deepseek-computer-use 0.1.1 → 0.2.0

package/native/wgcfw/Program.cs

@@ -1,249 +1,249 @@
-using System;
-using System.Drawing;
-using System.Drawing.Imaging;
-using System.Runtime.InteropServices;
-using System.Threading;
-using SharpDX.Direct3D;
-using SharpDX.Direct3D11;
-using SharpDX.DXGI;
-using Windows.Graphics.Capture;
-using Windows.Graphics.DirectX;
-using Windows.Graphics.DirectX.Direct3D11;
-using D3D11Device = SharpDX.Direct3D11.Device;
-using D3D11Texture2D = SharpDX.Direct3D11.Texture2D;
-using DXGIDevice = SharpDX.DXGI.Device;
-
-namespace DshCuCapture
-{
-    /// Windows.Graphics.Capture for .NET Framework 4.8 - the same API Codex uses,
-    /// with no .NET 8 runtime dependency (Framework 4.8 ships with Windows 10/11).
-    internal static class Program
-    {
-        [ComImport]
-        [Guid("A9B3D012-3DF2-4EE3-B8D1-8695F457D3C1")]
-        [InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
-        private interface IDirect3DDxgiInterfaceAccess
-        {
-            [PreserveSig] int GetInterface(ref Guid iid, out IntPtr result);
-        }
-
-        [DllImport("d3d11.dll", ExactSpelling = true)]
-        private static extern int CreateDirect3D11DeviceFromDXGIDevice(IntPtr dxgiDevice, out IntPtr graphicsDevice);
-
-        [DllImport("combase.dll", CharSet = CharSet.Unicode)]
-        private static extern int WindowsCreateString(string source, int length, out IntPtr hstring);
-
-        [DllImport("combase.dll")]
-        private static extern int RoGetActivationFactory(IntPtr activatableClassId, ref Guid iid, out IntPtr factory);
-
-        [DllImport("combase.dll")]
-        private static extern int WindowsDeleteString(IntPtr hstring);
-
-        [DllImport("kernel32.dll", EntryPoint = "RtlMoveMemory")]
-        private static extern void CopyMemory(IntPtr destination, IntPtr source, int length);
-
-        [UnmanagedFunctionPointer(CallingConvention.StdCall)]
-        private delegate int CreateForWindowDelegate(IntPtr thisPtr, IntPtr hwnd, ref Guid iid, out IntPtr item);
-        private delegate int CreateForMonitorDelegate(IntPtr thisPtr, IntPtr hmon, ref Guid iid, out IntPtr item);
-
-        [DllImport("user32.dll")] private static extern IntPtr MonitorFromPoint(POINT pt, uint flags);
-        [StructLayout(LayoutKind.Sequential)] private struct POINT { public int X, Y; }
-
-        private static bool _suppressBorder = true;
-
-        private sealed class Slot
-        {
-            public Bitmap Bitmap;
-            public Exception Error;
-        }
-
-        private static int Main(string[] args)
-        {
-            long hwnd = 0;
-            bool monitor = false;
-            bool redactAll = false;
-            bool suppressBorder = true;
-            var redactRects = new System.Collections.Generic.List<System.Drawing.Rectangle>();
-            string output = "capture.png";
-            int timeoutMs = 2000;
-            for (int i = 0; i < args.Length - 1; i++)
-            {
-                if (args[i] == "--redact")
-                {
-                    string[] parts = args[i + 1].Split(',');
-                    if (parts.Length == 4)
-                        redactRects.Add(new System.Drawing.Rectangle(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), int.Parse(parts[3])));
-                }
-                else if (args[i] == "--hwnd") long.TryParse(args[i + 1], out hwnd);
-                else if (args[i] == "--out") output = args[i + 1];
-                else if (args[i] == "--timeout") int.TryParse(args[i + 1], out timeoutMs);
-            }
-            for (int i = 0; i < args.Length; i++)
-            {
-                if (args[i] == "--monitor") monitor = true;
-                else if (args[i] == "--redact-all") redactAll = true;
-                else if (args[i] == "--keep-border") suppressBorder = false;
-            }
