Add an option to exlcude basic models from mappings

[heindrickdumdum0217]

Environment

Pydantic == 2.8.2
pyesorm == 0.4.8

I have several models, and they have common fields.
So I created a base model, but when call setup_mappings function, it tries to create an index for the base model too.

from esorm.model import ESModel


class _BaseModel(ESModel):
    id: unsigned_long


class A(_BaseModel):
    name: str

It throws the following error.

Index mappings error:
Traceback (most recent call last):
File \"/home/xxx/.pyenv/versions/3.11.5/envs/py3.11.5/lib/python3.11/site-packages/esorm/model.py\", line 1207, in setup_mappings
await es.indices.create(index=index,\n  File \"/home/xxx/.pyenv/versions/3.11.5/envs/py3.11.5/lib/python3.11/site-packages/elasticsearch/_async/client/indices.py\", line 553, in create
return await self.perform_request(  # type: ignore[return-value]     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File \"/home/xxx/.pyenv/versions/3.11.5/envs/py3.11.5/lib/python3.11/site-packages/elasticsearch/_async/client/_base.py\", line 423, in perform_request
return await self._client.perform_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File \"/home/xxx/.pyenv/versions/3.11.5/envs/py3.11.5/lib/python3.11/site-packages/elasticsearch/_async/client/_base.py\", line 271, in perform_request
response = await self._perform_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File \"/home/xxx/.pyenv/versions/3.11.5/envs/py3.11.5/lib/python3.11/site-packages/elasticsearch/_async/client/_base.py\", line 352, in _perform_request
raise HTTP_EXCEPTIONS.get(meta.status, ApiError)(
elasticsearch.BadRequestError: BadRequestError(400, 'invalid_index_name_exception', 'Invalid index name [myproject__BaseModel], must be lowercase')

The error doesn't affect to other indices, but I think still need an option to ignore some models from mappings.
For example:

class ESConfig:
    abstract = True

[wallneradam]

The ESBaseModel is exactly for this.

[wallneradam]

Oh, I misunderstood your question...
I'm going to test if the following works:

import esorm


class BaseModel(esorm.ESBaseModel):
    ...


class Model(BaseModel, esorm.ESModel):
    ...

I'm not sure if it works atm. but if not I will make it work this way.

[heindrickdumdum0217]

@wallneradam

Yes, we can use two base models.
But I prefer to use only one base model and can we add Config class to ESBaseModel?
For all models, I want to add an option to ignore extra.

class _BaseModel(ESModel):
    id: unsigned_long

    class Config:
        extra = Extra.ignore

[wallneradam]

Ok, I've tested it, this works, and it is more pythonic way. You have one base model, but two parents:

class BaseModelParent(esorm.ESBaseModel):
    class ESConfig:
        id_field = 'f_str'

    f_str: str
    f_int: int

class BaseModelParentModel(BaseModelParent, esorm.ESModel):
    f_float: float

ESBaseModel classes won't be incuded in mappings. So you can use them as a base structure for multiple models. Also ESConfig settings are inherited from the base model. I think this is what yout want.

[heindrickdumdum0217]

@wallneradam
I just tested, parent model config is not extended to child model.

class _BaseModel(ESBaseModel):
    id: unsigned_long

    class Config:
        extra = Extra.ignore


class A(_BaseModel, ESBaseModel)
    name: str


data = {"id": 3, "name": "Bob", "age": 17}
a = A.parse_obj(data)
await a.save()

It throws the following error.

Extra inputs are not permitted

I'm just giving an example.
In real situation data is from third party and it has very complex structure, this is why I'm using parse_obj function.

** Note **
I'm saying about Config class not ESConfig class.

[wallneradam]

1. There's no Config class in Pydantic v2.x. After that it will be inherited according to the documentation.
2. If you change extra to allow, i'm not sure it will work as intended. At least it is not tested. I don't think is a good idea to change model_config at all in Esorm models, it is fine tuned for ElasticSearch.
3. Why you would like to add extra? Esorm is for easily make "schema" for your model. Which can be used everywhere in the IDE and it is optimized in the DB. If you can add extra fields, the main goal will hurt.

[heindrickdumdum0217]

@wallneradam
ESBaseModel has model_config, I can't overwride the existing config.

https://github.com/wallneradam/esorm/blob/main/esorm/model.py#L237-L244

[wallneradam]

You can. I linked the documentation about it. Search this text: "If you add a model_config to the Model class, it will merge with the model_config from Parent".

It is the magic of Pydantic. I use similar magic with ESConfig class.