Subtitutiongroups cause more attributes than may be serialised

[skinkie]

SubstitutionGroups in xsData cause multiple attributes. Each of these attributes typically match with a type, allowing static typing. The problem appears when multiple of these attributes are filled in and the schema has a limitation, for example the element has a maxOccurs=1 (default), providing multiple variants will cause the serialised document to be invalid.

I would propose an export serialisation option (configurable) that would explicitly limit the exported attribute to the first set.

[skinkie]

The following example creates a document that has a ServiceJourneyPatternRef and ServiceJourneyRef where only one of this type are allowed (in the current schema).

from xsdata.formats.dataclass.serializers import XmlSerializer
from xsdata.formats.dataclass.serializers.config import SerializerConfig
from xsdata.models.datatype import XmlDateTime
import datetime

from netex import GeneralFrame, GeneralFrameMembersRelStructure, \
    ServiceJourneyPattern, PointsInJourneyPatternRelStructure, StopPointInJourneyPattern, ScheduledStopPointRef, \
    PublicationDelivery, DataObjectsRelStructure, ScheduledStopPoint, MultilingualString, ServiceJourneyPatternRef, \
    ServiceJourney, ServiceJourneyRef

serializer_config = SerializerConfig(ignore_default_attributes=True)
serializer_config.pretty_print = True
serializer = XmlSerializer(serializer_config)

ns_map = {'': 'http://www.netex.org.uk/netex', 'gml': 'http://www.opengis.net/gml/3.2'}

with open('/tmp/out.xml', 'w') as out:
    ssp1 = ScheduledStopPoint(id="SSP1", version="1", name=MultilingualString(value="SSP1"))
    ssp2 = ScheduledStopPoint(id="SSP2", version="1", name=MultilingualString(value="SSP2"))

    sj = ServiceJourney(id="SJ", version="1")

    sjp = ServiceJourneyPattern(id="SJP", version="1", points_in_sequence=
                                PointsInJourneyPatternRelStructure(point_in_journey_pattern_or_stop_point_in_journey_pattern_or_timing_point_in_journey_pattern=
                                                                   [StopPointInJourneyPattern(id="SPIJP1", version="1", order=1,
                                                                                              scheduled_stop_point_ref=ScheduledStopPointRef(ref="SSP1", version="1"),
                                                                                              service_journey_pattern_ref=ServiceJourneyPatternRef(ref="SJP", version="1"),
                                                                                              service_journey_ref=ServiceJourneyRef(ref="SJ", version="1")
                                                                                              ),
                                                                    StopPointInJourneyPattern(id="SPIJP2", version="1", order=2, scheduled_stop_point_ref=ScheduledStopPointRef(ref="SSP2", version="1"))]))

    general_frame = GeneralFrame(id="GF", version="1", members=GeneralFrameMembersRelStructure(choice=[ssp1, ssp2, sj, sjp]))

    publication_delivery = PublicationDelivery(
        publication_timestamp=XmlDateTime.from_datetime(datetime.datetime.now()))
    publication_delivery.version = "ntx:1.1"
    publication_delivery.participant_ref = "NDOV"
    publication_delivery.description = "NeTEx export"
    publication_delivery.data_objects = DataObjectsRelStructure(choice=[general_frame])

    serializer.write(out, publication_delivery, ns_map)

[tefra]

Hey @skinkie, effectively substitution groups are like choices, hrm ok make sense
https://www.w3.org/TR/xmlschema-1/#key-equivalenceClass

I think we can generate them then as compound fields, then
https://xsdata.readthedocs.io/en/latest/examples/compound-fields.html

[tefra]

This one is a really tough cookie, I am trying to resolve it for weeks now, that's why I am delaying the new release...

[tefra]

I don't have a lot of free time lately, but this ticket really took a lot to fix, give it a try @skinkie

I had to rewrite how we calculate sequential fields, compound fields and min/max occurs within nested containers before I could even apply the fix to treat substitutions as choice elements.

Hopefully the result is a lot, a lot more accurate now 😄

[skinkie]

@tefra looking forward to compare the generated code :)