-            if (hwnd == 0 && !monitor) { Console.Error.WriteLine("--hwnd or --monitor is required"); return 2; }
-
-            try
-            {
-                GraphicsCaptureItem item = monitor ? CreateMonitorItem() : CreateItem(new IntPtr(hwnd));
-                using (D3D11Device device = new D3D11Device(DriverType.Hardware, DeviceCreationFlags.BgraSupport))
-                using (DXGIDevice dxgi = device.QueryInterface<DXGIDevice>())
-                {
-                    int hr = CreateDirect3D11DeviceFromDXGIDevice(dxgi.NativePointer, out IntPtr winrtDevicePtr);
-                    if (hr < 0) { Console.Error.WriteLine("CreateDirect3D11DeviceFromDXGIDevice hr=0x" + hr.ToString("X8")); return 3; }
-                    IDirect3DDevice winrtDevice;
-                    try { winrtDevice = (IDirect3DDevice)Marshal.GetObjectForIUnknown(winrtDevicePtr); }
-                    finally { Marshal.Release(winrtDevicePtr); }
-
-                    _suppressBorder = suppressBorder;
-                    var slot = new Slot();
-                    using (var pool = Direct3D11CaptureFramePool.CreateFreeThreaded(
-                        winrtDevice, DirectXPixelFormat.B8G8R8A8UIntNormalized, 2, item.Size))
-                    using (var session = pool.CreateCaptureSession(item))
-                    using (var done = new ManualResetEventSlim(false))
-                    {
-                        try { session.IsCursorCaptureEnabled = false; } catch { }
-                        // Windows 11 draws a yellow frame around any captured window. Kill it
-                        // (Codex does the same through its capture-border access path) so the
-                        // border never shows up in the picture or on the user's desktop.
-                        if (_suppressBorder) { try { session.IsBorderRequired = false; } catch { } }
-
-                        Direct3D11CaptureFramePool owner = pool;
-                        D3D11Device gpu = device;
-                        pool.FrameArrived += (sender, e) =>
-                        {
-                            if (slot.Bitmap != null) return;
-                            try
-                            {
-                                using (var frame = owner.TryGetNextFrame())
-                                {
-                                    if (frame == null) return;
-                                    var access = (IDirect3DDxgiInterfaceAccess)frame.Surface;
-                                    Guid iid = new Guid("6f15aaf2-d208-4e89-9ab4-489535d34f9c"); // ID3D11Texture2D
-                                    IntPtr texPtr;
-                                    if (access.GetInterface(ref iid, out texPtr) < 0 || texPtr == IntPtr.Zero) return;
-                                    using (var texture = new D3D11Texture2D(texPtr))
-                                    {
-                                        var desc = texture.Description;
-                                        var stagingDesc = new Texture2DDescription
-                                        {
-                                            Width = desc.Width,
-                                            Height = desc.Height,
-                                            MipLevels = 1,
-                                            ArraySize = 1,
-                                            Format = desc.Format,
-                                            SampleDescription = new SampleDescription(1, 0),
-                                            Usage = ResourceUsage.Staging,
-                                            BindFlags = BindFlags.None,
-                                            CpuAccessFlags = CpuAccessFlags.Read,
-                                            OptionFlags = ResourceOptionFlags.None,
-                                        };
-                                        using (var staging = new D3D11Texture2D(gpu, stagingDesc))
