Validation Errors when InjectionMaterials is combination of models

[mekhlakapoor]

Describe the bug
When injection_materials contains both ViralMaterial and NonViralMaterial objects, the injection fails validation due to "extra input" and "field required" errors. It is incorrectly validating ViralMaterial objects against NonViralMaterials and vice versa.

To Reproduce

1. Create a ViralMaterial with some fields (ex: titer)
2. Create a NonViralMaterial with some fields (ex: concentration)
3. Create an Injection with injection_materials = [viral, nonviral]

from aind_data_schema.core.procedures import TarsVirusIdentifiers, NanojectInjection, ViralMaterial, NonViralMaterial
ids = TarsVirusIdentifiers.model_construct(virus_tars_id="AiV123", prep_lot_number="lot123")
viral = ViralMaterial.model_construct(name="example", tars_identifiers=ids, titer=7000000)
non_viral = NonViralMaterial.model_construct(concentration=0.5)
inj = NanojectInjection.model_construct(injection_materials=[viral,non_viral]
 inj.__class__.model_validate(inj.model_dump())

4. See validation errors.

Expected behavior
injection_materials is a union of ViralMaterial and NonViralMaterial. It should allow and validate correctly.

Desktop (please complete the following information):

• OS: [e.g. iOS]
• Browser [e.g. chrome, safari]
• Version [e.g. 22]

Smartphone (please complete the following information):

• Device: [e.g. iPhone6]
• OS: [e.g. iOS8.1]
• Browser [e.g. stock browser, safari]
• Version [e.g. 22]

Additional context
Add any other context about the problem here.

[bruno-f-cruz]

You are forcing the model to be built without validation by calling the model_construct method. See the relevant documentation here.

The trace you get is not from the final NanojectInjection per se, but instead because of the recursive validation of the NonViralMaterial. Here's a working example that should get you started:

from aind_data_schema.core.procedures import TarsVirusIdentifiers, NanojectInjection, ViralMaterial, NonViralMaterial
from aind_data_schema.models.organizations import Organization

ids = TarsVirusIdentifiers(virus_tars_id="AiV123", prep_lot_number="lot123")
viral = ViralMaterial(name="example", tars_identifiers=ids, titer=7000000)
non_viral = NonViralMaterial(name="example", source=Organization.ABCAM, rrid=None, lot_number="foo", expiration_date=None)
inj = NanojectInjection(injection_materials=[viral, non_viral],
                        injection_coordinate_ml=0, injection_coordinate_ap=0, injection_coordinate_depth=[0, +10, -10],
                        protocol_id="bar", injection_angle=0, injection_volume=[-10, 0, 0])
inj.model_validate(inj.model_dump())

[jtyoung84]

This signature is incorrect and may fix the bug if we fix it:

injection_materials: Annotated[
        List[Union[ViralMaterial, NonViralMaterial]],
        Field(title="Injection material", min_length=1, discriminator="material_type"),
    ]

[micahwoodard]

I think this bug might have already been fixed? I can't reproduce the bug when using Bruno's example and injection_materials already has the signature Jon suggested. Let me know if I'm missing something!

[mekhlakapoor]

Closing this issue because it works with Bruno's example. Opening issue in metadata-service to fix construction of models to not use model_construct