最近为了学习C#，决定自己做一个屏幕截图工具，来代替长久以来每次都要按下PrintScreen键然后到Clipboard里面寻找之麻烦。学以致用~
用C#做屏幕截图，大致有三种方法。
1、最managed大概就是使用Graphics.CopyFromScreen()方法，此方法有四个重载，不过经反编可见最后调用的都是：
CopyFromScreen(int sourceX, int sourceY, int destinationX, int destinationY, Size blockRegionSize, CopyPixelOperation copyPixelOperation)
调用该方法的优点是代码简单，不用平台调用。不过最大的缺点是：不能截取半透明窗体。你要想截图透明窗体的话，可以加一个参数，它就是：CopyPixelOperation.CaptureBlt。但是，这样的话，只能截图透明窗体，截不到屏幕的其他部分了。所以，该方法已经基本可以放弃了。不过我们还是来研究了一下CopyFromScreen()的内部实现吧，看看问题之所在。下面是该方法的具体实现，reflect自System.Drawing.dll
CopyFromScreen
public void CopyFromScreen(int sourceX, int sourceY, int destinationX, int destinationY, Size blockRegionSize, CopyPixelOperation copyPixelOperation)
{
switch (copyPixelOperation)
{
case CopyPixelOperation.NotSourceErase:
case CopyPixelOperation.NotSourceCopy:
case CopyPixelOperation.NoMirrorBitmap:
case CopyPixelOperation.Blackness:
case CopyPixelOperation.SourceErase:
...
case CopyPixelOperation.SourceCopy:
case CopyPixelOperation.SourceAnd:
case CopyPixelOperation.MergePaint:
case CopyPixelOperation.SourcePaint:
case CopyPixelOperation.PatCopy:
case CopyPixelOperation.PatPaint:
case CopyPixelOperation.Whiteness:
case CopyPixelOperation.CaptureBlt:
{
new UIPermission(UIPermissionWindow.AllWindows).Demand();
int width = blockRegionSize.Width;
int height = blockRegionSize.Height;
using (DeviceContext context = DeviceContext.FromHwnd(IntPtr.Zero))
{
HandleRef hSrcDC = new HandleRef(null, context.Hdc);
HandleRef hDC = new HandleRef(null, this.GetHdc());
try
{
if (SafeNativeMethods.BitBlt(hDC, destinationX, destinationY, width, height, hSrcDC, sourceX, sourceY, (int) copyPixelOperation) == 0)
{
throw new Win32Exception();
}
}
finally
{
this.ReleaseHdc();
}
}
return;
}
}
throw new InvalidEnumArgumentException("value", (int) copyPixelOperation, typeof(CopyPixelOperation));
}
通过上面代码可知，类库里面最后调用的是windows 的API :BitBlt(),这就是我们要说的第二种方法。
2、平台调用，即调用windows的API函数。其中要用到的最主要的就是BitBlt()。在此，本人先给各位推荐个网站，它就是：www.pinvoke.net，一个包含几乎所有API的wiki,里面一般还含有示例代码~为了便于没有网络的情况下查询，本人已经把该网站拷贝下来了。。耗费了一天时间做成CHM。。。
先来看看BitBlt()的description:The BitBlt function performs a bit-block transfer of the color data corresponding to a rectangle of pixels from the specified source device context into a destination device context.
这里，我们只关心它的最后一个参数/// <param name="dwRop">A raster-operation code.</param>
中文翻译大概叫作：光栅操作码。大概就是控制从source device context 到 destination device context的拷贝方式。它是一个DWORD类型，取值在
// Summary:
// Determines how the source color in a copy pixel operation is combined with
// the destination color to result in a final color.
public enum CopyPixelOperation
{
// Summary:
// The bitmap is not mirrored.
NoMirrorBitmap = -2147483648,
//
// Summary:
// The destination area is filled by using the color associated with index 0
// in the physical palette. (This color is black for the default physical palette.)
Blackness = 66,
//
// Summary:
// The source and destination colors are combined using the Boolean OR operator,
// and then resultant color is then inverted.
NotSourceErase = 1114278,
//
// Summary:
// The inverted source area is copied to the destination.
NotSourceCopy = 3342344,
//
// Summary:
// The inverted colors of the destination area are combined with the colors
// of the source area using the Boolean AND operator.
SourceErase = 4457256,
//
// Summary:
// The destination area is inverted.
DestinationInvert = 5570569,
//
// Summary:
// The colors of the brush currently selected in the destination device context
// are combined with the colors of the destination are using the Boolean XOR
// operator.
PatInvert = 5898313,
//
// Summary:
// The colors of the source and destination areas are combined using the Boolean
// XOR operator.
SourceInvert = 6684742,
//
// Summary:
// The colors of the source and destination areas are combined using the Boolean
// AND operator.
SourceAnd = 8913094,
//
// Summary:
// The colors of the inverted source area are merged with the colors of the
// destination area by using the Boolean OR operator.
