agentgui 1.0.301 → 1.0.303

package/lib/speech.js

@@ -1,278 +1,4 @@
 import { createRequire } from 'module';
-import fs from 'fs';
-import path from 'path';
-import os from 'os';
-import { fileURLToPath } from 'url';
-
 const require = createRequire(import.meta.url);
-const __dirname = path.dirname(fileURLToPath(import.meta.url));
-const ROOT = path.dirname(__dirname);
-
-// Load modules
-let serverTTS = null;
-let serverSTT = null;
-let audioDecode = null;
-let ttsUtils = null;
-
-try { serverTTS = require('webtalk/server-tts'); } catch(e) { console.warn('[TTS] webtalk/server-tts unavailable:', e.message); }
-try { serverSTT = require('webtalk/server-stt'); } catch(e) { console.warn('[STT] webtalk/server-stt unavailable:', e.message); }
-try { audioDecode = require('audio-decode'); } catch(e) { console.warn('[TTS] audio-decode unavailable:', e.message); }
-try { ttsUtils = require('webtalk/tts-utils'); } catch(e) {}
-
-// Detect webtalk API type: old (server-tts.js with getVoices/synthesizeViaPocket)
-// vs new ONNX (server-tts-onnx.js with encodeVoiceAudio)
-const isOnnxApi = serverTTS && typeof serverTTS.encodeVoiceAudio === 'function';
-const isPocketApi = serverTTS && typeof serverTTS.getVoices === 'function';
-
-// Voice directories to scan
-const VOICE_DIRS = [
-  path.join(os.homedir(), 'voices'),
-  path.join(ROOT, 'voices'),
-  '/config/voices',
-];
-
-const AUDIO_EXTENSIONS = ['.wav', '.mp3', '.ogg', '.flac', '.m4a'];
-
-const POCKET_TTS_VOICES = [
-  { id: 'default', name: 'Default', gender: 'female', accent: 'French' },
-  { id: 'alba', name: 'Alba', gender: 'female', accent: 'French' },
-  { id: 'marius', name: 'Marius', gender: 'male', accent: 'French' },
-  { id: 'javert', name: 'Javert', gender: 'male', accent: 'French' },
-  { id: 'jean', name: 'Jean', gender: 'male', accent: 'French' },
-  { id: 'fantine', name: 'Fantine', gender: 'female', accent: 'French' },
-  { id: 'cosette', name: 'Cosette', gender: 'female', accent: 'French' },
-  { id: 'eponine', name: 'Eponine', gender: 'female', accent: 'French' },
-  { id: 'azelma', name: 'Azelma', gender: 'female', accent: 'French' },
-];
-
-const SAMPLE_RATE = 24000;
-
-// Embedding cache: voiceId -> {data, shape}
-const voiceEmbeddingCache = new Map();
-
-function getModelDir() {
-  return path.join(os.homedir(), '.gmgui', 'models', 'tts');
-}
-
-function findVoiceFile(voiceId) {
-  if (!voiceId || voiceId === 'default') return null;
-  const baseName = voiceId.replace(/^custom_/, '');
-  for (const dir of VOICE_DIRS) {
-    for (const ext of AUDIO_EXTENSIONS) {
-      const p = path.join(dir, baseName + ext);
-      if (fs.existsSync(p)) return p;
-    }
-  }
-  return null;
-}
-
-function scanVoiceDir(dir) {
-  const voices = [];
-  try {
-    if (!fs.existsSync(dir)) return voices;
-    const seen = new Set();
-    for (const file of fs.readdirSync(dir)) {
-      const ext = path.extname(file).toLowerCase();
-      if (!AUDIO_EXTENSIONS.includes(ext)) continue;
-      const baseName = path.basename(file, ext);
-      if (seen.has(baseName)) continue;
-      seen.add(baseName);
-      voices.push({
-        id: 'custom_' + baseName.replace(/[^a-zA-Z0-9_-]/g, '_'),
-        name: baseName.replace(/_/g, ' '),
-        gender: 'custom', accent: 'custom', isCustom: true,
-      });
-    }
-  } catch (_) {}
-  return voices;
-}
-
-// Encode a voice WAV file to an ONNX voice embedding
-async function getVoiceEmbedding(voiceId) {
-  if (voiceEmbeddingCache.has(voiceId)) return voiceEmbeddingCache.get(voiceId);
