Merge branch 'master' of https://gitea.xiaowang.eu.org/wx44wx/AI-VideoAssistant

2026-02-27 17:38:48 +08:00
parent 531cf6080a ebb6f59878
commit e976f7d833
2 changed files with 133 additions and 29 deletions
--- a/engine/core/duplex_pipeline.py
+++ b/engine/core/duplex_pipeline.py
@@ -1673,6 +1673,13 @@ class DuplexPipeline:
    def _tool_wait_for_response(self, tool_name: str) -> bool:
        return bool(self._runtime_tool_wait_for_response.get(tool_name, False))

+    def _tool_wait_timeout_seconds(self, tool_name: str) -> float:
+        if tool_name == "text_msg_prompt" and self._tool_wait_for_response(tool_name):
+            # Keep engine wait slightly longer than UI auto-close (120s)
+            # to avoid race where engine times out before client emits timeout result.
+            return 125.0
+        return self._TOOL_WAIT_TIMEOUT_SECONDS
+
    def _tool_executor(self, tool_call: Dict[str, Any]) -> str:
        name = self._tool_name(tool_call)
        if name and name in self._runtime_tool_executor:
@@ -1792,10 +1799,17 @@ class DuplexPipeline:
            self._early_tool_results[call_id] = item
            self._completed_tool_call_ids.add(call_id)