MergePaint = 12255782,
//
// Summary:
// The colors of the source area are merged with the colors of the selected
// brush of the destination device context using the Boolean AND operator.
MergeCopy = 12583114,
//
// Summary:
// The source area is copied directly to the destination area.
SourceCopy = 13369376,
//
// Summary:
// The colors of the source and destination areas are combined using the Boolean
// OR operator.
SourcePaint = 15597702,
//
// Summary:
// The brush currently selected in the destination device context is copied
// to the destination bitmap.
PatCopy = 15728673,
//
// Summary:
// The colors of the brush currently selected in the destination device context
// are combined with the colors of the inverted source area using the Boolean
// OR operator. The result of this operation is combined with the colors of
// the destination area using the Boolean OR operator.
PatPaint = 16452105,
//
// Summary:
// The destination area is filled by using the color associated with index 1
// in the physical palette. (This color is white for the default physical palette.)
Whiteness = 16711778,
//
// Summary:
// Windows that are layered on top of your window are included in the resulting
// image. By default, the image contains only your window. Note that this generally
// cannot be used for printing device contexts.
CaptureBlt = 1073741824,
}
里面。
现在继续回来说截取透明窗体的问题，我们知道它跟dwRop参数有关，在codeproject上面可以找到的源码，调用BitBlt()时，都是这样的：
BitBlt( hMemDC, 0, 0, size.cx, size.cy, hDC, 0, 0, PlatformInvokeGDI32.TernaryRasterOperations.SourceCopy);
我想一般的屏幕截图软件也是这样做的，这样做的话，你是不能截取到透明窗体的，我们应该改成这个样子：
BitBlt( hMemDC, 0, 0, size.cx, size.cy, hDC, 0, 0, (uint)(PlatformInvokeGDI32.TernaryRasterOperations.SourceCopy | PlatformInvokeGDI32.TernaryRasterOperations.CaptureBlt) );
通过试验可知是实现了预期效果。CaptureBlt的作用在上面~
需要注意的是，当我们使用BitBlt的时候，涉及设备上下文句柄操作，需要用到另外几个API函数来创建和释放资源。贴个例子：
GetDesktopImage
public static Bitmap GetDesktopImage()
{
//In size variable we shall keep the size of the screen.
SIZE size;
//Variable to keep the handle to bitmap.
IntPtr hBitmap;
//Here we get the handle to the desktop device context.
IntPtr hDC = PlatformInvokeUSER32.GetDC(PlatformInvokeUSER32.GetDesktopWindow());
//Here we make a compatible device context in memory for screen device context.
IntPtr hMemDC = PlatformInvokeGDI32.CreateCompatibleDC(hDC);
//We pass SM_CXSCREEN constant to GetSystemMetrics to get the X coordinates of screen.
size.cx = PlatformInvokeUSER32.GetSystemMetrics(PlatformInvokeUSER32.SM_CXSCREEN);
//We pass SM_CYSCREEN constant to GetSystemMetrics to get the Y coordinates of screen.
size.cy = PlatformInvokeUSER32.GetSystemMetrics(PlatformInvokeUSER32.SM_CYSCREEN);
//We create a compatible bitmap of screen size using screen device context.
hBitmap = PlatformInvokeGDI32.CreateCompatibleBitmap(hDC, size.cx, size.cy);
//As hBitmap is IntPtr we can not check it against null. For this purspose IntPtr.Zero is used.
if (hBitmap!=IntPtr.Zero)
{
//Here we select the compatible bitmap in memeory device context and keeps the refrence to Old bitmap.
IntPtr hOld = (IntPtr) PlatformInvokeGDI32.SelectObject(hMemDC, hBitmap);
//We copy the Bitmap to the memory device context.
PlatformInvokeGDI32.BitBlt( hMemDC, 0, 0, size.cx, size.cy, hDC, 0, 0, (uint)(PlatformInvokeGDI32.TernaryRasterOperations.SRCCOPY | PlatformInvokeGDI32.TernaryRasterOperations.CAPTUREBLT) );
//We select the old bitmap back to the memory device context.
PlatformInvokeGDI32.SelectObject(hMemDC, hOld);
//We delete the memory device context.
PlatformInvokeGDI32.DeleteDC(hMemDC);
//We release the screen device context.
PlatformInvokeUSER32.ReleaseDC(PlatformInvokeUSER32.GetDesktopWindow(), hDC);
//Image is created by Image bitmap handle and stored in local variable.
Bitmap bmp = System.Drawing.Image.FromHbitmap(hBitmap);
//Release the memory to avoid memory leaks.
PlatformInvokeGDI32.DeleteObject(hBitmap);
//This statement runs the garbage collector manually.
GC.Collect();
//Return the bitmap
return bmp;
}
//If hBitmap is null retunrn null.
return null;
}
可见虽然直接的平台调用相比Graphics.CopyFromScreen()速度快，灵活，但是还是相当繁琐的。涉及到句柄操作。这些操作在Graphics.CopyFromScreen()已经替我们做了。
下面来说第三种方法：