Pydantic Input Types behavior is strange #4042
Description
Describe the Bug
I have a pydantic class
class UserModel(pydantic.BaseModel):
name: str
interests: list[str] | None = pydantic.Field(
default_factory=list, description="The list of user interests."
)
Now building a strawberry input type:
@strawberry.experimental.pydantic.input(
model=UserModel,
description="""Input type for updating a User.""",
)
class UpdateUserInput:
name: strawberry.auto
interests: strawberry.auto
and building the mutation:
@strawberry.mutation
async def update_user(
self,
root: Any,
info: strawberry.types.Info,
user_data: UpdateUserInput,
) :
....
assuming I have a user in the db that already has interests ["reading", "cooking"] and I want to update the name.
My mutation would look something like this:
mutation {
update_user(user_data: {name: "New Name"})
...
}
Now what I see when looking at the user_data in the update_user function is:
user_data=UpdateUserInput(name='New Name', interests=[])
I would have expected to only see the set values or have the option somehow to filter out unset values... Otherwise partial updates are not possible.
System Information
- Operating system: Debian
- Python version: 3.13
- Strawberry version (if applicable): strawberry-graphql==0.284.1