You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@pdfbox.apache.org by Branden Visser <mr...@gmail.com> on 2015/11/24 04:40:44 UTC

Getting visual height of a glyph

Hello!

I've seen lots of discussion about getting the height of a glyph,
those of which are saying that using the path accessor of the font
should result in getting an accurate visual height of a glyph, while
close, it has not been my experience.

I've attached a PDF with 7 different TrueType fonts, and I have a
(Scala) program [1] [2] that uses the font path to try and get the
bounding boxes of the characters, and then draws a copy of the PDF
with the boxes drawn around the characters.

You'll see that most boxes are double the height of the glyph.
However, for many other documents I've tried, the path is very
accurate (generally PDType1Font, I think).

I've seen discussion that says that the font size is the height of the
glyph, but that appears to be ~1.5x the height of the capital A's in
the attached document. Same with the CapHeight which IIUC should be
the height of those capital A's, but it seems to be around the same
height as the font size. I've also seen crazy documents where the font
size is 1.0 for all characters, and the size is controlled by the
Y-scale, so I'm not sure using font size for height is a great
alternative.

Presumably, if these letter A's can be displayed to me at this height,
there must be something that would indicate the height of the
characters, but I can't pinpoint what it is. I looked at the
PathIterator instructions, and indeed the upper-most point that is
drawn is one that appears to be exactly double the height of the
glyph.

So, is there a way to get the visual height of a given character? Is
there an element to the height that I'm missing?

Thanks,
Branden

[1] Function to get the height from a character code

private[this] def getCharacterHeightFromPath(font: PDFont, code: Int): Float = {
  val path = font match {
    case f: PDType1CFont => f.getPath(f.getEncoding.getName(code))
    case f: PDType1Font => f.getPath(f.getEncoding.getName(code))
    case f: PDType0Font => f.getPath(code)
    case f: PDType3Font => f.getPath(f.getEncoding.getName(code))
    case f: PDTrueTypeFont => f.getPath(f.getEncoding.getName(code))
  }

  path.getBounds2D.getHeight.toFloat
}

[2] Translate to user space, where `tp` is a TextPosition given by the
PDFTextStripper

val f = tp.getFont
val glyphHeight = getCharacterHeightFromPath(f, tp.getCharacterCodes.head)
val heightUser = f.getFontMatrix.transformPoint(0, glyphHeight).y

val scalingFactorY = tp.getTextMatrix.getScalingFactorY
val heightUserScaled = heightUser * scalingFactorY

Re: Getting visual height of a glyph

Posted by John Hewson <jo...@jahewson.com>.
Hi,

> On 23 Nov 2015, at 19:40, Branden Visser <mr...@gmail.com> wrote:
> 
> Hello!
> 
> I've seen lots of discussion about getting the height of a glyph,
> those of which are saying that using the path accessor of the font
> should result in getting an accurate visual height of a glyph, while
> close, it has not been my experience.
> 
> I've attached a PDF with 7 different TrueType fonts, and I have a
> (Scala) program [1] [2] that uses the font path to try and get the
> bounding boxes of the characters, and then draws a copy of the PDF
> with the boxes drawn around the characters.
> 
> You'll see that most boxes are double the height of the glyph.
> However, for many other documents I've tried, the path is very
> accurate (generally PDType1Font, I think).
> 
> I've seen discussion that says that the font size is the height of the
> glyph, but that appears to be ~1.5x the height of the capital A's in
> the attached document. Same with the CapHeight which IIUC should be
> the height of those capital A's, but it seems to be around the same
> height as the font size. I've also seen crazy documents where the font
> size is 1.0 for all characters, and the size is controlled by the
> Y-scale, so I'm not sure using font size for height is a great
> alternative.
> 
> Presumably, if these letter A's can be displayed to me at this height,
> there must be something that would indicate the height of the
> characters, but I can't pinpoint what it is. I looked at the
> PathIterator instructions, and indeed the upper-most point that is
> drawn is one that appears to be exactly double the height of the
> glyph.
> 
> So, is there a way to get the visual height of a given character? Is
> there an element to the height that I'm missing?
> 
> Thanks,
> Branden
> 
> [1] Function to get the height from a character code
> 
> private[this] def getCharacterHeightFromPath(font: PDFont, code: Int): Float = {
>  val path = font match {
>    case f: PDType1CFont => f.getPath(f.getEncoding.getName(code))
>    case f: PDType1Font => f.getPath(f.getEncoding.getName(code))
>    case f: PDType0Font => f.getPath(code)
>    case f: PDType3Font => f.getPath(f.getEncoding.getName(code))
>    case f: PDTrueTypeFont => f.getPath(f.getEncoding.getName(code))
>  }
> 
>  path.getBounds2D.getHeight.toFloat
> }
> 
That's correct.

> [2] Translate to user space, where `tp` is a TextPosition given by the
> PDFTextStripper
> 
> val f = tp.getFont
> val glyphHeight = getCharacterHeightFromPath(f, tp.getCharacterCodes.head)
> val heightUser = f.getFontMatrix.transformPoint(0, glyphHeight).y

You're trying to transform glyphHeight (a dimension) using a method for transforming points. That won't work.

> val scalingFactorY = tp.getTextMatrix.getScalingFactorY
> val heightUserScaled = heightUser * scalingFactorY

The matrix transforms are 2D, you can't extract just a single scale and use it to transform a point (even if that point is non-zero in only one dimension, because the resulting matrix-transformed point may be non-zero for *both* dimensions, x and y).

I'd recommend creating an AffineTransform from the Matrix, and then call transformShape to transform the glyph's bounding rectangle. Make sure you do all transforms using matrices, don't try to extract the various scaling factors because they can't be applied in isolation.

-- John

> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@pdfbox.apache.org
> For additional commands, e-mail: users-help@pdfbox.apache.org

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@pdfbox.apache.org
For additional commands, e-mail: users-help@pdfbox.apache.org

Re: Getting visual height of a glyph

Posted by Tilman Hausherr <TH...@t-online.de>.
Am 24.11.2015 um 04:40 schrieb Branden Visser:
> I've attached a PDF with 7 different TrueType fonts

Could you please upload it somewhere? PDFs are deleted by the mailing 
list server.

Tilman

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@pdfbox.apache.org
For additional commands, e-mail: users-help@pdfbox.apache.org