agentgui 1.0.507 → 1.0.508

package/lib/tool-manager.js

@@ -254,23 +254,74 @@ export async function checkForUpdates(toolId) {
 
 const spawnBunxProc = (pkg, onProgress) => new Promise((resolve) => {
   const cmd = isWindows ? 'bunx.cmd' : 'bunx';
-  const proc = spawn(cmd, [pkg], { stdio: ['pipe', 'pipe', 'pipe'], timeout: 300000, shell: isWindows });
   let completed = false, stderr = '', stdout = '';
-  const timer = setTimeout(() => { if (!completed) { completed = true; try { proc.kill('SIGKILL'); } catch (_) {} resolve({ success: false, error: 'Timeout (5min)' }); }}, 300000);
-  proc.stdout.on('data', (d) => { stdout += d.toString(); if (onProgress) onProgress({ type: 'progress', data: d.toString() }); });
-  proc.stderr.on('data', (d) => { stderr += d.toString(); if (onProgress) onProgress({ type: 'error', data: d.toString() }); });
+  let lastDataTime = Date.now();
+  let proc;
+
+  try {
+    proc = spawn(cmd, [pkg], { stdio: ['pipe', 'pipe', 'pipe'], timeout: 300000, shell: isWindows });
+  } catch (err) {
+    return resolve({ success: false, error: `Failed to spawn bunx: ${err.message}` });
+  }
+
+  if (!proc) {
+    return resolve({ success: false, error: 'Failed to spawn bunx process' });
+  }
+
+  const timer = setTimeout(() => {
+    if (!completed) {
+      completed = true;
+      try { proc.kill('SIGKILL'); } catch (_) {}
+      resolve({ success: false, error: 'Timeout (5min)' });
+    }
+  }, 300000);
+
+  const heartbeatTimer = setInterval(() => {
+    if (completed) { clearInterval(heartbeatTimer); return; }
+    const timeSinceLastData = Date.now() - lastDataTime;
+    if (timeSinceLastData > 30000) {
+      console.warn(`[tool-manager] No output from bunx ${pkg} for ${timeSinceLastData}ms - process may be hung`);
+    }
+  }, 30000);
+
+  const onData = (d) => {
+    lastDataTime = Date.now();
+    if (onProgress) onProgress({ type: 'progress', data: d.toString() });
+  };
+
+  if (proc.stdout) proc.stdout.on('data', (d) => { stdout += d.toString(); onData(d); });
+  if (proc.stderr) proc.stderr.on('data', (d) => { stderr += d.toString(); onData(d); });
+
   proc.on('close', (code) => {
     clearTimeout(timer);
+    clearInterval(heartbeatTimer);
     if (completed) return;
     completed = true;
     const output = stdout + stderr;
-    if (code === 0 || output.includes('upgraded') || output.includes('registered') || output.includes('Hooks registered')) {
+    const successPatterns = [
+      code === 0,
+      output.includes('upgraded'),
+      output.includes('registered'),
+      output.includes('Hooks registered'),
+      output.includes('successfully'),
+      output.includes('Done'),
+      code === 0 && !output.includes('error')
+    ];
+    if (successPatterns.some(p => p)) {
       resolve({ success: true, error: null });
     } else {
       resolve({ success: false, error: output.substring(0, 1000) || 'Failed' });
     }
   });
-  proc.on('error', (err) => { clearTimeout(timer); if (!completed) { completed = true; resolve({ success: false, error: err.message }); }});
+
+  proc.on('error', (err) => {
+    clearTimeout(timer);
+    clearInterval(heartbeatTimer);
+    if (!completed) {
+      completed = true;
+      resolve({ success: false, error: `Process error: ${err.message}` });
+    }
+  });
 });
 
 export async function install(toolId, onProgress) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentgui",
-  "version": "1.0.507",
+  "version": "1.0.508",
   "description": "Multi-agent ACP client with real-time communication",
   "type": "module",
   "main": "server.js",

package/static/index.html

@@ -3231,6 +3231,14 @@
             aria-label="Message input"
             rows="1"
           ></textarea>
+          <button class="voice-mic-btn" id="chatMicBtn" title="Record voice input" aria-label="Voice input">
+            <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
+              <path d="M12 1a3 3 0 0 0-3 3v8a3 3 0 0 0 6 0V4a3 3 0 0 0-3-3z"/>
+              <path d="M19 10v2a7 7 0 0 1-14 0v-2"/>
+              <line x1="12" y1="19" x2="12" y2="23"/>
+              <line x1="8" y1="23" x2="16" y2="23"/>
+            </svg>
+          </button>
           <button class="inject-btn" id="injectBtn" title="Inject instructions into running agent" aria-label="Inject instructions">
             <svg viewBox="0 0 24 24" fill="currentColor" width="18" height="18">
               <path d="M20 12l-1.41-1.41L13 16.17V4h-2v12.17l-5.58-5.59L4 12l8 8 8-8z"/>
@@ -3288,6 +3296,7 @@
   <script defer src="/gm/js/terminal.js"></script>
   <script defer src="/gm/js/script-runner.js"></script>
   <script defer src="/gm/js/tools-manager.js"></script>
+  <script defer src="/gm/js/stt-handler.js"></script>
   <script defer src="/gm/js/client.js"></script>
   <script type="module" src="/gm/js/voice.js"></script>
   <script defer src="/gm/js/features.js"></script>

