myagent-ai 1.16.2 → 1.16.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myagent-ai",
-  "version": "1.16.2",
+  "version": "1.16.4",
   "description": "本地桌面端执行型AI助手 - Open Interpreter 风格 | Local Desktop Execution-Oriented AI Assistant",
   "main": "main.py",
   "bin": {

package/skills/chromedev_mcp.py

@@ -124,6 +124,16 @@ class MCPClient:
         if self._slim:
             args.append("--slim")
 
+        # [v1.16.2] 自动检测可用的浏览器，支持 chromium 等替代方案
+        browser_path = self._detect_browser()
+        env = {**os.environ, "NO_COLOR": "1"}
+        if browser_path:
+            # 优先使用 CHROME_PATH 环境变量传递给 chrome-devtools-mcp
+            env["CHROME_PATH"] = browser_path
+            logger.info(f"使用浏览器: {browser_path}")
+        else:
+            logger.warning("未找到 Chrome/Chromium 浏览器，尝试使用默认路径")
+
         logger.info(f"启动 chrome-devtools-mcp: {' '.join(args)}")
 
         try:
@@ -135,7 +145,7 @@ class MCPClient:
                 stdin=subprocess.PIPE,
                 stdout=subprocess.PIPE,
                 stderr=subprocess.DEVNULL,
-                env={**os.environ, "NO_COLOR": "1"},
+                env=env,
             )
 
             # 启动读取线程
@@ -157,6 +167,50 @@ class MCPClient:
             self._cleanup()
             return False
 
+    @staticmethod
+    def _detect_browser() -> Optional[str]:
+        """[v1.16.2] 自动检测可用的浏览器（Chrome/Chromium/Brave 等）。
+
+        检测顺序:
+          1. 环境变量 CHROME_PATH / BROWSER_PATH
+          2. PATH 中的常见浏览器命令
+          3. Linux 常见安装路径
+        """
+        # 1. 环境变量
+        for env_key in ("CHROME_PATH", "BROWSER_PATH"):
+            env_val = os.environ.get(env_key, "").strip()
+            if env_val and os.path.isfile(env_val):
+                return env_val
+
+        # 2. PATH 中的浏览器命令
+        path_candidates = [
+            "google-chrome", "google-chrome-stable", "google-chrome-beta",
+            "chromium-browser", "chromium",
+            "brave-browser", "microsoft-edge",
+        ]
+        for cmd in path_candidates:
+            found = shutil.which(cmd)
+            if found:
+                return found
+
+        # 3. Linux 常见安装路径
+        linux_paths = [
+            "/usr/bin/google-chrome",
+            "/usr/bin/google-chrome-stable",
+            "/usr/bin/chromium-browser",
+            "/usr/bin/chromium",
+            "/snap/bin/chromium",
+            "/usr/bin/brave-browser",
+            "/usr/bin/microsoft-edge",
+            "/opt/google/chrome/chrome",
+            "/opt/microsoft/msedge/msedge",
+        ]
+        for p in linux_paths:
+            if os.path.isfile(p) and os.access(p, os.X_OK):
+                return p
+
+        return None
+
     def _read_stdout(self):
         """持续读取 MCP Server 的 stdout"""
         if not self._process or not self._process.stdout:

package/web/api_server.py

@@ -1093,10 +1093,19 @@ class ApiServer:
 
