Registering IPC handlers costs 0.4 ms; evaluating the modules behind them (axios, drizzle-orm, better-sqlite3, electron-updater, fix-path) before the window could load was the real cost. main.ts now keeps only the pre-paint wiring and loads the rest as the deferred-events.js chunk inside the main window's did-start-loading listener, where the import and its registrations complete before any renderer invoke can arrive.
Interleaved A/B on the performance build: app.whenReady 323 -> 265 ms, did-finish-load 499 -> 447 ms; J1 journey spawnToDidFinishLoad ~405 -> ~365 ms with identical counters. Packaging ships the chunk explicitly, verify:package-layout requires it, and the benchmark build identity hashes it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
main.js is now a small entry that enables Node's on-disk V8 compile cache under userData/v8-compile-cache and then requires the application bundle main.app.js. Warm launches reach app.whenReady about 13 ms sooner at the median; IPTVNATOR_DISABLE_COMPILE_CACHE=1 turns it off and IPTVNATOR_COMPILE_CACHE_DIR relocates it.
Packaging now ships main.app.js explicitly and verify:package-layout requires the main-process entry files in app.asar, because nx-electron copies the backend through an allowlist. The Xtream benchmark build identity hashes both the launcher and the bundle.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(performance): add the J1 launch-to-usable journey benchmark
Implements plan items A1, A3 (J1 only) and the minimal A4 from
.plans/2026-09-25-performance-journeys-ratchet.md.
- journey-renderer-probe.ts: init-script probe counting DOM mutations,
layout shifts and long tasks until the first source card is visible on
/workspace with the splash removed; unit-tested with jsdom fixtures.
- journey-main-ipc-capture.ts: counts bridge invocations from the preload's
renderer-API trace channel up to a sentinel call the probe fires, so the
IPC counter is exact without touching production code.
- launch.journey.ts + playwright.journeys.config.ts: seeded profile (one
M3U source, one Xtream portal on the loopback mock), one warm-up and five
measured iterations, fresh process and data directory each, writing
dist/performance/journeys/<timestamp>/summary.json with exact counters
and P50/P90 wall-clock.
- Nx target electron-backend-e2e:journeys and root script perf:journeys.
- docs/architecture/performance-journeys.md, README, context and
validation map entries.
renderer.cdTicksToFirstCard and main.sqlStatementsBeforeReadyToShow are
reported as unavailable with the reason instead of being faked.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(performance): gate the renderer load so the probes never race startup
Review follow-up for #1698.
- journey-renderer-gate.cjs: a main-process hook loaded with `-r` (the
mechanism Playwright uses for its own loader) makes the first
loadFile/loadURL navigate to about:blank and holds the real load until
the test releases it. Playwright reports no page before a navigation
commits, so this is what lets the renderer probe be registered on the
page before the real document exists; the IPC capture is installed
before the release too. A safety timeout releases the gate on its own
and marks the iteration invalid. Unit-tested with a fake BrowserWindow.
- launch-journey-app.ts: registers the probe on the parked page, releases
the gate, waits for the real document to commit, fails fast when the
probe is missing, and refuses an iteration whose gate timed out, saw a
second load, or released before the probe was in place.
- journey-renderer-probe.ts: entries delivered live after the terminal
batch are buffered and filtered by the same cutoff as queued ones, and
the cutoff is sampled in a timer queued from the first rAF, i.e. after
the card's frame is painted, so the render task's long task and layout
shift are consistently included.
- launch-journey-record.ts: layoutShiftScore rounded to three decimals; a
0.0001 shift flipped in and out of the cutoff between iterations.
- playwright.journeys.config.ts: reuse a mock server left on the journeys
port locally (its fixtures are deterministic); CI still starts its own.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(performance): validate the journey gate after the probe completes
Codex follow-up on #1698: the gate state returned by release() cannot see a
reload or recovery navigation that happens before the first card. Re-read the
live state once the renderer probe has finished and validate that instead,
so an iteration spanning an extra navigation is rejected.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(performance): fail an iteration whose performance observers were unavailable
Codex follow-up on #1698: a renderer that cannot observe layout-shift or
longtask entries used to pass the probe with zero counters, which a ratchet
could not tell apart from a genuine zero. The probe assertion now rejects
such an iteration and names the missing observer.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: 4gray <fourgray@proton.me>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* test(electron): reap Windows process trees and retain smoke diagnostics
* test(electron): require confirmed process exit before relaunch
* test(electron): retain process handle after application exit
* test(electron): bind captured processes to application instances
Xtream and Stalker live cards on the dashboard carried no XMLTV key, so the
rails never asked anything for them and showed only the LIVE chip. Their
programme now comes from the portal, one card at a time and only once the card
is on screen.
- `lib-dashboard-rail` reports the cards inside its viewport through an
IntersectionObserver rooted at the track; `DashboardLiveEpgPresenter` unions
them with the pinned hero row and hands the set to
`DashboardPortalLiveEpgPresenter`.
- `DashboardPortalLiveEpgService` runs the bounded queue — two requests in
flight, 200 ms apart, one card each — through
`StreamResolverService.loadEpgForItems`, publishing every answer the moment
it lands, so the page never waits and a slow portal delays no other card. A
card scrolled past before its turn is never requested.
- A programme is trusted for 60 s, an empty answer for 30 s (the resolver
reports a dead portal and a guide-less channel identically), and a
completion captures both the display offset and the EPG source revision,
requeueing itself when either moved.
- The presenter hands its wanted set back on destroy. Desktop only: the shared
collection resolver is gated on the local XMLTV bridge.
- A shimmer placeholder shows only before a card's first portal answer; M3U
cards keep the batched XMLTV lookup.
* feat(shell): zoom shortcuts on Windows and Linux (#1109)
Cmd/Ctrl and +/−/0 (numpad included) now zoom the app on every platform.
