I am trying to draw a word-wrapped string within centered both vertically and horizontally within a bitmap using WinAPI's DrawText function.
The problem is that if text is longer than the available space and "END ELLIPSIS" (...) is added to a cropped string, the reported drawing coordinates returned when using the "DT_CALCRECT" report the uncropped number of lines which messes with the vertical centering calculations.
I read many posts on this, and thought that "Delphi - Draw text multiline in the centre of a rect" may hold the answer, but it didn't (screenshot of the code output using the sample in the linked question http://zoomplayer.com/pix/font_vcenter.jpg). The author of the accepted answer suggested I create a new question so here it is.
For quick-reference, here is a slightly simplified (removing unrelated code) text rendering code from the linked accepted answer:
procedure DrawTextCentered(Canvas: TCanvas; const R: TRect; S: String);
var
DrawRect: TRect;
DrawFlags: Cardinal;
begin
DrawRect := R;
DrawFlags := DT_END_ELLIPSIS or DT_NOPREFIX or DT_WORDBREAK or
DT_EDITCONTROL or DT_CENTER;
DrawText(Canvas.Handle, PChar(S), -1, DrawRect, DrawFlags or DT_CALCRECT);
DrawRect.Right := R.Right;
if DrawRect.Bottom < R.Bottom then
OffsetRect(DrawRect, 0, (R.Bottom - DrawRect.Bottom) div 2)
else
DrawRect.Bottom := R.Bottom;
DrawTextEx(Canvas.Handle, PChar(S), -1, DrawRect, DrawFlags, nil);
end;
As you can see from the screenshot, the problem is after the initial call to DrawText with the "DT_CALCRECT" flag to measure the output height for later vertical centering, rendering the string "Trending in: Worldwide" returns a DrawRect.Bottom value representing 3 lines of text even though only 2 lines are drawn, breaking the vertical centering code.
Related
I have wrapping NSTextView instances stacked vertically, for example:
The quick brown fox
jumps over the lazy dog
Jackdaws love my big
sphinx of quartz
I need to move between them with up/down arrows. For example, when the cursor is positioned after the l in "lazy" and the user presses the down arrow, the cursor should jump right after the y in "my" – like it would do if these sentences were in the same text view.
By default, when the down arrow is pressed while the cursor is at the last wrapped line, a text view moves it to the end of that line. While I can use textView(_:doCommandBy:) in NSTextViewDelegate to detect the "arrow down" selector and override the default behavior, there are two problems:
I can determine if the cursor is at the last line by getting its position via the selectedRanges property and then checking for the newline character after this position, but it is not possible to know if it is at the last wrapped line, i.e. near the border of the current text view.
I need to know the X coordinate of the cursor to place it at approximately the same X coordinate in another text view (the fonts won't necessarily be fixed-width, so I can't rely on the character count).
I suppose both of them can be resolved via NSLayoutManager, but I can't wrap my head around all of its available methods.
It turned out to be relatively easy, here's what I've done (the examples are in C#). First, boundingRect(forGlyphRange:in:) gets the cursor's location in the current view:
var cursorLocation = new NSRange(CurrentTextView.SelectedRange.Location, 0);
var cursorCoordinates = CurrentTextView.LayoutManager.BoundingRectForGlyphRange(cursorLocation, CurrentTextView.TextContainer).Location;
Then if the second text view is below, the insertion point will be at 0 on the Y axis:
var insertionPoint = new CGPoint(cursorCoordinates.X, 0);
And if it is above, then another view's height should be used (reduced by 1, otherwise the resulting character index will be incorrect and the cursor will be placed at the end of the line):
var insertionPoint = new CGPoint(cursorCoordinates.X, AnotherTextView.Bounds.Size.Height - 1);
After getting the insertion point, another view needs to become the first responder and then characterIndexForInsertion(at:) does the job:
Window.MakeFirstResponder(AnotherTextView);
var index = AnotherTextView.CharacterIndex(insertionPoint);
AnotherTextView.SelectedRange = new NSRange(index, 0);
I need help from those who know the pascal very well.
I need (teachers requirement) to use a variant record. Those variant record are defined like this:
Temperature = record
case scale : TemperatureScale of
celsius : (celsius_value : ScaleCelsius);
kelvin : (kelvin_value : ScaleKelvin);
end;
According to primary sources I found during my research about this topic, I could not find how to use variant record, only how to declare it.
My primary sources: google search(found nothing), Case-Freepascal and Record-Freepascal
Edit for those who wonder what i must use in implementation it is exacly this:
type
UkPolozka = ^Polozka;
UkHodnota = ^Hodnota;
TypUdaj=(typretez, typcele, typrealne, typlogik, typpole, typobjekt);
VarZaznam = record
case Udaj: TypUdaj of
typretez: (retez: string);
typcele: (cele: word);
typrealne: (realne: single);
typlogik: (logik: boolean);
typpole: (pole: UkHodnota);
typobjekt: (objekt: UkPolozka);
end;
Polozka = record
Nazev: string;
Hodn: VarZaznam;
Dalsi: UkPolozka
end;
Hodnota = record
Hodn: VarZaznam;
Dalsi: UkHodnota
end;
Consider example record from Delphi Help. I modified it with tag.