-                                        {
-                                            var context = gpu.ImmediateContext;
-                                            context.CopyResource(texture, staging);
-                                            SharpDX.DataBox box = context.MapSubresource(staging, 0, MapMode.Read, SharpDX.Direct3D11.MapFlags.None);
-                                            try
-                                            {
-                                                Bitmap bmp = new Bitmap(desc.Width, desc.Height, PixelFormat.Format32bppArgb);
-                                                BitmapData data = bmp.LockBits(new Rectangle(0, 0, desc.Width, desc.Height), ImageLockMode.WriteOnly, PixelFormat.Format32bppArgb);
-                                                for (int y = 0; y < desc.Height; y++)
-                                                {
-                                                    IntPtr src = IntPtr.Add(box.DataPointer, y * box.RowPitch);
-                                                    IntPtr dst = IntPtr.Add(data.Scan0, y * data.Stride);
-                                                    CopyMemory(dst, src, desc.Width * 4);
-                                                }
-                                                bmp.UnlockBits(data);
-                                                slot.Bitmap = bmp;
-                                            }
-                                            finally { context.UnmapSubresource(staging, 0); }
-                                        }
-                                    }
-                                }
-                            }
-                            catch (Exception ex) { slot.Error = ex; }
-                            finally { done.Set(); }
-                        };
-
-                        session.StartCapture();
-                        done.Wait(timeoutMs);
-                        try { session.Dispose(); } catch { }
-                    }
-
-                    if (slot.Error != null) { Console.Error.WriteLine(slot.Error.GetType().Name + ": " + slot.Error.Message); return 6; }
-                    if (slot.Bitmap == null) { Console.Error.WriteLine("no frame arrived (timeout or empty surface)"); return 4; }
-
-                    if (redactAll || redactRects.Count > 0)
-                    {
-                        using (var g = Graphics.FromImage(slot.Bitmap))
-                        {
-                            g.Clear(redactAll ? Color.Black : Color.Transparent);
-                            foreach (var rect in redactRects) g.FillRectangle(Brushes.Black, rect);
-                        }
-                    }
-                    int w = slot.Bitmap.Width, h = slot.Bitmap.Height;
-                    slot.Bitmap.Save(output, ImageFormat.Png);
-                    slot.Bitmap.Dispose();
-                    Console.WriteLine("wgc " + w + "x" + h + " -> " + output);
-                    return 0;
-                }
-            }
-            catch (Exception ex)
-            {
-                Console.Error.WriteLine(ex.GetType().Name + ": " + ex.Message);
-                return 5;
-            }
-        }
-
-        private static GraphicsCaptureItem CreateMonitorItem()
-        {
-            POINT origin = new POINT { X = 0, Y = 0 };
-            IntPtr monitorHandle = MonitorFromPoint(origin, 2); // MONITOR_DEFAULTTONEAREST
-            const string className = "Windows.Graphics.Capture.GraphicsCaptureItem";
-            WindowsCreateString(className, className.Length, out IntPtr hstr);
-            try
-            {
-                Guid interopIid = new Guid("3628E81B-3CAC-4C60-B7F4-23CE0E0C3356");
-                int hr = RoGetActivationFactory(hstr, ref interopIid, out IntPtr factoryPtr);
-                if (hr < 0) throw new COMException("RoGetActivationFactory hr=0x" + hr.ToString("X8"), hr);
