CoW: Add lazy copy mechanism to DataFrame constructor for dict of Index (#52947)

Author: phofl

doc/source/whatsnew/v2.1.0.rst

@@ -20,6 +20,9 @@ Copy-on-Write improvements
 ^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 - Setting a :class:`Series` into a :class:`DataFrame` now creates a lazy instead of a deep copy (:issue:`53142`)
+- The :class:`DataFrame` constructor, when constructing a DataFrame from a dictionary
+  of Index objects and specifying ``copy=False``, will now use a lazy copy
+  of those Index objects for the columns of the DataFrame (:issue:`52947`)
 
 .. _whatsnew_210.enhancements.enhancement2:
 

pandas/core/indexes/base.py

@@ -22,7 +22,10 @@
 
 import numpy as np
 
-from pandas._config import get_option
+from pandas._config import (
+    get_option,
+    using_copy_on_write,
+)
 
 from pandas._libs import (
     NaT,
@@ -1635,7 +1638,7 @@ def to_frame(
 
         if name is lib.no_default:
             name = self._get_level_names()
-        result = DataFrame({name: self._values.copy()})
+        result = DataFrame({name: self}, copy=not using_copy_on_write())
 
         if index:
             result.index = self

pandas/core/internals/construction.py

@@ -460,13 +460,19 @@ def dict_to_mgr(
         keys = list(data.keys())
         columns = Index(keys) if keys else default_index(0)
         arrays = [com.maybe_iterable_to_list(data[k]) for k in keys]
-        arrays = [arr if not isinstance(arr, Index) else arr._data for arr in arrays]
 
     if copy:
         if typ == "block":
             # We only need to copy arrays that will not get consolidated, i.e.
             #  only EA arrays
-            arrays = [x.copy() if isinstance(x, ExtensionArray) else x for x in arrays]
+            arrays = [
+                x.copy()
+                if isinstance(x, ExtensionArray)
+                else x.copy(deep=True)
+                if isinstance(x, Index)
+                else x
+                for x in arrays
+            ]
         else:
             # dtype check to exclude e.g. range objects, scalars
             arrays = [x.copy() if hasattr(x, "dtype") else x for x in arrays]
@@ -573,10 +579,10 @@ def _homogenize(
     refs: list[Any] = []
 
     for val in data:
-        if isinstance(val, ABCSeries):
+        if isinstance(val, (ABCSeries, Index)):
             if dtype is not None:
                 val = val.astype(dtype, copy=False)
-            if val.index is not index:
+            if isinstance(val, ABCSeries) and val.index is not index:
                 # Forces alignment. No need to copy data since we
                 # are putting it into an ndarray later
                 val = val.reindex(index, copy=False)

pandas/tests/copy_view/index/test_index.py

@@ -153,3 +153,17 @@ def test_infer_objects(using_copy_on_write):
     view_.iloc[0, 0] = "aaaa"
     if using_copy_on_write:
         tm.assert_index_equal(idx, expected, check_names=False)
+
+
+def test_index_to_frame(using_copy_on_write):
+    idx = Index([1, 2, 3], name="a")
+    expected = idx.copy(deep=True)
+    df = idx.to_frame()
+    if using_copy_on_write:
+        assert np.shares_memory(get_array(df, "a"), idx._values)
+        assert not df._mgr._has_no_reference(0)
+    else:
+        assert not np.shares_memory(get_array(df, "a"), idx._values)
+
+    df.iloc[0, 0] = 100
+    tm.assert_index_equal(idx, expected)

pandas/tests/copy_view/test_constructors.py

@@ -340,3 +340,15 @@ def test_dataframe_from_records_with_dataframe(using_copy_on_write):
         tm.assert_frame_equal(df, df_orig)
     else:
         tm.assert_frame_equal(df, df2)
+
+
+def test_frame_from_dict_of_index(using_copy_on_write):
+    idx = Index([1, 2, 3])
+    expected = idx.copy(deep=True)
+    df = DataFrame({"a": idx}, copy=False)
+    assert np.shares_memory(get_array(df, "a"), idx._values)
+    if using_copy_on_write:
+        assert not df._mgr._has_no_reference(0)
+
+        df.iloc[0, 0] = 100
+        tm.assert_index_equal(idx, expected)

pandas/tests/indexes/test_common.py

@@ -31,7 +31,7 @@
 
 class TestCommon:
     @pytest.mark.parametrize("name", [None, "new_name"])
-    def test_to_frame(self, name, index_flat):
+    def test_to_frame(self, name, index_flat, using_copy_on_write):
         # see GH#15230, GH#22580
         idx = index_flat
 
@@ -45,7 +45,8 @@ def test_to_frame(self, name, index_flat):
         assert df.index is idx
         assert len(df.columns) == 1
         assert df.columns[0] == idx_name
-        assert df[idx_name].values is not idx.values
+        if not using_copy_on_write:
+            assert df[idx_name].values is not idx.values
 
         df = idx.to_frame(index=False, name=idx_name)
         assert df.index is not idx