Windows/Linux run without a menu (`setMenu(null)`), so the shortcuts are a
renderer key binding in `WorkspaceKeyboardShortcutsService` calling a new
synchronous, preload-local bridge method `adjustZoomLevel`, which steps the
frame-bound temporary level through `webFrame.setZoomLevel` — never a
main-process `webContents.setZoomLevel`, whose per-URL entry the app's
`file://` path routing resets. Step and limits live in
`libs/shared/interfaces` (`stepZoomLevel`: 0.5 per press like Electron's
zoomIn/zoomOut roles, clamped to levels −4…6). On macOS the renderer sees the
key before the application menu, and `preventDefault()` keeps the menu role
from stepping a second time (Electron only performs the menu key equivalent
in its unhandled-keyboard-event hook).
Persistence is unchanged: the main process still reads the live level back
on close, quit and reload. The zoom E2E now drives the real shortcuts (in,
out, numpad, reset). Help dialog entries added and translated for all
locales; contract updated in docs/architecture/workspace-shell.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(shell): never step a stored out-of-range zoom level against the request
A level persisted before the shortcuts existed (the macOS menu roles never
clamped, and the store restores any finite level) was clamped BEFORE the
step, so the first zoom-in from level 7 rendered smaller. Step from the raw
level instead: a press further out leaves an out-of-range level where it is,
a press back in lands on the limit.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(shell): state the zoom bridge's return contract precisely
`adjustZoomLevel` steps by `stepZoomLevel`'s rules; a stored out-of-range
level is never moved against the request, so the returned level is not
itself guaranteed to be within `ZOOM_LEVEL_MIN..ZOOM_LEVEL_MAX`.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(release): add a release note for the zoom shortcuts
The Release note gate requires an added `.changes/*.md` for runtime changes;
the shortcuts are a user-visible feature of their own, so they get their own
note and the persistence note stays about persistence.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Movie and series detail pages showed two arrow_back controls while the
inline player was open: the shell's sticky arrow (added in #1576 so Back
survives scrolling) meant "Close player" in watch state, while the
now-playing bar carried a second arrow that meant route-level Back — the
same icon with two meanings, next to a "Close player" button that
duplicated the first.
The shell's sticky arrow is now route-level Back in browse and watch
alike, and the bar carries no arrow of its own. Closing the player is
the bar's "Close player" button and Escape, which still unwinds one
level (close, then back). Hosts without a browse Back target (M3U,
downloads) render no arrow in either state.
Unit specs for the shell and the inline player cover the new contract;
the Electron and web E2E helpers that pressed Back from watch are
updated, and the M3U flow closes the player through the bar's button.
Docs, the mirrored CLAUDE.md/AGENTS.md paragraph and a release note
follow the change.
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
The packaged renderer is index.html over file:// with path routing, so
after in-app navigation the document URL names a path with no file behind
it. A main-process reload (macOS View > Reload, DevTools) failed with
ERR_FILE_NOT_FOUND and stranded the window on Chromium's error page; a
renderer-initiated reload (the settings unsaved-changes guard's confirmed
location.reload()) was cancelled by the will-navigate trust check and
silently did nothing.
Both legs now re-load the packaged index with the route in a restoreRoute
query parameter, which main.ts restores with history.replaceState before
Angular bootstraps. The did-fail-load recovery is deferred to the error
page's dom-ready: a load issued from inside the failure event yields a
document that never receives animation frames and never paints.
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(shell): persist and restore the app zoom level (#1109)
The app-wide zoom (Cmd/Ctrl and +/-) was never saved, so it reset to the
default on every restart. Save the webContents zoom level next to the window
bounds on window close and before quit, and reapply it once the renderer
finishes loading. webContents zoom is per-host, so the restored level then
holds across in-app section navigation (SPA route changes never reload).
* fix(shell): restore the zoom level as frame-bound temporary zoom (#1109)
Under file:// Chromium keys zoom by the full URL, and the packaged
renderer routes with pushState, so a level applied through
webContents.setZoomLevel belongs to index.html only: the first resize
after a section change snapped the renderer back to the default, and
the close handler read the current route's entry (usually 0) over the
user's choice. The preload now applies the persisted level with
webFrame.setZoomLevel, a temporary zoom bound to the frame that survives
in-page navigation and resizes; the main process hands it over through
the synchronous WINDOW:GET_ZOOM_LEVEL IPC and writes the live level back
on close, before-quit and before every cross-document navigation, since
a reload drops the temporary level.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(shell): apply the restored zoom level at DOMContentLoaded (#1109)
A webFrame.setZoomLevel at preload start left a hidden window without a
first frame on Linux and Windows: ready-to-show never fired, the window
never showed, and the renderer got no animation frames, so the startup
splash removed in a requestAnimationFrame stayed. macOS was unaffected
and CDP-driven tests force frames, which is why only the packaged
legacy-migration E2E asserting the splash is gone caught it. Applying
once the document is parsed is harmless and still lands before the
first Angular paint; ownership of the level now follows the preload's
WINDOW:ZOOM_LEVEL_APPLIED acknowledgement instead of the request. The
release note no longer advertises Ctrl shortcuts Windows and Linux do
not have.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Justin Willhite <5132924+thejdubb02@users.noreply.github.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* chore(deps): upgrade Angular to 22.1 and Nx to 23.2
* fix(deps): complete Angular migrations after rebasing on master
* fix(ci): use the Node pin for Windows runtime refresh
* docs(deps): synchronize the workspace-shell Node requirements
* feat(epg): accept local XMLTV files as EPG sources
Settings → EPG and the playlist dialog accepted `file://` in their form
pattern, but the main process rejected everything except http(s), so a local
XMLTV entry saved fine and then failed on import. Both surfaces now take a
remote link, a `file:` URL, an absolute POSIX path or a Windows drive/UNC
path (`classifyEpgSourceReference` in shared/interfaces), and the settings
section spells out the accepted formats with examples.
