LibraryItem never declared removed/_mtime so those fields were
silently dropped by Gson before Rust ever saw them, even though
library_state.rs already worked on raw JSON and could have used them.
Adds the fields, StremioRepository.getLibraryWatchlistWithTimestamps,
and wires it into the same mergeWatchlist reconciliation the other
providers use -- but pull-only (reconcileWatchlistPullOnly), not the
full push-capable path.
Deliberately not pushing local watchlist changes to Stremio: that
would need a "toggle membership" write that preserves whatever
progress/state fields already exist on the remote item (a
read-modify-write), which isn't built and isn't safe to guess at
without being able to verify Stremio's actual datastore PUT semantics
(full-replace vs partial-patch) against a live account.
pushWatchlist's AniList branch only ever called SaveMediaListEntry
(add/update) -- removing locally never propagated, since AniList's
DeleteMediaListEntry mutation needs the list-entry id, not the media
id, and nothing looked it up. Adds a lookup query (Media(id) {
mediaListEntry { id } }, which resolves against the authenticated
viewer implicitly, no separate viewer-id round trip needed) followed
by the delete mutation, and branches pushWatchlist's AniList call on
isInWatchlist same as the other providers.
Un-marking a single episode as watched is left out of this pass --
AniList tracks a show-level episode-progress counter, not discrete
per-episode watched flags, so "unwatch this one episode" doesn't map
cleanly onto its data model without risking corrupting progress.
The AniList GraphQL query only ever fetched status: CURRENT, so the
watchlist-add push added in the AniList phase had nothing to
reconcile against on the next sync -- pushed items would just get
re-pushed or silently diverge. Drops the CURRENT-only filter (Rust's
anilist_entries_to_sync already builds a timestamped watchlist array
from PLANNING entries, just needed the query to actually fetch them)
and wires it through the same mergeWatchlist reconciliation Trakt
uses. Generalizes reconcileTraktWatchlist -> reconcileWatchlist since
the logic is fully provider-agnostic.
Simkl mark-watched push existed but nothing was ever pulled back --
marking an episode watched on simkl.com never reached Fluxa. Simkl's
sync/all-items already returns per-episode watched_at when fetched
with extended=full (already the default), just unused until now.
Adds getSimklWatchedEpisodesWithTimestamps (watching+completed status,
shows+anime types) and generalizes the Trakt watched-episode
reconciliation (reconcileTraktWatched -> reconcileWatchedEpisodes)
since the merge logic and push path are provider-agnostic -- both
Trakt and Simkl now call the same function.
Simkl's sync/all-items endpoint has no "added to list" timestamp
(last_watched_at is watch-history, not list-membership, and is null
for plantowatch items) so a real timestamp merge like Trakt's isn't
possible without risking local removals being silently undone by a
fabricated timestamp. Per explicit decision: items on Simkl's
plantowatch list are treated as authoritative and always restored
locally each sync, overriding any local removal tombstone. Removing
something from Simkl's own list doesn't remove it locally -- that
would need a persisted prior-snapshot to detect, out of scope here.
Trakt's watched-state pull only ever wrote into the external mirror
table for display; local watched_episodes and Trakt's remote state
were never compared or reconciled, so un-marking an episode locally
never propagated and remote-only watched episodes never landed
locally. Trakt's watched-shows response already carries per-episode
last_watched_at, just unused until now.
Adds getWatchedEpisodesWithTimestamps (parses last_watched_at),
global (not per-series) local watched-episode/removal snapshot
queries, and reuses the mergeWatched core_invoke method (Phase 1) to
decide push vs. apply per composite episode id, mirroring the
watchlist reconciliation pattern.
AniList integration was structurally pull-only — no mutation was ever
sent, so adding to the watchlist or marking something watched in
Fluxa never propagated back, even though the stored OAuth token
already has full read/write access (AniList's OAuth has no scope
parameter). Wires pushWatchlist/pushMarkWatched to call the new
anilistSaveMediaListEntryVariables core_invoke method and POST a
SaveMediaListEntry mutation through the existing generic
anilistGraphQl endpoint.
