feedBack/static/js/session.js
Byron Gamatos 8d3db5f42c
Some checks are pending
ship-ci / ci (push) Waiting to run
fix(nav): nobody may monkey-patch window.showScreen — add screen:changing, make the shell listen (#924) (#925)
* fix(nav): the library sometimes showed the legacy screen — map 'home' inside showScreen

Testers: "randomly, when moving to the library from another menu option, the library shows the
old interface — never when a song ends."

━━━ WHAT WAS ACTUALLY HAPPENING ━━━

#home is the PRE-V3 library screen. The v3 shell replaced it with #v3-songs, and the mapping DID
exist — but only inside WRAPPERS on window.showScreen, and only for callers that go through
`window`. THREE independent parties monkey-patch it, each capturing whatever happens to be there
at the time:

    app.js publishes the raw function
      -> shell.js wraps it, adding the home -> v3-songs mapping
      -> the stems plugin wraps it AGAIN (src/main.js:1029), capturing the current value

Plugins load ASYNCHRONOUSLY. The chain links up in whatever order the race settles, and any
capture taken before shell.js installs — or any re-assignment after it — silently drops the
mapping. Hence "randomly".

AND THE INTERNAL CALLERS NEVER TOUCHED window.showScreen AT ALL. closeCurrentSong and the
Esc-from-settings shortcut call the IMPORTED showScreen, which no wrapper ever sees. Reproduced
in a browser: the unwrapped function with 'home' lands on the dead legacy screen EVERY time.

"Never when a song ends" is the tell, and it is what identified the mechanism: closeCurrentSong
resolves its target through _resolvePlayerOrigin(), which ALREADY applies this mapping. That one
path was fine — which is exactly why the bug looked random rather than total.

PRE-EXISTING, not a regression from the module carve: the onclick="showScreen('home')" links and
the wrapper-only mapping both date to 2026-06-22.

━━━ THE FIX ━━━

The guard lives inside showScreen now: ONE place, in the function every caller routes through,
instead of a chain of monkey-patches that must each remember. Wrapper order stops mattering, and
the module-internal callers are covered for the first time.

Verified in a browser: the raw, unwrapped showScreen('home') — which reproduced as #home — now
lands on #v3-songs, and cannot be undone by any wrapper order.

━━━ AND A [P1] I INTRODUCED, WHICH CODEX CAUGHT ━━━

My first cut mapped BOTH 'home' and 'v3-home', copied straight from _resolvePlayerOrigin.

That is correct THERE and wrong HERE. _resolvePlayerOrigin computes where to RETURN TO after a
song, and landing on the Songs list from the dashboard is the right behaviour. But #v3-home is
the v3 DASHBOARD — a real screen that the shell's Home nav, the onboarding tour and the dashboard
re-render listener all target. Redirecting it would have made Home unreachable.

A LEGACY ALIAS IS NOT THE SAME THING AS A RETURN TARGET. Only 'home' is mapped now, and a test
pins that: re-adding 'v3-home' to the guard fails it.

4 tests, bite-tested both ways.

node 1049, pytest 2425, ESLint 0, Codex 0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(nav): nobody may monkey-patch window.showScreen — add screen:changing, make the shell listen (#924)

window.showScreen was wrapped by THREE independent parties, each capturing whatever happened to be
there at the time:

    app.js publishes the raw function
      -> static/v3/shell.js wrapped it (to call syncActive, and to map home -> v3-songs)
      -> the stems plugin wrapped it AGAIN (to tear down on leaving the player)

Plugins load ASYNCHRONOUSLY, so the chain linked up in whatever order the race settled. A capture
taken before shell.js installed silently dropped the mapping it carried — and the library opened on
the dead legacy #home screen. Testers saw that as "randomly, the library shows the old interface"
(#923).

#923 fixed the symptom by moving the mapping inside showScreen. This removes the CAUSE: neither
wrapper ever needed to be one.

━━━ TWO EVENTS, AND THE DISTINCTION IS THE WHOLE POINT ━━━

    screen:changing  emitted BEFORE anything happens. "I am leaving `from`." Teardown/cancel here.
    screen:changed   emitted after the DOM and data settle. "I am on `id`." Now carries `from`.

screen:changing is new, and it exists because Codex caught me collapsing the two. The stems plugin
tore down its audio graph BEFORE showScreen did anything; screen:changed fires at the very END,
after core awaits library and provider loads — so moving the plugin onto it would have delayed
teardown behind a slow fetch, or skipped it entirely if that fetch threw, and stems would keep
playing on a non-player screen. A test pins the ordering: screen:changing must precede the first
await.

shell.js is a plain screen:changed listener now, like app.js, audio-mixer.js and tour-engine.js
already were. window.showScreen is an unwrapped function again, and tests/js/
no_showscreen_monkeypatch.test.js fails CI if anything in static/ ever assigns to it again — so the
hazard is structurally impossible rather than merely avoided.

