xml-dev - Re: required data

[Rick JELLIFFE <ricko@geotempo.com>]

Craig Miller wrote:
> 
> I currently have a dtd as follows:
> 
> <?xml version="1.0" encoding="UTF-8"?>
> <!ELEMENT authentication (sessionid, serialnumber, username)>
>         <!ELEMENT sessionid (#PCDATA)>
>         <!ELEMENT serialnumber (#PCDATA)>
>         <!ELEMENT username (#PCDATA)>
> 
> The xml file looks like:
> 
> <?xml version="1.0" encoding="UTF-8"?>
> <!DOCTYPE authentication SYSTEM "auth.dtd">
> <authentication>
>         <sessionid>12345</sessionid>
>         <serialnumber>67890</serialnumber>
>         <username>Hello</username>
> </authentication>
> 
> I was wondering if there is a way to make the data required, not just the
> tag.  For example, if I remove <sessionid>12345</sessionid> from the xml
> file then it fails, as it should.  I would however like it to fail if the
> tag looks like this also <sessionid></sessionid>.  Is there a way to mark
> the dtd to fail if the value is NULL.
 
The mania against DTDs has made developers disinclined to provide proper
notation handling in XML systems.  This is how it could work with DTDs :

<!NOTATION non-empty SYSTEM "javascript:currentNode.value.length() > 0;
">
<!ATTLIST session
   type NOTATION (non-empty) "non-empty" >

but then you need 
1) a simple dispatcher to check the current node against notation, as
part of validation.
2) a javascript interpreter (or whatever).

If you use Schematron schema language at
http://www.ascc.net/xml/resource/schematron/schematron.html
you can do
 <rule context="sessionid">
   <assert test="srting-length(text()) &gt; 0" />
 </rule>
and there are matches for specifying that certain characters can or
cannot be found in the contents too.

XML Schemas will be pretty good for this: you can specify regular
expressions for strings and numbers in ranges.

Rick Jelliffe