Tag usage is optional and usually is not used and doesn't provide useful information.
Note that the first line outputs radius with the same value as height - they share the same memory.
Then I explicitly set tag (but fields stay the same)
Also note size difference for no-tag record and tagged one.
Note again - in most cases programmers don't define or use tag value. It might be unsafe to rely on tag (until you follow strict rules in assigning tags). I did not set rectangle tag, but program consider zero tag as the first variant.
Just assign needed field by name. Just read needed field by name. For your example varVarZaznam.retez:= 'test';
type
TShapeList = (Rectangle, Triangle, Circle, Ellipse, Other);
TFigure = record
case shape: TShapeList of
Rectangle: (Height, Width: Real);
Triangle: (Side1, Side2, Angle: Real);
Circle: (Radius: Real);
Ellipse, Other: ();
end;
TFigureNoTag = record
case TShapeList of
Rectangle: (Height, Width: Real);
Triangle: (Side1, Side2, Angle: Real);
Circle: (Radius: Real);
Ellipse, Other: ();
end;
var
Fig: TFigure;
begin
Fig.Height := 5;
Fig.Width := 3;
//default tag is zero = Rectangle
if Fig.shape = Rectangle then
Writeln('Rect ', Fig.Height:3:0, Fig.Width:3:0, Fig.Radius:3:0)
else
Writeln('not Rect');
Fig.shape := Circle;
if Fig.shape <> Rectangle then
Writeln('not Rect');
Writeln('Fig size ', SizeOf(TFigure), ' noTag size ', SizeOf(TFigureNoTag));
Readln;
output in Delphi:
Rect 5 3 5
not Rect
Fig size 32 noTag size 24
Given the following code snippet:
procedure TPicture.PaintLine(_Canvas: TCanvas; _Left, _Top, _Right, _Bottom: Integer);
begin
IntersectClipRect(_Canvas.Handle, _Left, _Top, _Right, _Bottom);
try
_Canvas.MoveTo(_Left - 10, _Top - 10);
_Canvas.LineTo(_Right + 10, _Bottom + 10);
// (This is an example only, the actual drawing is much more complex.)
finally
SelectClipRgn(_Canvas.Handle, 0); // This does too much
end;
end;
I want to undo the clipping effected by the call to IntersectClipRect so the previously active clipping becomes active again. In the above code, this is done by SelectClipRgn(...,0) which turns off clipping altogether. This works, kind of, but afterwards there is no clipping active so any drawing that is executed after the above will paint to areas that should not be painted to.
So, what is the correct way to undo only the effect of IntersectClipRect?
EDIT: Removed the unnecessary CreateRectRgn and DeleteObject code after I understood the comment from Sertac, to make the question more readable for others that might stumble upon it later.
You can save and restore the state of the DC:
var
// RGN: HRGN;
SavedDC: Integer;
begin
// RGN := CreateRectRgn(_Left, _Top, _Right, _Bottom);
SavedDC := SaveDC(_Canvas.Handle);
try
IntersectClipRect(_Canvas.Handle, _Left, _Top, _Right, _Bottom);
_Canvas.MoveTo(_Left - 10, _Top - 10);
_Canvas.LineTo(_Right + 10, _Bottom + 10);
// (This is an example only, the actual drawing is much more complex.)
finally
RestoreDC(_Canvas.Handle, SavedDC);
end;
...
IIRC, first store the current clip region using GetClipRgn, and after you're done, SelectClipRgn the stored region again.
Looking at your code, it should be enough to SelectClipRgnyour RGN again, because:
The IntersectClipRect function creates a new clipping region from the intersection of the current clipping region and the specified rectangle.
I have a simple chart using only TLineSeries with a single Y axis. As the mouse pointer moves over the chart, I’d like to get the X and Y values associated with the pointer’s position independent of any series.
I can handle the MouseMove event and get the screen X and Y coordinates, but the only way I’ve found to convert them is via the Series->XValues->Locate and Series->YValue->Locate methods.
There are two problems with this:
1 - The value returned from Series->YValue->Locate is always -1 regardless of whether the pointer is over a series line or not.
2 – The value returned from Series->XValue->Locate is -1 unless the pointer is over a part of the chart containing a series line.
Why does Series->YValue->Locate always return -1?
More importantly, how can I get the values regardless of whether the pointer is over a part of the chart with series lines or not?
I’m using the version of TeeChart that ships with Rad Studio XE3.
Why does Series->YValue->Locate always return -1?
That's because Locate uses a series value and returns its point index in the series. OnMouseMove provides screen pixel coordinates, not series values.
More importantly, how can I get the values regardless of whether the
pointer is over a part of the chart with series lines or not?
You can use axes as a reference instead of series, for example:
procedure TForm2.Chart1MouseMove(Sender: TObject; Shift: TShiftState; X,
Y: Integer);
var
XVal: double;
YVal: double;
begin
XVal:=Chart1.Axes.Bottom.CalcPosPoint(X);
YVal:=Chart1.Axes.Left.CalcPosPoint(Y);
Chart1.Title.Text[0]:=FormatFloat('#.##', XVal) + ' - ' + FormatFloat('#.##', YVal);
end;
i am trying to render RichEdit to a bitmap. I want to make backgrond tranparent.
