EM_FORMATRANGE keeps returning smaller ranges

[AriBlack]

I'm using EM_FORMATRANGE to print from a RichEdit control to a printer. However, at a certain point, there are fewer and fewer lines on each page and the number of lines gets progressively smaller. This doesn't happen when I use EM_STREAMOUT on the RichEdit so I know it's the printing code I'm using.

Below is the code I'm currently using. Please note the call to OutputDebugString

  int left = GetDeviceCaps(dc, PHYSICALOFFSETX);
  int top = GetDeviceCaps(dc, PHYSICALOFFSETY);
  int width = GetDeviceCaps(dc, PHYSICALWIDTH);
  int height = GetDeviceCaps(dc, PHYSICALHEIGHT);
  int right = width - (left + GetDeviceCaps(dc, HORZRES));
  int bottom = height - (top + GetDeviceCaps(dc, VERTRES));

  FORMATRANGE format;

  format.hdc = dc;
  format.hdcTarget = dc;
  format.chrg.cpMin = 0;
  format.chrg.cpMax = -1;

  width = (int)(8.5 * 1440); // Letter width in TWIPS
  height = (int)(11 * 1440); // Letter height in TWIPS

  format.rcPage.left = 0;
  format.rcPage.top = 0;
  format.rcPage.right = width;
  format.rcPage.bottom = height;

  bool useMinimal = true;

  if (SendDlgItemMessage(mDialog, IDR_CUSTOMMARGIN, BM_GETCHECK, NULL, NULL) == BST_CHECKED){
    useMinimal = false;

    if (mCustLeft < left || mCustRight < right || mCustTop < top || mCustBottom < bottom){
      if (MessageBox(mDialog, "The custom margins are outside the selected printer's bounds. Would you like to use the anyway? Press No to use the printer's minimum margins.", "Story Folder", MB_YESNO|MB_APPLMODAL) == IDNO){
        useMinimal = true;
      } 
    }
  } 
  
  if (useMinimal){
    //  Minimal margins
    format.rc.left = left;
    format.rc.right = width - right;
    format.rc.top = top;
    format.rc.bottom = height - bottom;
  } else {
    format.rc.left = mCustLeft;
    format.rc.right = width - mCustRight;
    format.rc.top = mCustTop;
    format.rc.bottom = height - mCustBottom;
  }

  SendMessage(hwnd, EM_EXSETSEL, 0, (LPARAM)&format.chrg); // Select everything
  SendMessage(hwnd, EM_EXGETSEL, 0, (LPARAM)&format.chrg); // Fill in the charrange for the whole selection 

  SendMessage(hwnd, EM_SETTARGETDEVICE, (WPARAM)dc, width);
  
  bool success = true;
  int min;
  int page = 0;
  char buffer[100];

  while (format.chrg.cpMin < format.chrg.cpMax && success){
    success = StartPage(dc) > 0;

    if (!success) break;

    ++page;

    min = SendMessage(hwnd, EM_FORMATRANGE, TRUE, (LPARAM)&format);
    
    if (min <= format.chrg.cpMin){
      success = false;
      
      break;
    }
    
    sprintf(buffer, "\r\n(%d): %d", page, min - format.chrg.cpMin);
    OutputDebugString(buffer);

    format.chrg.cpMin = min;
    
    success = EndPage(dc) > 0;

    if (!success) break;
  }

  SendMessage(hwnd, EM_FORMATRANGE, FALSE, NULL);

  return success;

Here is the debug output from the loop:

(1): 1249
(2): 1148
(3): 1365
(4): 1193
(5): 1125
(6): 980
(7): 1053
(8): 1134
(9): 1083
(10): 1313
(11): 1199
(12): 1092
(13): 932
(14): 914
(15): 796
(16): 638
(17): 822
(18): 751
(19): 858
(20): 981
(21): 663
(22): 866
(23): 788
(24): 691
(25): 715
(26): 827
(27): 595
(28): 718
(29): 703
(30): 469
(31): 486
(32): 470
(33): 478
(34): 693
(35): 474
(36): 471
(37): 467
(38): 560
(39): 696
(40): 331
(41): 40

Any help with why my code is doing this would be very much appreciated.

[olivthill2]

Below

format.chrg.cpMin = min;

Try

format.chrg.cpMax = -1;

See http://support.microsoft.com/?scid=k...9860&x=10&y=12

[AriBlack]

Below

format.chrg.cpMin = min;

Try

format.chrg.cpMax = -1;

See http://support.microsoft.com/?scid=k...9860&x=10&y=12

Thank you for your suggestion. In the code I provided please note the lines

  format.chrg.cpMin = 0;
  format.chrg.cpMax = -1;

and

  SendMessage(hwnd, EM_EXSETSEL, 0, (LPARAM)&format.chrg); // Select everything
  SendMessage(hwnd, EM_EXGETSEL, 0, (LPARAM)&format.chrg); // Fill in the charrange for the whole selection

The final call to SendMessage fills in the CHARRANGE with the true character indexes for the RichEdit. Using 0 and -1 is a simply a way of avoiding having to find the absolute ranges. However, I made the change you suggested but it gave the same results.

[AriBlack]

Well, I discovered the source of the problem but not why it's happening. For fun, I watched the values of the FORMATRANGE struct that I pass to EM_FORMATRANGE after each call to SendMessage. As it turns out, the value of rc.bottom keeps shrinking and not by a consistent amount.

I updated the loop to the following code and this has corrected the problem but I can't imagine why this would happen in the first place. Any ideas?

  RECT saveRect;

  memcpy(&saveRect, &format.rc, sizeof(RECT));

  while (format.chrg.cpMin < format.chrg.cpMax && success){
    if (!(StartPage(dc) > 0)){
      OutputDebugString("\r\nPrintLoop failure on StartPage");

      success = false;
      
      break;
    }

    min = SendMessage(hwnd, EM_FORMATRANGE, TRUE, (LPARAM)&format);

    memcpy(&format.rc, &saveRect, sizeof(RECT));
    
    if (min <= format.chrg.cpMin){
      success = false;
      
      break;
    }
    
    ++page;

    sprintf(buffer, "\r\n(%d): %d", page, min - format.chrg.cpMin);
    OutputDebugString(buffer);

    format.chrg.cpMin = min;
    
    if (!(EndPage(dc) > 0)){
      OutputDebugString("\r\nPrintLoop failure on StartPage");

      success = false;
      
      break;
    }
  }

[AriBlack]

For anyone coming to this thread now, my "band-aid" solution was actually the correct solution. This behaviour is part of EM_FORMATRANGE.

Here's the answer I received from the MSDN forums:

EM_FORMATRANGE will set rc.bottom within the FORMATRANGE struct to be the
bottom of the area actually printed to, so this must be reset between calls
to make sure that the page boundaries are consistent between the pages

http://osdir.com/ml/wine-cvs/2010-01/msg00497.html