fetchall()}
+        if "content_preference" not in columns:
+            await self._conn.execute(
+                "ALTER TABLE users ADD COLUMN content_preference TEXT NOT NULL DEFAULT 'both'"
+            )
+
     async def close(self) -> None:
         if self._conn is not None:
             await self._conn.close()

app/main.py

@@ -45,7 +45,7 @@ async def lifespan(app: FastAPI) -> AsyncIterator[None]:
     await db.connect()
     await apply_source_base_urls()
     await download_manager.start()
-    warmup = asyncio.create_task(discover_service.latest())
+    warmup = asyncio.create_task(discover_service.latest_by_type())
     warmup.add_done_callback(_log_task_error)
     logger.info("%s prêt", settings.app_name)
     yield

app/routers/admin.py

@@ -28,7 +28,6 @@ from app.services.settings import (
     set_source_base_url,
     set_source_enabled,
     set_source_health,
-    set_update_config,
 )
 from app.services.sonarr import sonarr
 from app.services.update import UpdateError, fetch_latest_version, trigger_update
@@ -222,23 +221,12 @@ async def test_sonarr(admin: AdminUser) -> dict:
 # ---------------------------------------------------------------- mise à jour logicielle
 
 
-class UpdateConfig(BaseModel):
-    gitea_url: str
-    repo: str
-    token: str
-
-
 @router.get("/update")
 async def get_update(admin: AdminUser) -> dict:
-    """Version courante, dernière version disponible et configuration Gitea."""
+    """Version courante, dernière version disponible et patchnote."""
     return await update_status()
 
 
-@router.put("/update")
-async def save_update(payload: UpdateConfig, admin: AdminUser) -> dict:
-    await set_update_config(payload.gitea_url, payload.repo, payload.token)
-    return await update_status()
-
 @router.post("/update/check")
 async def check_update(admin: AdminUser) -> dict:
     """Force la re-vérification de la dernière version (ignore le cache)."""

app/routers/auth.py

@@ -3,7 +3,7 @@
 import logging
 from typing import Annotated
 
-from fastapi import APIRouter, Depends, Form, HTTPException, Request, Response, status
+from fastapi import APIRouter, Body, Depends, Form, HTTPException, Request, Response, status
 from fastapi.responses import RedirectResponse
 
 from app import auth
@@ -107,4 +107,22 @@ async def logout(user: CurrentUser) -> RedirectResponse:
 
 @router.get("/me")
 async def me(user: CurrentUser) -> dict:
-    return {"id": user.id, "username": user.username, "is_admin": user.is_admin}
+    return {
+        "id": user.id,
+        "username": user.username,
+        "is_admin": user.is_admin,
+        "content_preference": user.content_preference,
+    }
+
+
+@router.put("/preferences")
+async def update_preferences(
+    user: CurrentUser,
+    content_preference: Annotated[str, Body(embed=True)],
+) -> dict:
+    """Préférence de contenus : animés, séries, ou les deux."""
+    try:
+        await auth.set_content_preference(user.id, content_preference)
+    except ValueError as exc:
+        raise HTTPException(422, detail=str(exc)) from exc
+    return {"content_preference": content_preference}

app/routers/discover.py

@@ -1,4 +1,4 @@
-"""Découverte : nouveautés des sources, incontournables, recommandations."""
+"""Découverte : nouveautés par type, incontournables, recommandations, exploration."""
 
 import logging
 from typing import Annotated
@@ -6,7 +6,9 @@ from typing import Annotated
 from fastapi import APIRouter, Depends, Query
 
 from app.routers.auth import CurrentUser, current_user
-from app.services.discover import discover
+from app.routers.search import _allowed_media_types
+from app.scrapers.base import ScrapeError, get_source
+from app.services.discover import ANIME_GENRES, discover
 
 logger = logging.getLogger(__name__)
 
@@ -20,8 +22,61 @@ async def get_discover(
     must_watch_limit: Annotated[int, Query(ge=1, le=20)] = 20,
     for_you_limit: Annotated[int, Query(ge=1, le=20)] = 20,
 ) -> dict:
-    """Les trois sections de découverte en une requête (sections vides si source KO)."""
+    """Rails de découverte en une requête (sections vides si source KO).
+
+    « latest_anime » et « latest_serie » (séries + films réels) sont des rails
+    indépendants, filtrés par la préférence de contenu du compte.
+    """
+    allowed = _allowed_media_types(user.content_preference)
+    rails = await discover.latest_by_type(latest_limit)
+    latest_anime = rails["anime"] if "anime" in allowed else []
+    latest_serie = (rails["serie"] + rails["film"]) if {"serie", "film"} & allowed else []
+    # Incontournables et Pour toi sont issus de Kitsu (catalogue animés) :
+    # en mode séries, elles n'ont pas de sens — on ne les calcule même pas.
+    if "anime" in allowed:
         must_watch = await discover.must_watch(must_watch_limit)
         for_you = await discover.for_you(user.id, for_you_limit)
-    latest = await discover.latest(latest_limit)
-    return {"latest": latest, "must_watch": must_watch, "for_you": for_you}
+    else:
+        must_watch = []
+        for_you = {"based_on": [], "items": []}
+    return {
+        "latest_anime": latest_anime,
+        "latest_serie": latest_serie,
+        "must_watch": must_watch,
+        "for_you": for_you,
+    }
+
+
+@router.get("/discover/genres")
+async def get_genres(user: CurrentUser) -> dict:
+    """Catalogue des genres parcourables (page Explorer), selon la préférence."""
+    allowed = _allowed_media_types(user.content_preference)
+    catalog: dict[str, list[dict]] = {}
+    if "anime" in allowed:
+        catalog["anime"] = [{"key": key, "label": label} for key, label in ANIME_GENRES.items()]
+    try:
+        french_stream = get_source("french_stream")
+        available = french_stream.browse_catalog()
+    except ScrapeError:
+        available = {}
+    for kind in ("serie", "film"):
+        if kind in allowed and kind in available:
+            catalog[kind] = [
+                {"key": key, "label": label} for key, label in available[kind].items()
+            ]
+    return catalog
+
+
+@router.get("/discover/browse")
+async def get_browse(
+    user: CurrentUser,
+    type: Annotated[str, Query(pattern="^(anime|serie|film)$")],
+    genre: Annotated[str, Query(min_length=1, max_length=40)],
+    limit: Annotated[int, Query(ge=1, le=40)] = 24,
+) -> dict:
+    """Titres d'un genre : animés via Kitsu, séries/films via French-Stream."""
+    if type not in _allowed_media_types(user.content_preference):
+        return {"items": []}  # hors préférence du compte
+    if type == "anime":
+        return {"items": await discover.browse_anime(genre, limit)}
+    return {"items": await discover.browse_serie_film(type, genre, limit)}

app/routers/pages.py

@@ -12,6 +12,13 @@ protected = APIRouter(tags=["pages"], include_in_schema=False, dependencies=[Dep
 
 templates = Jinja2Templates(directory=BASE_DIR / "app" / "templates")
 
+# Cache-busting des assets : version dérivée de la date des fichiers statiques —
+# tout changement de CSS/JS invalide le cache navigateur sans intervention.
+_static_root = BASE_DIR / "app" / "static"
+templates.env.globals["asset_v"] = str(
+    max(int(p.stat().st_mtime) for p in _static_root.rglob("*") if p.is_file())
+)
+
 
 @router.get("/login", response_class=HTMLResponse)
 async def login_page(request: Request) -> HTMLResponse:

app/routers/search.py

@@ -7,7 +7,7 @@ from typing import Annotated
 
 from fastapi import APIRouter, Depends, HTTPException, Query
 
-from app.routers.auth import current_user
+from app.routers.auth import CurrentUser, current_user
 from app.scrapers.base import (
     ScrapeError,
     SourceScraper,
@@ -27,13 +27,35 @@ router = APIRouter(prefix="/api", tags=["search"], dependencies=[Depends(current
 import_all_scrapers()
 kitsu = KitsuService()
 
+# Types de médias gardés selon la préférence. Les films réels (French-Stream)
+# accompagnent les séries — le mode animés reste sur l'animation.
+_PREFERENCE_MEDIA_TYPES = {
+    "anime": {"anime"},
+    "serie": {"serie", "film"},
+    "both": {"anime", "serie", "film"},
+}
+
+
+def _allowed_media_types(preference: str) -> set[str]:
+    return _PREFERENCE_MEDIA_TYPES.get(preference, _PREFERENCE_MEDIA_TYPES["both"])
+
+
+async def enabled_sources(allowed: set[str] | None = None) -> list[SourceScraper]:
+    """Sources activées, limitées à celles pouvant servir les types de médias autorisés."""
+    sources = []
+    for source in all_sources():
+        if not await is_source_enabled(source.name):
+            continue
+        if allowed is not None and not set(source.media_types) & allowed:
+            continue
+        sources.append(source)
+    return sources
+
 
-async def enabled_sources() -> list[SourceScraper]:
     sources = []
     for source in all_sources():
         if await is_source_enabled(source.name):
             sources.append(source)
-    return sources
 
 
 @router.get("/sources")
@@ -52,9 +74,10 @@ async def list_sources() -> list[dict]:
 
 
 @router.get("/search")
-async def search(q: Annotated[str, Query(min_length=2)]) -> dict:
+async def search(q: Annotated[str, Query(min_length=2)], user: CurrentUser) -> dict:
     """Recherche unifiée : une requête interroge toutes les sources activées."""
-    sources = await enabled_sources()
+    allowed = _allowed_media_types(user.content_preference)
+    sources = await enabled_sources(allowed)
 
     async def safe_search(source: SourceScraper) -> tuple[list, str | None]:
         try:
@@ -66,6 +89,7 @@ async def search(q: Annotated[str, Query(min_length=2)]) -> dict:
 
     outcomes = await asyncio.gather(*(safe_search(s) for s in sources))
     results = [item for items, _ in outcomes for item in items]
+    results = [r for r in results if r.get("media_type", "anime") in allowed]
     failed = [name for _, name in outcomes if name]
     return {"query": q, "count": len(results), "results": results, "failed_sources": failed}
 

app/scrapers/configs/french_stream.yaml

@@ -0,0 +1,60 @@
+# Sélecteurs/endpoints French-Stream (french-stream.lat, DataLife Engine) — surchargeables
+# sans toucher au code. Structure vérifiée en live (films & séries VF/VOSTFR).
+
+endpoints:
+  search: "/engine/ajax/search.php"        # POST query=<q>&page=1
+  episodes: "/ep-data.php?id={newsid}&format=js"  # JSON {vf, vostfr, vo, info}
+  film: "/engine/ajax/film_api.php?id={newsid}"   # JSON {players: {hoster: {version: url}}}
+
+search:
+  item: "div.search-item"          # bloc résultat (lien dans onclick="location.href='...'")
+  title: ".search-title"
+  poster: ".search-poster img"
+
+details:
+  title: "h1#s-title"
+  synopsis: "div.fdesc"            # le boilerplate p.desc-text est retiré
+  synopsis_boilerplate: "p.desc-text"
+  genres: ".facts .genres"
+  year: ".facts .release"
+  poster_serie: ".fposter img"
+  poster_film: "#film-data"        # attribut data-affiche
+  serie_marker: "#serie-config"    # présent = fiche série (sinon film)
+
+latest:
+  path: "/series/"                 # mix films/séries ; « Saison » dans le titre = série
+  item: "div.short"
+  link: "a.short-poster"           # href = fiche (/index.php?newsid=N), alt = titre
+  image: "img"
+
+browse: # parcours par genre (page Explorer) — chemins vérifiés en live
+  serie: # pages /<genre>-series-/ (9 genres exposés par le site)
+    aventure:        {path: "/aventure-series-/",          label: "Aventure"}
+    familles:        {path: "/familles-series-/",          label: "Famille"}
+    fantastique:     {path: "/fantastique-series-/",       label: "Fantastique"}
+    judiciaire:      {path: "/judiciare-series-/",         label: "Judiciaire"}   # coquille du site
+    medical:         {path: "/medical-series-/",           label: "Médical"}
+    romance:         {path: "/romance-series-/",           label: "Romance"}
+    science-fiction: {path: "/science-fiction-series-/",   label: "Science-Fiction"}
+    thriller:        {path: "/thriller-series-/",          label: "Thriller"}
+    western:         {path: "/western-series-/",           label: "Western"}
+  film: # pages /films/<genre>/
+    actions:          {path: "/films/actions/",            label: "Action"}
+    animations:       {path: "/films/animations/",         label: "Animation"}
+    aventures:        {path: "/films/aventures/",          label: "Aventure"}
+    biopics:          {path: "/films/biopics/",            label: "Biopic"}
+    comedies:         {path: "/films/comedies/",           label: "Comédie"}
+    cultes:           {path: "/films/cultes/",             label: "Culte"}
+    documentaires:    {path: "/films/documentaires/",      label: "Documentaire"}
+    drames:           {path: "/films/drames/",             label: "Drame"}
+    epouvante-horreurs: {path: "/films/epouvante-horreurs/", label: "Épouvante-Horreur"}
+    espionnages:      {path: "/films/espionnages/",        label: "Espionnage"}
+    familles:         {path: "/films/familles/",           label: "Famille"}
+    fantastiques:     {path: "/films/fantastiques/",       label: "Fantastique"}
+    guerres:          {path: "/films/guerres/",            label: "Guerre"}
+    historiques:      {path: "/films/historiques/",        label: "Historique"}
+    policiers:        {path: "/films/policiers/",          label: "Policier"}
+    romances:         {path: "/films/romances/",           label: "Romance"}
+    science-fictions: {path: "/films/science-fictions/",   label: "Science-Fiction"}
+    thrillers:        {path: "/films/thrillers/",          label: "Thriller"}
+    westerns:         {path: "/films/westerns/",           label: "Western"}

app/scrapers/hosters/uqload.py

@@ -1,4 +1,4 @@
-"""Extracteur Uqload (uqload.to/.co/.com/.io) — page embed → mp4/m3u8.
+"""Extracteur Uqload (uqload.to/.co/.com/.io/.vc) — page embed → mp4/m3u8.
 
 La page embed contient un jwplayer configuré dans du JS packé
 (p,a,c,k,e,d) : `sources:[{file:"https://.../master.m3u8?..."}]`.
@@ -24,7 +24,7 @@ _PATTERNS = (
 @register_hoster
 class UqloadExtractor(HosterExtractor):
     name = "uqload"
-    domains = ("uqload.to", "uqload.co", "uqload.com", "uqload.io")
+    domains = ("uqload.to", "uqload.co", "uqload.com", "uqload.io", "uqload.vc")
 
     async def extract(self, embed_url: str) -> VideoLink:
         html = await fetch(embed_url)

app/scrapers/hosters/vidzy.py

@@ -1,4 +1,4 @@
-"""Extracteur Vidzy (vidzy.org) — page embed videojs → m3u8.
+"""Extracteur Vidzy (vidzy.org/.cc/.live) — page embed videojs → m3u8.
 
 La page embed ne contient pas l'URL vidéo en clair : le script videojs appelle
 une fonction de décodage inline `atob(s)` + reverse + XOR, avec une graine
@@ -45,7 +45,7 @@ def _decode(b64: str, hostname: str) -> str | None:
 @register_hoster
 class VidzyExtractor(HosterExtractor):
     name = "vidzy"
-    domains = ("vidzy.org",)
+    domains = ("vidzy.org", "vidzy.cc", "vidzy.live")
 
     async def extract(self, embed_url: str) -> VideoLink:
         html = await fetch(embed_url)

app/scrapers/hosters/voiranime.py

@@ -0,0 +1,111 @@
+"""Extracteur VoirAnime — endpoint « prepare » → URL lisible (directe ou proxy site).
+
+Chaque source d'un épisode voiranime.xyz se résout via
+`/lecteur/prepare/<source_id>?content=..&episode=..` (AJAX : en-têtes Accept JSON et
+X-Requested-With obligatoires) qui renvoie :
+`{"success": true, "media_type": "mp4"|"hls", "stream_url": "/proxy/media?payload=<b64>"}`.
+
+Le payload base64 est un JSON `{"url": ..., "referer": ..., "kind": ...}` où `url` est
+l'URL directe chez l'hébergeur d'origine. Deux cas (vérifiés en live) :
+- mp4 : l'URL d'origine est libre d'accès avec son Referer (Sibnet…) → on la retourne
+  directement (téléchargement/Range natifs, sans double saut).
+- hls : les CDN utilisés (SmoothPre & co) signent leurs playlists pour le backend du
+  site — accès direct 403. Le proxy `/proxy/media?payload=…` du site, lui, sert la
+  playlist ET ses segments (chemins re-hostés) → on retourne l'URL proxy absolue.
+"""
+
+import base64
+import binascii
+import json
+import logging
+from urllib.parse import parse_qs, urlparse
+
+from app.scrapers.base import HosterExtractor, ScrapeError, VideoLink, register_hoster
+from app.scrapers.http import fetch
+
+logger = logging.getLogger(__name__)
+
+
+@register_hoster
+class VoirAnimeExtractor(HosterExtractor):
+    name = "voiranime"
+    domains = ("voiranime.xyz",)
+
+    async def extract(self, embed_url: str) -> VideoLink:
+        origin = self._origin_of(embed_url)
+        raw = await fetch(
+            embed_url,
+            referer=self._referer_for(embed_url),
+            headers={
+                "Accept": "application/json",
+                "X-Requested-With": "XMLHttpRequest",
+            },
+        )
+        try:
+            data = json.loads(raw)
+        except json.JSONDecodeError as exc:
+            raise ScrapeError(f"voiranime : réponse prepare invalide : {raw[:120]}") from exc
+        if not data.get("success"):
+            raise ScrapeError(f"voiranime : {data.get('error') or 'échec de préparation du lecteur'}")
+
+        media_type = str(data.get("media_type") or "").lower()
+        stream_url = str(data.get("stream_url") or "")
+
+        if media_type == "hls":
+            if not stream_url:
+                raise ScrapeError(f"voiranime : flux HLS sans stream_url dans {embed_url}")
+            return VideoLink(
+                url=f"{origin}/{stream_url.lstrip('/')}",
+                hoster=self.name,
+                headers={"Referer": f"{origin}/"},
+                is_hls=True,
+            )
+
+        payload = self._extract_payload(stream_url)
+        url = str(payload.get("url") or "")
+        if not url:
+            raise ScrapeError(f"voiranime : payload sans URL de flux dans {embed_url}")
+        referer = str(payload.get("referer") or "")
+        return VideoLink(
+            url=url,
+            hoster=self.name,
+            headers={"Referer": referer} if referer else {},
+            is_hls=url.split("?")[0].endswith(".m3u8"),
+        )
+
+    @staticmethod
+    def _origin_of(embed_url: str) -> str:
+        parsed = urlparse(embed_url)
+        return f"{parsed.scheme}://{parsed.netloc}"
+
+    @staticmethod
+    def _referer_for(embed_url: str) -> str:
+        """Reconstruit la page lecteur d'où provient la requête prepare."""
+        origin = VoirAnimeExtractor._origin_of(embed_url)
+        query = parse_qs(urlparse(embed_url).query)
+        content = query.get("content", [""])[0]
+        episode = query.get("episode", [""])[0]
+        if content and episode:
+            return f"{origin}/lecteur/{content}/{episode}"
+        return f"{origin}/"
+
+    @staticmethod
+    def _extract_payload(stream_url: str) -> dict:
+        payload_b64 = parse_qs(urlparse(stream_url).query).get("payload", [""])[0]
+        if not payload_b64:
+            raise ScrapeError(f"voiranime : pas de payload dans stream_url ({stream_url[:80]})")
+        padded = payload_b64 + "=" * (-len(payload_b64) % 4)
+        try:
+            decoded = base64.b64decode(padded)
+        except (binascii.Error, ValueError):
+            try:
+                decoded = base64.urlsafe_b64decode(padded)
+            except (binascii.Error, ValueError) as exc:
+                raise ScrapeError(f"voiranime : payload illisible : {payload_b64[:60]}") from exc
+        try:
+            data = json.loads(decoded)
+        except (json.JSONDecodeError, UnicodeDecodeError) as exc:
+            raise ScrapeError(f"voiranime : payload non-JSON : {payload_b64[:60]}") from exc
+        if not isinstance(data, dict):
+            raise ScrapeError("voiranime : payload inattendu (pas un objet JSON)")
+        return data