The EPG worker opens every source through `openEpgSourceStream`: remote
links keep the validated-redirect client and trust policy, local files are
read from disk behind the signature-sniffing optional gunzip stage, so
.xml, .xml.gz and extension-less gzip all parse. Only hand-typed sources
may be local: `extractM3uEpgUrls` harvests http(s) links only from M3U
headers, since the local branch bypasses `validateRemoteUrl`.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): pick local XMLTV files with a native file dialog
A folder button beside each EPG source row (Settings → EPG and the playlist
dialog) opens the native open-file dialog and writes the chosen absolute
path into the row. New `EPG_OPEN_FILE_DIALOG` IPC behind
`ElectronBridgeApi.openEpgFileDialog`, gated in the renderer by
`RuntimeCapabilitiesService.supportsEpgFilePicker`.
The row's refresh/remove buttons carry `data-test-id`s now, and the EPG
e2e suites address them by id instead of index, since the folder button
became the first button in a row.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): authorize local XMLTV files in the main process
Review follow-up (Greptile P1, Codex P1). The renderer hands source strings
to FETCH_EPG/EPG_FORCE_FETCH unchanged, so the form validator alone could
not enforce the provenance rule: a compromised renderer, or a legacy
`file://` entry an older version stored from an M3U header, could name any
file on disk.
`EpgWorkerService.startFetch` now asks a main-process
`EpgLocalSourceAuthorizer` before a local path reaches the worker: a path
the native picker returned is trusted at once, a hand-typed path is
confirmed once in a native message box the renderer cannot fake, and a
refusal is reported in the progress panel. Allowed paths persist under
TRUSTED_LOCAL_EPG_SOURCES in the main-process config. The worker opens its
local branch only when main set `allowLocalFile`; the service defaults to
deny-all until epg.events installs the persisted authorizer.
`resolvePlaylistEpgSourceState` and `filterPlaylistEpgUrlsForFetch` drop a
stored non-remote entry unless it is also in `manualEpgUrls`.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): fail a refused local EPG fetch instead of resolving it
Review follow-up (Codex P2). A denied native confirmation now rejects the
fetch after reporting the error row, so handleFetchEpg and the renderer's
fetch result cannot claim the file was read. Also restores the unrelated
CLAUDE.md paragraph an earlier formatter pass had reflowed into a list.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): cancel a local source retired during its authorization prompt
Review follow-up (Codex P2). startFetch keeps the request generation
captured before awaiting the native confirmation and rechecks it
afterwards: a source retired meanwhile ends as cancelled instead of
starting an import that a pending clear would then have to await.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(epg): leave the local XMLTV e2e with a pristine settings form
The local-file test ended with the EPG source field still dirty, which
arms the main-process close guard: the app then waited for the unsaved
changes dialog instead of closing, the close timeout killed it, and on
Windows the killed process kept iptvnator.db busy (EBUSY on the data-dir
cleanup) and hung the Playwright worker teardown. Discarding the form
before the app closes takes the test from 15 s to 4 s locally.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: 4gray <fourgray@proton.me>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): render catch-up start times in the panel timezone
The `{Y-m-d:H-M}` segment of an Xtream timeshift URL is read by the panel
with `strtotime()` in ITS timezone (`server_info.timezone`), never the
viewer's. The timezone was learned in memory only, by the store's
`checkPortalStatus()`, so the Favorites / Recent catch-up resolver — which
reads the STORED playlist row — always fell back to the viewer's local
clock and asked the panel for the wrong programme (#1562).
- Normalize the panel's clock once (`resolveXtreamServerTimezone`): an
ICU-resolvable name is kept, otherwise a `UTC±HH:MM` offset is derived
from the `time_now` / `timestamp_now` clock pair, so spellings such as
`UTC+3` no longer silently mean "local time".
- Persist it on the playlist row through `transformPlaylistMeta` (no-op
when unchanged) and project it back from the payload in
`DB_GET_PLAYLIST`, so both catch-up entry points and a restart see it.
- Format with `hourCycle: 'h23'` (server midnight is `00`, never `24`) and
read timestamp-less EPG `start`/`end` strings in the panel's clock.
- Mock: `tzoffset:tzoffset` scenario with an unusable timezone name and a
+03:00 clock pair; Electron e2e covers Live TV, Favorites, a restart into
Global favorites, and the clock-pair derivation at a UTC-3 viewer.
Closes#1562
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): guard the account-info answer by playlist identity and reject rolled-over dates
Review follow-ups (Greptile):
- A source switch while `get_account_info` is in flight no longer hands
playlist A's status or clock to playlist B: the store is patched only
while the asking playlist is still selected, the timezone is persisted
under the asking playlist's id regardless, and a late failure cannot mark
the newly selected playlist unavailable.
- `parseNaiveUtcMs` reads the constructed date back, so out-of-range panel
strings (`2026-13-01 25:00:00`) are rejected instead of silently rolling
over into a real instant.
- Document that a clock-derived fixed offset is a DST-less snapshot, refreshed
by every account-info check and only ever used for non-standard servers.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): drop a panel clock that no longer belongs to the source
Review follow-ups (Codex + Greptile):
- A metadata update or DB_UPDATE_PLAYLIST that points the source at another
server drops the persisted `serverTimezone` (payload-only) until the next
account-info check, so Favorites / Recent cannot keep rendering the OLD
panel's clock; an update that supplies a clock keeps it.
- A late account-info answer is persisted only onto a row that still points
at the panel it came from — an edit that moved the source during the
request keeps the clock the edit flow dropped.
- The PWA data source and the route-session converter carry the persisted
timezone into the store playlist, so a later response without a usable
clock has a previous value to preserve.
