[feat] receive remote audio with panel playback and bounded WAV capture

This commit is contained in:
dijunkun committed 2026-11-06 21:55:58 +08:00
1 parent 02b10d0d8e
commit 82b8cc11bc
15 files changed
+287 -7

No files matched your search

+22 -2
View File
@@ -1,5 +1,6 @@
import type { App } from "@modelcontextprotocol/ext-apps";
import "./viewer.css";
import { AudioPlayer } from "./audio-player.js";
import { HumanInput } from "./human-input.js";
import { VideoPlayer } from "./video-player.js";
@@ -55,7 +56,7 @@ root.innerHTML = `
<div class="viewer-card">
<div class="viewer-toolbar">
<div class="viewer-title"><h2 id="device-name">远程画面</h2><span id="video-state">待连接</span></div>
<div class="viewer-controls"><button id="take-over" disabled>人工接管</button><button id="pause" disabled><span id="pause-label">暂停控制</span></button><button id="disconnect" class="danger" disabled>断开连接</button><button id="focus-view" aria-label="放大画面" aria-pressed="false" disabled>放大画面</button></div>
<div class="viewer-controls"><button id="audio-play" disabled>播放声音</button><label class="audio-volume">音量<input id="audio-volume" type="range" min="0" max="1" step="0.05" value="1"></label><button id="take-over" disabled>人工接管</button><button id="pause" disabled><span id="pause-label">暂停控制</span></button><button id="disconnect" class="danger" disabled>断开连接</button><button id="focus-view" aria-label="放大画面" aria-pressed="false" disabled>放大画面</button></div>
</div>
<div class="screen" id="screen">
<canvas id="frame" role="img" aria-label="远程桌面实时画面" hidden></canvas>
@@ -107,10 +108,13 @@ let imageAt = 0;
let busy = false,
polling = false,
started = false;
const audioPlayer = new AudioPlayer();
const humanInput = new HumanInput(element("screen"), element<HTMLTextAreaElement>("remote-input"),
(args) => invoke("input", args), showError);
function ownsInput() { return status.control_owner === "human" && status.human_view === context?.view_id; }
const messages: Record<string, string> = {
AUDIO_INVALID: "远端音频格式无效。",
AUDIO_UNAVAILABLE: "尚未收到远端声音,请检查远端音频捕获权限。",
HUMAN_CONTROL_ACTIVE: "远端正在由人工操作,请先交还 Agent。",
STALE_OBSERVATION: "画面已变化,请等待新画面后继续操作。",
STALE_CONTROL: "控制权已交接。",
@@ -386,6 +390,9 @@ function renderStatus(s: Data) {
if (lost && !disconnectedByUser && ["ready", "waiting_frame"].includes(status.state)) showError(new Error("CONNECTION_LOST"));
const changedSession = s.session_id !== status.session_id;
status = s;
element<HTMLButtonElement>("audio-play").disabled = !connected;
text("audio-play", audioPlayer.enabled ? "静音" : "播放声音");
if (!connected || changedSession) audioPlayer.flush();
humanInput.update(connected && ownsInput() ? s.human_input_epoch : undefined);
text("take-over", ownsInput() ? "交还 Agent" : "人工接管");
element<HTMLButtonElement>("take-over").disabled = !connected || !s.control_allowed ||
@@ -503,6 +510,17 @@ input("recovery").onchange = async () => {
renderStatus(await invoke("status"));
} catch (error) { showError(error); renderStatus(status); }
};
element("audio-play").onclick = async () => {
const enabled = !audioPlayer.enabled;
element<HTMLButtonElement>("audio-play").disabled = true;
try {
await audioPlayer.enable(enabled);
audioPlayer.volume(Number(input("audio-volume").value));
await invoke("audio", { enabled });
} catch (error) { await audioPlayer.enable(false); showError(error); }
finally { renderStatus(status); }
};
input("audio-volume").oninput = () => audioPlayer.volume(Number(input("audio-volume").value));
element("take-over").onclick = async () => {
if (controlBusy) return;
controlBusy = true;
@@ -550,6 +568,7 @@ element("disconnect").onclick = async () => {
}
};
async function closePreview() {
audioPlayer.flush();
++videoEpoch;
player?.stop();
player = undefined;
@@ -596,7 +615,7 @@ async function startPreview() {
element<HTMLButtonElement>("focus-view").disabled = false;
text("video-state", "实时");
element("video-state").dataset.state = "live";
}, failed, element<HTMLCanvasElement>("agent-cursor"));
}, failed, element<HTMLCanvasElement>("agent-cursor"), (bytes) => audioPlayer.accept(bytes));
player.start(grant.url, grant.ticket, grant.protocol, grant.decoder);
} catch (error) {
failed(error instanceof Error ? error : new Error("VIDEO_UNAVAILABLE"));
@@ -693,6 +712,7 @@ async function start() {
unlock: "crossdesk_ui_unlock",
recovery: "crossdesk_ui_recovery",
input: "crossdesk_ui_input",
audio: "crossdesk_ui_audio",
};
let connected = false,
initial: Data | undefined;