[REQ][Python] UUID support

[igorbanfi]

Currently in python generator UUID are mapped to str. Are there any plans to enable generation of type UUID?

https://github.com/OpenAPITools/openapi-generator/blob/dfc381c22fde85be83681199d39f754efe04061e/modules/openapi-generator/src/main/java/org/openapitools/codegen/languages/AbstractPythonCodegen.java#L160C9-L160C10

[c99SRS]

Hi @igorbanfi you can assign this to me under hacktobestfest tag. Will research on this

[igorbanfi]

Hello, @c99SRS I am not a maintainer of openapi-generator and I am unable to assign issues. You will have to contact the maintainers of this repo.

[to-bee]

To make this work, the most critical change is reordering the type check conditions. The UUID type must be checked before the string type, as UUID is a subtype of string (with the format: UUID). This ensures the UUID condition is correctly applied.

AbstractPythonCodegen.java:

DefaultCodegen.java:

api_client.mustache:

[wing328]

@to-bee I wonder if you can file a PR with the suggested change 🙏

[BigyaPradhan]

I wish some one would do the same for mapping date-time Openapi spec to timedelta python.

[EspenHa]

This can be closed now, it is fixed in 7.15

[ArtemIlinArammeem]

@wing328 this issue is not mentioned in release notes. Is it released?

[wing328]

@ArtemIlinArammeem the release note only shows highlights (some of the PRs).

There's a link to all merged PRs included in the release.

[alexiswl]

This is causing some problems in my IDE:

Where the job_id parameter in the api_instance.get_job method is of type Annotated[UUID, Field(description="The ID of the job.")],

Solutions on user end is to set top-level job definition to be a union type, compared to the screenshot above, we no longer get the error.

However, I wonder if there is a solution on the openapi-generator end (that would save me having to update all my method types).
From this discussion, could UUIDs instead be typed with UUID4 | Annotated[str, AfterValidator(lambda x: uuid.UUID(x, version=4))], I would be keen to know if the AfterValidator from pydantic v2 works and actually validates the parameter or this is for models only.

Alternatively, you can 'cast' your string variable as a UUID4 object, with the following

api_instance.get_job(cast(UUID4, cast(object, job_id)))

[alexiswl]

@to-bee in your statement as UUID is a subtype of string I'm not sure this is correct.

from uuid import UUID, uuid4

x = uuid4()  # x = UUID('1d111d69-48b1-47f0-8501-47102e37c771')


isinstance(x, UUID)
# True (x is of type UUID)

isinstance(x, str)
# False (x is not an instance of type str)

issubclass(UUID, str)
# False (UUID is not a subclass of str)

issubclass(str, UUID)
# False (str is not subclass of UUID)

Maybe I'm missing something here?

[EspenHa]

@alexiswl I think what was meant by "UUID is a subtype of string" is that this is how it is encoded in openapi specification.
I believe this is because there is not native UUID type in json.

Regarding your code, you should be able to only modify the function as follows, without changing the function definition:

import uuid

api_instance.get_job(uuid.UUID(job_id)) # convert str to UUID

[alexiswl]

Regarding your code, you should be able to only modify the function as follows, without changing the function definition:

Thanks @EspenHa, that does at least provide the assertion that the job_id variable is in-fact a UUID.

Another change I've noticed is that the response body attributes from the sdk are also now UUID objects (rather than strings) for our project. This means that they have different methods and need to be handled differently too.

project_id attribute of the object is now a UUID object, was previously a string

project_id attribute has UUID methods, rather than str methods