Files
edde746 05fd622968 feat(emby): add Emby as a MediaBrowser backend alongside Jellyfin
Emby is Jellyfin's upstream ancestor and speaks a near-identical MediaBrowser
API, so the existing Jellyfin stack is parameterised by a `MediaBrowserDialect`
rather than forked. `JellyfinClient`, its auth service, endpoint discovery, LAN
discovery, and the add/edit connection screens all take the dialect and keep one
implementation; `MediaBackend.emby` and `ConnectionKind.emby` carry it through
the neutral models, the Drift `kind` discriminator, downloads, and caches.

Every divergence below was measured against a live Emby 4.9.5 server, not
inferred from documentation, and each is documented at its capability getter.
Jellyfin's request strings stay byte-identical so nothing about its behaviour
changes.

Routes and auth
- Emby only accepts the pre-10.9 user-scoped item routes (`/Users/{id}/Items/…`,
  `/Users/{id}/PlayedItems/…`, `/Users/{id}/FavoriteItems/…`); the unprefixed
  forms Jellyfin 10.11 added return 404.
- The API is also served under a legacy `/emby` prefix, and both dialects accept
  the token as `X-Emby-Token` or `api_key=`.
- Emby answers only its own LAN discovery datagram ("who is EmbyServer?") and
  ignores Jellyfin's; its default HTTPS port is 8920.
- No `/QuickConnect` route exists, so Quick Connect stays Jellyfin-only.

Row fields Emby withholds
- `ProductionYear`, `OfficialRating`, `PremiereDate` and `DateCreated` are absent
  from list rows unless named in `Fields`, which would otherwise strip the year
  and age-rating badge from every card in the app.
- `UserData.LastPlayedDate` never appears on a list row under `Fields=UserData`,
  `EnableUserData=true` or the user-scoped `Ids=` form — only on the single-item
  detail route, or when the Emby-specific `UserDataLastPlayedDate` token is
  requested. Without it every recency-ordered surface silently degrades to
  library-add time, and `JellyfinApiCache.applyWatchState` stamps
  `DateTime.now()` on watched rows, so an offline watch-state pull would rewrite
  the cached play time of everything it walked.

Continue Watching and Next Up
- Emby computes Next Up per series only: the library-wide `/Shows/NextUp` query
  returns nothing under every parameter combination tried. The shelf is
  therefore reconstructed from a played-episode recency scan plus one
  `/Shows/NextUp?SeriesId=` per distinct series, bounded by a shared wall clock
  that covers the scan as well — per-request timeouts cannot bound the pass
  because `MediaServerHttpClient` times the connect and receive phases
  independently. Rows are stamped with their series' newest play from the same
  response that ordered them, so no per-series enrichment request is needed.
- `/Shows/NextUp` ignores `NextUpDateCutoff`, and no server-side played-date
  filter exists to delegate to (`MinDatePlayed` and `MinDateLastPlayed` are
  ignored; `MinDateLastSaved`, `MinDateCreated` and `MinPremiereDate` filter
  unrelated dates), so the 365-day window is applied to the scanned dates.
- The resume route returns items with no saved position, including plain next
  episodes, so the Emby resume leg reads from `/Items?Filters=IsResumable`.
- Emby is ahead of Jellyfin in one place: `/Users/{id}/Items/{id}/HideFromResume`
  makes Continue Watching removal a real capability.

Everything else
- `/Sessions/Playing` and `/Sessions/Playing/Progress` reject a body with no
  `PlaySessionId` (HTTP 400), so playback reporting always sends one.
- Passing any `MediaTypes` value to the playlist query returns an empty list.
- There is no aggregate `/Items/Filters` route; the four filter facets are
  reassembled from `/Genres`, `/OfficialRatings`, `/Studios` and `/Tags`.
- Metadata writes take name-pair lists (`Genres: [{'Name': 'Action'}]`); the
  plain string array is accepted and then silently discarded.
- Custom artwork uploads must be base64 text, not raw bytes — which was broken
  for Jellyfin too and is fixed for both.
- Trickplay, media segments and lyrics 404 on Emby, so scrub previews are absent
  and intro/credit markers fall back to chapter names.

