Merge pull request #8480 from behzadnouri/align-mi

BUG: sub-frame assignment of a multi-index frame breaks alignment

Author: jreback

doc/source/v0.15.0.txt

@@ -1066,3 +1066,4 @@ Bug Fixes
 - Bug in ``NDFrame.loc`` indexing when row/column names were lost when target was a list/ndarray (:issue:`6552`)
 - Regression in ``NDFrame.loc`` indexing when rows/columns were converted to Float64Index if target was an empty list/ndarray (:issue:`7774`)
 - Bug in Series that allows it to be indexed by a DataFrame which has unexpected results.  Such indexing is no longer permitted (:issue:`8444`)
+- Bug in item assignment of a DataFrame with multi-index columns where right-hand-side columns were not aligned (:issue:`7655`)

pandas/core/frame.py

@@ -2194,6 +2194,15 @@ def reindexer(value):
             value = reindexer(value)
 
         elif isinstance(value, DataFrame):
+            # align right-hand-side columns if self.columns
+            # is multi-index and self[key] is a sub-frame
+            if isinstance(self.columns, MultiIndex) and key in self.columns:
+                loc = self.columns.get_loc(key)
+                if isinstance(loc, (slice, Series, np.ndarray, Index)):
+                    cols = _maybe_droplevels(self.columns[loc], key)
+                    if len(cols) and not cols.equals(value.columns):
+                        value = value.reindex_axis(cols, axis=1)
+            # now align rows
             value = reindexer(value).T
 
         elif isinstance(value, Categorical):

pandas/tests/test_frame.py

@@ -224,6 +224,31 @@ def test_setitem_list_of_tuples(self):
         expected = Series(tuples, index=self.frame.index)
         assert_series_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
+        it = ['jim', 'joe', 'jolie'], ['first', 'last'], \
+             ['left', 'center', 'right']
+
+        cols = MultiIndex.from_product(it)
+        index = pd.date_range('20141006',periods=20)
+        vals = np.random.randint(1, 1000, (len(index), len(cols)))
+        df = pd.DataFrame(vals, columns=cols, index=index)
+
+        i, j = df.index.values.copy(), it[-1][:]
+
+        np.random.shuffle(i)
+        df['jim'] = df['jolie'].loc[i, ::-1]
+        assert_frame_equal(df['jim'], df['jolie'])
+
+        np.random.shuffle(j)
+        df[('joe', 'first')] = df[('jolie', 'last')].loc[i, j]
+        assert_frame_equal(df[('joe', 'first')], df[('jolie', 'last')])
+
+        np.random.shuffle(j)
+        df[('joe', 'last')] = df[('jolie', 'first')].loc[i, j]
+        assert_frame_equal(df[('joe', 'last')], df[('jolie', 'first')])
+
     def test_getitem_boolean(self):
         # boolean indexing
         d = self.tsframe.index[10]