cinny/src/app/plugins/call/CallControl.ts

import { ClientWidgetApi } from 'matrix-widget-api';
import { EventEmitter } from 'events';
import { CallControlState } from './CallControlState';
import { ElementMediaStateDetail, ElementMediaStatePayload, ElementWidgetActions } from './types';
import { canDelegateCapability } from './utils';

export enum CallControlEvent {
  StateUpdate = 'state_update',
}

/**
 * [lotus #7 / P5-31] Payload for the fork's `io.lotus.set_quality` action.
 * All fields optional; `null` clears that cap. Bits/sec for bitrates, fps for
 * framerate.
 */
export type LotusQualityPayload = {
  audioMaxBitrate?: number | null;
  screenshareMaxBitrate?: number | null;
  screenshareMaxFramerate?: number | null;
};

/** fromWidget `io.lotus.hotkey` (Gitea #43): a watched key in the call frame. */
export type ForkHotkeyReport =
  | {
      type: 'keydown' | 'keyup';
      code: string;
      repeat: boolean;
      ctrlKey: boolean;
      altKey: boolean;
      metaKey: boolean;
      shiftKey: boolean;
      /** Went to a text field in the frame. */
      editable: boolean;
      /** Went to a button/link in the frame. */
      interactive: boolean;
    }
  | { type: 'focus' };

export function parseForkHotkeyReport(data: unknown): ForkHotkeyReport | null {
  if (typeof data !== 'object' || data === null) return null;
  const d = data as Record<string, unknown>;
  if (d.type === 'focus') return { type: 'focus' };
  if ((d.type !== 'keydown' && d.type !== 'keyup') || typeof d.code !== 'string') return null;
  return {
    type: d.type,
    code: d.code,
    repeat: d.repeat === true,
    ctrlKey: d.ctrlKey === true,
    altKey: d.altKey === true,
    metaKey: d.metaKey === true,
    shiftKey: d.shiftKey === true,
    editable: d.editable === true,
    interactive: d.interactive === true,
  };
}

export class CallControl extends EventEmitter implements CallControlState {
  private state: CallControlState;

  private call: ClientWidgetApi;

  private iframe: HTMLIFrameElement;

  // [Gitea #56] Tracks the participant currently pinned via focusCameraParticipant(),
  // so callers (MemberGlance) can render a "Focus camera" / "Unfocus camera" toggle
  // instead of a one-way pin with no way back. null == no manual pin (speaker-follows).
  private _focusedUserId: string | null = null;

  private _focusedMediaId: string | null = null;

  // C-M3: last quality payload requested via setQuality(). Held so we can (re)send
  // it once joined (io.lotus.set_quality must not be sent before call-join — a
  // pre-join send pends to a 10s widget timeout, mirroring the deafen gate).
  private lastQuality: LotusQualityPayload | null = null;

  // C-M5: set true by CallControls while a push-to-talk key is held. A PTT hold
  // unmutes the mic transiently, and onMediaState() must NOT treat that as a
  // user-initiated unmute that auto-undeafens the user.
  public pttActive = false;

  // Deafen mutes the mic; undeafen should give it back ONLY if it was on
  // before (Discord semantics — verified against the real client, Gitea #173:
  // undeafen used to leave you muted). Cleared by any manual mic toggle.
  private micOnBeforeDeafen = false;

  // P6-2: mirrors CallEmbed.joined. Set true from forceState(), which CallEmbed
  // invokes only from onCallJoined(). Gates io.lotus.set_deafen so we never send
  // before the fork's widget handler mounts (pre-join sends pend to a 10s
  // timeout — io.lotus toWidget actions must only be sent after call-join).
  private joined = false;

  constructor(state: CallControlState, call: ClientWidgetApi, iframe: HTMLIFrameElement) {
    super();
    // [Gitea #316] More than 10 call features subscribe to state_update at once
    // (call bar, controls, hotkeys, AFK, muted-talk, join/leave sounds,
    // announcements, rejoin, policy toast, glance, desktop thumbbar/SMTC …).
    // Hitting the default limit logged a MaxListenersExceededWarning, and the
    // console then kept this CallControl — and through it the whole Element Call
    // document — alive: ~11 MB leaked per call. These are not leaks, so lift it.
    this.setMaxListeners(100);

