REF: consolidate numeric_only checks in GroupBy (pandas-dev#51185)

Author: jbrockmendel

pandas/core/groupby/generic.py

@@ -58,6 +58,7 @@
     is_dict_like,
     is_integer_dtype,
     is_interval_dtype,
+    is_numeric_dtype,
     is_scalar,
 )
 from pandas.core.dtypes.missing import (
@@ -172,9 +173,18 @@ def _wrap_agged_manager(self, mgr: Manager) -> Series:
         # NB: caller is responsible for setting ser.index
         return ser
 
-    def _get_data_to_aggregate(self) -> SingleManager:
+    def _get_data_to_aggregate(
+        self, *, numeric_only: bool = False, name: str | None = None
+    ) -> SingleManager:
         ser = self._selected_obj
         single = ser._mgr
+        if numeric_only and not is_numeric_dtype(ser.dtype):
+            # GH#41291 match Series behavior
+            kwd_name = "numeric_only"
+            raise TypeError(
+                f"Cannot use {kwd_name}=True with "
+                f"{type(self).__name__}.{name} and non-numeric dtypes."
+            )
         return single
 
     def _iterate_slices(self) -> Iterable[Series]:
@@ -1542,9 +1552,9 @@ def _cython_transform(
         #  test_transform_numeric_ret
         # With self.axis == 1, _get_data_to_aggregate does a transpose
         #  so we always have a single block.
-        mgr: Manager2D = self._get_data_to_aggregate()
-        if numeric_only:
-            mgr = mgr.get_numeric_data(copy=False)
+        mgr: Manager2D = self._get_data_to_aggregate(
+            numeric_only=numeric_only, name=how
+        )
 
         def arr_func(bvalues: ArrayLike) -> ArrayLike:
             return self.grouper._cython_operation(
@@ -1864,12 +1874,18 @@ def _gotitem(self, key, ndim: int, subset=None):
 
         raise AssertionError("invalid ndim for _gotitem")
 
-    def _get_data_to_aggregate(self) -> Manager2D:
+    def _get_data_to_aggregate(
+        self, *, numeric_only: bool = False, name: str | None = None
+    ) -> Manager2D:
         obj = self._obj_with_exclusions
         if self.axis == 1:
-            return obj.T._mgr
+            mgr = obj.T._mgr
         else:
-            return obj._mgr
+            mgr = obj._mgr
+
+        if numeric_only:
+            mgr = mgr.get_numeric_data(copy=False)
+        return mgr
 
     def _indexed_output_to_ndframe(
         self, output: Mapping[base.OutputKey, ArrayLike]

pandas/core/groupby/groupby.py

@@ -1540,22 +1540,9 @@ def _cython_agg_general(
         # Note: we never get here with how="ohlc" for DataFrameGroupBy;
         #  that goes through SeriesGroupBy
 
-        data = self._get_data_to_aggregate()
+        data = self._get_data_to_aggregate(numeric_only=numeric_only, name=how)
         is_ser = data.ndim == 1
 
-        if numeric_only:
-            if is_ser and not is_numeric_dtype(self._selected_obj.dtype):
-                # GH#41291 match Series behavior
-                kwd_name = "numeric_only"
-                if how in ["any", "all"]:
-                    kwd_name = "bool_only"
-                raise TypeError(
-                    f"Cannot use {kwd_name}={numeric_only} with "
-                    f"{type(self).__name__}.{how} and non-numeric types."
-                )
-            if not is_ser:
-                data = data.get_numeric_data(copy=False)
-
         def array_func(values: ArrayLike) -> ArrayLike:
             try:
                 result = self.grouper._cython_operation(
@@ -2034,15 +2021,6 @@ def std(
 
             return np.sqrt(self._numba_agg_general(sliding_var, engine_kwargs, ddof))
         else:
-            if (
-                numeric_only
-                and self.obj.ndim == 1
-                and not is_numeric_dtype(self.obj.dtype)
-            ):
-                raise TypeError(
-                    f"{type(self).__name__}.std called with "
-                    f"numeric_only={numeric_only} and dtype {self.obj.dtype}"
-                )
 
