@flowingspring/dsh-voco 0.3.7 → 0.3.8

package/README.zh.md

@@ -61,7 +61,7 @@ SILICONFLOW_API_KEY=sk-your-api-key
 
 1. 运行 `dsh web` 并打开 DSH Web UI。
 2. 新建或选择会话，点击麦克风并允许浏览器权限。
-3. 直接说话，连续静音约 1.5 秒后提交一句话。
+3. 直接说话，连续静音约 1.5 秒后提交一句话；也可以长按麦克风约 400 毫秒按住说话，松手后提交这一段。
 4. 普通聊天由前台回答，需要项目工具的工作会委派给后台 Agent。
 
 每个语音会话持续绑定自己的后台 Agent Session。上下文达到轮换阈值后，后续任务会自动进入新的子会话，但仍与原语音会话绑定。

package/lib/client.js

@@ -4,13 +4,13 @@ window.__ModuleLoader__.load({
 		var module = { exports: {} };
 		var exports = module.exports;
 		Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-		let react_jsx_runtime = require("react/jsx-runtime");
 		let react = require("react");
+		let react_jsx_runtime = require("react/jsx-runtime");
 		let react_dom_client = require("react-dom/client");
 		let _deepseek_ai_dsh_client_ui_primitives = require("@deepseek-ai/dsh-client-ui-primitives");
 		let _deepseek_ai_dsh_client_runtime_client = require("@deepseek-ai/dsh-client-runtime/client");
 		//#region \0dsh-css:D:\QUAN\DevCode\Python\Project\dsh\dsh-voco\packages\ui-voice\src\client\VoiceControl.module.css.mjs
-		const css$4 = ".TFZsua_controls{align-items:center;gap:2px;display:inline-flex}.TFZsua_button{width:28px;height:28px;color:var(--dsw-alias-label-secondary);cursor:pointer;background:0 0;border:0;border-radius:8px;justify-content:center;align-items:center;padding:0;display:inline-flex}.TFZsua_button:hover{background:var(--dsw-alias-interactive-bg-hover);color:var(--dsw-alias-label-primary)}.TFZsua_active{color:var(--dsw-alias-state-business-primary);background:var(--dsw-alias-interactive-bg-hover-solid)}.TFZsua_button[data-muted=true]{color:var(--dsw-alias-state-error-primary);background:color-mix(in srgb, currentColor 12%, transparent)}.TFZsua_interruptButton,.TFZsua_interruptButton:hover{color:var(--dsw-alias-state-error-primary)}.TFZsua_button[data-muted=true]:hover{background:color-mix(in srgb, currentColor 18%, transparent)}.TFZsua_button[data-state=speaking][data-muted=false]{animation:1.2s ease-in-out infinite TFZsua_voice-pulse}@keyframes TFZsua_voice-pulse{50%{opacity:.55;transform:scale(.94)}}@media (prefers-reduced-motion:reduce){.TFZsua_button[data-state=speaking][data-muted=false]{animation:none}}";
+		const css$4 = ".TFZsua_controls{align-items:center;gap:2px;display:inline-flex}.TFZsua_button{touch-action:manipulation;width:28px;height:28px;color:var(--dsw-alias-label-secondary);cursor:pointer;background:0 0;border:0;border-radius:8px;justify-content:center;align-items:center;padding:0;display:inline-flex}.TFZsua_button:hover{background:var(--dsw-alias-interactive-bg-hover);color:var(--dsw-alias-label-primary)}.TFZsua_active{color:var(--dsw-alias-state-business-primary);background:var(--dsw-alias-interactive-bg-hover-solid)}.TFZsua_button[data-muted=true]{color:var(--dsw-alias-state-error-primary);background:color-mix(in srgb, currentColor 12%, transparent)}.TFZsua_button[data-push-to-talk=true]{color:var(--dsw-alias-state-business-primary);background:var(--dsw-alias-interactive-bg-hover-solid);transform:scale(.94)}.TFZsua_interruptButton,.TFZsua_interruptButton:hover{color:var(--dsw-alias-state-error-primary)}.TFZsua_button[data-muted=true]:hover{background:color-mix(in srgb, currentColor 18%, transparent)}.TFZsua_button[data-state=speaking][data-muted=false]{animation:1.2s ease-in-out infinite TFZsua_voice-pulse}@keyframes TFZsua_voice-pulse{50%{opacity:.55;transform:scale(.94)}}@media (prefers-reduced-motion:reduce){.TFZsua_button[data-state=speaking][data-muted=false]{animation:none}}";