-                IntPtr vtable = Marshal.ReadIntPtr(factoryPtr);
-                IntPtr fn = Marshal.ReadIntPtr(vtable, 4 * IntPtr.Size); // CreateForMonitor
-                var createForMonitor = (CreateForMonitorDelegate)Marshal.GetDelegateForFunctionPointer(fn, typeof(CreateForMonitorDelegate));
-                Guid itemIid = new Guid("79C3F95B-31F7-4EC2-A464-632EF5D30760");
-                int hrItem = createForMonitor(factoryPtr, monitorHandle, ref itemIid, out IntPtr itemPtr);
-                Marshal.Release(factoryPtr);
-                if (hrItem < 0) throw new COMException("CreateForMonitor hr=0x" + hrItem.ToString("X8"), hrItem);
-                try { return (GraphicsCaptureItem)Marshal.GetObjectForIUnknown(itemPtr); }
-                finally { Marshal.Release(itemPtr); }
-            }
-            finally { WindowsDeleteString(hstr); }
-        }
-
-        private static GraphicsCaptureItem CreateItem(IntPtr hwnd)
-        {
-            const string className = "Windows.Graphics.Capture.GraphicsCaptureItem";
-            WindowsCreateString(className, className.Length, out IntPtr hstr);
-            try
-            {
-                Guid interopIid = new Guid("3628E81B-3CAC-4C60-B7F4-23CE0E0C3356");
-                int hr = RoGetActivationFactory(hstr, ref interopIid, out IntPtr factoryPtr);
-                if (hr < 0) throw new COMException("RoGetActivationFactory hr=0x" + hr.ToString("X8"), hr);
-
-                IntPtr vtable = Marshal.ReadIntPtr(factoryPtr);
-                IntPtr fn = Marshal.ReadIntPtr(vtable, 3 * IntPtr.Size);
-                var createForWindow = (CreateForWindowDelegate)Marshal.GetDelegateForFunctionPointer(fn, typeof(CreateForWindowDelegate));
-
-                Guid itemIid = new Guid("79C3F95B-31F7-4EC2-A464-632EF5D30760");
-                int hrItem = createForWindow(factoryPtr, hwnd, ref itemIid, out IntPtr itemPtr);
-                Marshal.Release(factoryPtr);
-                if (hrItem < 0) throw new COMException("CreateForWindow hr=0x" + hrItem.ToString("X8"), hrItem);
-                try { return (GraphicsCaptureItem)Marshal.GetObjectForIUnknown(itemPtr); }
-                finally { Marshal.Release(itemPtr); }
-            }
-            finally { WindowsDeleteString(hstr); }
-        }
-    }
-}
+using System;
+using System.Drawing;
+using System.Drawing.Imaging;
+using System.Runtime.InteropServices;
+using System.Threading;
+using SharpDX.Direct3D;
+using SharpDX.Direct3D11;
+using SharpDX.DXGI;
+using Windows.Graphics.Capture;
+using Windows.Graphics.DirectX;
+using Windows.Graphics.DirectX.Direct3D11;
+using D3D11Device = SharpDX.Direct3D11.Device;
+using D3D11Texture2D = SharpDX.Direct3D11.Texture2D;
+using DXGIDevice = SharpDX.DXGI.Device;
+
+namespace DshCuCapture
+{
+    /// Windows.Graphics.Capture for .NET Framework 4.8 - the same API Codex uses,
+    /// with no .NET 8 runtime dependency (Framework 4.8 ships with Windows 10/11).
+    internal static class Program
+    {
+        [ComImport]
+        [Guid("A9B3D012-3DF2-4EE3-B8D1-8695F457D3C1")]
+        [InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
+        private interface IDirect3DDxgiInterfaceAccess
+        {
+            [PreserveSig] int GetInterface(ref Guid iid, out IntPtr result);
+        }
+
+        [DllImport("d3d11.dll", ExactSpelling = true)]
+        private static extern int CreateDirect3D11DeviceFromDXGIDevice(IntPtr dxgiDevice, out IntPtr graphicsDevice);
+
+        [DllImport("combase.dll", CharSet = CharSet.Unicode)]
+        private static extern int WindowsCreateString(string source, int length, out IntPtr hstring);
+
+        [DllImport("combase.dll")]
+        private static extern int RoGetActivationFactory(IntPtr activatableClassId, ref Guid iid, out IntPtr factory);
+
+        [DllImport("combase.dll")]
+        private static extern int WindowsDeleteString(IntPtr hstring);
+
