Improve camera drawer UX: layout, device switching, default selection

- Grid layout for source controls: camera + switch on top row (equal
  width), upload full-width below; preview area uses aspect-ratio 4/3
- "使用摄像头" no longer changes label text; uses is-active highlight only
- "切换摄像头" now actually cycles through real video devices via
  enumerateDevices(); falls back to facingMode flip on mobile/single-cam;
  button disabled until camera is active and ≥2 devices are detected
- Default-selects first sample image when drawer opens so "拍摄完成"
  is immediately pressable without requiring a capture/upload first
- Fix: camera prompt text is now locked to CAMERA_STATE_PROMPTS[state]
  and no longer overwritten by assistant replies

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

examples/webpage/app.js

@@ -148,6 +148,8 @@ const state = {
   cameraStream: null,
   cameraActive: false,
   cameraFacing: "environment",
+  videoDevices: [],
+  videoDeviceIndex: 0,
   pendingImage: null,
   samplesRendered: false,
 
@@ -255,6 +257,7 @@ function syncCameraDrawer(value) {
     els.cameraState.textContent = `State ${value}`;
     els.cameraQuestion.textContent = prompt;
     renderSampleThumbnails();
+    selectDefaultImage();
   } else {
     els.cameraState.textContent = "State -";
     els.cameraQuestion.textContent = "";
@@ -912,15 +915,27 @@ function setPendingImage(payload) {
   setCameraButtonEnabled();
 }
 
-async function startCamera() {
+async function refreshVideoDevices() {
+  try {
+    const devices = await navigator.mediaDevices.enumerateDevices();
+    state.videoDevices = devices.filter((d) => d.kind === "videoinput");
+  } catch (_) {
+    state.videoDevices = [];
+  }
+}
+
+async function startCamera(deviceId) {
   if (!navigator.mediaDevices || !navigator.mediaDevices.getUserMedia) {
     addWsLog("system", "getUserMedia not available in this browser");
     return;
   }
   stopCameraStream();
+  const video = deviceId
+    ? { deviceId: { exact: deviceId } }
+    : { facingMode: state.cameraFacing };
   try {
     state.cameraStream = await navigator.mediaDevices.getUserMedia({
-      video: { facingMode: state.cameraFacing },
+      video,
       audio: false,
     });
   } catch (err) {
@@ -937,12 +952,33 @@ async function startCamera() {
   state.pendingImage = null;
   setPreviewMode("camera");
   els.cameraStartBtn.classList.add("is-active");
-  els.cameraStartBtn.textContent = "重新拍摄";
-  els.cameraFlipBtn.hidden = false;
   clearSampleSelection();
+
+  // Device labels become available only after permission is granted.
+  await refreshVideoDevices();
+  if (deviceId) {
+    const idx = state.videoDevices.findIndex((d) => d.deviceId === deviceId);
+    if (idx >= 0) state.videoDeviceIndex = idx;
+  }
+  els.cameraFlipBtn.disabled = state.videoDevices.length < 2;
   setCameraButtonEnabled();
 }
 
+// Cycle to the next available camera, or flip facingMode when devices are
+// unlabeled (e.g. before permission, or on mobile with front/back only).
+async function switchCamera() {
+  if (!state.cameraActive) return;
+  if (state.videoDevices.length > 1) {
+    state.videoDeviceIndex =
+      (state.videoDeviceIndex + 1) % state.videoDevices.length;
+    await startCamera(state.videoDevices[state.videoDeviceIndex].deviceId);
+  } else {
+    state.cameraFacing =
+      state.cameraFacing === "environment" ? "user" : "environment";
+    await startCamera();
+  }
+}
+
 function stopCameraStream() {
   if (state.cameraStream) {
     state.cameraStream.getTracks().forEach((track) => track.stop());
@@ -951,8 +987,7 @@ function stopCameraStream() {
   els.cameraVideo.srcObject = null;
   state.cameraActive = false;
   els.cameraStartBtn.classList.remove("is-active");
-  els.cameraStartBtn.textContent = "使用摄像头";
-  els.cameraFlipBtn.hidden = true;
+  els.cameraFlipBtn.disabled = true;
 }
 
 function captureFromCamera() {
@@ -1037,6 +1072,16 @@ function resetCameraInput() {
   setCameraButtonEnabled();
 }
 
+// Pre-select the first sample image so "拍摄完成" is immediately pressable when
+// the drawer opens, without requiring the user to capture or pick first.
+function selectDefaultImage() {
+  if (state.pendingImage || state.cameraActive) return;
+  const first = els.cameraSamples.querySelector(".camera-drawer__sample");
+  if (first && SAMPLE_IMAGES[0]) {
+    selectSampleImage(SAMPLE_IMAGES[0].src, first);
+  }
+}
+
 function sendImage(payload, text) {
   if (!payload) return false;
   if (!state.ws || state.ws.readyState !== WebSocket.OPEN) return false;
@@ -1415,9 +1460,7 @@ els.cameraStartBtn.addEventListener("click", () => {
 });
 
 els.cameraFlipBtn.addEventListener("click", () => {
-  state.cameraFacing =
-    state.cameraFacing === "environment" ? "user" : "environment";
-  if (state.cameraActive) startCamera();
+  switchCamera();
 });
 
 els.cameraUpload.addEventListener("change", (event) => {