GH1074 Add type hint Series[list[str]] for Series.str.split with expand=False

Author: pan-vlados

pandas-stubs/_typing.pyi

@@ -547,7 +547,8 @@ S1 = TypeVar(
     | Period
     | Interval
     | CategoricalDtype
-    | BaseOffset,
+    | BaseOffset
+    | list[str],
 )
 
 S2 = TypeVar(
@@ -566,7 +567,8 @@ S2 = TypeVar(
     | Period
     | Interval
     | CategoricalDtype
-    | BaseOffset,
+    | BaseOffset
+    | list[str],
 )
 
 IndexingInt: TypeAlias = (

pandas-stubs/core/series.pyi

@@ -242,15 +242,15 @@ class Series(IndexOpsMixin[S1], NDFrame):
         copy: bool = ...,
     ) -> Series[float]: ...
     @overload
-    def __new__(  # type: ignore[overload-overlap]
+    def __new__(
         cls,
-        data: Sequence[Never],
+        data: Sequence[str],
         index: Axes | None = ...,
         *,
         dtype: Dtype = ...,
         name: Hashable = ...,
         copy: bool = ...,
-    ) -> Series[Any]: ...
+    ) -> Series[str]: ...
     @overload
     def __new__(
         cls,

pandas-stubs/core/strings.pyi

@@ -65,6 +65,10 @@ class StringMethods(NoNewAttributesMixin, Generic[T, _TS, _TM]):
         self, pat: str = ..., *, n: int = ..., expand: Literal[True], regex: bool = ...
     ) -> _TS: ...
     @overload
+    def split(
+        self, pat: str = ..., *, n: int = ..., expand: Literal[False], regex: bool = ...
+    ) -> Series[list[str]]: ...
+    @overload
     def split(
         self, pat: str = ..., *, n: int = ..., expand: bool = ..., regex: bool = ...
     ) -> T: ...

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[Any]."""
+    """Test for the creation of a Series from an empty list GH571 to map to a Series[str]."""
     new_tab: Sequence[Never] = []  # need to be typehinted to please mypy
-    check(assert_type(pd.Series(new_tab), "pd.Series[Any]"), pd.Series)
-    check(assert_type(pd.Series([]), "pd.Series[Any]"), pd.Series)
+    check(assert_type(pd.Series(new_tab), "pd.Series[str]"), pd.Series)
+    check(assert_type(pd.Series([]), "pd.Series[str]"), 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_series.py

@@ -1556,6 +1556,9 @@ def test_string_accessors():
     check(assert_type(s.str.split("a"), pd.Series), pd.Series)
     # GH 194
     check(assert_type(s.str.split("a", expand=True), pd.DataFrame), pd.DataFrame)
+    check(
+        assert_type(s.str.split("a", expand=False), "pd.Series[list[str]]"), pd.Series
+    )
     check(assert_type(s.str.startswith("a"), "pd.Series[bool]"), pd.Series, np.bool_)
     check(
         assert_type(s.str.startswith(("a", "b")), "pd.Series[bool]"),