We Keep Coding

Is there a way to change the color of the NonClientFrameEdges or make them invisible (particularly with WPF/.NET)?

Goal Summary
I'm trying to make a WPF window with a custom window caption (so I can paint the title bar and add controls there). I've been successful in doing this with the WPF WindowChrome class, but it introduces a lot of bugs. I've been able to work around most of them by setting the WindowChrome NoneClientFrameEdges property to any value other than None, but this introduced a new bug in the process. There's an unsightly ~1px thick border where the NonClientFrameEdge property is set. You can see it in the video, but it's very faint. I want to either set the color to transparent or find a way to disable rendering it entirely. The problem is that I can't actually remove the NonClientFrameEdge because it's required to fix the bugs I mentioned earlier (more details below).
Detailed explanation of the problem (as best as I can understand it)
From what I understand, a standard window is separated into at least two parts: the client area and the non-client area. The client area is the part of the window you can easily customize, where you add your controls and other window content. The non-client area is the part of the window that includes the frame, which consists of the resize borders and the caption. Under normal conditions you can’t modify the non-client area.
The WindowChrome class in WPF provides an easy way to customize the non-client area of a WPF window. This enables you to design your own caption and place controls there like you see in modern enterprise apps such as Visual Studio and Microsoft Office. The problem is that using WindowChrome causes many bugs. Here’s a list of the ones I’ve found so far:
When you minimize the window from a maximized state and hover over it in the taskbar, the resulting window preview has ~8 px of empty space on the top and top-left sides. If you click it to bring it to the front, it snaps back to the proper space. Even Visual Studio and Microsoft Office have this bug. (caused by having GlassFrameThickness value of 0 set in WindowChrome class)
The transparent glass rectangle effect you get when hovering over the aero peek button for “peek at desktop” no longer works. You just see empty space, no outline of the app. (caused by having GlassFrameThickness value of 0 set in WindowChrome class)
The window is jittery when resizing from any corner other than the bottom right. I’ve read this is because resizing from these corners forces a position change of the window. (caused by having NonClientFrameEdges value of 0, at least 1 edge needs to be set to avoid this)
The window is distorted and blurry during the focus zoom effect when selecting a minimized window from task view. Even many enterprise apps have this problem. (Adobe Photoshop, Visual Studio, and several game launchers, to name a few). (caused either by GlassFrameThickness value of 0, NonClientFrameEdges value of None, or a combination of the two)
Hacky solution and video before and after
I noticed that issue 1 and 2 can be fixed by setting the GlassFrameThickness to a non-zero value. Issues 3 and 4 can be fixed by setting the NonClientFrameEdges property of the WindowChrome class to any value other than None. This video shows all of the problems listed above, along with how they look before and after changing that property: https://www.youtube.com/watch?v=z7O28aEPygg
Note: The video ends abruptly due to my poor editing skills, but it should show all the necessary issues. You will have to look very closely at the bottom of the app in the second half of the video to see the 1px grey/white NonClientFrameEdge, but it's noticeable. This is what I'm trying to fix.
The problem is that setting NonClientFrameEdges to any value other than None literally adds a 2-3 px edge to your window, and it's visually obvious. Setting GlassFrameThickness to 1 on the same edge as the NonClientFrameEdge reduces the visibility significantly, but it's still a noticeable 1px or so eyesore. For example if NonClientFrameEdge is set to Bottom, GlassFrameThickness should be set to "0, 0, 0, 1".
Minimal Example
You can reproduce the problem by creating a .NET or .NET Core WPF project and adding the following code to the window's XAML view file:
<WindowChrome.WindowChrome>
<WindowChrome GlassFrameThickness="0 0 0 1" CornerRadius="0" CaptionHeight="38" UseAeroCaptionButtons="False" ResizeBorderThickness="5" NonClientFrameEdges="Bottom" />
</WindowChrome.WindowChrome>
Here's a Git Repo with a minimal project for convenience if you want it to test but don't feel like typing all the code. It includes some extra stuff like the boilerplate for the ViewModel, command firing, buttons for min/max/close, and a button to add a border for testing (makes things like the resize jitter more apparent). It also includes a hook into WndProc with Pinvoke just in case you want to experiment with the WindowsAPI: https://github.com/cjfcode/WindowProject

