False Unused Import Statement in PyCharm?

Each Answer to this Q is separated by one/two green lines.

Given this scenario:

b.py:

import A
# A is unused here

c.py:

from b import A
# A is used here

PyCharm complains in b.py that import A is an unused import and Optimize imports deletes it, breaking import in c.py.

I know these chained imports are not a good practice (although you may use it to implement a facade module), but is it me or is it a PyCharm fail?

You can actually use the PyUnresolvedReferences marker to deactivate the inspection for your import statement:

# noinspection PyUnresolvedReferences
import A

Reference: PyCharm bug PY-2240

As far as I can tell this behaviour is not handled as an inspection or some other configurable option, which means there is no #noinspection UnusedImport (or equivalent) that can be placed before the imports.

If you don’t want to define an unused block where you use those variables there’s an other simple and probably better way to achieve what you want:

#b.py code
import A

# [...] your code


__all__ = ['A', ...]  # *all* the names you want to export

PyCharm is smart enough to look at __all__ and avoid removing A as unused import.
However there’s a limitation that __all__ must be a simple list literal. You cannot do things like:

__all__ = ['A'] + [name for name in iterable if condition(name)]

Not even:

x = 'b'
__all__ = ['A', x]

Defining __all__ is a best-practice to make your module *-import safe anyway, so is something you should already do.

from C import A, B
_ = (A, B); del _

Works for me. I don’t like

# noinspection PyUnresolvedReferences

as it would give false negatives in case A cannot be imported. And

__all__ = ['A', 'B', ...]

is cryptic and is not convenient for refactoring.


The answers/resolutions are collected from stackoverflow, are licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0 .