━━━ AND A FALLBACK THAT COULD NEVER FIRE ━━━

My retry-if-the-bus-is-late path listened for `slopsmith:capabilities:ready`. Core dispatches
`feedBack:capabilities:ready` (capabilities.js:1536) — the slopsmith: name is the PRE-DMCA event
and nothing has emitted it since the rename. Codex caught it. A guard that cannot fire is worse
than no guard: it reads as protection and is decoration.

(The same dead-event bug turned out to be sitting in THREE of the stems plugin's fallbacks, where
it has silently disabled its lifecycle wiring whenever the bus was late. Fixed in
feedback-plugin-stems#38.)

VERIFIED. A/B against origin/main: the nav highlight and topbar title follow IDENTICALLY with
shell.js as a listener; screen:changing -> screen:changed fire in order with the right {id, from};
window.showScreen is unwrapped; and showScreen('home') still lands on v3-songs.

node 1053, pytest 2425, ESLint 0, Codex 0.

Closes #924

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-12 16:06:28 +02:00

749 lines
36 KiB
JavaScript

//
// ━━━ THIS WAS THE UNCUTTABLE HEART, AND IT IS 359 LINES ━━━
//
// At the start of the app.js carve, seeding a dependency closure from count-in, from loops, from
// section-practice or from the JUCE seek shim all returned the SAME 178-function, 3,360-line
// set. playSong and showScreen called each other; everything called them; nothing could be cut
// anywhere. The conclusion — correct at the time — was that no closure-based carve could touch
// it at any seed, and the answer was a HOST SEAM.
//
// That was true THEN. It is not true now. Every slice taken out since (transport, loops,
// count-in, section-practice, the library, the edit modal, settings) removed edges, and the
// strongly-connected component DISSOLVED. This closure is 36 declarations with an interface
// width of FOUR.
//
// The lesson is not that the seam was wrong. The seam is what MADE this possible: it let the
// carves proceed against a cyclic core instead of stalling on it. The lesson is to RE-MEASURE.
// An SCC is a fact about a graph at a moment, not a property of the code.
//
// ━━━ THE GATE STATEMENTS AT THE BOTTOM, AND WHY NO SCAN FOUND THEM ━━━
//
// window.feedBack.holdAutoplay / holdAutoExit and their two event handlers are TOP-LEVEL
// STATEMENTS, not declarations. They WRITE this module's state (_autoplayHeld, _autoExitTimer,
// …), and an imported binding is READ-ONLY — so left behind in app.js, every one of them threw
// "Assignment to constant variable" the instant this module existed.
//
// A dependency scan that walks DECLARATIONS cannot see them. Only the browser A/B did. It is the
// same blind spot that nearly shipped a dead library A-Z rail (#896): app.js keeps its public
// API in top-level statements, and those are invisible to a call-graph.
//
// ━━━ ZERO OUTSIDE WRITES, BY MOVING THE BOUNDARY RATHER THAN BUILDING MACHINERY ━━━
//
// Autoplay scalars and the wake-lock state were written from outside — which would have forced a
// setter or a container. But the writers (_releaseAutoplay, _acquireWakeLock) plainly belong
// here. Pulling them in left ZERO outside writes, so every export is a plain import. Same move as
// settings (#920): measure the writers before you reach for a container.
import {
loadSettings,
} from './settings.js';
import {
clearLoop,
loadSavedLoops,
} from './loops.js';
import {
audio,
} from './audio-el.js';
import {
_snapshotResumeSession,
} from './resume-session.js';
import {
_resetJuceAudioShimChain,
} from './juce-audio.js';
import {
_hideSectionPracticeBar,
_resetSectionPracticeLog,
_scheduleSectionPracticeRetries,
} from './section-practice.js';
import {
_cancelCountIn,
armCreditsHideOnPlay,
hideSongCreditsOverlay,
holdCreditsThen,
scheduleCreditsHide,
showSongCreditsOverlay,
startSongCountIn,
} from './count-in.js';
import {
_autoplayExitEnabled,
_countdownBeforeSongEnabled,
_resetPlaybackSpeedForNewSong,
} from './player-controls.js';
import {
_audioTime,
_resetAudioSeekState,
_songEventPayload,
jucePlayer,
setPlayButtonState,
togglePlay,
} from './transport.js';
import {
_activeLibraryProviderId,
_bumpLibNavGeneration,
_getArrangementNamingMode,
_libScrollOnNextRender,
_resetLibraryProviderViewState,
loadFavorites,
loadLibrary,
loadLibraryProviders,
stopInfiniteScroll,
} from './library.js';
import {
S,
} from './player-state.js';
import {
L,
} from './library-state.js';
// Tracks which list screen launched the player so Esc-from-player
// returns the user to that screen instead of always defaulting to
// the Library (feedBack#126). Reset on every `playSong` call so a
// song launched from a deep-link / plugin screen still gets a sane
// fallback ('home').
export let _playerOriginScreen = 'home';
export let _settingsOriginScreen = 'home';
// ── Screen Navigation ─────────────────────────────────────────────────────