The key to solving this is to wire up a window hook with a handler for the WM_NCCALCSIZE (0x83) and WM_NCPAINT (0x85) messages.
WM_NCPAINT will allow you to remove the single pixel bottom border by calling DwmExtendFrameIntoClientArea. In the code below I've wrapped that call in a method called RemoveFrame.
WM_NCCALCSIZE will allow you to change the size of the client area of the window restoring the extra space that WindowChrome set by using GlassFrameThickness="0,0,0,1" and NonClientFrameEdges="Bottom".
I've wrapped up this functionality into a XAML behavior.
Here is final code that will solve your issue:
WindowChromeLoadedBehavior
using System;
using System.Drawing;
using System.Runtime.InteropServices;
using System.Threading.Tasks;
using System.Windows;
using System.Windows.Interop;
using System.Windows.Shell;
using Microsoft.Xaml.Behaviors;
namespace WpfApp1
{
public class WindowChromeLoadedBehavior : Behavior<FrameworkElement>
{
private Window window;
protected override void OnAttached()
{
base.OnAttached();
AssociatedObject.Loaded += OnLoaded;
}
protected override void OnDetaching()
{
base.OnDetaching();
AssociatedObject.Loaded -= OnLoaded;
}
private void OnLoaded(object sender, RoutedEventArgs e)
{
window = Window.GetWindow(AssociatedObject);
if (window == null) return;
Task.Delay(5).ContinueWith(_ =>
{
Dispatcher.Invoke(() =>
{
var oldWindowChrome = WindowChrome.GetWindowChrome(window);
if (oldWindowChrome == null) return;
var newWindowChrome = new WindowChrome
{
CaptionHeight = oldWindowChrome.CaptionHeight,
CornerRadius = oldWindowChrome.CornerRadius,
GlassFrameThickness = new Thickness(0, 0, 0, 1),
NonClientFrameEdges = NonClientFrameEdges.Bottom,
ResizeBorderThickness = oldWindowChrome.ResizeBorderThickness,
UseAeroCaptionButtons = oldWindowChrome.UseAeroCaptionButtons
};
WindowChrome.SetWindowChrome(window, newWindowChrome);
});
});
var hWnd = new WindowInteropHelper(window).Handle;
HwndSource.FromHwnd(hWnd)?.AddHook(WndProc);
}
private IntPtr WndProc(IntPtr hwnd, int msg, IntPtr wParam, IntPtr lParam, ref bool handled)
{
switch (msg)
{
case NativeMethods.WM_NCPAINT:
RemoveFrame();
handled = false;
break;
case NativeMethods.WM_NCCALCSIZE:
handled = false;
var rcClientArea = (RECT)Marshal.PtrToStructure(lParam, typeof(RECT));
rcClientArea.Bottom += (int)(WindowChromeHelper.WindowResizeBorderThickness.Bottom / 2);
Marshal.StructureToPtr(rcClientArea, lParam, false);
var retVal = IntPtr.Zero;
if (wParam == new IntPtr(1))
{
retVal = new IntPtr((int)NativeMethods.WVR.REDRAW);
}
return retVal;
}
return IntPtr.Zero;
}
private void RemoveFrame()
{
if (Environment.OSVersion.Version.Major >= 6 && NativeMethods.IsDwmAvailable())
{
if (NativeMethods.DwmIsCompositionEnabled() && SystemParameters.DropShadow)
{
NativeMethods.MARGINS margins;
margins.bottomHeight = -1;
margins.leftWidth = 0;
margins.rightWidth = 0;
margins.topHeight = 0;
var helper = new WindowInteropHelper(window);
NativeMethods.DwmExtendFrameIntoClientArea(helper.Handle, ref margins);
}
}
}
[Serializable]
[StructLayout(LayoutKind.Sequential)]
public struct RECT
{
public int Left;
public int Top;
public int Right;
public int Bottom;
public static RECT Empty;
public int Width => Math.Abs(Right - Left);
public int Height => (Bottom - Top);
public RECT(int left, int top, int right, int bottom)
{
Left = left;
Top = top;
Right = right;
Bottom = bottom;
}
public RECT(RECT rcSrc)
{
Left = rcSrc.Left;
Top = rcSrc.Top;
Right = rcSrc.Right;
Bottom = rcSrc.Bottom;
}
public RECT(Rectangle rectangle) : this(rectangle.Left, rectangle.Top, rectangle.Right, rectangle.Bottom)
{
}
public bool IsEmpty
{
get
{
if (Left < Right)
{
return (Top >= Bottom);
}
return true;
}
}
public override string ToString()
{
if (this == Empty)
{
return "RECT {Empty}";
}
return string.Concat("RECT { left : ", Left, " / top : ", Top, " / right : ", Right, " / bottom : ", Bottom, " }");
}
public override bool Equals(object obj)
{