 		const tagId$4 = "@flowingspring/dsh-voco/VoiceControl.module.css";
 		if (typeof document !== "undefined" && document.querySelector("style[data-plugin-css=" + JSON.stringify(tagId$4) + "]") === null) {
 			const tag = document.createElement("style");
@@ -20,8 +20,8 @@ window.__ModuleLoader__.load({
 			document.head.appendChild(tag);
 		}
 		var VoiceControl_module_css_default = {
-			"button": "TFZsua_button",
 			"controls": "TFZsua_controls",
+			"button": "TFZsua_button",
 			"active": "TFZsua_active",
 			"interruptButton": "TFZsua_interruptButton",
 			"voice-pulse": "TFZsua_voice-pulse"
@@ -29,10 +29,38 @@ window.__ModuleLoader__.load({
 		//#endregion
 		//#region ../ui-voice/src/client/VoiceControl.tsx
 		/** Start a fresh Voice Session from this Session's location or control the active connection. */
-		function VoiceControl({ sessionId, useVoice, startVoice, retryVoice, setVoiceMuted, t, interruptResponse }) {
+		function VoiceControl({ sessionId, useVoice, startVoice, retryVoice, beginPushToTalk, endPushToTalk, setVoiceMuted, t, interruptResponse }) {
 			const state = useVoice((snapshot) => snapshot.state);
 			const activeSessionId = useVoice((snapshot) => snapshot.sessionId);
 			const inputMuted = useVoice((snapshot) => snapshot.inputMuted);
+			const pushToTalkActive = useVoice((snapshot) => snapshot.pushToTalkActive);
+			const timer = (0, react.useRef)(void 0);
+			const pointerDown = (0, react.useRef)(false);
+			const longPress = (0, react.useRef)(false);
+			const suppressClick = (0, react.useRef)(false);
+			const clearTimer = () => {
+				if (timer.current !== void 0) {
+					window.clearTimeout(timer.current);
+					timer.current = void 0;
+				}
+			};
+			const releasePushToTalk = () => {
+				clearTimer();
+				pointerDown.current = false;
+				if (longPress.current) {
+					longPress.current = false;
+					endPushToTalk();
+				}
+			};
+			(0, react.useEffect)(() => {
+				window.addEventListener("blur", releasePushToTalk);
+				document.addEventListener("visibilitychange", releasePushToTalk);
+				return () => {
+					window.removeEventListener("blur", releasePushToTalk);
+					document.removeEventListener("visibilitychange", releasePushToTalk);
+					releasePushToTalk();
+				};
+			}, []);
 			const toggle = async () => {
 				if (state === "error" && activeSessionId !== void 0) {
 					try {
@@ -55,6 +83,35 @@ window.__ModuleLoader__.load({
 					interruptResponse();
 				} catch {}
 			};
+			const onPointerDown = (event) => {
+				if (event.pointerType === "mouse" && event.button !== 0) return;
+				pointerDown.current = true;
+				longPress.current = false;
+				suppressClick.current = false;
+				if (typeof event.currentTarget.setPointerCapture === "function") event.currentTarget.setPointerCapture(event.pointerId);
+				clearTimer();
+				timer.current = window.setTimeout(() => {
+					timer.current = void 0;
+					if (!pointerDown.current) return;
+					longPress.current = true;
+					suppressClick.current = true;
+					beginPushToTalk(sessionId);
+				}, 400);
+			};
+			const onPointerUp = (event) => {
+				releasePushToTalk();
+				if (event.currentTarget.hasPointerCapture?.(event.pointerId)) event.currentTarget.releasePointerCapture(event.pointerId);
