Fund: [one|any|all]Of+required schema rules produce numerous near-identical models which are then quite hard to differentiate

Describe the bug
Combining [one|all|any]Of + required generates excessive near-identical models instead of treating them as root validation rules. This makes the output relatively quite verbose and significantly more difficult to debug.

To Reproduce

Example schema:

type: object
properties:
  a:
    type: string
  b:
    type: string
  c:
    type: string
  d:
    type: string
anyOf:
  - required:
    - a
  - required:
    - b

Used commandline:

$ datamodel-codegen --reuse-model --output-model-type pydantic_v2.BaseModel --target-python-version 3.9 --use-subclass-enum --collapse-root-models --allow-extra-fields --input openapi/components/schemas/test.yaml --output test.py

Generated output:

from __future__ import annotations

from typing import Optional, Union

from pydantic import BaseModel, ConfigDict, RootModel


class Model1(BaseModel):
    model_config = ConfigDict(
        extra='allow',
    )
    a: str
    b: Optional[str] = None
    c: Optional[str] = None
    d: Optional[str] = None


class Model2(BaseModel):
    model_config = ConfigDict(
        extra='allow',
    )
    a: Optional[str] = None
    b: str
    c: Optional[str] = None
    d: Optional[str] = None


class Model(RootModel[Union[Model1, Model2]]):
    root: Union[Model1, Model2]

Expected behavior

I believe it would be better in cases like this if the output were as follows with the rules recorded as rules instead of model clones

from pydantic import BaseModel, root_validator, ValidationError
from typing import Optional

class Model(BaseModel):
    model_config = ConfigDict(
        extra='allow',
    )
    a: Optional[str] = None
    b: Optional[str] = None
    c: Optional[str] = None
    d: Optional[str] = None

    @root_validator(pre=True)
    def at_least_one_field(cls, values):
        if not any(values.get(field) for field in ['a', 'b']):
            raise ValueError('At least one of "a", "b" must be provided')

        return values

Keep in mind that the above is a stripped-down compact test. In practice the output is substantially more difficult to evaluate, e.g....

class Model1(BaseModel):
    model_config = ConfigDict(
        extra='allow',
    )
    a: str = Field(
        ...,
        description='...',
        examples=['...'],
    )
    b: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    c: Optional[str] = Field(
        None, description='...',
    )
    d:: Optional[str] = Field(
        None, description='...',
    )
    e: Optional[str] = Field(
        None, description='...',
    )
    f: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    g: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    h: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    i: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    j: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    k: Optional[str] = Field(
        None,
        description='...',
    )
    l: Optional[bool] = Field(
        None,
        description='...',
        examples=[False],
    )
    m: Optional[str] = Field(
        None, description='...',
    )
    n: Optional[str] = Field(
        None, description='...',
    )


class Model2(BaseModel):
    model_config = ConfigDict(
        extra='allow',
    )
    a: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    b: str = Field(
        ...,
        description='...',
        examples=['...'],
    )
    c: Optional[str] = Field(
        None, description='...',
    )
    d:: Optional[str] = Field(
        None, description='...',
    )
    e: Optional[str] = Field(
        None, description='...',
    )
    f: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    g: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    h: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    i: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    j: Optional[str] = Field(
        None,
        description='...',
        examples=['...'],
    )
    k: Optional[str] = Field(
        None,
        description='...',
    )
    l: Optional[bool] = Field(
        None,
        description='...',
        examples=[False],
    )
    m: Optional[str] = Field(
        None, description='...',
    )
    n: Optional[str] = Field(
        None, description='...',
    )

Version:

OS: Linux
Python version: 3.9.12
datamodel-code-generator version: 0.25.8

Koudai Aono/datamodel-code-generator

[one|any|all]Of+required schema rules produce numerous near-identical models which are then quite hard to differentiate

How does funding with Polar work?

Backer

Contributor

Maintainer

Koudai Aono/datamodel-code-generator

[one|any|all]Of+required schema rules produce numerous near-identical models which are then quite hard to differentiate

How does funding with Polar work?

Backer

Why does "Fund on completion" require GitHub login?

When is the invoice due for "Fund on completion"?

What happens if the issue is never completed?

Do I get any extra benefits by funding?

Do I get progress updates?

Contributor

Do I get a reward?

Is rewards guaranteed?

Maintainer

How can I get funding like this for my open source initiatives?