return ((obj is RECT) && (this == ((RECT)obj)));
}
public override int GetHashCode()
{
return ((Left.GetHashCode() + Top.GetHashCode()) + Right.GetHashCode()) + Bottom.GetHashCode();
}
public static bool operator ==(RECT rect1, RECT rect2)
{
return ((((rect1.Left == rect2.Left) && (rect1.Top == rect2.Top)) && (rect1.Right == rect2.Right)) && (rect1.Bottom == rect2.Bottom));
}
public static bool operator !=(RECT rect1, RECT rect2)
{
return !(rect1 == rect2);
}
static RECT()
{
Empty = new RECT();
}
}
}
}
WindowChromeHelper
using System;
using System.Runtime.InteropServices;
using System.Windows;
namespace WpfApp1
{
public static class WindowChromeHelper
{
public static Thickness LayoutOffsetThickness => new Thickness(0d, 0d, 0d, SystemParameters.WindowResizeBorderThickness.Bottom);
/// <summary>
/// Gets the properly adjusted window resize border thickness from system parameters.
/// </summary>
public static Thickness WindowResizeBorderThickness
{
get
{
var dpix = GetDpi(GetDeviceCapsIndex.LOGPIXELSX);
var dpiy = GetDpi(GetDeviceCapsIndex.LOGPIXELSY);
var dx = GetSystemMetrics(GetSystemMetricsIndex.CXFRAME);
var dy = GetSystemMetrics(GetSystemMetricsIndex.CYFRAME);
// This adjustment is needed since .NET 4.5
var d = GetSystemMetrics(GetSystemMetricsIndex.SM_CXPADDEDBORDER);
dx += d;
dy += d;
var leftBorder = dx / dpix;
var topBorder = dy / dpiy;
return new Thickness(leftBorder, topBorder, leftBorder, topBorder);
}
}
[DllImport("user32.dll")]
private static extern int ReleaseDC(IntPtr hWnd, IntPtr hDC);
[DllImport("user32.dll")]
private static extern IntPtr GetDC(IntPtr hwnd);
[DllImport("gdi32.dll")]
private static extern int GetDeviceCaps(IntPtr hdc, int nIndex);
private static float GetDpi(GetDeviceCapsIndex index)
{
var desktopWnd = IntPtr.Zero;
var dc = GetDC(desktopWnd);
float dpi;
try
{
dpi = GetDeviceCaps(dc, (int)index);
}
finally
{
ReleaseDC(desktopWnd, dc);
}
return dpi / 96f;
}
private enum GetDeviceCapsIndex
{
LOGPIXELSX = 88,
LOGPIXELSY = 90
}
[DllImport("user32.dll")]
private static extern int GetSystemMetrics(GetSystemMetricsIndex nIndex);
private enum GetSystemMetricsIndex
{
CXFRAME = 32,
CYFRAME = 33,
SM_CXPADDEDBORDER = 92
}
}
}
NativeMethods
using System;
using System.Runtime.InteropServices;
namespace WpfApp1
{
public static class NativeMethods
{
public const int WM_NCCALCSIZE = 0x83;
public const int WM_NCPAINT = 0x85;
[DllImport("kernel32", SetLastError = true, CharSet = CharSet.Ansi, BestFitMapping = false, ThrowOnUnmappableChar = true)]
private static extern IntPtr LoadLibrary(string lpFileName);
[DllImport("dwmapi.dll", PreserveSig = false)]
public static extern bool DwmIsCompositionEnabled();
[DllImport("kernel32", SetLastError = true, ExactSpelling = true, CharSet = CharSet.Ansi, BestFitMapping = false, ThrowOnUnmappableChar = true)]
private static extern IntPtr GetProcAddress(IntPtr hModule, string procName);
[StructLayout(LayoutKind.Sequential)]
public struct MARGINS
{
public int leftWidth;
public int rightWidth;
public int topHeight;
public int bottomHeight;
}
private delegate int DwmExtendFrameIntoClientAreaDelegate(IntPtr hwnd, ref MARGINS margins);
public static int DwmExtendFrameIntoClientArea(IntPtr hwnd, ref MARGINS margins)
{
var hModule = LoadLibrary("dwmapi");
if (hModule == IntPtr.Zero)
{
return 0;
}
var procAddress = GetProcAddress(hModule, "DwmExtendFrameIntoClientArea");
if (procAddress == IntPtr.Zero)
{
return 0;
}
var delegateForFunctionPointer = (DwmExtendFrameIntoClientAreaDelegate)Marshal.GetDelegateForFunctionPointer(procAddress, typeof(DwmExtendFrameIntoClientAreaDelegate));
return delegateForFunctionPointer(hwnd, ref margins);
}
public static bool IsDwmAvailable()
{
if (LoadLibrary("dwmapi") == IntPtr.Zero)
{
return false;
}
return true;
}
internal enum WVR
{
ALIGNTOP = 0x0010,
ALIGNLEFT = 0x0020,
ALIGNBOTTOM = 0x0040,
ALIGNRIGHT = 0x0080,
HREDRAW = 0x0100,
VREDRAW = 0x0200,
VALIDRECTS = 0x0400,
REDRAW = HREDRAW | VREDRAW
}
}
}