-  const voicePath = findVoiceFile(voiceId);
-  if (!voicePath) return null;
-  if (!audioDecode || !serverTTS || !isOnnxApi) return null;
-
-  const modelDir = getModelDir();
-  if (serverTTS.loadModels) await serverTTS.loadModels(modelDir);
-
-  const raw = fs.readFileSync(voicePath);
-  const decoded = await audioDecode.default(raw);
-  let pcm = decoded.getChannelData(0);
-  if (decoded.sampleRate !== SAMPLE_RATE) {
-    pcm = ttsUtils ? ttsUtils.resample(pcm, decoded.sampleRate, SAMPLE_RATE)
-      : (() => {
-        const ratio = decoded.sampleRate / SAMPLE_RATE;
-        const out = new Float32Array(Math.round(pcm.length / ratio));
-        for (let i = 0; i < out.length; i++) out[i] = pcm[Math.floor(i * ratio)];
-        return out;
-      })();
-  }
-
-  const embedding = await serverTTS.encodeVoiceAudio(pcm);
-  voiceEmbeddingCache.set(voiceId, embedding);
-  return embedding;
-}
-
-// Convert Float32Array PCM to WAV buffer
-function pcmToWav(samples, sampleRate = SAMPLE_RATE) {
-  const numSamples = samples.length;
-  const numChannels = 1;
-  const bitsPerSample = 16;
-  const byteRate = sampleRate * numChannels * bitsPerSample / 8;
-  const blockAlign = numChannels * bitsPerSample / 8;
-  const dataSize = numSamples * blockAlign;
-  const buf = Buffer.alloc(44 + dataSize);
-
-  buf.write('RIFF', 0); buf.writeUInt32LE(36 + dataSize, 4);
-  buf.write('WAVE', 8); buf.write('fmt ', 12);
-  buf.writeUInt32LE(16, 16); buf.writeUInt16LE(1, 20);
-  buf.writeUInt16LE(numChannels, 22); buf.writeUInt32LE(sampleRate, 24);
-  buf.writeUInt32LE(byteRate, 28); buf.writeUInt16LE(blockAlign, 32);
-  buf.writeUInt16LE(bitsPerSample, 34); buf.write('data', 36);
-  buf.writeUInt32LE(dataSize, 40);
-
-  for (let i = 0; i < numSamples; i++) {
-    const s = Math.max(-1, Math.min(1, samples[i]));
-    buf.writeInt16LE(Math.round(s * 32767), 44 + i * 2);
-  }
-  return buf;
-}
-
-function getSttOptions() {
-  if (process.env.PORTABLE_EXE_DIR) {
-    return { cacheDir: path.join(process.env.PORTABLE_EXE_DIR, 'models') };
-  }
-  if (process.env.PORTABLE_DATA_DIR) {
-    return { cacheDir: path.join(process.env.PORTABLE_DATA_DIR, 'models') };
-  }
-  return {};
-}
-
-async function getEmbeddingForVoice(voiceId) {
-  if (voiceId && voiceId !== 'default') {
-    const emb = await getVoiceEmbedding(voiceId);
-    if (emb) return emb;
-  }
-  // Fall back to first available voice file
-  for (const dir of VOICE_DIRS) {
-    for (const ext of AUDIO_EXTENSIONS) {
-      const entries = fs.existsSync(dir) ? fs.readdirSync(dir).filter(f => f.endsWith(ext)) : [];
-      if (entries.length) {
-        const emb = await getVoiceEmbedding('custom_' + entries[0].replace(new RegExp(`\\${ext}$`), ''));
-        if (emb) return emb;
-      }
-    }
-  }
-  return null;
-}
-
-async function synthesize(text, voiceId) {
-  if (isOnnxApi) {
-    // Node.js ONNX TTS - no Python required
-    const modelDir = getModelDir();
-    const embedding = await getEmbeddingForVoice(voiceId);
-    if (!embedding) throw new Error('No voice file available for TTS - add a WAV file to ~/voices/');
-    const pcm = await serverTTS.synthesize(text, embedding, modelDir);
-    return pcmToWav(pcm);
-  }
-
-  if (isPocketApi) {
-    // Old server-tts.js with pocket-tts sidecar
-    return serverTTS.synthesize(text, voiceId, VOICE_DIRS);
-  }
-
-  throw new Error('No TTS backend available');
-}
-
-async function* synthesizeStream(text, voiceId) {
-  if (isOnnxApi) {
