Bug/gh62522

doc/source/whatsnew/v3.0.0.rst

@@ -1166,6 +1166,7 @@ ExtensionArray
 ^^^^^^^^^^^^^^
 - Bug in :class:`Categorical` when constructing with an :class:`Index` with :class:`ArrowDtype` (:issue:`60563`)
 - Bug in :meth:`.arrays.ArrowExtensionArray.__setitem__` which caused wrong behavior when using an integer array with repeated values as a key (:issue:`58530`)
+- Bug in :meth:`.arrays.ArrowExtensionArray._cast_pointwise_result` causing ``Decimal("NaN")`` inputs to downcast results to ``null[pyarrow]`` instead of preserving decimal dtype (:issue:`62522`)
 - Bug in :meth:`ArrowExtensionArray.factorize` where NA values were dropped when input was dictionary-encoded even when dropna was set to False(:issue:`60567`)
 - Bug in :meth:`api.types.is_datetime64_any_dtype` where a custom :class:`ExtensionDtype` would return ``False`` for array-likes (:issue:`57055`)
 - Bug in comparison between object with :class:`ArrowDtype` and incompatible-dtyped (e.g. string vs bool) incorrectly raising instead of returning all-``False`` (for ``==``) or all-``True`` (for ``!=``) (:issue:`59505`)

pandas/core/arrays/arrow/array.py

@@ -4,6 +4,7 @@
     date,
     datetime,
 )
+from decimal import Decimal
 import functools
 import operator
 from pathlib import Path
@@ -481,6 +482,31 @@ def _cast_pointwise_result(self, values) -> ArrayLike:
             except pa.lib.ArrowInvalid:
                 # e.g. test_combine_add if we can't cast
                 pass
+        elif pa.types.is_null(arr.type):
+            # ``pa.array`` will produce null-dtype arrays when every value is a
+            # Decimal NaN. Try to preserve decimal storage by rebuilding the array
+            # with an explicit decimal type derived from the input values.
+            decimals = [val for val in values if isinstance(val, Decimal)]
+            if decimals and all(
+                isinstance(val, Decimal) or isna(val) for val in values
+            ):
+                decimal_type: pa.DataType | None = None
+                for dec in decimals:
+                    if getattr(dec, "is_nan", None) and dec.is_nan():
+                        continue
+                    try:
+                        decimal_type = pa.scalar(dec).type
+                        break
+                    except pa.ArrowInvalid:
+                        continue
+                if decimal_type is None:
+                    # All decimals were NaN -> fall back to a wide decimal so we
+                    # can retain the decimal dtype even though values stay null.
+                    decimal_type = pa.decimal128(38, 18)
+                try:
+                    arr = pa.array(values, type=decimal_type, from_pandas=True)
+                except (pa.ArrowInvalid, pa.ArrowTypeError):
+                    pass
 
         if isinstance(self.dtype, StringDtype):
             if pa.types.is_string(arr.type) or pa.types.is_large_string(arr.type):