Re: structure of extracted index page

[Dan Mueth <dan eazel com>]

On Wed, 25 Apr 2001, Alexander Kirillov wrote:

> Sorry if I am missing something, but the example given here doesn't
> seem to be legal docbook: AFAIK, there is no such tag as <indexdoc> in
> DocBook, and <indexentry> *must* contain <primaryie>, and there is no
> such attribute as "linkid". Thus, the correct markup would be
> 
> <index> 
> <indexentry> 
>  <primaryie> Apple</primaryie>
>  <secondaryie>Big </secondaryie>
> </indexentry>
> etc. I am not sure what to do with the linking id. 
> 
> Of course, this xml document is only intended as a temporary one to be
> used by help system internally, so in theory we can introduce our own
> tags - but I'd rather try avoiding this. 

Mary is creating a new DTD to describe indexing information which is used
internally to ScrollKeeper, and possibly to a help browser.  It is not
meant to be a part of a document, so it isn't necessary or even
appropriate to use DocBook.

> For example, this would mean we can't use usual (Norman Walsh's)
> stylesheets for xml->html translation, and would also cause many new
> problems.

We would process the DocBook document with this stylesheet when rendering
the document, but the internal data Mary is showing would not be handled
by a DocBook stylesheet.

> More seriously, I'd consider it important that: 
> 
>  a. generated index contains <indexdiv> (which separate entries
>  starting with diffrent letters), if it is not too much of a problem 
> 
>  b. it correctly puts together multiple occurences of a word: if the
>  document contains 
> 
>    <indexterm id="idx-a1">
>      <primary>Apple</primary><secondary>Big </secondary>
>    </indexterm>
> 
>   and 
> 
>    <indexterm id="idx-a5">
>      <primary>Apple</primary><secondary>Small </secondary>
>    </indexterm>
> 
>   then the produced index contains 
> 
>   <indexentry> 
>    <primaryie> Apple</primaryie>
>     <secondaryie>Big </secondaryie>
>     <secondaryie>Small </secondaryie>
>   </indexentry>
>  rather than  two separate entries "Apple, Big" and "Apple, Small"
> 
>  c. it correctly sorts words alphabetically, even for alphabets other
>  than Latin, using "lang" attribute of the document - e.g., if the
>  document is written in Greek as indicated by lang="gr" attribute,
>  then  index should be sorted accroding to Greek alphabet. 

I'm not sure how good the existing DocBook stylesheet is at doing
this.  Perhaps somebody else on this list knows?

Dan