-    async def _wait_for_single_tool_result(self, call_id: str) -> Dict[str, Any]:
+    async def _wait_for_single_tool_result(
+        self,
+        call_id: str,
+        *,
+        tool_name: str = "unknown_tool",
+        timeout_seconds: Optional[float] = None,
+    ) -> Dict[str, Any]:
        if call_id in self._completed_tool_call_ids and call_id not in self._early_tool_results:
            return {
                "tool_call_id": call_id,
+                "name": tool_name,
                "status": {"code": 208, "message": "tool_call result already handled"},
                "output": "",
            }
@@ -1806,12 +1820,14 @@ class DuplexPipeline:
        loop = asyncio.get_running_loop()
        future = loop.create_future()
        self._pending_tool_waiters[call_id] = future
+        wait_timeout = float(timeout_seconds if isinstance(timeout_seconds, (int, float)) and timeout_seconds > 0 else self._TOOL_WAIT_TIMEOUT_SECONDS)
        try:
-            return await asyncio.wait_for(future, timeout=self._TOOL_WAIT_TIMEOUT_SECONDS)
+            return await asyncio.wait_for(future, timeout=wait_timeout)
        except asyncio.TimeoutError:
            self._completed_tool_call_ids.add(call_id)
            return {
                "tool_call_id": call_id,
+                "name": tool_name,
                "status": {"code": 504, "message": "tool_call timeout"},
                "output": "",
            }
@@ -1900,7 +1916,9 @@ class DuplexPipeline:
                        tool_id = self._tool_id_for_name(tool_name)
                        tool_display_name = self._tool_display_name(tool_name) or tool_name
                        wait_for_response = self._tool_wait_for_response(tool_name)
+                        wait_timeout_seconds = self._tool_wait_timeout_seconds(tool_name)
                        enriched_tool_call["wait_for_response"] = wait_for_response
+                        enriched_tool_call["wait_timeout_ms"] = int(wait_timeout_seconds * 1000)
                        call_id = str(enriched_tool_call.get("id") or "").strip()
                        fn_payload = (
                            dict(enriched_tool_call.get("function"))
@@ -1935,9 +1953,10 @@ class DuplexPipeline:
                                    tool_id=tool_id,
                                    tool_display_name=tool_display_name,
                                    wait_for_response=wait_for_response,
+                                    wait_timeout_ms=int(wait_timeout_seconds * 1000),
                                    arguments=tool_arguments,
                                    executor=executor,
-                                    timeout_ms=int(self._TOOL_WAIT_TIMEOUT_SECONDS * 1000),
+                                    timeout_ms=int(wait_timeout_seconds * 1000),
                                    tool_call=enriched_tool_call,
                                )
                            },
@@ -2075,7 +2094,11 @@ class DuplexPipeline:
                    tool_id = self._tool_id_for_name(tool_name)
                    logger.info(f"[Tool] execute start name={tool_name} call_id={call_id} executor={executor}")
                    if executor == "client":
-                        result = await self._wait_for_single_tool_result(call_id)
+                        result = await self._wait_for_single_tool_result(
+                            call_id,
+                            tool_name=tool_name,
+                            timeout_seconds=self._tool_wait_timeout_seconds(tool_name),
+                        )
                        await self._emit_tool_result(result, source="client")
                        tool_results.append(result)
                        continue
--- a/web/pages/Assistants.tsx
+++ b/web/pages/Assistants.tsx
@@ -1901,27 +1901,68 @@ const TranscriptionLog: React.FC<{
  messages: DebugTranscriptMessage[];
  isLoading: boolean;
  className?: string;
-}> = ({ scrollRef, messages, isLoading, className = '' }) => (
-  <div ref={scrollRef} className={`overflow-y-auto overflow-x-hidden space-y-4 p-2 border border-white/5 rounded-md bg-black/20 min-h-0 custom-scrollbar ${className}`}>
-    {messages.length === 0 && <div className="text-center text-muted-foreground text-xs py-4">暂无转写记录</div>}
-    {messages.map((m, i) => (
-      <div key={i} className={`flex ${m.role === 'user' ? 'justify-end' : 'justify-start'}`}>
-        <div className={`max-w-[85%] rounded-lg px-3 py-2 text-sm ${m.role === 'user' ? 'bg-primary text-primary-foreground' : m.role === 'tool' ? 'bg-amber-500/10 border border-amber-400/30 text-amber-100' : 'bg-card border border-white/10 shadow-sm text-foreground'}`}>
-          <div className="mb-0.5 flex items-center gap-1.5">
-            <span className="text-[10px] opacity-70 uppercase tracking-wider">{m.role === 'user' ? 'Me' : m.role === 'tool' ? 'Tool' : 'AI'}</span>
-            {m.role === 'model' && typeof m.ttfbMs === 'number' && Number.isFinite(m.ttfbMs) && (
-              <span className="rounded border border-cyan-300/40 bg-cyan-500/10 px-1.5 py-0.5 text-[10px] text-cyan-200">
-                TTFB {Math.round(m.ttfbMs)}ms
-              </span>
+}> = ({ scrollRef, messages, isLoading, className = '' }) => {
+  const [copiedIndex, setCopiedIndex] = useState<number | null>(null);
+
+  const copyText = async (text: string, index: number) => {
+    const value = String(text || '');
+    if (!value) return;
+    try {
+      if (typeof navigator !== 'undefined' && navigator.clipboard?.writeText) {
+        await navigator.clipboard.writeText(value);
+      } else {
+        const textarea = document.createElement('textarea');
+        textarea.value = value;
+        textarea.style.position = 'fixed';
+        textarea.style.opacity = '0';
+        document.body.appendChild(textarea);
+        textarea.select();
+        document.execCommand('copy');
+        document.body.removeChild(textarea);
+      }
+      setCopiedIndex(index);
+      window.setTimeout(() => {
+        setCopiedIndex((prev) => (prev === index ? null : prev));
+      }, 1200);
+    } catch {
+      // no-op
+    }
+  };
+
+  return (
+    <div ref={scrollRef} className={`overflow-y-auto overflow-x-hidden space-y-4 p-2 border border-white/5 rounded-md bg-black/20 min-h-0 custom-scrollbar ${className}`}>
+      {messages.length === 0 && <div className="text-center text-muted-foreground text-xs py-4">暂无转写记录</div>}
+      {messages.map((m, i) => (
+        <div key={i} className={`flex ${m.role === 'user' ? 'justify-end' : 'justify-start'}`}>
+          <div className="flex items-center gap-2">
+            {m.role === 'user' && (
+              <button
+                type="button"
+                onClick={() => void copyText(m.text, i)}
+                className="h-6 w-6 shrink-0 rounded border border-white/20 bg-black/20 text-primary-foreground/80 hover:bg-black/35 transition-colors flex items-center justify-center"
+                title="复制此消息"
+              >
+                {copiedIndex === i ? <ClipboardCheck className="h-3.5 w-3.5" /> : <Copy className="h-3.5 w-3.5" />}
+              </button>
            )}
+            <div className={`max-w-[85%] rounded-lg px-3 py-2 text-sm ${m.role === 'user' ? 'bg-primary text-primary-foreground' : m.role === 'tool' ? 'bg-amber-500/10 border border-amber-400/30 text-amber-100' : 'bg-card border border-white/10 shadow-sm text-foreground'}`}>
+              <div className="mb-0.5 flex items-center gap-1.5">
+                <span className="text-[10px] opacity-70 uppercase tracking-wider">{m.role === 'user' ? 'Me' : m.role === 'tool' ? 'Tool' : 'AI'}</span>
+                {m.role === 'model' && typeof m.ttfbMs === 'number' && Number.isFinite(m.ttfbMs) && (
+                  <span className="rounded border border-cyan-300/40 bg-cyan-500/10 px-1.5 py-0.5 text-[10px] text-cyan-200">
+                    TTFB {Math.round(m.ttfbMs)}ms
+                  </span>
+                )}
+              </div>
+              {m.text}
+            </div>
          </div>
-          {m.text}
        </div>
-      </div>
-    ))}
-    {isLoading && <div className="text-xs text-muted-foreground ml-2 animate-pulse">Thinking...</div>}
-  </div>
-);
+      ))}
+      {isLoading && <div className="text-xs text-muted-foreground ml-2 animate-pulse">Thinking...</div>}
+    </div>
+  );
+};