app/scrapers/http.py

@@ -41,15 +41,30 @@ async def fetch(
     *,
     referer: str | None = None,
     retries: int = 2,
+    headers: dict[str, str] | None = None,
+    data: dict[str, str] | None = None,
 ) -> str:
-    """GET d'une page avec retries ; lève ScrapeError en cas d'échec définitif."""
-    headers = {"Referer": referer} if referer else {}
+    """GET (ou POST si `data` est fourni) avec retries ; lève ScrapeError en cas d'échec définitif."""
+    request_headers = dict(headers) if headers else {}
+    if referer:
+        request_headers.setdefault("Referer", referer)
     last_error: Exception | None = None
     for attempt in range(retries + 1):
         try:
-            response = await get_client().get(url, headers=headers)
+            if data is None:
+                response = await get_client().get(url, headers=request_headers)
+            else:
+                response = await get_client().post(url, headers=request_headers, data=data)
             response.raise_for_status()
             return response.text
+        except httpx.HTTPStatusError as exc:
+            status = exc.response.status_code
+            if 400 <= status < 500 and status not in (408, 429):
+                raise ScrapeError(f"Échec de récupération de {url} : HTTP {status} (définitif)") from exc
+            last_error = exc
+            logger.warning("fetch %s — HTTP %d, tentative %d/%d", url, status, attempt + 1, retries + 1)
+            if attempt < retries:
+                await asyncio.sleep(1.0 * (attempt + 1))
         except (httpx.HTTPError, httpx.InvalidURL) as exc:
             last_error = exc
             logger.warning("fetch %s — tentative %d/%d : %s", url, attempt + 1, retries + 1, exc)

app/scrapers/sources/french_stream.py

@@ -0,0 +1,429 @@
+"""Source French-Stream (french-stream.lat) — films & séries VF/VOSTFR, moteur DataLife Engine.
+
+Faits structurels (vérifiés en live) :
+- Recherche : POST /engine/ajax/search.php (query, page) → blocs `div.search-item`
+  (lien dans onclick="location.href='...'", poster `.search-poster img`).
+- Fiche : `/index.php?newsid=<id>` (ou la jolie URL `/<id>-<slug>.html`).
+  `source_id` = newsid numérique.
+- Séries : une fiche par saison (« Titre - Saison N »). Épisodes via
+  GET /ep-data.php?id=<newsid>&format=js → JSON
+  `{"vf": {"1": {"vidzy": url, "uqload": url, ...}}, "vostfr": {...}, "info": {...}}`.
+- Films : lecteurs via GET /engine/ajax/film_api.php?id=<newsid> → JSON
+  `{"players": {"vidzy": {"default": url, "vff": url, "vostfr": url, ...}}}`.
+- URL d'épisode interne : `<fiche>#vf-3` (série) ou `<fiche>#film` (film).
+- Nouveautés : /series/ → blocs `div.short` (mix films/séries, « Saison » dans le
+  titre = série).
+"""
+
+import json
+import logging
+import re
+from copy import deepcopy
+
+from bs4 import BeautifulSoup, Tag
+
+from app.scrapers.base import (
+    Episode,
+    ScrapeError,
+    SearchResult,
+    SourceScraper,
+    TitleDetails,
+    register_source,
+)
+from app.scrapers.config_loader import load_scraper_config
+from app.scrapers.http import fetch, fetch_soup
+
+logger = logging.getLogger(__name__)
+
+DEFAULT_CONFIG: dict = {
+    "endpoints": {
+        "search": "/engine/ajax/search.php",
+        "episodes": "/ep-data.php?id={newsid}&format=js",
+        "film": "/engine/ajax/film_api.php?id={newsid}",
+    },
+    "search": {
+        "item": "div.search-item",
+        "title": ".search-title",
+        "poster": ".search-poster img",
+    },
+    "details": {
+        "title": "h1#s-title",
+        "synopsis": "div.fdesc",
+        "synopsis_boilerplate": "p.desc-text",
+        "genres": ".facts .genres",
+        "year": ".facts .release",
+        "poster_serie": ".fposter img",
+        "poster_film": "#film-data",
+        "serie_marker": "#serie-config",
+    },
+    "latest": {
+        "path": "/series/",
+        "item": "div.short",
+        "link": "a.short-poster",
+        "image": "img",
+    },
+    # Parcours par genre (page Explorer) : type → clé → {path, label}.
+    # Chemins vérifiés en live : films = /films/<genre>/, séries = /<genre>-series-/.
+    "browse": {
+        "serie": {
+            "aventure": {"path": "/aventure-series-/", "label": "Aventure"},
+            "familles": {"path": "/familles-series-/", "label": "Famille"},
+            "fantastique": {"path": "/fantastique-series-/", "label": "Fantastique"},
+            "judiciaire": {"path": "/judiciare-series-/", "label": "Judiciaire"},  # coquille du site
+            "medical": {"path": "/medical-series-/", "label": "Médical"},
+            "romance": {"path": "/romance-series-/", "label": "Romance"},
+            "science-fiction": {"path": "/science-fiction-series-/", "label": "Science-Fiction"},
+            "thriller": {"path": "/thriller-series-/", "label": "Thriller"},
+            "western": {"path": "/western-series-/", "label": "Western"},
+        },
+        "film": {
+            "actions": {"path": "/films/actions/", "label": "Action"},
+            "animations": {"path": "/films/animations/", "label": "Animation"},
+            "aventures": {"path": "/films/aventures/", "label": "Aventure"},
+            "biopics": {"path": "/films/biopics/", "label": "Biopic"},
+            "comedies": {"path": "/films/comedies/", "label": "Comédie"},
+            "cultes": {"path": "/films/cultes/", "label": "Culte"},
+            "documentaires": {"path": "/films/documentaires/", "label": "Documentaire"},
+            "drames": {"path": "/films/drames/", "label": "Drame"},
+            "epouvante-horreurs": {"path": "/films/epouvante-horreurs/", "label": "Épouvante-Horreur"},
+            "espionnages": {"path": "/films/espionnages/", "label": "Espionnage"},
+            "familles": {"path": "/films/familles/", "label": "Famille"},
+            "fantastiques": {"path": "/films/fantastiques/", "label": "Fantastique"},
+            "guerres": {"path": "/films/guerres/", "label": "Guerre"},
+            "historiques": {"path": "/films/historiques/", "label": "Historique"},
+            "policiers": {"path": "/films/policiers/", "label": "Policier"},
+            "romances": {"path": "/films/romances/", "label": "Romance"},
+            "science-fictions": {"path": "/films/science-fictions/", "label": "Science-Fiction"},
+            "thrillers": {"path": "/films/thrillers/", "label": "Thriller"},
+            "westerns": {"path": "/films/westerns/", "label": "Western"},
+        },
+    },
+}
+
+_NEWSID_RE = re.compile(r"(?:newsid=|/)(\d+)(?:-[^/]*)?\.?html?$|(?:newsid=)(\d+)")
+_ONCLICK_URL_RE = re.compile(r"location\.href='([^']+)'")
+_SEASON_RE = re.compile(r"saison\s*:?\s*(\d+)", re.IGNORECASE)
+_YEAR_RE = re.compile(r"(\d{4})")
+_FRAGMENT_RE = re.compile(r"^(vf|vostfr|vo)-(\d+(?:\.\d+)?)$")
+_VERSION_ORDER = {"vf": 0, "vostfr": 1, "vo": 2}
+
+
+def _merged_config() -> dict:
+    merged = deepcopy(DEFAULT_CONFIG)
+    for key, values in load_scraper_config("french_stream").items():
+        if isinstance(values, dict) and isinstance(merged.get(key), dict):
+            merged[key].update(values)
+        else:
+            merged[key] = values
+    return merged
+
+
+def _newsid_from_url(url: str) -> str | None:
+    match = _NEWSID_RE.search(url)
+    if not match:
+        return None
+    return match.group(1) or match.group(2)
+
+
+def _media_type(title: str, url: str) -> str:
+    if _SEASON_RE.search(title) or "-saison-" in url:
+        return "serie"
+    return "film"
+
+
+@register_source
+class FrenchStreamScraper(SourceScraper):
+    name = "french_stream"
+    label = "French-Stream"
+    base_url = "https://french-stream.lat"
+    media_types = ("serie", "film")
+
+    # ------------------------------------------------------------- helpers
+
+    def _title_url(self, source_id: str) -> str:
+        return f"{self.base_url}/index.php?newsid={source_id}"
+
+    @staticmethod
+    def _text(element: Tag | None) -> str:
+        return element.get_text(" ", strip=True) if element else ""
+
+    def _is_serie(self, soup: BeautifulSoup, title: str, url: str) -> bool:
+        config = _merged_config()
+        if soup.select_one(config["details"]["serie_marker"]):
+            return True
+        return _media_type(title, url) == "serie"
+
+    # ------------------------------------------------------------- search
+
+    async def search(self, query: str) -> list[SearchResult]:
+        config = _merged_config()
+        url = f"{self.base_url}{config['endpoints']['search']}"
+        html = await fetch(
+            url, referer=f"{self.base_url}/", data={"query": query, "page": "1"}
+        )
+        soup = BeautifulSoup(html, "lxml")
+        results: list[SearchResult] = []
+        for block in soup.select(config["search"]["item"]):
+            onclick = block.get("onclick", "")
+            url_match = _ONCLICK_URL_RE.search(onclick)
+            if not url_match:
+                logger.warning("french_stream : résultat sans lien, ignoré")
+                continue
+            href = url_match.group(1)
+            source_id = _newsid_from_url(href)
+            if not source_id:
+                logger.warning("french_stream : newsid introuvable dans %s", href)
+                continue
+            title = self._text(block.select_one(config["search"]["title"]))
+            image = block.select_one(config["search"]["poster"])
+            results.append(
+                SearchResult(
+                    source=self.name,
+                    source_id=source_id,
+                    title=title,
+                    url=f"{self.base_url}{href}" if href.startswith("/") else href,
+                    image_url=image.get("src") if image else None,
+                    media_type=_media_type(title, href),
+                )
+            )
+        logger.info("french_stream : %d résultats pour %r", len(results), query)
+        return results
+
+    # ------------------------------------------------------------- latest
+
+    async def latest(self) -> list[SearchResult]:
+        """Ajouts récents — page /series/ (mix films & séries)."""
+        config = _merged_config()
+        soup = await fetch_soup(f"{self.base_url}{config['latest']['path']}")
+        results = self._short_blocks(soup)
+        logger.info("french_stream : %d nouveautés récupérées", len(results))
+        return results
+
+    async def browse(self, media_type: str, category: str) -> list[SearchResult]:
+        """Parcours par genre — pages /films/<genre>/ et /<genre>-series-/.
+
+        ``media_type`` (« serie » | « film ») et ``category`` (clé du catalogue
+        YAML, ex. « thriller »). Le type est forcé sur les résultats : une page
+        genre séries ne liste que des séries, une page genre films que des films.
+        """
+        catalog = _merged_config().get("browse", {}).get(media_type, {})
+        entry = catalog.get(category)
+        if entry is None:
+            raise ScrapeError(f"Catégorie inconnue : {media_type}/{category}")
+        soup = await fetch_soup(f"{self.base_url}{entry['path']}")
+        results = self._short_blocks(soup, force_type=media_type)
+        logger.info("french_stream : %d titres dans %s/%s", len(results), media_type, category)
+        return results
+
+    def _short_blocks(self, soup: BeautifulSoup, force_type: str | None = None) -> list[SearchResult]:
+        """Bloc `div.short` → SearchResult (structure partagée nouveautés/genres)."""
+        config = _merged_config()
+        latest_cfg = config["latest"]
+        results: list[SearchResult] = []
+        for block in soup.select(latest_cfg["item"]):
+            link = block.select_one(latest_cfg["link"])
+            href = link.get("href") if link else None
+            if not href:
+                continue
+            source_id = _newsid_from_url(href)
+            if not source_id:
+                logger.warning("french_stream : newsid introuvable dans %s", href)
+                continue
+            image = link.select_one(latest_cfg["image"])
+            title = link.get("alt") or self._text(link)
+            results.append(
+                SearchResult(
+                    source=self.name,
+                    source_id=source_id,
+                    title=title,
+                    url=f"{self.base_url}{href}" if href.startswith("/") else href,
+                    image_url=image.get("src") if image else None,
+                    media_type=force_type or _media_type(title, href),
+                )
+            )
+        return results
+
+    def browse_catalog(self) -> dict[str, dict[str, str]]:
+        """Catalogue des genres parcourables : ``{type: {clé: libellé}}``."""
+        catalog = _merged_config().get("browse", {})
+        return {
+            media_type: {key: entry.get("label", key) for key, entry in genres.items()}
+            for media_type, genres in catalog.items()
+        }
+
+
+    async def genres_of_page(self, page_url: str) -> list[str]:
+        """Genres (libellés FR) de la fiche pointée par l'URL d'un téléchargement.
+
+        Utilisé par les recommandations « Pour toi » pour les séries/films,
+        absents de Kitsu (et donc sans signal de genre sans Sonarr).
+        """
+        if self.base_url.split("//")[-1].split("/")[0] not in page_url:
+            return []  # fiche d'une autre source (animé) : rien à lire ici
+        source_id = _newsid_from_url(page_url)
+        try:
+            details = await self.get_details(source_id)
+        except ScrapeError:
+            return []
+        return details.genres or []
+    # ------------------------------------------------------------- details
+
+    async def get_details(self, source_id: str) -> TitleDetails:
+        config = _merged_config()
+        details_cfg = config["details"]
+        url = self._title_url(source_id)
+        soup = await fetch_soup(url)
+
+        title = " ".join(self._text(soup.select_one(details_cfg["title"])).split())
+        if not title:
+            raise ScrapeError(f"french_stream : fiche introuvable pour {source_id} ({url})")
+
+        synopsis_el = soup.select_one(details_cfg["synopsis"])
+        if synopsis_el is not None:
+            for boilerplate in synopsis_el.select(details_cfg["synopsis_boilerplate"]):
+                boilerplate.decompose()
+        synopsis = self._text(synopsis_el) or None
+
+        genres_el = soup.select_one(details_cfg["genres"])
+        if genres_el and genres_el.select("a"):
+            genres = [a.get_text(strip=True) for a in genres_el.select("a") if a.get_text(strip=True)]
+        else:
+            genres = [g.strip() for g in self._text(genres_el).split(",") if g.strip()]
+
+        year_match = _YEAR_RE.search(self._text(soup.select_one(details_cfg["year"])))
+        if year_match is None:
+            year_match = _YEAR_RE.search(title)
+
+        poster_el = soup.select_one(details_cfg["poster_serie"])
+        poster = poster_el.get("src") if poster_el else None
+        if not poster:
+            film_data = soup.select_one(details_cfg["poster_film"])
+            poster = film_data.get("data-affiche") if film_data else None
+
+        is_serie = self._is_serie(soup, title, url)
+        episodes = await self._fetch_episodes(source_id, url, title) if is_serie else []
+
+        return TitleDetails(
+            source=self.name,
+            source_id=source_id,
+            title=title,
+            url=url,
+            synopsis=synopsis,
+            image_url=poster,
+            genres=genres,
+            year=int(year_match.group(1)) if year_match else None,
+            episode_count=len(episodes) if is_serie else 1,
+            episodes=episodes if is_serie else [self._film_episode(url)],
+            media_type="serie" if is_serie else "film",
+        )
+
+    # ------------------------------------------------------------ episodes
+
+    async def list_episodes(self, source_id: str) -> list[Episode]:
+        url = self._title_url(source_id)
+        soup = await fetch_soup(url)
+        title = " ".join(
+            self._text(soup.select_one(_merged_config()["details"]["title"])).split()
+        )
+        if not self._is_serie(soup, title, url):
+            return [self._film_episode(url)]
+        episodes = await self._fetch_episodes(source_id, url, title)
+        if not episodes:
+            raise ScrapeError(f"french_stream : aucun épisode trouvé pour {source_id} ({url})")
+        return episodes
+
+    @staticmethod
+    def _film_episode(page_url: str) -> Episode:
+        return Episode(number=1, title="Film", url=f"{page_url}#film", season=1)
+
+    async def _fetch_episodes(
+        self, source_id: str, page_url: str, title: str
+    ) -> list[Episode]:
+        """Épisodes d'une saison via l'API JSON du site (versions vf/vostfr/vo)."""
+        config = _merged_config()
+        endpoint = config["endpoints"]["episodes"].format(newsid=source_id)
+        raw = await fetch(f"{self.base_url}{endpoint}", referer=page_url)
+        try:
+            data = json.loads(raw)
+        except json.JSONDecodeError as exc:
+            raise ScrapeError(
+                f"french_stream : JSON d'épisodes invalide pour {source_id}"
+            ) from exc
+        if not isinstance(data, dict):
+            return []
+
+        season_match = _SEASON_RE.search(title)
+        season = int(season_match.group(1)) if season_match else 1
+        info = data.get("info") if isinstance(data.get("info"), dict) else {}
+
+        episodes: list[Episode] = []
+        for version, eps in data.items():
+            if version == "info" or not isinstance(eps, dict):
+                continue
+            for number_key in eps:
+                number_match = re.match(r"^(\d+(?:\.\d+)?)$", number_key)
+                if not number_match:
+                    continue
+                number = float(number_key)
+                ep_info = info.get(number_key) or info.get(str(int(number))) or {}
+                episodes.append(
+                    Episode(
+                        number=number,
+                        title=ep_info.get("title") if isinstance(ep_info, dict) else None,
+                        url=f"{page_url}#{version}-{number_key}",
+                        season=season,
+                        version=version,
+                    )
+                )
+        episodes.sort(
+            key=lambda e: (e.number, _VERSION_ORDER.get(e.version or "", 99))
+        )
+        return episodes
+
+    # -------------------------------------------------------------- embeds
+
+    async def extract_embed_links(self, episode_url: str) -> list[str]:
+        config = _merged_config()
+        page_url, _, fragment = episode_url.partition("#")
+        source_id = _newsid_from_url(page_url)
+        if not source_id:
+            raise ScrapeError(f"french_stream : newsid introuvable dans {episode_url}")
+
+        if fragment == "film":
+            endpoint = config["endpoints"]["film"].format(newsid=source_id)
+            raw = await fetch(f"{self.base_url}{endpoint}", referer=page_url)
+            try:
+                data = json.loads(raw)
+            except json.JSONDecodeError as exc:
+                raise ScrapeError(
+                    f"french_stream : JSON film invalide pour {source_id}"
+                ) from exc
+            links: list[str] = []
+            for hoster_urls in (data.get("players") or {}).values():
+                if not isinstance(hoster_urls, dict):
+                    continue
+                for embed in hoster_urls.values():
+                    if isinstance(embed, str) and embed.startswith("http") and embed not in links:
+                        links.append(embed)
+        else:
+            fragment_match = _FRAGMENT_RE.match(fragment)
+            if not fragment_match:
+                raise ScrapeError(
+                    f"french_stream : fragment d'épisode invalide dans {episode_url}"
+                )
+            version, number = fragment_match.group(1), fragment_match.group(2)
+            endpoint = config["endpoints"]["episodes"].format(newsid=source_id)
+            raw = await fetch(f"{self.base_url}{endpoint}", referer=page_url)
+            try:
+                data = json.loads(raw)
+            except json.JSONDecodeError as exc:
+                raise ScrapeError(
+                    f"french_stream : JSON d'épisodes invalide pour {source_id}"
+                ) from exc
+            hosters = (data.get(version) or {}).get(number) or {}
+            links = [u for u in hosters.values() if isinstance(u, str) and u.startswith("http")]
+
+        if not links:
+            raise ScrapeError(f"french_stream : aucun lien embed extrait de {episode_url}")
+        logger.info("french_stream : %d liens embed pour %s", len(links), episode_url)
+        return links

