[XeTeX] anti-xunicode ;-)

[Ross Moore]

Hi Jonathan, François, and others,

On 22/07/2006, at 12:12 AM, Jonathan Kew wrote:

> On 21 Jul 2006, at 2:58 pm, François Charette wrote:

> I'd be pretty wary of fallbacks that involve getting glyphs from
> another font;

Agreed, about getting from another font, automatically.
The author of a document needs to have explicit input
into handling awkward characters that way.

> what I think is much more useful, in general, is the
> ability to fall back to the use of \accent, etc., when the font
> doesn't include all the precomposed glyphs (or full positioning
> support so that decomposed Unicode sequences of base + diacritic(s)
> render well).

  xunicode.sty  currently implements fall-backs to combining characters
when the precomposed glyph isn't declared to be available.

You can use the following LaTeX coding to test this feature.

\documentclass{article}
\usepackage{fontspec}
\usepackage{xunicode}
\setromanfont[Mapping=tex-text]{Lucida Grande}
\begin{document}
\"u
\UndeclareUTFcomposite[U]{x00FC}{\"}{u}
\"u
\end{document}

Here the first \"u gets set as  Ux00A8
while the 2nd gives the combination  u Ux0308 .


  xunicode.sty  could (and perhaps should) be modified to also include
a check for the precomposed glyph, then use the fall-back if not found.


But this thread is about having yet another kind of fall-back,
whereby TeX positions the accent-glyph over a letter, as it
used to do, and still does with OT1 encoding.

Here is a way to achieve this.


\documentclass[11pt,a4paper]{article}
\usepackage{fontspec}
\usepackage{xunicode}
\setromanfont[Mapping=tex-text]{Lucida Grande}
\makeatletter
   \def\changefencoding#1{\def\cf at encoding{#1}}
\makeatother
\let\realaccent\"   %%  this is important ...
   \DeclareTextAccent{\"}{UX}{"00A8}
\let\"\realaccent   %%  ... to avoid an infinite loop
\begin{document}
{%\tracingall
\"u
\UndeclareUTFcomposite[U]{x00FC}{\"}{u}
\"u
\changefencoding{UX}%
\"u
}
\end{document}


Now the 3rd instance of \"u is set by placing Ux00A8
appropriately above the letter u .


One way to check these claims is to uncomment before
\tracingall , and check the .log file  or Console window.
There you can find that the 3 instances result in,
respectively:

1st instance:
\\U\"-u ->\char "00FC\relax


2nd instance:
\@secondoftwo #1#2->#2
#1<-\\U\"-u
#2<-\add at encoded@accent {0308}{u}{00A8}

\add at encoded@accent #1#2#3->\ifx \relax #2\relax \char "#3\else \ifx  
\ #2\relax
\char "#3\else \expandafter \ifx \UTF at space #2\relax \char "#3\else  
\ifx ~#2\c
har "#3\else #2\char "#1\fi \fi \fi \fi \relax
#1<-0308
#2<-u
#3<-00A8
(... ultimately gets to the final branch:  #2\char "#1  )


3rd instance:
\UX\" ->\add at accent {"00A8}

\add at accent #1#2->\hmode at bgroup \let \hmode at start@before at group  
\@firstofone \se
tbox \@tempboxa \hbox {#2\global \mathchardef \accent at spacefactor  
\spacefactor
}\accent #1 #2\egroup \spacefactor \accent at spacefactor
#1<-"00A8
#2<-u

Note that this latter method *is* the OT1 method of placing
accents.



For accents that "go under", as with \d  (dot-under accent),
there is a slight variation, since the OT1 method doesn't
use \add at accent  for these.


However, this works:

\let\realaccent\d
\DeclareTextAccent{\d}{UX}{"002E}
\expandafter\let\csname UX\string\d\expandafter\endcsname\csname OT1 
\string\d\endcsname
\let\d\realaccent



>
> This could be a little like a generalized extension of MLTeX's
> \charsubdef, which in theory exists in xetex, but is not fully
> Unicode-enabled. This is something I'd like to enhance at some point,
> as suggested by Mojca, but I can't offer a timescale of when it might
> happen.

The above examples show that XeLaTeX already has the ability to do
what the OP requested; namely to have a sequence of fall-backs
available for accented characters, according to what is available
in the font, or font-encoding.
It just takes a little bit of careful exploring to find the best
way to hook into LaTeX's font-handling mechanism.


If you want the same ability in (Plain) XeTeX or XeConTeXt
well, that's a different story.


>
> JK


Hope this helps,

	Ross

------------------------------------------------------------------------
Ross Moore                                         ross at maths.mq.edu.au
Mathematics Department                             office: E7A-419
Macquarie University                               tel: +61 +2 9850 8955
Sydney, Australia  2109                            fax: +61 +2 9850 8114
------------------------------------------------------------------------