Backport PR #51411 on branch 2.0.x (ENH: Optimize CoW for fillna with…

… ea dtypes) (#51639)

Backport PR #51411: ENH: Optimize CoW for fillna with ea dtypes

Co-authored-by: Patrick Hoefler <61934744+phofl@users.noreply.github.com>

pandas/core/internals/blocks.py

@@ -1758,9 +1758,14 @@ def fillna(
                 downcast=downcast,
                 using_cow=using_cow,
             )
-        new_values = self.values.fillna(value=value, method=None, limit=limit)
-        nb = self.make_block_same_class(new_values)
-        return nb._maybe_downcast([nb], downcast)
+        if using_cow and self._can_hold_na and not self.values._hasna:
+            refs = self.refs
+            new_values = self.values
+        else:
+            refs = None
+            new_values = self.values.fillna(value=value, method=None, limit=limit)
+        nb = self.make_block_same_class(new_values, refs=refs)
+        return nb._maybe_downcast([nb], downcast, using_cow=using_cow)
 
     @cache_readonly
     def shape(self) -> Shape:

pandas/tests/copy_view/test_interp_fillna.py

@@ -2,6 +2,7 @@
 import pytest
 
 from pandas import (
+    NA,
     DataFrame,
     Interval,
     NaT,
@@ -271,3 +272,48 @@ def test_fillna_series_empty_arg_inplace(using_copy_on_write):
     assert np.shares_memory(get_array(ser), arr)
     if using_copy_on_write:
         assert ser._mgr._has_no_reference(0)
+
+
+def test_fillna_ea_noop_shares_memory(
+    using_copy_on_write, any_numeric_ea_and_arrow_dtype
+):
+    df = DataFrame({"a": [1, NA, 3], "b": 1}, dtype=any_numeric_ea_and_arrow_dtype)
+    df_orig = df.copy()
+    df2 = df.fillna(100)
+
+    assert not np.shares_memory(get_array(df, "a"), get_array(df2, "a"))
+
+    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)
+    else:
+        assert not np.shares_memory(get_array(df, "b"), get_array(df2, "b"))
+
+    tm.assert_frame_equal(df_orig, df)
+
+    df2.iloc[0, 1] = 100
+    if using_copy_on_write:
+        assert not np.shares_memory(get_array(df, "b"), get_array(df2, "b"))
+        assert df2._mgr._has_no_reference(1)
+        assert df._mgr._has_no_reference(1)
+    tm.assert_frame_equal(df_orig, df)
+
+
+def test_fillna_inplace_ea_noop_shares_memory(
+    using_copy_on_write, any_numeric_ea_and_arrow_dtype
+):
+    df = DataFrame({"a": [1, NA, 3], "b": 1}, dtype=any_numeric_ea_and_arrow_dtype)
+    df_orig = df.copy()
+    view = df[:]
+    df.fillna(100, inplace=True)
+
+    assert not np.shares_memory(get_array(df, "a"), get_array(view, "a"))
+
+    if using_copy_on_write:
+        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)
+    else:
+        assert not np.shares_memory(get_array(df, "b"), get_array(view, "b"))
+    df.iloc[0, 1] = 100
+    tm.assert_frame_equal(df_orig, view)

pandas/tests/extension/base/methods.py

@@ -243,24 +243,25 @@ def test_factorize_empty(self, data):
     def test_fillna_copy_frame(self, data_missing):
         arr = data_missing.take([1, 1])
         df = pd.DataFrame({"A": arr})
+        df_orig = df.copy()
 
         filled_val = df.iloc[0, 0]
         result = df.fillna(filled_val)
 
-        assert df.A.values is not result.A.values
+        result.iloc[0, 0] = filled_val
 
-    def test_fillna_copy_series(self, data_missing, no_op_with_cow: bool = False):
+        self.assert_frame_equal(df, df_orig)
+
+    def test_fillna_copy_series(self, data_missing):
         arr = data_missing.take([1, 1])
         ser = pd.Series(arr)
+        ser_orig = ser.copy()
 
         filled_val = ser[0]
         result = ser.fillna(filled_val)
+        result.iloc[0] = filled_val
 
-        if no_op_with_cow:
-            assert ser._values is result._values
-        else:
-            assert ser._values is not result._values
-        assert ser._values is arr
+        self.assert_series_equal(ser, ser_orig)
 
     def test_fillna_length_mismatch(self, data_missing):
         msg = "Length of 'value' does not match."

pandas/tests/extension/conftest.py

@@ -2,7 +2,10 @@
 
 import pytest
 
-from pandas import Series
+from pandas import (
+    Series,
+    options,
+)
 
 
 @pytest.fixture
@@ -193,3 +196,11 @@ def invalid_scalar(data):
     If the array can hold any item (i.e. object dtype), then use pytest.skip.
     """
     return object.__new__(object)
+
+
+@pytest.fixture
+def using_copy_on_write() -> bool:
+    """
+    Fixture to check if Copy-on-Write is enabled.
+    """
+    return options.mode.copy_on_write and options.mode.data_manager == "block"

pandas/tests/extension/json/test_json.py

@@ -301,17 +301,9 @@ def test_searchsorted(self, data_for_sorting):
     def test_equals(self, data, na_value, as_series):
         super().test_equals(data, na_value, as_series)
 