export async function showScreen(id) {
// ── 'home' is the LEGACY library screen. Always route it to the v3 Songs list. ──
//
// The v3 shell replaced #home with #v3-songs. That mapping DID exist — but only inside
// wrappers on `window.showScreen`, and only for callers that go through `window`:
//
// app.js publishes the raw fn -> shell.js wraps it (adding the mapping)
// -> the stems plugin wraps it AGAIN, capturing whatever
// happened to be there at the time
//
// Two ways that fails, and testers hit both:
//
// 1. ORDER. Three independent parties monkey-patch window.showScreen, each capturing the
// current value. Plugins load ASYNCHRONOUSLY, so the chain links up in whatever order
// the race settles — and any capture taken before shell.js installs, or any
// re-assignment after it, silently drops the mapping.
//
// 2. THE INTERNAL CALLERS NEVER TOUCHED window.showScreen AT ALL. closeCurrentSong and the
// Esc-from-settings shortcut call the IMPORTED showScreen directly, so no wrapper ever
// sees them. Verified in a browser: the unwrapped function with 'home' lands on the dead
// legacy screen every single time.
//
// Hence "randomly, when moving to the library from another menu option" — and "never when a
// song ends", because closeCurrentSong resolves its target through _resolvePlayerOrigin(),
// which already applies this mapping.
//
// So it lives HERE now: ONE guard in the function every caller routes through, rather than a
// chain of monkey-patches that must each remember.
//
// ONLY 'home'. NOT 'v3-home'. _resolvePlayerOrigin() maps BOTH — correctly, because it
// computes where to RETURN TO after a song, and coming back to the Songs list from the
// dashboard is the right behaviour. Copying that condition here was a [P1] (Codex caught it):
// #v3-home is the v3 DASHBOARD, a real screen the shell's Home nav, the onboarding tour and
// the dashboard re-render listener all target. Redirecting it would make Home unreachable.
//
// A legacy alias is not the same thing as a return target.
if (id === 'home' && document.getElementById('v3-songs')) {
id = 'v3-songs';
}
// Capture the previous screen before changing active classes
const prevScreenId = document.querySelector('.screen.active')?.id;
// ── screen:changing — emitted BEFORE any of the work below ──────────────────
//
// Timing matters here, and Codex caught me getting it wrong. The stems plugin used to
// monkey-patch window.showScreen so it could tear down its audio graph BEFORE navigation
// began. screen:changed fires at the very END of this function — after awaiting library and
// provider loads — so moving that plugin onto it would have delayed teardown behind a slow
// fetch, or skipped it entirely if the fetch threw. Stems would keep playing on a non-player
// screen.
//
// So there are two events, and the distinction is the whole point:
// screen:changing — before anything happens. "I am leaving `from`." Cancel/teardown here.
// screen:changed — after the DOM and data are settled. "I am on `id`."
if (window.feedBack) window.feedBack.emit('screen:changing', { id, from: prevScreenId || null });
document.querySelectorAll('.screen').forEach(s => s.classList.remove('active'));
document.getElementById(id).classList.add('active');
// Mark the next render as a screen-entry so it scrolls the
// restored selection into view exactly once. Routine renders
// (search / sort / filter typing) won't have this flag set and
// so won't yank the viewport. Also bump the nav-items
// generation so the next keypress doesn't reuse a cache built
// against a now-hidden screen's container.
_bumpLibNavGeneration();
if (id === 'home') {
_libScrollOnNextRender.home = true;
const beforeProviderId = _activeLibraryProviderId();
await loadLibraryProviders({ restoreSaved: true });
if (_activeLibraryProviderId() !== beforeProviderId) {
_resetLibraryProviderViewState();
} else {
L.libEpoch++;
L.currentPage = 0;
L.treeStats = null;
stopInfiniteScroll();
}
loadLibrary(0);
}
if (id === 'favorites') { _libScrollOnNextRender.favorites = true; loadFavorites(); }
if (id === 'settings') {
// Record where we came from so Esc can go back. The player screen
// is torn down by the `id !== 'player'` branch below, so
// re-entering it via showScreen() would land on a dead screen —
// fall back to the player's own origin (or 'home') instead.
if (prevScreenId && prevScreenId !== 'settings') {
_settingsOriginScreen = prevScreenId === 'player'
? (_playerOriginScreen || 'home')
: prevScreenId;
}
loadSettings();
}
if (id !== 'player') {
const audio = document.getElementById('audio');
const stopTime = _audioTime();
const hadPlayableSong = !!audio.src || !!window._juceAudioUrl || S.isPlaying;
// Snapshot where we were so leaving the player — especially by accident
// — is recoverable instead of dumping the user back at bar 1 next time.