app/scrapers/sources/voiranime.py

@@ -0,0 +1,258 @@
+"""Source VoirAnime (voiranime.xyz) — plateforme VODSPHERE, animes VOSTFR.
+
+Faits structurels (vérifiés en live) :
+- Recherche : GET /catalogue?q=<q> → `article.catalogue-card` (lien `a.catalogue-poster`,
+  poster et titre dans `img[src]` / `img[alt]`).
+- Fiche : /catalogue/<slug> — titre `h1`, poster `img[alt^="Affiche de"]`, compteur
+  d'épisodes dans `.media-detail-meta span` (« 1192 épisodes »), synopsis `.media-detail-story p`.
+- Épisodes : `a.detail-episode-card` avec `data-season` et libellé `<small>S1 · E1</small>`.
+- Lecteur : /lecteur/<content>/<episode> — attributs data-prepare-url (résolution AJAX)
+  et data-fallback-source-url (source suivante) ; en fin de chaîne le fallback repointe
+  vers une source déjà vue → dédoublonnage par id de source obligatoire.
+- Accueil : carrousels #carousel-new (nouveautés) et #carousel-added (ajouts), 24 cartes.
+- La résolution d'un lien direct se fait dans l'extracteur hoster dédié (voir
+  app/scrapers/hosters/voiranime.py) : cette source ne renvoie que les URLs prepare.
+"""
+
+import logging
+import re
+from copy import deepcopy
+from urllib.parse import quote_plus, urljoin
+
+from bs4 import BeautifulSoup, Tag
+
+from app.scrapers.base import (
+    Episode,
+    ScrapeError,
+    SearchResult,
+    SourceScraper,
+    TitleDetails,
+    register_source,
+)
+from app.scrapers.config_loader import load_scraper_config
+from app.scrapers.http import fetch_soup
+
+logger = logging.getLogger(__name__)
+
+DEFAULT_CONFIG: dict = {
+    "search": {
+        "endpoint": "/catalogue?q={query}",
+        "result": "article.catalogue-card",
+        "link": "a.catalogue-poster",
+    },
+    "details": {
+        "title": "h1",
+        "poster": 'img[alt^="Affiche de"]',
+        "meta_count": ".media-detail-meta span",
+        "synopsis": ".media-detail-story p",
+    },
+    "episodes": {
+        "card": "a.detail-episode-card",
+    },
+    "player": {
+        "root": "[data-prepare-url]",
+        "prepare_attr": "data-prepare-url",
+        "fallback_attr": "data-fallback-source-url",
+        "max_sources": 4,
+    },
+    "latest": {
+        "path": "/",
+        "carousels": ("#carousel-new", "#carousel-added"),
+        "item": "article.content-card",
+        "link": "a.card-poster",
+    },
+}
+
+_SLUG_RE = re.compile(r"/catalogue/([a-z0-9-]+)", re.IGNORECASE)
+_EPISODE_LABEL_RE = re.compile(r"S\s*(\d+)\s*·\s*E\s*(\d+(?:[.,]\d+)?)", re.IGNORECASE)
+_EPISODE_COUNT_RE = re.compile(r"(\d+)\s*épisodes", re.IGNORECASE)
+_SOURCE_ID_RE = re.compile(r"/lecteur/prepare/(\d+)")
+
+_FALLBACK_SOURCE_RE = re.compile(r"[?&]source=(\d+)")
+def _merged_config() -> dict:
+    merged = deepcopy(DEFAULT_CONFIG)
+    for key, values in load_scraper_config("voiranime").items():
+        if isinstance(values, dict) and isinstance(merged.get(key), dict):
+            merged[key].update(values)
+        else:
+            merged[key] = values
+    return merged
+
+
+@register_source
+class VoirAnimeScraper(SourceScraper):
+    name = "voiranime"
+    label = "VoirAnime"
+    base_url = "https://voiranime.xyz"
+    media_types = ("anime",)
+
+    # ------------------------------------------------------------- helpers
+
+    def _title_url(self, source_id: str) -> str:
+        return f"{self.base_url}/catalogue/{source_id}"
+
+    @staticmethod
+    def _slug_from_url(url: str) -> str | None:
+        match = _SLUG_RE.search(url)
+        return match.group(1) if match else None
+
+    @staticmethod
+    def _text(element: Tag | None) -> str:
+        return element.get_text(" ", strip=True) if element else ""
+
+    def _result_from_card(self, link: Tag) -> SearchResult | None:
+        href = link.get("href")
+        if not href:
+            return None
+        source_id = self._slug_from_url(str(href))
+        if not source_id:
+            return None
+        img = link.select_one("img")
+        title = (img.get("alt") if img else None) or source_id.replace("-", " ")
+        image = img.get("src") if img else None
+        return SearchResult(
+            source=self.name,
+            source_id=source_id,
+            title=str(title).strip(),
+            url=self._title_url(source_id),
+            image_url=str(image) if image else None,
+        )
+
+    # ------------------------------------------------------------- search
+
+    async def search(self, query: str) -> list[SearchResult]:
+        config = _merged_config()
+        search_cfg = config["search"]
+        url = self.base_url + search_cfg["endpoint"].format(query=quote_plus(query))
+        soup = await fetch_soup(url)
+        results: list[SearchResult] = []
+        seen: set[str] = set()
+        for card in soup.select(search_cfg["result"]):
+            link = card.select_one(search_cfg["link"])
+            if link is None:
+                logger.warning("voiranime : carte de résultat sans lien, ignorée")
+                continue
+            result = self._result_from_card(link)
+            if result and result.source_id not in seen:
+                seen.add(result.source_id)
+                results.append(result)
+        return results
+
+    # ------------------------------------------------------------- détails
+
+    async def get_details(self, source_id: str) -> TitleDetails:
+        config = _merged_config()
+        details_cfg = config["details"]
+        url = self._title_url(source_id)
+        soup = await fetch_soup(url)
+
+        title = self._text(soup.select_one(details_cfg["title"])) or source_id.replace("-", " ")
+        poster = soup.select_one(details_cfg["poster"])
+        image = str(poster["src"]) if poster and poster.get("src") else None
+        synopsis = self._text(soup.select_one(details_cfg["synopsis"])) or None
+
+        episode_count: int | None = None
+        count_el = soup.select_one(details_cfg["meta_count"])
+        if count_el:
+            match = _EPISODE_COUNT_RE.search(self._text(count_el))
+            if match:
+                episode_count = int(match.group(1))
+
+        episodes = self._parse_episodes(soup, config)
+        return TitleDetails(
+            source=self.name,
+            source_id=source_id,
+            title=title,
+            url=url,
+            synopsis=synopsis,
+            image_url=image,
+            episode_count=episode_count if episode_count is not None else len(episodes),
+            episodes=episodes,
+        )
+
+    # ------------------------------------------------------------- épisodes
+
+    async def list_episodes(self, source_id: str) -> list[Episode]:
+        config = _merged_config()
+        soup = await fetch_soup(self._title_url(source_id))
+        return self._parse_episodes(soup, config)
+
+    def _parse_episodes(self, soup: BeautifulSoup, config: dict) -> list[Episode]:
+        episodes: list[Episode] = []
+        for card in soup.select(config["episodes"]["card"]):
+            href = card.get("href")
+            if not href:
+                continue
+            label = self._text(card)
+            match = _EPISODE_LABEL_RE.search(label)
+            if not match:
+                logger.warning("voiranime : libellé d'épisode illisible : %r", label[:60])
+                continue
+            episodes.append(
+                Episode(
+                    number=float(match.group(2).replace(",", ".")),
+                    title=None,
+                    url=urljoin(self.base_url + "/", str(href)),
+                    season=int(match.group(1)),
+                )
+            )
+        return episodes
+
+    # ------------------------------------------------------------- lecteurs
+
+    async def extract_embed_links(self, episode_url: str) -> list[str]:
+        """Parcourt la chaîne Lecteur 1 → 2 → … et renvoie les URLs prepare absolues."""
+        config = _merged_config()
+        player_cfg = config["player"]
+        prepare_urls: list[str] = []
+        seen_sources: set[str] = set()
+        page_url = episode_url
+        for _ in range(int(player_cfg["max_sources"])):
+            soup = await fetch_soup(page_url)
+            root = soup.select_one(player_cfg["root"])
+            if root is None:
+                break
+            prepare = root.get(player_cfg["prepare_attr"])
+            if not prepare:
+                break
+            prepare_url = urljoin(self.base_url + "/", str(prepare))
+            match = _SOURCE_ID_RE.search(prepare_url)
+            source_key = match.group(1) if match else prepare_url
+            if source_key in seen_sources:
+                break
+            seen_sources.add(source_key)
+            prepare_urls.append(prepare_url)
+            fallback = root.get(player_cfg["fallback_attr"])
+            if not fallback:
+                break
+            page_url = urljoin(self.base_url + "/", str(fallback))
+            next_source = _FALLBACK_SOURCE_RE.search(page_url)
+            if next_source and next_source.group(1) in seen_sources:
+                break
+        if not prepare_urls:
+            raise ScrapeError(f"voiranime : aucun lecteur trouvé sur {episode_url}")
+        return prepare_urls
+
+    # ------------------------------------------------------------- découverte
+
+    async def latest(self) -> list[SearchResult]:
+        """Nouveautés + ajouts récents — carrousels de la page d'accueil."""
+        config = _merged_config()
+        latest_cfg = config["latest"]
+        soup = await fetch_soup(self.base_url + latest_cfg["path"])
+        results: list[SearchResult] = []
+        seen: set[str] = set()
+        for carousel in latest_cfg["carousels"]:
+            section = soup.select_one(carousel)
+            if section is None:
+                logger.debug("voiranime : carrousel %s absent de l'accueil", carousel)
+                continue
+            for card in section.select(latest_cfg["item"]):
+                link = card.select_one(latest_cfg["link"])
+                if link is None:
+                    continue
+                result = self._result_from_card(link)
+                if result and result.source_id not in seen:
+                    seen.add(result.source_id)
+                    results.append(result)
+        return results

app/services/discover.py

@@ -30,6 +30,7 @@ from app.scrapers.base import (
     ScrapeError,
     SourceScraper,
     all_sources,
+    get_source,
     import_all_scrapers,
 )
 from app.services.kitsu import KitsuService, normalize_title
@@ -44,11 +45,56 @@ import_all_scrapers()
 _MAX_HISTORY_TITLES = 12  # titres récents analysés (téléchargements + favoris)
 _MAX_GENRES = 4           # genres retenus pour la requête Kitsu
 _KITSU_PAGE_MAX = 20      # limite dure de l'API Kitsu (page[limit] > 20 → 400)
-_ENRICH_CONCURRENCY = 6   # enrichissements Kitsu parallèles max (nouveautés)
+# (les TTL des sections suivent, plus bas)
+
+# Genres animés proposés à l'exploration — slugs des catégories Kitsu officielles.
+ANIME_GENRES: dict[str, str] = {
+    "action": "Action",
+    "adventure": "Aventure",
+    "comedy": "Comédie",
+    "drama": "Drame",
+    "fantasy": "Fantasy",
+    "science-fiction": "Science-Fiction",
+    "romance": "Romance",
+    "slice-of-life": "Tranche de vie",
+    "sports": "Sport",
+    "supernatural": "Surnaturel",
+    "mystery": "Mystère",
+    "psychological": "Psychologique",
+    "horror": "Horreur",
+    "mecha": "Mecha",
+    "isekai": "Isekai",
+    "music": "Musique",
+}
+
+# Genres des fiches séries/films (French-Stream, libellés FR, slugifiés) →
+# catégorie Kitsu équivalente. Les genres sans équivalent (Médical, Western…)
+# sont ignorés : mieux vaut un signal incomplet qu'une catégorie inexistante.
+_GENRE_FR_TO_KITSU: dict[str, str] = {
+    "action": "Action",
+    "aventure": "Adventure",
+    "aventures": "Adventure",
+    "comedie": "Comedy",
+    "comedies": "Comedy",
+    "drame": "Drama",
+    "drames": "Drama",
+    "epouvante-horreur": "Horror",
+    "horreur": "Horror",
+    "fantastique": "Fantasy",
+    "fantastiques": "Fantasy",
+    "romance": "Romance",
+    "romances": "Romance",
+    "science-fiction": "Science Fiction",
+    "science-fictions": "Science Fiction",
+    "surnaturel": "Supernatural",
+    "thriller": "Thriller",
+    "thrillers": "Thriller",
+}
 
 _LATEST_TTL_SECONDS = 600        # nouveautés : re-scrape au bout de 10 min
 _MUST_WATCH_TTL_SECONDS = 21600  # incontournables : quasi statique, 6 h
 _FOR_YOU_TTL_SECONDS = 3600      # recommandations : 1 h (l'historique évolue lentement)
+_ENRICH_CONCURRENCY = 6          # enrichissements Kitsu parallèles max (nouveautés)
 
 
 class _TTLCache:
@@ -92,14 +138,18 @@ class DiscoverService:
 
     # ------------------------------------------------------------ nouveautés
 
-    async def latest(self, limit: int = 24) -> list[dict]:
-        """Nouveautés toutes sources confondues, triées par date de sortie réelle.
+    async def latest_by_type(self, limit: int = 24) -> dict[str, list[dict]]:
+        """Nouveautés par type de média : ``{"anime": [...], "serie": [...], "film": [...]}``.
 
-        Les « récemment ajoutés » de chaque source sont fusionnés (doublons retirés),
-        enrichis via Kitsu (date de début, statut de diffusion) puis triés du plus
-        récent au plus ancien — ce qui sort / vient de sortir en premier.
+        Les « récemment ajoutés » de chaque source activée sont fusionnés (doublons
+        retirés) puis répartis en rails indépendants — chaque type garde sa place,
+        aucun n'évince les autres. Le rail animés est enrichi via Kitsu (date de
+        sortie, statut) et trié du plus récent au plus ancien ; les séries et films
+        réels, absents de Kitsu, gardent l'ordre du site (déjà « récents d'abord »).
+        Chaque rail est tronqué à ``limit``.
         """
-        cached = self._cache.get(f"latest:{limit}")
+        cache_key = f"latest_by_type:{limit}"
+        cached = self._cache.get(cache_key)
         if cached is not None:
             return cached  # type: ignore[return-value]
 
@@ -113,17 +163,28 @@ class DiscoverService:
             existing = merged.get(key)
             if existing is None or (not existing.get("image_url") and item.get("image_url")):
                 merged[key] = item
+
+        by_type: dict[str, list[dict]] = {"anime": [], "serie": [], "film": []}
+        for item in merged.values():
+            by_type.setdefault(item.get("media_type", "anime"), []).append(item)
+
+        # Enrichissement Kitsu limité aux animés (seul catalogue couvert) —
+        # inutile de bombarder l'API pour des titres qui n'y figurent pas.
         semaphore = asyncio.Semaphore(_ENRICH_CONCURRENCY)
 
         async def bounded(item: dict) -> dict:
             async with semaphore:
                 return await self._with_release_info(item)
 
-        enriched = await asyncio.gather(*(bounded(item) for item in merged.values()))
-        result = sorted(enriched, key=lambda it: it.get("start_date") or "", reverse=True)
-        result = result[:limit]
-        self._cache.set(f"latest:{limit}", result, _LATEST_TTL_SECONDS)
-        return result
+        enriched = await asyncio.gather(*(bounded(item) for item in by_type["anime"]))
+        by_type["anime"] = sorted(
+            enriched, key=lambda it: it.get("start_date") or "", reverse=True
+        )[:limit]
+        for kind in ("serie", "film"):
+            by_type[kind] = by_type[kind][:limit]
+
+        self._cache.set(cache_key, by_type, _LATEST_TTL_SECONDS)
+        return by_type
 
     async def _latest_of(self, source: SourceScraper) -> list[dict] | None:
         """Items latest() d'une source, aplatis avec les infos de source ([] si KO)."""
@@ -164,6 +225,46 @@ class DiscoverService:
         self._cache.set(key, items, _MUST_WATCH_TTL_SECONDS)
         return items
 
+
+    async def browse_serie_film(self, media_type: str, genre: str, limit: int = 24) -> list[dict]:
+        """Séries/films d'un genre French-Stream, aplatis comme les nouveautés.
+
+        Liste vide si la catégorie est inconnue ou la source en échec.
+        """
+        key = f"browse_fs:{media_type}:{genre}:{limit}"
+        cached = self._cache.get(key)
+        if cached is not None:
+            return cached  # type: ignore[return-value]
+        try:
+            scraper = get_source("french_stream")
+            results = await scraper.browse(media_type, genre)
+        except ScrapeError as exc:
+            logger.warning("Parcours %s/%s indisponible : %s", media_type, genre, exc)
+            return []
+        items = [
+            {**dataclasses.asdict(r), "source": scraper.name, "label": scraper.label}
+            for r in results
+        ][:limit]
+        self._cache.set(key, items, _LATEST_TTL_SECONDS)
+        return items
+    async def browse_anime(self, genre: str, limit: int = _KITSU_PAGE_MAX) -> list[dict]:
+        """Animés populaires d'une catégorie Kitsu (genre = slug du catalogue).
+
+        Liste vide si le genre est inconnu ou si Kitsu échoue (dégradation gracieuse).
+        """
+        limit = min(limit, _KITSU_PAGE_MAX)
+        if genre not in ANIME_GENRES:
+            return []
+        key = f"browse_anime:{genre}:{limit}"
+        cached = self._cache.get(key)
+        if cached is not None:
+            return cached  # type: ignore[return-value]
+        items = await self._kitsu_anime(
+            {"filter[categories]": genre, "sort": "-userCount", "page[limit]": limit}
+        )
+        self._cache.set(key, items, _FOR_YOU_TTL_SECONDS)
+        return items
+
     # ------------------------------------------------------------ pour toi
 
     async def for_you(self, user_id: int, limit: int = _KITSU_PAGE_MAX) -> dict:
@@ -179,49 +280,109 @@ class DiscoverService:
         if cached is not None:
             return cached  # type: ignore[return-value]
 
