Merge pull request #3632 from jreback/assign_df

BUG: (GH3626) issue with alignment of a DataFrame setitem with a piece of another DataFrame

Author: jreback

RELEASE.rst

@@ -126,6 +126,7 @@ pandas 0.11.1
   - Correctly parse date columns with embedded (nan/NaT) into datetime64[ns] dtype in ``read_csv``
     when ``parse_dates`` is specified (GH3062_)
   - Fix not consolidating before to_csv (GH3624_)
+  - Fix alignment issue when setitem in a DataFrame with a piece of a DataFrame (GH3626_)
 
 .. _GH3164: https://github.com/pydata/pandas/issues/3164
 .. _GH2786: https://github.com/pydata/pandas/issues/2786
@@ -178,6 +179,7 @@ pandas 0.11.1
 .. _GH3611: https://github.com/pydata/pandas/issues/3611
 .. _GH3062: https://github.com/pydata/pandas/issues/3062
 .. _GH3624: https://github.com/pydata/pandas/issues/3624
+.. _GH3626: https://github.com/pydata/pandas/issues/3626
 .. _GH1512: https://github.com/pydata/pandas/issues/1512
 
 

pandas/core/indexing.py

@@ -138,14 +138,14 @@ def setter(item, v):
                         # align to
                         if item in value:
                             v = value[item]
-                            v = v.reindex(self.obj[item].reindex(v.index).dropna().index)
+                            v = v.reindex(self.obj[item].index & v.index)
                             setter(item, v.values)
                         else:
                             setter(item, np.nan)
 
                 # we have an equal len ndarray
-                elif isinstance(value, np.ndarray) and value.ndim > 1:
-                    if len(labels) != len(value):
+                elif isinstance(value, np.ndarray) and value.ndim == 2:
+                    if len(labels) != value.shape[1]:
                         raise ValueError('Must have equal len keys and value when'
                                          ' setting with an ndarray')
 

pandas/tests/test_indexing.py

@@ -853,6 +853,41 @@ def test_iloc_panel_issue(self):
         self.assert_(p.iloc[1, :3, 1].shape == (3,))
         self.assert_(p.iloc[:3, 1, 1].shape == (3,))
 
+    def test_multi_assign(self):
+
+        # GH 3626, an assignement of a sub-df to a df
+        df = DataFrame({'FC':['a','b','a','b','a','b'],
+                        'PF':[0,0,0,0,1,1],
+                        'col1':range(6),
+                        'col2':range(6,12)})
+        df.ix[1,0]=np.nan
+        df2 = df.copy()
+
+        mask=~df2.FC.isnull()
+        cols=['col1', 'col2']
+
+        dft = df2 * 2
+        dft.ix[3,3] = np.nan
+
+        expected = DataFrame({'FC':['a',np.nan,'a','b','a','b'],
+                              'PF':[0,0,0,0,1,1],
+                              'col1':Series([0,1,4,6,8,10],dtype='float64'),
+                              'col2':[12,7,16,np.nan,20,22]})
+        
+
+        # frame on rhs
+        df2.ix[mask, cols]= dft.ix[mask, cols]
+        assert_frame_equal(df2,expected)
+        df2.ix[mask, cols]= dft.ix[mask, cols]
+        assert_frame_equal(df2,expected)
+
+        # with an ndarray on rhs
+        df2 = df.copy()
+        df2.ix[mask, cols]= dft.ix[mask, cols].values
+        assert_frame_equal(df2,expected)
+        df2.ix[mask, cols]= dft.ix[mask, cols].values
+        assert_frame_equal(df2,expected)
+
 
 if __name__ == '__main__':
     import nose