Question

I’m still early on in learning XML schema, so hopefully this isn’t too easy or naive a question.
_{(I’m perusing the specifications, and various other resources, but if anyone can suggest further reading, I’m more than open to suggestion)}

Anyway, what I’m after doesn’t appear to be possible with my novice understanding, so I’m hoping someone can correct me (thus proving it possible)

I’m trying to create a complex type that can contain either any simple type, or another node of the containing complex type, but not both; thus (with the addition of a string “key” attribute) mimicking a multidimensional associative array.

Problem is, I can’t seem to figure out how to allow a choice between the two (of course, I can’t seem to figure out if this is possible either, though my reading is leading me to believe not)

So, essentially, what would I need to do in order to have a target document validate with:

<node key="string">
    <node key="string">
        arbitrary value
    </node>
    <node key="string">
        arbitrary value
    </node>
</node>

But invalidate with:

<node key="string">
    <node key="string">
        arbitrary value
        <node key="string">
            arbitrary value
        </node>
    </node>
</node>

(Note the mixed content node, I don’t want to allow that)

This is where I’m at:

<xs:attribute name="key">
    <xs:simpleType>
        <xs:restriction base="xs:string">
            <xs:whiteSpace value="collapse" />
            <xs:pattern value="[a-zA-Z_][a-zA-Z0-9_]*" />
        </xs:restriction>
    </xs:simpleType>
</xs:attribute>
<xs:complexType name="data">
    <!-- what do i do? -->
    <xs:attribute ref="key" use="required" />
</xs:complexType>

If this isn’t in fact possible, I suppose that’s also fine, but this seems like a reasonably common data structure, one which I would think XML would accommodate.

Answer 1

I think you would find the XML easier to process (not just with XSD, but also with XSLT and XQuery) if you changed the design to:

<node key="string">
    <node key="string">
        <value>arbitrary value</value>
    </node>
    <node key="string">
        <value>arbitrary value</value>
    </node>
</node>

It would then be easy to specify in your schema that the content model for node is a choice: either a sequence of nodes, or a single value.