    this.state = state;
    this.call = call;
    this.iframe = iframe;
  }

  public getState(): CallControlState {
    return this.state;
  }

  public get microphone(): boolean {
    return this.state.microphone;
  }

  public get video(): boolean {
    return this.state.video;
  }

  public get sound(): boolean {
    return this.state.sound;
  }

  public get screenshare(): boolean {
    return this.state.screenshare;
  }

  public get spotlight(): boolean {
    return this.state.spotlight;
  }

  public get screenshareAudioMuted(): boolean {
    return this.state.screenshareAudioMuted;
  }

  public async applyState() {
    await this.setMediaState({
      audio_enabled: this.microphone,
      video_enabled: this.video,
    });
    this.setSound();
    this.emitStateUpdate();
  }

  public async forceState(desired: CallControlState) {
    this.state = new CallControlState(
      desired.microphone,
      desired.video,
      desired.sound,
      this.screenshare,
      this.spotlight,
      this.screenshareAudioMuted,
    );
    await this.applyState();
    // P6-2: CallEmbed calls forceState() only from onCallJoined(), so this is
    // the join transition. Flip the gate open, then push the current deafen
    // state to the fork's freshly-mounted handler. (setSound() above ran while
    // this.joined was still false, so it was gated — this is the first send.)
    this.joined = true;
    this.sendDeafenState();
    this.sendQuality();
    this.sendHotkeyCodes();
    this.sendFrameScreenshareAllowed();
  }

  /**
   * C-H1 / C-M3: re-push the sticky fork-side state (deafen + quality) after an
   * EC reconnect. Unlike forceState() this does NOT touch mic/video, so a
   * reconnect can't clobber the user's live media state — it only re-arms the
   * fork handlers that remount on reconnect.
   */
  public resendForkState(): void {
    this.sendDeafenState();
    this.sendQuality();
    if (this._audioOutputId !== undefined) this.sendAudioOutput(this._audioOutputId);
    // [Gitea #17] The pin lives fork-side and is dropped on a handler remount.
    if (this._focusedUserId !== null) this.sendFocus(this._focusedUserId, this._focusedMediaId);
    this.sendHotkeyCodes();
    this.sendFrameScreenshareAllowed();
  }

  private async setMediaState(state: ElementMediaStatePayload) {
    // transport.send resolves once EC has ACK'd the command, which is enough to
    // consider the mute applied. We deliberately do NOT gate completion on a
    // follow-up DeviceMute state-echo: EC may elide it (e.g. when the requested
    // state already matches its current state) or skip it during teardown,
    // which would strand this promise forever and block applyState(). The echo,
    // when it does arrive, is still handled authoritatively by onMediaState().
    return this.call.transport.send(ElementWidgetActions.DeviceMute, state);
  }

  // [Gitea #209] Deafen is applied by the fork (io.lotus.set_deafen). The
  // iframe-DOM `.muted` fallback that used to live here (plus the per-member
  // re-apply in useCallMemberSoundSync) fought EC's audio renderer and is gone
  // now that the pin is ≥ 0.25.0-lotus.2 — late joiners are the fork's job.
  private setSound(): void {
    this.sendDeafenState();
  }

  private applyScreenshareAudioMuted(): void {
    if (!this.sound) return;
    this.sendDeafenState();
  }

  // P6-2: send deafen state to the fork (io.lotus.set_deafen). Join-gated: the
  // fork's handler only exists once joined; onCallJoined() re-sends the current
  // state so a pre-join deafen is not lost.
  // [Gitea #119] Output device (headset ↔ speakers) chosen from the host's
  // call bar; the fork applies it with mediaDevices.audioOutput.select().
  private _audioOutputId: string | undefined;

  public get audioOutputId(): string | undefined {
    return this._audioOutputId;
  }

  public setAudioOutput(deviceId: string): void {
    this._audioOutputId = deviceId;
    this.sendAudioOutput(deviceId);
  }

  private sendAudioOutput(deviceId: string): void {
    if (!this.joined) return;
    this.call.transport.send('io.lotus.set_audio_output', { deviceId }).catch(() => undefined);
  }

  private sendDeafenState(): void {
    if (!this.joined) return;
    this.call.transport
      .send('io.lotus.set_deafen', {
        deafened: !this.sound,
        screenshareAudioMuted: this.screenshareAudioMuted,
      })
      .catch(() => undefined);
  }

  public onMediaState(evt: CustomEvent<ElementMediaStateDetail>) {
    const { data } = evt.detail;
    if (!data) return;

    const state = new CallControlState(
      data.audio_enabled ?? this.microphone,
      data.video_enabled ?? this.video,
      this.sound,
      this.screenshare,
      this.spotlight,
      this.screenshareAudioMuted,
    );

    this.state = state;
    this.emitStateUpdate();