+			};
+			const onPointerCancel = () => {
+				releasePushToTalk();
+			};
+			const onClick = () => {
+				if (suppressClick.current) {
+					suppressClick.current = false;
+					return;
+				}
+				toggle();
+			};
 			return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("span", {
 				className: VoiceControl_module_css_default.controls,
 				children: [state === "speaking" && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("button", {
@@ -88,14 +145,17 @@ window.__ModuleLoader__.load({
 				}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)("button", {
 					type: "button",
 					className: `${VoiceControl_module_css_default.button} ${active ? VoiceControl_module_css_default.active : ""}`,
-					"aria-label": label,
+					"aria-label": pushToTalkActive ? t("control.pushToTalk") : label,
 					"aria-pressed": active ? inputMuted : void 0,
 					title: label,
 					"data-state": state,
 					"data-muted": inputMuted,
-					onClick: () => {
-						toggle();
-					},
+					"data-push-to-talk": pushToTalkActive,
+					onPointerDown,
+					onPointerUp,
+					onPointerCancel,
+					onLostPointerCapture: onPointerCancel,
+					onClick,
 					children: /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("svg", {
 						viewBox: "0 0 16 16",
 						width: "16",
@@ -163,23 +223,23 @@ window.__ModuleLoader__.load({
 			document.head.appendChild(tag);
 		}
 		var VoiceNodeViews_module_css_default = {
-			"taskDot": "Kt2smW_taskDot",
-			"taskCard": "Kt2smW_taskCard",
-			"meta": "Kt2smW_meta",
+			"taskSummary": "Kt2smW_taskSummary",
 			"taskTitle": "Kt2smW_taskTitle",
-			"utterance": "Kt2smW_utterance",
 			"bubble": "Kt2smW_bubble",
-			"taskDetails": "Kt2smW_taskDetails",
-			"taskSummary": "Kt2smW_taskSummary",
-			"miniWave": "Kt2smW_miniWave",
+			"meta": "Kt2smW_meta",
 			"taskCancel": "Kt2smW_taskCancel",
-			"taskUpdate": "Kt2smW_taskUpdate",
-			"voiceBadge": "Kt2smW_voiceBadge",
 			"taskStatus": "Kt2smW_taskStatus",
-			"taskLink": "Kt2smW_taskLink",
+			"taskUpdate": "Kt2smW_taskUpdate",
+			"utterance": "Kt2smW_utterance",
 			"taskInput": "Kt2smW_taskInput",
+			"taskLink": "Kt2smW_taskLink",
 			"taskChevron": "Kt2smW_taskChevron",
-			"taskActions": "Kt2smW_taskActions"
+			"miniWave": "Kt2smW_miniWave",
+			"taskCard": "Kt2smW_taskCard",
+			"taskDot": "Kt2smW_taskDot",
+			"taskActions": "Kt2smW_taskActions",
+			"taskDetails": "Kt2smW_taskDetails",
+			"voiceBadge": "Kt2smW_voiceBadge"
 		};
 		//#endregion
 		//#region ../ui-voice/src/client/VoiceNodeViews.tsx
@@ -318,10 +378,10 @@ window.__ModuleLoader__.load({
 		}
 		var VoiceOverlay_module_css_default = {
 			"root": "GPPB2G_root",
-			"voice-wave": "GPPB2G_voice-wave",
 			"wave": "GPPB2G_wave",
 			"status": "GPPB2G_status",
-			"stop": "GPPB2G_stop"
+			"stop": "GPPB2G_stop",
+			"voice-wave": "GPPB2G_voice-wave"
 		};
 		//#endregion
 		//#region ../ui-voice/src/client/VoiceOverlay.tsx
@@ -380,19 +440,19 @@ window.__ModuleLoader__.load({
 			document.head.appendChild(tag);
 		}
 		var VoiceSessionMarkers_module_css_default = {
-			"child": "_0brRca_child",
 			"toggle": "_0brRca_toggle",
-			"childDot": "_0brRca_childDot",
-			"childLabel": "_0brRca_childLabel",
-			"marker": "_0brRca_marker",
-			"childTime": "_0brRca_childTime",
-			"deleteStatus": "_0brRca_deleteStatus",
 			"children": "_0brRca_children",
-			"childMain": "_0brRca_childMain",
