Fix result type of Series.sum(...) (#604)

* Fix result type of `Series.sum(...)`

* Fix note comment in `test_types_sum()`

* Make tests in `test_types_sum()` a bit stricter

pandas-stubs/core/series.pyi

@@ -70,7 +70,10 @@ from pandas.core.window.rolling import (
     Rolling,
     Window,
 )
-from typing_extensions import TypeAlias
+from typing_extensions import (
+    Never,
+    TypeAlias,
+)
 import xarray as xr
 
 from pandas._libs.interval import Interval
@@ -1811,6 +1814,30 @@ class Series(IndexOpsMixin, NDFrame, Generic[S1]):
         fill_value: float | None = ...,
         axis: AxisIndex | None = ...,
     ) -> Series[S1]: ...
+    # ignore needed because of mypy, for using `Never` as type-var.
+    @overload
+    def sum(
+        self: Series[Never],  # type: ignore[type-var]
+        axis: AxisIndex | None = ...,
+        skipna: _bool | None = ...,
+        level: None = ...,
+        numeric_only: _bool = ...,
+        min_count: int = ...,
+        **kwargs,
+    ) -> Any: ...
+    # ignore needed because of mypy, for overlapping overloads
+    # between `Series[bool]` and `Series[int]`.
+    @overload
+    def sum(  # type: ignore[misc]
+        self: Series[bool],
+        axis: AxisIndex | None = ...,
+        skipna: _bool | None = ...,
+        level: None = ...,
+        numeric_only: _bool = ...,
+        min_count: int = ...,
+        **kwargs,
+    ) -> int: ...
+    @overload
     def sum(
         self: Series[S1],
         axis: AxisIndex | None = ...,

tests/test_series.py

@@ -332,6 +332,36 @@ def test_types_sum() -> None:
     s.sum(numeric_only=False)
     s.sum(min_count=4)
 
+    # Note:
+    # 1. Return types of `series.groupby(...).sum(...)` are NOT tested
+    #    (waiting for stubs).
+    # 2. Runtime return types of `series.sum(min_count=...)` are NOT
+    #    tested (because of potential `nan`s).
+
+    s0 = assert_type(pd.Series([1, 2, 3, np.nan]), "pd.Series")
+    check(assert_type(s0.sum(), "Any"), np.float64)
+    check(assert_type(s0.sum(skipna=False), "Any"), np.float64)
+    check(assert_type(s0.sum(numeric_only=False), "Any"), np.float64)
+    assert_type(s0.sum(min_count=4), "Any")
+
+    s1 = assert_type(pd.Series([False, True], dtype=bool), "pd.Series[bool]")
+    check(assert_type(s1.sum(), "int"), np.int64)
+    check(assert_type(s1.sum(skipna=False), "int"), np.int64)
+    check(assert_type(s1.sum(numeric_only=False), "int"), np.int64)
+    assert_type(s1.sum(min_count=4), "int")
+
+    s2 = assert_type(pd.Series([0, 1], dtype=int), "pd.Series[int]")
+    check(assert_type(s2.sum(), "int"), np.int64)
+    check(assert_type(s2.sum(skipna=False), "int"), np.int64)
+    check(assert_type(s2.sum(numeric_only=False), "int"), np.int64)
+    assert_type(s2.sum(min_count=4), "int")
+
+    s3 = assert_type(pd.Series([1, 2, 3, np.nan], dtype=float), "pd.Series[float]")
+    check(assert_type(s3.sum(), "float"), np.float64)
+    check(assert_type(s3.sum(skipna=False), "float"), np.float64)
+    check(assert_type(s3.sum(numeric_only=False), "float"), np.float64)
+    assert_type(s3.sum(min_count=4), "float")
+
 
 def test_types_cumsum() -> None:
     s = pd.Series([1, 2, 3, np.nan])