- Mirror the catch-up timezone contract into AGENTS.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): drop the stale panel clock inside the UPDATE statement
Review follow-up (Codex): the database worker interleaves requests, so a
read-modify-write of the playlist payload could hand a concurrent upsert's
newer payload back to the past. The `serverTimezone` removal on a server
URL change is now one `CASE … json_remove(payload, '$.serverTimezone')`
expression inside the same UPDATE, guarded by `json_valid`; the spec runs
the real statement against Electron's SQLite on the actual `playlists`
table (moved, renamed, clock-less, malformed-payload and NULL-URL rows).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(xtream): split the server-clock primitives out of the timezone util
Review follow-up (Greptile): `xtream-server-timezone.util.ts` had grown past
the 300-line file guideline. The zone-agnostic wall-clock primitives (stored
forms, Intl parts, naive parsing) now live in `xtream-server-clock.util.ts`;
the timezone util keeps the Xtream policy and re-exports the public helpers,
so every import and the spec are unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): offer the learned panel clock to storage on every check
Review follow-up (Codex): a transient storage failure left the clock in the
store but not on the row, and the next check compared the answer with the
in-memory value and never retried. The resolved timezone is now always
handed to `transformPlaylistMeta`, whose row-level equality check keeps the
common case a read without a write; a failed write is retried by the next
account-info check.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): apply an account-info answer only to the panel it came from
Review follow-up (Codex): an in-place edit keeps the playlist id while
moving the source, so an answer already on the wire for the OLD panel
passed the id-only guard and patched the new panel's status and clock into
the store. One `answersFor(candidate, credentials)` predicate now gates the
store patch, the error path and the persisted-row transform alike.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): never report another panel's status for the selected playlist
Review follow-up (Greptile): callers gate content initialization on the
value `checkPortalStatus()` returns for whatever is selected NOW. When the
answer no longer describes the selected playlist (source switch or in-place
edit during the request), the store's own verdict about the current
selection is returned instead of the old panel's status — on success and on
failure alike.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): persist the panel clock with one conditional UPDATE
Review follow-up (Codex): `transformPlaylistMeta` reads the row and then
upserts it whole, while the Xtream edit dialog saves through
`DB_UPDATE_PLAYLIST` outside `PlaylistsService`'s queue and the database
worker interleaves requests — an edit landing between that read and the
upsert was silently undone.
Persistence now goes through `IXtreamDataSource.rememberServerTimezone`:
- Electron: new `DB_SET_PLAYLIST_SERVER_TIMEZONE` worker op — one UPDATE
that `json_set`s the payload only while the row still points at the
request's connection and does not already carry the value; a malformed
payload is never rewritten (CASE, not AND, so json_extract cannot run
before json_valid). Wired through the worker types, main handler,
preload, bridge interface, both IPC contract tables and
`DatabaseService.setXtreamPlaylistServerTimezone`.
- PWA: `transformPlaylistMeta`, whose read and write share one IndexedDB
readwrite cursor transaction, plus the localStorage copy.
The store no longer injects `PlaylistsService`; it offers the resolved clock
to the data source and keeps only its in-memory guards. Real-SQLite coverage
for the op (fresh / same / moved / NULL / malformed / missing rows),
delegation specs for both data sources, docs updated.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): keep the stored panel clock across clockless full upserts
Review follow-up (Codex): a `PlaylistsService` mutation that read the row
before `DB_SET_PLAYLIST_SERVER_TIMEZONE` landed and upserted afterwards
replaced the payload with its clockless snapshot. `DB_UPSERT_APP_PLAYLIST(S)`
now carry the STORED clock into a snapshot that has none while the row still
points at the same connection (`playlistConflictUpdate`, nested CASE so the
json_* readers never run on a malformed payload); a snapshot with its own
clock, or one that moves the source, wins as is. Real-SQLite coverage for
kept / moved / own-clock / batch rows.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(release): split capture-navigation under the max-lines cap
`tools/release/capture-navigation.ts` had grown to 567 counted lines, past
the 400-line rule, which failed `release-tools:lint` and — because the file
was not in the baseline — the max-lines baseline test on master and on
every PR branched from it. The 19 named setup actions are now grouped by
subject over one leaf module of shared page helpers:
- `capture-navigation-helpers.ts`: playlist-id registry, dialog handling,
navigation moves, `settleUi`
- `capture-navigation-setup-actions.ts`: add-playlist dialogs, settings
sections, remote control
- `capture-navigation-portal-actions.ts`: portal catalogs, live lists,
alternative sources (the two identical live-category flows share one
helper)
- `capture-navigation-download-actions.ts`: the download manager shots
- `capture-navigation.ts`: the `runAction` dispatcher, theme switching and
the re-exported API the seeding driver and the capture script import
Actions call their siblings directly instead of recursing through
`runAction`, so no module depends on the dispatcher. The action vocabulary
is unchanged (same 19 names, same waits and timeouts); every file is under
300 lines and the new modules are listed in the `release-tools` lint target.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(electron): move the panel-clock SQL into its own operations module
Review follow-up (Greptile): the timezone persistence, invalidation,
upsert-preservation and row projection had landed in
`playlist.operations.ts`, a baselined 1,000-line file. They now live in
`playlist-server-timezone.operations.ts` (155 lines) — the three SQL
shapes plus the payload projection — and the playlist operations compose
them; the baselined file shrinks by 107 lines. Behaviour and the
real-SQLite coverage are unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): add programme guide redesign spec for the M3U host
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): add programme guide implementation plan
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): add window-scoped guide programme queries
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): harden guide query scoping, caps and row mapping
- Scoped guide programme/coverage queries now include legacy
(unsourced) rows via source_url IN (...) OR IS NULL OR '',
mirroring EpgQueryService's legacy fallback.
- getProgramsForChannels/getProgramCoverage build their result from
the normalized, capped window.channelIds instead of the raw
request, so a key cut by the cap is absent rather than [] — an
invalid window now returns {}. Truncation logs counts only.
