Fix assignment to multiple columns when some column do not exist

Author: How Si Wei

doc/source/whatsnew/v1.0.0.rst

@@ -37,6 +37,37 @@ Other enhancements
 Backwards incompatible API changes
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
+.. _whatsnew_1000.api_breaking.multicolumn_assignment:
+
+Assignment to multiple columns of a DataFrame when some columns do not exist
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Assignment to multiple columns of a :class:`DataFrame` when some of the columns do not exist would previously assign the values to the last column. Now, new columns would be constructed withe the right values. (:issue:`13658`)
+
+.. ipython:: python
+    df = pd.DataFrame({'a': [0, 1, 2], 'b': [3, 4, 5]})
+    df
+
+*Previous behavior*:
+
+.. code-block:: ipython
+
+    In [3]: df[['a', 'c']] = 1
+    In [4]: df
+    Out[4]:
+       a  b
+    0  1  1
+    1  1  1
+    2  1  1
+
+*New behavior*:
+
+.. ipython:: python
+
+    df[['a', 'c']] = 1
+    df
+
+
 .. _whatsnew_1000.api.other:
 
 - :class:`pandas.core.groupby.GroupBy.transform` now raises on invalid operation names (:issue:`27489`).

pandas/core/frame.py

@@ -3006,6 +3006,12 @@ def _setitem_array(self, key, value):
                 for k1, k2 in zip(key, value.columns):
                     self[k1] = value[k2]
             else:
+                if all(is_hashable(k) for k in key):
+                    for k in key:
+                        try:
+                            self[k]
+                        except KeyError:
+                            self[k] = np.nan
                 indexer = self.loc._get_listlike_indexer(
                     key, axis=1, raise_missing=False
                 )[1]

pandas/core/indexing.py

@@ -12,6 +12,7 @@
 from pandas.core.dtypes.common import (
     ensure_platform_int,
     is_float,
+    is_hashable,
     is_integer,
     is_integer_dtype,
     is_iterator,
@@ -197,6 +198,19 @@ def _get_setitem_indexer(self, key):
     def __setitem__(self, key, value):
         if isinstance(key, tuple):
             key = tuple(com.apply_if_callable(x, self.obj) for x in key)
+            if (
+                self.name == "loc"
+                and len(key) > 1
+                and is_list_like_indexer(key[1])
+                and not isinstance(key[1], tuple)
+                and not com.is_bool_indexer(key[1])
+                and all(is_hashable(k) for k in key[1])
+            ):
+                for k in key[1]:
+                    try:
+                        self.obj[k]
+                    except KeyError:
+                        self.obj[k] = np.nan
         else:
             key = com.apply_if_callable(key, self.obj)
         indexer = self._get_setitem_indexer(key)

pandas/tests/frame/test_indexing.py

@@ -208,6 +208,47 @@ def test_setitem_list_of_tuples(self, float_frame):
         expected = Series(tuples, index=float_frame.index, name="tuples")
         assert_series_equal(result, expected)
 
+    def test_setitem_list_all_missing_columns_scalar(self, float_frame):
+        # GH 26534
+        result = float_frame.copy()
+        result[["E", "F"]] = 1
+        expected = float_frame.copy()
+        # force the dtypes to be float as currently multcolumn assignment does not
+        # change column dtype from float to int even when it's being assigned an int
+        expected["E"] = 1.0
+        expected["F"] = 1.0
+        assert_frame_equal(result, expected)
+
+    def test_setitem_list_some_missing_columns_list(self, float_frame):
+        # GH 26534
+        result = float_frame.copy()
+        result[["A", "E"]] = [1, 2]
+        expected = float_frame.copy()
+        # force the dtypes to be float as currently multcolumn assignment does not
+        # change column dtype from float to int even when it's being assigned an int
+        expected["A"] = 1.0
+        expected["E"] = 2.0
+        assert_frame_equal(result, expected)
+
+    def test_setitem_list_some_missing_columns_dataframe(self, float_frame):
+        # GH 26534
+        result = float_frame.copy()
+        result[["A", "E"]] = float_frame[["B", "C"]]
+        expected = float_frame.copy()
+        expected["A"] = float_frame["B"]
+        expected["E"] = float_frame["C"]
+        assert_frame_equal(result, expected)
+
+    def test_setitem_list_some_missing_columns_2dlist(self):
+        # GH 26534
+        result = pd.DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
+        result[["B", "C", "D"]] = [[7, 8, 9], [10, 11, 12], [13, 14, 15]]
+        expected = pd.DataFrame(
+            [[1, 7, 8, 9], [3, 10, 11, 12], [5, 13, 14, 15]],
+            columns=["A", "B", "C", "D"],
+        )
+        tm.assert_frame_equal(result, expected)
+
     def test_setitem_mulit_index(self):
         # GH7655, test that assigning to a sub-frame of a frame
         # with multi-index columns aligns both rows and columns
@@ -501,13 +542,6 @@ def test_setitem(self, float_frame):
         float_frame["col6"] = series
         tm.assert_series_equal(series, float_frame["col6"], check_names=False)
 
