[XeTeX] XeTeX 1.0 - request for comments

[=?GB2312?Q?Pavel_Stra=A8=BE=A8=A2k?=]

On 20.10.2005, at 0:11, Jonathan Kew wrote:

>> In CsLaTeX I can say \csprimeson and \csprimesoff to typeset Czech 
>> quotation marks.
>> In XeLaTeX with Fontspec I would love "mapping=tex-text" to change 
>> the behavoir depending on locale specified by babel or \XeTeXlocale 
>> "cs", or something like that. CsLaTeX uses \usepackage{czech} for 
>> this.
>
> You can create alternate font-mapping files that change the quotation 
> marks to different characters; "tex-text" is not in any way built in 
> to the XeTeX program, it's just a mapping file created with TECkit 
> (http://scripts.sil.org/teckit) and stored in 
> texmf/fonts/misc/xetex/fontmapping. So create a variant called 
> "tex-text-cs.tec" and then use fontspec to load this mapping when 
> typesetting Czech.

I have looked at the mapping tex-text, it looks simple enough. I have 
just installed TECkit.
The only real problem in my Czech mapping is this:
I need to happen only when not between two letters. The apostroph 
should be U+2019
U+0027			<>	U+2018	; ' -> Czech right single quote

I will read the documentation and see whether there is some option to 
specify rules via regex.
It would allow me to make even rules like this (perl regex): 
s/(\d+)%/$1\x{200A}%/g
In other words, Czech typography insists on "20\,%" etc.
I am not sure that U+200A is equal to "\," but I just wanted to make my 
point.
Another case like this is the rule stating that single-letter 
prepositions should not be last words on the line. We use external 
program called "vlna" ("wave") to input non-braking space (~) before 
such prepositions in tex source.

Anyway, I am sure I will soon have at least some solution soon. Thanks 
for the advice.

Pavel