- Split the 100-channel guide cap from a new 2000-key coverage cap,
and cap sourceUrls at 50; normalizeGuideWindow takes the cap as a
parameter and moved (with guideWindowOverlapSqlText) into
epg-guide-window.util.ts.
- Extracted shared row mapping (toEpgProgramFromRow/isValidEpgProgram)
into epg-program-row.util.ts, used by both EpgQueryService and
EpgGuideQueryService so invalid start/stop rows are dropped
identically in both.
- Added a real-SQLite-backed test for the overlap predicate's exact
text, plus per-key array copies in the response.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): render the guide predicate in tests and document its scope
Correct the guide query's JSDoc: it runs one query accepting the union
of requested-source and unsourced legacy rows, unlike EpgQueryService's
two-query scoped-then-legacy fallback. Replace the hand-maintained
plain-SQL twin of the Drizzle overlap predicate with a rendered copy of
the real predicate (SQLiteSyncDialect().sqlToQuery) in the spec, add a
source-scoping case, and drop the now-redundant operator-sequence test.
warnIfTruncated reuses uniqueTrimmedStrings and names which read
(programme/coverage) was truncated. Rename epg-query.service.ts's local
EpgProgramRow to EpgProgramSelectRow so it isn't confused with the
shared EpgProgramRow type, and document getProgramCoverage like its
sibling.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): expose guide programme and coverage reads over the bridge
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): separate coverage chunk size in the guide plan
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): add guide source contract, day layout maths and preferences
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): key guide IPC answers by trimmed, present keys only
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): guide search hits carry a row id
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): make guide geometry DST-safe and tighten the contract
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): cache guide programmes per day with batched loading
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): add guide keyboard navigation controller
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): make guide programme cache robust to first-run effects and coverage failures
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): add the programme guide grid components
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(epg): add a Guide button to the timeline toolbar
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(m3u): adapt the playlist channel list to the guide contract
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(m3u): guard the guide's initial group scope and track language changes
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* feat(m3u): open the programme guide in place with a docked player
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): scope guide keys to the grid, clip the now-line and re-measure on resize
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(epg): remove the multi-EPG overlay and the channel-range IPC
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): document the programme guide and its release note
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* i18n(epg): translate the programme guide
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(m3u): let the guide own the keyboard and gate its entry points
While the programme guide is open the docked player carries
`data-player-shortcuts-suspended`, which `ControlsShortcuts` now honours
alongside `[inert]` — the arrows moved the player's volume instead of the
guide's row focus. The external-player strip loses its Collapse toggle
(nothing to reveal, no preference to write), the header action and its
palette command report `disabled` when the guide cannot open, the docked
strip derives its programme from the active channel's own schedule instead
of the retained NgRx value, switching playlists closes the guide, and the
collapsed strip can reach 48 px on phones.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(m3u): keep the sidebar mounted while the guide is open
Guide mode wrapped the sidebar in `@if (!guideOpen())`, so opening the
guide destroyed `app-channel-list-container`, whose `ngOnDestroy`
dispatches `resetActiveChannel()`. That cleared the active channel, which
unmounted the block hosting `app-epg-guide` and tripped the
`!canOpenGuide()` effect into closing the guide again: the guide never
appeared and the page dropped to "Please select a channel".
The sidebar now stays mounted and is hidden with
`.sidebar--guide-hidden` plus `inert`, so it is neither focusable nor read
by assistive technology while the guide owns the layout. Hiding also
preserves the channel list's scroll position across guide toggles.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(e2e): cover the programme guide flow
Imports a two-channel playlist with XMLTV, opens the guide from the
timeline toolbar and asserts the row list, the "Only with EPG" filter, a
channel switch that keeps the guide open, the hidden-but-mounted sidebar,
and that the player element survives both the mode and channel switches.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* chore(epg): tidy guide docs, palette gating and the unbound output
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): match guide favorites by channel URL and skip re-activating the playing row
Favorites are persisted by channel URL (FavoritesActions.updateFavorites),
so the Favorites scope compared the wrong key; the id stays as a legacy
fallback. A double-click arrives as click, click, dblclick and each
activate restarts playback, so the guide now leaves the already-playing row
alone and the commit path only closes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* perf(epg): let the guide window predicate use the programme time index
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(m3u): stabilise guide row identity, seed the sidebar group and provide translations in every player fixture
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(epg): split the guide shell, add a roving focus model and offset-aware search times
The shell component now owns rows, focus and the viewport only: the day,
zoom, density, filters, clock and day geometry move to EpgGuideViewState,
and every programme-dialog entry point to EpgGuideDialogController.
Keyboard navigation is reachable by assistive technology: exactly one grid
cell carries tabindex="0" (the focused cell, else the playing row's channel
cell, else the first row's), the guide moves DOM focus with it after each
handled key, a click hands the roving index to the clicked cell, and the
viewport, rows and cells expose grid/row/gridcell roles.
Search results were formatting raw provider instants, so they ignored the
EPG display offset; they go through getProgramTimeMs like every other time
the guide renders.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(m3u): make guide row ids collision-proof and gate the G shortcut
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): keep guide keys on the grid, reconcile focus with filtered rows and wrap the toolbar
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(epg): describe guide row ids as scope-local
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): clear guide search on scope change, match the active duplicate by url, keep failed coverage unknown
Search hits carry scope-local row ids, so a scope change drops them.
