Backport PR #46335: BUG: replace with value also being replaced (#46340)

Co-authored-by: Luke Manley <[email protected]>

Author: meeseeksmachine

doc/source/whatsnew/v1.4.2.rst

@@ -18,6 +18,7 @@ Fixed regressions
 - Fixed regression in :func:`read_csv` killing python process when invalid file input was given for ``engine="c"`` (:issue:`45957`)
 - Fixed memory performance regression in :meth:`Series.fillna` when called on a :class:`DataFrame` column with ``inplace=True`` (:issue:`46149`)
 - Provided an alternative solution for passing custom Excel formats in :meth:`.Styler.to_excel`, which was a regression based on stricter CSS validation. Examples available in the documentation for :meth:`.Styler.format` (:issue:`46152`)
+- Fixed regression in :meth:`DataFrame.replace` when a replacement value was also a target for replacement (:issue:`46335`)
 - Fixed regression in :meth:`DataFrame.loc.__setitem__` losing :class:`MultiIndex` names if :class:`DataFrame`  was empty before (:issue:`46317`)
 -
 

pandas/core/internals/blocks.py

@@ -708,9 +708,18 @@ def replace(
 
         else:
             # split so that we only upcast where necessary
-            return self.split_and_operate(
-                type(self).replace, to_replace, value, inplace=True
-            )
+            blocks = []
+            for i, nb in enumerate(self._split()):
+                blocks.extend(
+                    type(self).replace(
+                        nb,
+                        to_replace=to_replace,
+                        value=value,
+                        inplace=True,
+                        mask=mask[i : i + 1],
+                    )
+                )
+            return blocks
 
     @final
     def _replace_regex(

pandas/tests/frame/methods/test_replace.py

@@ -1519,3 +1519,10 @@ def test_replace_regex_dtype_frame(self, regex):
         expected_df2 = DataFrame({"A": [1], "B": ["1"]})
         result_df2 = df2.replace(to_replace="0", value=1, regex=regex)
         tm.assert_frame_equal(result_df2, expected_df2)
+
+    def test_replace_with_value_also_being_replaced(self):
+        # GH46306
+        df = DataFrame({"A": [0, 1, 2], "B": [1, 0, 2]})
+        result = df.replace({0: 1, 1: np.nan})
+        expected = DataFrame({"A": [1, np.nan, 2], "B": [np.nan, 1, 2]})
+        tm.assert_frame_equal(result, expected)