PERF: ArrowExtensionArray.fillna when array does not contains any nulls (#51635)

* PERF: ArrowExtensionArray.fillna when array does not have any nulls

* whatsnew

* fix test

* fix test

* Update array.py

* adjust CoW tests for arrow dtypes

---------

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

Author: lukemanley

doc/source/whatsnew/v2.1.0.rst

@@ -120,6 +120,7 @@ Performance improvements
 - Performance improvement in :func:`read_parquet` and :meth:`DataFrame.to_parquet` when reading a remote file with ``engine="pyarrow"`` (:issue:`51609`)
 - Performance improvement in :meth:`MultiIndex.sortlevel` when ``ascending`` is a list (:issue:`51612`)
 - Performance improvement in :meth:`~arrays.ArrowExtensionArray.isna` when array has zero nulls or is all nulls (:issue:`51630`)
+- Performance improvement in :meth:`~arrays.ArrowExtensionArray.fillna` when array does not contain nulls (:issue:`51635`)
 - Performance improvement when parsing strings to ``boolean[pyarrow]`` dtype (:issue:`51730`)
 - Performance improvement when searching an :class:`Index` sliced from other indexes (:issue:`51738`)
 - Performance improvement in :meth:`Series.combine_first` (:issue:`51777`)

pandas/core/arrays/arrow/array.py

@@ -766,6 +766,10 @@ def fillna(
     ) -> Self:
         value, method = validate_fillna_kwargs(value, method)
 
+        if not self._hasna:
+            # TODO(CoW): Not necessary anymore when CoW is the default
+            return self.copy()
+
         if limit is not None:
             return super().fillna(value=value, method=method, limit=limit)
 

pandas/tests/copy_view/test_interp_fillna.py

@@ -3,6 +3,7 @@
 
 from pandas import (
     NA,
+    ArrowDtype,
     DataFrame,
     Interval,
     NaT,
@@ -286,6 +287,9 @@ def test_fillna_ea_noop_shares_memory(
     if using_copy_on_write:
         assert np.shares_memory(get_array(df, "b"), get_array(df2, "b"))
         assert not df2._mgr._has_no_reference(1)
+    elif isinstance(df.dtypes[0], ArrowDtype):
+        # arrow is immutable, so no-ops do not need to copy underlying array
+        assert np.shares_memory(get_array(df, "b"), get_array(df2, "b"))
     else:
         assert not np.shares_memory(get_array(df, "b"), get_array(df2, "b"))
 
@@ -313,6 +317,9 @@ def test_fillna_inplace_ea_noop_shares_memory(
         assert np.shares_memory(get_array(df, "b"), get_array(view, "b"))
         assert not df._mgr._has_no_reference(1)
         assert not view._mgr._has_no_reference(1)
+    elif isinstance(df.dtypes[0], ArrowDtype):
+        # arrow is immutable, so no-ops do not need to copy underlying array
+        assert np.shares_memory(get_array(df, "b"), get_array(view, "b"))
     else:
         assert not np.shares_memory(get_array(df, "b"), get_array(view, "b"))
     df.iloc[0, 1] = 100

pandas/tests/extension/test_arrow.py

@@ -691,8 +691,8 @@ def test_fillna_no_op_returns_copy(self, data):
         result = data.fillna(valid)
         assert result is not data
         self.assert_extension_array_equal(result, data)
-        with tm.assert_produces_warning(PerformanceWarning):
-            result = data.fillna(method="backfill")
+
+        result = data.fillna(method="backfill")
         assert result is not data
         self.assert_extension_array_equal(result, data)
 

pandas/tests/extension/test_string.py

@@ -165,10 +165,7 @@ def test_fillna_no_op_returns_copy(self, data):
         assert result is not data
         self.assert_extension_array_equal(result, data)
 
-        with tm.maybe_produces_warning(
-            PerformanceWarning, data.dtype.storage == "pyarrow"
-        ):
-            result = data.fillna(method="backfill")
+        result = data.fillna(method="backfill")
         assert result is not data
         self.assert_extension_array_equal(result, data)