Watchlist removal isn't covered: AniList's delete mutation needs the
list-entry id (not the media id), which requires an extra lookup
query — left as follow-up rather than scope creep here. Same for
un-marking watched.
pushWatchlist's Simkl branch was guarded by `if (isInWatchlist && ...)`
with no else — removing an item locally never removed it on Simkl,
since only sync/add-to-list was ever called. Adds the mirroring
sync/remove-from-list endpoint and routes the push based on
isInWatchlist, matching how the Trakt branch already handles both
directions.
Durability (WorkManager-backed) for watchlist push in general —
Trakt's branch has the same gap — is left as follow-up; this only
fixes the missing Simkl remove call.
StremioRepository.savePlaybackProgress was fire-and-forget on the
calling coroutine at both call sites (HomePlaybackController and the
writePlaybackProgress headless effect) — an app kill mid-request
silently dropped the push with no retry, unlike Trakt/Simkl scrobble
and Nuvio's progress push which are all worker-backed.
Adds StremioPlaybackProgressPushWorker on the ProviderSyncPushWorker
base, keyed by profileId+contentId same as Nuvio's so rapid saves
coalesce via ExistingWorkPolicy.REPLACE. savePlaybackProgress now
returns Boolean instead of swallowing failure, since the worker needs
to know whether to retry.
TraktScrobbleWorker, SimklScrobbleWorker, and NuvioPlaybackProgressPushWorker
had hand-copied WorkManager boilerplate: identical Constraints/backoff
policy, enqueueUniqueWork wiring, and profile-lookup/failure-bookkeeping
against ProfileManager. Behavior-preserving refactor onto a common base
so the next durable workers (Stremio, Simkl watchlist, AniList) don't
duplicate it a fourth and fifth time.
Trakt's watchlist pull previously fed only a live, un-timestamped,
display-only union in HomeLibraryCoordinator/AndroidLibraryDataSource
(distinctBy id, recomputed on every emission, never persisted or
pushed back). This wires it into the Phase 0/1 groundwork instead:
- TraktSyncItem now parses listed_at, exposed via
TraktSyncClient.getWatchlistWithListedAt / TraktRepository
- ExternalSyncMergeBridge wraps the mergeWatchlistTimestamped
core_invoke call for Kotlin callers
- WatchlistManager gains a local membership snapshot (active entries +
removal tombstones) and applyRemoteWatchlistAdd
- HomeLibraryCoordinator.load() now reconciles Trakt's remote
watchlist against the local one after each fetch: items the merge
says are locally newer get pushed via the existing
ExternalSyncPushCoordinator.pushWatchlist, items remote is newer on
get applied into the local watchlist_entries table
Simkl and Stremio are follow-up work — their fetch shapes differ
(Simkl has no true watchlist endpoint, Stremio's is a unified
datastore blob) and need separate integration passes.
Both only ever read id/type/seasonsCount from the 39-field Meta model,
in violation of the leaf-UI-gets-narrow-models rule in CLAUDE.md.
Adds seasonsCount to the existing PlayerContentUiModel (already
computed once via meta.toPlayerContentUiModel() at the top of
PlayerScreenContent) and threads that through instead.
The other Player files that take a Meta parameter (PlayerLoadingEffects,
PlayerTrackMemoryEffects, PlayerPipAndEpisodes, etc.) are LaunchedEffect
wrappers that hand Meta straight to ViewModel/repository calls for
persistence and scrobbling — that's the "effects/coordinators" layer
CLAUDE.md explicitly allows to hold domain objects, not a violation.
Lays the groundwork for timestamp-based conflict resolution against
Trakt/Simkl/Stremio (matching the pattern Nuvio already uses), by
giving watchlist entries and watched episodes an updatedAt column and
a soft-delete tombstone table so removals carry a comparable
timestamp instead of vanishing on hard delete. Progress already had
updatedAt; this brings watchlist/watched up to the same level.
Adds a per-profile continue-watching source preference (Fluxa/Stremio/
Nuvio/Trakt/Simkl/AniList), a background worker to push playback
progress to Nuvio instead of pushing inline, dedup of mark-watched
pushes against already-watched episodes, and general Nuvio account
import/merge fixes.
When a provider lists the requested season but not the requested
episode, fall through to no-match instead of an episode-only match
that could silently grab a different season's same-numbered episode.
