Merge pull request #3464 from jreback/GH3461

BUG: GH3461 Fix sorting in a frame with a list of columns which contains datetime64[ns]

Author: jreback

RELEASE.rst

@@ -54,6 +54,7 @@ pandas 0.12.0
   - ``.loc`` was not raising when passed an integer list (GH3449_)
   - Unordered time series selection was misbehaving when using label slicing (GH3448_)
   - Duplicate indexes with getitem will return items in the correct order (GH3455_, GH3457_)
+  - Fix sorting in a frame with a list of columns which contains datetime64[ns] dtypes (GH3461_)
 
 .. _GH3164: https://github.com/pydata/pandas/issues/3164
 .. _GH3251: https://github.com/pydata/pandas/issues/3251
@@ -64,6 +65,7 @@ pandas 0.12.0
 .. _GH3437: https://github.com/pydata/pandas/issues/3437
 .. _GH3455: https://github.com/pydata/pandas/issues/3455
 .. _GH3457: https://github.com/pydata/pandas/issues/3457
+.. _GH3461: https://github.com/pydata/pandas/issues/3461
 .. _GH3448: https://github.com/pydata/pandas/issues/3448
 .. _GH3449: https://github.com/pydata/pandas/issues/3449
 

pandas/core/common.py

@@ -1477,6 +1477,8 @@ def is_timedelta64_dtype(arr_or_dtype):
         tipo = arr_or_dtype.dtype.type
     return issubclass(tipo, np.timedelta64)
 
+def needs_i8_conversion(arr_or_dtype):
+    return is_datetime64_dtype(arr_or_dtype) or is_timedelta64_dtype(arr_or_dtype)
 
 def is_float_dtype(arr_or_dtype):
     if isinstance(arr_or_dtype, np.dtype):

pandas/core/frame.py

@@ -3144,7 +3144,12 @@ def sort_index(self, axis=0, by=None, ascending=True, inplace=False):
                                          % str(x))
                     keys.append(k)
 
-                keys = [self[x].values for x in by]
+                def trans(v):
+                    if com.needs_i8_conversion(v):
+                        return v.view('i8')
+                    return v
+
+                keys = [trans(self[x].values) for x in by]
                 indexer = _lexsort_indexer(keys, orders=ascending)
                 indexer = com._ensure_platform_int(indexer)
             else:

pandas/tests/test_frame.py

@@ -7737,6 +7737,25 @@ def test_sort_index_duplicates(self):
         except Exception, e:
             self.assertTrue('duplicate' in str(e))
 
+    def test_sort_datetimes(self):
+
+        # GH 3461, argsort / lexsort differences for a datetime column
+        df = DataFrame(['a','a','a','b','c','d','e','f','g'],
+                       columns=['A'],
+                       index=date_range('20130101',periods=9))
+        dts = [ Timestamp(x) for x in  ['2004-02-11','2004-01-21','2004-01-26','2005-09-20','2010-10-04','2009-05-12','2008-11-12','2010-09-28','2010-09-28'] ]
+        df['B'] = dts[::2] + dts[1::2]
+        df['C'] = 2.
+        df['A1'] = 3.
+
+        df1 = df.sort(columns='A')
+        df2 = df.sort(columns=['A'])
+        assert_frame_equal(df1,df2)
+
+        df1 = df.sort(columns='B')
+        df2 = df.sort(columns=['B'])
+        assert_frame_equal(df1,df2)
+
     def test_frame_column_inplace_sort_exception(self):
         s = self.frame['A']
         self.assertRaises(Exception, s.sort)