+			"marker": "_0brRca_marker",
+			"child": "_0brRca_child",
+			"childLabel": "_0brRca_childLabel",
 			"childActions": "_0brRca_childActions",
 			"childMenuButton": "_0brRca_childMenuButton",
+			"deleteStatus": "_0brRca_deleteStatus",
 			"deleteAction": "_0brRca_deleteAction",
+			"childDot": "_0brRca_childDot",
+			"childTime": "_0brRca_childTime",
 			"deleteError": "_0brRca_deleteError",
+			"childMain": "_0brRca_childMain",
 			"slot": "_0brRca_slot"
 		};
 		//#endregion
@@ -868,29 +928,29 @@ window.__ModuleLoader__.load({
 			document.head.appendChild(tag);
 		}
 		var VocoSettingsCard_module_css_default = {
+			"save": "G5LlXq_save",
+			"chevron": "G5LlXq_chevron",
+			"hint": "G5LlXq_hint",
+			"header": "G5LlXq_header",
+			"body": "G5LlXq_body",
 			"pending": "G5LlXq_pending",
-			"badgeMuted": "G5LlXq_badgeMuted",
-			"input": "G5LlXq_input",
-			"footer": "G5LlXq_footer",
-			"name": "G5LlXq_name",
-			"chevronOpen": "G5LlXq_chevronOpen",
 			"headText": "G5LlXq_headText",
-			"cardOpen": "G5LlXq_cardOpen",
-			"hint": "G5LlXq_hint",
-			"discard": "G5LlXq_discard",
+			"chevronOpen": "G5LlXq_chevronOpen",
+			"description": "G5LlXq_description",
+			"readOnly": "G5LlXq_readOnly",
+			"label": "G5LlXq_label",
 			"providerDetails": "G5LlXq_providerDetails",
-			"save": "G5LlXq_save",
 			"failed": "G5LlXq_failed",
+			"badgeMuted": "G5LlXq_badgeMuted",
+			"fieldHead": "G5LlXq_fieldHead",
+			"providerDetail": "G5LlXq_providerDetail",
+			"cardOpen": "G5LlXq_cardOpen",
+			"footer": "G5LlXq_footer",
 			"field": "G5LlXq_field",
-			"description": "G5LlXq_description",
+			"input": "G5LlXq_input",
 			"badge": "G5LlXq_badge",
-			"label": "G5LlXq_label",
-			"providerDetail": "G5LlXq_providerDetail",
-			"header": "G5LlXq_header",
-			"body": "G5LlXq_body",
-			"fieldHead": "G5LlXq_fieldHead",
-			"readOnly": "G5LlXq_readOnly",
-			"chevron": "G5LlXq_chevron",
+			"name": "G5LlXq_name",
+			"discard": "G5LlXq_discard",
 			"card": "G5LlXq_card"
 		};
 		//#endregion
@@ -1012,6 +1072,7 @@ window.__ModuleLoader__.load({
 		const INITIAL_SNAPSHOT = {
 			state: "off",
 			inputMuted: false,
+			pushToTalkActive: false,
 			textById: {}
 		};
 		/** Own the single browser microphone, provider WebSocket, playback queue, and live transcript feed. */
@@ -1026,6 +1087,9 @@ window.__ModuleLoader__.load({
 			pendingText = [];
 			onConversationStarted;
 			conversationStarted = false;
+			pushToTalkOrigin;
+			pushToTalkPending = false;
+			pushToTalkActive = false;
 			constructor(options = {}) {
 				this.onConversationStarted = options.onConversationStarted;
 			}
@@ -1045,6 +1109,101 @@ window.__ModuleLoader__.load({
 			async start(sessionId) {
 				await this.open(sessionId, false);
 			}
+			/** Begin a temporary manually committed utterance after a long press. */
+			beginPushToTalk(sessionId) {
+				if (this.pushToTalkPending || this.pushToTalkActive) return;
+				const state = this.snapshot.state;
+				if (state === "off") {
+					this.pushToTalkOrigin = "off";
+					this.pushToTalkPending = true;
+					this.open(sessionId, true).catch(() => {});
+					return;
+				}
+				if (state === "error") {
+					const activeSessionId = this.snapshot.sessionId;
+					if (activeSessionId === void 0) return;
+					this.pushToTalkOrigin = this.snapshot.inputMuted ? "muted" : "hands-free";
+					this.pushToTalkPending = true;