// Must run BEFORE window.highway.stop()/audio unload, while getSongInfo() and
// the position (stopTime) are still live.
if (hadPlayableSong) _snapshotResumeSession(stopTime);
window.highway.stop();
// Cancel any queued seeks, in-flight shim closures, AND active
// count-in timers before stopping playback so none of these paths
// can mutate the torn-down session (mirrors the same triple reset
// in playSong()).
_cancelCountIn();
_resetJuceAudioShimChain();
_resetAudioSeekState();
if (window._juceMode) {
// HTML5 emits 'pause' via the media-element listener below;
// JUCE doesn't, so plugins would stay stuck in "playing".
// Snapshot the canonical payload BEFORE stop() resets _pos
// to 0, then emit AFTER stop completes. Mirrors the HTML5
// pause contract via _songEventPayload (audioT/chartT/perfNow).
const payload = _songEventPayload();
const wasPlaying = S.isPlaying;
await jucePlayer.stop().catch(() => {});
if (wasPlaying && window.feedBack) {
window.feedBack.isPlaying = false;
window.feedBack.emit('song:pause', payload);
}
window._juceMode = false;
window._juceAudioUrl = null;
}
if (hadPlayableSong) window.feedBack.emit('song:stop', { time: stopTime || 0, screen: id });
audio.pause();
audio.src = '';
window._currentSongAudio = null;
// Reloading any song later should get a fresh JUCE routing attempt.
window._clearJuceRerouteMemo?.();
S.isPlaying = false;
setPlayButtonState(false);
}
window.scrollTo(0, 0);
// `from` is the screen we just LEFT. Without it, "I am leaving the player" is not
// expressible from an event, and the only way to express it was to WRAP window.showScreen —
// which is what shell.js and the stems plugin both did, and why the library intermittently
// showed the legacy screen (#923, #924): three parties patching one global, each capturing
// whatever was there at the time, in whatever order the plugin loads settled.
//
// Additive: every existing listener (app.js, audio-mixer.js, tour-engine.js) reads `id` and
// is unaffected.
if (window.feedBack) window.feedBack.emit('screen:changed', { id, from: prevScreenId || null });
}
export let currentFilename = '';
export function _playbackApi() {
return window.feedBack && window.feedBack.playback && window.feedBack.playback.version === 1
? window.feedBack.playback
: null;
}
// Bridge hits are a "this legacy surface is still in use" signal, not a call
// counter — but recordBridgeHit is not cheap (compat-shim bookkeeping, a
// playback:bridge-hit event, and a diagnostics snapshot rebuild per call).
// Plugins legitimately poll read surfaces like window.feedBack.getLoop() from
// HUD ticks (note_detect polled at ~30 Hz), which turned every tick into a
// snapshot serialization on the main thread and saturated the inspector's
// hitCount. Throttle per surface: the first call records immediately, repeats
// within the window are dropped.
export const _bridgeRecordLast = new Map();
export const _BRIDGE_RECORD_MIN_MS = 5000;
export function _recordPlaybackBridge(bridgeId, legacySurface, reason) {
const playback = _playbackApi();
if (!playback || typeof playback.recordBridgeHit !== 'function') return;
const key = `${bridgeId}|${legacySurface}`;
const now = Date.now();
const last = _bridgeRecordLast.get(key);
if (last != null && now - last < _BRIDGE_RECORD_MIN_MS) return;
_bridgeRecordLast.set(key, now);
playback.recordBridgeHit({
bridgeId,
legacySurface,
source: 'core.app',
reason: reason || 'legacy playback surface used',
});
}
// Screen Wake Lock — keep the display awake while a song is playing so the
// OS screensaver doesn't kick in during windowed-mode playback (only audio +
// the highway animation are active, so the input-idle timer otherwise fires).
// Engaged only while playing (acquire on play/resume, release on
// pause/ended/stop) per issue #686. In a plain browser this uses the W3C
// Screen Wake Lock API; inside feedBack-desktop (Electron) navigator.wakeLock
// is unreliable, so we also drive the native powerSaveBlocker bridge when it
// is exposed — both calls are best-effort and degrade silently elsewhere.
export let _screenWakeLock = null;
export let _wakeLockPending = false;
// Desired state: true while a song should be keeping the screen awake. This is
// the source of truth that survives the async gap of navigator.wakeLock.request
// — set synchronously by acquire/release so an in-flight request that resolves
// after playback already stopped can release itself instead of leaking a lock.
export let _wakeLockWanted = false;
// Set when an acquire is requested while one is already in flight (e.g. a quick
// hide→show during the first request); the in-flight request retries once on
// settle so a transient NotAllowedError doesn't leave the song unprotected.
export let _wakeLockRetry = false;
// Last value handed to the desktop bridge. This is the value we *requested*,
// not one confirmed by the IPC round trip: the Electron main-process side
