Backport PR pandas-dev#52422: BUG: merge with arrow and numpy dtypes raises

Author: phofl

doc/source/whatsnew/v2.0.1.rst

@@ -20,7 +20,7 @@ Fixed regressions
 
 Bug fixes
 ~~~~~~~~~
--
+- Fixed bug in :func:`merge` when merging with ``ArrowDtype`` one one and a NumPy dtype on the other side (:issue:`52406`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_201.other:

pandas/core/reshape/merge.py

@@ -78,13 +78,15 @@
 )
 
 from pandas import (
+    ArrowDtype,
     Categorical,
     Index,
     MultiIndex,
     Series,
 )
 import pandas.core.algorithms as algos
 from pandas.core.arrays import (
+    ArrowExtensionArray,
     BaseMaskedArray,
     ExtensionArray,
 )
@@ -2372,7 +2374,11 @@ def _factorize_keys(
         rk = ensure_int64(rk.codes)
 
     elif isinstance(lk, ExtensionArray) and is_dtype_equal(lk.dtype, rk.dtype):
-        if not isinstance(lk, BaseMaskedArray):
+        if not isinstance(lk, BaseMaskedArray) and not (
+            # exclude arrow dtypes that would get cast to object
+            isinstance(lk.dtype, ArrowDtype)
+            and is_numeric_dtype(lk.dtype.numpy_dtype)
+        ):
             lk, _ = lk._values_for_factorize()
 
             # error: Item "ndarray" of "Union[Any, ndarray]" has no attribute
@@ -2387,6 +2393,16 @@ def _factorize_keys(
         assert isinstance(rk, BaseMaskedArray)
         llab = rizer.factorize(lk._data, mask=lk._mask)
         rlab = rizer.factorize(rk._data, mask=rk._mask)
+    elif isinstance(lk, ArrowExtensionArray):
+        assert isinstance(rk, ArrowExtensionArray)
+        # we can only get here with numeric dtypes
+        # TODO: Remove when we have a Factorizer for Arrow
+        llab = rizer.factorize(
+            lk.to_numpy(na_value=1, dtype=lk.dtype.numpy_dtype), mask=lk.isna()
+        )
+        rlab = rizer.factorize(
+            rk.to_numpy(na_value=1, dtype=lk.dtype.numpy_dtype), mask=rk.isna()
+        )
     else:
         # Argument 1 to "factorize" of "ObjectFactorizer" has incompatible type
         # "Union[ndarray[Any, dtype[signedinteger[_64Bit]]],
@@ -2445,6 +2461,8 @@ def _convert_arrays_and_get_rizer_klass(
             #  Invalid index type "type" for "Dict[Type[object], Type[Factorizer]]";
             #  expected type "Type[object]"
             klass = _factorizers[lk.dtype.type]  # type: ignore[index]
+        elif isinstance(lk.dtype, ArrowDtype):
+            klass = _factorizers[lk.dtype.numpy_dtype.type]
         else:
             klass = _factorizers[lk.dtype.type]
 

pandas/tests/reshape/merge/test_merge.py

@@ -2735,3 +2735,18 @@ def test_merge_ea_and_non_ea(any_numeric_ea_dtype, join_type):
         }
     )
     tm.assert_frame_equal(result, expected)
+
+
+@pytest.mark.parametrize("dtype", ["int64", "int64[pyarrow]"])
+def test_merge_arrow_and_numpy_dtypes(dtype):
+    # GH#52406
+    pytest.importorskip("pyarrow")
+    df = DataFrame({"a": [1, 2]}, dtype=dtype)
+    df2 = DataFrame({"a": [1, 2]}, dtype="int64[pyarrow]")
+    result = df.merge(df2)
+    expected = df.copy()
+    tm.assert_frame_equal(result, expected)
+
+    result = df2.merge(df)
+    expected = df2.copy()
+    tm.assert_frame_equal(result, expected)