How to set the client area (ClientRectangle) in a borderless form?

I want to know if it's possible to set the client area of a borderless form. Say for example I define a form like so:
Code
public class MyForm : Form
{
public MyForm()
{
this.FormBorderStyle = FormBorderStyle.None;
}
}
Result
What I want to do is specify the client area, so that the form has a frame (like the standard windows frame, but custom drawn).
Result
Essentially, the blue area would become the non client area, and the gray area would remain as the client area.
I have tried to set the client area, but this just seems to resize the entire form, thus, is does not leave behind a "non-client" area
Is this possible?

This is possible, however I don't know how well this works with a Windows Form with the WindowStyle set to Borderless. Using PInvoke (Platform Invoke) Functions, you can remove window themes which will give you a very basic looking Windows Form. You can then use various PInvoke functions to manipulate the Non-client area of the windows form.
I recommend that you read through these topics. They're designed for Win32 Applications using C++, but PInvoke is the process of calling these native APIs using Managed Code (C#)
WM_NCCALCSIZE: https://msdn.microsoft.com/en-us/library/windows/desktop/ms632634(v=vs.85).aspx
WM_NCPAINT: https://msdn.microsoft.com/en-us/library/windows/desktop/dd145212(v=vs.85).aspx
GetDCEx: https://msdn.microsoft.com/en-us/library/windows/desktop/dd144873(v=vs.85).aspx
GetWindowDC: https://msdn.microsoft.com/en-us/library/windows/desktop/dd144947(v=vs.85).aspx
SetWindowTheme: https://msdn.microsoft.com/en-us/library/windows/desktop/bb759827(v=vs.85).aspx
This example is very very crude, but it provides basic functionality. I don't know how SetWindowTheme works on Windows 8 or 8.1, but in Windows 7, it gives windows the "classic" theme.
public partial class MyForm : Form
{
//Window Messages
public const uint WM_NCPAINT = 0x85;
public const uint WM_NCCALCSIZE = 0x83;
public const uint WM_NCHITTEST = 0x84;
//GetDCEx Flags
public const int DCX_WINDOW = 0x00000001;
public const int DCX_CACHE = 0x00000002;
public const int DCX_PARENTCLIP = 0x00000020;
public const int DCX_CLIPSIBLINGS = 0x00000010;
public const int DCX_CLIPCHILDREN = 0x00000008;
public const int DCX_NORESETATTRS = 0x00000004;
public const int DCX_LOCKWINDOWUPDATE = 0x00000400;
public const int DCX_EXCLUDERGN = 0x00000040;
public const int DCX_INTERSECTRGN = 0x00000080;
public const int DCX_INTERSECTUPDATE = 0x00000200;
public const int DCX_VALIDATE = 0x00200000;
//RECT Structure
[System.Runtime.InteropServices.StructLayout(System.Runtime.InteropServices.LayoutKind.Sequential)]
public struct RECT
{
public int left, top, right, bottom;
}
//WINDOWPOS Structure
[System.Runtime.InteropServices.StructLayout(System.Runtime.InteropServices.LayoutKind.Sequential)]
public struct WINDOWPOS
{
public IntPtr hwnd;
public IntPtr hwndinsertafter;
public int x, y, cx, cy;
public int flags;
}
//NCCALCSIZE_PARAMS Structure
[System.Runtime.InteropServices.StructLayout(System.Runtime.InteropServices.LayoutKind.Sequential)]
public struct NCCALCSIZE_PARAMS
{
public RECT rgrc0, rgrc1, rgrc2;
public WINDOWPOS lppos;
}
//SetWindowTheme UXtheme Function
[System.Runtime.InteropServices.DllImport("uxtheme.dll", ExactSpelling = true, CharSet = System.Runtime.InteropServices.CharSet.Unicode)]
public static extern int SetWindowTheme(
IntPtr hWnd,
String pszSubAppName,
String pszSubIdList);
//GetWindowRect User32 Function
[System.Runtime.InteropServices.DllImport("user32.dll", ExactSpelling = true)]
[return: System.Runtime.InteropServices.MarshalAs(System.Runtime.InteropServices.UnmanagedType.Bool)]
public static extern bool GetWindowRect(
IntPtr hwnd,
out RECT lpRect
);
//GetWindowDC User32 Function
[System.Runtime.InteropServices.DllImport("user32.dll", ExactSpelling = true)]
public static extern IntPtr GetWindowDC(
IntPtr hWnd
);
//GetDCEx User32 Function
[System.Runtime.InteropServices.DllImport("user32.dll", ExactSpelling = true)]
public static extern IntPtr GetDCEx(
IntPtr hWnd,
IntPtr hrgnClip,
int flags
);
//Window Procedure Hook
protected override void WndProc(ref Message m)
{
//Don't style window in designer...
if (DesignMode)
base.WndProc(ref m);
//Handle Message
switch ((uint)m.Msg)
{
case WM_NCCALCSIZE: WmNCCalcSize(ref m); break;
case WM_NCPAINT: WmNCPaint(ref m); break;
default: base.WndProc(ref m); break;
}
}
//Handle Creation
protected override void OnHandleCreated(EventArgs e)
{
//Base Procedure...
base.OnHandleCreated(e);
//Remove Theme
SetWindowTheme(this.Handle, string.Empty, string.Empty);
}
//WM_NCCALCSIZE
private void WmNCCalcSize(ref Message m)
{
//Get Window Rect
RECT formRect = new RECT();
GetWindowRect(m.HWnd, out formRect);
//Check WPARAM
if (m.WParam != IntPtr.Zero) //TRUE
{
//When TRUE, LPARAM Points to a NCCALCSIZE_PARAMS structure
var nccsp = (NCCALCSIZE_PARAMS)System.Runtime.InteropServices.Marshal.PtrToStructure(m.LParam, typeof(NCCALCSIZE_PARAMS));
//We're adjusting the size of the client area here. Right now, the client area is the whole form.
//Adding to the Top, Bottom, Left, and Right will size the client area.
nccsp.rgrc0.top += 30; //30-pixel top border
nccsp.rgrc0.bottom -= 4; //4-pixel bottom (resize) border
nccsp.rgrc0.left += 4; //4-pixel left (resize) border
nccsp.rgrc0.right -= 4; //4-pixel right (resize) border
//Set the structure back into memory
System.Runtime.InteropServices.Marshal.StructureToPtr(nccsp, m.LParam, true);
}
else //FALSE
{
//When FALSE, LPARAM Points to a RECT structure
var clnRect = (RECT)System.Runtime.InteropServices.Marshal.PtrToStructure(m.LParam, typeof(RECT));
//Like before, we're adjusting the rectangle...
//Adding to the Top, Bottom, Left, and Right will size the client area.
clnRect.top += 30; //30-pixel top border
clnRect.bottom -= 4; //4-pixel bottom (resize) border
clnRect.left += 4; //4-pixel left (resize) border
clnRect.right -= 4; //4-pixel right (resize) border
//Set the structure back into memory
System.Runtime.InteropServices.Marshal.StructureToPtr(clnRect, m.LParam, true);
}
//Return Zero
m.Result = IntPtr.Zero;
}
//WM_NCPAINT
private void WmNCPaint(ref Message m)
{
//Store HDC
IntPtr HDC = IntPtr.Zero;
Graphics gfx = null;
//Check the WPARAM
if(m.WParam == (IntPtr)1)
{
//For reasons unknown to me, the update region doesn't contain valid data and calling GetDCEx will do nothing.
//So I call GetWindowDC and exclude the area using System.Drawing.Graphics instead.
//Graphics Object from HDC
HDC = GetWindowDC(m.HWnd);
gfx = Graphics.FromHdc(HDC);
//Exclude Client Area
gfx.ExcludeClip(new Rectangle(4, 30, Width - 8, 34)); //Exclude Client Area (GetWindowDC grabs the WHOLE window's graphics handle)
}
else
{
//Graphics Object from HDC
HDC = GetDCEx(m.HWnd, m.WParam, DCX_WINDOW | DCX_INTERSECTRGN);
gfx = Graphics.FromHdc(HDC);
}
//Call Paint
using (PaintEventArgs ncPaintArgs = new PaintEventArgs(gfx, new Rectangle(0, 0, Width, Height)))
MyForm_NCPaint(this, ncPaintArgs);
//Return Zero
m.Result = IntPtr.Zero;
}
public MyForm()
{
InitializeComponent();
}
private void MyForm_NCPaint(object sender, PaintEventArgs e)
{
//Clear
e.Graphics.Clear(Color.Green);
}
}

WPF Window fixed width mouse cursor

I am building a WPF 4.5 Application that has controls that enable the User to "Lock" and "Unlock" the Application's Height.
In order to lock the Height, I am following this StackOverflow answer regarding setting the MinHeight and MaxHeight to the same value.
In order to unlock the Height, I set MinHeight=0 and MaxHeight=double.PositiveInfinity
This all appears to be working fine.
The problem I'm encountering that I haven't been able to solve is that when the height is "Locked", when I mouseover the right edge of the Application Window, the cursor turns into the horizontal resize cursor.
Is there a way I can disable that so that the cursor stays as the regular pointer in WPF?
I am on WPF 4.5.
I saw this post that has answers showing how to do it in Win32: WPF: Make window unresizeable, but keep the frame?.
This post is over 3 years old, and I was just wondering (hoping) maybe WPF has evolved since then.
Thank you very much in advance!
Philip

On your startup Window (MainWindow.xaml), try making a binding for the Window's ResizeMode property and then modifying it to 'NoResize' when you don't want it to be resizable. To make it resizable, change it to 'CanResize'.
Hope that helps!