// effect (powerSaveBlocker start/stop) happens when the message is received,
// before its promise resolves, so deduping on the requested value lets opposite
// transitions (true↔false) always go through promptly while still suppressing
// redundant repeats (e.g. the synchronous song:play + song:resume pair). A
// rejected/throwing call invalidates the marker (the side effect never landed)
// so the next song:* / visibilitychange retries — without an inline re-sync,
// which would tight-loop on a persistently failing bridge.
// Last value handed to the bridge: false (off) / true (on) / null (unknown —
// a call failed, so the real blocker state can't be assumed). null never equals
// a boolean `want`, so the next sync always re-sends and recovers.
export let _desktopAwakeReq = false;
// Monotonic id of the most recent bridge call, so a stale (out-of-order)
// rejection from a superseded call can be ignored rather than corrupting the
// marker — a boolean alone can't tell "my request failed" from "an older
// same-valued request failed after a newer one already succeeded".
export let _desktopAwakeGen = 0;
// Drive the native feedBack-desktop blocker to exactly (wanted && visible),
// mirroring the browser wake lock which is only held while the page is visible.
// Gating on visibility stops a minimized Electron window from keeping the whole
// display awake. No-op in a plain browser; isolated from the wakeLock path so a
// flaky bridge can't abort it.
export function _syncDesktopBridge() {
const want = _wakeLockWanted && document.visibilityState === 'visible';
if (want === _desktopAwakeReq) return; // already requested this value
const bridge = window.feedBackDesktop?.power?.setScreenAwake;
if (typeof bridge !== 'function') return; // plain browser — nothing to sync
_desktopAwakeReq = want;
const gen = ++_desktopAwakeGen;
let r;
try {
r = bridge(want);
} catch (e) {
console.debug('desktop wake bridge failed:', e?.name || e);
if (gen === _desktopAwakeGen) _desktopAwakeReq = null; // unknown — force a re-send next event
return;
}
if (r && typeof r.then === 'function') {
r.catch((e) => {
console.debug('desktop wake bridge rejected:', e);
// The IPC didn't take effect; we can't assume which state the blocker
// is in (a prior call may also have failed), so mark it unknown and
// let the next song:* / visibilitychange re-send. Only if this is
// still the latest request — a stale rejection from a superseded call
// must not clobber a newer request's marker.
if (gen === _desktopAwakeGen) _desktopAwakeReq = null;
});
}
}
export async function _acquireWakeLock() {
_wakeLockWanted = true;
_syncDesktopBridge();
if (_screenWakeLock) return; // already held — nothing to do
// A request is already in flight (song:play and song:resume fire
// synchronously from the audio 'play' listener, and visibilitychange can
// re-enter): don't issue a duplicate, but remember to retry on settle so a
// visibility bounce during the request can't strand us without a lock.
if (_wakeLockPending) { _wakeLockRetry = true; return; }
if (!navigator.wakeLock?.request) return;
_wakeLockPending = true;
_wakeLockRetry = false;
try {
const sentinel = await navigator.wakeLock.request('screen');
if (!_wakeLockWanted) {
// Playback stopped while the request was in flight — release the
// just-granted lock immediately rather than holding it stale.
try { await sentinel.release(); } catch (e) { /* already released */ }
return;
}
_screenWakeLock = sentinel;
sentinel.addEventListener('release', () => {
_screenWakeLock = null;
// The UA auto-releases on tab hide, but may also release for its own
// reasons (power policy) while the page stays visible. Re-acquire if
// a song is still playing and we're visible — the visibilitychange
// handler covers the hidden→visible case.
if (_wakeLockWanted && document.visibilityState === 'visible') {
_acquireWakeLock();
}
});
} catch (e) {
// NotAllowedError (page hidden / no user activation) or unsupported.
console.debug('wakeLock request failed:', e?.name || e);
} finally {
_wakeLockPending = false;
// A re-acquire arrived while the request was in flight (typically a
// hide→show bounce). If we still want the lock, are visible, and didn't
// get one (the request raced a hidden window and rejected), try once
// more now that the page state has settled. Bounded: only fires when a
// bounce actually occurred, so a permanently-denied request can't loop.
if (_wakeLockRetry && _wakeLockWanted && !_screenWakeLock
&& document.visibilityState === 'visible') {
_wakeLockRetry = false;
_acquireWakeLock();
}
}
}
export async function _releaseWakeLock() {
_wakeLockWanted = false;
_syncDesktopBridge();
if (!_screenWakeLock) return;
try { await _screenWakeLock.release(); } catch (e) { /* already released */ }
_screenWakeLock = null;
}
// Resolve where the player should return on Esc / close / auto-exit.
