BUG: is_string_dtype returns True for ArrowDtype(pa.string()) (#50963)

Author: mroeschke

doc/source/whatsnew/v2.0.0.rst

@@ -1021,7 +1021,7 @@ Conversion
 
 Strings
 ^^^^^^^
-- Bug in :func:`pandas.api.dtypes.is_string_dtype` that would not return ``True`` for :class:`StringDtype` (:issue:`15585`)
+- Bug in :func:`pandas.api.dtypes.is_string_dtype` that would not return ``True`` for :class:`StringDtype` or :class:`ArrowDtype` with ``pyarrow.string()`` (:issue:`15585`)
 - Bug in converting string dtypes to "datetime64[ns]" or "timedelta64[ns]" incorrectly raising ``TypeError`` (:issue:`36153`)
 -
 

pandas/core/arrays/arrow/dtype.py

@@ -95,6 +95,9 @@ def name(self) -> str:  # type: ignore[override]
     @cache_readonly
     def numpy_dtype(self) -> np.dtype:
         """Return an instance of the related numpy dtype"""
+        if pa.types.is_string(self.pyarrow_dtype):
+            # pa.string().to_pandas_dtype() = object which we don't want
+            return np.dtype(str)
         try:
             return np.dtype(self.pyarrow_dtype.to_pandas_dtype())
         except (NotImplementedError, TypeError):

pandas/tests/extension/test_arrow.py

@@ -46,6 +46,7 @@
     is_integer_dtype,
     is_numeric_dtype,
     is_signed_integer_dtype,
+    is_string_dtype,
     is_unsigned_integer_dtype,
 )
 from pandas.tests.extension import base
@@ -651,6 +652,24 @@ def test_groupby_extension_agg(self, as_index, data_for_grouping, request):
         ):
             super().test_groupby_extension_agg(as_index, data_for_grouping)
 
+    def test_in_numeric_groupby(self, data_for_grouping):
+        if is_string_dtype(data_for_grouping.dtype):
+            df = pd.DataFrame(
+                {
+                    "A": [1, 1, 2, 2, 3, 3, 1, 4],
+                    "B": data_for_grouping,
+                    "C": [1, 1, 1, 1, 1, 1, 1, 1],
+                }
+            )
+
+            expected = pd.Index(["C"])
+            with pytest.raises(TypeError, match="does not support"):
+                df.groupby("A").sum().columns
+            result = df.groupby("A").sum(numeric_only=True).columns
+            tm.assert_index_equal(result, expected)
+        else:
+            super().test_in_numeric_groupby(data_for_grouping)
+
 
 class TestBaseDtype(base.BaseDtypeTests):
     def test_construct_from_string_own_name(self, dtype, request):
@@ -730,7 +749,6 @@ def test_get_common_dtype(self, dtype, request):
                 and (pa_dtype.unit != "ns" or pa_dtype.tz is not None)
             )
             or (pa.types.is_duration(pa_dtype) and pa_dtype.unit != "ns")
-            or pa.types.is_string(pa_dtype)
             or pa.types.is_binary(pa_dtype)
         ):
             request.node.add_marker(
@@ -743,6 +761,13 @@ def test_get_common_dtype(self, dtype, request):
             )
         super().test_get_common_dtype(dtype)
 
+    def test_is_not_string_type(self, dtype):
+        pa_dtype = dtype.pyarrow_dtype
+        if pa.types.is_string(pa_dtype):
+            assert is_string_dtype(dtype)
+        else:
+            super().test_is_not_string_type(dtype)
+
 
 class TestBaseIndex(base.BaseIndexTests):
     pass