Removes the standalone inline audio/subtitle delay adjust pill and
its state, and adds MediaTrack.subtitleFormatLabel/containerTrackId
so subtitle tracks can be labeled by codec (SRT/ASS/VTT/TTML/PGS/DVB/
VOBSUB).
Every stateless method on FluxaCoreNative now calls the corresponding
core_invoke route via FluxaCoreUniFfi instead of a hand-written
external fun; only the handle-lifecycle functions (headless engine,
app core state) keep their dedicated JNI bindings. NuvioCoreBridge and
ExternalLibraryClient's generic coreInvoke calls move onto the same
UniFFI path. Public method signatures are unchanged, so no other call
site needed to change.
FluxaCoreUniFfi.kt and the UniFFI-generated com.fluxa.core.uniffi
bindings only existed in the app Gradle module, but FluxaCoreNative.kt
(data module) needs to call into them and data cannot depend on app.
Moves the generateFluxaCoreUniFfiBindings task and the Kotlin facade
into data, which app already depends on.
Season-candidate selection, widget row formatting, and release
notification filtering now call the equivalent Rust calendar_plan
functions through core_invoke instead of duplicating that logic in
Kotlin.
The row under the seekbar exited with fade and slide only, so its
layout space stayed reserved until the animation finished and then
snapped away. While scrubbing, the seekbar was therefore held up in
its buttons-visible position and the time and thumbnail preview
anchored above it floated high over empty space. The row now shrinks
vertically as it fades, taking its spacing with it, so the seekbar
settles to the bottom and the preview stays snug above it.
Restores the live seek used to capture preview thumbnails during
scrubbing, but hides its side effect: when a drag starts, the current
surface frame is captured via PixelCopy and drawn full-screen under the
controls until the drag ends. The real player can jump between buckets
for thumbnail capture without the visible video changing; the frame you
started dragging on stays on screen until release commits the seek.
Opacity sliders wrote the profile on every drag frame, which is what
made them stutter; they now hold a local live value that drives the
sample cue and only commit to the profile on release. With the panel
usable during drags, the fade-to-15% behavior is removed and the sample
cue moves lower (20dp from the bottom) instead.
The seek preview thumbnail system seeked the real player whenever a
drag paused over an uncached bucket, so scrubbing caused audible
jumps mid-drag. It now serves cached frames only, keeping the
background capture loop; the actual seek happens solely on seekbar
release, which the seekbar already implemented.
The adjust card stacked its slider blocks two lines tall, making the
content roughly the full height of a landscape phone screen, so even
top-anchored it buried the sample cue. Slider items collapse to single
rows (title, slider, percent) and the card anchors top-right instead of
top-center, clearing the bottom-center subtitle zone both vertically
and horizontally.
Drags on a sheet, card, or the speed popup were only consumed when a
child (slider, list) handled them; anything else bubbled up to the
player's volume/brightness drag layer, so swiping on the subtitle
adjust panel changed the volume. The shell root and the speed popup now
consume all unhandled drag gestures, while child scrollables and
sliders keep priority. The adjust page content also becomes vertically
scrollable since it can overflow the top-anchored card on short
landscape screens.
The adjust entry no longer scrolls with the track lists: each category
header (column label in the two-column card, the tab row in portrait)
carries a small tune icon. Audio's icon opens the inline delay pill
directly; the subtitles icon opens the adjust panel.
The adjust panel now anchors to the top of the screen with a near-zero
scrim so the bottom-center subtitle area stays visible, and while a
slider is being dragged the whole card fades to 15% opacity. A sample
cue styled with the live text/background/outline opacity values renders
at the subtitle position whenever the panel is open, so there is always
something to judge adjustments against.
Every player menu previously shared one full-chrome sidebar shell. The
shell now renders a translucent 24dp-radius card (0xE6101418, no
border) that fades in with a short upward drift, centered in landscape,
with no subtitle line and no close button; portrait keeps the bottom
sheet with its drag handle. Headers only render when a title or back
action exists.
Speed is no longer a sheet: it is a compact popup near the controls
that scales and fades in from its corner over a light scrim, listing
plain speed rows with the current one checked.
Audio and subtitles merge into a single Netflix-style two-column card
in landscape (portrait keeps the segmented tabs). Audio delay and
subtitle delay leave the card entirely: selecting them collapses the
menu to a floating pill over the video with -/value/+ controls so the
sync effect is visible live while adjusting. Subtitle styling keeps the
in-card adjust page.
