Merge pull request #6135 from hayd/mi_partial_sorting

ENH partial sorting for mi in sortlevel

Author: jreback

doc/source/v0.14.0.txt

@@ -510,6 +510,8 @@ Enhancements
 
 - ``quotechar``, ``doublequote``, and ``escapechar`` can now be specified when
   using ``DataFrame.to_csv`` (:issue:`5414`, :issue:`4528`)
+- Partially sort by only the specified levels of a MultiIndex with the
+  ``sort_remaining`` boolean kwarg. (:issue:`3984`)
 - Added a ``to_julian_date`` function to ``TimeStamp`` and ``DatetimeIndex``
   to convert to the Julian Date used primarily in astronomy. (:issue:`4041`)
 - ``DataFrame.to_stata`` will now check data for compatibility with Stata data types

pandas/core/frame.py

@@ -2648,7 +2648,8 @@ def trans(v):
         else:
             return self._constructor(new_data).__finalize__(self)
 
-    def sortlevel(self, level=0, axis=0, ascending=True, inplace=False):
+    def sortlevel(self, level=0, axis=0, ascending=True,
+                  inplace=False, sort_remaining=True):
         """
         Sort multilevel index by chosen axis and primary level. Data will be
         lexicographically sorted by the chosen level followed by the other
@@ -2661,6 +2662,8 @@ def sortlevel(self, level=0, axis=0, ascending=True, inplace=False):
         ascending : boolean, default True
         inplace : boolean, default False
             Sort the DataFrame without creating a new instance
+        sort_remaining : boolean, default True
+            Sort by the other levels too.
 
         Returns
         -------
@@ -2671,7 +2674,8 @@ def sortlevel(self, level=0, axis=0, ascending=True, inplace=False):
         if not isinstance(the_axis, MultiIndex):
             raise TypeError('can only sort by level with a hierarchical index')
 
-        new_axis, indexer = the_axis.sortlevel(level, ascending=ascending)
+        new_axis, indexer = the_axis.sortlevel(level, ascending=ascending,
+                                               sort_remaining=sort_remaining)
 
         if self._is_mixed_type and not inplace:
             ax = 'index' if axis == 0 else 'columns'

pandas/core/index.py

@@ -3096,17 +3096,19 @@ def reorder_levels(self, order):
     def __getslice__(self, i, j):
         return self.__getitem__(slice(i, j))
 
-    def sortlevel(self, level=0, ascending=True):
+    def sortlevel(self, level=0, ascending=True, sort_remaining=True):
         """
         Sort MultiIndex at the requested level. The result will respect the
         original ordering of the associated factor at that level.
 
         Parameters
         ----------
-        level : int or str, default 0
+        level : list-like, int or str, default 0
             If a string is given, must be a name of the level
+            If list-like must be names or ints of levels.
         ascending : boolean, default True
             False to sort in descending order
+        sort_remaining : sort by the remaining levels after level.
 
         Returns
         -------
@@ -3115,24 +3117,35 @@ def sortlevel(self, level=0, ascending=True):
         from pandas.core.groupby import _indexer_from_factorized
 
         labels = list(self.labels)
+        shape = list(self.levshape)
 
-        level = self._get_level_number(level)
-        primary = labels.pop(level)
+        if isinstance(level, (str, int)):
+            level = [level]
+        level = [self._get_level_number(lev) for lev in level]
 
-        shape = list(self.levshape)
-        primshp = shape.pop(level)
+        # partition labels and shape
+        primary = tuple(labels.pop(lev - i) for i, lev in enumerate(level))
+        primshp = tuple(shape.pop(lev - i) for i, lev in enumerate(level))
 
