Fix last remaining MyPy errors

[potiuk]

Closes: #19891

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.

[potiuk]

Seems like this one will also be a collaborative effort eventually, There are a few errors around "airflow.models.decorators" package which I am not entirely sure why they are there and whether I made the right fix.

The one remaining - after I think I solved all other is

airflow/models/dag.py:2118: error: Argument 1 to "partial" has incompatible
type "_TaskDecoratorFactory"; expected "Callable[..., Any]"  [arg-type]
    ...return cast("TaskDecoratorFactory", functools.partial(task, dag=self))
                                                             ^
airflow/models/dag.py:2118: note: "_TaskDecoratorFactory.__call__" has type "staticmethod"
Found 1 error in 1 file (checked 2704 source files)

ERROR: The previous step completed with error. Please take a look at output above 

But I am not sure if the fixes are added are good. Seems like mypy sometimes uses the stub and sometimes not and I have no idea why.

[potiuk]

Any help appreciated @uranusjr @ashb :D

[uranusjr]

The naming and structuring of airflow/decorators/__init__.py is a small disaster. Let’s leave it out of this PR, I can fix them separately and perhaps try to improve the situation a bit.

[potiuk]

Ehh. so you will be the last one :)

[potiuk]

@uranusjr - removed it. Let's do it this way - since I also want to re-enable mypy and we should wait a few days with some "older PRs" to get "no-rebase" warning, I will continue rebasing that one until it gets "green". So feel free to unentangle the "decorators" disaster in the meantime

[potiuk]

I have a slight doubt about this one. This is "technically" breaking change with the experimental API, but one that should be very welcomed by any user. So I am tempted to leave it like that (I would treat it as a bugfix).

[ashb]

What bit throws the value error here? (It's not clear to from just looking at the code)

Oh, that state=dummy results in a ValueError -- wouldn't that result in a 500 returned, when instead it should be a 400?

[potiuk]

All right @uranusjr . Decorators are all yours:

No need to rebuild the image: none of the important files changed

airflow/models/dag.py:91: error: Module "airflow.decorators" has no attribute
"TaskDecoratorFactory"; maybe "_TaskDecoratorFactory" or "TaskDecorator"? 
[attr-defined]
        from airflow.decorators import TaskDecoratorFactory
        ^
airflow/models/dag.py:2118: error: Argument 1 to "partial" has incompatible
type "_TaskDecoratorFactory"; expected "Callable[..., Any]"  [arg-type]
    ...return cast("TaskDecoratorFactory", functools.partial(task, dag=self))
                                                             ^
airflow/models/dag.py:2118: note: "_TaskDecoratorFactory.__call__" has type "staticmethod"
Found 2 errors in 1 file (checked 2704 source files)

ERROR: The previous step completed with error. Please take a look at output above 


No need to rebuild the image: none of the important files changed

airflow/example_dags/tutorial_taskflow_api_etl.py:25: error: Module
"airflow.decorators" has no attribute "dag"  [attr-defined]
    from airflow.decorators import dag, task
    ^
Found 1 error in 1 file (checked 403 source files)

ERROR: The previous step completed with error. Please take a look at output above 

[potiuk]

Right - rebased and fixed last two errors: @uranusjr @ashb - I fixed two small thing (not even errors but inconsistencies and MyPy complaints in the new Operator Interface - I am not sure if adding "owner" to interface is a good idea - I think think we do not need to add dag though (as get_dag() shoudl return it anyway):

The errors were:

airflow/models/base.py:67: error: "Operator" has no attribute "dag" 
[attr-defined]
                return self.dag.dag_id
                       ^
airflow/models/base.py:68: error: "Operator" has no attribute "owner" 
[attr-defined]
            return f"adhoc_{self.owner}"

[ashb]

I'm surprised this needs type ignoring.

[ashb]

Oh, t.task could be None as far as typing is concenred?

[github-actions]

The PR most likely needs to run full matrix of tests because it modifies parts of the core of Airflow. However, committers might decide to merge it quickly and take the risk. If they don't merge it quickly - please rebase it to the latest main at your convenience, or amend the last commit of the PR, and push it with --force-with-lease.

[uranusjr]

Those two are easy, use get_dag instead. Oh, those are already fixed in #20945 (my myself lol), so we should be good now!