Use to_numpy in ImplicitToExplicitIndexingAdapter.__array__ method

doc/whats-new.rst

@@ -42,7 +42,9 @@ Bug Fixes
 - Fix :py:func:`decode_cf` failing on integer-encoded time arrays that contain
   NaT when running against numpy 2.5+.
   By `Ian Hunt-Isaak <https://github.com/ianhi>`_.
-
+- Fix ``TypeError: Implicit conversion to a NumPy array is not allowed`` when trying to
+  use :py:func:`open_mfdataset` with a backend engine reading to CuPy arrays.
+  By `Wei Ji Leong <https://github.com/weiji14>`_.
 
 Documentation
 ~~~~~~~~~~~~~

xarray/core/indexing.py

@@ -35,7 +35,12 @@
     to_0d_array,
 )
 from xarray.namedarray.parallelcompat import get_chunked_array_type
-from xarray.namedarray.pycompat import array_type, integer_types, is_chunked_array
+from xarray.namedarray.pycompat import (
+    array_type,
+    integer_types,
+    is_chunked_array,
+    to_numpy,
+)
 
 if TYPE_CHECKING:
     from xarray.core.extension_array import PandasExtensionArray
@@ -683,9 +688,10 @@ def __array__(
         self, dtype: DTypeLike | None = None, /, *, copy: bool | None = None
     ) -> np.ndarray:
         if Version(np.__version__) >= Version("2.0.0"):
-            return np.asarray(self.get_duck_array(), dtype=dtype, copy=copy)
+            return np.asarray(to_numpy(self.get_duck_array()), dtype=dtype, copy=copy)
+
         else:
-            return np.asarray(self.get_duck_array(), dtype=dtype)
+            return np.asarray(to_numpy(self.get_duck_array()), dtype=dtype)
 
     def get_duck_array(self):
         return self.array.get_duck_array()

xarray/namedarray/pycompat.py

@@ -97,7 +97,7 @@ def is_0d_dask_array(x: duckarray[Any, Any]) -> bool:
 
 
 def to_numpy(
-    data: duckarray[Any, Any], **kwargs: dict[str, Any]
+    data: duckarray[Any, Any], **kwargs: Any
 ) -> np.ndarray[Any, np.dtype[Any]]:
     from xarray.core.indexing import ExplicitlyIndexed
     from xarray.namedarray.parallelcompat import get_chunked_array_type

xarray/tests/test_indexing.py

@@ -902,6 +902,14 @@ def test_implicit_indexing_adapter_copy_on_write() -> None:
     assert isinstance(implicit[:], indexing.ImplicitToExplicitIndexingAdapter)
 
 
+def test_implicit_indexing_adapter_duck_array() -> None:
+    array = DuckArrayWrapper(array=np.arange(10))
+    implicit = indexing.ImplicitToExplicitIndexingAdapter(
+        indexing.ArrayApiIndexingAdapter(array), indexing.BasicIndexer
+    )
+    np.testing.assert_array_equal(np.asarray(implicit), np.arange(10))
+
+
 def test_outer_indexer_consistency_with_broadcast_indexes_vectorized() -> None:
     def nonzero(x):
         if isinstance(x, np.ndarray) and x.dtype.kind == "b":