I could achieve that by the code above. The problem is that when i set bitmap as tranparent
only the transparent color becomes transparent. Is there any way to handle the pixels on the edge
which are diffrent from background color and also different from font color. i mean making the pixels
semi-tranparent on the edge which makes a smooth view.
Graphics::TBitmap *bitmap = new Graphics::TBitmap();
bitmap->Width = RichEdit1->Width ;
bitmap->Height = RichEdit1->Height ;
TRect BoundingBox(0,0,RichEdit1->Width, RichEdit1->Height) ;
// Render RichEdit to bitmap
TFormatRange formatRange;
int twipsPerPixel = 1440 / Screen->PixelsPerInch;
formatRange.hdc = bitmap->Canvas->Handle;
formatRange.hdcTarget = bitmap->Canvas->Handle;
formatRange.chrg.cpMin = 0;
formatRange.chrg.cpMax = -1;
formatRange.rc.top = 2 * twipsPerPixel;
formatRange.rc.bottom = (BoundingBox.Height() - 4) * twipsPerPixel + formatRange.rc.top;
formatRange.rc.left = 2 * twipsPerPixel;
formatRange.rc.right = (BoundingBox.Width() - 4) * twipsPerPixel;
// Measure text's height.
RichEdit1->Perform(EM_FORMATRANGE, 0, 0);
RichEdit1->Perform(EM_FORMATRANGE, 0, (LPARAM) &formatRange);
formatRange.rc.bottom = (BoundingBox.Height() - 4) * twipsPerPixel + formatRange.rc.top;
formatRange.rc.left = 2 * twipsPerPixel;
formatRange.rc.right = (BoundingBox.Width() - 4) * twipsPerPixel;
formatRange.rcPage = formatRange.rc;
/**
* Draw..
**************************************************************************/
RichEdit1->Perform(EM_FORMATRANGE, 1, (LPARAM) &formatRange);
RichEdit1->Perform(EM_FORMATRANGE, 0, 0);
// Draw background
// Use different background color to see the trasparency problem
this->Canvas->Brush->Color = clRed ;
this->Canvas->Rectangle(0,0,RichEdit1->Width , RichEdit1->Height );
// Draw the transparent bitmap
bitmap->Transparent = true ;
bitmap->TransparentColor = RichEdit1->Color ;
this->Canvas->Draw(0,0,bitmap);
Thanx.
Font smoothing works with partial transparency using an alpha channel. The Transparent and TransparentColor properties of TBitmap are therefore not applicable.
You haven't said which version of the C++ Builder/VCL you are using, but more modern versions have better support for partial transparency than some of the older ones.
To get this to work you will need to set the PixelFormat of your bitmap to be pf32bit. You may also need to set AlphaFormat to afDefined.
If you can't get TBitmap to do what you need then you'll have to revert to GDI commands to create a suitable HBITMAP. You can at least assign that to the Handle property of a TBitmap and usually from there everything behaves.
Note that I am not a user of C++ Builder but do know the VCL from Delphi.
UPDATE
I tried this out in Delphi and the following worked fine for me:
procedure TForm4.Button1ClickBMP(Sender: TObject);
var
BMP: TBitmap;
fmtRange: TFormatRange;
intPPI, Flags: Integer;
begin
BMP := TBitmap.Create;
Try
BMP.PixelFormat := pf32bit;
BMP.SetSize(RichEdit1.Width, RichEdit1.Height);
FillChar(fmtRange, SizeOf(fmtRange), 0);
with fmtRange do begin
hDC := BMP.Canvas.Handle;
hdcTarget := hDC;
intPPI := Screen.PixelsPerInch;
rc := Rect(
0,
0,
RichEdit1.Width*1440 div intPPI,
RichEdit1.Height*1440 div intPPI
);
rcPage := rc;
chrg.cpMin := 0;
chrg.cpMax := -1;
end;
Flags := 1;
RichEdit1.Perform(EM_FORMATRANGE, Flags, Longint(#fmtRange));
RichEdit1.Perform(EM_FORMATRANGE, 0, 0);
BMP.SaveToFile('c:\desktop\test.bmp');
Finally
FreeAndNil(BMP);
End;
end;
The output looks like this, blown up somewhat to see the anti-aliasing:
I hope this helps, because it looks like you are very nearly there!
As David has answered already, sub-pixel antialiasing requires knowledge of the background colour. This is explained in more detailed in this answer. In essence, when you do sub-pixel anti-aliasing you treat the three colour channels as having different spatial offsets (which is where the apparent increase in resolution comes from). This means they need different alpha values, but of course there is only one alpha channel.
You can of course do regular full-pixel grayscale antialiasing over a transparent background. Perhaps this would be good enough? Some of the other answers in the question linked above suggest ways to achieve this. Have a look at ANTIALIASED_QUALITY (vs. CLEARTYPE_QUALITY) in the LOGFONT structure. (I haven't tried this.)