Backport PR #52591 on branch 2.0.x (BUG: pd.array raising with NumPy array and large dtype) (#52951)

Backport PR #52591: BUG: pd.array raising with NumPy array and large dtype

Co-authored-by: Patrick Hoefler <[email protected]>

Author: meeseeksmachine

doc/source/whatsnew/v2.0.2.rst

@@ -22,6 +22,7 @@ Bug fixes
 ~~~~~~~~~
 - Bug in :func:`api.interchange.from_dataframe` was returning :class:`DataFrame`'s of incorrect sizes when called on slices (:issue:`52824`)
 - Bug in :func:`api.interchange.from_dataframe` was unnecessarily raising on bitmasks (:issue:`49888`)
+- Bug in :meth:`pd.array` raising for ``NumPy`` array and ``pa.large_string`` or ``pa.large_binary`` (:issue:`52590`)
 -
 
 .. ---------------------------------------------------------------------------

pandas/core/arrays/arrow/array.py

@@ -247,6 +247,16 @@ def _from_sequence(cls, scalars, *, dtype: Dtype | None = None, copy: bool = Fal
         Construct a new ExtensionArray from a sequence of scalars.
         """
         pa_dtype = to_pyarrow_type(dtype)
+        if (
+            isinstance(scalars, np.ndarray)
+            and isinstance(dtype, ArrowDtype)
+            and (
+                pa.types.is_large_binary(pa_dtype) or pa.types.is_large_string(pa_dtype)
+            )
+        ):
+            # See https://github.com/apache/arrow/issues/35289
+            scalars = scalars.tolist()
+
         if isinstance(scalars, cls):
             scalars = scalars._data
         elif not isinstance(scalars, (pa.Array, pa.ChunkedArray)):

pandas/tests/extension/test_arrow.py

@@ -2620,6 +2620,20 @@ def test_setitem_boolean_replace_with_mask_segfault():
     assert arr._data == expected._data
 
 
+@pytest.mark.parametrize(
+    "data, arrow_dtype",
+    [
+        ([b"a", b"b"], pa.large_binary()),
+        (["a", "b"], pa.large_string()),
+    ],
+)
+def test_conversion_large_dtypes_from_numpy_array(data, arrow_dtype):
+    dtype = ArrowDtype(arrow_dtype)
+    result = pd.array(np.array(data), dtype=dtype)
+    expected = pd.array(data, dtype=dtype)
+    tm.assert_extension_array_equal(result, expected)
+
+
 @pytest.mark.parametrize("pa_type", tm.ALL_INT_PYARROW_DTYPES + tm.FLOAT_PYARROW_DTYPES)
 def test_describe_numeric_data(pa_type):
     # GH 52470