views:

462

answers:

2

I'm trying to implement a very simple XML schema constraint.

The idref attribute on elements of type <batz> should only be allowed to have a value that matches the id attribute on at least one element <bar>.

If that doesn't make any sense to you then please just look at the example XML-document below, I think it actually explains it better than my attempt to put it in words.

So, question: Why does xmllint let the below schema/xml combination pass (it says the document is valid)? How to fix it to achieve the desired constraint?

The schema:

<?xml version="1.0"?>
<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema"
    targetNamespace="test" xmlns="test" elementFormDefault="qualified">

    <xs:element name="foo">
        <xs:complexType>
            <xs:sequence>
                <xs:element name="bar" minOccurs="0" maxOccurs="unbounded">
                    <xs:complexType>
                        <xs:attribute name="id" use="required" type="xs:string" />
                    </xs:complexType>
                </xs:element>
                <xs:element name="batz" minOccurs="0" maxOccurs="unbounded">
                    <xs:complexType>
                        <xs:attribute name="idref" use="required" type="xs:string" />
                    </xs:complexType>
                </xs:element>
            </xs:sequence>
        </xs:complexType>

        <xs:key name="ID">
            <xs:selector xpath="./bar" />
            <xs:field xpath="@id" />
        </xs:key>

        <xs:keyref name="IDREF" refer="ID">
            <xs:selector xpath="./batz" />
            <xs:field xpath="@idref" />
        </xs:keyref>

    </xs:element>
</xs:schema>

The document:

<?xml version="1.0"?>
<foo xmlns="test">
    <bar id="1" />
    <bar id="2" />
    <batz idref="1" /> <!-- this should succeed because <bar id="1"> exists -->
    <batz idref="3" /> <!-- this should FAIL -->
</foo>
A: 

Your XML document, as shown, doesn't include a schemaLocation. When an XML document doesn't reference a schema or DTD, it may pass validation simply by being well-formed XML. (This once happened to a co-worker, using a different validator. I think it's a bug that the validator didn't at least give a warning that it was missing a schema or DTD. But I digress.)

Anyway, it should probably be something like:

<?xml version="1.0"?>
<foo
  xmlns="test" <!-- This is bad form, by the way... -->
  xsi:schemaLocation="test /path/to/schema/document"
    <bar id="1" />
    <bar id="2" />
    <batz idref="1" /> <!-- this should succeed because <bar id="1"> exists -->
    <batz idref="3" /> <!-- this should FAIL -->
</foo>
Dan Breslau
Thank you! That was the problem.
razorline
Using this it still validates in certain validators. But of course you need to reference the schema.
Peter Lindqvist
+1  A: 

Even with an assigned schema location this will not work in all parsers.

<?xml version="1.0"?>
<foo xmlns="test"   
     xsi:schemaLocation="test test.xsd"  
     xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"&gt;
    <bar id="1" />
    <bar id="2" />
    <batz idref="1" /> <!-- this should succeed because <bar id="1"> exists -->
    <batz idref="3" /> <!-- this should FAIL -->
</foo>

This will validate as well, because the key is not referencing the target namespace.

Changes that need to be made in the XSD are

<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema"
    targetNamespace="test" 
    xmlns:t="test"
    xmlns="test" elementFormDefault="qualified">

And

<xs:key name="ID">
    <xs:selector xpath="./t:bar" />
    <xs:field xpath="@id" />
</xs:key>

<xs:keyref name="IDREF" refer="ID">
    <xs:selector xpath="./t:batz" />
    <xs:field xpath="@idref" />
</xs:keyref>

For a discussion regarding this behaviour see #1545101

Peter Lindqvist