// A one-shot setReturnScreen() override wins (consumed here) — used by the
// lessons catalog so a lesson returns to the lessons screen rather than the
// library, even though the external tutorials plugin owns the playSong call.
// Otherwise remember the actual launch screen; the element-exists guard
// keeps the classic v2 UI (no #v3-* ids) from being stranded on a missing
// screen, and unknown launches fall back to 'home'. The dashboard — classic
// 'home' and the v3 shell's 'v3-home' — returns to the Songs list when it
// exists (dashboard actions call playSong() directly, so its id is the
// active screen at launch).
export function _resolvePlayerOrigin() {
const override = window.feedBack && window.feedBack._nextReturnScreen;
if (window.feedBack) window.feedBack._nextReturnScreen = null;
if (override && document.getElementById(override)) return override;
const launchFrom = document.querySelector('.screen.active');
const launchId = launchFrom && launchFrom.id;
if (launchId && launchId !== 'player' && document.getElementById(launchId)) {
return ((launchId === 'home' || launchId === 'v3-home') && document.getElementById('v3-songs'))
? 'v3-songs' : launchId;
}
return 'home';
}
// Autoplay: one-shot flag armed by each fresh playSong(), consumed by the
// next song:ready. song:ready also fires on arrangement switches / seeks,
// which never arm the flag, so those don't auto-restart.
export let _pendingAutostart = false;
// Autoplay gate (window.feedBack.holdAutoplay): a plugin (the tuner) can defer the
// auto-start of a freshly-loaded song until it's cleared — "tune before you play".
// The hold is claimed synchronously on song:loading (so it beats this song:ready
// autostart); release() — or a fail-open backstop — runs the deferred start.
// Generation-guarded so a newer song invalidates a stale hold. Manual Play never
// flows through here, so Play always wins.
export let _autoplayHeld = false;
export let _autoplayStart = null;
export let _autoplayGen = 0;
export let _autoplayBackstop = null;
export const AUTOPLAY_HOLD_BACKSTOP_MS = 12000;
export function _clearAutoplayHold() {
if (_autoplayBackstop) { clearTimeout(_autoplayBackstop); _autoplayBackstop = null; }
_autoplayHeld = false;
_autoplayStart = null;
_autoplayGen++;
}
export function _releaseAutoplay(gen) {
if (gen !== _autoplayGen) return; // a newer song superseded this hold
if (_autoplayBackstop) { clearTimeout(_autoplayBackstop); _autoplayBackstop = null; }
_autoplayHeld = false;
const start = _autoplayStart;
_autoplayStart = null;
if (typeof start === 'function') start();
}
export let _autoplayHoldToken = 0;
window.feedBack.holdAutoplay = function () {
const gen = _autoplayGen;
const token = ++_autoplayHoldToken; // this hold's identity — a stale release from an earlier hold is a no-op
_autoplayHeld = true;
if (_autoplayBackstop) clearTimeout(_autoplayBackstop);
// Fail-open: a hold that's never released (a plugin that claimed but wedged before
// it could decide) must never permanently block the song. Once the holder commits
// to an intentional, user-dismissable hold it calls release.settle() to cancel this
// — so the backstop can't cut off e.g. a user still tuning past the timeout.
_autoplayBackstop = setTimeout(() => _releaseAutoplay(gen), AUTOPLAY_HOLD_BACKSTOP_MS);
let released = false;
function release() {
if (released || gen !== _autoplayGen || token !== _autoplayHoldToken) return;
released = true;
_releaseAutoplay(gen);
}
// Cancel the fail-open backstop WITHOUT releasing: the holder has taken explicit
// responsibility for releasing (on dismiss), and a song switch clears the hold anyway.
release.settle = function () {
if (gen !== _autoplayGen || token !== _autoplayHoldToken) return;
if (_autoplayBackstop) { clearTimeout(_autoplayBackstop); _autoplayBackstop = null; }
};
return release;
};
window.feedBack.on('song:ready', () => {
if (!_pendingAutostart) return;
_pendingAutostart = false;
if (S.isPlaying) return;
// Feedpak contributor credits: only real feedpak plays carry authors
// (loose/archive and minigames get []), so a non-empty list is the gate.
// Shown over the highway and dismissed the moment real playback begins
// (song:play). This fresh-load path is the only place it fires —
// arrangement switches / seeks / manual replays never arm _pendingAutostart,
// and minigames never get here. Decoupled from autoplay below so credits
// show on load even when autoplay-exit is disabled.
const authors = (window.feedBack.currentSong && window.feedBack.currentSong.authors) || [];
if (authors.length) {
showSongCreditsOverlay(authors);
armCreditsHideOnPlay();
}
// Autoplay-exit disabled: don't auto-start. Still let the credits dwell a
// couple seconds on the freshly-loaded song, then clear them (they also
// clear early if the user manually presses Play, via _creditsHideOnPlay).
