Some time ago, Will Partain <partain@mekb2.sps.mot.com> wrote:
> Folks, I wasn't pleased with my initial skirmish with XML::Parser,
> trying to do the task below, so I thought I would ask how people here
> would go about it. (I had no difficulty with clunky solutions, but
> nothing even slightly elegant sprang from my wee brain.)
> 
> The input is from an xml-as-interchange-format world; imagine I have
> records something like...
> 
>     <vendorinfo>
>       <id> netscape.com
>       </id> 
>       <name> Netscape
>       </name> 
>       <status> active
>       </status> 
>       <packages> communicator-4.07, communicator-4.5, navigator-4.07
>       </packages> 
>     </vendorinfo>
> 
> I want some code to go slurp!, such that I end up with a VendorInfo
> object so that, in the rest of my code, I can just
> do... $vendor->name(), $vendor->packages(), ...
> 
> The slurping code (what I'm missing) needs to do the obvious things:
> well-formedness, check that <name> isn't given twice, ...
> 
> Actually, it should all be abstracted up a level, given that I'll
> have many more types of "infos", besides just vendors.  So I'd
> really like a slurper that, given a list of mandatory and optional
> tags, would give me back what I wanted. Well, you get the
> picture... Ideas?

use XML::DT;

%handler = ( -default => sub{$c},                # $c has the contents...
             packages => sub{ [ split(",",$c)] },
             -type    => { vendorinfo => 'MAP' }
           );

$a = dt("vendorfilename",%handler);

which makes $a a ref to

   { 'name' => ' Netscape ',
     'status' => ' active ',
     'id' => ' netscape.com',
     'packages' => [ ' communicator-4.07',
                     ' communicator-4.5',
                     ' navigator-4.07 ' ] };


Other possible types: SEQ, STR, MULTIMAP

More information in pod documentation if the module.
   http://www.di.uminho.pt/~jj/perl/XML/DT.html

Feedback about this would be precious

Um abraco
J.Joao