+        [DllImport("kernel32.dll", EntryPoint = "RtlMoveMemory")]
+        private static extern void CopyMemory(IntPtr destination, IntPtr source, int length);
+
+        [UnmanagedFunctionPointer(CallingConvention.StdCall)]
+        private delegate int CreateForWindowDelegate(IntPtr thisPtr, IntPtr hwnd, ref Guid iid, out IntPtr item);
+        private delegate int CreateForMonitorDelegate(IntPtr thisPtr, IntPtr hmon, ref Guid iid, out IntPtr item);
+
+        [DllImport("user32.dll")] private static extern IntPtr MonitorFromPoint(POINT pt, uint flags);
+        [StructLayout(LayoutKind.Sequential)] private struct POINT { public int X, Y; }
+
+        private static bool _suppressBorder = true;
+
+        private sealed class Slot
+        {
+            public Bitmap Bitmap;
+            public Exception Error;
+        }
+
+        private static int Main(string[] args)
+        {
+            long hwnd = 0;
+            bool monitor = false;
+            bool redactAll = false;
+            bool suppressBorder = true;
+            var redactRects = new System.Collections.Generic.List<System.Drawing.Rectangle>();
+            string output = "capture.png";
+            int timeoutMs = 2000;
+            for (int i = 0; i < args.Length - 1; i++)
+            {
+                if (args[i] == "--redact")
+                {
+                    string[] parts = args[i + 1].Split(',');
+                    if (parts.Length == 4)
+                        redactRects.Add(new System.Drawing.Rectangle(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), int.Parse(parts[3])));
+                }
+                else if (args[i] == "--hwnd") long.TryParse(args[i + 1], out hwnd);
+                else if (args[i] == "--out") output = args[i + 1];
+                else if (args[i] == "--timeout") int.TryParse(args[i + 1], out timeoutMs);
+            }
+            for (int i = 0; i < args.Length; i++)
+            {
+                if (args[i] == "--monitor") monitor = true;
+                else if (args[i] == "--redact-all") redactAll = true;
+                else if (args[i] == "--keep-border") suppressBorder = false;
+            }
+            if (hwnd == 0 && !monitor) { Console.Error.WriteLine("--hwnd or --monitor is required"); return 2; }
+
+            try
+            {
+                GraphicsCaptureItem item = monitor ? CreateMonitorItem() : CreateItem(new IntPtr(hwnd));
+                using (D3D11Device device = new D3D11Device(DriverType.Hardware, DeviceCreationFlags.BgraSupport))
+                using (DXGIDevice dxgi = device.QueryInterface<DXGIDevice>())
+                {
+                    int hr = CreateDirect3D11DeviceFromDXGIDevice(dxgi.NativePointer, out IntPtr winrtDevicePtr);
+                    if (hr < 0) { Console.Error.WriteLine("CreateDirect3D11DeviceFromDXGIDevice hr=0x" + hr.ToString("X8")); return 3; }
+                    IDirect3DDevice winrtDevice;
+                    try { winrtDevice = (IDirect3DDevice)Marshal.GetObjectForIUnknown(winrtDevicePtr); }
+                    finally { Marshal.Release(winrtDevicePtr); }
+
+                    _suppressBorder = suppressBorder;
+                    var slot = new Slot();
+                    using (var pool = Direct3D11CaptureFramePool.CreateFreeThreaded(
+                        winrtDevice, DirectXPixelFormat.B8G8R8A8UIntNormalized, 2, item.Size))
+                    using (var session = pool.CreateCaptureSession(item))
+                    using (var done = new ManualResetEventSlim(false))
+                    {
+                        try { session.IsCursorCaptureEnabled = false; } catch { }
+                        // Windows 11 draws a yellow frame around any captured window. Kill it
+                        // (Codex does the same through its capture-border access path) so the
+                        // border never shows up in the picture or on the user's desktop.