-        owned, favorite_genres = await self._owned(user_id)
+        owned, favorite_genres, series_pages = await self._owned(user_id)
         genre_counts = await self._genres_from_downloads(owned)
         for genre, count in favorite_genres.items():
             genre_counts[genre] = genre_counts.get(genre, 0) + count
+        if series_pages:
+            # Séries/films téléchargés : genres lus sur leur fiche source (sans Sonarr)
+            for genre, count in (await self._genres_from_series_pages(series_pages)).items():
+                genre_counts[genre] = genre_counts.get(genre, 0) + count
         sonarr_owned, sonarr_genres = await sonarr.profile()
         for genre, count in sonarr_genres.items():
             genre_counts[genre] = genre_counts.get(genre, 0) + count
+
+        if not (owned or favorite_genres or sonarr_owned):
+            # Aucun historique : proposition d'amorçage plutôt qu'une section muette
+            return {"based_on": [], "items": [], "cold_start": True}
         owned |= sonarr_owned
+
         if not genre_counts:
             result: dict = {"based_on": [], "items": []}
             self._cache.set(cache_key, result, _FOR_YOU_TTL_SECONDS)
             return result
 
         top_genres = sorted(genre_counts, key=genre_counts.get, reverse=True)[:_MAX_GENRES]
-        slugs = [category_slug(genre) for genre in top_genres]
-        items = await self._kitsu_anime(
-            {
-                "filter[categories]": ",".join(slugs),
-                "sort": "-userCount",
-                "page[limit]": limit,  # le déjà-possédé est filtré après
-            }
+        # Une requête par genre (Kitsu cumule les catégories en ET : quatre genres
+        # ensemble ne laissent que quelques titres — parfois le déjà-possédé !).
+        # Fusion entrelacée : chaque genre contribue, doublons retirés.
+        pools = [
+            list(pool)
+            for pool in await asyncio.gather(*(
+                self._kitsu_anime(
+                    {"filter[categories]": category_slug(g), "sort": "-userCount", "page[limit]": limit}
                 )
-        kept = [item for item in items if item["title"] and item["title"].casefold() not in owned]
+                for g in top_genres
+            ))
+            if pool
+        ]
+        candidates: list[dict] = []
+        seen: set[str] = set()
+        while pools:
+            for pool in pools[:]:
+                item = pool.pop(0)
+                key = str(item.get("kitsu_id") or item.get("title", "").casefold())
+                if key not in seen:
+                    seen.add(key)
+                    candidates.append(item)
+                if not pool:
+                    pools.remove(pool)
+        kept = [
+            item for item in candidates
+            if item["title"] and normalize_title(item["title"]).casefold() not in owned
+        ][:limit]
         result = {"based_on": top_genres, "items": kept}
         self._cache.set(cache_key, result, _FOR_YOU_TTL_SECONDS)
         return result
 
+    async def _genres_from_series_pages(self, pages: dict[str, str]) -> dict[str, int]:
+        """Genres Kitsu des séries/films téléchargés, lus sur leur fiche source.
+
+        Les libellés français des fiches sont convertis vers les catégories Kitsu
+        (anglais) ; les genres sans équivalent sont ignorés.
+        """
+        if not pages:
+            return {}
+        try:
+            scraper = get_source("french_stream")
+        except ScrapeError:
+            return {}
+        counts: dict[str, int] = {}
+        for page_url in list(pages.values())[:_MAX_HISTORY_TITLES]:
+            for genre_fr in await scraper.genres_of_page(page_url):
+                canonical = _GENRE_FR_TO_KITSU.get(category_slug(genre_fr))
+                if canonical:
+                    counts[canonical] = counts.get(canonical, 0) + 1
+        return counts
+
     def invalidate_for_you(self) -> None:
         """Recommandations recalculées au prochain appel (réglages Sonarr modifiés)."""
         self._cache.clear("for_you:")
 
-    async def _owned(self, user_id: int) -> tuple[set[str], dict[str, int]]:
-        """Titres possédés (normalisés) + genres directement connus via les favoris."""
+    async def _owned(self, user_id: int) -> tuple[set[str], dict[str, int], dict[str, str]]:
+        """Titres possédés (normalisés), genres des favoris, fiches des téléchargements.
+
+        ``series_pages`` relie un titre possédé à l'URL de sa fiche chez la source
+        (séries/films French-Stream) : sans Sonarr, c'est la seule source de genres.
+        """
         rows = await db.fetchall(
-            "SELECT DISTINCT title FROM downloads ORDER BY created_at DESC LIMIT ?",
+            "SELECT DISTINCT title, page_url FROM downloads ORDER BY created_at DESC LIMIT ?",
             (_MAX_HISTORY_TITLES,),
         )
         fav_rows = await db.fetchall(
             "SELECT payload FROM favorites WHERE user_id = ? ORDER BY created_at DESC LIMIT ?",
             (user_id, _MAX_HISTORY_TITLES),
         )
-        owned = {normalize_title(row["title"]).casefold() for row in rows}
-        owned.discard("")
+        owned: set[str] = set()
+        series_pages: dict[str, str] = {}
+        for row in rows:
+            key = normalize_title(row["title"]).casefold()
+            if not key:
+                continue
+            owned.add(key)
+            page_url = row["page_url"] or ""
+            if page_url and key not in series_pages:
+                series_pages[key] = page_url
         genre_counts: dict[str, int] = {}
         for row in fav_rows:
             try:
@@ -231,7 +392,7 @@ class DiscoverService:
             for genre in payload.get("genres") or []:
                 if isinstance(genre, str) and genre.strip():
                     genre_counts[genre.strip()] = genre_counts.get(genre.strip(), 0) + 1
-        return owned, genre_counts
+        return owned, genre_counts, series_pages
 
     async def _genres_from_downloads(self, titles: set[str]) -> dict[str, int]:
         """Genres Kitsu des titres téléchargés (cache DB puis recherche)."""

app/services/downloads.py

@@ -533,6 +533,10 @@ class DownloadManager:
         )
         await self._emit(download_id)
         logger.info("Téléchargement terminé (id=%s, %d octets)", download_id, size)
+        # L'historique a changé : les recommandations « Pour toi » se recalculent
+        from app.services.discover import discover
+
+        discover.invalidate_for_you()
 
     # ------------------------------------------------------------ helpers
 

app/services/kitsu.py

@@ -21,15 +21,26 @@ logger = logging.getLogger(__name__)
 _NOISE_WORDS_RE = re.compile(
     r"\b(?:VOSTFR\d*|VOST|VF[IV]?|TRUEFRENCH|FRENCH|MULTI|SUBFR?)\b", re.IGNORECASE
 )
-_TRAILING_SEASON_RE = re.compile(r"[\s\-–—:.]*\s*(?:saison|season)\s*\d+\s*$", re.IGNORECASE)
-_TRAILING_CODE_RE = re.compile(r"[\s\-–—:.]*\s*S\d+(?:E\d+)?\s*$", re.IGNORECASE)
+_TRAILING_SEASON_RE = re.compile(r"[\s\-–—:.]*\b(?:saison|season)\s*\d+\s*$", re.IGNORECASE)
+_TRAILING_CODE_RE = re.compile(r"[\s\-–—:.]*\bS\d+(?:\s*E\d+)?\s*$", re.IGNORECASE)
+_TRAILING_EPISODE_RE = re.compile(r"[\s\-–—:.]*\b(?:e|ep|episode)\s*\d+\s*$", re.IGNORECASE)
 
 
 def normalize_title(title: str) -> str:
-    """Nettoie un titre de scraping avant recherche Kitsu (bruit, saison, tirets)."""
+    """Nettoie un titre de scraping avant comparaison/recherche Kitsu.
+
+    Retire le bruit (VF/VOSTFR…), les marqueurs de saison/épisode en fin de titre
+    (« - Saison 1 - E3 », « S1 E1 ») — appliqués en boucle : un titre canonique
+    émergera identique d'un téléchargement (« Titre - Saison 1 - E3 ») ou d'une
+    fiche Kitsu (« Titre »), condition de l'exclusion du déjà-possédé.
+    """
     cleaned = _NOISE_WORDS_RE.sub(" ", title)
+    previous = None
+    while previous != cleaned:
+        previous = cleaned
         cleaned = _TRAILING_SEASON_RE.sub("", cleaned)
         cleaned = _TRAILING_CODE_RE.sub("", cleaned)
+        cleaned = _TRAILING_EPISODE_RE.sub("", cleaned)
     cleaned = re.sub(r"\s*[-–—_]+\s*", " ", cleaned)
     cleaned = re.sub(r"\s+", " ", cleaned)
     return cleaned.strip(" -–—:.")

app/services/settings.py

@@ -114,23 +114,3 @@ async def set_sonarr_config(url: str, apikey: str) -> None:
     await set_setting(SONARR_APIKEY_KEY, apikey.strip())
     logger.info("Configuration Sonarr enregistrée (%s)", url)
 
-# ---------------------------------------------------------------- mise à jour logicielle
-
-UPDATE_GITEA_URL_KEY = "update:gitea_url"
-UPDATE_REPO_KEY = "update:repo"
-UPDATE_TOKEN_KEY = "update:token"
-
-
-async def get_update_config() -> dict[str, str]:
-    return {
-        "gitea_url": await get_setting(UPDATE_GITEA_URL_KEY, "https://git.lanro.eu"),
-        "repo": await get_setting(UPDATE_REPO_KEY, "Roman/ohm_streaming"),
-        "token": await get_setting(UPDATE_TOKEN_KEY, ""),
-    }
-
-
-async def set_update_config(gitea_url: str, repo: str, token: str) -> None:
-    await set_setting(UPDATE_GITEA_URL_KEY, gitea_url.rstrip("/"))
-    await set_setting(UPDATE_REPO_KEY, repo.strip().strip("/"))
-    await set_setting(UPDATE_TOKEN_KEY, token.strip())
-    logger.info("Configuration de mise à jour enregistrée (%s)", repo)

app/services/update.py

@@ -1,6 +1,7 @@
 """Mises à jour logicielles.
 
-- Détection : dernier tag semver du dépôt Gitea via son API (jeton requis, repo privé).
+- Détection : dernier tag semver du dépôt Gitea public via son API (lecture anonyme),
+  avec le message du tag comme patchnote.
 - Application : POST à Watchtower (compagnon docker-compose) qui tire la nouvelle
   image et recrée le conteneur — quelques secondes d'indisponibilité.
 """
@@ -12,7 +13,6 @@ import time
 import httpx
 
 from app.config import get_settings
-from app.services.settings import get_update_config
 from app.version import get_version
 
 logger = logging.getLogger(__name__)
@@ -20,11 +20,11 @@ logger = logging.getLogger(__name__)
 _TAG_RE = re.compile(r"^v?(\d+)\.(\d+)\.(\d+)$")
 _CACHE_TTL = 300.0  # secondes
 
-_cache: dict[str, object] = {"checked_at": 0.0, "latest": None}
+_cache: dict[str, object] = {"checked_at": 0.0, "latest": None, "notes": None}
 
 
 class UpdateError(Exception):
-    """Erreur de mise à jour (config absente, Gitea injoignable, Watchtower KO)."""
+    """Erreur de mise à jour (Gitea injoignable, Watchtower KO)."""
 
 
 def parse_tag(tag: str) -> tuple[int, int, int] | None:
@@ -41,51 +41,48 @@ def is_newer(latest: str, current: str) -> bool:
 
 
 async def fetch_latest_version(*, force: bool = False) -> str | None:
-    """Dernier tag semver du dépôt (cache 5 min). None si non configuré ou erreur."""
+    """Dernier tag semver du dépôt + patchnote (cache 5 min). None si erreur."""
     now = time.monotonic()
     latest_cache = _cache["latest"]
     if not force and latest_cache and now - float(_cache["checked_at"]) < _CACHE_TTL:
         return str(latest_cache)  # type: ignore[arg-type]
 
-    config = await get_update_config()
-    if not config["repo"]:
-        return None
-
-    url = f"{config['gitea_url']}/api/v1/repos/{config['repo']}/tags?limit=20"
-    # Jeton requis uniquement pour un dépôt privé (public en lecture : inutile)
-    headers = {"Authorization": f"token {config['token']}"} if config["token"] else {}
+    settings = get_settings()
+    url = f"{settings.gitea_url}/api/v1/repos/{settings.gitea_repo}/tags?limit=20"
     try:
         async with httpx.AsyncClient(timeout=10, follow_redirects=True) as client:
-            resp = await client.get(url, headers=headers)
+            resp = await client.get(url)
             resp.raise_for_status()
-            tags = [t["name"] for t in resp.json() if parse_tag(t.get("name", ""))]
+            tags = [
+                (t["name"], (t.get("message") or "").strip())
+                for t in resp.json()
+                if parse_tag(t.get("name", ""))
+            ]
     except (httpx.HTTPError, ValueError, KeyError) as exc:
         logger.warning("Vérification de mise à jour impossible : %s", exc)
         return None
 
-    latest = max(tags, key=parse_tag) if tags else None  # type: ignore[arg-type]
-    _cache.update(checked_at=now, latest=latest)
+    latest, notes = max(tags, key=lambda t: parse_tag(t[0])) if tags else (None, None)
+    _cache.update(checked_at=now, latest=latest, notes=notes or None)
     if latest and is_newer(latest, get_version()):
         logger.info("Nouvelle version disponible : %s (courante %s)", latest, get_version())
     return latest
 
 
 def invalidate_cache() -> None:
-    _cache.update(checked_at=0.0, latest=None)
+    _cache.update(checked_at=0.0, latest=None, notes=None)
 
 
 async def status() -> dict[str, object]:
-    """État complet : version courante, dernière dispo, config."""
-    config = await get_update_config()
+    """État complet : version courante, dernière dispo, patchnote."""
     current = get_version()
     latest = await fetch_latest_version()
     return {
         "current": current,
         "latest": latest,
+        "notes": _cache["notes"],
         "update_available": bool(latest and is_newer(latest, current)),
-        "configured": bool(config["repo"]),
         "docker": bool(get_settings().watchtower_url),
-        "config": config,
     }
 
 

app/static/css/style.css

@@ -89,6 +89,14 @@ button { font-family: inherit; }
 }
 .icon-btn:hover { color: var(--text); background: var(--surface-2); }
 
+.pref-select {
+  background: var(--surface); color: var(--text-dim);
+  border: 1px solid var(--border); border-radius: 8px;
+  font: inherit; font-size: 0.85rem;
+  padding: 0.35rem 0.5rem; cursor: pointer;
+}
+.pref-select:hover { color: var(--text); border-color: var(--accent); }
+
 /* ------------------------------------------------------------ layout */
 
 .main { padding: 2rem clamp(1rem, 3.5vw, 4rem) 4rem; }
@@ -544,6 +552,52 @@ button { font-family: inherit; }
 
 .rail-section { margin-bottom: 2.4rem; animation: rise 0.5s ease backwards; }
 
+/* ------------------------------------------------------------- Explorer */
+
+.explore-section { animation-delay: 0s; }
+.explore-filters { display: flex; flex-direction: column; gap: 0.5rem; margin-bottom: 1rem; }
+.chip-row { display: flex; flex-wrap: wrap; gap: 0.4rem; }
+.genre-chip {
+  background: var(--surface); color: var(--text-dim);
+  border: 1px solid var(--border); border-radius: 99px;
+  font: inherit; font-size: 0.82rem; font-weight: 600;
+  padding: 0.32rem 0.85rem; cursor: pointer;
+  transition: color 0.15s, border-color 0.15s, background 0.15s;
+}
+.genre-chip:hover { color: var(--text); border-color: var(--accent); }
+.genre-chip.active {
+  background: var(--accent); border-color: var(--accent); color: #fff;
+}
+.explore-skel .rail-skel { margin-top: 0.2rem; }
+
+/* CTA d'amorçage « Pour toi » (aucun historique) */
+.cta-card {
+  display: flex; align-items: center; gap: 1.6rem;
+  background: var(--surface); border: 1px solid var(--border);
+  border-radius: 14px; padding: 1.6rem 1.8rem;
+}
+.cta-emoji { font-size: 2.6rem; }
+.cta-body { display: flex; flex-direction: column; gap: 0.4rem; }
+.cta-title { font-weight: 700; font-size: 1.05rem; }
+.cta-body p { margin: 0; color: var(--text-dim); font-size: 0.92rem; }
+.cta-actions { display: flex; gap: 0.6rem; margin-top: 0.6rem; }
+.cta-btn {
+  background: var(--accent); color: #fff; border: none; border-radius: 8px;
+  font: inherit; font-size: 0.88rem; font-weight: 600;
+  padding: 0.5rem 1rem; cursor: pointer; text-decoration: none;
+  transition: filter 0.15s;
+}
+.cta-btn:hover { filter: brightness(1.12); }
+.cta-btn.ghost {
+  background: transparent; color: var(--text);
+  border: 1px solid var(--border);
+}
+.cta-btn.ghost:hover { border-color: var(--accent); color: var(--accent); filter: none; }
+@media (max-width: 640px) {
+  .cta-card { flex-direction: column; text-align: center; }
+  .cta-actions { justify-content: center; flex-wrap: wrap; }
+}
+
 .rail-title {
   font-size: 1.15rem;
   font-weight: 700;

app/templates/admin.html

@@ -111,28 +111,23 @@
     </p>
 
     <div style="display:flex;gap:0.5rem;flex-wrap:wrap;align-items:center">
-      <input class="input" x-model="upd.config.gitea_url" placeholder="https://git.lanro.eu"
-             style="flex:2;min-width:220px" @change="upd._dirty = true">
-      <input class="input" x-model="upd.config.repo" placeholder="Roman/ohm_streaming"
-             style="flex:1;min-width:180px" @change="upd._dirty = true">
-      <input class="input" x-model="upd.config.token" type="password" placeholder="Jeton d'accès Gitea"
-             style="flex:1;min-width:180px" @change="upd._dirty = true">
-      <button class="btn btn-sm" @click="saveUpdate()" :disabled="!upd._dirty">Enregistrer</button>
       <button class="btn btn-sm btn-ghost" @click="checkUpdate()" :disabled="upd._checking"
               x-text="upd._checking ? '…' : 'Vérifier'"></button>
       <button class="btn btn-sm btn-accent" x-show="upd.update_available" @click="applyUpdate()"
               :disabled="upd._applying" x-text="upd._applying ? 'Mise à jour…' : '⬆ Mettre à jour maintenant'"></button>
     </div>
-    <p x-show="!upd.configured" style="margin:0.5rem 0 0;font-size:0.85rem;color:var(--text-dim)">
-      Renseignez le dépôt Gitea pour activer la détection des mises à jour.
-      Jeton d'accès (droit « lecture ») nécessaire uniquement si le dépôt est privé.
-    </p>
-    <p x-show="upd.configured && !upd.docker" style="margin:0.5rem 0 0;font-size:0.85rem;color:var(--text-dim)">
+    <details x-show="upd.notes" style="margin:0.6rem 0 0">
+      <summary style="cursor:pointer;font-size:0.85rem;color:var(--text-dim)">
+        Patchnote <span x-text="upd.latest"></span>
+      </summary>
+      <pre style="white-space:pre-wrap;font-size:0.82rem;margin:0.5rem 0 0;color:var(--text-dim)"
+           x-text="upd.notes"></pre>
+    </details>
+    <p x-show="!upd.docker" style="margin:0.5rem 0 0;font-size:0.85rem;color:var(--text-dim)">
       ⚠ Mise à jour automatique disponible uniquement en déploiement Docker
       (<code>docker compose pull &amp;&amp; docker compose up -d</code> sinon).
     </p>
   </div>
-+
 
