Backport PR pandas-dev#56644 on branch 2.2.x (BUG: Series.to_numpy ra…

…ising for arrow floats to numpy floats) (pandas-dev#56648) Backport PR pandas-dev#56644: BUG: Series.to_numpy raising for arrow floats to numpy floats Co-authored-by: Patrick Hoefler <[email protected]>
meeseeksmachine · Dec 28, 2023 · f8e9892 · f8e9892
1 parent 95cc200
commit f8e9892
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 1 deletion.
diff --git a/pandas/core/arrays/arrow/array.py b/pandas/core/arrays/arrow/array.py
@@ -37,6 +37,7 @@
     CategoricalDtype,
     is_array_like,
     is_bool_dtype,
+    is_float_dtype,
     is_integer,
     is_list_like,
     is_numeric_dtype,
@@ -1320,6 +1321,7 @@ def to_numpy(
         copy: bool = False,
         na_value: object = lib.no_default,
     ) -> np.ndarray:
+        original_na_value = na_value
         dtype, na_value = to_numpy_dtype_inference(self, dtype, na_value, self._hasna)
         pa_type = self._pa_array.type
         if not self._hasna or isna(na_value) or pa.types.is_null(pa_type):
@@ -1345,7 +1347,14 @@ def to_numpy(
             if dtype is not None and isna(na_value):
                 na_value = None
             result = np.full(len(data), fill_value=na_value, dtype=dtype)
-        elif not data._hasna or (pa.types.is_floating(pa_type) and na_value is np.nan):
+        elif not data._hasna or (
+            pa.types.is_floating(pa_type)
+            and (
+                na_value is np.nan
+                or original_na_value is lib.no_default
+                and is_float_dtype(dtype)
+            )
+        ):
             result = data._pa_array.to_numpy()
             if dtype is not None:
                 result = result.astype(dtype, copy=False)

diff --git a/pandas/tests/extension/test_arrow.py b/pandas/tests/extension/test_arrow.py
@@ -3153,6 +3153,14 @@ def test_string_to_time_parsing_cast():
     tm.assert_series_equal(result, expected)
 
 
+def test_to_numpy_float():
+    # GH#56267
+    ser = pd.Series([32, 40, None], dtype="float[pyarrow]")
+    result = ser.astype("float64")
+    expected = pd.Series([32, 40, np.nan], dtype="float64")
+    tm.assert_series_equal(result, expected)
+
+
 def test_to_numpy_timestamp_to_int():
     # GH 55997
     ser = pd.Series(["2020-01-01 04:30:00"], dtype="timestamp[ns][pyarrow]")