package/static/js/client.js

@@ -394,6 +394,8 @@ class AgentGUIClient {
       this.ui.sendButton.addEventListener('click', () => this.startExecution());
     }
 
+    this.setupChatMicButton();
+
     this.ui.stopButton = document.getElementById('stopBtn');
     this.ui.injectButton = document.getElementById('injectBtn');
 
@@ -486,6 +488,52 @@ class AgentGUIClient {
     });
   }
 
+  setupChatMicButton() {
+    const chatMicBtn = document.getElementById('chatMicBtn');
+    if (!chatMicBtn) return;
+
+    let isRecording = false;
+
+    chatMicBtn.addEventListener('mousedown', async (e) => {
+      e.preventDefault();
+      if (isRecording) return;
+      isRecording = true;
+      chatMicBtn.classList.add('recording');
+      const result = await window.STTHandler.startRecording();
+      if (!result.success) {
+        isRecording = false;
+        chatMicBtn.classList.remove('recording');
+        alert('Microphone access denied: ' + result.error);
+      }
+    });
+
+    chatMicBtn.addEventListener('mouseup', async (e) => {
+      e.preventDefault();
+      if (!isRecording) return;
+      isRecording = false;
+      chatMicBtn.classList.remove('recording');
+      const result = await window.STTHandler.stopRecording();
+      if (result.success) {
+        if (this.ui.messageInput) {
+          this.ui.messageInput.value = result.text;
+        }
+      } else {
+        alert('Transcription failed: ' + result.error);
+      }
+    });
+
+    chatMicBtn.addEventListener('mouseleave', async (e) => {
+      if (isRecording) {
+        isRecording = false;
+        chatMicBtn.classList.remove('recording');
+        const result = await window.STTHandler.stopRecording();
+        if (result.success && this.ui.messageInput) {
+          this.ui.messageInput.value = result.text;
+        }
+      }
+    });
+  }
+
   /**
    * Connect to WebSocket
    */

