TYP: Arraylike alias change follow up #40398
Conversation
pandas/io/parsers/base_parser.py
Outdated
| # error: Argument 2 to "astype_nansafe" has incompatible type | ||
| # "Type[str]"; expected "Union[dtype[Any], ExtensionDtype]" | ||
| values = astype_nansafe(values, str) # type: ignore[arg-type] | ||
| values = astype_nansafe(values, np.string_) |
There was a problem hiding this comment.
I think this should be np.dtype(str). will change shortly.
| @@ -1190,14 +1207,10 @@ def astype_nansafe( | |||
| flags = arr.flags | |||
| flat = arr.ravel("K") | |||
| result = astype_nansafe(flat, dtype, copy=copy, skipna=skipna) | |||
| order = "F" if flags.f_contiguous else "C" | |||
| order: Literal["C", "F"] = "F" if flags.f_contiguous else "C" | |||
| # error: Item "ExtensionArray" of "Union[ExtensionArray, ndarray]" has no | |||
| # attribute "reshape" | |||
There was a problem hiding this comment.
@jbrockmendel this is not necessarily a false positive. I think this maybe assuming that if ndim > 1, we have a numpy array. but a 2d EA will have additional methods such as reshape.
There was a problem hiding this comment.
yah even if we got here with NDArrayBackedExtensionArray (which does have reshape) itd break bc it doesn't have flags. ignoring for the time being seems reasonable
pandas/core/arrays/sparse/dtype.py
Outdated
| fill_value = astype_nansafe( # type: ignore[union-attr] | ||
| np.array(self.fill_value), dtype | ||
| ).item() | ||
| dtype = cast(np.dtype, dtype) |
There was a problem hiding this comment.
@jbrockmendel rather than cast, should I change the is_extension_array_dtype above to an isinstance check for consistency with the other PRs to remove ignores?
There was a problem hiding this comment.
as long as you've reliably got a DtypeObj, then yes. the isinstance checks are more performant
There was a problem hiding this comment.
cool. done that. i'll take your LGTM as an approval and self merge this soon.
follow-on from #40379 removing ignores that either changed or were added with the change of the ArrayLike alias to a Union