jbot: fix voice button stuck state + add global fallback handlers

Fixes:
1. Race condition: after await startRec(), check isSpaceDownRef again —
   fast-tap released before mic ready was leaving phase stuck at 'listening'
2. Global window mouseup/touchend/touchcancel listeners as fallback so
   releasing outside the button boundary always triggers handleSpaceUp
3. Show ✕ cancel button during ALL non-idle phases (was only thinking/speaking)
4. handleCancel now stops the recorder if called during listening/acquiring
5. Safety auto-cancel timeout: if stuck in 'listening' > 60s, auto-stop
6. Add stopRec to handleSpaceDown dep array

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

static_src/jbot/App.jsx

@@ -47,7 +47,9 @@ function speakText(text) {
 
 // ── Hold-to-talk voice button (defined outside App to keep stable identity) ──
 function VoiceButton({ phase, onStart, onEnd, onCancel }) {
-  const busy = phase === 'thinking' || phase === 'speaking';
+  const busy       = phase === 'thinking' || phase === 'speaking';
+  // Show cancel any time we're not fully idle
+  const showCancel = phase !== 'idle';
   let label = '🎙 按住说话';
   let mod   = '';
   if (phase === 'acquiring')      { label = '⏳ 准备中…';  mod = ' vbtn--acquiring'; }
@@ -56,6 +58,7 @@ function VoiceButton({ phase, onStart, onEnd, onCancel }) {
   else if (phase === 'speaking')  { label = '🔊 播放中…';  mod = ' vbtn--busy'; }
 
   const startEvt = (e) => { e.preventDefault(); if (!busy) onStart(); };
+  // Button-level end — global window listener is the primary fallback
   const endEvt   = (e) => { e.preventDefault(); onEnd(); };
 
   return (
@@ -72,7 +75,7 @@ function VoiceButton({ phase, onStart, onEnd, onCancel }) {
       >
         {label}
       </button>
-      {busy && (
+      {showCancel && (
         <button className="vbtn-cancel" onClick={onCancel} title="取消">✕</button>
       )}
     </div>
@@ -147,13 +150,19 @@ export default function App() {
   const abortRef = useRef(null);
 
   const handleCancel = useCallback(() => {
+    isSpaceDownRef.current = false;
+    // Stop recorder if actively recording
+    const phase = voicePhaseRef.current;
+    if (phase === 'listening' || phase === 'acquiring') {
+      stopRec().catch(() => {});
+    }
     abortRef.current?.abort();
     abortRef.current = null;
     setVoicePhase('idle');
     setVoiceStatus(null);
     setExprId('idle');
     scheduleNext();
-  }, [scheduleNext]);
+  }, [scheduleNext, stopRec]);
 
   // Load backend mode from server config on mount
   useEffect(() => {
@@ -192,11 +201,20 @@ export default function App() {
       return;
     }
 
+    // Race condition guard: user released before mic was ready (fast tap)
+    if (!isSpaceDownRef.current) {
+      stopRec().catch(() => {}); // discard the just-acquired recording
+      setVoicePhase('idle');
+      setVoiceStatus(null);
+      scheduleNext();
+      return;
+    }
+
     // Only show "listening" once the mic is actually recording
     setExprId('listening');
     setVoicePhase('listening');
     setVoiceStatus({ type: 'listening', text: '🎙 请讲话 · 松开结束' });
-  }, [clearAuto, startRec, scheduleNext]);
+  }, [clearAuto, startRec, stopRec, scheduleNext]);
 
   const handleSpaceUp = useCallback(async () => {
     if (!isSpaceDownRef.current) return;
@@ -364,6 +382,33 @@ export default function App() {
     };
   }, [handleSpaceDown, handleSpaceUp, pick]);
 
+  // ── Global mouse/touch fallback — catches release outside button boundary ──
+  useEffect(() => {
+    const onUp = () => handleSpaceUp();
+    window.addEventListener('mouseup',    onUp);
+    window.addEventListener('touchend',   onUp, { passive: true });
+    window.addEventListener('touchcancel', onUp, { passive: true });
+    return () => {
+      window.removeEventListener('mouseup',    onUp);
+      window.removeEventListener('touchend',   onUp);
+      window.removeEventListener('touchcancel', onUp);
+    };
+  }, [handleSpaceUp]);
+
+  // ── Safety timeout: auto-cancel if stuck in listening > 60 s ──
+  const recTimerRef = useRef(null);
+  useEffect(() => {
+    if (voicePhase === 'listening') {
+      recTimerRef.current = setTimeout(() => {
+        console.warn('[PTT] Auto-stop: max recording time reached');
+        handleSpaceUp();
+      }, 60_000);
+    } else {
+      clearTimeout(recTimerRef.current);
+    }
+    return () => clearTimeout(recTimerRef.current);
+  }, [voicePhase, handleSpaceUp]);
+
   // ── Render ─────────────────────────────────────────────────
 
   return (