-    def test_fillna_copy_frame(self, data_missing, using_copy_on_write):
-        arr = data_missing.take([1, 1])
-        df = pd.DataFrame({"A": arr})
-
-        filled_val = df.iloc[0, 0]
-        result = df.fillna(filled_val)
-
-        if using_copy_on_write:
-            assert df.A.values is result.A.values
-        else:
-            assert df.A.values is not result.A.values
+    @pytest.mark.skip("fill-value is interpreted as a dict of values")
+    def test_fillna_copy_frame(self, data_missing):
+        super().test_fillna_copy_frame(data_missing)
 
 
 class TestCasting(BaseJSON, base.BaseCastingTests):

pandas/tests/extension/test_datetime.py

@@ -116,11 +116,6 @@ def test_combine_add(self, data_repeated):
         # Timestamp.__add__(Timestamp) not defined
         pass
 
-    def test_fillna_copy_series(self, data_missing, using_copy_on_write):
-        super().test_fillna_copy_series(
-            data_missing, no_op_with_cow=using_copy_on_write
-        )
-
 
 class TestInterface(BaseDatetimeTests, base.BaseInterfaceTests):
     pass

pandas/tests/extension/test_interval.py

@@ -132,11 +132,6 @@ def test_combine_add(self, data_repeated):
     def test_fillna_length_mismatch(self, data_missing):
         super().test_fillna_length_mismatch(data_missing)
 
-    def test_fillna_copy_series(self, data_missing, using_copy_on_write):
-        super().test_fillna_copy_series(
-            data_missing, no_op_with_cow=using_copy_on_write
-        )
-
 
 class TestMissing(BaseInterval, base.BaseMissingTests):
     # Index.fillna only accepts scalar `value`, so we have to xfail all

pandas/tests/extension/test_period.py

@@ -105,11 +105,6 @@ def test_diff(self, data, periods):
         else:
             super().test_diff(data, periods)
 
-    def test_fillna_copy_series(self, data_missing, using_copy_on_write):
-        super().test_fillna_copy_series(
-            data_missing, no_op_with_cow=using_copy_on_write
-        )
-
 
 class TestInterface(BasePeriodTests, base.BaseInterfaceTests):
     pass

pandas/tests/extension/test_sparse.py

@@ -272,25 +272,32 @@ def test_fillna_frame(self, data_missing):
 class TestMethods(BaseSparseTests, base.BaseMethodsTests):
     _combine_le_expected_dtype = "Sparse[bool]"
 
-    def test_fillna_copy_frame(self, data_missing):
+    def test_fillna_copy_frame(self, data_missing, using_copy_on_write):
         arr = data_missing.take([1, 1])
         df = pd.DataFrame({"A": arr}, copy=False)
 
         filled_val = df.iloc[0, 0]
         result = df.fillna(filled_val)
 
         if hasattr(df._mgr, "blocks"):
-            assert df.values.base is not result.values.base
+            if using_copy_on_write:
+                assert df.values.base is result.values.base
+            else:
+                assert df.values.base is not result.values.base
         assert df.A._values.to_dense() is arr.to_dense()
 
-    def test_fillna_copy_series(self, data_missing):
+    def test_fillna_copy_series(self, data_missing, using_copy_on_write):
         arr = data_missing.take([1, 1])
         ser = pd.Series(arr)
 
         filled_val = ser[0]
         result = ser.fillna(filled_val)
 
-        assert ser._values is not result._values
+        if using_copy_on_write:
+            assert ser._values is result._values
+
+        else:
+            assert ser._values is not result._values
         assert ser._values.to_dense() is arr.to_dense()
 
     @pytest.mark.xfail(reason="Not Applicable")

pandas/tests/groupby/test_raises.py

@@ -303,7 +303,9 @@ def test_groupby_raises_datetime_np(how, by, groupby_series, groupby_func_np):
 
 
 @pytest.mark.parametrize("how", ["method", "agg", "transform"])
-def test_groupby_raises_category(how, by, groupby_series, groupby_func):
+def test_groupby_raises_category(
+    how, by, groupby_series, groupby_func, using_copy_on_write
+):
     # GH#50749
     df = DataFrame(
         {
@@ -370,7 +372,9 @@ def test_groupby_raises_category(how, by, groupby_series, groupby_func):
             TypeError,
             r"Cannot setitem on a Categorical with a new category \(0\), "
             + "set the categories first",
-        ),
+        )
+        if not using_copy_on_write
+        else (None, ""),  # no-op with CoW
         "first": (None, ""),
         "idxmax": (None, ""),
         "idxmin": (None, ""),
@@ -491,7 +495,7 @@ def test_groupby_raises_category_np(how, by, groupby_series, groupby_func_np):
 
 @pytest.mark.parametrize("how", ["method", "agg", "transform"])
 def test_groupby_raises_category_on_category(
-    how, by, groupby_series, groupby_func, observed
+    how, by, groupby_series, groupby_func, observed, using_copy_on_write
 ):
     # GH#50749
     df = DataFrame(
@@ -562,7 +566,9 @@ def test_groupby_raises_category_on_category(
             TypeError,
             r"Cannot setitem on a Categorical with a new category \(0\), "
             + "set the categories first",
-        ),
+        )
+        if not using_copy_on_write
+        else (None, ""),  # no-op with CoW
         "first": (None, ""),
         "idxmax": (ValueError, "attempt to get argmax of an empty sequence")
         if empty_groups