Python 如何使用建议值设置不区分大小写的模型

将类型指定为文字或某个字符串，并保持验证:

class Command(BaseModel):
    action: Literal["jump", "walk", "sleep"] | str

    @field_validator('action')
    @classmethod
    def validate_command(cls, v: str) -> str:
        """
        Checks if command is valid and converts it to lower.
        """
        if v.lower() not in {'jump', 'walk', 'sleep'}:
            raise HTTPException(status_code=422, detail="Action must be either 'jump', 'walk', or 'sleep'")
        return v.lower()

纯粹主义者可能会说Literal["jump", "walk", "sleep"] | str等同于str，但在这种情况下，人和计算机都可以推断"跳"和其他两个值是特殊输入.它的作用是提供示例值.

您可能不想重复文字值--传递mypy的解决方案是将它们设置为枚举:

from enum import Enum

class Action(Enum):
    jump = "jump"
    walk = "walk"
    sleep = "sleep"


class Command(BaseModel):
    action: Action | str

    @field_validator('action')
    @classmethod
    def validate_command(cls, v: str) -> str:
        """
        Checks if command is valid and converts it to lower.
        """
        if v.lower() not in Action.__members__:
            raise HTTPException(status_code=422, detail="Action must be either 'jump', 'walk', or 'sleep'")
        return v.lower()

另一种解决方案是继续使用Literal而不是枚举，并使用get_args来提取文本类型的值，如in this answer所述