You need to set MinWidth = MaxWidth = Width = your desired width as mentioned in this StackOverflow answer regarding setting the MinHeight and MaxHeight to the same value.
In addition you need to hook the winproc for your window and process the WM_NCHITTEST message.
#region Vertical Resize Only
// ReSharper disable InconsistentNaming
private const int WM_NCHITTEST = 0x0084;
private const int HTBORDER = 18;
private const int HTBOTTOM = 15;
private const int HTBOTTOMLEFT = 16;
private const int HTBOTTOMRIGHT = 17;
private const int HTLEFT = 10;
private const int HTRIGHT = 11;
private const int HTTOP = 12;
private const int HTTOPLEFT = 13;
private const int HTTOPRIGHT = 14;
[DllImport("user32.dll", CharSet = CharSet.Auto)]
public static extern IntPtr DefWindowProc(
IntPtr hWnd,
int msg,
IntPtr wParam,
IntPtr lParam);
// ReSharper restore InconsistentNaming
#endregion Vertical Resize Only
public CanConfigurationDialog()
{
InitializeComponent();
Loaded += MainWindowLoaded;
}
#region Vertical Resize Only
private void MainWindowLoaded(object sender, RoutedEventArgs e)
{
try
{
// Obtain the window handle for WPF application
var mainWindowPtr = new WindowInteropHelper(this).Handle;
var mainWindowSrc = HwndSource.FromHwnd(mainWindowPtr);
mainWindowSrc?.AddHook(WndProc);
}
catch (Exception)
{
;
}
}
private static IntPtr WndProc(IntPtr hwnd, int msg, IntPtr wParam, IntPtr lParam, ref bool handled)
{
// Override the window hit test
// and if the cursor is over a resize border,
// return a standard border result instead.
if (msg == WM_NCHITTEST)
{
handled = true;
var htLocation = DefWindowProc(hwnd, msg, wParam, lParam).ToInt32();
switch (htLocation)
{
case HTTOP:
case HTTOPLEFT:
case HTTOPRIGHT:
htLocation = HTTOP;
break;
case HTBOTTOM:
case HTBOTTOMLEFT:
case HTBOTTOMRIGHT:
htLocation = HTBOTTOM;
break;
case HTLEFT:
case HTRIGHT:
htLocation = HTBORDER;
break;
}
return new IntPtr(htLocation);
}
return IntPtr.Zero;
}
#endregion Vertical Resize Only
This will prevent the horizontal resize cursor from being displayed!
Q.E.D.

Best way to take screenshot of a web page

What is the best way to take screenshot of a web page?
At the moment I just start an selenium instance of firefox and using winapi bring it to the front and make a screenshot.
I ask similar question already.
There is two points:
Slowness.
If any window occurently gets higher than our web browser's window, this window will imprint in our screenshot.
Is there any method to take screenshot more 'programmly'?
Here is some code I use now:
class FirefoxDriverEx : FirefoxDriver
{
public Process GetFirefoxProcess()
{
var fi = typeof(FirefoxBinary).GetField("process", BindingFlags.NonPublic | BindingFlags.Instance);
return fi.GetValue(this.Binary) as Process;
}
}
Here is the code illustrating process of taking screenshot itself:
using (FirefoxDriverEx driver = new FirefoxDriverEx())
{
driver.Navigate().GoToUrl(url);
var process = driver.GetFirefoxProcess();
if (process != null)
{
var screenCapture = new ScreenCapture();
Win.SetForegroundWindow(process.MainWindowHandle.ToInt32());
}
}
Right now, I'm thinking about some manager that will control a queue of windows to take the screenshots from.
Question edit.
I'm not looking for a solution to just get screenshot 'in memory' and return it back to HTTP stream. So any ways to save screenshot and save it to file and then get it from there is very ambiguous for that purpose.
Question edit #2.
I forgot to mention. Needed screenshot should be made as it seen by user. So, screenshot should have browser window and a site inside of web browser window's bounds. I can't find any way to change mode of taking a screenshot in WebDriver of selenium. WebDriver just take screenshot of a page without any browser window.

I'd recommend getScreenshotAs. It gets even the 'out of view' part of the screen.
Here is some sample code in gr0ovy.
import java.io.IOException
import java.net.URL
import java.nio.file.Path
import java.nio.file.Paths
import java.text.SimpleDateFormat
import org.openqa.selenium.Capabilities
import org.openqa.selenium.TakesScreenshot
import org.openqa.selenium.WebDriverException
import org.openqa.selenium.remote.CapabilityType
import org.openqa.selenium.remote.DriverCommand
import org.openqa.selenium.remote.RemoteWebDriver
import org.openqa.selenium.OutputType
import org.openqa.selenium.WebDriver
public class Selenium2Screenshot {
private WebDriver driver
private String browserType
private boolean skipScreenshots
public Selenium2Screenshot(WebDriver webDriver, String browserType, boolean skipScreenshots) {
this.driver = webDriver
this.browserType = browserType
this.skipScreenshots = skipScreenshots
}
public void takeScreenshot(String filenameBase) {
if (!skipScreenshots) {
Date today
String formattedDate
SimpleDateFormat formatter
Locale currentLocale
File scrFile
currentLocale = new Locale("en", "US")
formatter = new SimpleDateFormat("yyyy_MM_dd_HH_mm_ss_SSS", currentLocale)
today = new Date()
formattedDate = formatter.format(today)
String filename = getUiAutomationDir() + filenameBase + "_" + browserType + formattedDate + ".png"
Log.logger.info("Screenshot filename = " + filename)
try {
scrFile = ((TakesScreenshot)driver).getScreenshotAs(OutputType.FILE)
JavaIO.copy(scrFile.getAbsolutePath(), filename)
} catch (Exception e) {
Log.logger.error(e.message, e)
}
} else {
Log.logger.info("Skipped Screenshot")
}
}
private String getUiAutomationDir()
{
String workingDir = System.getProperty("user.dir")
Path workingDirPath = Paths.get(workingDir)
String returnString = workingDirPath.toString() + "\\"
return returnString
}
}
Edited on 8/1/12:
Get application handle code. I am surely duplicating code that is on stackoverflow several times, but hopefully this is not the exact same code as in other posts :-)
public static IntPtr FindWindowByPartialCaption(String partialCaption)
{
var desktop = User32.GetDesktopWindow();
var children = EnumerateWindows.GetChildWindows(desktop);
foreach (var intPtr in children)
{
var current = GetText(intPtr);
if (current.Contains(partialCaption))
return intPtr;
}
return IntPtr.Zero;
}
[DllImport("user32.dll", EntryPoint = "GetDesktopWindow")]
public static extern IntPtr GetDesktopWindow();
[DllImport("user32.dll")]
public static extern bool EnumChildWindows(IntPtr hWndParent, EnumWindowProc lpEnumFunc, IntPtr lParam);
public delegate bool EnumWindowProc(IntPtr hWnd, IntPtr parameter);
public static List<IntPtr> GetChildWindows(IntPtr parent)
{
return GetChildWindows(parent, false);
}
public static List<IntPtr> GetChildWindows(IntPtr parent, bool reverse)
{
List<IntPtr> result = new List<IntPtr>();
GCHandle listHandle = GCHandle.Alloc(result);
try
{
EnumWindowProc childProc = new EnumWindowProc(EnumWindow);
EnumChildWindows(parent, childProc, GCHandle.ToIntPtr(listHandle));
}
finally
{
if (listHandle.IsAllocated)
listHandle.Free();
}
if (reverse)
{
List<IntPtr> resultList = result.Reverse<IntPtr>().ToList();
return resultList;
}
else
return result;
}
private static bool EnumWindow(IntPtr handle, IntPtr pointer)
{
GCHandle gch = GCHandle.FromIntPtr(pointer);
List<IntPtr> list = gch.Target as List<IntPtr>;
if (list == null)
{
throw new InvalidCastException("GCHandle Target could not be cast as List<IntPtr>");
}
list.Add(handle);
// You can modify this to check to see if you want to cancel the operation, then return a null here
return true;
}
}
http://www.pinvoke.net/ is also a great resource.

