Validation error with str Enum field and numpy's str_ type

[maxsch3]

Initial Checks

• I confirm that I'm using Pydantic V2

Description

I'm having this issue in the API that uses machine learning (numpy+scikit-learn+xgboost) to generate predictions. The predictions are text labels. The ML part returns predictions as numpy-specific strings of type str_.

These values trigger validation error when used in the ENUM fields in pydantic model:

animal
  Input should be 'dog' or 'cat' [type=enum, input_value='cat', input_type=str_]
    For further information visit https://errors.pydantic.dev/2.7/v/enum

Example Code

from enum import Enum
from pydantic import BaseModel, Field
import numpy as np


class Animal(Enum):

    dog = "dog"
    cat = "cat"


class ModelStr(BaseModel):
    animal: str = Field(..., description="A type of animal")


class ModelEnum(BaseModel):
    animal: Animal = Field(..., description="A type of animal")


numpy_array = np.array(["cat", "dog"])

# this works
ModelStr(animal=numpy_array[0])

# this triggers error
ModelEnum(animal=numpy_array[0])

Python, Pydantic & OS Version

pydantic version: 2.7.0
        pydantic-core version: 2.18.1
          pydantic-core build: profile=release pgo=true
                 install path: /home/max/Code/Drax/dev/role-classification/venv/lib/python3.10/site-packages/pydantic
               python version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0]
                     platform: Linux-5.15.0-102-generic-x86_64-with-glibc2.35
             related packages: fastapi-0.101.1 typing_extensions-4.8.0
                       commit: unknown

[samuelcolvin]

Can you check if this is new in V2.7?

[maxsch3]

Yes, the example works fine in 2.6.4

[belinux]

I'm having exactly the same issue, with the only difference that instead of numpy my enums are evaluated on django models.IntegerChoices.

Code example:

from django.db import models

class Status(models.IntegerChoices):
    UNKNOWN = -1, _("unknown")
    ACCEPTED = 0, _("accepted")
    REJECTED = 1, _("rejected")

Output:

foo_status
  Input should be -1, 0 or 1 [type=enum, input_value=Status.ACCEPTED, input_type=Status]
    For further information visit https://errors.pydantic.dev/2.7/v/enum

[sydney-runkle]

@maxsch3,

If you inherit from str in your Animal type, all works as expected:

from enum import Enum
from pydantic import BaseModel, Field
import numpy as np


class Animal(str, Enum):

    dog = "dog"
    cat = "cat"


class ModelStr(BaseModel):
    animal: str = Field(..., description="A type of animal")


class ModelEnum(BaseModel):
    animal: Animal = Field(..., description="A type of animal")


numpy_array = np.array(["cat", "dog"])

# this works
ModelStr(animal=numpy_array[0])

# this works
ModelEnum(animal=numpy_array[0])

This workaround should work for now, but I'll add this to the 2.7 fixes milestone, as I don't think this extra specification should be required

[sydney-runkle]

Interesting that this still works, we must be missing a str conversion attempt somewhere:

from enum import Enum
from pydantic import BaseModel, Field

class Animal(Enum):

    dog = "dog"
    cat = "cat"


class ModelEnum(BaseModel):
    animal: Animal = Field(..., description="A type of animal")

# this works
ModelEnum(animal="cat")

[sydney-runkle]

Hi -- still working on a fix for this. Will wait on 2.7.2 most likely, hopefully early next week :). See pydantic/pydantic-core#1273 (comment)