   <div class="panel">
     <h2>👥 Utilisateurs</h2>
@@ -171,9 +166,8 @@ function adminPage() {
   return {
     users: [], sources: [], stats: {}, forbidden: false,
     upd: {
-      current: '', latest: null, update_available: false, configured: false, docker: false,
-      config: { gitea_url: '', repo: '', token: '' },
-      _dirty: false, _checking: false, _applying: false,
+      current: '', latest: null, notes: null, update_available: false, docker: false,
+      _checking: false, _applying: false,
     },
     integrations: {
       torznab: { apikey: '', endpoint: '' }, sonarr: { url: '', apikey: '' },
@@ -287,18 +281,9 @@ function adminPage() {
     applyUpdateData(data) {
       this.upd.current = data.current;
       this.upd.latest = data.latest;
+      this.upd.notes = data.notes;
       this.upd.update_available = data.update_available;
-      this.upd.configured = data.configured;
       this.upd.docker = data.docker;
-      this.upd.config = { ...data.config, _dirty: false };
-    },
-
-    async saveUpdate() {
-      const res = await fetch('/api/admin/update', {
-        method: 'PUT', headers: { 'Content-Type': 'application/json' },
-        body: JSON.stringify(this.upd.config),
-      });
-      if (res.ok) { this.applyUpdateData(await res.json()); toast('✔ Configuration enregistrée'); }
     },
 
     async checkUpdate() {

app/templates/base.html

@@ -4,14 +4,14 @@
   <meta charset="utf-8">
   <meta name="viewport" content="width=device-width, initial-scale=1">
   <title>{% block title %}Ohm Stream{% endblock %}</title>
-  <link rel="stylesheet" href="/static/css/style.css">
+  <link rel="stylesheet" href="/static/css/style.css?v={{ asset_v }}">
   <link rel="preconnect" href="https://fonts.googleapis.com">
   <link href="https://fonts.googleapis.com/css2?family=Outfit:wght@400;500;600;700;800&display=swap" rel="stylesheet">
   <script src="https://unpkg.com/htmx.org@2.0.4"></script>
   <script defer src="https://cdn.jsdelivr.net/npm/alpinejs@3.14.1/dist/cdn.min.js"></script>
 </head>
 <body>
-<header class="topbar" x-data="{ username: '' }" x-init="fetch('/auth/me').then(r => r.json()).then(u => username = u.username).catch(() => {})">
+<header class="topbar" x-data="{ username: '', pref: 'both' }" x-init="fetch('/auth/me').then(r => r.json()).then(u => { username = u.username; pref = u.content_preference || 'both'; }).catch(() => {})">
   <a class="logo" href="/">OHM<span>STREAM</span></a>
   <nav class="topnav">
     <a class="{{ 'active' if active == 'discover' }}" href="/discover">Découvrir</a>
@@ -22,6 +22,12 @@
     <a class="{{ 'active' if active == 'admin' }}" href="/admin">Admin</a>
   </nav>
   <div class="topbar-right">
+    <select class="pref-select" x-model="pref" x-show="username" title="Types de contenus"
+            @change="fetch('/auth/preferences', { method: 'PUT', headers: {'Content-Type': 'application/json'}, body: JSON.stringify({ content_preference: pref }) }).then(() => location.reload())">
+      <option value="anime">⛩ Animés</option>
+      <option value="serie">📺 Séries</option>
+      <option value="both">✨ Les deux</option>
+    </select>
     <div class="avatar" x-text="username ? username[0] : '·'" :title="username"></div>
     <form method="post" action="/auth/logout" style="display:inline">
       <button class="icon-btn" type="submit" title="Déconnexion">⏻</button>
@@ -31,8 +37,8 @@
 <main class="main">
   {% block content %}{% endblock %}
 </main>
-<script src="/static/js/app.js"></script>
-<script src="/static/js/update-watcher.js"></script>
+<script src="/static/js/app.js?v={{ asset_v }}"></script>
+<script src="/static/js/update-watcher.js?v={{ asset_v }}"></script>
 {% block scripts %}{% endblock %}
 </body>
 </html>

app/templates/discover.html

@@ -4,10 +4,61 @@
 
 {% block content %}
 <h1 class="page-title">Découvrir</h1>
-<p class="page-sub">Nouveautés de tes sources, incontournables et suggestions basées sur tes téléchargements.</p>
+<p class="page-sub">Explore par genre, nouveautés de tes sources, incontournables et suggestions basées sur tes téléchargements.</p>
 
 <div x-data="discoverPage()" x-init="load()" x-cloak>
 
+  <!-- ------------------------------------------------ Explorer par genre -->
+  <section class="rail-section explore-section" x-show="!loading && types.length > 0">
+    <h2 class="rail-title">🎭 Explorer
+      <span class="rail-source" x-text="exploreSubtitle"></span>
+    </h2>
+    <div class="explore-filters">
+      <div class="chip-row" role="tablist" aria-label="Type de contenu">
+        <template x-for="t in types" :key="t">
+          <button class="genre-chip" :class="{ active: exploreType === t }" role="tab"
+                  :aria-selected="exploreType === t" @click="setType(t)"
+                  x-text="typeLabels[t]"></button>
+        </template>
+      </div>
+      <div class="chip-row" role="tablist" aria-label="Genre" x-show="currentGenres.length > 0">
+        <template x-for="g in currentGenres" :key="g.key">
+          <button class="genre-chip" :class="{ active: exploreGenre === g.key }" role="tab"
+                  :aria-selected="exploreGenre === g.key" @click="setGenre(g.key)"
+                  x-text="g.label"></button>
+        </template>
+      </div>
+    </div>
+    <div class="rail-wrap" x-show="!exploreLoading">
+      <button class="rail-nav rail-prev" type="button" aria-label="Défiler à gauche"
+              @click="scrollRail($el, -1)">‹</button>
+      <div class="rail">
+        <template x-for="(item, i) in exploreItems" :key="'ex' + i + (item.source_id || item.kitsu_id || item.title)">
+          <a class="card rail-card" :style="`--i:${i}`" :href="exploreHref(item)"
+             :title="item.source ? item.title : `Rechercher « ${item.title} »`">
+            <img class="card-poster" :src="item.image_url || '/static/img/placeholder.svg'" :alt="item.title"
+                 loading="lazy" onerror="this.src='/static/img/placeholder.svg'">
+            <span class="card-chip" x-show="item.label" x-text="item.label"></span>
+            <span class="card-chip" x-show="!item.label && item.rating" x-text="item.rating ? '★ ' + item.rating : ''"></span>
+            <div class="card-overlay">
+              <div class="card-title" x-text="item.title"></div>
+              <div class="card-meta">
+                <span class="badge badge-type" x-text="item.start_date ? item.start_date.slice(0, 4) : (item.year || '')"></span>
+              </div>
+            </div>
+          </a>
+        </template>
+      </div>
+      <button class="rail-nav rail-next" type="button" aria-label="Défiler à droite"
+              @click="scrollRail($el, 1)">›</button>
+    </div>
+    <div class="explore-skel" x-show="exploreLoading">
+      <div class="rail rail-skel">
+        <template x-for="i in 8" :key="'sk' + i"><div class="rail-card skel skel-card"></div></template>
+      </div>
+    </div>
+  </section>
+
   <!-- ------------------------------------------------ Skeletons de chargement -->
   <div x-show="loading">
     <section class="rail-section" x-data="{ n: 8 }">
@@ -28,14 +79,14 @@
     <div class="login-error" x-text="error"></div>
   </template>
 
-  <!-- ------------------------------------------------ Nouveautés -->
-  <section class="rail-section" x-show="!loading && latest.length > 0">
-    <h2 class="rail-title">🆕 Nouveautés <span class="rail-source">de tes sources, triées par date de sortie</span></h2>
+  <!-- ------------------------------------------------ Nouveautés animés -->
+  <section class="rail-section" x-show="!loading && latestAnime.length > 0">
+    <h2 class="rail-title">🆕 Nouveautés animés <span class="rail-source">triées par date de sortie</span></h2>
     <div class="rail-wrap">
       <button class="rail-nav rail-prev" type="button" aria-label="Défiler à gauche"
               @click="scrollRail($el, -1)">‹</button>
       <div class="rail">
-        <template x-for="(item, i) in latest" :key="item.source + item.source_id">
+        <template x-for="(item, i) in latestAnime" :key="item.source + item.source_id">
           <a class="card rail-card" :style="`--i:${i}`" :href="`/title/${item.source}/${encodeURIComponent(item.source_id)}`">
             <img class="card-poster" :src="item.image_url || '/static/img/placeholder.svg'" :alt="item.title"
                  loading="lazy" onerror="this.src='/static/img/placeholder.svg'">
@@ -56,9 +107,33 @@
     </div>
   </section>
 
+  <!-- ------------------------------------------------ Nouveautés séries & films -->
+  <section class="rail-section" x-show="!loading && latestSerie.length > 0">
+    <h2 class="rail-title">🆕 Nouveautés séries &amp; films <span class="rail-source">ajouts récents de French-Stream</span></h2>
+    <div class="rail-wrap">
+      <button class="rail-nav rail-prev" type="button" aria-label="Défiler à gauche"
+              @click="scrollRail($el, -1)">‹</button>
+      <div class="rail">
+        <template x-for="(item, i) in latestSerie" :key="item.source + item.source_id">
+          <a class="card rail-card" :style="`--i:${i}`" :href="`/title/${item.source}/${encodeURIComponent(item.source_id)}`">
+            <img class="card-poster" :src="item.image_url || '/static/img/placeholder.svg'" :alt="item.title"
+                 loading="lazy" onerror="this.src='/static/img/placeholder.svg'">
+            <span class="card-chip" x-text="item.label"></span>
+            <span class="card-year" x-text="item.media_type === 'film' ? 'Film' : ''"></span>
+            <div class="card-overlay">
+              <div class="card-title" x-text="item.title"></div>
+            </div>
+          </a>
+        </template>
+      </div>
+      <button class="rail-nav rail-next" type="button" aria-label="Défiler à droite"
+              @click="scrollRail($el, 1)">›</button>
+    </div>
+  </section>
+
   <!-- ------------------------------------------------ Incontournables -->
   <section class="rail-section" x-show="!loading && mustWatch.length > 0">
-    <h2 class="rail-title">🔥 Incontournables <span class="rail-source">les classiques les mieux notés</span></h2>
+    <h2 class="rail-title">🔥 Incontournables animés <span class="rail-source">les classiques les mieux notés</span></h2>
     <div class="rail-wrap">
       <button class="rail-nav rail-prev" type="button" aria-label="Défiler à gauche"
               @click="scrollRail($el, -1)">‹</button>
@@ -81,11 +156,26 @@
   </section>
 
   <!-- ------------------------------------------------ Pour toi -->
-  <section class="rail-section" x-show="!loading && forYou.items.length > 0">
+  <section class="rail-section" x-show="!loading && (forYou.items.length > 0 || forYou.cold_start)">
     <h2 class="rail-title">✨ Pour toi
       <span class="rail-source" x-show="forYou.based_on.length"
             x-text="forYou.based_on.length ? 'parce que tu aimes : ' + forYou.based_on.join(', ') : ''"></span>
     </h2>
+
+
+    <!-- Aucun historique : carte d'amorçage au lieu d'une section muette -->
+    <div class="cta-card" x-show="!loading && forYou.cold_start">
+      <div class="cta-emoji">🧭</div>
+      <div class="cta-body">
+        <div class="cta-title">On ne te connaît pas encore…</div>
+        <p>Télécharge tes premiers animés ou séries, ou ajoute des favoris :
+           « Pour toi » s'ajustera automatiquement à tes genres préférés.</p>
+        <div class="cta-actions">
+          <a class="cta-btn" href="/">Chercher un premier titre</a>
+          <a class="cta-btn ghost" href="/discover">Explorer par genre</a>
+        </div>
+      </div>
+    </div>
     <div class="rail-wrap">
       <button class="rail-nav rail-prev" type="button" aria-label="Défiler à gauche"
               @click="scrollRail($el, -1)">‹</button>
@@ -107,7 +197,7 @@
     </div>
   </section>
 
-  <template x-if="!loading && !error && latest.length === 0 && mustWatch.length === 0 && forYou.items.length === 0">
+  <template x-if="!loading && !error && latestAnime.length === 0 && latestSerie.length === 0 && mustWatch.length === 0 && forYou.items.length === 0 && exploreItems.length === 0">
     <div class="empty-state"><div class="big">🏜️</div>Rien à afficher pour le moment — réessaie plus tard.</div>
   </template>
 </div>
@@ -117,23 +207,89 @@
 <script>
 function discoverPage() {
   return {
-    latest: [], mustWatch: [], forYou: { based_on: [], items: [] },
+    latestAnime: [], latestSerie: [], mustWatch: [], forYou: { based_on: [], items: [] },
+    genres: {}, exploreType: null, exploreGenre: null,
+    exploreItems: [], exploreLoading: false,
     loading: true, error: null,
+    typeLabels: { anime: '⛩ Animés', serie: '📺 Séries', film: '🎬 Films' },
+
+    get types() { return Object.keys(this.genres); },
+    get currentGenres() { return this.genres[this.exploreType] || []; },
+    get exploreSubtitle() {
+      const genre = (this.currentGenres.find(g => g.key === this.exploreGenre) || {}).label;
+      return genre ? `${this.typeLabels[this.exploreType] || ''} · ${genre}` : 'choisis un genre';
+    },
+
     scrollRail(el, dir) {
       const rail = el.closest('.rail-wrap').querySelector('.rail');
       rail.scrollBy({ left: dir * rail.clientWidth * 0.85, behavior: 'smooth' });
     },
+
+    exploreHref(item) {
+      return item.source
+        ? `/title/${item.source}/${encodeURIComponent(item.source_id)}`
+        : `/?q=${encodeURIComponent(item.title)}`;
+    },
+
+    setType(type) {
+      this.exploreType = type;
+      this.exploreGenre = (this.genres[type][0] || {}).key || null;
+      this.browse();
+    },
+
+    setGenre(genre) {
+      this.exploreGenre = genre;
+      this.browse();
+    },
+
+    async browse() {
+      if (!this.exploreType || !this.exploreGenre) return;
+      this.exploreLoading = true;
+      const params = new URLSearchParams({ type: this.exploreType, genre: this.exploreGenre });
+      const url = `${location.pathname}?${params}`;
+      history.replaceState(null, '', url);  // état partageable
+      try {
+        const res = await fetch(`/api/discover/browse?${params}`);
+        if (!res.ok) throw new Error('Erreur ' + res.status);
+        this.exploreItems = (await res.json()).items;
+      } catch (e) {
+        this.exploreItems = [];
+        console.error('Explorer :', e);
+      } finally {
+        this.exploreLoading = false;
+      }
+    },
+
     async load() {
       this.loading = true; this.error = null;
       try {
-        const res = await fetch('/api/discover');
-        if (!res.ok) throw new Error('Erreur ' + res.status);
-        const data = await res.json();
-        this.latest = data.latest;
+        const [discoverRes, genresRes] = await Promise.all([
+          fetch('/api/discover'), fetch('/api/discover/genres'),
+        ]);
+        if (!discoverRes.ok) throw new Error('Erreur ' + discoverRes.status);
+        const data = await discoverRes.json();
+        this.latestAnime = data.latest_anime;
+        this.latestSerie = data.latest_serie;
         this.mustWatch = data.must_watch;
         this.forYou = data.for_you;
-      } catch (e) { this.error = e.message; }
-      finally { this.loading = false; }
+        this.genres = await genresRes.json();
+
+        // État initial : paramètres d'URL (?t=&g=) sinon premier type/genre dispo
+        const params = new URLSearchParams(location.search);
+        let type = params.get('t');
+        if (!this.genres[type]) type = this.types[0];
+        let genre = params.get('g');
+        if (!(this.genres[type] || []).some(g => g.key === genre)) {
+          genre = ((this.genres[type] || [])[0] || {}).key;
+        }
+        this.exploreType = type || null;
+        this.exploreGenre = genre || null;
+        await this.browse();
+      } catch (e) {
+        this.error = e.message;
+      } finally {
+        this.loading = false;
+      }
     },
   };
 }

app/templates/login.html

@@ -4,7 +4,7 @@
   <meta charset="utf-8">
   <meta name="viewport" content="width=device-width, initial-scale=1">
   <title>Connexion — Ohm Stream</title>
-  <link rel="stylesheet" href="/static/css/style.css">
+  <link rel="stylesheet" href="/static/css/style.css?v={{ asset_v }}">
   <link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800&display=swap" rel="stylesheet">
   <script defer src="https://cdn.jsdelivr.net/npm/alpinejs@3.14.1/dist/cdn.min.js"></script>
 </head>

docker-compose.podman.yml

@@ -0,0 +1,51 @@
+# Déploiement OhmStreaming sous Podman — variante de docker-compose.yml
+#
+#   1. Activer le socket Podman (une seule fois) :
+#        rootless : systemctl --user enable --now podman.socket
+#                   sudo loginctl enable-linger $USER   # persister après déconnexion
+#                   export PODMAN_SOCKET=/run/user/$(id -u)/podman/podman.sock
+#        rootful  : sudo systemctl enable --now podman.socket
+#   2. cp .env.example .env && $EDITOR .env   (OHM_SECRET_KEY + WATCHTOWER_TOKEN obligatoires)
+#   3. podman compose -f docker-compose.podman.yml up -d
+#      (ou podman-compose -f docker-compose.podman.yml up -d)
+#
+# Mise à jour : page Admin → « Mise à jour », ou manuellement :
+#   podman compose -f docker-compose.podman.yml pull && podman compose -f docker-compose.podman.yml up -d
+
+services:
+  ohm:
+    image: ${OHM_IMAGE:-ghcr.io/aulyriusohm/ohm-streaming:latest}
+    restart: unless-stopped
+    ports:
+      - "${OHM_PORT:-8777}:8777"
+    env_file: .env
+    environment:
+      # Chemins internes (volumes ci-dessous) — ne pas changer
+      OHM_DATA_DIR: /data
+      OHM_DOWNLOAD_DIR: /downloads
+      OHM_DATABASE_PATH: /data/ohm.db
+      # Watchtower compagnon sur le réseau interne compose
+      OHM_WATCHTOWER_URL: http://watchtower:8080
+      OHM_WATCHTOWER_TOKEN: ${WATCHTOWER_TOKEN:?WATCHTOWER_TOKEN manquant — voir .env.example}
+    volumes:
+      # Fedora/SELinux : suffixer de « :z » en cas d'erreur « Permission denied »
+      - ./data:/data
+      - ./downloads:/downloads
+
+  # Compagnon de mise à jour : recrée le conteneur ohm quand l'admin le demande
+  # (API HTTP interne uniquement, aucun port publié sur l'hôte)
+  watchtower:
+    image: containrrr/watchtower
+    restart: unless-stopped
+    volumes:
+      # Socket Podman (compatible API Docker) au lieu de /var/run/docker.sock
+      - ${PODMAN_SOCKET:-/run/podman/podman.sock}:/var/run/docker.sock
+    environment:
+      # Pas de vérification périodique : uniquement sur appel de l'admin
+      WATCHTOWER_HTTP_API_UPDATE: "true"
+      WATCHTOWER_HTTP_API_TOKEN: ${WATCHTOWER_TOKEN:?WATCHTOWER_TOKEN manquant — voir .env.example}
+      # Ne mettre à jour que les conteneurs étiquetés (ohm), pas watchtower lui-même
+      WATCHTOWER_LABEL_ENABLE: "true"
+      # Pas de DOCKER_API_VERSION : Podman annonce lui-même la version d'API
+      # Supprimer les anciennes images après mise à jour
+      WATCHTOWER_CLEANUP: "true"

docker-compose.yml

@@ -2,16 +2,18 @@
 #
 #   Installation guidée (recommandée) :  ./scripts/install.sh
 #   Ou à la main :
-#     1. docker login git.lanro.eu          (compte Gitea avec accès lecture au repo)
-#     2. cp .env.example .env && $EDITOR .env   (OHM_SECRET_KEY + WATCHTOWER_TOKEN obligatoires)
-#     3. docker compose up -d
+#     1. cp .env.example .env && $EDITOR .env   (OHM_SECRET_KEY + WATCHTOWER_TOKEN obligatoires)
+#     2. docker compose up -d
+#
+# Image par défaut : GHCR (publique). Pour utiliser le registre Gitea à la place :
+#   OHM_IMAGE=git.lanro.eu/roman/ohm_streaming:latest dans .env (+ docker login git.lanro.eu)
 #
 # Mise à jour : page Admin → « Mise à jour », ou manuellement :
 #   docker compose pull && docker compose up -d
 
 services:
   ohm:
-    image: git.lanro.eu/roman/ohm_streaming:latest
+    image: ${OHM_IMAGE:-ghcr.io/aulyriusohm/ohm-streaming:latest}
     # Build local (dev) : décommenter ces lignes et commenter « image: » ci-dessus
     # build:
     #   context: .