             def _preprocessing(values):
                 if isinstance(values, BaseMaskedArray):
@@ -3114,11 +3092,6 @@ def quantile(
         a    2.0
         b    3.0
         """
-        if numeric_only and self.obj.ndim == 1 and not is_numeric_dtype(self.obj.dtype):
-            raise TypeError(
-                f"{type(self).__name__}.quantile called with "
-                f"numeric_only={numeric_only} and dtype {self.obj.dtype}"
-            )
 
         def pre_processor(vals: ArrayLike) -> tuple[np.ndarray, DtypeObj | None]:
             if is_object_dtype(vals):
@@ -3258,8 +3231,7 @@ def blk_func(values: ArrayLike) -> ArrayLike:
 
         obj = self._obj_with_exclusions
         is_ser = obj.ndim == 1
-        mgr = self._get_data_to_aggregate()
-        data = mgr.get_numeric_data() if numeric_only else mgr
+        data = self._get_data_to_aggregate(numeric_only=numeric_only, name="quantile")
         res_mgr = data.grouped_reduce(blk_func)
 
         if is_ser:
@@ -3716,10 +3688,7 @@ def blk_func(values: ArrayLike) -> ArrayLike:
 
         # Operate block-wise instead of column-by-column
         is_ser = obj.ndim == 1
-        mgr = self._get_data_to_aggregate()
-
-        if numeric_only:
-            mgr = mgr.get_numeric_data()
+        mgr = self._get_data_to_aggregate(numeric_only=numeric_only, name=how)
 
         res_mgr = mgr.grouped_reduce(blk_func)
 

pandas/tests/groupby/aggregate/test_cython.py

@@ -92,7 +92,8 @@ def test_cython_agg_boolean():
 def test_cython_agg_nothing_to_agg():
     frame = DataFrame({"a": np.random.randint(0, 5, 50), "b": ["foo", "bar"] * 25})
 
-    with pytest.raises(TypeError, match="Cannot use numeric_only=True"):
+    msg = "Cannot use numeric_only=True with SeriesGroupBy.mean and non-numeric dtypes"
+    with pytest.raises(TypeError, match=msg):
         frame.groupby("a")["b"].mean(numeric_only=True)
 
     with pytest.raises(TypeError, match="Could not convert (foo|bar)*"):
@@ -117,7 +118,8 @@ def test_cython_agg_nothing_to_agg_with_dates():
             "dates": pd.date_range("now", periods=50, freq="T"),
         }
     )
-    with pytest.raises(TypeError, match="Cannot use numeric_only=True"):
+    msg = "Cannot use numeric_only=True with SeriesGroupBy.mean and non-numeric dtypes"
+    with pytest.raises(TypeError, match=msg):
         frame.groupby("b").dates.mean(numeric_only=True)
 
 

pandas/tests/groupby/test_function.py

@@ -1555,11 +1555,10 @@ def test_deprecate_numeric_only_series(dtype, groupby_func, request):
     elif dtype is object:
         msg = "|".join(
             [
-                "Cannot use numeric_only=True",
-                "called with numeric_only=True and dtype object",
+                "SeriesGroupBy.sem called with numeric_only=True and dtype object",
                 "Series.skew does not allow numeric_only=True with non-numeric",
-                "got an unexpected keyword argument 'numeric_only'",
-                "is not supported for object dtype",
+                "cum(sum|prod|min|max) is not supported for object dtype",
+                r"Cannot use numeric_only=True with SeriesGroupBy\..* and non-numeric",
             ]
         )
         with pytest.raises(TypeError, match=msg):

pandas/tests/resample/test_resample_api.py

@@ -908,7 +908,8 @@ def test_series_downsample_method(method, numeric_only, expected_data):
 
     func = getattr(resampled, method)
     if numeric_only and numeric_only is not lib.no_default:
-        with pytest.raises(TypeError, match="Cannot use numeric_only=True"):
+        msg = rf"Cannot use numeric_only=True with SeriesGroupBy\.{method}"
+        with pytest.raises(TypeError, match=msg):
             func(**kwargs)
     elif method == "prod":
         with pytest.raises(TypeError, match="can't multiply sequence by non-int"):