Pydantic v2：优雅处理含逗号的浮点数字符串解析

from pydantic import BaseModel

class User(BaseModel):
    name: str
    balance: float
    weight: float
    # ... 其他浮点数字段

from pydantic import BaseModel, model_validator
from typing import Dict, Any

class User(BaseModel):
    name: str
    balance: float
    weight: float
    # ... 其他浮点数字段

    @model_validator(mode='before')
    @classmethod
    def fix_float_comma_strings(cls, data: Dict[str, Any]) -> Dict[str, Any]:
        """
        在Pydantic模型验证之前，将输入数据中所有声明为float类型的字段
        且其值为含逗号的字符串时，将逗号替换为点。
        """
        if not isinstance(data, dict):
            # 如果输入不是字典，Pydantic会在后续处理中报错，这里可选择跳过或提前处理
            return data

        for k, v in data.items():
            # 1. 检查字段k是否在模型中被声明
            if k in cls.__annotations__:
                # 2. 检查字段k的声明类型是否为float
                # issubclass(float, cls.__annotations__[k]) 检查 cls.__annotations__[k] 是否是 float 或其子类
                # 对于 Union[float, None] 等复杂类型，需要更复杂的判断
                # 简单起见，这里假设直接就是 float 类型
                if cls.__annotations__[k] is float:
                    # 3. 检查当前值是否为字符串
                    if isinstance(v, str):
                        # 4. 执行替换操作
                        data[k] = v.replace(',', '.')
        return data

# 示例使用
if __name__ == "__main__":
    # 正常数据
    user1 = User(name="Alice", balance=123.45, weight="60.5")
    print(f"User 1 (正常数据): {user1.model_dump()}")
    # Output: User 1 (正常数据): {'name': 'Alice', 'balance': 123.45, 'weight': 60.5}

    # 含逗号的浮点数字符串数据
    bad_data = {
        "name": "Bob",
        "balance": "13,7",
        "weight": "75,25"
    }
    user2 = User(**bad_data)
    print(f"User 2 (含逗号数据): {user2.model_dump()}")
    # Output: User 2 (含逗号数据): {'name': 'Bob', 'balance': 13.7, 'weight': 75.25}

    # 混合数据
    mixed_data = {
        "name": "Charlie",
        "balance": "100.5", # 点分隔
        "weight": "80,1"   # 逗号分隔
    }
    user3 = User(**mixed_data)
    print(f"User 3 (混合数据): {user3.model_dump()}")
    # Output: User 3 (混合数据): {'name': 'Charlie', 'balance': 100.5, 'weight': 80.1}

    # 验证非字符串数据不会被修改
    non_string_data = {
        "name": "David",
        "balance": 50,
        "weight": "65,0"
    }
    user4 = User(**non_string_data)
    print(f"User 4 (非字符串数据): {user4.model_dump()}")
    # Output: User 4 (非字符串数据): {'name': 'David', 'balance': 50.0, 'weight': 65.0}