             # ── 尝试 faster-whisper ──
             try:
+                # [v1.16.3] 抑制 HuggingFace 未认证警告和 pydub ffmpeg 警告
+                import warnings
+                warnings.filterwarnings("ignore", message=".*HF_TOKEN.*", category=UserWarning)
+                warnings.filterwarnings("ignore", message=".*huggingface_hub.*token.*", category=UserWarning)
+                warnings.filterwarnings("ignore", message=".*ffmpeg or avconv.*", category=RuntimeWarning)
+                # 抑制 HF Hub telemetry 和进度条
+                os.environ.setdefault("HF_HUB_DISABLE_TELEMETRY", "1")
+                os.environ.setdefault("HF_HUB_DISABLE_PROGRESS_BARS", "1")
+                os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
+
                 from faster_whisper import WhisperModel
                 whisper_model = getattr(self, '_whisper_model', None)
                 if whisper_model is None:
-                    import os
                     model_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'models', 'whisper')
                     # 使用 tiny 模型（最轻量，~39MB），CPU int8 量化
                     self._whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8",

package/web/ui/chat/chat.css

@@ -47,7 +47,7 @@ button{font:inherit;cursor:pointer;border:none;background:none;color:inherit}
 input,textarea,select{font:inherit}
 
 /* ── Layout ── */
-.app{position:fixed;top:0;left:0;width:100vw;height:100vh;display:flex;overflow:hidden;background:var(--bg)}
+.app{position:fixed;top:0;left:0;width:100vw;height:100vh;height:100dvh;display:flex;overflow:hidden;background:var(--bg)}
 
 /* Layout panels */
 .sidebar,
@@ -720,9 +720,21 @@ input,textarea,select{font:inherit}
   line-height: 1.6;
   white-space: pre-wrap;
   word-break: break-word;
-  min-height: 24px;
+  min-height: 48px;
   max-height: 120px;
-  overflow-y: auto;
+  width: 100%;
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  padding: 8px 10px;
+  background: var(--bg);
+  resize: none;
+  outline: none;
+  font-family: inherit;
+  box-sizing: border-box;
+  transition: border-color 0.15s;
+}
+.voice-preview-text:focus {
+  border-color: var(--accent);
 }
 .voice-preview-actions {
   display: flex;
@@ -2145,7 +2157,7 @@ body.popout-mode #popoutBtn{display:none !important}
   /* ── Input Area ── */
   .input-area{
     padding:10px 12px;
-    padding-bottom: max(10px, env(safe-area-inset-bottom));
+    padding-bottom: max(12px, env(safe-area-inset-bottom));
   }
   .input-wrapper{
     max-width:100%;

package/web/ui/chat/chat.js

@@ -25,25 +25,26 @@
 })();
 
 // ── Mobile keyboard / visualViewport handling ──
+// [v1.16.2] CSS 已使用 100dvh 动态视口高度，浏览器会自动处理键盘弹出时的视口收缩。
+// 此处仅作为兼容性补充：对不支持 100dvh 的旧浏览器，使用 visualViewport API 调整。
 function onAllScriptsLoaded() {
   if (!window.visualViewport) return;
 
-  var inputArea = document.getElementById('inputBox');
-  if (!inputArea) return;
+  // 检测是否支持 dvh（现代浏览器都支持）
+  var testEl = document.createElement('div');
+  var supportsDvh = CSS.supports && CSS.supports('height', '100dvh');
+
+  // 如果支持 dvh，CSS 已经处理了，不需要 JS 干预
+  if (supportsDvh) return;
+
+  var app = document.querySelector('.app');
+  if (!app) return;
 
   var vv = window.visualViewport;
 
-  // When the virtual keyboard appears (viewport shrinks), the input area
-  // needs extra bottom padding so the send button stays above the keyboard.
   function onViewportResize() {
-    var diff = window.innerHeight - vv.height;
-    if (diff > 100) {
-      // Keyboard is open — add padding equal to keyboard height
-      inputArea.style.marginBottom = diff + 'px';
-    } else {
-      // Keyboard is closed
-      inputArea.style.marginBottom = '0px';
-    }
+    // 对不支持 dvh 的浏览器，用 visualViewport.height 手动设置高度
+    app.style.height = vv.height + 'px';
   }
 
   vv.addEventListener('resize', onViewportResize);

