Pydanticスキーマ検証

なぜ Pydantic なのか

Defining a Schema

from pydantic import BaseModel, Field

class Product(BaseModel):
    name: str
    price: float = Field(gt=0, description='Positive price in USD')
    in_stock: bool
    tags: list[str] = []

Validating a Dict

data = {'name': 'Mug', 'price': 9.99, 'in_stock': True, 'tags': ['kitchen']}
product = Product.model_validate(data)
print(product.price)        # 9.99

Validating JSON

json_str = '{"name":"Mug","price":9.99,"in_stock":true}'
product = Product.model_validate_json(json_str)

Validation Errors

from pydantic import ValidationError

try:
    Product.model_validate({'name': 'Mug', 'price': -1, 'in_stock': True})
except ValidationError as e:
    print(e)
# [{'type': 'greater_than', 'loc': ('price',), 'msg': 'Input should be greater than 0', ...}]

Field-Level Validators

from pydantic import field_validator

class User(BaseModel):
    email: str

    @field_validator('email')
    @classmethod
    def must_have_at(cls, v):
        if '@' not in v:
            raise ValueError('Invalid email')
        return v.lower()

Nested Models

class Address(BaseModel):
    city: str
    zip: str

class User(BaseModel):
    name: str
    address: Address

data = {'name': 'Alice', 'address': {'city': 'Berlin', 'zip': '10115'}}
user = User.model_validate(data)
print(user.address.city)

Generating JSON Schema

schema = Product.model_json_schema()
# Pass straight to OpenAI tools or response_format

Using With OpenAI SDK

from openai import OpenAI
client = OpenAI()

response = client.beta.chat.completions.parse(
    model='gpt-4o-2024-08-06',
    messages=[{'role': 'user', 'content': 'Give me a product.'}],
    response_format=Product
)
product = response.choices[0].message.parsed

Optional Fields

from typing import Optional
class Order(BaseModel):
    id: int
    note: Optional[str] = None

Defaults

class Settings(BaseModel):
    units: str = 'celsius'
    notify: bool = True

エイリアス

class User(BaseModel):
    user_id: int = Field(alias='userId')

u = User.model_validate({'userId': 42})
print(u.user_id)   # 42

判別可能なユニオン

from typing import Literal, Union
from pydantic import Field

class TextMessage(BaseModel):
    type: Literal['text']
    content: str

class ImageMessage(BaseModel):
    type: Literal['image']
    url: str

Message = Annotated[Union[TextMessage, ImageMessage], Field(discriminator='type')]

バリデーションの目的

まとめ