BUG: fix combine_first reorders columns (#60791)

* Add test

* Fix combine_first reorders columns

* Add whatsnew

* Fix corner case when self is empty and future.infer_string is True

* Update

Author: yuanx749

doc/source/whatsnew/v3.0.0.rst

@@ -758,6 +758,7 @@ Groupby/resample/rolling
 Reshaping
 ^^^^^^^^^
 - Bug in :func:`qcut` where values at the quantile boundaries could be incorrectly assigned (:issue:`59355`)
+- Bug in :meth:`DataFrame.combine_first` not preserving the column order (:issue:`60427`)
 - Bug in :meth:`DataFrame.join` inconsistently setting result index name (:issue:`55815`)
 - Bug in :meth:`DataFrame.join` when a :class:`DataFrame` with a :class:`MultiIndex` would raise an ``AssertionError`` when :attr:`MultiIndex.names` contained ``None``. (:issue:`58721`)
 - Bug in :meth:`DataFrame.merge` where merging on a column containing only ``NaN`` values resulted in an out-of-bounds array access (:issue:`59421`)

pandas/core/frame.py

@@ -8671,6 +8671,7 @@ def combine(
         2  NaN  3.0 1.0
         """
         other_idxlen = len(other.index)  # save for compare
+        other_columns = other.columns
 
         this, other = self.align(other)
         new_index = this.index
@@ -8681,8 +8682,8 @@ def combine(
         if self.empty and len(other) == other_idxlen:
             return other.copy()
 
-        # sorts if possible; otherwise align above ensures that these are set-equal
-        new_columns = this.columns.union(other.columns)
+        # preserve column order
+        new_columns = self.columns.union(other_columns, sort=False)
         do_fill = fill_value is not None
         result = {}
         for col in new_columns:

pandas/tests/frame/methods/test_combine_first.py

@@ -380,7 +380,7 @@ def test_combine_first_with_asymmetric_other(self, val):
         df2 = DataFrame({"isBool": [True]})
 
         res = df1.combine_first(df2)
-        exp = DataFrame({"isBool": [True], "isNum": [val]})
+        exp = DataFrame({"isNum": [val], "isBool": [True]})
 
         tm.assert_frame_equal(res, exp)
 
@@ -555,3 +555,13 @@ def test_combine_first_empty_columns():
     result = left.combine_first(right)
     expected = DataFrame(columns=["a", "b", "c"])
     tm.assert_frame_equal(result, expected)
+
+
+def test_combine_first_preserve_column_order():
+    # GH#60427
+    df1 = DataFrame({"B": [1, 2, 3], "A": [4, None, 6]})
+    df2 = DataFrame({"A": [5]}, index=[1])
+
+    result = df1.combine_first(df2)
+    expected = DataFrame({"B": [1, 2, 3], "A": [4.0, 5.0, 6.0]})
+    tm.assert_frame_equal(result, expected)