Homogeneous element of tagged unions stops on first failure

[DenSinH]

So I have a model which contains a list of a tagged union. However, I would really like to have some sort of catch-all or failure whenever a tagged-union element cannot be determined. The current behavior is that the loading just stops. At the very least, I would like it to keep loading, but a catch-all would be best. Consider the following example:

from pydantic_xml import BaseXmlModel, element, attr
from typing import Literal, Annotated
from pydantic import Field


class Pet(BaseXmlModel, tag="pet"):
    pet_type: str = Field()  # must be filled by child class


class Cat(Pet):
    pet_type: Literal['cat'] = attr(default="cat")


class Dog(Pet):
    pet_type: Literal['dog'] = attr(default="dog")


class Lizard(Pet):
    pet_type: Literal['lizard'] = attr(default="lizard")


AnyPet = Annotated[
    Cat | Dog | Lizard,
    Field(discriminator="pet_type")
]


class House(BaseXmlModel, tag="house"):
    pets: list[AnyPet] = element()


house = House.from_xml("""
<house>
    <pet pet_type="cat"/>
    <pet pet_type="dog"/>
    <pet pet_type="cow"/>
    <pet pet_type="lizard"/>
</house>
""")
print(house)
# prints
# pets=[Cat(pet_type='cat'), Dog(pet_type='dog')]

Note that the loading stops as soon as it hits an unknown tag ("cow"). I would prefer it to either crash, or at least continue loading.

To get it to continue loading, I think simply removing the break from the following line would work:0

pydantic-xml/pydantic_xml/serializers/factories/homogeneous.py

Line 87 in e5f21f5

break

but an error would be best.

[dapper91]

@DenSinH Hi,

that is the intended behavior, for the case when a model consists of multiple fields (for example a list followed by a submodel):

class House(BaseXmlModel, tag="house"):
    pets: list[AnyPet] = element()
    owner: Owner

If the parser finds an element that doesn't match the current field it tries the next one and so on.

In you case, when the cow element is reached the parser tries to match it with the AnyPet field. When it fails it tries next field (which doesn't exists) so the parsing process stops.

If you desire to get an error you may add extra=forbid option to the model:

class House(BaseXmlModel, tag="house", extra='forbid'):
    pets: list[AnyPet] = element()

[DenSinH]

I figured this would be the best solution. Sadly, there are other fields which I do want to ignore, as the model I am trying to parse is quite large, and I don't want to just add all the fields I don't know yet.