The landscape sheet stretched full height and flush to the screen edge,
so short menus like Audio or Speed left a huge empty void below their
rows. Side sheets are now 340dp floating cards (400dp on TV) rounded on
all corners, vertically centered with a 16dp edge margin, sized to
their content with a viewport-bounded max height. The slide-in
animation accounts for the margin, and the header is tightened with a
smaller, quieter close button.
The sheets opened with an all-caps title, an instruction sentence, a
panel spanning almost half the screen, and two-line rows where every
speed carried a redundant Slower/Faster label. Titles are now title
case (Audio, Subtitles, Playback speed), the instruction line is gone,
the landscape sheet narrows from 46% to 38% width capped at 400dp,
rows tighten to 46dp with single-line speed entries (only 1x keeps a
Normal label), and mono/stereo channel counts render as words instead
of 1.0/2.0.
Rows in the player sheets were flat full-bleed text with a 4dp inset;
the only selection cue was the trailing check. Rows are now rounded
12dp containers with a subtle white highlight behind the selected one,
matching ripple bounds, and the header subtitle is slightly larger.
The sidebar previously entered with a fade plus a 44px nudge and
disappeared instantly on close. It now slides the full distance from
its edge (right for side sheets, bottom for bottom sheets) with the
scrim fading in sync, and close requests play the exit animation
before the panel is removed, including swipe-to-dismiss. Panel height
changes between tabs are animated instead of snapping.
The mobile speed menu drops its bordered chip grid in favor of the
same checkmarked rows used by the audio, subtitle, and source lists.
Opening the external player fires ON_PAUSE on MainActivity, and
PlayerLifecycleEffect auto-entered PiP because the player screen was
still active and the profile PiP setting was on. Gate the auto-enter
behind a suppression flag set for the duration of the external launch,
cleared on activity result and ON_RESUME.
Reworks the player overlay chrome, sidebar shell, settings and
source sidebars, mark-segment sidebar, and mobile dialogs, plus
new dimension/icon tokens and i18n strings for the redesign.
Addon catalogs can return overlapping items across paginated skip
offsets, so the discover results list could contain the same
type:id twice. LazyVerticalGrid requires unique keys and crashed
with IllegalArgumentException once a duplicate scrolled into view.
Dedupe results in the coordinator where they're assigned.
The 5-dot sliding window re-centered on the current page every swipe,
so the highlighted dot's position within the window stayed constant
while paging through the middle of the list. Show one dot per item
instead since the hero caps at 8 items.
Wrap the screen AnimatedContent in a SaveableStateHolder keyed by
screen so each destination's LazyColumn/LazyList state survives being
switched away from and back to, instead of resetting to the top.
Bottom bar now overlays content edge-to-edge with a scrim background
and top hairline instead of reserving fixed space, gains labels,
FluxaIcons, selection animation, and a calendar badge for new
releases. Home screen's filter/notification/avatar bar is pinned and
fades in as the hero scrolls past instead of scrolling away with it.
Hero auto-swipe now pauses during user interaction, gained a Play
button, capped height, and windowed dot indicators. Row "view all"
now navigates to the category grid instead of silently loading more
into the row.
Settings:
- Replace the single-level category state with a real back stack
(lifted into FluxaAppState) so back navigation goes up one level
instead of always jumping to the hub, with hardware back wired
through MainActivity.
- Scroll position is now keyed per category instead of shared/leaking
across pages, and nested pages show their own titles.
- Split Notifications and TMDB feature toggles out of Account into
their own pages instead of an accordion and an overloaded
TmdbAccountChanged action bundling unrelated fields.
- Downloads settings page no longer duplicates Library's live download
list; it links out to it instead.
- Add a lightweight settings search (SettingsSearch.kt) over a static
index of rows.
- TV renders Settings as a two-pane list-detail layout.
- Introduce SettingsGroupCard and restructure every settings page
from a flat row list into grouped inset cards, restyle the search
field and choice-picker dialog to match the app's black/white
aesthetic instead of default Material styling, and merge the
redundant avatar/account rows on the hub.
