Commit 2b25842

authored

String dtype: keep select_dtypes(include=object) selecting string columns (#62323)

1 parent e817930 commit 2b25842Copy full SHA for 2b25842

File tree

+31

-13

lines changed

+31

-13

lines changed

Lines changed: 10 additions & 0 deletions

Original file line number	Diff line number	Diff line change
@@ -18,6 +18,16 @@ Most changes in this release are related to :class:`StringDtype` which will
`18`	`18`	`become the default string dtype in pandas 3.0. See`
`19`	`19`	:ref:`whatsnew_230.upcoming_changes` for more details.
`20`	`20`
	`21`	`+.. _whatsnew_233.string_fixes.improvements:`
	`22`	`+`
	`23`	`+Improvements`
	`24`	`+^^^^^^^^^^^^`
	`25`	+- Update :meth:`DataFrame.select_dtypes` to keep selecting ``str`` columns when
	`26`	+ specifying ``include=["object"]`` for backwards compatibility. In a future
	`27`	`+ release, this will be deprecated and code for pandas 3+ should be updated to`
	`28`	+ do ``include=["str"]`` (:issue:`61916`)
	`29`	`+`
	`30`	`+`
`21`	`31`	`.. _whatsnew_233.string_fixes.bugs:`
`22`	`32`
`23`	`33`	`Bug fixes`

Lines changed: 3 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -865,7 +865,9 @@ def invalidate_string_dtypes(dtype_set: set[DtypeObj]) -> None:`
`865`	`865`	`np.dtype("<U").type, # type: ignore[arg-type]`
`866`	`866`	`}`
`867`	`867`	`if non_string_dtypes != dtype_set:`
`868`		`- raise TypeError("string dtypes are not allowed, use 'object' instead")`
	`868`	`+ raise TypeError(`
	`869`	`+ "numpy string dtypes are not allowed, use 'str' or 'object' instead"`
	`870`	`+ )`
`869`	`871`
`870`	`872`
`871`	`873`	`def coerce_indexer_dtype(indexer, categories) -> np.ndarray:`

Lines changed: 8 additions & 4 deletions

Original file line number	Diff line number	Diff line change
`@@ -5150,10 +5150,14 @@ def check_int_infer_dtype(dtypes):`
`5150`	`5150`	`def dtype_predicate(dtype: DtypeObj, dtypes_set) -> bool:`
`5151`	`5151`	`# GH 46870: BooleanDtype._is_numeric == True but should be excluded`
`5152`	`5152`	`dtype = dtype if not isinstance(dtype, ArrowDtype) else dtype.numpy_dtype`
`5153`		`- return issubclass(dtype.type, tuple(dtypes_set)) or (`
`5154`		`- np.number in dtypes_set`
`5155`		`- and getattr(dtype, "_is_numeric", False)`
`5156`		`- and not is_bool_dtype(dtype)`
	`5153`	`+ return (`
	`5154`	`+ issubclass(dtype.type, tuple(dtypes_set))`
	`5155`	`+ or (`
	`5156`	`+ np.number in dtypes_set`
	`5157`	`+ and getattr(dtype, "_is_numeric", False)`
	`5158`	`+ and not is_bool_dtype(dtype)`
	`5159`	`+ )`
	`5160`	`+ or (dtype.type is str and np.object_ in dtypes_set)`
`5157`	`5161`	`)`
`5158`	`5162`
`5159`	`5163`	`def predicate(arr: ArrayLike) -> bool:`

Lines changed: 10 additions & 8 deletions

Original file line number	Diff line number	Diff line change
`@@ -102,6 +102,10 @@ def test_select_dtypes_include_using_list_like(self, using_infer_string):`
`102`	`102`	`ri = df.select_dtypes(include=[str])`
`103`	`103`	`tm.assert_frame_equal(ri, ei)`
`104`	`104`
	`105`	`+ ri = df.select_dtypes(include=["object"])`
	`106`	`+ ei = df[["a"]]`
	`107`	`+ tm.assert_frame_equal(ri, ei)`
	`108`	`+`
`105`	`109`	`def test_select_dtypes_exclude_using_list_like(self):`
`106`	`110`	`df = DataFrame(`
`107`	`111`	`{`
`@@ -309,17 +313,15 @@ def test_select_dtypes_not_an_attr_but_still_valid_dtype(self, using_infer_strin`
`309`	`313`	`df["g"] = df.f.diff()`
`310`	`314`	`assert not hasattr(np, "u8")`
`311`	`315`	`r = df.select_dtypes(include=["i8", "O"], exclude=["timedelta"])`
`312`		`- if using_infer_string:`
`313`		`- e = df[["b"]]`
`314`		`- else:`
`315`		`- e = df[["a", "b"]]`
	`316`	`+ # if using_infer_string:`
	`317`	`+ # TODO warn`
	`318`	`+ e = df[["a", "b"]]`
`316`	`319`	`tm.assert_frame_equal(r, e)`
`317`	`320`
`318`	`321`	`r = df.select_dtypes(include=["i8", "O", "timedelta64[ns]"])`
`319`		`- if using_infer_string:`
`320`		`- e = df[["b", "g"]]`
`321`		`- else:`
`322`		`- e = df[["a", "b", "g"]]`
	`322`	`+ # if using_infer_string:`
	`323`	`+ # TODO warn`
	`324`	`+ e = df[["a", "b", "g"]]`
`323`	`325`	`tm.assert_frame_equal(r, e)`
`324`	`326`
`325`	`327`	`def test_select_dtypes_empty(self):`

Comments

(0)