[reportlab-users] vertically centered text

[Aaron Spike]

Robin Becker wrote:

> The y coordinate of a drawString type operation ends up as being the 

> baseline of the text. Typically text is not vertically symmetric so is 

> not visually centered. 


Yes, and this works great because it is the natural visual anchor point
for most uses of text. But if you want to center text in a tightly
containing shape it is difficult. In my particular application I want to
center text in a circle. If it isn't close to exactly centered it
distracts the eye.


> A good rule of thumb is that the part below the 

> base line is 20% of the point size ie the vertical middle of the text is 

> 40% of the pointsize above the normal base line. 


I tried using some magic numbers like these and found the results
unsatisfactory.


> However, these 

> assumptions are clearly dependent on exactly which glyphs are being 

> used. I know this is not a decent answer, but you can obtain the curves 

> being used for a particular string using 

> reportlab.graphics.charts.textlabels._text2path

> 

> def _text2Path(text, x=0, y=0, fontName='Times-Roman', fontSize=1000,

>                 anchor='start', truncate=1, pathReverse=0):

> 

> 

> the returned path can then be used to obtain a bounding box ie

> 

>  >>> from reportlab.graphics.charts.textlabels import _text2Path

>  >>> p=_text2Path('a min')

>  >>> p.getBounds()

> (34, -14, 2246, 694)

>  >>> p=_text2Path('A min')

>  >>> p.getBounds()

> (10, 0, 2524, 694)

>  >>>

> 

> so you can see that A is taller than a


_text2path looks like exactly what I need. When I'm working with Cairo I
use text_extents to find the bounds of a particular piece of text:

http://cairographics.org/manual/cairo-Scaled-Fonts.html#cairo-text-extents-t 



My output with Cairo looks like:

http://www.ekips.org/misc/arpeggio/all.pdf

I'd like to make my diagrams flowable and use reportlab's layout
features for titles and annotations.

Should I be afraid that _text2path looks like a private part of the API?
or am I free to use that? Is there no public way to get this info? Maybe
something which can look into the current text rendering context?

Aaron Spike