- Add-ons settings page now only launches the real add-ons screen
directly (dropped the redundant "Manage Add-ons" middle page and
the torrent preferences that lived there).
Add-ons store:
- Validate pasted manifest URLs by actually fetching them before
installing, instead of accepting any string that pattern-matched
as a manifest path; show an inline error on failure and a
confirmation dialog on success.
- Replace the bare "+" icon button with a labeled Add button.
Also strips emoji baked into a few i18n strings (CloudStream repo/
plugin headers, premiere notification) per project convention.
Tapping an episode card only updated the selected-episode state without
emitting a navigation event, so nothing visibly happened until Play was
pressed. DetailStore now emits SelectSources on EpisodeSelected, and
selectedEpisodeId is tracked via a StateFlow instead of a plain var so
the selection highlight updates immediately instead of lagging behind
an unrelated state emission.
FluxaAppHost already owns real navigation internally via FluxaAppState
(tab switching, detail selection, category browsing, profile editing,
source selection, notifications) - Screen/AppNavigator/AppRoutesHost's
12-variant sealed class and full backstack were solving a problem
FluxaAppHost had already solved. FluxaNavigationStore/FluxaRoute (the
type originally targeted for this) turned out to be orphaned scaffolding
with no real consumer.
Replaces the Android-side Screen stack with the handful of things
FluxaAppHost actually needs seeded from outside: current FluxaDestination
(+ a one-level previousDestination for the three non-clearing pushes:
search, addon store, connect-account), authStartOnNuvio, and a
PlayerLaunchRequest since Player isn't part of the shared UI yet. Deletes
AppNavigator.kt, AndroidFluxaNavigator.kt (confirmed dead - zero external
references), the Screen sealed class, and ~340 lines of AppRoutesHost's
Screen->FluxaDestination translation plus its dead legacy AnimatedContent
dispatch (Screen.Detail was never actually constructed anywhere outside
that dead navigator, confirming Detail navigation already flows entirely
through FluxaAppHost's internal state).
Verification gap: compiles clean and existing tests pass, but this is
the app's core navigation/back-button/PiP-lifecycle wiring and there is
no emulator available this session to click through it. Needs a manual
pass before shipping: cold launch with/without profiles, Home-Detail-
Player-back, hardware back from every top-level tab, PiP entry, connect-
account auth flow, profile switch, and download-item tap to Player.
Relocates skip-segment/transient overlays, the sidebar shell + track
list primitives, mark-segment/settings/track sidebars, and the
playerInputControls gesture modifier from app-module Android files
into shared/commonMain. Promotes coil3/coil3-compose to commonMain
(genuinely multiplatform; only coil3-network-okhttp stays
Android-only) so the overlay cards using AsyncImage can move too, and
promotes TorrentStreamStatus to player/commonMain as a plain data
holder.
Where a composable used to reach into UserProfile's JNI-backed
`safeXxx` properties (holdSpeed, holdToSpeedEnabled, language) or an
Android-only Locale API, it now takes the resolved primitive as a
parameter instead, with the app-side caller supplying it - keeping
the commonMain/androidMain boundary honest rather than papering over
it.
EpisodeSidebar (Hilt ViewModel-bound) and TrackSelectionState (direct
FluxaCoreNative JNI calls) stay Android-only, along with the actual
video surface, playback effects, and setup effects - those are the
next sub-steps in the player-extraction migration.
Relocates TvPlayerUIContent/MobilePlayerUIContent and their supporting
composables (seekbars, transport buttons, bottom action row) from
app-module Android files into shared/commonMain as
PlayerControlsChrome.kt. Both already consumed only lightweight params
(PlayerContentUiModel, Chapter, primitives, callbacks); the one real
platform coupling was the seek-thumbnail capture (SurfaceView/ExoPlayer
frame grab), which now happens at the PlayerOverlays.kt call site on
the Android side and gets passed down as a plain ImageBitmap instead of
being computed inside the shared composables.
First sub-step of the player-extraction migration step; overlays,
sidebars, and setup/effects follow as separate passes.
Sweeps addon repositories, watchlist/profile storage, Trakt/Nuvio
sync, discovery models, and player policy classes out of Android-only
packages into data/commonMain and player/commonMain, with thin
androidMain wrappers where platform APIs are still required.
