Re: Formatting DocBookXML with Lout / Latin-1 and Latin-2

[Yves Forkl]

Michael Piotrowski <address@hidden> writes:

> By switching between the Latin 1-encoded and
> the Latin 2-encoded fonts you can mix Latin 1 and Latin 2 characters in
> a single document, no problem.  You could extend this to, say, Cyrillic
> as well.
>
> You probably shouldn't use a single byte to represent two different
> characters, e.g., assuming your initial font is Times (i.e., Latin
> 1-encoded):
>
>   { TimesCE Base } @Font { £ód¼ } Fænø
>
> and write this instead:
>
>   { TimesCE Base } @Font { address@hidden address@hidden address@hidden zacute} 
> } Fænø

Hi,

that is fine for the "Base" face, but getting italics is a bit more 
tricky as soon as you are generating Lout code from XML rather than 
keying it in. Say I have the above Polish city name in an emphasis 
element like this:

<para><emphasis>The city of &#x0141;&#xf3;d&#x017A; is in 
Poland.</emphasis></para>

By just replacing the special characters as shown above this would leave 
me with something like

@PP address@hidden {The city of { TimesCE Base } @Font { address@hidden address@hidden 
address@hidden zacute} } is in Poland.} }

which of course does not work, the correct code being, I suppose,

@PP address@hidden {The city of} { TimesCE Slope } @Font { address@hidden address@hidden 
address@hidden zacute} } @I {is in Poland.} }

Generating this Lout code from the above XML sample, however, requires 
analyzing the contents of each text node and conditionally transforming 
its substrings, to split up the @I object as needed. That is very hard 
to do, especially in XSLT 1. Even with the string-matching functionality 
of XSLT 2 it would be quite a lot of work.

If all of the Latin-2 words were wrapped into an element of their own (a 
preprocessor could be used to do this kind of post-tagging), this would 
at least simplify the transformation, as it would allow to focus on that 
element (and its neighbouring text nodes).

Some ideas on this problem?

  Yves