 // --- Debug Drawer Component ---
 export const DebugDrawer: React.FC<{
@@ -2086,6 +2127,7 @@ export const DebugDrawer: React.FC<{
  const micProcessorRef = useRef<ScriptProcessorNode | null>(null);
  const micGainRef = useRef<GainNode | null>(null);
  const micFrameBufferRef = useRef<Uint8Array>(new Uint8Array(0));
+  const textPromptTimeoutRef = useRef<number | null>(null);
  const userDraftIndexRef = useRef<number | null>(null);
  const lastUserFinalRef = useRef<string>('');
  const debugVolumePercentRef = useRef<number>(50);
@@ -2172,6 +2214,7 @@ export const DebugDrawer: React.FC<{
        stopVoiceCapture();
        stopMedia();
        closeWs();
+        clearTextPromptTimeout();
        setTextPromptDialog({ open: false, message: '' });
        if (audioCtxRef.current) {
          void audioCtxRef.current.close();
@@ -2215,6 +2258,12 @@ export const DebugDrawer: React.FC<{
    }
  }, [clientToolEnabledMap]);

+  useEffect(() => {
+    return () => {
+      clearTextPromptTimeout();
+    };
+  }, []);
+
  // Auto-scroll logic
  useEffect(() => {
      if (scrollRef.current) {
@@ -2399,6 +2448,13 @@ export const DebugDrawer: React.FC<{
    clearPlaybackQueue();
  };

+  const clearTextPromptTimeout = () => {
+    if (textPromptTimeoutRef.current !== null) {
+      window.clearTimeout(textPromptTimeoutRef.current);
+      textPromptTimeoutRef.current = null;
+    }
+  };
+
  const emitClientToolResult = (resultPayload: any, toolDisplayName?: string) => {
    const ws = wsRef.current;
    if (ws && ws.readyState === WebSocket.OPEN) {
@@ -2425,7 +2481,8 @@ export const DebugDrawer: React.FC<{
    ]);
  };