+                        if (_suppressBorder) { try { session.IsBorderRequired = false; } catch { } }
+
+                        Direct3D11CaptureFramePool owner = pool;
+                        D3D11Device gpu = device;
+                        pool.FrameArrived += (sender, e) =>
+                        {
+                            if (slot.Bitmap != null) return;
+                            try
+                            {
+                                using (var frame = owner.TryGetNextFrame())
+                                {
+                                    if (frame == null) return;
+                                    var access = (IDirect3DDxgiInterfaceAccess)frame.Surface;
+                                    Guid iid = new Guid("6f15aaf2-d208-4e89-9ab4-489535d34f9c"); // ID3D11Texture2D
+                                    IntPtr texPtr;
+                                    if (access.GetInterface(ref iid, out texPtr) < 0 || texPtr == IntPtr.Zero) return;
+                                    using (var texture = new D3D11Texture2D(texPtr))
+                                    {
+                                        var desc = texture.Description;
+                                        var stagingDesc = new Texture2DDescription
+                                        {
+                                            Width = desc.Width,
+                                            Height = desc.Height,
+                                            MipLevels = 1,
+                                            ArraySize = 1,
+                                            Format = desc.Format,
+                                            SampleDescription = new SampleDescription(1, 0),
+                                            Usage = ResourceUsage.Staging,
+                                            BindFlags = BindFlags.None,
+                                            CpuAccessFlags = CpuAccessFlags.Read,
+                                            OptionFlags = ResourceOptionFlags.None,
+                                        };
+                                        using (var staging = new D3D11Texture2D(gpu, stagingDesc))
+                                        {
+                                            var context = gpu.ImmediateContext;
+                                            context.CopyResource(texture, staging);
+                                            SharpDX.DataBox box = context.MapSubresource(staging, 0, MapMode.Read, SharpDX.Direct3D11.MapFlags.None);
+                                            try
+                                            {
+                                                Bitmap bmp = new Bitmap(desc.Width, desc.Height, PixelFormat.Format32bppArgb);
+                                                BitmapData data = bmp.LockBits(new Rectangle(0, 0, desc.Width, desc.Height), ImageLockMode.WriteOnly, PixelFormat.Format32bppArgb);
+                                                for (int y = 0; y < desc.Height; y++)
+                                                {
+                                                    IntPtr src = IntPtr.Add(box.DataPointer, y * box.RowPitch);
+                                                    IntPtr dst = IntPtr.Add(data.Scan0, y * data.Stride);
+                                                    CopyMemory(dst, src, desc.Width * 4);
+                                                }
+                                                bmp.UnlockBits(data);
+                                                slot.Bitmap = bmp;
+                                            }
+                                            finally { context.UnmapSubresource(staging, 0); }
+                                        }
+                                    }
+                                }
+                            }
+                            catch (Exception ex) { slot.Error = ex; }
+                            finally { done.Set(); }
+                        };
+
+                        session.StartCapture();
+                        done.Wait(timeoutMs);
+                        try { session.Dispose(); } catch { }
