Pydantic 2.* validator to add all extra fields to unique dict field

Question

Both validators and extra columns have different ways of configuration in pydantic 2.*, I would like to know if I can still do a validation to take all the extra fields and put them in a single dictionary field

The expected behavior is:

class A(BaseModel):
    name: str
    extra_columns: Optional[Dict]

a = A(name="john", age=24, address="Brazil")

print(a)

>>> A(name="john", extra_columns={"age": 24, "address": "Brazil"})

My previous code:

    @root_validator(pre=True)
    def build_extra(cls, values: Dict[str, Any]) -> Dict[str, Any]:
        all_required_field_names = {
            field.alias for field in cls.__fields__.values() if field.alias != "extra_columns"
        }

        extra: Dict[str, Any] = {}
        for field_name in list(values):
            if field_name not in all_required_field_names:
                extra[field_name] = values.pop(field_name)
        values["extra_columns"] = extra
        return values

Is possible to take the same behavior in pydantic 2.*?

Answer 1

This can be solved with model_validator:

from typing import Any
from pydantic import BaseModel, model_validator

class A(BaseModel):
    name: str
    extra_columns: dict | None = None
    
    @model_validator(mode="before")
    @classmethod
    def set_extra_columns(cls, data: Any):
        if isinstance(data, dict):
            extra_fields = data.keys() - cls.model_fields.keys()
            if extra_fields:
                if "extra_columns" not in data:
                    data["extra_columns"] = {}
                data["extra_columns"].update(
                    {field_name: data[field_name] for field_name in extra_fields}
                )
        return data

a = A(name="john", age=24, address="Brazil")

print(a)

# >>> name='john' extra_columns={'age': 24, 'address': 'Brazil'}