Extend .model_validate function, accepting some fields not in object

[mmzeynalli]

Initial Checks

• I have searched Google & GitHub for similar requests and couldn't find anything
• I have read and followed the docs and still think this feature is missing

Description

Why am I asking this:

I have a SQLAlchemy 2.0 query, which has User fields selected, as well as two aggregated fields, which are labeled. As, it is not possible to access those labelled attributes when used scalars(), I am forced to retrieve result as Row object. The query also contains joinedload of other models, and they have corresponding field in pydantic schema. However, my two labelled aggregate data comes as separate, detached from User model:

async def get_full_user_profile_query(user_id: int):
    return (
        select(
            User,
            cast(func.array_agg(func.distinct(Skill.name)), String).label('skills'),
            cast(func.array_agg(func.distinct(Tool.name)), String).label('tools'),
        )
        .where(User.id == user_id)
        .outerjoin(UserSkill, User.id == UserSkill.user_id)
        .outerjoin(Skill, UserSkill.skill_id == Skill.id)
        .outerjoin(UserTool, User.id == UserTool.user_id)
        .outerjoin(Tool, UserTool.tool_id == Tool.id)
        .options(joinedload(User.profile))
        .group_by(User)
    )

class UserProfileSchema(BaseModel):
    id: int
    first_name: Optional[str]
    last_name: Optional[str]
    avatar: Optional[str]
    email: EmailStr
    profile: 'UserProfileMinimalSchema'
    skills: str
    tools: str

    
async def get_user(user_id: int, db: DbDependency):
    """get user route"""
    query = await get_full_user_profile_query(user_id)
    res = db.execute(query).unique()

    item: Row = unwrap_scalars(res)[0]. # result is: Row(User, skills, tools)
    extra = item._asdict()  # main_obj + labelled aggs
    obj = extra.pop(item._parent._keys[0])
    
    ###### PROBLEM HERE ######
    return UserProfileSchema.model_validate(obj, **extra, from_attributes=True)

As seen, skills and tools are not part of User object in Row result, however, I want to pass them directly to model_validate as a part of User model. I can use for loop and setattr, however, I find that inefficient, especially If I have many aggregated/labelled fields in select.

Affected Components

• Compatibility between releases
• Data validation/parsing
• Data serialization - .model_dump() and .model_dump_json()
• JSON Schema
• Dataclasses
• Model Config
• Field Types - adding or changing a particular data type
• Function validation decorator
• Generic Models
• Other Model behaviour - model_construct(), pickling, private attributes, ORM mode
• Plugins and integration with other tools - mypy, FastAPI, python-devtools, Hypothesis, VS Code, PyCharm, etc.

[sydney-runkle]

@mmzeynalli,

Thanks for your question. Could you get your obj and extra data in dictionary format, then just unpack both of those? Moving this to a discussion, as it falls under the question category!

[mmzeynalli]

I tried that, however, considering the user object has its complex structures (a.k.a. profile), it gives an error, as user.profile is a UserProfile object, not, UserProfileMinimalSchema or dict. So, that is not a solution. However, I was able to find a workaround, not a good one though (I mentioned this in a previous one):

# The beginning of the code
obj.__dict__.update(extra)
return UserProfileSchema.model_validate(obj, from_attributes=True)