    // C-M5: auto-undeafen when the mic turns on, but NOT for a transient
    // push-to-talk unmute — a PTT tap while deafened must not silently
    // un-deafen the user.
    if (this.microphone && !this.sound && !this.pttActive) {
      this.toggleSound();
    }
  }

  // [Gitea #43] The fork has reported io.lotus.controls_state, so its Lotus
  // action handlers are mounted. Screenshare and layout state come from that
  // report; the host never reads or clicks EC's DOM.
  private forkReady = false;

  // [Gitea #43] The fork shows EC's own screenshare button inside the frame
  // (lotusFrameScreenshare: no Capability Delegation here), so the host bar
  // hides its own.
  private _frameScreenshare = false;

  private frameScreenshareListeners = new Set<() => void>();

  // [Gitea #43] The fork draws the "Share your screen?" prompt inside the frame
  // on request (io.lotus.prompt_screenshare), so where the click can't be
  // delegated our call-bar button asks for that instead of starting the share.
  private _screensharePrompt = false;

  // [Gitea #43] The fork's in-frame prompt is showing: anything the host lays
  // over the frame (the picture-in-picture overlay) must let clicks through.
  private _screensharePromptOpen = false;

  // [Gitea #43] Whether the room's call policy allows screensharing; the fork
  // hides its in-frame button when it doesn't.
  private frameScreenshareAllowed = true;

  private hotkeyCodes = new Map<string, string[]>();

  private hotkeyListeners = new Set<(report: ForkHotkeyReport) => void>();

  /** EC's own screenshare button is shown in the frame instead of ours. */
  public get frameScreenshare(): boolean {
    return this._frameScreenshare;
  }

  /**
   * Starting a share must go through the fork's in-frame prompt: this engine
   * can't delegate the click (Firefox, Safari, WebKitGTK) and the fork has
   * the prompt. Stopping never needs it.
   */
  public get screenshareNeedsPrompt(): boolean {
    return this._screensharePrompt && !canDelegateCapability();
  }

  /** The fork's in-frame "Share your screen?" prompt is showing. */
  public get screensharePromptOpen(): boolean {
    return this._screensharePromptOpen;
  }

  /**
   * Ask the fork to show "Share your screen?" inside the call frame; its Share
   * button starts the share with the frame's own click.
   */
  public promptScreenshare(): void {
    this.sendForkAction('io.lotus.prompt_screenshare', {});
  }

  /**
   * Subscribe to `frameScreenshare` / `screenshareNeedsPrompt` /
   * `screensharePromptOpen` changes. Returns an unsubscribe.
   */
  public onFrameScreenshareChange(cb: () => void): () => void {
    this.frameScreenshareListeners.add(cb);
    return () => {
      this.frameScreenshareListeners.delete(cb);
    };
  }

  /** Room call policy for the in-frame screenshare button. */
  public setFrameScreenshareAllowed(allowed: boolean): void {
    this.frameScreenshareAllowed = allowed;
    this.sendFrameScreenshareAllowed();
  }

  private sendFrameScreenshareAllowed(): void {
    if (!this.joined || !this.forkReady) return;
    if (!this._frameScreenshare && !this._screensharePrompt) return;
    this.sendForkAction('io.lotus.set_frame_screenshare', {
      visible: this.frameScreenshareAllowed,
    });
  }

  /** Key codes `source` (e.g. 'ptt', 'deafen') wants reported from the frame. */
  public setHotkeyCodes(source: string, codes: string[]): void {
    this.hotkeyCodes.set(source, codes);
    this.sendHotkeyCodes();
  }

  private sendHotkeyCodes(): void {
    if (!this.joined || !this.forkReady) return;
    const codes = [...new Set([...this.hotkeyCodes.values()].flat())];
    this.call.transport.send('io.lotus.set_hotkeys', { codes }).catch(() => undefined);
  }

