enum.auto should use capitalised enumerable strings

[craigcrawfordcts]

Bug report

Bug description:

Most languages dictate that the value and the name of the enumerable should be uppercase, for example with Java: https://docs.oracle.com/javase/tutorial/java/javaOO/enum.html

In python, a convenience method is provided auto, but neither in the description of StrEnum nor auto is it described that the enumerable values will be the lowercased variant of the enumerable's name.

This results in a failure to parse JSON objects using pydantic where the enumerable uses auto instead of directly defining the string, for example if those enumerable values were serialised in Java they are likely to be uppercased.

This results in a shooting-yourself-in-the-foot scenario if you believe you can just swap out your enum values for auto...

from enum import StrEnum, auto
from pydantic import BaseModel
import pytest

class StringEnumType(StrEnum):
    SOME_STRING = "SOME_STRING"


class AutoEnumType(StrEnum):
    SOME_STRING = auto()


class BaseModelWithStringEnumType(BaseModel):
    some_enum: StringEnumType


class BaseModelWithAutoEnumType(BaseModel):
    some_enum: AutoEnumType

@pytest.mark.parametrize(
    "model",
    [
        (BaseModelWithStringEnumType),(BaseModelWithAutoEnumType)
    ]
)
def test(model: BaseModel):
    model = model.model_validate({
        "some_enum": "SOME_STRING"
    })

tests/unit/test_fail.py::test[BaseModelWithAutoEnumType] FAILED
tests/unit/test_fail.py::test[BaseModelWithStringEnumType] PASSED

================================================================================== FAILURES ===================================================================================
_______________________________________________________________________ test[BaseModelWithAutoEnumType] _______________________________________________________________________

model = <class 'unit.test_fail.BaseModelWithAutoEnumType'>

    @pytest.mark.parametrize(
        "model",
        [
            (BaseModelWithStringEnumType),(BaseModelWithAutoEnumType)
        ]
    )
    def test(model: BaseModel):
>       model = model.model_validate({
            "some_enum": "SOME_STRING"
        })
E       pydantic_core._pydantic_core.ValidationError: 1 validation error for BaseModelWithAutoEnumType
E       some_enum
E         Input should be 'some_string' [type=enum, input_value='SOME_STRING', input_type=str]

tests/unit/test_fail.py:27: ValidationError

CPython versions tested on:

3.11

Operating systems tested on:

Linux

[ethanfurman]

@craigcrawfordcts :

but neither in the description of StrEnum nor auto is it described that the enumerable values will be the lowercased variant of the enumerable's name

enum docs for auto:

StrEnum defaults to the lower-cased version of the member name

enum docs for StrEnum:

Note Using auto with StrEnum results in the lower-cased member name as the value.

If you want a different case, you can create your own base class:

class UpperStrEnum(StrEnum):
    @staticmethod
    def _generate_next_value_(name, *args):
        return name.upper()

[craigcrawfordcts]

Disappointing. Correct those are in the online version of the document, but they are not in the doc strings:

One could argue that the lowercased property value of an enum property name is not appropriate.

[alexdelorenzo]

The StrEnum package on PyPI preserves case when using auto()