+                    }
+
+                    if (slot.Error != null) { Console.Error.WriteLine(slot.Error.GetType().Name + ": " + slot.Error.Message); return 6; }
+                    if (slot.Bitmap == null) { Console.Error.WriteLine("no frame arrived (timeout or empty surface)"); return 4; }
+
+                    if (redactAll || redactRects.Count > 0)
+                    {
+                        using (var g = Graphics.FromImage(slot.Bitmap))
+                        {
+                            g.Clear(redactAll ? Color.Black : Color.Transparent);
+                            foreach (var rect in redactRects) g.FillRectangle(Brushes.Black, rect);
+                        }
+                    }
+                    int w = slot.Bitmap.Width, h = slot.Bitmap.Height;
+                    slot.Bitmap.Save(output, ImageFormat.Png);
+                    slot.Bitmap.Dispose();
+                    Console.WriteLine("wgc " + w + "x" + h + " -> " + output);
+                    return 0;
+                }
+            }
+            catch (Exception ex)
+            {
+                Console.Error.WriteLine(ex.GetType().Name + ": " + ex.Message);
+                return 5;
+            }
+        }
+
+        private static GraphicsCaptureItem CreateMonitorItem()
+        {
+            POINT origin = new POINT { X = 0, Y = 0 };
+            IntPtr monitorHandle = MonitorFromPoint(origin, 2); // MONITOR_DEFAULTTONEAREST
+            const string className = "Windows.Graphics.Capture.GraphicsCaptureItem";
+            WindowsCreateString(className, className.Length, out IntPtr hstr);
+            try
+            {
+                Guid interopIid = new Guid("3628E81B-3CAC-4C60-B7F4-23CE0E0C3356");
+                int hr = RoGetActivationFactory(hstr, ref interopIid, out IntPtr factoryPtr);
+                if (hr < 0) throw new COMException("RoGetActivationFactory hr=0x" + hr.ToString("X8"), hr);
+                IntPtr vtable = Marshal.ReadIntPtr(factoryPtr);
+                IntPtr fn = Marshal.ReadIntPtr(vtable, 4 * IntPtr.Size); // CreateForMonitor
+                var createForMonitor = (CreateForMonitorDelegate)Marshal.GetDelegateForFunctionPointer(fn, typeof(CreateForMonitorDelegate));
+                Guid itemIid = new Guid("79C3F95B-31F7-4EC2-A464-632EF5D30760");
+                int hrItem = createForMonitor(factoryPtr, monitorHandle, ref itemIid, out IntPtr itemPtr);
+                Marshal.Release(factoryPtr);
+                if (hrItem < 0) throw new COMException("CreateForMonitor hr=0x" + hrItem.ToString("X8"), hrItem);
+                try { return (GraphicsCaptureItem)Marshal.GetObjectForIUnknown(itemPtr); }
+                finally { Marshal.Release(itemPtr); }
+            }
+            finally { WindowsDeleteString(hstr); }
+        }
+
+        private static GraphicsCaptureItem CreateItem(IntPtr hwnd)
+        {
+            const string className = "Windows.Graphics.Capture.GraphicsCaptureItem";
+            WindowsCreateString(className, className.Length, out IntPtr hstr);
+            try
+            {
+                Guid interopIid = new Guid("3628E81B-3CAC-4C60-B7F4-23CE0E0C3356");
+                int hr = RoGetActivationFactory(hstr, ref interopIid, out IntPtr factoryPtr);
+                if (hr < 0) throw new COMException("RoGetActivationFactory hr=0x" + hr.ToString("X8"), hr);
+
+                IntPtr vtable = Marshal.ReadIntPtr(factoryPtr);
+                IntPtr fn = Marshal.ReadIntPtr(vtable, 3 * IntPtr.Size);
+                var createForWindow = (CreateForWindowDelegate)Marshal.GetDelegateForFunctionPointer(fn, typeof(CreateForWindowDelegate));
+
+                Guid itemIid = new Guid("79C3F95B-31F7-4EC2-A464-632EF5D30760");
+                int hrItem = createForWindow(factoryPtr, hwnd, ref itemIid, out IntPtr itemPtr);
+                Marshal.Release(factoryPtr);
+                if (hrItem < 0) throw new COMException("CreateForWindow hr=0x" + hrItem.ToString("X8"), hrItem);
+                try { return (GraphicsCaptureItem)Marshal.GetObjectForIUnknown(itemPtr); }
+                finally { Marshal.Release(itemPtr); }