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ohm-stream"
-version = "0.1.0"
+version = "0.2.0"
 description = "Ohm Stream Downloader — centre de contrôle auto-hébergé pour animes et séries VOSTFR"
 requires-python = ">=3.13"
 dependencies = [

scripts/install.sh

@@ -5,7 +5,7 @@
 #   bash scripts/install.sh                        (depuis un clone du dépôt)
 #   curl -fsSL <gitea>/raw/branch/main/scripts/install.sh | bash
 #
-# Options non interactives : --dir <chemin> --port <n> --skip-login
+# Options non interactives : --dir <chemin> --port <n>
 # (les valeurs passées en option court-circuitent les questions)
 #
 # Pose les questions (destination des animés, port), génère les secrets,
@@ -23,12 +23,11 @@ ok()   { printf "  ${GREEN}✔${RESET} %s\n" "$1"; }
 warn() { printf "  ${YELLOW}⚠${RESET} %s\n" "$1"; }
 die()  { printf "  ${RED}✖ %s${RESET}\n" "$1" >&2; exit 1; }
 
-ARG_DIR="" ARG_PORT="" ARG_SKIP_LOGIN=0
+ARG_DIR="" ARG_PORT=""
 while [ $# -gt 0 ]; do
     case "$1" in
         --dir)        ARG_DIR="$2"; shift 2;;
         --port)       ARG_PORT="$2"; shift 2;;
-        --skip-login) ARG_SKIP_LOGIN=1; shift;;
         *) shift;;
     esac
 done
@@ -74,21 +73,6 @@ if [ ! -f "$COMPOSE_FILE" ]; then
 fi
 ok "Projet prêt dans $(pwd)"
 
-# ---------------------------------------------------------------- registre privé
-step "Accès au registre privé ($REGISTRY_HOST)"
-if [ "$ARG_SKIP_LOGIN" = "1" ]; then
-    warn "--skip-login : connexion au registre ignorée"
-elif docker login "$REGISTRY_HOST" 2>/dev/null; then
-    ok "Connecté au registre"
-else
-    warn "Pas encore connecté — un compte Gitea avec accès lecture au dépôt est requis."
-    if [ "$(ask "Se connecter maintenant (docker login) ?" "o")" = "o" ]; then
-        docker login "$REGISTRY_HOST" || die "docker login a échoué — relance le script après t'être connecté."
-        ok "Connecté au registre"
-    else
-        die "Sans docker login, l'image ne pourra pas être tirée."
-    fi
-fi
 
 # ---------------------------------------------------------------- destination des épisodes
 step "Destination des épisodes téléchargés"

scripts/release.sh

@@ -1,13 +1,14 @@
 #!/usr/bin/env bash
 # Publie une version d'OhmStreaming :
-#   bump de version → commit + tag git → build de l'image Docker → push registre Gitea.
+#   bump de version → commit + tag git → build de l'image Docker → push GHCR (public) + Gitea.
 #
-# Usage : ./scripts/release.sh 0.2.0
+# Usage : ./scripts/release.sh 0.2.0 ["patchnote multi-lignes"]
 # Option : OHM_REGISTRY=git.lanro.eu/roman/ohm_streaming (défaut) pour viser un autre registre.
 set -euo pipefail
 
-VERSION="${1:?Usage : ./scripts/release.sh <version>   (ex. 0.2.0)}"
+VERSION="${1:?Usage : ./scripts/release.sh <version> [patchnote]   (ex : 0.2.0)}"
 REGISTRY="${OHM_REGISTRY:-git.lanro.eu/roman/ohm_streaming}"
+GHCR="${OHM_GHCR:-ghcr.io/aulyriusohm/ohm-streaming}"
 cd "$(dirname "$0")/.."
 
 [[ "$VERSION" =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]] || { echo "✖ « $VERSION » n'est pas un semver X.Y.Z"; exit 1; }
@@ -30,15 +31,17 @@ git add pyproject.toml uv.lock
 git commit -m "v$VERSION"
 
 echo "▶ 2/3  Tag et push git"
-git tag "$TAG"
+git tag -a "$TAG" -m "${2:-$TAG}"
 git push origin HEAD
 git push origin "$TAG"
 
-echo "▶ 3/3  Image Docker → $REGISTRY"
-docker build --build-arg VERSION="$VERSION" -t "$REGISTRY:$VERSION" -t "$REGISTRY:latest" .
-docker push "$REGISTRY:$VERSION"
-docker push "$REGISTRY:latest"
+echo "▶ 3/3  Image Docker → $GHCR (public) + $REGISTRY (Gitea)"
+docker build --build-arg VERSION="$VERSION" -t "$GHCR:$VERSION" -t "$GHCR:latest" .
+docker push "$GHCR:$VERSION" && docker push "$GHCR:latest"
+docker tag "$GHCR:$VERSION" "$REGISTRY:$VERSION"
+docker tag "$GHCR:latest" "$REGISTRY:latest"
+docker push "$REGISTRY:$VERSION" && docker push "$REGISTRY:latest"
 
 echo
-echo "✔ v$VERSION publiée ($REGISTRY:$VERSION et :latest)"
+echo "✔ v$VERSION publiée ($GHCR:$VERSION et :latest — public, miroir Gitea $REGISTRY)"
 echo "  Les instances la détecteront via Admin → Mise à jour → « Vérifier »."

tests/test_core.py

@@ -1,9 +1,12 @@
 """Tests unitaires : format interne, sanitisation, auth, registres scrapers."""
 
+import httpx
 import pytest
 
 from app import auth
+from app.scrapers import http as http_module
 from app.scrapers.base import (
+    ScrapeError,
     decode_internal_url,
     encode_internal_url,
     get_source,
@@ -133,3 +136,43 @@ def test_hoster_resolution():
     assert resolve_hoster("https://vidmoly.to/embed-x.html").name == "vidmoly"
     assert resolve_hoster("https://sendvid.com/embed/x").name == "sendvid"
     assert resolve_hoster("https://inconnu.example.com/v/1") is None
+
+
+# ------------------------------------------------------------ client HTTP
+
+
+class _StatusClient:
+    """Fake client HTTP : sert une série de codes statut puis un corps 200."""
+
+    def __init__(self, statuses: list[int]):
+        self._statuses = statuses
+        self.calls = 0
+
+    async def get(self, url, headers=None):
+        self.calls += 1
+        request = httpx.Request("GET", url)
+        status = self._statuses.pop(0) if self._statuses else 200
+        response = httpx.Response(status, request=request, text="ok")
+        response.raise_for_status()
+        return response
+
+
+async def test_fetch_fails_fast_on_definitive_4xx(monkeypatch):
+    client = _StatusClient([422])
+    monkeypatch.setattr("app.scrapers.http.get_client", lambda: client)
+    with pytest.raises(ScrapeError, match="422"):
+        await http_module.fetch("https://site.example/lecteur/prepare/1?content=2&episode=3")
+    assert client.calls == 1  # aucun retry sur erreur définitive
+
+
+@pytest.mark.parametrize("status", [429, 500, 503])
+async def test_fetch_retries_transitory_errors(monkeypatch, status):
+    client = _StatusClient([status, 200])
+    monkeypatch.setattr("app.scrapers.http.get_client", lambda: client)
+
+    async def no_sleep(_delay):
+        return None
+
+    monkeypatch.setattr("app.scrapers.http.asyncio.sleep", no_sleep)
+    assert await http_module.fetch("https://site.example/page", retries=2) == "ok"
+    assert client.calls == 2

tests/test_discover.py

@@ -154,7 +154,7 @@ async def test_latest_merges_and_sorts_by_release_date(monkeypatch):
 
     service = DiscoverService()
     monkeypatch.setattr(service, "_kitsu_match_for_title", fake_match)
-    result = await service.latest(limit=10)
+    result = (await service.latest_by_type(limit=10))["anime"]
 
     titles = [item["title"] for item in result]
     assert titles.count("Frieren VOSTFR") + titles.count("Frieren - Saison 1") == 1  # dédoublonné
@@ -186,11 +186,52 @@ async def test_latest_orders_by_kitsu_start_date(monkeypatch):
 
     service = DiscoverService()
     monkeypatch.setattr(service, "_kitsu_match_for_title", fake_match)
-    result = await service.latest(limit=10)
+    result = (await service.latest_by_type(limit=10))["anime"]
     assert [item["title"] for item in result] == ["Anime 1", "Anime 0", "Anime 2"]
     assert result[0]["rating"] is None
 
 
+async def test_latest_by_type_keeps_every_type(monkeypatch):
+    """Chaque type garde son rail complet : les animés nombreux ne vicient pas
+    le rail séries (les séries n'existent pas chez Kitsu, donc sans date)."""
+    import app.services.discover as discover_module
+
+    animes = [
+        SearchResult(source="vostfree", source_id=str(n), title=f"Animé {n}", url=f"https://a/{n}")
+        for n in range(1, 31)
+    ]
+    series = [
+        SearchResult(
+            source="french_stream", source_id=str(n), title=f"Série {n}",
+            url=f"https://s/{n}", media_type="serie",
+        )
+        for n in range(1, 10)
+    ]
+
+    class AnimeSource(_FakeSource):
+        name, label, base_url, media_types = "vostfree", "Vostfree", "https://a", ("anime",)
+
+    class SerieSource(_FakeSource):
+        name, label, base_url, media_types = "french_stream", "French-Stream", "https://s", ("serie", "film")
+
+    monkeypatch.setattr(discover_module, "all_sources", lambda: [AnimeSource(animes), SerieSource(series)])
+
+    async def enabled(name: str) -> bool:
+        return True
+
+    monkeypatch.setattr(discover_module, "is_source_enabled", enabled)
+
+    async def fake_match(title: str):
+        return None  # rien chez Kitsu → aucune date de sortie
+
+    service = DiscoverService()
+    monkeypatch.setattr(service, "_kitsu_match_for_title", fake_match)
+
+    rails = await service.latest_by_type(limit=24)
+    assert len(rails["anime"]) == 24  # tronqué à la limite
+    assert len(rails["serie"]) == 9  # les 9 séries intactes, malgré 30 animés
+    assert all(item["media_type"] == "serie" for item in rails["serie"])
+
 async def test_latest_skips_broken_source_and_uses_cache(monkeypatch):
     """Une source en échec disparaît sans erreur, et le TTL évite les re-scrapes."""
     import app.services.discover as discover_module
@@ -214,8 +255,8 @@ async def test_latest_skips_broken_source_and_uses_cache(monkeypatch):
     monkeypatch.setattr(discover_module, "is_source_enabled", enabled)
 
     service = DiscoverService()
-    assert await service.latest() == []
-    assert await service.latest() == []
+    assert await service.latest_by_type() == {"anime": [], "serie": [], "film": []}
+    assert await service.latest_by_type() == {"anime": [], "serie": [], "film": []}
     assert calls["n"] == 1  # deuxième appel servi depuis le cache TTL
 
 
@@ -256,16 +297,15 @@ async def history(monkeypatch):
         return ["Fantasy", "Adventure"] if anime_id == "46474" else []
 
     async def fake_kitsu_anime(params: dict) -> list[dict]:
-        slugs = frozenset(params["filter[categories]"].split(","))
         assert params["sort"] == "-userCount"
-        catalog = {
-            frozenset(["fantasy", "adventure", "action", "comedy"]): [
+        # Une catégorie par requête (le cumul Kitsu est un ET trop restrictif)
+        assert "," not in params["filter[categories]"]
+        return [
             {"title": "Helck", "kitsu_id": "999"},
             {"title": "Sousou no Frieren", "kitsu_id": "46474"},
             {"title": "Konosuba", "kitsu_id": "1"},
-            ],
-        }
-        return catalog.get(slugs, [])
+        ]
+
 
     service = DiscoverService()
     service.test_user_id = user_id  # pour les assertions du test
@@ -284,11 +324,78 @@ async def test_for_you_aggregates_genres_and_excludes_owned(history):
     assert "Sousou no Frieren" in titles
     assert "Konosuba" in titles
 
+async def test_for_you_excludes_owned_with_episode_markers(monkeypatch):
+    """Régression : « Titre - Saison 1 - E3 » possédé doit exclure le « Titre »
+    canonique recommandé (la normalisation retire saison ET épisode, en boucle)."""
+    cursor = await db.execute(
+        "INSERT INTO users (username, password_hash) VALUES (?, ?)",
+        ("reg-user", "x" * 64),
+    )
+    user_id = cursor.lastrowid
+    await db.execute(
+        "INSERT INTO downloads (source_key, video_url, title, status) VALUES (?,?,?,?)",
+        ("kr", "https://v/r", "The Eminence in Shadow - Saison 1 - E3", "done"),
+    )
+    await db.execute(
+        "INSERT INTO favorites (user_id, source, source_id, title, image_url, payload) "
+        "VALUES (?,?,?,?,?,?)",
+        (user_id, "vostfree", "r", "Favori R", None, '{"genres": ["Fantasy"]}'),
+    )
 
-async def test_for_you_empty_without_history(monkeypatch):
+    async def fake_match(title: str):
+        return None  # genres apportés par le favori
+
+    async def fake_kitsu_anime(params: dict) -> list[dict]:
+        return [
+            {"title": "The Eminence in Shadow", "kitsu_id": "1"},
+            {"title": "Autre Anime", "kitsu_id": "2"},
+        ]
+
+    service = DiscoverService()
+    monkeypatch.setattr(service, "_kitsu_match_for_title", fake_match)
+    monkeypatch.setattr(service, "_kitsu_anime", fake_kitsu_anime)
+    result = await service.for_you(user_id=user_id, limit=10)
+    titles = [item["title"] for item in result["items"]]
+    assert "The Eminence in Shadow" not in titles  # possédé (« - Saison 1 - E3 »)
+    assert "Autre Anime" in titles
+
+async def test_for_you_cold_start_without_history(monkeypatch):
+    """Aucun téléchargement/favori/Sonarr : amorçage signalé (non caché)."""
     service = DiscoverService()
     result = await service.for_you(user_id=42)
-    assert result == {"based_on": [], "items": []}
+    assert result == {"based_on": [], "items": [], "cold_start": True}
+    assert service._cache.get("for_you:42:20") is None  # pas de cache : bon marché, change au 1er téléchargement
+
+async def test_for_you_uses_series_page_genres_without_sonarr(monkeypatch):
+    """Sans Sonarr, les séries téléchargées donnent leurs genres via leur fiche
+    source (libellés FR convertis en catégories Kitsu)."""
+    cursor = await db.execute(
+        "INSERT INTO users (username, password_hash) VALUES (?, ?)",
+        ("serie-user", "x" * 64),
+    )
+    user_id = cursor.lastrowid
+    await db.execute(
+        "INSERT INTO downloads (source_key, video_url, page_url, title, status) VALUES (?,?,?,?,?)",
+        ("k9", "https://v/9", "https://french-stream.lat/15138132-stat-saison-5.html",
+         "STAT - Saison 5", "done"),
+    )
+
+    async def fake_no_match(title: str):
+        return None  # série réelle : aucun match Kitsu
+
+    async def fake_genres_of_page(self, page_url: str):
+        assert "15138132" in page_url
+        return ["Drame", "Médical"]  # Médical : sans équivalent Kitsu → ignoré
+
+    service = DiscoverService()
+    monkeypatch.setattr(service._kitsu, "search_anime", fake_no_match)
+    monkeypatch.setattr(
+        "app.scrapers.sources.french_stream.FrenchStreamScraper.genres_of_page",
+        fake_genres_of_page,
+    )
+    result = await service.for_you(user_id=user_id, limit=10)
+    assert result["based_on"] == ["Drama"]  # « Drame » converti, « Médical » écarté
+    assert result["items"]  # des recommandations issues de la catégorie drama
 
 
 
@@ -336,11 +443,18 @@ async def test_api_discover_requires_auth(client):
 
 
 async def test_api_discover_sections(client, admin_cookies, monkeypatch):
-    async def fake_latest(limit: int = 24):
-        return [
+    async def fake_latest_by_type(limit: int = 24):
+        return {
+            "anime": [
                 {"source": "vostfree", "label": "Vostfree", "source_id": "a", "title": "T",
-             "start_date": "2026-01-01", "status": "current", "rating": 8.1},
-        ]
+                 "media_type": "anime", "start_date": "2026-01-01", "status": "current", "rating": 8.1},
+            ],
+            "serie": [
+                {"source": "french_stream", "label": "French-Stream", "source_id": "b",
+                 "title": "S", "media_type": "serie"},
+            ],
+            "film": [],
+        }
 
     async def fake_must_watch(limit: int = 20):
         return [{"kitsu_id": "1", "title": "Attack on Titan", "rating": 8.5}]
@@ -348,19 +462,63 @@ async def test_api_discover_sections(client, admin_cookies, monkeypatch):
     async def fake_for_you(user_id: int, limit: int = 20):
         return {"based_on": ["Action"], "items": [{"kitsu_id": "2", "title": "X"}]}
 