Two playlist entries can share an id but not a stream, so the active row
is matched by id + url before falling back to the id. A failed coverage
query now rejects instead of answering an empty set, which the guide
already treats as "coverage unknown" (every row stays visible).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): tell duplicate guide rows apart by group, keep G out of dialogs, use prototype-safe answers
The store spreads the selected channel, so the active row is matched by
id, url, group and name before widening; G no longer closes the guide from
a dialog or menu; guide answers use null-prototype records so a key named
__proto__ stays an own property.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): let coverage reject on lookup failures and compare whole entries for the active guide row
EpgQueryService.getChannelMetadata swallowed database errors into {}, so the
guide's coverage read could publish an empty set after a transient failure;
the guide now uses the strict resolveChannelMetadata (getChannelMetadata is
the fail-soft wrapper around it). The active guide row is matched on the
whole channel entry (all fields except the reducer-rewritten epgParams)
before widening to url and id, so copies that differ only in playback
headers or logo are told apart.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): let the guide return catch-up to live and normalise programme-search rows
The guide source contract gains an optional livePlayback signal: while the
host plays a catch-up URL, the active row may be activated again, which is
how the M3U host returns to live. EPG_DB_SEARCH_PROGRAMS now maps the raw
snake_case rows to the EpgProgram shape the bridge promises (plus the joined
channel name), so search hits resolve their channel and keep descriptions.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): name search hits, keep guide coverage strict on mapping failures
- Search results and the unresolved programme dialog show the channel's
display name (playlist row name, else the XMLTV display name the search
joined in) instead of the raw XMLTV id.
- The guide coverage read resolves manual mappings through a strict variant
that rejects on database failure, so a mapped channel can never be reported
as uncovered and hidden by "Only with EPG".
- Architecture doc describes the tiered active-row resolution.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): offer the Guide action in the list view too
The EPG list view mirrors the timeline's input/output contract, but the Guide
action was bound only in the timeline branch, so Settings → EPG → Guide view =
List lost the in-panel entry point. The list toolbar now carries the same
icon-only Guide button behind `guideAvailable`/`openGuide`, and the M3U
host binds it in both branches.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
## Summary
Live TV panels now fold from the outside in, in three nested levels, instead of one toggle that hid the categories rail and the channel list together:
1. **Categories + channels + player** (browse, unchanged).
2. **Channels + player** — a new `chevron_left` in the categories rail header hides only that rail. The channels header then turns its title into a **category dropdown** that opens the same shell panel as a popover (search, sort, counts, selection are one implementation), plus a `chevron_right` that brings the rail back.
3. **Player only** — the channels header chevron, as before. The floating restore handle and `Cmd/Ctrl+B` return to the level the user collapsed from, not always to level 1.
Every level is restored as stored, per surface (`live-sidebar-state:<surface>`, from #1555): a hidden rail is discoverable through the workspace header toggle and the hidden-list empty state that #1555 added, so this PR no longer needs its original "player-only never restores" rule. The level `Cmd/Ctrl+B` comes back to is seeded from the restored level and kept for the session.
## Design notes
- Nested levels rather than two independent booleans: "channels hidden, categories visible" makes no sense since a category click has to bring the channels back anyway. The model follows the outside-in collapse of three-pane apps (Mail, Slack, Plex).
- The categories rail folds at level 2 **only while a category is selected**: the live root ("All Items" grid) has no channels header to host the way back, so folding there would strand the user. Level 3 folds it regardless, because the floating restore handle lives in the content area.
- `LIVE_CATEGORIES_POPOVER` (`@iptvnator/portal/shared/util`) is the DI bridge: the workspace shell provides `WorkspaceLiveCategoriesPopoverService` (CDK overlay hosting `WorkspaceContextPanelComponent` in `presentation="popover"`), the Xtream and Stalker live layouts inject it optionally and keep their plain heading without a provider.
- M3U and the unified live tab have no categories rail and treat level 2 like level 1; their code is untouched.
## Merged with #1555 (per-surface rail state)
#1555 landed while this PR was open and reworked the same service: state per surface (`m3u` / `portal` / `collection`), a workspace header toggle, the hidden-list empty state, and the legacy shared key forgotten on startup. This PR keeps that model and layers the three levels onto the `portal` surface (`areCategoriesHiddenFor`, `hideCategories` / `showCategories` / `collapse` / `expand` per surface; `toggle(surface)` returns to the level the surface collapsed from). "Show playing channel" uses `expand('portal')` so it keeps a deliberately hidden categories rail folded, and the category sort preference moved to `PortalCategorySortStateService` so the popover copy of the context panel and the retained rail agree.
## Also fixed along the way
- The channels header showed "Channels" instead of the category name: provider category ids are strings, the selection is numeric. Compared via `String()` now.
- A collapsed context panel left a 22px padding strip beside the channels rail.
- The panel toggle labels said "Hide channels list" while also hiding categories; labels and tooltips are honest now (8 new i18n keys, all 18 locales).
Docs: `docs/architecture/iptvnator-ui-guidelines.md` ("Collapsible Live Sidebar" rewritten), `docs/architecture/workspace-shell.md`. Release note: `.changes/portals-live-panel-collapse-levels.md`.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(live-tv): keep a hidden channel list discoverable and scoped per surface
The second report in #1458 ("all channels disappear after clearing the
playback history, reset does not bring them back") was not data loss: the
history write never touches playlist items. The reporter's screenshot shows
a collapsed channel rail, a state persisted under one localStorage key
shared by the M3U player, the Xtream/Stalker live layouts and the
favorites/recent live tab. It survived restart, "Remove all playlists" and
re-import, and the only way back was a 32px chevron or Ctrl/Cmd+B.
- LiveLayoutSidebarStateService keeps the state per surface (m3u / portal /
collection) under live-sidebar-state:<surface>; the M3U player now goes
through the service instead of its own signal. The legacy shared key is
forgotten on startup and never read, so the update itself restores the
list for everyone who got stuck.
- The workspace header renders a view_sidebar toggle on every route that
renders its own rail (M3U all/groups, Xtream live, Stalker itv/radio), so
the control exists in both states instead of disappearing with the rail.
Collection pages keep their own toggle beside the content switch.
- While the rail is collapsed and nothing plays, every live host shows
app-channel-list-hidden-state (title, shortcut hint, full-size "Show
channels list" button) instead of asking to pick from a list that is not
on screen. app-portal-empty-state gained optional hint/action inputs.
- New LAYOUT.CHANNELS_LIST_HIDDEN(_HINT) strings in en plus 18 locales.