+            }
+            finally { WindowsDeleteString(hstr); }
+        }
+    }
+}

package/package.json

@@ -1,13 +1,14 @@
 {
   "name": "deepseek-computer-use",
-  "version": "0.1.1",
-  "description": "Windows Computer Use for DeepSeek Harness: UIA observation, SendInput actions, Windows.Graphics.Capture — modeled on Codex's window2 API. Community plugin, not an official DeepSeek package.",
+  "version": "0.2.0",
+  "description": "在 DSH 中查看和操作 Windows 桌面软件，支持窗口截图、鼠标键盘与控件操作。 欢迎提交 issues 和 PR。",
   "type": "module",
   "main": "lib/index.js",
   "exports": {
     ".": "./lib/index.js",
     "./cordis.patch.yml": "./cordis.patch.yml",
-    "./package.json": "./package.json"
+    "./package.json": "./package.json",
+    "./client": "./lib/update-client.js"
   },
   "files": [
     "lib",
@@ -20,23 +21,36 @@
     "native/wgcfw/Capture.csproj",
     "native/wgcfw/Program.cs",
     "skills",
+    "assets/cover-whale-girl.png",
     "cordis.patch.yml",
     "README.md",
+    "CHANGELOG.md",
     "LICENSE"
   ],
   "scripts": {
-    "build": "node lib/build.mjs",
+    "build": "node lib/build.mjs && node scripts/build-updates.mjs",
     "build:helper": "node lib/build.mjs",
-    "test": "node test/integration.mjs",
-    "test:all": "node test/integration.mjs && node test/parity.mjs && node test/parity-js.mjs"
+    "test": "node scripts/build-updates.mjs && node --test test/contracts.test.mjs test/bridge.test.mjs && node test/integration.mjs",
+    "test:contracts": "node --test test/contracts.test.mjs test/bridge.test.mjs",
+    "test:harness": "node test/harness.mjs",
+    "test:all": "npm test && node test/parity.mjs && node test/parity-js.mjs"
   },
   "dsh": {
     "bundle": {
       "patch": "./cordis.patch.yml"
+    },
+    "client": {
+      "platform": "web",
+      "inject": [
+        "@deepseek-ai/dsh-client-ui-settings",
+        "@deepseek-ai/dsh-client-ui-layout",
+        "@deepseek-ai/dsh-client-ui-sidebar"
+      ]
     }
   },
   "engines": {
-    "node": ">=18.17"
+    "node": "^22.19.0 || >=24.0.0",
+    "dsh": ">=0.2.1-rc.3"
   },
   "keywords": [
     "dsh",
@@ -64,5 +78,9 @@
   ],
   "publishConfig": {
     "access": "public"
-  }
+  },
+  "dependencies": {
+    "@stardustlc/dsh-updates": "0.1.0"
+  },
+  "packageManager": "pnpm@11.7.0"
 }

package/skills/computer-use/SKILL.md

@@ -5,7 +5,8 @@ description: Drive Windows desktop apps through DSH Computer Use tools (UIA obse
 
 # Computer Use (Windows)
 
-The `deepseek-computer-use` plugin exposes the Codex window2 API on Windows.
+The `deepseek-computer-use` plugin provides Windows desktop tools for DSH,
+using window2-style action and observation semantics.
 Prefer shell, files and web tools whenever they can do the job: GUI automation is
 the slowest, most fragile option and it moves the user's real mouse and keyboard.
 
@@ -81,10 +82,12 @@ submit, publish, install, 删除, 卸载, 支付, 提交 …) asks for confirmat
 |---|---|---|
 | `stale_snapshot` | element indexes expired | `get_window_state include_text:true` again |
 | `stale_screenshot` | coordinate screenshot expired | observe again and use the new id |
+| `fresh_observation_required` | an earlier input or capture failed | observe again; discard every old index, screenshot id and coordinate before deciding the next action |
+| `timeout` | the old helper was stopped and its snapshots discarded | take a fresh observation; never repeat an input whose outcome is unknown |
+| `cancelled` / `turn_ended` | the active native request was stopped | stop issuing input and wait for a new turn |
 | `window_gone` | app closed / handle changed | `list_windows` again |
 | `window_minimized` | nothing to observe | `activate_window`, then observe |
 | `focus_failed` | another window holds the foreground | stop, ask the user; do not retry in a loop |
 | `screen_locked` | the desktop is locked (LockApp) | stop and ask the user to unlock |
 | `hard_denied` | the surface is never automatable | tell the user, do not route around it |
-| `turn_ended` | the turn is over | wait for the user to continue |
 | empty tree | Electron/Canvas surface | observe with the screenshot and act by coordinates |