  /** Subscribe to the fork's `io.lotus.hotkey` reports. Returns an unsubscribe. */
  public onHotkey(cb: (report: ForkHotkeyReport) => void): () => void {
    this.hotkeyListeners.add(cb);
    return () => {
      this.hotkeyListeners.delete(cb);
    };
  }

  /** [Gitea #43] The fork's `io.lotus.hotkey` report. */
  public onHotkeyReport(data: unknown): void {
    const report = parseForkHotkeyReport(data);
    if (report) this.hotkeyListeners.forEach((l) => l(report));
  }

  /** [Gitea #43] The fork's `io.lotus.controls_state` report. */
  public onControlsState(data: unknown) {
    if (typeof data !== 'object' || data === null) return;
    const { screensharing, layout, frameScreenshare, screensharePrompt, screensharePromptOpen } =
      data as {
        screensharing?: unknown;
        layout?: unknown;
        frameScreenshare?: unknown;
        screensharePrompt?: unknown;
        screensharePromptOpen?: unknown;
      };
    const firstReport = !this.forkReady;
    this.forkReady = true;
    const frame = frameScreenshare === true;
    const prompt = screensharePrompt === true;
    const promptOpen = screensharePromptOpen === true;
    if (promptOpen !== this._screensharePromptOpen) {
      this._screensharePromptOpen = promptOpen;
      if (frame === this._frameScreenshare && prompt === this._screensharePrompt) {
        this.frameScreenshareListeners.forEach((l) => l());
      }
    }
    if (frame !== this._frameScreenshare || prompt !== this._screensharePrompt) {
      const promptTurnedOn = prompt && !this._screensharePrompt;
      this._frameScreenshare = frame;
      this._screensharePrompt = prompt;
      this.frameScreenshareListeners.forEach((l) => l());
      if (promptTurnedOn && !firstReport) this.sendFrameScreenshareAllowed();
    }
    if (firstReport) {
      this.sendHotkeyCodes();
      this.sendFrameScreenshareAllowed();
    }
    this.applyControls(
      typeof screensharing === 'boolean' ? screensharing : this.screenshare,
      layout === 'spotlight' || layout === 'grid' ? layout === 'spotlight' : this.spotlight,
    );
  }

  private applyControls(screenshare: boolean, spotlight: boolean) {
    const wasScreensharing = this.screenshare;

    // C-M6: when a screenshare stops, clear the screenshare-audio mute so a
    // later screenshare doesn't start pre-muted.
    const screenshareAudioMuted =
      wasScreensharing && !screenshare ? false : this.screenshareAudioMuted;

    // C-H3: the body observer now watches subtree:true, so this fires on any DOM
    // churn in EC's controls. Only re-emit (→ re-render every consumer) when one
    // of the values this method derives actually changed — microphone/video/sound
    // are copied unchanged from the current state here.
    if (
      this.state.screenshare === screenshare &&
      this.state.spotlight === spotlight &&
      this.state.screenshareAudioMuted === screenshareAudioMuted
    ) {
      return;
    }

    this.state = new CallControlState(
      this.microphone,
      this.video,
      this.sound,
      screenshare,
      spotlight,
      screenshareAudioMuted,
    );
    this.emitStateUpdate();
  }

  public setMicrophone(enabled: boolean) {
    const payload: ElementMediaStatePayload = {
      audio_enabled: enabled,
      video_enabled: this.video,
    };
    return this.setMediaState(payload);
  }

  public toggleMicrophone() {
    const payload: ElementMediaStatePayload = {
      audio_enabled: !this.microphone,
      video_enabled: this.video,
    };
    return this.setMediaState(payload);
  }

  public toggleVideo() {
    const payload: ElementMediaStatePayload = {
      audio_enabled: this.microphone,
      video_enabled: !this.video,
    };
    return this.setMediaState(payload);
  }

  public toggleSound() {
    const sound = !this.sound;