-        indexer = _indexer_from_factorized((primary,) + tuple(labels),
-                                           (primshp,) + tuple(shape),
+        if sort_remaining:
+            primary += primary + tuple(labels)
+            primshp += primshp + tuple(shape)
+            sortorder = None
+        else:
+            sortorder = level[0]
+
+        indexer = _indexer_from_factorized(primary,
+                                           primshp,
                                            compress=False)
+
         if not ascending:
             indexer = indexer[::-1]
 
         indexer = com._ensure_platform_int(indexer)
         new_labels = [lab.take(indexer) for lab in self.labels]
 
         new_index = MultiIndex(labels=new_labels, levels=self.levels,
-                               names=self.names, sortorder=level,
+                               names=self.names, sortorder=sortorder,
                                verify_integrity=False)
 
         return new_index, indexer

pandas/core/series.py

@@ -1784,7 +1784,7 @@ def _try_kind_sort(arr):
         else:
             return result.__finalize__(self)
 
-    def sortlevel(self, level=0, ascending=True):
+    def sortlevel(self, level=0, ascending=True, sort_remaining=True):
         """
         Sort Series with MultiIndex by chosen level. Data will be
         lexicographically sorted by the chosen level followed by the other
@@ -1802,7 +1802,8 @@ def sortlevel(self, level=0, ascending=True):
         if not isinstance(self.index, MultiIndex):
             raise TypeError('can only sort by level with a hierarchical index')
 
-        new_index, indexer = self.index.sortlevel(level, ascending=ascending)
+        new_index, indexer = self.index.sortlevel(level, ascending=ascending,
+                                                 sort_remaining=sort_remaining)
         new_values = self.values.take(indexer)
         return self._constructor(new_values,
                                  index=new_index).__finalize__(self)

pandas/tests/test_frame.py

@@ -10116,6 +10116,15 @@ def test_sort_index_duplicates(self):
         result = df.sort_index(by=('a',1))
         assert_frame_equal(result, expected)
 
+    def test_sortlevel(self):
+        mi = MultiIndex.from_tuples([[1, 1, 3], [1, 1, 1]], names=list('ABC'))
+        df = DataFrame([[1, 2], [3, 4]], mi)
+        res = df.sortlevel('A', sort_remaining=False)
+        assert_frame_equal(df, res)
+
+        res = df.sortlevel(['A', 'B'], sort_remaining=False)
+        assert_frame_equal(df, res)
+
     def test_sort_datetimes(self):
 
         # GH 3461, argsort / lexsort differences for a datetime column

pandas/tests/test_index.py

@@ -2420,6 +2420,11 @@ def test_sortlevel(self):
         sorted_idx, _ = index.sortlevel(1, ascending=False)
         self.assert_(sorted_idx.equals(expected[::-1]))
 
+    def test_sortlevel_not_sort_remaining(self):
+        mi = MultiIndex.from_tuples([[1, 1, 3], [1, 1, 1]], names=list('ABC'))
+        sorted_idx, _ = mi.sortlevel('A', sort_remaining=False)
+        self.assert_(sorted_idx.equals(mi))
+
     def test_sortlevel_deterministic(self):
         tuples = [('bar', 'one'), ('foo', 'two'), ('qux', 'two'),
                   ('foo', 'one'), ('baz', 'two'), ('qux', 'one')]

pandas/tests/test_series.py

@@ -5302,6 +5302,23 @@ def test_unstack(self):
         unstacked = s.unstack(0)
         assert_frame_equal(unstacked, expected)
 
+    def test_sortlevel(self):
+        mi = MultiIndex.from_tuples([[1, 1, 3], [1, 1, 1]], names=list('ABC'))
+        s = Series([1, 2], mi)
+        backwards = s.iloc[[1, 0]]
+
+        res = s.sortlevel('A')
+        assert_series_equal(backwards, res)
+
+        res = s.sortlevel(['A', 'B'])
+        assert_series_equal(backwards, res)
+
+        res = s.sortlevel('A', sort_remaining=False)
+        assert_series_equal(s, res)
+
+        res = s.sortlevel(['A', 'B'], sort_remaining=False)
+        assert_series_equal(s, res)
+
     def test_head_tail(self):
         assert_series_equal(self.series.head(), self.series[:5])
         assert_series_equal(self.series.tail(), self.series[-5:])