mypy errors when using Pydantic integration

[olivier-thatch]

Hello,

I'm trying to use the Pydantic integration and getting a few mypy errors.

Simplified version of the code:

import strawberry
from pydantic import BaseModel, EmailStr


class UserBase(BaseModel):
    email: Optional[EmailStr] = None


class UserCreate(PatientBase):
    email: EmailStr


@strawberry.experimental.pydantic.input(model=UserCreate, all_fields=True)
class UserCreateInput:
    pass


@strawberry.mutation
def create_user(self, info: Info, input_data: UserCreateInput) -> PatientType:
    instance = input_data.to_pydantic()
    # ...

This causes 2 errors:

1. @strawberry.experimental.pydantic.input(model=UserCreate, all_fields=True) causes this error:

   error: model argument in decorator failed to be parsed
   

2. instance = input_data.to_pydantic() causes this error:

   error: "UserCreateInput" has no attribute "to_pydantic"
   

Package versions:

• Python 3.10.3
• strawberry-graphql 0.106.3
• pydantic 1.9.0

Upvote & Fund

• We're using Polar.sh so you can upvote and help fund this issue.
• We receive the funding once the issue is completed & confirmed by you.
• Thank you in advance for helping prioritize & fund our backlog.

[motherofcoconuts]

Experiencing the same. Furthermore I have similar issues with from_pydantic. Ex: error: "Type[User]" has no attribute "from_pydantic"

[thejaminator]

hey @motherofcoconuts and @olivier-thatch are you using our mypy plugin? i fixed that

[mypy]
plugins = strawberry.ext.mypy_plugin

[motherofcoconuts]

hey @thejaminator i am indeed:

pyproject.toml

[tool.mypy]
plugins = ["pydantic.mypy", "strawberry.ext.mypy_plugin", "sqlalchemy.ext.mypy.plugin"]

[thejaminator]

@motherofcoconuts uhoh. i've figured out why

[motherofcoconuts]

Pulled from my poetry.lock file:

[[package]]
name = "strawberry-graphql"
version = "0.97.0"
description = "A library for creating GraphQL APIs"
category = "main"
optional = false
python-versions = ">=3.7,<4.0"

[[package]]
name = "mypy"
version = "0.942"
description = "Optional static typing for Python"
category = "dev"
optional = false
python-versions = ">=3.6"

[thejaminator]

@motherofcoconuts as a temporary workaround use the is_input=True parameter instead for strawberry.experimental.pydantic.type

from typing import Optional
import strawberry
from pydantic import BaseModel, EmailStr


class UserBase(BaseModel):
    email: Optional[EmailStr] = None


class UserCreate(BaseModel):
    email: EmailStr


@strawberry.experimental.pydantic.type(model=UserCreate, all_fields=True, is_input=True)
class UserCreateInput:
    pass


@strawberry.mutation
def create_user(self, info, input_data: UserCreateInput) -> None:
    instance = input_data.to_pydantic()
    # ...

basically

@strawberry.experimental.pydantic.type(model=UserCreate, all_fields=True, is_input=True)

instead of

@strawberry.experimental.pydantic.input(model=UserCreate, all_fields=True)

[motherofcoconuts]

@thejaminator @olivier-thatch is using the pydantic.input type. I actually use the following partials:

type = partial(strawberry.experimental.pydantic.type, all_fields=True)
input = partial(strawberry.experimental.pydantic.type, is_input=True, all_fields=True)

and experience it for both types with the @type decorator as well as the @input decoratior.

[thejaminator]

@thejaminator @olivier-thatch is using the pydantic.input type. I actually use the following partials:

type = partial(strawberry.experimental.pydantic.type, all_fields=True)
input = partial(strawberry.experimental.pydantic.type, is_input=True, all_fields=True)

and experience it for both types with the @type decorator as well as the @input decoratior.

ah ok, yeah seems like the mypy plugin isn't working well with those partials :(

See python/mypy#1484

[thejaminator]

@motherofcoconuts for your case, where you manually define your own decorator. unfortunately there are issues with mypy that i write here #1833