http://msdn.microsoft.com/en-us/library/windows/desktop/dd162869(v=vs.85).aspx
I personally love this API. Create a bitmap with width and height calculated from the returned rectangle of GetWindowRect API and for HDC parameter use (for example):
thebitmap.GetHdc()
You should be fine.
Edit: also check this.
Btw, you can take screenshot of any window you like, even if they fall back.(note that this will not work for minimized windows. However, if you really need, there are some way arounds for that too.)

If you're looking for a programmatic way to get a screenshot of the main window of a given process, here is a function that does it:
public static Bitmap TakeScreenshot(Process process)
{
// may need a process Refresh before
return TakeScreenshot(process.MainWindowHandle);
}
public static Bitmap TakeScreenshot(IntPtr handle)
{
RECT rc = new RECT();
GetWindowRect(handle, ref rc);
Bitmap bitmap = new Bitmap(rc.right - rc.left, rc.bottom - rc.top);
using (Graphics graphics = Graphics.FromImage(bitmap))
{
PrintWindow(handle, graphics.GetHdc(), 0);
}
return bitmap;
}
[DllImport("user32.dll")]
private static extern bool GetWindowRect(IntPtr hWnd, ref RECT rect);
[DllImport("user32.dll")]
private static extern bool PrintWindow(IntPtr hWnd, IntPtr hDC, int flags);
[StructLayout(LayoutKind.Sequential)]
private struct RECT
{
public int left;
public int top;
public int right;
public int bottom;
}
Unfortunately, on Aero-equipped OS (Vista/Win7/Win8) it will not capture the full transparent border. The usual transparent border will be blacked instead. Maybe it's enough for what you're trying to acomplish.

I've been using webshotcmd (the paid version is also command line) in a production app for years. It can be configured to wait for the page to load, to wait n seconds after page load, etc. It uses Internet Explorer and works on Windows. Starts pretty fast (in my experience, the msie activex has always been instant to load).
Other than the above, I would recommend something based on a Webkit libray, it would be so much smaller than Firefox, and would start very fast (wkhtmltoimage is for now only available on Linux, but when it will be available for Windows, I would go for it - also command line). Right now just google for webkit screenshot (the huge number of available screenshotters using webkit makes me believe using that DLL would be easy to port to C#).
Edit: Considering your 2nd edit, take a look at Chrome Screen Capture source.
To try it, the extension is available in the store/extension gallery.

I was able to accomplish this by copying the window (piece by piece) into a bitmap that is set to the size of the ScrollRectangle for my webBrowser control. While it is certainly not the most elegant way of achieving this goal, I wanted to share the code in case anyone might be able to use it. Once I had something that was mostly working, I was then able to add some args, and I can now execute this utility from the command line:
Executable_Path URL Filename
/// <summary>
/// This method is called to start the process of copying the webpage to the bitmap
/// this should be called after the page has fully loaded (use DocumentCompleted event to determine
/// if the page has completed loading if calling from the command line.)
/// </summary>
private void copyWebpageToImage()
{
//these two vars will house the current position in the bmp file (starting at 0,0)
int currXPosition = 0;
int currYPosition = 0;
//we need to set the height and width of our bitmap to the scrollrectangle of the webbrowser document object
int width = webBrowser1.Document.Body.ScrollRectangle.Width;
int height = webBrowser1.Document.Body.ScrollRectangle.Height;
//instantiate the bitmap
bm = new Bitmap(wd, ht);
//Instantiate our graphics object
Graphics gfx = Graphics.FromImage((Image)bm);
//this point is used throughout the process, and helps to determine where the form is at on the screen
Point formPoint = Form1.ActiveForm.Location;
formPoint.X = formPoint.X + webBrowser1.Location.X;
formPoint.Y = formPoint.Y + webBrowser1.Location.Y;
formPoint.X = formPoint.X + 8; //offsets for my form (may be different for yours)
formPoint.Y = formPoint.Y + 33; //offsets for my form
//begin our recursive call that will stop when it reaches the end of the page
copyEverythingToBitmap(bm, currXPosition, currYPosition, formPoint, gfx);
}
private void copyEverythingToBitmap(Bitmap bm, int currXPosition, int currYPosition, Point formPoint, Graphics gfx)
{
//check to see if currXPosition and currYPosition are both 0, if so we just began, call the zero copy method
if (currXPosition == 0 && currYPosition == 0)
{
performZeroCopy(bm, currXPosition, currYPosition, formPoint, gfx);
}
//if the current x position is less than the total width of the scrollrectangle - the width of the webbrowser,
//then we need to scroll the window, and copy the contents, y stays the same
else if (currXPosition < bm.Width - webBrowser1.Width)
{
AlterXPosition(bm, ref currXPosition, ref currYPosition, ref formPoint, gfx);
}
//if we are no longer at the zero, zero, and we cannot increase the x position anymore,
//then we need to scroll the window down and copy the contents, x is reset back to zero
else if(currYPosition < bm.Height - webBrowser1.Height)
{
currYPosition = currYPosition + webBrowser1.Height - 20;
currXPosition = 0;
performZeroCopy(bm, currXPosition, currYPosition, formPoint, gfx);
}
}
/// <summary>
/// The name of this method is slightly misleading. It inherently means that X is zero.
/// </summary>
private void performZeroCopy(Bitmap bm, int currXPosition, int currYPosition, Point formPoint, Graphics gfx)
{
webBrowser1.Document.Window.ScrollTo(currXPosition, currYPosition);
gfx.CopyFromScreen(formPoint, new Point(currXPosition, currYPosition), new Size(webBrowser1.Width - 20, webBrowser1.Height - 20));
if (currXPosition < bm.Width - webBrowser1.Width)
{
AlterXPosition(bm, ref currXPosition, ref currYPosition, ref formPoint, gfx);
}
else if(currYPosition < bm.Height - webBrowser1.Height)
{
currYPosition = currYPosition + webBrowser1.Height - 20;
currXPosition = 0;
performZeroCopy(bm, currXPosition, currYPosition, formPoint, gfx);
}
}
private void AlterXPosition(Bitmap bm, ref int currXPosition, ref int currYPosition, ref Point formPoint, Graphics gfx)
{
currXPosition = currXPosition + webBrowser1.Width - 20;
webBrowser1.Document.Window.ScrollTo(bm.Width - currXPosition, currYPosition);
gfx.CopyFromScreen(formPoint, new Point(bm.Width - currXPosition - 3, currYPosition), new Size(webBrowser1.Width - 20, webBrowser1.Height - 20));
if (currXPosition + webBrowser1.Width < bm.Width)
{
//we still have not traversed the full width of the page, call to alterxposition again...
}
else
{
copyEverythingToBitmap(bm, currXPosition, currYPosition, formPoint, gfx);
}
}
private void saveImageToFile(string p)
{
bm.Tag = DateTime.Now;
bm.Save(p, ImageFormat.Jpeg);
}