package/web/ui/chat/chat_container.html

@@ -4,7 +4,7 @@
 <meta charset="UTF-8">
 <meta name="viewport" content="width=device-width, initial-scale=1.0, viewport-fit=cover">
 <title>MyAgent - AI 助手</title>
-<link rel="stylesheet" href="chat.css?v=7">
+<link rel="stylesheet" href="chat.css?v=8">
 </head>
 <body>
 <div class="app">
@@ -177,7 +177,7 @@
           </div>
           <div class="voice-preview" id="voicePreview" style="display:none">
             <div class="voice-preview-label">语音输入 · <span id="voicePreviewHint">识别中...</span></div>
-            <div class="voice-preview-text" id="voicePreviewText"></div>
+            <textarea class="voice-preview-text" id="voicePreviewText" placeholder="识别结果..." rows="2"></textarea>
             <div class="voice-preview-actions">
               <button class="voice-preview-cancel" onclick="cancelVoicePreview()">取消</button>
               <button class="voice-preview-send" id="voicePreviewSend" onclick="sendVoiceMessage()">发送</button>
@@ -275,6 +275,6 @@
 <div id="groupModalContainer"></div>
 
 <!-- Load JS scripts (no HTML fragments to fetch) -->
-<script src="chat.js?v=8"></script>
+<script src="chat.js?v=9"></script>
 </body>
 </html>

package/web/ui/chat/chat_main.js

@@ -4590,7 +4590,7 @@ var VoiceInput = {
     // 显示预览区域
     if (voiceArea) voiceArea.style.display = 'none';
     if (voicePreview) voicePreview.style.display = 'block';
-    if (previewText) previewText.textContent = '识别中...';
+    if (previewText) previewText.value = '识别中...';
     if (previewHint) previewHint.textContent = '正在发送到本地STT引擎';
     if (previewSend) previewSend.disabled = true;
 
@@ -4614,11 +4614,11 @@ var VoiceInput = {
       if (data && data.text && data.text.trim()) {
         this.rawText = data.text.trim();
         this._sttEngine = data.engine || 'unknown';
-        if (previewText) previewText.textContent = this.rawText;
-        if (previewHint) previewHint.textContent = '已识别 (' + (this._sttEngine || 'local') + ')';
+        if (previewText) previewText.value = this.rawText;
+        if (previewHint) previewHint.textContent = '已识别 (' + (this._sttEngine || 'local') + ')，可编辑后发送';
       } else if (data && data.error) {
         if (previewHint) previewHint.textContent = '识别失败';
-        if (previewText) previewText.textContent = data.error;
+        if (previewText) previewText.value = data.error;
         console.warn('Voice STT error:', data.error);
         // 如果没有STT引擎，给出提示
         if (data.available === false) {
@@ -4629,12 +4629,12 @@ var VoiceInput = {
         }
       } else {
         if (previewHint) previewHint.textContent = '未识别到文字';
-        if (previewText) previewText.textContent = '未识别到文字，请重试';
+        if (previewText) previewText.value = '未识别到文字，请重试';
       }
     } catch (e) {
       console.error('Voice STT request error:', e);
       if (previewHint) previewHint.textContent = '网络错误';
-      if (previewText) previewText.textContent = 'STT请求失败，请重试';
+      if (previewText) previewText.value = 'STT请求失败，请重试';
     }
 
     this._cleanupStream();
@@ -4669,11 +4669,12 @@ var VoiceInput = {
     this.audioChunks = [];
   },
 
-  /** Send the voice text as a message（直接发送原始STT文本，由后端usersays_correct纠错） */
+  /** Send the voice text as a message（支持用户编辑识别结果后再发送） */
   sendMessage: function() {
-    if (!this.rawText || !this.rawText.trim()) return;
-
-    var text = this.rawText.trim();
+    // 从 textarea 读取编辑后的文本
+    var previewText = document.getElementById('voicePreviewText');
+    var text = previewText ? previewText.value.trim() : '';
+    if (!text) return;
 
     // 切回文本模式
     this.switchMode('text');