Verified against a local Emby 4.9.5 and a Jellyfin 10.11.11 control server:
onboarding, browse, detail, playable stream URLs serving real bytes, subtitle
sidecars, watch-state write and restore, hubs, cross-server aggregation and
search across both backends simultaneously.
2026-08-05 06:09:26 +02:00

254 lines
10 KiB
Dart

import 'dart:convert';
import '../media/ids.dart';
import 'package:drift/drift.dart';
import '../database/app_database.dart';
import '../media/media_backend.dart';
import '../media/media_item.dart';
import '../utils/isolate_helper.dart';
/// Backend-agnostic key-value cache for API responses.
///
/// Stores raw JSON keyed by `serverId:endpoint` in the shared `ApiCache`
/// Drift table. `serverId` values are globally unique across connected
/// backends, so Plex and MediaBrowser entries never collide despite sharing
/// the same table.
///
/// Plex- and MediaBrowser-specific helpers (item-id pinning, metadata parsing)
/// live on subclasses [PlexApiCache] / [JellyfinApiCache], which also
/// implement the abstract [getMetadata] / [pinForOffline] / [deleteForItem]
/// methods so callers can dispatch via [forBackend] instead of switching on
/// the backend type at every call site.
class ApiCacheSingleton<T extends ApiCache> {
ApiCacheSingleton(this.backends, this.typeName);
/// Every backend this cache answers for. Jellyfin and Emby share one
/// instance: their DTO shapes are identical and cache rows are keyed by the
/// compound `machineId/userId` scope, so there is nothing to isolate.
final Set<MediaBackend> backends;
final String typeName;
T? _instance;
T get instance {
final value = _instance;
if (value == null) {
throw StateError('$typeName not initialized. Call $typeName.initialize() first.');
}
return value;
}
void install(T instance) {
_instance = instance;
ApiCache.registerInstance(instance, backends);
}
}
/// Decodes independent cached JSON rows, dropping only the malformed row.
Map<String, MediaItem> decodeCachedMediaRows<T>(
Iterable<T> rows, {
required String Function(T row) serializedData,
required MapEntry<String, MediaItem>? Function(T row, Map<String, dynamic> json) decode,
}) {
final result = <String, MediaItem>{};
for (final row in rows) {
try {
final json = jsonDecode(serializedData(row)) as Map<String, dynamic>;
final decoded = decode(row, json);
if (decoded != null) {
result[decoded.key] = decoded.value;
}
} catch (_) {
// A malformed cache row does not invalidate its siblings.
}
}
return result;
}
abstract class ApiCache {
static final Map<MediaBackend, ApiCache> _byBackend = {};
/// Registers [instance] for each requested backend. A new database marks a
/// new application/test lifecycle, so registrations tied to the previous
/// database are discarded instead of leaving backend dispatch pointed at a
/// closed connection.
static void registerInstance(ApiCache instance, Set<MediaBackend> backends) {
if (_byBackend.values.any((registered) => !identical(registered.database, instance.database))) {
_byBackend.clear();
}
for (final backend in backends) {
_byBackend[backend] = instance;
}
}
/// Pick the cache for [backend]. Plex is the legacy default — covers items
/// predating the Connections table where the backend can't be resolved.
static ApiCache forBackend(MediaBackend? backend) {
final picked = _byBackend[backend ?? MediaBackend.plex] ?? _byBackend[MediaBackend.plex];
if (picked == null) {
throw StateError('No ApiCache registered for backend $backend');
}
return picked;
}
/// Clears volatile rows for every distinct registered database.
///
/// Production backend caches share one [AppDatabase], while focused tests
/// may register only one backend. This operation is therefore independent
/// of backend initialization order and is a no-op before registration.
static Future<void> clearRegisteredVolatile() async {
final cleared = <AppDatabase>{};
for (final cache in _byBackend.values) {
if (cleared.add(cache.database)) {
await cache.clearVolatile();
}
}
}
final AppDatabase _db;
ApiCache(this._db);
/// Direct database access for services that need to query the cache table
/// outside the standard get/put surface (e.g. playback initialisation that
/// joins on adjacent tables).
AppDatabase get database => _db;
String _buildKey(ServerId serverId, String endpoint) {
return '$serverId:$endpoint';
}
Future<Map<String, dynamic>?> get(ServerId serverId, String endpoint) async {
final key = _buildKey(serverId, endpoint);
final result = await (_db.select(_db.apiCache)..where((t) => t.cacheKey.equals(key))).getSingleOrNull();
if (result != null) {
return await tryIsolateRun(() => jsonDecode(result.data) as Map<String, dynamic>);
}
return null;
}
Future<void> put(ServerId serverId, String endpoint, Map<String, dynamic> data) async {
final key = _buildKey(serverId, endpoint);
final encoded = await tryIsolateRun(() => jsonEncode(data));
await _db
.into(_db.apiCache)
.insertOnConflictUpdate(
ApiCacheCompanion(cacheKey: Value(key), data: Value(encoded), cachedAt: Value(DateTime.now())),
);
}
