BUG: combine_first ignoring others columns if other is empty (pandas-dev#53792)

* BUG: combine_first ignoring others columns if other is empty

* Fix

Author: phofl

doc/source/whatsnew/v2.1.0.rst

@@ -491,6 +491,7 @@ Reshaping
 - Bug in :func:`merge_asof` raising ``KeyError`` for extension dtypes (:issue:`52904`)
 - Bug in :func:`merge_asof` raising ``ValueError`` for data backed by read-only ndarrays (:issue:`53513`)
 - Bug in :meth:`DataFrame.agg` and :meth:`Series.agg` on non-unique columns would return incorrect type when dist-like argument passed in (:issue:`51099`)
+- Bug in :meth:`DataFrame.combine_first` ignoring other's columns if ``other`` is empty (:issue:`53792`)
 - Bug in :meth:`DataFrame.idxmin` and :meth:`DataFrame.idxmax`, where the axis dtype would be lost for empty frames (:issue:`53265`)
 - Bug in :meth:`DataFrame.merge` not merging correctly when having ``MultiIndex`` with single level (:issue:`52331`)
 - Bug in :meth:`DataFrame.stack` losing extension dtypes when columns is a :class:`MultiIndex` and frame contains mixed dtypes (:issue:`45740`)

pandas/core/frame.py

@@ -8344,7 +8344,13 @@ def combiner(x, y):
 
             return expressions.where(mask, y_values, x_values)
 
-        combined = self.combine(other, combiner, overwrite=False)
+        if len(other) == 0:
+            combined = self.reindex(
+                self.columns.append(other.columns.difference(self.columns)), axis=1
+            )
+            combined = combined.astype(other.dtypes)
+        else:
+            combined = self.combine(other, combiner, overwrite=False)
 
         dtypes = {
             col: find_common_type([self.dtypes[col], other.dtypes[col]])

pandas/tests/frame/methods/test_combine_first.py

@@ -538,3 +538,11 @@ def test_midx_losing_dtype():
     )
     expected = DataFrame({"a": [np.nan, 4, 3, 3]}, index=expected_midx)
     tm.assert_frame_equal(result, expected)
+
+
+def test_combine_first_empty_columns():
+    left = DataFrame(columns=["a", "b"])
+    right = DataFrame(columns=["a", "c"])
+    result = left.combine_first(right)
+    expected = DataFrame(columns=["a", "b", "c"])
+    tm.assert_frame_equal(result, expected)