Tests: service, empty-state, hidden-state and header component specs, a
separate video-player-sidebar spec (the main M3U spec sits at the test
line budget), and an Electron E2E covering history clearing, restore via
button/header/shortcut across restart and re-import, per-surface scoping
against an Xtream portal, and legacy-key cleanup.
Refs #1458
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(live-tv): mirror the EPG offset setting in the sidebar spec mock
Master's player reads `resolvedEpgOffsetMinutes` from the settings store; the
new sidebar spec was cloned from the movie-gate harness before that field
landed, so its playing-channel case threw inside the EPG effect.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(web-e2e): scope the Stalker radio rail toggles to the rail
The workspace header now carries a second "Hide/Show channels list" toggle,
so the role+name locators matched more than one button and tripped
Playwright's strict mode. Target the rail's own chevron and the floating
restore button, and assert the header toggle mirrors the state.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(live-tv): honour Cmd/Ctrl+B on collection pages and hide the header rail toggle on phones
Codex review follow-ups on #1555:
- The hidden-list state advertises Cmd/Ctrl+B, but the favorites/recent
collection page had no handler; only the routed M3U/Xtream/Stalker live
layouts did. The page now toggles the collection surface while its live
tab is on screen, with the same typing/inert guards as the other hosts.
- At the phone breakpoint the header already holds the drawer toggle,
switcher, search and Add; the live rail is a bottom drawer with its own
toggle there, so the header rail toggle is hidden below 640px.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(live-tv): migrate the live navigation helpers to the per-surface sidebar API
master (#1554) added `XtreamLiveChannelNavigationService` and
`stalker-live-navigation.ts`, which expand the rail through
`sidebar.setState('expanded')` on the pre-split signature. Point them at the
`portal` surface and update their specs; drop the now-unused hidden-state
stub from the Xtream layout spec, which master pushed to the max-lines
budget.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(xtream): try advertised TS after initial web HLS HTTP failure
* refactor(playback): extract fullscreen channel panel state
* test(xtream): keep synthetic media within the mock project
* feat(epg): compact timeline toolbar with icon-only Now and cycling zoom
Replace the 96px zoom range slider with a single icon button that cycles
day overview → by hour → detailed, make the Now button icon-only (label in
tooltip + aria-label), and add Ctrl/⌘ + wheel zoom around the cursor on the
ribbon (preventDefault keeps Chromium from page-zooming the same gesture).
Zoom math moves into TimelineZoomController so the component stays under
the max-lines limit. The channel/programme heading no longer carries a 60%
width cap and takes the space the controls free up.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(epg): coalesce wheel-zoom anchoring and make the e2e zoom reset deterministic
- TimelineZoomController keeps a logical pending scrollLeft across a burst
of wheel/pinch events and applies it in one animation frame, so every
event anchors on the same minute instead of the stale DOM position
- the e2e settles on the detail preset from whichever band the wheel
landed in (contrast checks need the block's time line, hidden at
narrower tiers)
- zoom tooltip advertises ⌘/Ctrl per the repo's shortcut convention
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(portals): preserve live channel navigation while browsing
* test(portals): await media source assertion in remote E2E
* fix(xtream): capture destination queue for live auto-open
* fix(stalker): keep live search within the selected category
* test(stalker): assert retained video ownership without source timing
* test(stalker): distinguish paged All Items from the initial cache grid
* fix(stalker): reveal remote selections in uncached search results
* test(stalker): wait for category rows and retain settled playback
* fix(migration): recover legacy desktop sources without replacing current data
* test(migration): cover legacy recovery IPC contracts
* test(migration): use static legacy Electron bootstrap
* fix(playback): apply themes to player and EPG panels
* test(playback): verify active recording icon theme
* fix(epg): keep loading shimmer visible in both themes
* fix(playback): close legacy picture-in-picture on video replacement
* test(playback): wait for the selected video before PiP setup
* test(playback): await changed settings before PiP navigation
* fix(playback): release legacy WebKit picture-in-picture
Adds a global EPG display-time offset (Settings → EPG, whole minutes, ±720) for guides whose provider labels programme times with the wrong timezone. Display-only: parsed XMLTV values, SQLite rows, catch-up URLs and recording snapshots keep the provider's own times, so changing it needs no guide refresh. Closes the global part of #50.
The contract lives in `libs/shared/interfaces/src/lib/epg-display-offset.util.ts` with two equivalent forms: `epgDisplayTimeMs` shifts a programme for display, `epgProviderClockMs` shifts "now" into the provider's clock for every "currently airing" decision — the batched `GET_CURRENT_PROGRAMS_BATCH` lookup takes an explicit `nowMs`, and the channel lists, the Xtream/Stalker previews, the M3U player's current-programme mirror, the unified collection resolver, the dashboard live cards and the recording overlap all pick the same programme the guide renders as "now". Portal short-EPG windows start at the provider's own "now", so under a non-zero offset the Xtream preview surfaces cut their window from the full guide at the provider clock, Stalker short-EPG requests are widened for negative offsets, and every per-stream memory of the previous offset is retired together when the setting changes.
Co-authored-by: Mark Jardine <markjardine27@gmail.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(playback): seek Embedded MPV steps relative to mpv's own position
Arrow keys and the ±10 s buttons in the Embedded MPV player advanced only
about a second per press when pressed repeatedly or held. The shortcuts
already asked for 5 s steps, but `EmbeddedMpvCommandRunner.seekBy` turned
each step into an absolute `seek` computed from `session.positionSeconds`,
which is floored to whole seconds, polled every 500 ms (helper snapshots at
most every 250 ms) and not refreshed by the seek reply. Every press inside
that window therefore landed on the same target.
