[Qucs-devel] Re: proposed xfigfrag patch for psfrag replacement in pdflatex

[Eric C. de V. <Eri...@en...>]

Hello Transfig team, Raimund, and Qucs team!

Raimund 'Raimi' Jacob <ra...@lk...> wrote:

[ looking for a way to use psfrag-based formulas in xfig images with 
pdflatex ]
> Looking for a solution we found Eric's great approach which we think is 
> the way to go: http://www.di.ens.fr/~colin/goodies/xfigfrag.html
>
> (To be honest we have to add that the current implementation could
> be improved. All text attributes (font, size, angle, justification,
> color, etc) are lost - which is a problem for us already.)
>
> For this solution to work one requires a patched fig2dev program.

Let me describe in more details how the thing works.  I added drivers in 
fig2dev that act in a similar way as pstex and pstex_t:

(1) The drivers `epsfrg' and `pdffrg' export the picture with all the text 
removed.  This is the same as the driver pstex or pdftex, except that, in 
these drivers, the non-special text is shown.  This could also be done on 
all current fig2dev drivers by specifying an option saying `Don't export 
any text of this figure'.

(2) The driver `frg' creates a LaTeX-readable description of the bounding 
box of the figure, together with the information of the text objects (text 
itself, location, and attributes).  This is more or less what the pstex_t 
or pdftex_t drivers achieve, except that:
    - these drivers only describe special text;
    - they really include LaTeX code that creates a box of appropriate size 
and inserts the text.  What I need is an output like
       \@xfigfrag@init{u}{b1}{b2}{b3}{b4}, where u is the unit
           length and b1...b4 are the bounding box
       \@xfigfrag@put{x}{y}{rot}{justif}{text} (ideally also with the other
           attributes, hidden, special, scaling, etc.), for each text on
           the figure.

Then a LaTeX style xfigfrag.sty defines these \@xfigfrag@init and 
\@xfigfrag@put commands, and, using hacks in graphicx.sty, everything 
works.

I'd be happy if this code were included in xfig.  But I know that it can 
clearly be improved; in particular, the \@xfigfrag@put command should 
really take as argument all the attributes of the text, so that these 
attributes can be used by the xfigfrag package if desired.

Any comment is welcome!

Best regards,

             Eric