Dataclasses for arguments of methods in dependency injection principle

Question

I have the following project structure:

|__workspaces
|  |__db
|  |  |__create.py
|  |__graphql
|  |  |__mutations.py
|  |  |__interfaces.py

In graphql.mutations I have create mutation which create multiple workspaces:

class MutationWorkspacesCreate:
    def __init__(self, workspaces_creator: WorkspacesCreatorI):
        self._workspaces_creator = workspaces_creator

    def __call__(self, ...) -> list[UUID]
        ...
        await self._workspaces_creator.create(...)

This mutation depends on the class that implements the WorkspacesCreatorI interface:

class WorkspaceCreatorI(abc.ABC):
    @abc.abstractmethod
    async def create(self, schemas: WorkspaceCreateSchemaSet) -> list[UUID]
        raise NotImplementedError

In db.create I have class which implements this interface:

class DbWorkspacesCreator(WorkspacesCreatorI):
    async def create(self, schemas: WorkspaceCreateSchemaSet) -> list[UUID]
        ...

But since I need to create several workspaces, I need to have some two dataclasses:

class WorkspaceCreateSchema(pydantic.BaseModel):
   ...

class WorkspaceCreateSchemaSet(pydantic.BaseModel):
    items: list[WorkspaceCreateSchema]

If I put these dataclasses in the interfaces.py file, it looks kind of weird, because they are not abstract classes, but fully implemented, plus they can be used in a module in the database from some other modules, so rather they belong to the database. But on the other hand, if I put them in the database, then it turns out strange that the graphql package depends on the database, and the database on the graphql.

Please advise where to put these dataclasses and why

Dataclasses for arguments of methods in dependency injection principle

0 Answers0