Steps now go through a new `EMBEDDED_MPV_SEEK_BY` IPC / `seekEmbeddedMpvBy`
bridge method that every backend forwards as mpv `seek <delta>
relative+exact`: `seekBy` exports in the macOS addon and the Windows/Linux
`wid` addon (Linux over its JSON IPC socket), and a `seek-by` stdin command
in the frame-copy helper. mpv resolves the delta against its own position
and merges queued relative seeks, so presses accumulate as in mpv itself.
The absolute form survives only as a fallback for a preload without the
method or an addon binary without `seekBy`; the timeline scrub still
commits an absolute target.
Validated with a real mpv 0.39 IPC probe: three relative seeks in a burst
advance +15 s, three absolute seeks from one stale base advance +5 s.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(playback): drop speculative position update from relative Embedded MPV seeks
Review follow-up for the relative seek path.
The macOS and Windows/Linux `seekBy` exports advanced `snapshot.positionSeconds`
by the delta after dispatching the mpv command. That is not idempotent the way
the absolute seek's optimistic write is: the observer (mpv event thread, or
the Linux IPC poll) can already have stored the post-seek `time-pos` under the
same mutex, so adding the delta on top counted the step twice, and while paused
nothing corrected it. On Linux it also advertised a position that a failed
socket delivery never reached. Relative steps now leave the snapshot alone;
only the observed `time-pos` updates the position.
The packaged Linux frame-copy smoke now drives `seekEmbeddedMpvBy` through the
built app: a burst of three +2 s steps issued without waiting for snapshots has
to land on 6 s, and a -60 s step has to clamp at 0. The generated Y4M fixture
grows from 2 s to 12 s (about 415 KB) so the burst and the playing section that
follows stay inside the clip. Replayed against a local mpv 0.39 with the same
fixture and media server: burst -> 6.0, -60 -> 0.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs(agents): mirror the Embedded MPV relative-seek contract into AGENTS.md
Review follow-up: the Shared Player Controls section documents the frame-copy
commands and shortcuts, so the relative seekEmbeddedMpvBy invariant lives
there too, next to the CLAUDE.md note.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(playback): reject a Linux relative seek the mpv IPC socket did not accept
Review follow-up: the Linux branch of SeekBy discarded the socket transaction
result and returned normally, so a step that never reached mpv looked like a
seek still awaiting observation. It now throws like a failed mpv_command_async
on the in-process engines; the renderer swallows the rejection and resyncs
from the next snapshot, and the main process logs it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Settings > General gains "Window on startup" (normal / maximized /
fullscreen), Electron only, mirrored into the main-process config by
SETTINGS_UPDATE and applied at the next window creation. `--fullscreen`
forces one fullscreen launch (consumed by the first window). F11 toggles
OS-level fullscreen through WINDOW:TOGGLE_FULLSCREEN — the exit path on
Windows/Linux where the title bar is hidden — and is skipped while the
player owns document.fullscreenElement.
attachWindowStateEvents tracks native and HTML fullscreen as two flags,
since Electron leaves only the HTML state when the window was already
natively fullscreen. macOS ignores the constructor `fullscreen` option on a
hidden window, so ready-to-show repeats the request after show(). Toggles
are decided by an observe-only, event-fed tracker
(native-fullscreen-transitions.ts), never against isFullScreen().
Closes#1455
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* perf(database): commit catalog writes in row-budgeted transactions
Refreshing or deleting a large Xtream playlist spent most of its time in
"Removing cached content": every 100 rows were deleted in their own
transaction, so a 300k-row catalog cost ~3,000 commits, each flushing an
FTS5 segment, re-appending dirty index pages to the WAL and, about every
4 MB, running an fsync-ing auto-checkpoint. Measured on a 900k-row copy of
a real database the delete took 13 s where a single set-based statement
takes 3 s, with 2 GB of WAL traffic instead of 140 MB. The re-import wrote
its rows the same way.
Deletes now read content row counts per category from the covering
indexes, pack categories into groups of about 5,000 rows and issue one
set-based DELETE per group, then drop the categories (and, for playlist
removal, the user-data tables) with one scoped statement each. Inserts keep
100-row statements but commit fifty of them at a time. Cancellation still
lands between commits and progress still reports after each one; the
worker additionally throttles progress events to one per 100 ms with
summed increments, so a large operation no longer floods the renderer.
Same subset, same machine: 13.0 s -> 5.6 s for the delete, 16.7 s -> 8.6 s
for the insert; the fsync-bound share is larger on Windows and spinning
disks.
Closes#1292
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(database): pin the row budget and the worker-side progress flush
Review follow-up: the default 5,000-row budget and the 50-statement insert
commit were only exercised with explicit overrides or sub-budget inputs, and
nothing covered the worker controller flushing a coalesced progress report
before its terminal event. Both are now pinned, and the docs no longer claim
the insert path reports SQLite `changes` or binds 1,600 parameters per
statement (it binds the eleven columns a value supplies).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test(e2e): keep the stress suites on 100-row commits via a test-only budget knob
The Xtream responsiveness and playlist-switcher suites slow the database
worker down with IPTVNATOR_DB_WORKER_BATCH_DELAY_MS so they can observe an
import mid-flight; the stress catalog is 1,920 rows per type, which at the
production budget of 5,000 rows per commit is a single commit per type and
too few progress events for their assertions. The new companion knob
IPTVNATOR_DB_WORKER_ROWS_PER_TRANSACTION restores 100-row commits for those
runs only; unset or invalid it leaves the default untouched.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(database): scope refresh and cache-clear deletes to the captured category ids
The row-budgeted rewrite deleted a refreshed playlist's categories with a
playlist-wide predicate. The worker serves other requests between commits,
so a newer import of the same playlist could create categories in that
window and lose them to the older refresh, after which its content inserts
fail their foreign keys. Count and delete now use the ids the collection
step read, as the chunked code did; playlist removal keeps its playlist
scope because its final playlist-row delete cascades the same set.
deleteCategoriesWhere runs every filter through requireScopedFilter.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Kate Markson