    // P6-2: commit state before setSound()/applyScreenshareAudioMuted() so
    // sendDeafenState() (which reads this.sound) reports the new value.
    const state = new CallControlState(
      this.microphone,
      this.video,
      sound,
      this.screenshare,
      this.spotlight,
      this.screenshareAudioMuted,
    );
    this.state = state;

    this.setSound();
    // After un-deafening, re-apply screenshare audio mute if active
    if (sound) this.applyScreenshareAudioMuted();

    this.emitStateUpdate();

    if (!sound) {
      this.micOnBeforeDeafen = this.microphone;
      if (this.microphone) this.toggleMicrophone();
    } else if (this.micOnBeforeDeafen && !this.microphone) {
      this.micOnBeforeDeafen = false;
      this.toggleMicrophone();
    }
  }

  public toggleScreenshareAudio() {
    const screenshareAudioMuted = !this.screenshareAudioMuted;
    this.state = new CallControlState(
      this.microphone,
      this.video,
      this.sound,
      this.screenshare,
      this.spotlight,
      screenshareAudioMuted,
    );
    this.emitStateUpdate();
    this.applyScreenshareAudioMuted();
  }

  /**
   * Must be called synchronously inside the user's click: starting a share
   * calls getDisplayMedia in the frame, which needs that click. Where the
   * browser can hand the click over (Capability Delegation, Chromium incl.
   * WebView2) it is sent with delegation. Elsewhere the host bar hides its
   * button and EC's own shows in the frame (`frameScreenshare`); a plain send
   * still stops a share, which needs no click.
   */
  public toggleScreenshare() {
    const data = { on: !this.screenshare };
    if (canDelegateCapability()) {
      this.sendDelegated('io.lotus.set_screenshare', data, 'display-capture');
      return;
    }
    this.sendForkAction('io.lotus.set_screenshare', data);
  }

  /**
   * Send a widget action with `postMessage(…, { delegate })` so the frame
   * receives the user's activation for `capability`. matrix-widget-api has no
   * postMessage options, so swap its `sendInternal` for this one synchronous
   * send (the request is posted before `send()` returns). Delegation refuses a
   * `*` target origin, so the frame's real origin is used.
   */
  private sendDelegated(action: string, data: Record<string, unknown>, capability: string): void {
    const target = this.iframe.contentWindow;
    if (!target) return;
    const targetOrigin = new URL(this.iframe.src, window.location.href).origin;
    const transport = this.call.transport as unknown as {
      sendInternal: (message: unknown) => void;
    };
    const hadOwn = Object.prototype.hasOwnProperty.call(transport, 'sendInternal');
    const original = transport.sendInternal;
    transport.sendInternal = (message: unknown) =>
      target.postMessage(message, {
        targetOrigin,
        delegate: capability,
      } as WindowPostMessageOptions);
    try {
      this.call.transport.send(action, data).catch(() => undefined);
    } finally {
      if (hadOwn) transport.sendInternal = original;
      else delete (transport as { sendInternal?: unknown }).sendInternal;
    }
  }

  public toggleSpotlight() {
    this.sendForkAction('io.lotus.set_layout', {
      layout: this.spotlight ? 'grid' : 'spotlight',
    });
  }

  public toggleReactions() {
    this.sendForkAction('io.lotus.toggle_reactions', {});
  }

  public toggleSettings() {
    // Same as clicking EC's settings button: opens the modal.
    this.sendForkAction('io.lotus.open_settings', { open: true });
  }

  private sendForkAction(action: string, data: Record<string, unknown>): void {
    this.call.transport.send(action, data).catch(() => undefined);
  }

  /**
   * Focus a specific participant's camera tile in Element Call.
   *
   * EC renders video tiles as `[data-testid="videoTile"]`. Each tile wraps a
   * mute-status indicator with `aria-label` set to the participant's Matrix
   * user ID. We find the tile containing that user, switch to spotlight mode
   * if needed, then click the tile so EC's internal focus handler runs.
   *
   * Falls back to a plain spotlight toggle if the tile is not found (e.g. the
   * participant has their camera off and EC didn't render a video tile for
   * them yet).
   */
  public get focusedUserId(): string | null {
    return this._focusedUserId;
  }

