Backport PR #54721 on branch 2.1.x (BUG/WARN: Passing EA object to dt…

…ype instead of an instance) (#54726)

Backport PR #54721: BUG/WARN: Passing EA object to dtype instead of an instance

Co-authored-by: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>

doc/source/whatsnew/v2.1.0.rst

@@ -837,6 +837,7 @@ ExtensionArray
 - Bug in :meth:`Series.rank` returning wrong order for small values with ``Float64`` dtype (:issue:`52471`)
 - Bug in :meth:`Series.unique` for boolean ``ArrowDtype`` with ``NA`` values (:issue:`54667`)
 - Bug in :meth:`~arrays.ArrowExtensionArray.__iter__` and :meth:`~arrays.ArrowExtensionArray.__getitem__` returning python datetime and timedelta objects for non-nano dtypes (:issue:`53326`)
+- Bug when passing an :class:`ExtensionArray` subclass to ``dtype`` keywords. This will now raise a ``UserWarning`` to encourage passing an instance instead (:issue:`31356`, :issue:`54592`)
 - Bug where the :class:`DataFrame` repr would not work when a column had an :class:`ArrowDtype` with a ``pyarrow.ExtensionDtype`` (:issue:`54063`)
 - Bug where the ``__from_arrow__`` method of masked ExtensionDtypes (e.g. :class:`Float64Dtype`, :class:`BooleanDtype`) would not accept PyArrow arrays of type ``pyarrow.null()`` (:issue:`52223`)
 

pandas/core/dtypes/common.py

@@ -1614,6 +1614,15 @@ def pandas_dtype(dtype) -> DtypeObj:
     # registered extension types
     result = registry.find(dtype)
     if result is not None:
+        if isinstance(result, type):
+            # GH 31356, GH 54592
+            warnings.warn(
+                f"Instantiating {result.__name__} without any arguments."
+                f"Pass a {result.__name__} instance to silence this warning.",
+                UserWarning,
+                stacklevel=find_stack_level(),
+            )
+            result = result()
         return result
 
     # try a numpy dtype

pandas/tests/dtypes/test_common.py

@@ -782,3 +782,9 @@ def test_pandas_dtype_numpy_warning():
         match="Converting `np.integer` or `np.signedinteger` to a dtype is deprecated",
     ):
         pandas_dtype(np.integer)
+
+
+def test_pandas_dtype_ea_not_instance():
+    # GH 31356 GH 54592
+    with tm.assert_produces_warning(UserWarning):
+        assert pandas_dtype(CategoricalDtype) == CategoricalDtype()