package/static/js/stt-handler.js

@@ -0,0 +1,138 @@
+(function() {
+  var BASE = window.__BASE_URL || '';
+  var isRecording = false;
+  var mediaStream = null;
+  var audioContext = null;
+  var workletNode = null;
+  var recordedChunks = [];
+  var TARGET_SAMPLE_RATE = 16000;
+
+  window.STTHandler = {
+    isRecording: function() { return isRecording; },
+
+    startRecording: async function() {
+      if (isRecording) return;
+      try {
+        mediaStream = await navigator.mediaDevices.getUserMedia({ audio: true });
+        audioContext = new (window.AudioContext || window.webkitAudioContext)();
+        var source = audioContext.createMediaStreamSource(mediaStream);
+        recordedChunks = [];
+        await audioContext.audioWorklet.addModule(BASE + '/js/audio-recorder-processor.js');
+        workletNode = new AudioWorkletNode(audioContext, 'recorder-processor');
+        workletNode.port.onmessage = function(e) {
+          recordedChunks.push(e.data);
+        };
+        source.connect(workletNode);
+        isRecording = true;
+        return { success: true };
+      } catch (e) {
+        isRecording = false;
+        return { success: false, error: e.message };
+      }
+    },
+
+    stopRecording: async function() {
+      if (!isRecording) return { success: false, error: 'Not recording' };
+      isRecording = false;
+
+      if (workletNode) {
+        workletNode.port.postMessage('stop');
+        workletNode.disconnect();
+        workletNode = null;
+      }
+
+      if (mediaStream) {
+        mediaStream.getTracks().forEach(function(t) { t.stop(); });
+        mediaStream = null;
+      }
+
+      var sourceSampleRate = audioContext ? audioContext.sampleRate : 48000;
+      if (audioContext) {
+        audioContext.close().catch(function() {});
+        audioContext = null;
+      }
+
+      if (recordedChunks.length === 0) {
+        return { success: false, error: 'No audio recorded' };
+      }
+
+      var totalLen = 0;
+      for (var i = 0; i < recordedChunks.length; i++) totalLen += recordedChunks[i].length;
+      var merged = new Float32Array(totalLen);
+      var offset = 0;
+      for (var j = 0; j < recordedChunks.length; j++) {
+        merged.set(recordedChunks[j], offset);
+        offset += recordedChunks[j].length;
+      }
+      recordedChunks = [];
+
+      var resampled = resampleBuffer(merged, sourceSampleRate, TARGET_SAMPLE_RATE);
+      var wavBuffer = encodeWav(resampled, TARGET_SAMPLE_RATE);
+
+      try {
+        var resp = await fetch(BASE + '/api/stt', {
+          method: 'POST',
+          headers: { 'Content-Type': 'audio/wav' },
+          body: wavBuffer
+        });
+        var data = await resp.json();
+        if (data.text) {
+          return { success: true, text: data.text };
+        } else if (data.error) {
+          return { success: false, error: data.error };
+        } else {
+          return { success: false, error: 'No transcription returned' };
+        }
+      } catch (e) {
+        return { success: false, error: 'Transcription failed: ' + e.message };
+      }
+    }
+  };
+
+  function resampleBuffer(inputBuffer, fromRate, toRate) {
+    if (fromRate === toRate) return inputBuffer;
+    var ratio = fromRate / toRate;
+    var newLen = Math.round(inputBuffer.length / ratio);
+    var result = new Float32Array(newLen);
+    for (var i = 0; i < newLen; i++) {
+      var srcIdx = i * ratio;
+      var lo = Math.floor(srcIdx);
+      var hi = Math.min(lo + 1, inputBuffer.length - 1);
+      var frac = srcIdx - lo;
+      result[i] = inputBuffer[lo] * (1 - frac) + inputBuffer[hi] * frac;
+    }
+    return result;
+  }
+
+  function encodeWav(float32Audio, sampleRate) {
+    var numSamples = float32Audio.length;
+    var bytesPerSample = 2;
+    var dataSize = numSamples * bytesPerSample;
+    var buffer = new ArrayBuffer(44 + dataSize);
+    var view = new DataView(buffer);
+
+    function writeStr(off, str) {
+      for (var i = 0; i < str.length; i++) view.setUint8(off + i, str.charCodeAt(i));
+    }
+
+    writeStr(0, 'RIFF');
+    view.setUint32(4, 36 + dataSize, true);
+    writeStr(8, 'WAVE');
+    writeStr(12, 'fmt ');
+    view.setUint32(16, 16, true);
+    view.setUint16(20, 1, true);
+    view.setUint16(22, 1, true);
+    view.setUint32(24, sampleRate, true);
+    view.setUint32(28, sampleRate * bytesPerSample, true);
+    view.setUint16(32, bytesPerSample, true);
+    view.setUint16(34, 16, true);
+    writeStr(36, 'data');
+    view.setUint32(40, dataSize, true);
+
+    for (var i = 0; i < numSamples; i++) {
+      var s = Math.max(-1, Math.min(1, float32Audio[i]));
+      view.setInt16(44 + i * 2, s < 0 ? s * 32768 : s * 32767, true);
+    }
+    return buffer;
+  }
+})();

package/static/js/voice.js

@@ -271,49 +271,6 @@
     }
   }
 
-  function resampleBuffer(inputBuffer, fromRate, toRate) {
-    if (fromRate === toRate) return inputBuffer;
-    var ratio = fromRate / toRate;
-    var newLen = Math.round(inputBuffer.length / ratio);
-    var result = new Float32Array(newLen);
-    for (var i = 0; i < newLen; i++) {
-      var srcIdx = i * ratio;
-      var lo = Math.floor(srcIdx);
-      var hi = Math.min(lo + 1, inputBuffer.length - 1);
-      var frac = srcIdx - lo;
-      result[i] = inputBuffer[lo] * (1 - frac) + inputBuffer[hi] * frac;
-    }
-    return result;
-  }
-
-  function encodeWav(float32Audio, sampleRate) {
-    var numSamples = float32Audio.length;
-    var bytesPerSample = 2;
-    var dataSize = numSamples * bytesPerSample;
-    var buffer = new ArrayBuffer(44 + dataSize);
-    var view = new DataView(buffer);
-    function writeStr(off, str) {
-      for (var i = 0; i < str.length; i++) view.setUint8(off + i, str.charCodeAt(i));
-    }
-    writeStr(0, 'RIFF');
-    view.setUint32(4, 36 + dataSize, true);
-    writeStr(8, 'WAVE');
-    writeStr(12, 'fmt ');
-    view.setUint32(16, 16, true);
-    view.setUint16(20, 1, true);
-    view.setUint16(22, 1, true);
-    view.setUint32(24, sampleRate, true);
-    view.setUint32(28, sampleRate * bytesPerSample, true);
-    view.setUint16(32, bytesPerSample, true);
-    view.setUint16(34, 16, true);
-    writeStr(36, 'data');
-    view.setUint32(40, dataSize, true);
-    for (var i = 0; i < numSamples; i++) {
-      var s = Math.max(-1, Math.min(1, float32Audio[i]));
-      view.setInt16(44 + i * 2, s < 0 ? s * 32768 : s * 32767, true);
-    }
-    return buffer;
-  }
 
