Re: [docbook] Footnote element, label attributes, DocBook 5

[Nic Gibson <nicg@corbas.net>]

On 7 Jan 2008, at 16:29, Dave Pawson wrote:

> Nic Gibson wrote:
>> Hi
>> I've been working with some data converters to get a large chunk of  
>> the
>> Penguin backlist converted into XML. We've had some feedback from the
>> converters about footnotes. The label attribute for footnotes is  
>> typed as
>> NMTOKEN in the schema. As far as I can see, any other element which  
>> has a
>> label attribute has the type set as 'text'.
>> Have our converters hit a bug here or is there a subtle reason I  
>> haven't
>> worked out?
>> The problem for the converters is that some of our older texts  
>> don't use
>> alphanumerics as footnote labels, they use things like asterisks and
>> crosses. Any ideas or suggestions would be highly appreciated.
>
>
> Quite possibly because it's an id/idref pair which produce the link?
> That requires the nmtoken IIRC.

I hadn't thought of it that way. TDG says 'Identifies the desired  
footnote mark' and I had always assumed a literal label. We use the  
xml:id attribute to generate links and just use the label to generate  
the visible portion of the link.

>
>
> Possibly pre-process your input documents to exchange (unicode  
> oddities) for fn1, fn2 in pairs?

Sadly the majority of our input documents are pdfs containing 300-400  
TIFFs or actual hardcopy.

>
>
> Pre-process with xslt and within any given scope matching your  
> footnotes (say chapters) do the same swap in an identity transform,  
> leaving everything else alone?

See above :(

nic