Enums are not created in new type alias syntax about sqlalchemy

[CaselIT]

Also TypeAliasType is not a valid type here, since it's likely that everything defined with type ... becomes one. Indeed:

>>> type x = int
>>> type(x)
<class 'typing.TypeAliasType'>
>>> type y = int | str | float
>>> type(y)
<class 'typing.TypeAliasType'>

[zzzeek]

OK mostly this just builds on 8772041 / #10807 which is very recent . great thanks for the PR

[CaselIT]

Hi,

I fairly sure this has already been reported but I cannot find it.
Sadly python decided that for the typing feature it was the best thing to have a thousand different ways of doing anything, so it's quite difficult to support all the various incantations.

thanks for reporting

[CaselIT]

It seems that one could use __value__ to access what the type alias points to

[Alc-Alc]

I fairly sure this has already been reported but I cannot find it.

I came across #10807 which felt similar but the issue was closed and it seems it was released in 2.0.25

Added preliminary support for Python 3.12 pep-695 type alias structures, when resolving custom type maps for ORM Annotated Declarative mappings.

but I was still running into this so I opened a new one.

[CaselIT]

Ok, Thanks for finding that. Then I guess this is an unsupported case

[Alc-Alc]

@CaselIT I have made an attempt here at fixing this, if you feel this is a good starting point I am happy to raise a PR and apply any relevant suggestions. I added one new test (modified based on an existing test) and it passes.

8fd36a4 shows my attempted changes

[CaselIT]

That seems to work but I wonder if it could be easier to just do at the start of _resolve_type

if is_pep695(python_type):
  return _resolve_type(python_type.__value__)

This would handle anything is declared with the new type syntax.
It would also allow for alias of alias (not sure if value in this case is also a typealiastype or if it's resolved)

Feel free to open a PR. It's a good start (if you do can you also add a change log?)

[sqla-tester]

Alc-Alc has proposed a fix for this issue in the main branch:

Enums are not created with pep695 literal https://gerrit.sqlalchemy.org/c/sqlalchemy/sqlalchemy/+/5260

[zzzeek]

was there another issue or discussion where we went over this previously?

[sqla-tester]

Alc-Alc has proposed a fix for this issue in the rel_2_0 branch:

improve pep-695 inference including Enum support https://gerrit.sqlalchemy.org/c/sqlalchemy/sqlalchemy/+/5278

[Alc-Alc]

was there another issue or discussion where we went over this previously?

Yes, I noticed you figured it in the next comment. Just for completion sake, #11305 (comment) states my reasoning behind opening this issue.