-    monkeypatch.setattr(discover, "latest", fake_latest)
+    monkeypatch.setattr(discover, "latest_by_type", fake_latest_by_type)
     monkeypatch.setattr(discover, "must_watch", fake_must_watch)
     monkeypatch.setattr(discover, "for_you", fake_for_you)
 
     r = await client.get("/api/discover", cookies=admin_cookies)
     assert r.status_code == 200
     data = r.json()
-    assert data["latest"][0]["label"] == "Vostfree"
-    assert data["latest"][0]["start_date"] == "2026-01-01"
+    assert data["latest_anime"][0]["label"] == "Vostfree"
+    assert data["latest_anime"][0]["start_date"] == "2026-01-01"
+    assert data["latest_serie"][0]["label"] == "French-Stream"
     assert data["must_watch"][0]["title"] == "Attack on Titan"
     assert data["for_you"]["based_on"] == ["Action"]
 
 
+async def test_api_browse(client, admin_cookies, monkeypatch):
+    """Explorer : animés via Kitsu, séries via French-Stream, hors préférence → vide."""
+
+    async def fake_browse_anime(genre: str, limit: int = 20):
+        assert genre == "comedy"
+        return [{"kitsu_id": "1", "title": "Nichijou"}]
+
+    async def fake_browse_serie_film(media_type: str, genre: str, limit: int = 24):
+        assert (media_type, genre) == ("serie", "medical")
+        return [{"source": "french_stream", "source_id": "9", "title": "STAT - Saison 5",
+                 "media_type": "serie"}]
+
+    monkeypatch.setattr(discover, "browse_anime", fake_browse_anime)
+    monkeypatch.setattr(discover, "browse_serie_film", fake_browse_serie_film)
+
+    r = await client.get(
+        "/api/discover/browse", params={"type": "anime", "genre": "comedy"}, cookies=admin_cookies
+    )
+    assert r.json()["items"][0]["title"] == "Nichijou"
+
+    r = await client.get(
+        "/api/discover/browse", params={"type": "serie", "genre": "medical"}, cookies=admin_cookies
+    )
+    assert r.json()["items"][0]["title"] == "STAT - Saison 5"
+
+    # Préférence animés : le parcours séries renvoie vide sans même scraper
+    await client.put("/auth/preferences", json={"content_preference": "anime"}, cookies=admin_cookies)
+    r = await client.get(
+        "/api/discover/browse", params={"type": "serie", "genre": "medical"}, cookies=admin_cookies
+    )
+    assert r.json()["items"] == []
+
+
+async def test_api_genres_catalog(client, admin_cookies):
+    """Catalogue de genres : animés + séries/films French-Stream en mode « les deux »."""
+    r = await client.get("/api/discover/genres", cookies=admin_cookies)
+    assert r.status_code == 200
+    catalog = r.json()
+    assert {"anime", "serie", "film"} <= set(catalog)
+    assert {"key": "thriller", "label": "Thriller"} in catalog["serie"]
+    assert any(g["key"] == "comedies" for g in catalog["film"])
+
+
 async def test_discover_page_renders(client, admin_cookies):
     r = await client.get("/discover", cookies=admin_cookies)
     assert r.status_code == 200

tests/test_french_stream.py

@@ -0,0 +1,257 @@
+"""Tests du scraper French-Stream (séries/films VF-VOSTFR) sur fixtures réelles."""
+
+import json
+
+import pytest
+
+from app.scrapers.base import ScrapeError, get_source, import_all_scrapers
+from app.scrapers.sources.french_stream import FrenchStreamScraper
+
+# ------------------------------------------------------------ fixtures
+
+SEARCH_HTML = """
+<div class='search-item' onclick="location.href='/9562-breaking-bad-saison-2-streaming-complet-vf-vostfr.html'">
+  <div class='search-poster'><img src='https://image.tmdb.org/t/p/w400/bb.jpg' alt='Breaking Bad - Saison 2'></div>
+  <div class='search-info'><div class='search-title'>Breaking Bad - Saison 2</div></div>
+</div>
+<div class='search-item' onclick="location.href='/1022-inception-streaming-complet-vf-vostfr.html'">
+  <div class='search-poster'><img src='https://image.tmdb.org/t/p/w400/inc.jpg' alt='Inception'></div>
+  <div class='search-info'><div class='search-title'>Inception (2010)</div></div>
+</div>
+<div class='search-item'><div class='search-info'><div class='search-title'>Sans lien</div></div></div>
+"""
+
+SERIE_DETAILS_HTML = """
+<div id="film-data" data-newsid="9562" data-affiche="https://image.tmdb.org/t/p/w300/bb.jpg"></div>
+<div class="fmain"><div class="fcols fx-row"><div class="fmid">
+  <h1 id="s-title">Breaking Bad - Saison 2 </h1>
+  <div class="facts">
+    <span class="certification">-12</span> - <span class="release">2008 -</span>
+    <span class="genres">Drame, Crime</span> - <span class="runtime">48 min</span>
+  </div>
+  <div class="flist clearfix"><div class="fdesc"><p>Walter et Jesse montent leur propre affaire.</p></div></div>
+</div></div></div>
+<div class="fleft"><div class="fposter"><img src="https://image.tmdb.org/t/p/w400/bb.jpg" alt="Breaking Bad - Saison 2"></div></div>
+<div id="serie-config" style="display:none" data-title="Breaking Bad - Saison 2" data-news-id="9562"></div>
+"""
+
+FILM_DETAILS_HTML = """
+<div id="film-data" data-newsid="1022" data-affiche="https://image.tmdb.org/t/p/w300/inc.jpg"></div>
+<div class="fmain"><div class="fcols fx-row"><div class="fmid">
+  <h1 id="s-title"> Inception - 2010 <span class="tag release_date"></span> </h1>
+  <div class="facts">
+    <span class="genres"><a href="/index.php?do=xfsearch&amp;xf=Action">Action</a><a href="/index.php?do=xfsearch&amp;xf=Science-Fiction">Science-Fiction</a></span>
+    <span class="runtime">- 2h28</span>
+  </div>
+  <div class="fdesc clearfix slice-this" id="s-desc">
+    <p class="desc-text">Résumé du film Inception en streaming complet vf et vostfr hd vod gratuit</p>
+    Dom Cobb est un voleur expérimenté.
+  </div>
+</div></div></div>
+"""
+
+EP_DATA = {
+    "vf": {
+        "1": {"vidzy": "https://vidzy.cc/embed-aaa.html", "uqload": "https://uqload.vc/embed-bbb.html"},
+        "2": {"vidzy": "https://vidzy.cc/embed-ccc.html"},
+    },
+    "vostfr": {
+        "1": {"vidzy": "https://vidzy.cc/embed-ddd.html"},
+        "2": {"vidzy": "https://vidzy.cc/embed-eee.html"},
+    },
+    "vo": {},
+    "info": {
+        "1": {"title": "Traqués", "synopsis": "Walt et Jesse...", "poster": "https://img/ep1.jpg"},
+        "2": {"title": "Chasse à l'homme"},
+    },
+}
+
+FILM_API = {
+    "players": {
+        "vidzy": {
+            "default": "https://vidzy.live/embed-fff.html",
+            "vff": "https://vidzy.live/embed-fff.html",
+            "vostfr": "https://vidzy.live/embed-ggg.html",
+        }
+    }
+}
+
+LATEST_HTML = """
+<div class="short"><div class="short-in nl">
+  <a class="short-poster img-box with-mask" href="/index.php?newsid=15137274" alt="The Drop - Saison 1">
+    <img src="https://image.tmdb.org/t/p/w300/drop.jpg">
+  </a>
+</div></div>
+<div class="short"><div class="short-in nl">
+  <a class="short-poster img-box with-mask" href="/index.php?newsid=15139056" alt="La Rumeur">
+    <img src="https://image.tmdb.org/t/p/w300/rumeur.jpg">
+  </a>
+</div></div>
+"""
+
+
+def _soup(html: str):
+    from bs4 import BeautifulSoup
+
+    return BeautifulSoup(html, "lxml")
+
+
+# ------------------------------------------------------------ source
+
+
+async def test_search(monkeypatch):
+    async def fake_fetch(url, **kwargs):
+        assert url.endswith("/engine/ajax/search.php")
+        assert kwargs["data"] == {"query": "breaking bad", "page": "1"}
+        return SEARCH_HTML
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch", fake_fetch)
+    results = await FrenchStreamScraper().search("breaking bad")
+    assert len(results) == 2
+    serie, film = results
+    assert serie.source_id == "9562"
+    assert serie.media_type == "serie"
+    assert serie.title == "Breaking Bad - Saison 2"
+    assert serie.url == "https://french-stream.lat/9562-breaking-bad-saison-2-streaming-complet-vf-vostfr.html"
+    assert serie.image_url == "https://image.tmdb.org/t/p/w400/bb.jpg"
+    assert film.media_type == "film"
+    assert film.source_id == "1022"
+
+
+async def test_latest(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        assert url.endswith("/series/")
+        return _soup(LATEST_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    results = await FrenchStreamScraper().latest()
+    assert [(r.source_id, r.media_type) for r in results] == [
+        ("15137274", "serie"),
+        ("15139056", "film"),
+    ]
+    assert results[0].title == "The Drop - Saison 1"
+
+
+async def test_browse(monkeypatch):
+    """Parcours par genre : page du genre scrapée, type forcé sur les résultats."""
+
+    async def fake_fetch_soup(url, **kwargs):
+        assert url.endswith("/medical-series-/")
+        return _soup(LATEST_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    results = await FrenchStreamScraper().browse("serie", "medical")
+    # LATEST_HTML contient un film sans « Saison » : le type est forcé à serie
+    assert [r.media_type for r in results] == ["serie", "serie"]
+    assert results[0].title == "The Drop - Saison 1"
+
+
+async def test_browse_unknown_category():
+    with pytest.raises(ScrapeError, match="Catégorie inconnue"):
+        await FrenchStreamScraper().browse("film", "inexistant")
+
+async def test_get_details_serie(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        assert "newsid=9562" in url
+        return _soup(SERIE_DETAILS_HTML)
+
+    async def fake_fetch(url, **kwargs):
+        assert "ep-data.php?id=9562" in url
+        return json.dumps(EP_DATA)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch", fake_fetch)
+    details = await FrenchStreamScraper().get_details("9562")
+    assert details.media_type == "serie"
+    assert details.title == "Breaking Bad - Saison 2"
+    assert details.year == 2008
+    assert details.genres == ["Drame", "Crime"]
+    assert details.synopsis == "Walter et Jesse montent leur propre affaire."
+    assert details.image_url == "https://image.tmdb.org/t/p/w400/bb.jpg"
+    assert details.episode_count == 4
+    vf1 = details.episodes[0]
+    assert (vf1.season, vf1.number, vf1.version, vf1.title) == (2, 1.0, "vf", "Traqués")
+    assert vf1.url.endswith("#vf-1")
+    versions = {(e.number, e.version) for e in details.episodes}
+    assert versions == {(1.0, "vf"), (1.0, "vostfr"), (2.0, "vf"), (2.0, "vostfr")}
+
+
+async def test_get_details_film(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        return _soup(FILM_DETAILS_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    details = await FrenchStreamScraper().get_details("1022")
+    assert details.media_type == "film"
+    assert details.title == "Inception - 2010"
+    assert details.year == 2010
+    assert details.genres == ["Action", "Science-Fiction"]
+    assert "Résumé du film" not in details.synopsis
+    assert details.episode_count == 1
+    assert details.episodes[0].url.endswith("#film")
+
+
+async def test_list_episodes_film(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        return _soup(FILM_DETAILS_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    episodes = await FrenchStreamScraper().list_episodes("1022")
+    assert len(episodes) == 1
+    assert episodes[0].number == 1
+
+
+async def test_extract_embed_links_episode(monkeypatch):
+    async def fake_fetch(url, **kwargs):
+        return json.dumps(EP_DATA)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch", fake_fetch)
+    links = await FrenchStreamScraper().extract_embed_links(
+        "https://french-stream.lat/index.php?newsid=9562#vf-1"
+    )
+    assert links == ["https://vidzy.cc/embed-aaa.html", "https://uqload.vc/embed-bbb.html"]
+
+
+async def test_extract_embed_links_film_dedup(monkeypatch):
+    async def fake_fetch(url, **kwargs):
+        assert "film_api.php?id=1022" in url
+        return json.dumps(FILM_API)
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch", fake_fetch)
+    links = await FrenchStreamScraper().extract_embed_links(
+        "https://french-stream.lat/index.php?newsid=1022#film"
+    )
+    assert links == ["https://vidzy.live/embed-fff.html", "https://vidzy.live/embed-ggg.html"]
+
+
+async def test_extract_embed_links_bad_fragment():
+    with pytest.raises(ScrapeError):
+        await FrenchStreamScraper().extract_embed_links(
+            "https://french-stream.lat/index.php?newsid=9562#nimportequoi"
+        )
+
+
+async def test_extract_embed_links_invalid_json(monkeypatch):
+    async def fake_fetch(url, **kwargs):
+        return "<html>oops</html>"
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch", fake_fetch)
+    with pytest.raises(ScrapeError):
+        await FrenchStreamScraper().extract_embed_links(
+            "https://french-stream.lat/index.php?newsid=9562#vf-1"
+        )
+
+
+async def test_unknown_detail_page(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        return _soup("<html><body>404</body></html>")
+
+    monkeypatch.setattr("app.scrapers.sources.french_stream.fetch_soup", fake_fetch_soup)
+    with pytest.raises(ScrapeError):
+        await FrenchStreamScraper().get_details("999999999")
+
+
+def test_registered():
+    import_all_scrapers()
+    source = get_source("french_stream")
+    assert source.media_types == ("serie", "film")

tests/test_preferences.py

@@ -0,0 +1,163 @@
+"""Tests de la préférence de contenus par utilisateur (animés / séries / les deux)."""
+
+from app.scrapers.base import SearchResult
+
+
+class MixedSource:
+    """Source factice renvoyant animés, séries et films."""
+
+    name = "mixed"
+    label = "Mixed"
+    base_url = "https://mixed.example"
+    media_types = ("anime", "serie")
+
+    async def search(self, query: str) -> list[SearchResult]:
+        return [
+            SearchResult(source="mixed", source_id="a1", title="Animé X", url="https://mixed.example/a1"),
+            SearchResult(source="mixed", source_id="s1", title="Série Y", url="https://mixed.example/s1", media_type="serie"),
+            SearchResult(source="mixed", source_id="f1", title="Film Z", url="https://mixed.example/f1", media_type="film"),
+        ]
+
+    async def latest(self) -> list[SearchResult]:
+        return await self.search("")
+
+
+class AnimeOnlySource:
+    name = "animeonly"
+    label = "AnimeOnly"
+    base_url = "https://animeonly.example"
+    media_types = ("anime",)
+
+    async def search(self, query: str) -> list[SearchResult]:
+        return [
+            SearchResult(source="animeonly", source_id="a2", title="Animé W", url="https://animeonly.example/a2")
+        ]
+
+    async def latest(self) -> list[SearchResult]:
+        return await self.search("")
+
+
+async def _patch_sources(monkeypatch, *sources):
+    import app.routers.search as search_module
+
+    monkeypatch.setattr(search_module, "all_sources", lambda: list(sources))
+    import app.services.discover as discover_module
+
+    async def enabled(name):
+        return True
+
+    monkeypatch.setattr(discover_module, "is_source_enabled", enabled)
+    discover_module.discover._cache.clear()
+    monkeypatch.setattr(discover_module, "all_sources", lambda: list(sources))
+
+
+# ---------------------------------------------------------------- endpoints
+
+
+async def test_me_exposes_default_preference(client, admin_cookies):
+    r = await client.get("/auth/me", cookies=admin_cookies)
+    assert r.status_code == 200
+    assert r.json()["content_preference"] == "both"
+
+
+async def test_preference_roundtrip(client, admin_cookies):
+    r = await client.put(
+        "/auth/preferences", json={"content_preference": "serie"}, cookies=admin_cookies
+    )
+    assert r.status_code == 200
+    assert (await client.get("/auth/me", cookies=admin_cookies)).json()["content_preference"] == "serie"
+
+
+async def test_preference_rejects_invalid_value(client, admin_cookies):
+    r = await client.put(
+        "/auth/preferences", json={"content_preference": "documentaire"}, cookies=admin_cookies
+    )
+    assert r.status_code == 422
+
+
+# ---------------------------------------------------------------- filtrage recherche
+
+
+async def test_search_both_keeps_everything(client, admin_cookies, monkeypatch):
+    await _patch_sources(monkeypatch, MixedSource(), AnimeOnlySource())
+    r = await client.get("/api/search", params={"q": "test"}, cookies=admin_cookies)
+    assert r.status_code == 200
+    assert r.json()["count"] == 4  # animés + série + film
+
+
+async def test_search_anime_filters_serie(client, admin_cookies, monkeypatch):
+    await _patch_sources(monkeypatch, MixedSource(), AnimeOnlySource())
+    await client.put("/auth/preferences", json={"content_preference": "anime"}, cookies=admin_cookies)
+    r = await client.get("/api/search", params={"q": "test"}, cookies=admin_cookies)
+    types = {x["media_type"] for x in r.json()["results"]}
+    assert types == {"anime"}  # le mode animés reste sur l'animation (films réels → côté séries)
+
+
+async def test_search_serie_skips_anime_sources(client, admin_cookies, monkeypatch):
+    """Préférence séries : les sources anime-only ne sont même pas interrogées."""
+    await _patch_sources(monkeypatch, MixedSource(), AnimeOnlySource())
+    await client.put("/auth/preferences", json={"content_preference": "serie"}, cookies=admin_cookies)
+
+    queried = []
+
+    class SpySource(AnimeOnlySource):
+        async def search(self, query: str) -> list[SearchResult]:
+            queried.append(self.name)
+            return await super().search(query)
+
+    import app.routers.search as search_module
+
+    monkeypatch.setattr(search_module, "all_sources", lambda: [MixedSource(), SpySource()])
+
+    r = await client.get("/api/search", params={"q": "test"}, cookies=admin_cookies)
+    assert r.status_code == 200
+    assert queried == []  # source anime-only ignorée
+    assert {x["media_type"] for x in r.json()["results"]} == {"serie", "film"}
+
+
+# ---------------------------------------------------------------- filtrage découverte
+
+
+async def test_discover_latest_filtered_by_preference(client, admin_cookies, monkeypatch):
+    """Rails par type : chaque mode n'expose que ses rails (les films suivent les séries)."""
+    await _patch_sources(monkeypatch, MixedSource())
+
+    r = await client.get("/api/discover", cookies=admin_cookies)
+    data = r.json()
+    assert {i["media_type"] for i in data["latest_anime"]} == {"anime"}
+    assert {i["media_type"] for i in data["latest_serie"]} == {"serie", "film"}
+
+    await client.put("/auth/preferences", json={"content_preference": "serie"}, cookies=admin_cookies)
+    data = (await client.get("/api/discover", cookies=admin_cookies)).json()
+    assert data["latest_anime"] == []
+    assert {i["media_type"] for i in data["latest_serie"]} == {"serie", "film"}
+
+    await client.put("/auth/preferences", json={"content_preference": "anime"}, cookies=admin_cookies)
+    data = (await client.get("/api/discover", cookies=admin_cookies)).json()
+    assert {i["media_type"] for i in data["latest_anime"]} == {"anime"}
+    assert data["latest_serie"] == []
+
+async def test_discover_hides_kitsu_sections_in_serie_mode(client, admin_cookies, monkeypatch):
+    """Incontournables et Pour toi (Kitsu = animés) disparaissent en mode séries."""
+    import app.services.discover as discover_module
+
+    await _patch_sources(monkeypatch, MixedSource())
+
+    async def fake_must_watch(limit=20):
+        return [{"kitsu_id": 1, "title": "Animé top"}]
+
+    async def fake_for_you(user_id=0, limit=20):
+        return {"based_on": ["Action"], "items": [{"kitsu_id": 2}]}
+
+    monkeypatch.setattr(discover_module.discover, "must_watch", fake_must_watch)
+    monkeypatch.setattr(discover_module.discover, "for_you", fake_for_you)
+
+    r = await client.get("/api/discover", cookies=admin_cookies)
+    assert len(r.json()["must_watch"]) == 1
+    assert len(r.json()["for_you"]["items"]) == 1
+
+    await client.put("/auth/preferences", json={"content_preference": "serie"}, cookies=admin_cookies)
+    data = (await client.get("/api/discover", cookies=admin_cookies)).json()
+    assert data["must_watch"] == []
+    assert data["for_you"] == {"based_on": [], "items": []}
+

tests/test_torznab.py

@@ -128,9 +128,26 @@ async def test_torznab_tvsearch_all_episodes(client, apikey, fake_source):
     assert "Frieren S02E01 VOSTFR WEB-DL" in titles
 