Transparent window layer that is click-through and always stays on top

This is some code that I picked up which I tried to implement. Its purpose is to create a form layer which is transparent, full screen, borderless, clickthrough, and always on top of other windows. It then lets you draw using directx over the top of it remaining otherwise transparent.
The parts that don't work are the click-through part, and the directx render. When I run it I basically have an invisible force field in front of all other windows and have to alt-tab around to visual studio to quickly press ALT F5 and end the debug (so at least the always on top and transparency works). I have been trying to figure out why those parts don't work, but my newbie c# skills fail me. hopefully someone can spot why and provide a modification.
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Text;
using System.Windows.Forms;
using System.Diagnostics;
using System.Runtime.InteropServices;
using System.Globalization;
using Microsoft.DirectX;
using Microsoft.DirectX.Direct3D;
using System.Threading;
namespace MinimapSpy
{
public partial class Form1 : Form
{
private Margins marg;
//this is used to specify the boundaries of the transparent area
internal struct Margins
{
public int Left, Right, Top, Bottom;
}
[DllImport("user32.dll", SetLastError = true)]
private static extern UInt32 GetWindowLong(IntPtr hWnd, int nIndex);
[DllImport("user32.dll")]
static extern int SetWindowLong(IntPtr hWnd, int nIndex, IntPtr dwNewLong);
[DllImport("user32.dll")]
static extern bool SetLayeredWindowAttributes(IntPtr hwnd, uint crKey, byte bAlpha, uint dwFlags);
public const int GWL_EXSTYLE = -20;
public const int WS_EX_LAYERED = 0x80000;
public const int WS_EX_TRANSPARENT = 0x20;
public const int LWA_ALPHA = 0x2;
public const int LWA_COLORKEY = 0x1;
[DllImport("dwmapi.dll")]
static extern void DwmExtendFrameIntoClientArea(IntPtr hWnd, ref Margins pMargins);
private Device device = null;
public Form1()
{
//Make the window's border completely transparant
SetWindowLong(this.Handle, GWL_EXSTYLE,
(IntPtr)(GetWindowLong(this.Handle, GWL_EXSTYLE) ^ WS_EX_LAYERED ^ WS_EX_TRANSPARENT));
//Set the Alpha on the Whole Window to 255 (solid)
SetLayeredWindowAttributes(this.Handle, 0, 255, LWA_ALPHA);
//Init DirectX
//This initializes the DirectX device. It needs to be done once.
//The alpha channel in the backbuffer is critical.
PresentParameters presentParameters = new PresentParameters();
presentParameters.Windowed = true;
presentParameters.SwapEffect = SwapEffect.Discard;
presentParameters.BackBufferFormat = Format.A8R8G8B8;
this.device = new Device(0, DeviceType.Hardware, this.Handle,
CreateFlags.HardwareVertexProcessing, presentParameters);
Thread dx = new Thread(new ThreadStart(this.dxThread));
dx.IsBackground = true;
dx.Start();
InitializeComponent();
}
protected override void OnPaint(PaintEventArgs e)
{
//Create a margin (the whole form)
marg.Left = 0;
marg.Top = 0;
marg.Right = this.Width;
marg.Bottom = this.Height;
//Expand the Aero Glass Effect Border to the WHOLE form.
// since we have already had the border invisible we now
// have a completely invisible window - apart from the DirectX
// renders NOT in black.
DwmExtendFrameIntoClientArea(this.Handle, ref marg);
}
private void Form1_Load(object sender, EventArgs e)
{
}
private void dxThread()
{
while (true)
{
//Place your update logic here
device.Clear(ClearFlags.Target, Color.FromArgb(0, 0, 0, 0), 1.0f, 0);
device.RenderState.ZBufferEnable = false;
device.RenderState.Lighting = false;
device.RenderState.CullMode = Cull.None;
device.Transform.Projection = Matrix.OrthoOffCenterLH(0, this.Width, this.Height, 0, 0, 1);
device.BeginScene();
//Place your rendering logic here
device.EndScene();
//device.Present();
}
this.device.Dispose();
Application.Exit();
}
}

