dataclass generator improvements

[anis-campos]

Use apply_discriminator_type for dataclasses
Allow dataclass models to be properly generated with discriminator field

Fix dataclass inheritance
Thanks to keyword only, dataclass models can use inheritance and no have issues with default values

Support datetime types in dataclass fields

applying --output-datetime-class from #2100 to dataclass to map date, time and date time to the python datetime objects instead of strings.

fixes #1966

[codspeed-hq]

CodSpeed Performance Report

Merging #2102 will not alter performance

_{Comparing anis-campos:dataclass-generator-improvements (8cd56d4) with main (2df133c)}

Summary

✅ 31 untouched benchmarks

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (2df133c) to head (8cd56d4).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #2102   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           38        38           
  Lines         4222      4250   +28     
  Branches       979       984    +5     
=========================================
+ Hits          4222      4250   +28     

Flag	Coverage Δ
unittests	99.67% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[koxudaxi]

@anis-campos
Thank you for submitting this pull request. I have two points that need to be addressed:

Since this change alters both the generated code and its behavior, could you please implement the kw_only feature as a new CLI option?
The kw_only parameter is a feature introduced in Python 3.10. We need to ensure that the output version is compatible with this feature. Please refer to the official documentation for more details:
https://docs.python.org/3/library/dataclasses.html#dataclasses.KW_ONLY

I would appreciate it if you could address these points. Thank you for your contribution, and I look forward to reviewing the updated version.

[koxudaxi]

I left the comment.

[anis-campos]

@koxudaxi I added a cli flag, let me know if the naming is right and if I could have changed less signatures with a different strategy to pass the info down to the model template.

[anis-campos]

@koxudaxi I'm also adding the support for datetime in dataclass, I'm using the recently merged --output-datetime-class to do so

[meliache]

I think this might fix also #2002 where --use-one-literal-as-default didn't work for discriminators. At least in my organizations codebase I thought that #2002 is the problem and this PR-branch seems to fix it. Thanks a lot, looking forward very much to see this merged and eventually in a release!

[anis-campos]

I think this might fix also #2002 where --use-one-literal-as-default didn't work for discriminators. At least in my organizations codebase I thought that #2002 is the problem and this PR-branch seems to fix it. Thanks a lot, looking forward very much to see this merged and eventually in a release!

thanks @meliache , I appreciate your comment.

I don't think it is linked, the issues is referencing pyantic output, I'm reworking only the dataclass generator.

side note, I would close the issue you linked, executing the example I see no issue with the pydantic output I have the default values just fine

# generated by datamodel-codegen:
#   filename:  openapi.yaml
#   timestamp: 2024-10-15T09:02:19+00:00

from __future__ import annotations

from typing import Literal, Union

from pydantic import BaseModel, Field, RootModel


class StartEvent(BaseModel):
    type: Literal['start'] = 'start'


class EndEvent(BaseModel):
    type: Literal['end'] = 'end'


class ResponseEvent(RootModel[Union[StartEvent, EndEvent]]):
    root: Union[StartEvent, EndEvent] = Field(..., discriminator='type')

[meliache]

@anis-campos

I don't think it is linked, the issues is referencing pyantic output, I'm reworking only the dataclass generator.

You are right, the linked issue seems to be already resolved in v0.26.1. In my project I had been using v0.25.9 and I had seen the same issue. In the release notes to v0.26.0 and v0.26.1 I saw nothing that indicated that explicitly indicated that the issue might be fixed. In this PR you at least mentioned that you worked on discriminator support, so on a whim I decided to try this branch, but then I tried v0.26.1 and there it also worked already.