 
-async def test_torznab_search_requires_query(client, apikey):
+async def test_torznab_search_without_query_serves_latest(client, apikey, fake_source):
+    # Sans q : flux RSS des nouveautés (RSS sync / test Sonarr), pas une erreur 400.
+    async def fake_latest():
+        return [
+            SearchResult(
+                source="fake",
+                source_id="frieren-1",
+                title="Frieren",
+                url="https://fake.example/frieren",
+            )
+        ]
+
+    fake_source.latest = fake_latest
     r = await client.get("/torznab/api", params={"t": "search", "apikey": apikey})
-    assert r.status_code == 400
+    assert r.status_code == 200
+    root = ET.fromstring(r.text)
+    items = root.findall(".//item")
+    assert len(items) == 1
+    # seul le dernier épisode de la nouveauté est publié (S02E01 > S01E02 > S01E01)
+    assert items[0].find("title").text == "Frieren S02E01 VOSTFR WEB-DL"
 
 
 # ---------------------------------------------------------------- grab

tests/test_update.py

@@ -3,8 +3,8 @@
 import httpx
 import pytest
 
+from app.config import get_settings
 from app.services import update as update_service
-from app.services.settings import get_update_config
 from app.version import get_version
 
 
@@ -15,6 +15,12 @@ def reset_update_cache():
     update_service.invalidate_cache()
 
 
+def _patch_gitea(monkeypatch, url: str = "https://git.example", repo: str = "roman/ohm") -> None:
+    settings = get_settings()
+    monkeypatch.setattr(settings, "gitea_url", url)
+    monkeypatch.setattr(settings, "gitea_repo", repo)
+
+
 # ---------------------------------------------------------------- semver / version
 
 
@@ -41,18 +47,9 @@ def test_get_version_env_override(monkeypatch):
 # ---------------------------------------------------------------- détection Gitea
 
 
-async def test_fetch_latest_without_repo():
-    from app.services.settings import set_update_config
-
-    await set_update_config("https://git.example", "", "tok")
-    assert await update_service.fetch_latest_version(force=True) is None
-
-
-async def test_fetch_latest_public_repo_no_token(monkeypatch):
-    """Dépôt public : pas de jeton → appel sans en-tête Authorization."""
-    from app.services.settings import set_update_config
-
-    await set_update_config("https://git.example", "roman/ohm", "")
+async def test_fetch_latest_no_auth_header(monkeypatch):
+    """Dépôt public : appel sans en-tête Authorization."""
+    _patch_gitea(monkeypatch)
 
     class FakeResponse:
         def raise_for_status(self) -> None:
@@ -73,17 +70,20 @@ async def test_fetch_latest_public_repo_no_token(monkeypatch):
 
 
 async def test_fetch_latest_picks_highest_semver(monkeypatch):
-    """Avec jeton : en-tête Authorization + plus haut semver retenu, puis cache."""
-    from app.services.settings import set_update_config
-
-    await set_update_config("https://git.example", "roman/ohm", "tok")
+    """Plus haut semver retenu + patchnote du tag, puis cache."""
+    _patch_gitea(monkeypatch)
 
     class FakeResponse:
         def raise_for_status(self) -> None:
             pass
 
         def json(self) -> list[dict]:
-            return [{"name": "v0.1.0"}, {"name": "v0.2.3"}, {"name": "v1.0.0-rc"}, {"name": "divers"}]
+            return [
+                {"name": "v0.1.0", "message": "ancien"},
+                {"name": "v0.2.3", "message": "correctifs\n"},
+                {"name": "v1.0.0-rc"},
+                {"name": "divers"},
+            ]
 
     calls: list[tuple[str, dict]] = []
 
@@ -92,21 +92,18 @@ async def test_fetch_latest_picks_highest_semver(monkeypatch):
         return FakeResponse()
 
     monkeypatch.setattr(httpx.AsyncClient, "get", fake_get)
-    latest = await update_service.fetch_latest_version(force=True)
+    status = await update_service.status()
 
-    assert latest == "v0.2.3"
-    assert calls == [
-        ("https://git.example/api/v1/repos/roman/ohm/tags?limit=20", {"Authorization": "token tok"})
-    ]
+    assert status["latest"] == "v0.2.3"
+    assert status["notes"] == "correctifs"
+    assert calls == [("https://git.example/api/v1/repos/roman/ohm/tags?limit=20", {})]
     # puis servi par le cache (plus d'appel réseau)
     assert await update_service.fetch_latest_version() == "v0.2.3"
     assert len(calls) == 1
 
 
 async def test_fetch_latest_network_error_degrades(monkeypatch):
-    from app.services.settings import set_update_config
-
-    await set_update_config("https://git.example", "roman/ohm", "tok")
+    _patch_gitea(monkeypatch)
 
     async def fake_get(self: httpx.AsyncClient, url: str, headers: dict | None = None):
         raise httpx.ConnectError("injoignable")
@@ -148,33 +145,13 @@ async def test_update_status_defaults(client, admin_cookies, monkeypatch):
     r = await client.get("/api/admin/update", cookies=admin_cookies)
     assert r.status_code == 200
     data = r.json()
-    assert data["configured"] is True and data["docker"] is False  # repo par défaut présent
     assert data["latest"] is None and data["update_available"] is False
-    assert data["config"]["gitea_url"] == "https://git.lanro.eu"
-    assert data["config"]["repo"] == "Roman/ohm_streaming"
+    assert data["docker"] is False and data["notes"] is None
 
 
-async def test_update_save_config_normalizes(client, admin_cookies):
-    r = await client.put(
-        "/api/admin/update",
-        json={"gitea_url": "https://git.example/", "repo": "/roman/ohm/", "token": " tok "},
-        cookies=admin_cookies,
-    )
-    assert r.status_code == 200
-    assert await get_update_config() == {
-        "gitea_url": "https://git.example",
-        "repo": "roman/ohm",
-        "token": "tok",
-    }
-
-
-async def test_update_check_degrades_gracefully(client, admin_cookies):
-    # Gitea configuré mais injoignable : le check répond quand même (latest=None)
-    await client.put(
-        "/api/admin/update",
-        json={"gitea_url": "https://git.inexistant", "repo": "roman/ohm", "token": "tok"},
-        cookies=admin_cookies,
-    )
+async def test_update_check_degrades_gracefully(client, admin_cookies, monkeypatch):
+    # Gitea injoignable : le check répond quand même (latest=None)
+    _patch_gitea(monkeypatch, url="https://git.inexistant", repo="roman/ohm")
     r = await client.post("/api/admin/update/check", cookies=admin_cookies)
     assert r.status_code == 200 and r.json()["latest"] is None
 

tests/test_voiranime.py

@@ -0,0 +1,238 @@
+"""Tests du scraper VoirAnime (source + extracteur prepare) sur fixtures HTML réelles."""
+
+import base64
+import json
+
+import pytest
+
+from app.scrapers.base import ScrapeError, get_source, import_all_scrapers, resolve_hoster
+from app.scrapers.hosters.voiranime import VoirAnimeExtractor
+from app.scrapers.sources.voiranime import VoirAnimeScraper
+
+# ------------------------------------------------------------ fixtures HTML
+
+SEARCH_HTML = """
+<div class="catalogue-grid">
+  <article class="catalogue-card catalogue-media-card">
+    <a href="/catalogue/one-piece" class="catalogue-poster">
+      <img src="https://img.example/one-piece.jpg" alt="one piece">
+    </a>
+  </article>
+  <article class="catalogue-card catalogue-media-card">
+    <a href="/catalogue/one-punch-man" class="catalogue-poster">
+      <img src="https://img.example/opm.jpg" alt="one punch man">
+    </a>
+  </article>
+  <article class="catalogue-card catalogue-media-card">
+    <a href="/catalogue/one-outs" class="catalogue-poster">
+      <img src="/assets/img/placeholders/content-placeholder.svg" alt="one outs">
+    </a>
+  </article>
+</div>
+"""
+
+DETAILS_HTML = """
+<h1>solo leveling</h1>
+<div class="media-detail-meta"><span>26 épisodes</span></div>
+<img src="https://img.example/solo.jpg" alt="Affiche de solo leveling">
+<div class="media-detail-story">
+  <span class="media-detail-section-label">Synopsis</span>
+  <p> Sung Jinwoo, chasseur le plus faible. </p>
+</div>
+<a href="/lecteur/965/29054" class="detail-episode-card" data-season="1">
+  <small>S1 · E1</small><span>Episode 1</span>
+</a>
+<a href="/lecteur/965/29055" class="detail-episode-card" data-season="1">
+  <small>S1 · E2</small><span>Episode 2</span>
+</a>
+<a href="/lecteur/965/29060" class="detail-episode-card" data-season="2">
+  <small>S2 · E1</small><span>Episode 1</span>
+</a>
+<a href="/lecteur/965/29061" class="detail-episode-card" data-season="2">
+  <small>S2 · E2.5</small><span>Episode 2.5</span>
+</a>
+"""
+
+PLAYER_PAGE_1 = """
+<main class="tv-player" data-player
+  data-content-id="1626" data-episode-id="46416" data-source-id="169261"
+  data-prepare-url="/lecteur/prepare/169261?content=1626&amp;episode=46416"
+  data-fallback-source-url="/lecteur/1626/46416?lang=vostfr&amp;source=169262&amp;previous_source=169261">
+</main>
+"""
+
+# fin de chaîne : le fallback repointe vers la source 1 (aller-retour observé en live)
+PLAYER_PAGE_2 = """
+<main class="tv-player" data-player
+  data-content-id="1626" data-episode-id="46416" data-source-id="169262"
+  data-prepare-url="/lecteur/prepare/169262?content=1626&amp;episode=46416"
+  data-fallback-source-url="/lecteur/1626/46416?lang=vostfr&amp;source=169261&amp;previous_source=169262">
+</main>
+"""
+
+HOME_HTML = """
+<div class="carousel" id="carousel-new">
+  <article class="content-card">
+    <a href="/catalogue/you-and-i-are-polar-opposites" class="card-poster">
+      <img src="https://img.example/polar.jpg" alt="you and i are polar opposites">
+    </a>
+  </article>
+</div>
+<div class="carousel" id="carousel-added">
+  <article class="content-card">
+    <a href="/catalogue/you-and-i-are-polar-opposites" class="card-poster">
+      <img src="https://img.example/polar.jpg" alt="you and i are polar opposites">
+    </a>
+  </article>
+  <article class="content-card">
+    <a href="/catalogue/solo-leveling" class="card-poster">
+      <img src="https://img.example/solo.jpg" alt="solo leveling">
+    </a>
+  </article>
+</div>
+"""
+
+
+def _soup(html: str):
+    from bs4 import BeautifulSoup
+
+    return BeautifulSoup(html, "lxml")
+
+
+def _prepare_payload(url: str, referer: str) -> str:
+    raw = json.dumps({"url": url, "referer": referer, "kind": "file"}).encode()
+    return base64.b64encode(raw).decode()
+
+
+# ------------------------------------------------------------ source
+
+
+async def test_search_parses_catalogue_cards(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        assert "q=one+piece" in url
+        return _soup(SEARCH_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.voiranime.fetch_soup", fake_fetch_soup)
+    results = await VoirAnimeScraper().search("one piece")
+    assert [r.source_id for r in results] == ["one-piece", "one-punch-man", "one-outs"]
+    assert results[0].title == "one piece"
+    assert results[0].url == "https://voiranime.xyz/catalogue/one-piece"
+    assert results[0].image_url == "https://img.example/one-piece.jpg"
+    assert results[2].image_url == "/assets/img/placeholders/content-placeholder.svg"
+
+
+async def test_get_details_and_episodes(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        assert url.endswith("/catalogue/solo-leveling")
+        return _soup(DETAILS_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.voiranime.fetch_soup", fake_fetch_soup)
+    scraper = VoirAnimeScraper()
+    details = await scraper.get_details("solo-leveling")
+    assert details.title == "solo leveling"
+    assert details.episode_count == 26
+    assert details.image_url == "https://img.example/solo.jpg"
+    assert details.synopsis.startswith("Sung Jinwoo")
+
+    episodes = details.episodes
+    assert len(episodes) == 4
+    assert (episodes[0].season, episodes[0].number) == (1, 1.0)
+    assert (episodes[2].season, episodes[2].number) == (2, 1.0)
+    assert (episodes[3].season, episodes[3].number) == (2, 2.5)
+    assert episodes[0].url == "https://voiranime.xyz/lecteur/965/29054"
+
+    standalone = await scraper.list_episodes("solo-leveling")
+    assert len(standalone) == len(episodes)
+
+
+async def test_extract_embed_links_follows_fallback_chain(monkeypatch):
+    pages = {
+        "https://voiranime.xyz/lecteur/1626/46416": PLAYER_PAGE_1,
+        "https://voiranime.xyz/lecteur/1626/46416?lang=vostfr&source=169262&previous_source=169261": PLAYER_PAGE_2,
+    }
+
+    async def fake_fetch_soup(url, **kwargs):
+        return _soup(pages[url])
+
+    monkeypatch.setattr("app.scrapers.sources.voiranime.fetch_soup", fake_fetch_soup)
+    urls = await VoirAnimeScraper().extract_embed_links("https://voiranime.xyz/lecteur/1626/46416")
+    assert urls == [
+        "https://voiranime.xyz/lecteur/prepare/169261?content=1626&episode=46416",
+        "https://voiranime.xyz/lecteur/prepare/169262?content=1626&episode=46416",
+    ]
+
+
+async def test_extract_embed_links_no_player_raises(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        return _soup("<html><body>pas de lecteur</body></html>")
+
+    monkeypatch.setattr("app.scrapers.sources.voiranime.fetch_soup", fake_fetch_soup)
+    with pytest.raises(ScrapeError):
+        await VoirAnimeScraper().extract_embed_links("https://voiranime.xyz/lecteur/1/2")
+
+
+async def test_latest_merges_carousels_dedup(monkeypatch):
+    async def fake_fetch_soup(url, **kwargs):
+        assert url == "https://voiranime.xyz/"
+        return _soup(HOME_HTML)
+
+    monkeypatch.setattr("app.scrapers.sources.voiranime.fetch_soup", fake_fetch_soup)
+    results = await VoirAnimeScraper().latest()
+    assert [r.source_id for r in results] == ["you-and-i-are-polar-opposites", "solo-leveling"]
+
+
+# ------------------------------------------------------------ extracteur hoster
+
+
+async def test_hoster_decodes_prepare_payload(monkeypatch):
+    payload = _prepare_payload(
+        "https://video.sibnet.ru/v/abc123/6235096.mp4", "https://video.sibnet.ru/"
+    )
+
+    async def fake_fetch(url, **kwargs):
+        assert "X-Requested-With" in kwargs.get("headers", {})
+        assert kwargs.get("referer") == "https://voiranime.xyz/lecteur/1626/46416"
+        return json.dumps(
+            {"success": True, "media_type": "mp4", "stream_url": f"/proxy/media?payload={payload}"}
+        )
+
+    monkeypatch.setattr("app.scrapers.hosters.voiranime.fetch", fake_fetch)
+    link = await VoirAnimeExtractor().extract(
+        "https://voiranime.xyz/lecteur/prepare/169261?content=1626&episode=46416"
+    )
+    assert link.url == "https://video.sibnet.ru/v/abc123/6235096.mp4"
+    assert link.hoster == "voiranime"
+    assert link.headers == {"Referer": "https://video.sibnet.ru/"}
+    assert link.is_hls is False
+
+
+async def test_hoster_marks_hls(monkeypatch):
+    payload = _prepare_payload("https://cdn.example/hls/master.m3u8", "https://cdn.example/")
+
+    async def fake_fetch(url, **kwargs):
+        return json.dumps(
+            {"success": True, "media_type": "hls", "stream_url": f"/proxy/media?payload={payload}"}
+        )
+
+    monkeypatch.setattr("app.scrapers.hosters.voiranime.fetch", fake_fetch)
+    link = await VoirAnimeExtractor().extract("https://voiranime.xyz/lecteur/prepare/1?content=2&episode=3")
+    # HLS signé pour le backend du site → on sert l'URL proxy du site, pas le CDN
+    assert link.url == f"https://voiranime.xyz/proxy/media?payload={payload}"
+    assert link.is_hls is True
+    assert link.headers == {"Referer": "https://voiranime.xyz/"}
+
+
+async def test_hoster_prepare_failure_raises(monkeypatch):
+    async def fake_fetch(url, **kwargs):
+        return json.dumps({"success": False, "error": "Flux extrait inaccessible"})
+
+    monkeypatch.setattr("app.scrapers.hosters.voiranime.fetch", fake_fetch)
+    with pytest.raises(ScrapeError, match="Flux extrait inaccessible"):
+        await VoirAnimeExtractor().extract("https://voiranime.xyz/lecteur/prepare/1?content=2&episode=3")
+
+
+def test_source_and_hoster_registered():
+    import_all_scrapers()
+    assert get_source("voiranime").label == "VoirAnime"
+    extractor = resolve_hoster("https://voiranime.xyz/lecteur/prepare/169261?content=1&episode=2")
+    assert extractor is not None and extractor.name == "voiranime"

uv.lock

@@ -774,7 +774,7 @@ wheels = [
 
 [[package]]
 name = "ohm-stream"
-version = "0.1.0"
+version = "0.2.0"
 source = { virtual = "." }
 dependencies = [
     { name = "aiosqlite" },