+					this.open(activeSessionId, true).catch(() => {});
+					return;
+				}
+				this.pushToTalkOrigin = this.snapshot.inputMuted ? "muted" : "hands-free";
+				this.pushToTalkPending = true;
+				if (state === "connecting") this.setInputMutedWithoutCommit(true);
+				this.activatePushToTalk();
+			}
+			/** Finish or cancel the temporary manually committed utterance. */
+			endPushToTalk() {
+				if (!this.pushToTalkPending && !this.pushToTalkActive) return;
+				if (!this.pushToTalkActive) {
+					this.pushToTalkPending = false;
+					this.restorePendingInputState();
+					return;
+				}
+				const transport = this.transport;
+				if (transport === void 0) {
+					this.pushToTalkActive = false;
+					this.pushToTalkPending = false;
+					this.publish({
+						...this.snapshot,
+						pushToTalkActive: false
+					});
+					return;
+				}
+				const origin = this.pushToTalkOrigin;
+				const restoreMuted = origin === "off" || origin === "muted";
+				try {
+					transport.commitAudio();
+					transport.setInputMuted(restoreMuted, false);
+				} catch {
+					this.handleUnexpectedClose();
+					return;
+				}
+				this.pushToTalkActive = false;
+				this.pushToTalkPending = false;
+				this.pushToTalkOrigin = void 0;
+				this.publish({
+					...this.snapshot,
+					inputMuted: restoreMuted,
+					pushToTalkActive: false
+				});
+			}
+			activatePushToTalk() {
+				if (!this.pushToTalkPending || this.pushToTalkActive || this.transport === void 0) return;
+				try {
+					this.transport.setInputMuted(false, false);
+					this.transport.beginManualUtterance();
+				} catch {
+					this.handleUnexpectedClose();
+					return;
+				}
+				this.pushToTalkActive = true;
+				this.publish({
+					...this.snapshot,
+					inputMuted: false,
+					pushToTalkActive: true
+				});
+			}
+			restorePendingInputState() {
+				const origin = this.pushToTalkOrigin;
+				if (origin === void 0) return;
+				const restoreMuted = origin === "off" || origin === "muted";
+				(this.transport ?? this.openingTransport)?.setInputMuted(restoreMuted, false);
+				this.pushToTalkOrigin = void 0;
+				this.publish({
+					...this.snapshot,
+					inputMuted: restoreMuted,
+					pushToTalkActive: false
+				});
+			}
+			setInputMutedWithoutCommit(muted) {
+				this.openingTransport?.setInputMuted(muted, false);
+				this.transport?.setInputMuted(muted, false);
+				if (this.snapshot.inputMuted !== muted) this.publish({
+					...this.snapshot,
+					inputMuted: muted
+				});
+			}
 			async open(sessionId, inputMuted) {
 				const generation = ++this.generation;
 				this.pendingText.splice(0);
@@ -1060,6 +1219,7 @@ window.__ModuleLoader__.load({
 				this.publish({
 					state: "connecting",
 					inputMuted,
+					pushToTalkActive: false,
 					sessionId,
 					textById: {}
 				});
@@ -1090,7 +1250,8 @@ window.__ModuleLoader__.load({
 					this.openingTransport = void 0;
 					this.transport = transport;
 					for (const text of this.pendingText.splice(0)) transport.submitText(text);
-					this.setState("listening");
+					if (this.pushToTalkPending) this.activatePushToTalk();
+					if (this.snapshot.state !== "error") this.setState("listening");
 				} catch (cause) {
 					if (generation !== this.generation) return;
 					this.setState("error");
@@ -1146,6 +1307,9 @@ window.__ModuleLoader__.load({
 				this.openingAbort = void 0;
