MFC的打印是如何工作的？
Before we go further, we need to learn how OnPrintPreview and OnPrint work in MFC. Here is the scratch of the code.
- Calls
OnPreparePrint to prompt dialog to setup actual printer and number of pages. Usually we will let MFC do the work for us here. But if you want to skip the dialog and setup printer DC directly, you’d better do it here or totally rewrite OnPrint/OnPrintPreview.
- Calls
OnBeginPrint. If we can decide how many page we need to print, it is better to set the number of pages here if possible.
- Print each page in a loop.
- In the loop, Calls
OnPrepareDC. If you need to setup map mode and conversion ratio between logical pixel and physical pixel, such as like one inch in screen equals one inch in printer, you'd better do it here. One important thing to bear in mind is that if you can’t decide how many pages you need to print on step 2, you can still use CPrintInfo::m_bContinuePrinting member variable to terminate printing.
- Calls
OnPrint to do actually printing
打印CFormView有两种方法：
1.Capturing screen image of CFormView.
Like injecting any code into a framework, first you need to know where to add your code. Such type of question is always the toughest one when programming in MFC. In this case, the question is when to grab the image. How about doing it in OnBeginPrint? Not bad idea at first glance. Well, it turns out there is catch here. As MFC prompts a window to emulate Printer DC in preview mode, you could end up capturing wrong image in this mode. It is better to do it in OnFilePrint and OnFilePrintPreview. The actual code looks like this:
void CFormViewPrintView::_grapImage( )
{
//Grap Image
CPoint oldPoint = GetScrollPosition( );
//scroll to top left corner as CFormView is a Scroll view
CPoint pt( 0, 0 );
this->ScrollToPosition( pt );
CClientDC dc(this);
CRect rect;
this->GetClientRect( rect );
m_dib.Attach( GDIUtil::GrabDIB( &dc, rect ) );
ScrollToPosition( oldPoint );
}
void CFormViewPrintView::OnFilePrintPreview()
{
// TODO: Add your command handler code here
_grapImage( );
CFormView::OnFilePrintPreview() ;
}
void CFormViewPrintView::OnFilePrint()
{
_grapImage( );
CFormView::OnFilePrint() ;
}
Hmm, what does the GDIUtil::GradDIB do? It grabs Bitmap from the screen and converts it to DIB. Why DIB, not Bitmap directly? A bitmap always depends on DC and screen DC is different than Printer DC. Without such conversion, we are under the mercy of Printer Driver. It may work fine in some printer, but badly on the other. Seen Roger Allen’s article on this.
Next, we need to deal with how to preserve something the same size as displayed on screen. Ever wondered why something turns terribly small when printing? Here is the reason, let’s say the resolution in printer is 600 pixel per inch, while we usually have 96 or 120 pixel per inch in the screen. If you simply print something “the same size” in pixel, it is not hard to imagine what will happen. That is also the reason why you should change font size when printing text. What we really want, is to print something the same size in inch, not pixel. “Point taken, but where to put the code of such conversion?” You ask yourself and realize this is the same old “where” question again. This can be done by overriding the method OnPrepareDC. What Microsoft really means by the name is “Setup map mode here if needed”. This is also the place to decide whether to terminate printing or not, if you haven’t figured out the number of printing pages previously. Our OnPrepareDC looks like this.
void CFormViewPrintView::OnPrepareDC(CDC* pDC,
CPrintInfo* pInfo /* = NULL */)
{
// TODO: Add your specialized code here and/or call the base class
if( pInfo )
{
CClientDC dc( this );
pDC->SetMapMode(MM_ANISOTROPIC);
CSize sz( dc.GetDeviceCaps(LOGPIXELSX),
dc.GetDeviceCaps(LOGPIXELSY) );
pDC->SetWindowExt( sz );
sz = CSize( pDC->GetDeviceCaps(LOGPIXELSX),
pDC->GetDeviceCaps(LOGPIXELSY) );
pDC->SetViewportExt( sz );
}
}
What does this code mean? It means one inch in screen, dc in this case, equals one inch in printer (could be pseudo one) and we don’t care about actual pixel size varies, say 120 ppi in screen vs 600 ppi in printer.
Last, the actual printing.
void CFormViewPrintView::OnPrint(CDC* pDC, CPrintInfo* pInfo)
{
// TODO: add customized printing code here
if( pInfo == NULL )
return;
if( m_dib.GetHandle( ) == NULL )
return;
{
//Call GlobalLock in constructor, call Unlock when exists the block
GLock lock( m_dib );
BITMAPINFOHEADER *pBMI = (BITMAPINFOHEADER*)(LPVOID)lock;
int nColors = 0;
if( pBMI->biBitCount <= 8 )
nColors = ( 1<< pBMI->biBitCount );
::StretchDIBits( pDC->GetSafeHdc( ),
pInfo->m_rectDraw.left,
pInfo->m_rectDraw.top,
pBMI->biWidth,
pBMI->biHeight,
0,
0,
pBMI->biWidth,
pBMI->biHeight,
(LPBYTE)pBMI + (pBMI->biSize + nColors * sizeof(RGBQUAD)),
(BITMAPINFO*)pBMI,
DIB_RGB_COLORS,
SRCCOPY);
}
}
One thing to mention is that GLock in GUtil follows the same idea as AutoPtr in STD. I have no idea why Microsoft does right thing in CClientDC and CPaintDC, while turning blind when dealing something like GlobalLock/Unlock or the notorious SelectObject. How many times have we scratched our head to detect GDI object resource leak, only finding out that we select something in, but forget to select it out.
2.Another way WM_PRINT message
Ever heard of WM_PRINT message? It is not even in Visual C++ class wizard, but it seems promising everything we need for printing CFormView. Here is another way to print CFormView:
void CFormViewPrint2View::_print( )
{
CRect rect;
this->GetClientRect( rect );
CDC memDC;
CClientDC dc( this );
memDC.CreateCompatibleDC( &dc );
CBitmap bitmap;
bitmap.CreateCompatibleBitmap( &dc, rect.Width(), rect.Height() );
{
//This will force bitmap selected out of DC when exit this block
LocalGDI local( &memDC, &bitmap );
this->Print( &memDC, PRF_ERASEBKGND|PRF_CLIENT|PRF_CHILDREN );
}
m_dib.Attach( GDIUtil::DDBToDIB( bitmap ) );
}
void CFormViewPrint2View::OnFilePrintPreview()
{
// TODO: Add your command handler code here
_print( );
CFormView::OnFilePrintPreview( );
}
void CFormViewPrint2View::OnFilePrint()
{
// TODO: Add your command handler code here
_print( );
CFormView::OnFilePrint( );
}
结论:
So, what is the strength and weakness of each method? The first one doesn’t care about how many individual child controls you have and how to print each of them on Printer, but it can only print visual part of the screen. While second one seems much better and cleaner than the first one, it even allows you to print all client area without displaying them on the screen. Unfortunately, there is a catch for it too. Some sub-classed Windows controls and user custom controls may forget to process WM_PRINT message at all, which is amazingly easy to implement if you can process WM_PAINT message.