-  const closeTextPromptDialog = (action: 'confirm' | 'dismiss') => {
+  const closeTextPromptDialog = (action: 'confirm' | 'dismiss' | 'timeout') => {
+    clearTextPromptTimeout();
    let pending:
      | {
          toolCallId: string;
@@ -2441,22 +2498,45 @@ export const DebugDrawer: React.FC<{
      return { open: false, message: '' };
    });
    if (pending?.waitForResponse) {
+      const isTimeout = action === 'timeout';
      emitClientToolResult(
        {
          tool_call_id: pending.toolCallId,
          name: pending.toolName,
          output: {
-            message: 'text_prompt_closed',
+            message: isTimeout ? 'text_prompt_timeout' : 'text_prompt_closed',
            action,
            msg: message,
          },
-          status: { code: 200, message: 'ok' },
+          status: isTimeout ? { code: 504, message: 'tool_call timeout' } : { code: 200, message: 'ok' },
        },
        pending.toolDisplayName
      );
    }
  };

+  const openTextPromptDialog = (args: {
+    message: string;
+    pendingResult?: {
+      toolCallId: string;
+      toolName: string;
+      toolDisplayName: string;
+      waitForResponse: boolean;
+    };
+    autoCloseMs?: number;
+  }) => {
+    clearTextPromptTimeout();
+    setTextPromptDialog({
+      open: true,
+      message: args.message,
+      pendingResult: args.pendingResult,
+    });
+    const timeoutMs = Math.max(1000, Number(args.autoCloseMs || 120000));
+    textPromptTimeoutRef.current = window.setTimeout(() => {
+      closeTextPromptDialog('timeout');
+    }, timeoutMs);
+  };
+
  const scheduleQueuedPlayback = (ctx: AudioContext) => {
    const queue = queuedAudioBuffersRef.current;
    if (queue.length === 0) return;
@@ -2587,6 +2667,8 @@ export const DebugDrawer: React.FC<{
  const handleHangup = () => {
    if (textPromptDialog.open) {
      closeTextPromptDialog('dismiss');
+    } else {
+      clearTextPromptTimeout();
    }
    stopVoiceCapture();
    stopMedia();
@@ -2594,7 +2676,6 @@ export const DebugDrawer: React.FC<{
    setCallStatus('idle');
    clearResponseTracking();
    setMessages([]);
-    setTextPromptDialog({ open: false, message: '' });
    lastUserFinalRef.current = '';
    setIsLoading(false);
  };
@@ -3236,8 +3317,7 @@ export const DebugDrawer: React.FC<{
                  resultPayload.output = { message: "Missing required argument 'msg'" };
                  resultPayload.status = { code: 422, message: 'invalid_arguments' };
                } else {
-                  setTextPromptDialog({
-                    open: true,
+                  openTextPromptDialog({
                    message: msg,
                    pendingResult: {
                      toolCallId: toolCallId,
@@ -3245,6 +3325,7 @@ export const DebugDrawer: React.FC<{
                      toolDisplayName,
                      waitForResponse,
                    },
+                    autoCloseMs: 120000,
                  });
                  if (!waitForResponse) {
                    resultPayload.output = { message: 'text_prompt_shown', msg };
@@ -3649,7 +3730,7 @@ export const DebugDrawer: React.FC<{
                    title={enabled ? '点击关闭' : '点击开启'}
                  >
                    <span
-                      className={`absolute top-0.5 h-5 w-5 rounded-full bg-white shadow transition-transform ${enabled ? 'translate-x-5' : 'translate-x-0.5'}`}
+                      className={`absolute left-0.5 top-0.5 h-5 w-5 rounded-full bg-white shadow transition-transform ${enabled ? 'translate-x-5' : 'translate-x-0'}`}
                    />
                  </button>
                </div>