if (!_autoplayExitEnabled()) {
if (authors.length) scheduleCreditsHide();
return;
}
// The actual auto-start: a count-in (which handles HTML5 + _juceMode) or the
// Play path directly. Guarded so a manual Play during a gate / credits hold
// can't double-toggle, and so a stale (released-after-leaving) start never
// begins playback off the player.
const start = () => {
if (S.isPlaying) return;
if (!document.getElementById('player')?.classList.contains('active')) { hideSongCreditsOverlay(); return; }
if (_countdownBeforeSongEnabled()) {
Promise.resolve(startSongCountIn()).catch((err) => console.warn('[app] song count-in failed:', err));
} else {
Promise.resolve(togglePlay())
.then(() => { if (!S.isPlaying) hideSongCreditsOverlay(); })
.catch((err) => { console.warn('[app] autoplay failed:', err); hideSongCreditsOverlay(); });
}
};
// A plugin (the tuner) may gate playback until it's cleared. The hold was
// claimed on song:loading; stash the start and let release()/the backstop run
// it. _cancelCountIn()/changeArrangement() clear _creditsTimer below, so a
// teardown during the credits dwell still cancels a non-gated play.
if (_autoplayHeld) { _autoplayStart = start; return; }
// Not gated: a count-in starts now (it owns its on-screen dwell); otherwise
// let the credits dwell a couple seconds first, then start.
if (_countdownBeforeSongEnabled() || !authors.length) start();
else holdCreditsThen(start);
});
// Auto-exit: when the song ends, return to the launching menu. A scoring
// plugin that shows an end-of-song results screen calls holdAutoExit() to
// defer this; the user closing that screen (its Close button calls
// window.closeCurrentSong()) performs the exit. With no results screen the
// grace timer returns to the menu on its own.
export const AUTO_EXIT_GRACE_MS = 1500;
export let _autoExitTimer = null;
export let _autoExitHeld = false;
// Bumped every time the auto-exit state is reset (new song via playSong, and
// each song:ended). A hold's release() captures the generation at hold time
// and no-ops once it changes, so a plugin that drops or fires its release
// handle after the player has moved on can never navigate a fresh session —
// callers don't need to balance the handle.
export let _autoExitGen = 0;
export function _clearAutoExit() {
if (_autoExitTimer) { clearTimeout(_autoExitTimer); _autoExitTimer = null; }
_autoExitHeld = false;
_autoExitGen++;
}
// Heuristic safety net for score-screen plugins that don't (yet) call
// holdAutoExit(): if a visible full-screen results/dialog overlay is on top
// when the grace timer fires, defer the auto-return and let that screen's
// own close button drive the exit (its Close should call closeCurrentSong).
// getClientRects() is used for the visibility test because it reports
// position:fixed overlays correctly, unlike offsetParent.
export function _resultsOverlayVisible() {
let nodes;
try {
nodes = document.querySelectorAll('[role="dialog"][aria-modal="true"], .fixed.inset-0');
} catch (_) { return false; }
for (const el of nodes) {
if (!el || el.id === 'player') continue; // never the player itself
if (el.classList && el.classList.contains('hidden')) continue;
if (el.getClientRects && el.getClientRects().length > 0) return true;
}
return false;
}
// Plugins call this synchronously from their own song:ended handler (core
// runs first, so the timer is already pending) to claim the exit.
window.feedBack.holdAutoExit = function () {
if (_autoExitTimer) { clearTimeout(_autoExitTimer); _autoExitTimer = null; }
_autoExitHeld = true;
const gen = _autoExitGen;
let released = false;
return function release() {
// No-op once released, or once the session has moved on (a newer
// playSong / song:ended bumped the generation) — so a stale handle
// never navigates away from a fresh song.
if (released || gen !== _autoExitGen) return;
released = true;
if (typeof window.closeCurrentSong === 'function') window.closeCurrentSong();
};
};
window.feedBack.on('song:ended', () => {
_clearAutoExit();
if (!_autoplayExitEnabled()) return;
// Only auto-exit from the player screen (ignore stale/duplicate ends).
const active = document.querySelector('.screen.active');
if (!active || active.id !== 'player') return;
_autoExitTimer = setTimeout(() => {
_autoExitTimer = null;
if (_autoExitHeld) return; // a plugin explicitly claimed the exit
if (_resultsOverlayVisible()) return; // a score/results overlay is up; let it drive the exit
const cur = document.querySelector('.screen.active');
if (cur && cur.id === 'player' && typeof window.closeCurrentSong === 'function') {
window.closeCurrentSong();
}
}, AUTO_EXIT_GRACE_MS);
});
// Abort controller for cancelling pending requests when entering player
export let artAbortController = null;
export async function playSong(filename, arrangement, options) {
console.log('playSong called:', filename);