-        msg = (
-            r"\"None of \[Float64Index\(\[.*dtype='float64'\)\] are in the"
-            r" \[columns\]\""
-        )
-        with pytest.raises(KeyError, match=msg):
-            float_frame[np.random.randn(len(float_frame) + 1)] = 1
-
         # set ndarray
         arr = np.random.randn(len(float_frame))
         float_frame["col9"] = arr
@@ -1143,17 +1177,6 @@ def test_fancy_index_int_labels_exceptions(self, float_frame):
             )
             with pytest.raises(KeyError, match=msg):
                 float_frame.ix[["foo", "bar", "baz"]] = 1
-            msg = (
-                r"None of \[Index\(\['E'\], dtype='object'\)\] are in the"
-                r" \[columns\]"
-            )
-            with pytest.raises(KeyError, match=msg):
-                float_frame.ix[:, ["E"]] = 1
-
-            # FIXME: don't leave commented-out
-            # partial setting now allows this GH2578
-            # pytest.raises(KeyError, float_frame.ix.__setitem__,
-            #               (slice(None, None), 'E'), 1)
 
     def test_setitem_fancy_mixed_2d(self, float_string_frame):
 

pandas/tests/indexing/test_loc.py

@@ -808,6 +808,46 @@ def test_loc_setitem_with_scalar_index(self, indexer, value):
 
         assert is_scalar(result) and result == "Z"
 
+    def test_loc_setitem_missing_columns_scalar_index_list_value(self):
+        # GH 26534
+        df = pd.DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
+        df.loc[1, ["C", "D"]] = [7, 8]
+        expected = pd.DataFrame(
+            [[1, 2, np.nan, np.nan], [3, 4, 7, 8], [5, 6, np.nan, np.nan]],
+            columns=["A", "B", "C", "D"],
+        )
+        tm.assert_frame_equal(df, expected)
+
+    def test_loc_setitem_missing_columns_full_index_dataframe_value(self):
+        # GH 26534
+        df = pd.DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
+        df2 = pd.DataFrame([[7, 8], [9, 10], [11, 12]], columns=["A", "C"])
+        df.loc[:, ["A", "C"]] = df2
+        expected = pd.DataFrame(
+            [[7, 2, 8], [9, 4, 10], [11, 6, 12]], columns=["A", "B", "C"]
+        )
+        tm.assert_frame_equal(df, expected)
+
+    def test_loc_setitem_missing_columns_list_index_scalar_value(self):
+        # GH 26534
+        df = pd.DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
+        df.loc[[0, 2], ["B", "C", "D"]] = 9
+        expected = pd.DataFrame(
+            [[1, 9, 9, 9], [3, 4, np.nan, np.nan], [5, 9, 9, 9]],
+            columns=["A", "B", "C", "D"],
+        )
+        tm.assert_frame_equal(df, expected)
+
+    def test_loc_setitem_missing_columns_range_index_2dlist_value(self):
+        # GH 26534
+        df = pd.DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
+        df.loc[1:3, ["B", "C", "D"]] = [[7, 8, 9], [10, 11, 12]]
+        expected = pd.DataFrame(
+            [[1, 2, np.nan, np.nan], [3, 7, 8, 9], [5, 10, 11, 12]],
+            columns=["A", "B", "C", "D"],
+        )
+        tm.assert_frame_equal(df, expected)
+
     def test_loc_coercion(self):
 
         # 12411