Pydantic with AsyncAttrs SQLAlchemy

Question

I want to use AsyncAttrs of SQLAlchemy with Pydantic .from_obj method

I have a simple SQLAlchemy Model, with inherited AsyncAttrs as said in sqlalchemy doc: AsyncAttrs doc

class ModelBase(AsyncAttrs, DeclarativeBase):
    ...

class A(ModelBase):
    __tablename__ = "a"

    bs = relationship()

and pydantic schemas:

class BsRelationSchema(BaseModel):
    id: int
    name: str
    class Config:
        orm_mode = True


    
class ASchema(BaseModel):
    bs: BsRelationSchema
    class Config:
        orm_mode = True

I can create instance of schema in this way:

a = (await session.scalars(select(A))).one()
a_schema = ASchema(bs=await a.awaitable_attrs.bs)

But how can i do this?

a = (await session.scalars(select(A))).one()
a_schema = ASchema.from_obj(a1)

Answer 1

I've been doing something similar lately, and there are two methods that I've used so far. It touches upon some of the things mentioned by DurandA, but a little less complex. The methods described below are taken from SQLAlchemy documentation

(Please also note I'll be using the new model_validate instead of from_orm, as described in Pydantic migration guide, but it works the same with from_orm, just trying to avoid the deprecated method.)

Anyway, method 1: after you get each object, you do an await on their awaitable_attrs before creating the schema. Yes, I am aware that is what you posted, I just wanted to write it here for completeness.

a = (await session.scalars(select(A))).one()
await a.awaitable_attrs.bs
a_schema = ASchema.model_validate(a)

Method 2: using the selectinload eager loader. You can modify your query to:

a = (await session.scalars(select(A).options(selectinload(A.bs)))).one()
a_schema = ASchema.model_validate(a)

Hope this helps!

Update: please be aware that this is just one way of relationship loading styles and they correspond to different SQL queries. Please have a look at Relationship Loading Techniques to learn more.