 				this.publish(INITIAL_SNAPSHOT);
 				this.pendingText.splice(0);
+				this.pushToTalkPending = false;
+				this.pushToTalkActive = false;
+				this.pushToTalkOrigin = void 0;
 				await this.closeTransport();
 			}
 			async closeTransport() {
@@ -1165,6 +1329,13 @@ window.__ModuleLoader__.load({
 			}
 			handleUnexpectedClose() {
 				this.openingAbort?.abort();
+				this.pushToTalkPending = false;
+				this.pushToTalkActive = false;
+				this.pushToTalkOrigin = void 0;
+				if (this.snapshot.pushToTalkActive) this.publish({
+					...this.snapshot,
+					pushToTalkActive: false
+				});
 				this.setState("error");
 				this.closeTransport().catch((cause) => {
 					console.error("voice transport cleanup failed:", cause);
@@ -1295,6 +1466,14 @@ window.__ModuleLoader__.load({
 			const chunks = [];
 			const playing = /* @__PURE__ */ new Set();
 			const transport = {
+				beginManualUtterance: () => {
+					if (socket.readyState !== WebSocket.OPEN) throw new Error("voice transport is not open");
+					socket.send(JSON.stringify({ type: "audio.push-to-talk.start" }));
+				},
+				commitAudio: () => {
+					if (socket.readyState !== WebSocket.OPEN) throw new Error("voice transport is not open");
+					socket.send(JSON.stringify({ type: "audio.commit" }));
+				},
 				cancelTask: (taskId) => {
 					if (socket.readyState !== WebSocket.OPEN) throw new Error("voice transport is not open");
 					socket.send(JSON.stringify({
@@ -1307,12 +1486,12 @@ window.__ModuleLoader__.load({
 					socket.send(JSON.stringify({ type: "response.interrupt" }));
 					stopPlayback();
 				},
-				setInputMuted: (muted) => {
+				setInputMuted: (muted, commit = true) => {
 					if (inputMuted === muted) return;
 					inputMuted = muted;
 					chunks.splice(0);
 					for (const track of stream?.getTracks() ?? []) track.enabled = !muted;
-					if (muted && !stopped && socket.readyState === WebSocket.OPEN) socket.send(JSON.stringify({ type: "audio.commit" }));
+					if (muted && commit && !stopped && socket.readyState === WebSocket.OPEN) socket.send(JSON.stringify({ type: "audio.commit" }));
 				},
 				submitText: (text) => {
 					if (socket.readyState !== WebSocket.OPEN) throw new Error("voice transport is not open");
@@ -1992,6 +2171,7 @@ registerProcessor('dsh-pcm-capture', DshPcmCaptureProcessor)
 			"control.interrupt": "打断播放",
 			"control.mute": "静音麦克风",
 			"control.unmute": "解除静音",
+			"control.pushToTalk": "按住说话中，松开发送",
 			"overlay.connecting": "正在连接语音",
 			"overlay.listening": "正在聆听",
 			"overlay.speaking": "正在回复",
@@ -2041,6 +2221,7 @@ registerProcessor('dsh-pcm-capture', DshPcmCaptureProcessor)
 			"control.interrupt": "Stop playback",
 			"control.mute": "Mute microphone",
 			"control.unmute": "Unmute microphone",
+			"control.pushToTalk": "Push to talk, release to send",
 			"overlay.connecting": "Connecting voice",
 			"overlay.listening": "Listening",
 			"overlay.speaking": "Playing response",
@@ -2265,6 +2446,12 @@ registerProcessor('dsh-pcm-capture', DshPcmCaptureProcessor)
 					const sessionId = controller.getSnapshot().sessionId;
 					if (sessionId !== void 0) history.record(sessionId);
 				},
+				beginPushToTalk: (sessionId) => {
+					controller.beginPushToTalk(sessionId);
+				},
+				endPushToTalk: () => {
+					controller.endPushToTalk();
+				},
 				interruptResponse: () => {
 					controller.interruptResponse();
 				},