-    const modelDir = getModelDir();
-    const embedding = await getEmbeddingForVoice(voiceId);
-    if (!embedding) throw new Error('No voice file available for TTS - add a WAV file to ~/voices/');
-    const pcm = await serverTTS.synthesize(text, embedding, modelDir);
-    yield pcmToWav(pcm);
-    return;
-  }
-
-  if (isPocketApi) {
-    for await (const chunk of serverTTS.synthesizeStream(text, voiceId, VOICE_DIRS)) {
-      yield chunk;
-    }
-    return;
-  }
-
-  throw new Error('No TTS backend available');
-}
-
-function transcribe(audioBuffer) {
-  if (!serverSTT) throw new Error('STT not available');
-  return serverSTT.transcribe(audioBuffer, getSttOptions());
-}
-
-function getSTT() {
-  if (!serverSTT) throw new Error('STT not available');
-  return serverSTT.getSTT(getSttOptions());
-}
-
-function getVoices() {
-  const seen = new Set();
-  const custom = [];
-  for (const dir of VOICE_DIRS) {
-    for (const v of scanVoiceDir(dir)) {
-      if (seen.has(v.id)) continue;
-      seen.add(v.id);
-      custom.push(v);
-    }
-  }
-  // Include built-in voices from old server-tts if available
-  if (isPocketApi) {
-    const upstream = serverTTS.getVoices(VOICE_DIRS).filter(v => v.isCustom);
-    for (const v of upstream) {
-      if (!seen.has(v.id)) { seen.add(v.id); custom.push(v); }
-    }
-  }
-  return [...POCKET_TTS_VOICES, ...custom];
-}
-
-function getStatus() {
-  const sttStatus = serverSTT ? serverSTT.getStatus() : { ready: false, loading: false, error: 'STT unavailable' };
-  const ttsBackend = isOnnxApi ? 'onnx-node' : isPocketApi ? 'pocket-tts' : 'none';
-  return {
-    sttReady: sttStatus.ready,
-    ttsReady: isOnnxApi || isPocketApi,
-    sttLoading: sttStatus.loading,
-    ttsLoading: false,
-    sttError: sttStatus.error,
-    ttsError: (!isOnnxApi && !isPocketApi) ? 'No TTS backend available' : null,
-    ttsBackend,
-  };
-}
-
-function preloadTTS() {
-  if (isOnnxApi) {
-    // Pre-load ONNX models in background
-    const modelDir = getModelDir();
-    if (serverTTS.loadModels) {
-      serverTTS.loadModels(modelDir).catch(e => console.warn('[TTS] ONNX preload failed:', e.message));
-    }
-  } else if (isPocketApi && serverTTS.preload) {
-    serverTTS.preload(null, {});
-  }
-}
-
-function ttsCacheKey(text, voiceId) {
-  return isPocketApi && serverTTS.ttsCacheKey ? serverTTS.ttsCacheKey(text, voiceId) : null;
-}
-
-function ttsCacheGet(key) {
-  return isPocketApi && serverTTS.ttsCacheGet ? serverTTS.ttsCacheGet(key) : null;
-}
-
-function splitSentences(text) {
-  if (isPocketApi && serverTTS.splitSentences) return serverTTS.splitSentences(text);
-  return text.match(/[^.!?]+[.!?]*/g)?.map(s => s.trim()).filter(Boolean) || [text];
-}
-
-export { transcribe, synthesize, synthesizeStream, getSTT, getStatus, getVoices, preloadTTS, ttsCacheKey, ttsCacheGet, splitSentences };
+const speech = require('webtalk/speech');
+export const { transcribe, synthesize, synthesizeStream, getSTT, getStatus, getVoices, preloadTTS, ttsCacheKey, ttsCacheGet, splitSentences } = speech;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentgui",
-  "version": "1.0.301",
+  "version": "1.0.303",
   "description": "Multi-agent ACP client with real-time communication",
   "type": "module",
   "main": "server.js",
@@ -30,7 +30,7 @@
     "fsbrowse": "^0.2.18",
     "google-auth-library": "^10.5.0",
     "onnxruntime-node": "^1.24.1",
-    "webtalk": "^1.0.17",
+    "webtalk": "file:../webtalk",
     "ws": "^8.14.2"
   },
   "overrides": {