Here's a refined full sample code for making a window topmost - click through - transparent (= alpha blended). The sample makes a rotating color wheel which is rendered with DirectX, or actually with XNA 4.0, because I believe Microsoft has discontinued developing the managed directx and favours XNA today.
using System;
using System.Windows.Forms;
using System.Runtime.InteropServices;
using Microsoft.Xna.Framework.Graphics;
namespace ClickThroughXNA
{
public partial class Form1 : Form
{
// Directx graphics device
GraphicsDevice dev = null;
BasicEffect effect = null;
// Wheel vertexes
VertexPositionColor[] v = new VertexPositionColor[100];
// Wheel rotation
float rot = 0;
public Form1()
{
InitializeComponent();
StartPosition = FormStartPosition.CenterScreen;
Size = new System.Drawing.Size(500, 500);
FormBorderStyle = System.Windows.Forms.FormBorderStyle.None; // no borders
TopMost = true; // make the form always on top
Visible = true; // Important! if this isn't set, then the form is not shown at all
// Set the form click-through
int initialStyle = GetWindowLong(this.Handle, -20);
SetWindowLong(this.Handle, -20, initialStyle | 0x80000 | 0x20);
// Create device presentation parameters
PresentationParameters p = new PresentationParameters();
p.IsFullScreen = false;
p.DeviceWindowHandle = this.Handle;
p.BackBufferFormat = SurfaceFormat.Vector4;
p.PresentationInterval = PresentInterval.One;
// Create XNA graphics device
dev = new GraphicsDevice(GraphicsAdapter.DefaultAdapter, GraphicsProfile.Reach, p);
// Init basic effect
effect = new BasicEffect(dev);
// Extend aero glass style on form init
OnResize(null);
}
protected override void OnResize(EventArgs e)
{
int[] margins = new int[] { 0, 0, Width, Height };
// Extend aero glass style to whole form
DwmExtendFrameIntoClientArea(this.Handle, ref margins);
}
protected override void OnPaintBackground(PaintEventArgs e)
{
// do nothing here to stop window normal background painting
}
protected override void OnPaint(PaintEventArgs e)
{
// Clear device with fully transparent black
dev.Clear(new Microsoft.Xna.Framework.Color(0, 0, 0, 0.0f));
// Rotate wheel a bit
rot+=0.1f;
// Make the wheel vertexes and colors for vertexes
for (int i = 0; i < v.Length; i++)
{
if (i % 3 == 1)
v[i].Position = new Microsoft.Xna.Framework.Vector3((float)Math.Sin((i + rot) * (Math.PI * 2f / (float)v.Length)), (float)Math.Cos((i + rot) * (Math.PI * 2f / (float)v.Length)), 0);
else if (i % 3 == 2)
v[i].Position = new Microsoft.Xna.Framework.Vector3((float)Math.Sin((i + 2 + rot) * (Math.PI * 2f / (float)v.Length)), (float)Math.Cos((i + 2 + rot) * (Math.PI * 2f / (float)v.Length)), 0);
v[i].Color = new Microsoft.Xna.Framework.Color(1 - (i / (float)v.Length), i / (float)v.Length, 0, i / (float)v.Length);
}
// Enable position colored vertex rendering
effect.VertexColorEnabled = true;
foreach (EffectPass pass in effect.CurrentTechnique.Passes) pass.Apply();
// Draw the primitives (the wheel)
dev.DrawUserPrimitives(PrimitiveType.TriangleList, v, 0, v.Length / 3, VertexPositionColor.VertexDeclaration);
// Present the device contents into form
dev.Present();
// Redraw immediatily
Invalidate();
}
[DllImport("user32.dll", SetLastError = true)]
static extern int GetWindowLong(IntPtr hWnd, int nIndex);
[DllImport("user32.dll")]
static extern int SetWindowLong(IntPtr hWnd, int nIndex, int dwNewLong);
[DllImport("dwmapi.dll")]
static extern void DwmExtendFrameIntoClientArea(IntPtr hWnd, ref int[] pMargins);
}
}

A little extension/modification to Jaska's code, which the form is transparent
public partial class Form1 : Form
{
public Form1()
{
InitializeComponent();
this.TopMost = true; // make the form always on top
this.FormBorderStyle = System.Windows.Forms.FormBorderStyle.None; // hidden border
this.WindowState = FormWindowState.Maximized; // maximized
this.MinimizeBox = this.MaximizeBox = false; // not allowed to be minimized
this.MinimumSize = this.MaximumSize = this.Size; // not allowed to be resized
this.TransparencyKey = this.BackColor = Color.Red; // the color key to transparent, choose a color that you don't use
}
protected override CreateParams CreateParams
{
get
{
CreateParams cp = base.CreateParams;
// Set the form click-through
cp.ExStyle |= 0x80000 /* WS_EX_LAYERED */ | 0x20 /* WS_EX_TRANSPARENT */;
return cp;
}
}
protected override void OnPaint(PaintEventArgs e)
{
base.OnPaint(e);
// draw what you want
e.Graphics.FillEllipse(Brushes.Blue, 30, 30, 100, 100);
}
}

Change your extended window style to only WS_EX_LAYERED, window style to only WS_POPUP (NO WS_SIZEBOX) and make sure to use DwmExtendFrameIntoClientArea with all -1's and this will produce transparent windows with layered support: downside is you need to bltbit with GDI from an offscreen directx rendering. Not optimal but it works. This gives mouse click throughs + directx rendering + transparency. Downside is you'll need to inform GDI anytime, pull the directx buffer (all of it or just the damaged portions) and write them to the screem with bltbit.
Setting the extended window style to WS_EX_COMPOSITED and DwmExtendedFrameIntoClientArea with all -1's (similar as above, WS_POPUP on the regular window style). This you can run directx from but no mouse clickthroughs. You can at this point define irregular paths for the hit mask and pass it to windows, its not perfect but if you know a general (non regular) area that can pass-through it'll work.
Still trying to find a true way of using opengl/directx on mac or windows platforms that can pass through mouse clicks with out having to do a bitblt to a legacy rendering system.

I have a simple way use TransparentKey property and a 1x1 pixel label with the color of Form TransparentKey.
On Form and all control MouseMouse event. Set label position to Mouse location.
private void MoveHole()
{
var newLocation = PointToClient(MousePosition);
lblHole.Location = newLocation;
}