  private sendFocus(userId: string, id: string | null): void {
    // [EC#30] `id` is the fork's per-device media id (userId:deviceId) taken
    // from io.lotus.call_state, so a multi-device user pins the right device;
    // the fork falls back to userId (preferring the speaking device) when absent.
    this.call.transport
      .send('io.lotus.focus_participant', id ? { userId, id } : { userId })
      .catch(() => undefined);
  }

  public focusCameraParticipant(userId: string, id: string | null = null): void {
    // [lotus #4] Pin the participant via the fork's widget action instead of
    // DOM-poking tiles. EC's layout honors it — including surfacing the camera
    // alongside a screenshare (A5) — and it's version-stable. The fork always
    // acks, so the promise resolves regardless.
    this._focusedUserId = userId;
    this._focusedMediaId = id;
    this.sendFocus(userId, id);
    // [Gitea #56] Notify state-update listeners so the menu can flip to "Unfocus camera".
    this.emitStateUpdate();
  }

  /** [lotus #4] Clear any manual spotlight pin and return to speaker-follows. */
  public clearFocusParticipant(): void {
    // [Gitea #56] No-op (and no redundant widget send) if nothing is pinned —
    // dispose() calls this unconditionally on every call teardown.
    if (this._focusedUserId === null) return;
    this._focusedUserId = null;
    this._focusedMediaId = null;
    this.call.transport.send('io.lotus.focus_participant', { userId: null }).catch(() => undefined);
    this.emitStateUpdate();
  }

  /**
   * [lotus #3 / P5-15] Inject a soundboard clip into the call so other
   * participants hear it. The fork publishes it as a separate LiveKit audio
   * track (`io.lotus.inject_audio`) rather than splicing the mic. `url` must be
   * an https/blob URL the widget can fetch WITHOUT credentials — the host
   * resolves an mxc clip to a `blob:` object URL first (authenticated media
   * can't be fetched cross-realm by the widget) — and `audio` the same clip's
   * bytes, which the fork prefers. `volume` is 0–1.
   *
   * The local user does not hear their own published track, so callers should
   * also play the clip locally for feedback.
   */
  public injectAudio(
    url: string,
    volume = 1,
    audio?: ArrayBuffer,
  ): Promise<{ played: boolean; reason?: string }> {
    // [EC#13] The fork now refuses while the local mic is muted and replies
    // { played:false, reason:"muted" }; older forks reply {} (treated as played).
    // [Gitea #43] `audio` carries the clip's bytes: a `blob:` URL only works on
    // this origin, so a call page on its own origin can't fetch it. Forks that
    // predate `audio` ignore it and use `url`.
    return this.call.transport
      .send<
        { url: string; volume: number; audio?: ArrayBuffer },
        { played?: boolean; reason?: string }
      >('io.lotus.inject_audio', audio ? { url, volume, audio } : { url, volume })
      .then((r) => ({ played: r?.played !== false, reason: r?.reason }))
      .catch(() => ({ played: true }));
  }

  /**
   * [lotus #7 / P5-31] Apply audio/screenshare encoding limits to the local
   * published tracks (the fork's `io.lotus.set_quality` action, via
   * `RTCRtpSender.setParameters` — no republish). Bitrates are bits/sec,
   * framerate is fps. A field set to `null` clears that cap. Settings are
   * sticky fork-side (re-applied on every re-publish / reconnect). Values are
   * clamped fork-side, so out-of-range input can't brick the encoder.
   */
  public setQuality(settings: LotusQualityPayload): void {
    // C-M3: remember the request and only send once joined; sendQuality() gates
    // on this.joined so a pre-join call is a no-op that we replay on join.
    this.lastQuality = settings;
    this.sendQuality();
  }

  // C-M3: push the last-requested quality to the fork. Gated on this.joined so
  // we never send io.lotus.set_quality before the fork's handler mounts (a
  // pre-join send would pend to a 10s widget timeout).
  private sendQuality(): void {
    if (!this.joined || !this.lastQuality) return;
    this.call.transport.send('io.lotus.set_quality', this.lastQuality).catch(() => undefined);
  }

  public dispose() {
    // [Gitea #56] Don't let a manual focus pin outlive the call.
    this.clearFocusParticipant();
  }

  private emitStateUpdate() {
    this.emit(CallControlEvent.StateUpdate);
  }
}