   async function startRecording() {
     if (isRecording) return;
@@ -326,23 +283,13 @@
         el.setAttribute('data-final', '');
       }
     }
-    try {
-      mediaStream = await navigator.mediaDevices.getUserMedia({ audio: true });
-      audioContext = new (window.AudioContext || window.webkitAudioContext)();
-      var source = audioContext.createMediaStreamSource(mediaStream);
-      recordedChunks = [];
-      await audioContext.audioWorklet.addModule(BASE + '/js/audio-recorder-processor.js');
-      workletNode = new AudioWorkletNode(audioContext, 'recorder-processor');
-      workletNode.port.onmessage = function(e) {
-        recordedChunks.push(e.data);
-      };
-      source.connect(workletNode);
+    var result = await window.STTHandler.startRecording();
+    if (result.success) {
       isRecording = true;
       var micBtn = document.getElementById('voiceMicBtn');
       if (micBtn) micBtn.classList.add('recording');
-    } catch (e) {
-      isRecording = false;
-      if (el) el.textContent = 'Mic access denied or unavailable: ' + e.message;
+    } else {
+      if (el) el.textContent = 'Mic access denied: ' + result.error;
     }
   }
 
@@ -352,24 +299,7 @@
     var micBtn = document.getElementById('voiceMicBtn');
     if (micBtn) micBtn.classList.remove('recording');
     var el = document.getElementById('voiceTranscript');
-    if (workletNode) { workletNode.port.postMessage('stop'); workletNode.disconnect(); workletNode = null; }
-    if (mediaStream) {
-      mediaStream.getTracks().forEach(function(t) { t.stop(); });
-      mediaStream = null;
-    }
-    var sourceSampleRate = audioContext ? audioContext.sampleRate : 48000;
-    if (audioContext) { audioContext.close().catch(function() {}); audioContext = null; }
-    if (recordedChunks.length === 0) return;
-    var totalLen = 0;
-    for (var i = 0; i < recordedChunks.length; i++) totalLen += recordedChunks[i].length;
-    var merged = new Float32Array(totalLen);
-    var offset = 0;
-    for (var j = 0; j < recordedChunks.length; j++) {
-      merged.set(recordedChunks[j], offset);
-      offset += recordedChunks[j].length;
-    }
-    recordedChunks = [];
-    var resampled = resampleBuffer(merged, sourceSampleRate, TARGET_SAMPLE_RATE);
+
     if (el) {
       if (el.value !== undefined) {
         el.value = 'Transcribing...';
@@ -377,46 +307,23 @@
         el.textContent = 'Transcribing...';
       }
     }
-    try {
-      var wavBuffer = encodeWav(resampled, TARGET_SAMPLE_RATE);
-      var resp = await fetch(BASE + '/api/stt', {
-        method: 'POST',
-        headers: { 'Content-Type': 'audio/wav' },
-        body: wavBuffer
-      });
-      var data = await resp.json();
-      if (data.text) {
-        if (el) {
-          if (el.value !== undefined) {
-            el.value = data.text;
-          } else {
-            el.textContent = data.text;
-            el.setAttribute('data-final', data.text);
-          }
-        }
-      } else if (data.error) {
-        if (el) {
-          if (el.value !== undefined) {
-            el.value = 'Error: ' + data.error;
-          } else {
-            el.textContent = 'Error: ' + data.error;
-          }
-        }
-      } else {
-        if (el) {
-          if (el.value !== undefined) {
-            el.value = '';
-          } else {
-            el.textContent = '';
-          }
+
+    var result = await window.STTHandler.stopRecording();
+    if (result.success) {
+      if (el) {
+        if (el.value !== undefined) {
+          el.value = result.text;
+        } else {
+          el.textContent = result.text;
+          el.setAttribute('data-final', result.text);
         }
       }
-    } catch (e) {
+    } else {
       if (el) {
         if (el.value !== undefined) {
-          el.value = 'Transcription failed: ' + e.message;
+          el.value = 'Error: ' + result.error;
         } else {
-          el.textContent = 'Transcription failed: ' + e.message;
+          el.textContent = 'Error: ' + result.error;
         }
       }
     }