Future<void> deleteForServer(ServerId serverId) async {
await (_db.delete(_db.apiCache)..where((t) => t.cacheKey.like('$serverId:%'))).go();
}
/// Pin an endpoint's response so the row survives cache eviction.
Future<void> pin(ServerId serverId, String endpoint) async {
final key = _buildKey(serverId, endpoint);
await (_db.update(
_db.apiCache,
)..where((t) => t.cacheKey.equals(key))).write(const ApiCacheCompanion(pinned: Value(true)));
}
Future<void> unpin(ServerId serverId, String endpoint) async {
final key = _buildKey(serverId, endpoint);
await (_db.update(
_db.apiCache,
)..where((t) => t.cacheKey.equals(key))).write(const ApiCacheCompanion(pinned: Value(false)));
}
Future<bool> isPinned(ServerId serverId, String endpoint) async {
final key = _buildKey(serverId, endpoint);
final result = await (_db.select(_db.apiCache)..where((t) => t.cacheKey.equals(key))).getSingleOrNull();
return result?.pinned ?? false;
}
/// Clear every cached row (debugging / sign-out).
Future<void> clearAll() async {
await _db.delete(_db.apiCache).go();
}
/// Clear volatile cached rows while preserving pinned offline metadata.
Future<void> clearVolatile() async {
await (_db.delete(_db.apiCache)..where((t) => t.pinned.equals(false))).go();
}
/// Pull pinned rows for [serverId] and extract the first capture group of
/// [keyPattern] from each `cacheKey`. Returns the unique set of captured
/// ids — backend subclasses use this to enumerate their pinned items
/// (Plex ratingKeys, MediaBrowser item ids).
Future<Set<String>> extractPinnedIds(ServerId serverId, RegExp keyPattern) async {
final rows = await (_db.select(
_db.apiCache,
)..where((t) => t.cacheKey.like('$serverId:%') & t.pinned.equals(true))).get();
final ids = <String>{};
for (final row in rows) {
final match = keyPattern.firstMatch(row.cacheKey);
if (match != null) ids.add(match.group(1)!);
}
return ids;
}
/// Walk every pinned row, extract `(serverId, capturedId, rawData)` tuples
/// from rows whose `cacheKey` matches [keyPattern]. The serverId is parsed
/// from the prefix before the first colon. Backend subclasses use this to
/// batch-load all pinned metadata into their own model type without
/// re-implementing the row walker.
Future<List<({ServerId serverId, String id, String data})>> listPinnedRowsByPattern(RegExp keyPattern) async {
final rows = await (_db.select(_db.apiCache)..where((t) => t.pinned.equals(true))).get();
final out = <({ServerId serverId, String id, String data})>[];
for (final row in rows) {
final colon = row.cacheKey.indexOf(':');
if (colon < 0) continue;
final match = keyPattern.firstMatch(row.cacheKey);
if (match == null) continue;
out.add((serverId: ServerId(row.cacheKey.substring(0, colon)), id: match.group(1)!, data: row.data));
}
return out;
}
/// Fetch and parse cached [MediaItem] for [itemId] on [serverId]. Returns
/// `null` when the item isn't cached.
Future<MediaItem?> getMetadata(ServerId serverId, String itemId);
/// Pin the cached metadata row(s) for [itemId] so they survive cache
/// eviction (used by the offline-download pipeline).
Future<void> pinForOffline(ServerId serverId, String itemId);
/// Delete cached metadata for [itemId] (used when removing a download).
Future<void> deleteForItem(ServerId serverId, String itemId);
/// Persist a watched/unwatched flip into the cached metadata JSON for
/// [itemId] so reloads (`getMetadata` / `getAllPinnedMetadata`) reflect the
/// state without having to refetch from the server. No-op when the row
/// isn't cached. Backend subclasses know which JSON fields to mutate
/// (Plex `viewCount`, MediaBrowser `UserData.PlayCount` / `Played`).
///
/// Optional positional progress fields ([viewOffsetMs], [lastViewedAt],
/// [viewedLeafCount]) let the offline-watch-sync service mirror richer
/// snapshots from the server's episode-list response without having to
/// fall back to a per-backend mutation. When omitted, the watched flip
/// uses the same defaults as before (zero-out `viewOffset`, stamp
/// `lastViewedAt` only when transitioning to watched).
///
/// **Drift discipline:** the inputs are backend-neutral but the JSON
/// shape + units are not. Adding a new watch-state input here means
/// updating *both* concrete impls ([PlexApiCache.applyWatchState],
/// [JellyfinApiCache.applyWatchState]) — Plex stores epoch-seconds and
/// flat fields, while Jellyfin and Emby store ISO-8601 + ticks under `UserData`.
/// The mutations are too short (~3 lines per backend) for a shared
/// adapter to be a net win, so they live duplicated by design.
Future<void> applyWatchState({
required ServerId serverId,
required String itemId,
required bool isWatched,
int? viewOffsetMs,
int? lastViewedAt,
int? viewedLeafCount,
});
/// Bulk-load pinned metadata whose private cache namespace is included in
/// [cacheServerIds]. A null set retains the backend's complete diagnostic
/// view; profile-visible hydration must always pass exact allowed scopes.
Future<Map<String, MediaItem>> getAllPinnedMetadata({Set<ServerId>? cacheServerIds});
}