Fix type issue in worker.py

[hirosassa]

What I did

Resolve all Pyright type errors in gokart/worker.py (23 → 0):

• Move DISABLED, DONE, FAILED, PENDING, UNKNOWN from luigi.scheduler to luigi.task_status
• Use dict literals for config_path — Pyright treats dict() calls as dict[str, str], which is not assignable to ConfigPath (TypedDict). Dict literals {...} are correctly inferred as TypedDict-compatible
• Add type narrowing for flatten() results — Luigi's flatten() returns list[Any], so elements can be str. Added cast(Task, ...) and isinstance guards before accessing Task-specific attributes

[Copilot]

Pull request overview

This PR focuses on eliminating Pyright type errors by tightening/clarifying typings around Luigi task status constants, config path typing, and flatten/mapping utilities used across worker/task execution and GCS metadata serialization.

Changes:

• Updated gokart/worker.py to import task status constants from luigi.task_status, adjusted config_path to TypedDict-friendly literals, and added type narrowing/casts around Luigi flatten()/dependency iteration.
• Refined FlattenableItems typing in gokart/utils.py (explicit list | tuple | dict recursion) and aligned mapping logic with the narrowed union.
• Added casts/narrowing in gokart/task.py and gokart/gcs_obj_metadata_client.py to satisfy static typing for nested flattenable structures.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
gokart/worker.py	Adjusts Luigi imports and adds type narrowing/casts around dependencies/outputs; updates Config config_path typing.
gokart/utils.py	Narrows FlattenableItems TypeAlias and updates map_flattenable_items branching accordingly.
gokart/task.py	Adds explicit narrowing/casts for requires() and required_task_outputs construction to satisfy static typing.
gokart/gcs_obj_metadata_client.py	Updates flattenable traversal typing for required task outputs serialization and adjusts accepted container types.

Comments suppressed due to low confidence (1)

gokart/gcs_obj_metadata_client.py:136

• The TypeError message says "dict, or iterable" but this branch only accepts list | tuple. Either broaden the implementation back to collections.abc.Iterable (while still excluding str) or update the error message to match the actually supported types so callers aren’t misled.

            return {k: GCSObjectMetadataClient._get_serialized_string(v) for k, v in required_task_outputs.items()}
        elif isinstance(required_task_outputs, list | tuple):
            return [GCSObjectMetadataClient._get_serialized_string(ro) for ro in required_task_outputs]
        else:
            raise TypeError(
                f'Unsupported type for required_task_outputs: {type(required_task_outputs)}. '
                'It should be RequiredTaskOutput, dict, or iterable of RequiredTaskOutput.'
            )

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[hiro-o918]

I also agree copilot changes, because this change will drop some tasks unexpectedly

[hirosassa]

Thanks for the feedback! I agree that silently dropping non-Task elements is risky. I've changed this to use cast(list[Task], requires.flat_requirements) instead of the isinstance filter.
This keeps the runtime behavior identical to the original code (a non-Task element would still raise an AttributeError immediately) while satisfying the type checker.

0170810

[hiro-o918]

I also agree copilot changes, because this change will drop some tasks unexpectedly

[hiro-o918]

LGTM!
thank you for your PRs!