Incorrect Validation Error for Substitution Group with Abstract Head

Description:

The xmlschema library incorrectly invalidates XML documents that use elements from a substitution group whose head is abstract. According to the W3C XML Schema specifications, substitution group members can appear in the instance document even if the head element is abstract.

To reproduce:

XML Schema (schema.xsd):

<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema">
<xs:element name="abstractSubgroupHead" type="xs:integer" abstract="true"/>
<xs:element name="subgroupMember" type="xs:positiveInteger" substitutionGroup="abstractSubgroupHead"/>
</xs:schema>

XML document (document.xml):
```
<subgroupMember>1</subgroupMember>
```

Validate:

xmlschema-validate --schema schema.xsd document.xml -v

Expected behavior:

The XML document should be validated successfully because subgroupMember is a global element; thus, a valid root within the instance document.

Actual behavior:

The xmlschema library reports the following error:

document.xml is not valid
failed validating <Element 'subgroupMember' at 0x7f6285136110> with XsdElement(name='abstractSubgroupHead', occurs=[1, 1]):

Reason: cannot use an abstract element for validation

Schema component:

  <xs:element xmlns:xs="http://www.w3.org/2001/XMLSchema" name="abstractSubgroupHead" type="xs:integer" abstract="true" />

Instance type: <class 'xml.etree.ElementTree.Element'>

Instance:

  <subgroupMember>1</subgroupMember>

Path: /subgroupMember

Environment:

xmlschema version: 3.4.0
Python version: 3.11.8

Additional notes:

The bug happens regardless of whether one uses the CLI or the API (is_valid and validate).
I believe the problem occurs only when the substitution element is used as a root element of the instance document, i.e., if there is another global element with complex type, which contains a reference to the abstract subgroup head, the validation is correct.

Hi, I will insert a fix in the next release of the package.

The fix requires a change of these lines:

https://github.com/sissaschool/xmlschema/blob/059fd3bfb305809d5ebdede1bd68dff9dbaf3348/xmlschema/validators/elements.py#L597-L599

adding a part for delegating the validation/decode to a substitute:

        if self.abstract:
            if self.name not in self.maps.substitution_groups:
                reason = _("can't use an abstract element for validation "
                           "unless it's the head of a substitution group")
                yield self.validation_error(validation, reason, obj, **kwargs)
            else:
                for xsd_element in self.iter_substitutes():
                    if obj.tag == xsd_element.name:
                        yield from xsd_element.iter_decode(obj, validation, **kwargs)
                return

I will extend also encoding part to process abstract elements and substitution (the tag matching is not so simple, because it depends by the convention used for decoded data).

Thank you for the contribution

sissaschool / xmlschema