TYP/CI: bump mypy&pyright #46905
TYP/CI: bump mypy&pyright #46905
Conversation
| yield ".".join(normalized_locale) | ||
| # error: Argument 1 to "join" of "str" has incompatible type | ||
| # "Tuple[Optional[str], Optional[str]]"; expected "Iterable[str]" | ||
| yield ".".join(normalized_locale) # type: ignore[arg-type] |
There was a problem hiding this comment.
mypy doesn't seem to understand all(...) to narrow the type.
There was a problem hiding this comment.
this works, but let's not do this yet! (until we can use it elsewhere)
def all_not_none(arg: Iterable[str | None]) -> TypeGuard[Iterable[str]]:
return all(x is not None for x in arg)
if all_not_none(normalized_locale):
There was a problem hiding this comment.
looking through the code there are a couple of other places this pattern is used that do not use all_not_none from pandas/core/common.py
maybe change this one here, then in the future we only need the TypeGuard in one place.
try:
import pandas.core.common as com
locale.setlocale(lc_var, new_locale)
normalized_locale = locale.getlocale()
if com.all_not_none(*normalized_locale):
note need to prevent circular import.
There was a problem hiding this comment.
Have there been discussions about using typing_extensions? Then we could already use TypeGuard.
There was a problem hiding this comment.
no we don't use typing_extensions but we can import from typing in pandas._typing with a guard so that it resolves to Any in older python versions.
There was a problem hiding this comment.
Have there been discussions about using
typing_extensions?
As long as typing_extensions is not a run-time dependency, I would not be adverse to it being added as a dev dependency.
|
Marking as draft until mypy 0.950 is available through conda |
Co-authored-by: Simon Hawkins <[email protected]>
|
mypy 0.950 is now available on conda-forge :) |
| @@ -957,7 +957,7 @@ def equals(self, other) -> bool: | |||
| return array_equivalent(left, right, dtype_equal=True) | |||
|
|
|||
| def _quantile( | |||
| self: BaseMaskedArrayT, qs: npt.NDArray[np.float64], interpolation: str | |||
| self, qs: npt.NDArray[np.float64], interpolation: str | |||
| ) -> BaseMaskedArray: | |||
There was a problem hiding this comment.
Pyright didn't previously find this. I assume the return value can be any BaseMaskedArray.
There was a problem hiding this comment.
yep. maybe add type annotations to _maybe_mask_result also.
There was a problem hiding this comment.
can be follow-on. I suspect that _maybe_mask_result should probably return the same type, but since we use explicit Subclass constructors this would not be the case if we subclass further.
There was a problem hiding this comment.
I will leave this out - it would require quite a few ignore comments if we do something like:
@overload
def _maybe_mask_result(self, result: ArrayLike, mask: ArrayLike) -> ArrayLike:
...
@overload
def _maybe_mask_result(
self, result: tuple[ArrayLike, ArrayLike], mask: ArrayLike
) -> tuple[ArrayLike, ArrayLike]:
...
def _maybe_mask_result(
self, result: ArrayLike | tuple[ArrayLike, ArrayLike], mask: ArrayLike
) -> ArrayLike | tuple[ArrayLike, ArrayLike]:For example FloatingArray and BooleanArray require an ndarray not an ExtensionArray.
|
doc-test failure seems to be unrelated but mypy&pyright passed :) |
|
@twoertwein planning to restart the ci here once #47015 is green and merged, so nothing more to do here as far as i'm concerned. |
Mypy aims to have a monthly release.