// A manual (non-queue) play abandons any active play-queue, so a stale queue
// can't hijack the next song's end. The queue passes fromQueue to keep itself.
if ((!options || !options.fromQueue) && window.feedBack && window.feedBack.playQueue) {
window.feedBack.playQueue.clear();
}
if (!options || options.bridge !== false) {
_recordPlaybackBridge('playback.window-play-song', 'window.playSong', 'legacy playSong entry point used');
}
// Invalidate any prior song's autoplay gate before plugins re-claim it on the
// song:loading emit below.
_clearAutoplayHold();
window.feedBack.emit('song:loading', { filename, arrangement: arrangement ?? null });
// Cancel any pending art/metadata requests
if (artAbortController) artAbortController.abort();
artAbortController = null;
window.highway.stop();
// Cancel any active count-in: clear timers/RAF and bump the gen so
// delayed callbacks (rewind frames, post-seek then, count-in ticks,
// post-count play) bail before mutating the new session.
_cancelCountIn();
// Reset the JUCE shim BEFORE awaiting jucePlayer.stop() so any in-flight
// shim closures see a stale generation after their await and bail out
// before mutating isPlaying / button label / song:* events for the
// outgoing song.
_resetJuceAudioShimChain();
// Cancel queued _audioSeek calls from the previous song: bumping the
// generation makes their chained callbacks bail out.
_resetAudioSeekState();
if (window._juceMode) {
// Mirror the showScreen teardown: emit song:pause for the JUCE
// path so plugins don't see a stale "playing" state on song
// change. (HTML5 fires it via the audio element 'pause' event.)
// Snapshot payload BEFORE stop() resets _pos so audioT/chartT
// capture the actual paused position.
const payload = _songEventPayload();
const wasPlaying = S.isPlaying;
await jucePlayer.stop().catch(() => {});
if (wasPlaying && window.feedBack) {
window.feedBack.isPlaying = false;
window.feedBack.emit('song:pause', payload);
}
window._juceMode = false;
window._juceAudioUrl = null;
}
audio.pause();
audio.src = '';
// Stale until the incoming song's WS handler (window.highway.js) sets it again.
window._currentSongAudio = null;
// Fresh JUCE routing attempt for whatever song loads next.
window._clearJuceRerouteMemo?.();
S.isPlaying = false;
setPlayButtonState(false);
_resetPlaybackSpeedForNewSong();
clearLoop();
_resetSectionPracticeLog();
_hideSectionPracticeBar();
// Reset so the jump-fix (setInterval, ~line 8979) doesn't mistake the new
// song starting at t=0 for an unexpected seek from the previous song's
// position. audio.currentTime may not reset synchronously when src is cleared.
S.lastAudioTime = 0;
currentFilename = filename;
// A fresh load arms autoplay; a pending auto-exit from the previous
// song is no longer relevant. A *resume* load (options.resume) instead
// arms _pendingResume — consumed at song:ready to restore speed + seek to
// the saved position, then start — so autostart and resume don't both try
// to begin playback from different positions.
if (options && options.resume && Number(options.resume.position) > 0) {
S.pendingResume = options.resume;
_pendingAutostart = false;
} else {
S.pendingResume = null;
_pendingAutostart = true;
}
_clearAutoExit();
// Remember which screen the player was launched from so Esc /
// navigation back from the player (and auto-exit) returns the user
// there (feedBack#126).
_playerOriginScreen = _resolvePlayerOrigin();
showScreen('player');
// Wait for previous WebSocket to fully close before opening new one
await new Promise(r => setTimeout(r, 500));
window.highway.init(document.getElementById('highway'));
const wsParams = new URLSearchParams();
if (arrangement !== undefined) wsParams.set('arrangement', arrangement);
wsParams.set('naming_mode', _getArrangementNamingMode());
const wsUrl = `${location.protocol === 'https:' ? 'wss:' : 'ws:'}//${location.host}/ws/highway/${decodeURIComponent(filename)}?${wsParams.toString()}`;
window.highway.connect(wsUrl);
_resetSectionPracticeLog();
_scheduleSectionPracticeRetries();
loadSavedLoops();
document.getElementById('quality-select').value = window.highway.getRenderScale();
const _minScaleSel = document.getElementById('min-scale-select');
if (_minScaleSel && window.highway.getMinRenderScale) _minScaleSel.value = String(window.highway.getMinRenderScale());
}
// Leave the player and return to the screen the song was launched from
// (Esc shortcut uses the same origin-aware target). showScreen() owns the
// full teardown: song:stop, audio unload, window.highway.stop(), count-in cancel.
export function closeCurrentSong() {
// A real close (user Escape/✕, or the queue-aware wrapper once the queue is
// exhausted) abandons any play-queue so a stale one can't advance later.
if (window.feedBack && window.feedBack.playQueue) window.feedBack.playQueue.clear();
return showScreen(_playerOriginScreen || 'home');
}