Re: [xsl] recognize character entities

[Owen Rees <owen.rees@xxxxxx>]

--On 30 August 2006 07:48 +0200 Frank Marent wrote:

>    <m:mo>&divide;</m:mo>
>
>    to
>
>    <m:mo fontfamily="Symbol">&divide;</m:mo>
>
> this is to address the correct font in the output equation eps files
> done by mfequation server from design science. right now we're taking  a
> sar (search and replace - i like that abbrevation) tool to do that:
>
>    [Search /x]
>    <m:mo>&*[];</mo>
>    [Replace]
>    <m:mo fontfamily='Symbol'>&%1;</mo>
>
> that's working. but means additional work since we have no continuous
> workflow and have to do manual work. an xslt process should identify  all
> m:mo elements that consist of one entity reference. that is the
> background of my question.

Given that you mention EPS and Symbol, do you need the eventual output to 
be the Type 1 Symbol set code for the character as well as being wrapped in 
an element that specifies the 'font'?

If you have some downstream process that translates &divide; to Symbol code 
270 (octal - as given in the back of the PDF reference manual), &Alpha; to 
Symbol code 101 etc. then this is overloading the entity reference syntax 
in a way that is potentially very confusing.

I suspect that the problem here is not so much recognising entity 
references, but that the syntax has been used to represent characters that 
are not to be handled as Unicode code points (so interpreting &divide; as 
meaning U+00F7 will give the wrong result.)


--
Owen Rees
Hewlett Packard Laboratories, Bristol, UK