Updates:

    - fix Index.str.split method return wrong result;
    - add test for Index.str.split method with expand=False;
    - return changes performed in pull request pandas-dev#1029.

Author: pan-vlados

pandas-stubs/core/indexes/base.pyi

@@ -261,7 +261,9 @@ class Index(IndexOpsMixin[S1]):
         **kwargs,
     ) -> Self: ...
     @property
-    def str(self) -> StringMethods[Self, MultiIndex, np_ndarray_bool]: ...
+    def str(
+        self,
+    ) -> StringMethods[Self, MultiIndex, np_ndarray_bool, Index[list[str]]]: ...
     def is_(self, other) -> bool: ...
     def __len__(self) -> int: ...
     def __array__(self, dtype=...) -> np.ndarray: ...

pandas-stubs/core/series.pyi

@@ -242,6 +242,26 @@ class Series(IndexOpsMixin[S1], NDFrame):
         copy: bool = ...,
     ) -> Series[float]: ...
     @overload
+    def __new__(  # type: ignore[overload-overlap]
+        cls,
+        data: Sequence[Never],
+        index: Axes | None = ...,
+        *,
+        dtype: Dtype = ...,
+        name: Hashable = ...,
+        copy: bool = ...,
+    ) -> Series[Any]: ...
+    @overload
+    def __new__(
+        cls,
+        data: Sequence[list[str]],
+        index: Axes | None = ...,
+        *,
+        dtype: Dtype = ...,
+        name: Hashable = ...,
+        copy: bool = ...,
+    ) -> Series[list[str]]: ...
+    @overload
     def __new__(
         cls,
         data: Sequence[str],
@@ -1199,7 +1219,9 @@ class Series(IndexOpsMixin[S1], NDFrame):
     ) -> Series[S1]: ...
     def to_period(self, freq: _str | None = ..., copy: _bool = ...) -> DataFrame: ...
     @property
-    def str(self) -> StringMethods[Series, DataFrame, Series[bool]]: ...
+    def str(
+        self,
+    ) -> StringMethods[Series, DataFrame, Series[bool], Series[list[str]]]: ...
     @property
     def dt(self) -> CombinedDatetimelikeProperties: ...
     @property

pandas-stubs/core/strings.pyi

@@ -15,6 +15,7 @@ import numpy as np
 import pandas as pd
 from pandas import (
     DataFrame,
+    Index,
     MultiIndex,
     Series,
 )
@@ -28,10 +29,12 @@ from pandas._typing import (
 
 # The _TS type is what is used for the result of str.split with expand=True
 _TS = TypeVar("_TS", DataFrame, MultiIndex)
+# The _TS2 type is what is used for the result of str.split with expand=False
+_TS2 = TypeVar("_TS2", Series[list[str]], Index[list[str]])
 # The _TM type is what is used for the result of str.match
 _TM = TypeVar("_TM", Series[bool], np_ndarray_bool)
 
-class StringMethods(NoNewAttributesMixin, Generic[T, _TS, _TM]):
+class StringMethods(NoNewAttributesMixin, Generic[T, _TS, _TM, _TS2]):
     def __init__(self, data: T) -> None: ...
     def __getitem__(self, key: slice | int) -> T: ...
     def __iter__(self) -> T: ...
@@ -67,7 +70,7 @@ class StringMethods(NoNewAttributesMixin, Generic[T, _TS, _TM]):
     @overload
     def split(
         self, pat: str = ..., *, n: int = ..., expand: Literal[False], regex: bool = ...
-    ) -> Series[list[str]]: ...
+    ) -> _TS2: ...
     @overload
     def split(
         self, pat: str = ..., *, n: int = ..., expand: bool = ..., regex: bool = ...

tests/test_frame.py

@@ -3767,9 +3767,9 @@ class MyDict(TypedDict):
 
 
 def test_series_empty_dtype() -> None:
-    """Test for the creation of a Series from an empty list GH571 to map to a Series[str]."""
+    """Test for the creation of a Series from an empty list GH571 to map to a Series[Any]."""
     new_tab: Sequence[Never] = []  # need to be typehinted to please mypy
-    check(assert_type(pd.Series(new_tab), "pd.Series[str]"), pd.Series)
-    check(assert_type(pd.Series([]), "pd.Series[str]"), pd.Series)
+    check(assert_type(pd.Series(new_tab), "pd.Series[Any]"), pd.Series)
+    check(assert_type(pd.Series([]), "pd.Series[Any]"), pd.Series)
     # ensure that an empty string does not get matched to Sequence[Never]
     check(assert_type(pd.Series(""), "pd.Series[str]"), pd.Series)

tests/test_indexes.py

@@ -113,6 +113,9 @@ def test_str_split() -> None:
     ind = pd.Index(["a-b", "c-d"])
     check(assert_type(ind.str.split("-"), "pd.Index[str]"), pd.Index)
     check(assert_type(ind.str.split("-", expand=True), pd.MultiIndex), pd.MultiIndex)
+    check(
+        assert_type(ind.str.split("-", expand=False), "pd.Index[list[str]]"), pd.Index
+    )
 
 
 def test_str_match() -> None: