Skip to content

non-empty-init-module (RUF067)

Preview (since 0.14.11) · Related issues · View source

This rule is unstable and in preview. The --preview flag is required for use.

What it does

Detects the presence of code in __init__.py files.

Why is this bad?

__init__.py files are often empty or only contain simple code to modify a module's API. As such, it's easy to overlook them and their possible side effects when debugging.

Example

Instead of defining MyClass directly in __init__.py:

"""My module docstring."""


class MyClass:
    def my_method(self): ...

move the definition to another file, import it, and include it in __all__:

"""My module docstring."""

from submodule import MyClass

__all__ = ["MyClass"]

Code in __init__.py files is also run at import time and can cause surprising slowdowns. To disallow any code in __init__.py files, you can enable the lint.ruff.strictly-empty-init-modules setting. In this case:

from submodule import MyClass

__all__ = ["MyClass"]

the only fix is entirely emptying the file:

Details

In non-strict mode, this rule allows several common patterns in __init__.py files:

  • Imports
  • Assignments to __all__, __path__, __version__, and __author__
  • Module-level and attribute docstrings
  • if TYPE_CHECKING blocks
  • PEP-562 module-level __getattr__ and __dir__ functions

Options

References