package/static/index.html

@@ -2257,35 +2257,28 @@
     }
 
     /* --- Inline Tool Result (nested inside tool_use) --- */
-    .tool-result-inline {
-      border-top: 1px solid #bbf7d0;
-      overflow: hidden;
+    /* Tool Result Blocks - Folded by default with status icons */
+    .folded-tool.block-type-tool_result {
+      margin-bottom: 0.375rem;
     }
-    html.dark .tool-result-inline { border-top-color: #166534; }
-    .tool-result-status {
-      display: flex;
-      align-items: center;
-      gap: 0.375rem;
-      padding: 0.3rem 0.625rem;
-      font-size: 0.75rem;
-      line-height: 1.3;
-      cursor: default;
-      user-select: none;
+    .folded-tool.folded-tool-success {
+      background: #f0fdf4;
+    }
+    html.dark .folded-tool.folded-tool-success {
+      background: #0a1f0f;
+    }
+    .folded-tool.folded-tool-success > .folded-tool-bar {
+      background: #dcfce7;
+    }
+    html.dark .folded-tool.folded-tool-success > .folded-tool-bar {
+      background: #0f2b1a;
+    }
+    .folded-tool.folded-tool-success > .folded-tool-bar::before {
+      color: #16a34a;
+    }
+    html.dark .folded-tool.folded-tool-success > .folded-tool-bar::before {
+      color: #4ade80;
     }
-    .tool-result-status::-webkit-details-marker { display: none; }
-    .tool-result-status::marker { display: none; content: ''; }
-    .tool-result-inline > .folded-tool-body { border-top: 1px solid #bbf7d0; }
-    html.dark .tool-result-inline > .folded-tool-body { border-top-color: #166534; }
-    .tool-result-error { background: #fef2f2; border-top-color: #fecaca; }
-    html.dark .tool-result-error { background: #1f0a0a; border-top-color: #991b1b; }
-    .tool-result-error .folded-tool-icon { color: #dc2626; }
-    html.dark .tool-result-error .folded-tool-icon { color: #f87171; }
-    .tool-result-error .folded-tool-name { color: #991b1b; }
-    html.dark .tool-result-error .folded-tool-name { color: #fca5a5; }
-    .tool-result-error .folded-tool-desc { color: #b91c1c; }
-    html.dark .tool-result-error .folded-tool-desc { color: #f87171; }
-    .tool-result-error > .folded-tool-body { border-top-color: #fecaca; }
-    html.dark .tool-result-error > .folded-tool-body { border-top-color: #991b1b; }
 
     /* --- Consecutive Block Joining --- */
     .streaming-blocks > * + *,

package/static/js/streaming-renderer.js

@@ -1210,35 +1210,31 @@ class StreamingRenderer {
     const isError = block.is_error || false;
     const content = block.content || '';
     const contentStr = typeof content === 'string' ? content : JSON.stringify(content, null, 2);
-    const parentIsOpen = context.parentIsOpen !== undefined ? context.parentIsOpen : true;
 
-    const wrapper = document.createElement('div');
-    wrapper.className = 'tool-result-inline' + (isError ? ' tool-result-error' : ' tool-result-success');
-    wrapper.dataset.eventType = 'tool_result';
-    if (block.tool_use_id) wrapper.dataset.toolUseId = block.tool_use_id;
-    wrapper.classList.add(this._getBlockTypeClass('tool_result'));
+    const details = document.createElement('details');
+    details.className = 'folded-tool' + (isError ? ' folded-tool-error' : ' folded-tool-success');
+    details.dataset.eventType = 'tool_result';
+    if (block.tool_use_id) details.dataset.toolUseId = block.tool_use_id;
+    details.classList.add(this._getBlockTypeClass('tool_result'));
 
-    const header = document.createElement('div');
-    header.className = 'tool-result-status';
+    const summary = document.createElement('summary');
+    summary.className = 'folded-tool-bar';
     const iconSvg = isError
       ? '<svg viewBox="0 0 20 20" fill="currentColor"><path fill-rule="evenodd" d="M10 18a8 8 0 100-16 8 8 0 000 16zM8.707 7.293a1 1 0 00-1.414 1.414L8.586 10l-1.293 1.293a1 1 0 101.414 1.414L10 11.414l1.293 1.293a1 1 0 001.414-1.414L11.414 10l1.293-1.293a1 1 0 00-1.414-1.414L10 8.586 8.707 7.293z" clip-rule="evenodd"/></svg>'
       : '<svg viewBox="0 0 20 20" fill="currentColor"><path fill-rule="evenodd" d="M10 18a8 8 0 100-16 8 8 0 000 16zm3.707-9.293a1 1 0 00-1.414-1.414L9 10.586 7.707 9.293a1 1 0 00-1.414 1.414l2 2a1 1 0 001.414 0l4-4z" clip-rule="evenodd"/></svg>';
-    header.innerHTML = `
+    summary.innerHTML = `
       <span class="folded-tool-icon">${iconSvg}</span>
       <span class="folded-tool-name">${isError ? 'Error' : 'Success'}</span>
     `;
-    wrapper.appendChild(header);
+    details.appendChild(summary);
 
     const renderedContent = StreamingRenderer.renderSmartContentHTML(contentStr, this.escapeHtml.bind(this), true);
     const body = document.createElement('div');
     body.className = 'folded-tool-body';
-    if (!parentIsOpen) {
-      body.style.display = 'none';
-    }
     body.innerHTML = renderedContent;
-    wrapper.appendChild(body);
+    details.appendChild(body);
 
-    return wrapper;
+    return details;
   }
 
   /**