Support `graphql_type` for `strawberry.argument`

[thearchitector]

Resolvers support different return types via the graphql_type argument to strawberry.field. This is super useful when the GraphQL semantics deal in one scalar, but the python semantics deal in another (for static typing purposes or otherwise).

If I have a custom scalar that (de)serializes to a built-in type, for ex., i can decorate the resolver with the built-in but tell GQL to use the custom scalar. Like so:

BigInt = strawberry.scalar(
   NewType("BigInt", int),
   serialize=lambda v: str(v),
   parse_value=lambda v: int(v)
)

@strawberry.field(graphql_type=BigInt)
def blah(info) -> int:
   return 1238423948234

I cannot do the same for arguments to custom resovlers, though, which causes some static typing inconveniences. For ex, Pyright complains with the following:

def some_num(a: list[int]) -> str:
   return "hi".join(a)


@strawberry.field
def blah(info, nums: list[BigInt]) -> str:
    return some_num(nums)

In the same manner as the returned type, it would be useful to be able to define the type for the argument:

@strawberry.field
def blah(
    info,
    nums: Annotated[list[int], strawberry.argument(graphql_type=list[BigInt])],
) -> str:
    return some_num(nums)

Feature Request Type

• Core functionality
• Alteration (enhancement/optimization) of existing feature(s)
• New behavior

[erikwrede]

Sounds useful! Would you like to contribute a PR for the feature?