BUG: from_dict ignored order of OrderedDict (#8425)

Author: mazayo

doc/source/whatsnew/v0.25.0.rst

@@ -633,6 +633,7 @@ Indexing
 - Bug in which :meth:`DataFrame.to_csv` caused a segfault for a reindexed data frame, when the indices were single-level :class:`MultiIndex` (:issue:`26303`).
 - Fixed bug where assigning a :class:`arrays.PandasArray` to a :class:`pandas.core.frame.DataFrame` would raise error (:issue:`26390`)
 - Allow keyword arguments for callable local reference used in the :meth:`DataFrame.query` string (:issue:`26426`)
+- Bug in which :meth:`DataFrame.from_dict` ignored order of OrderedDict when orient='index' (:issue:`8425`).
 
 
 Missing

pandas/_libs/lib.pyx

@@ -240,7 +240,8 @@ def fast_unique_multiple(list arrays, sort: bool=True):
 
 @cython.wraparound(False)
 @cython.boundscheck(False)
-def fast_unique_multiple_list(lists: list, sort: bool=True) -> list:
+def fast_unique_multiple_list(lists: list, sort: bool=True,
+                              ordered: bool=False) -> list:
     cdef:
         list buf
         Py_ssize_t k = len(lists)
@@ -257,7 +258,7 @@ def fast_unique_multiple_list(lists: list, sort: bool=True) -> list:
             if val not in table:
                 table[val] = stub
                 uniques.append(val)
-    if sort:
+    if sort and not ordered:
         try:
             uniques.sort()
         except Exception:

pandas/core/indexes/api.py

@@ -125,7 +125,7 @@ def _get_combined_index(indexes, intersect=False, sort=False):
     return index
 
 
-def _union_indexes(indexes, sort=True):
+def _union_indexes(indexes, sort=True, ordered=False):
     """
     Return the union of indexes.
 
@@ -136,6 +136,8 @@ def _union_indexes(indexes, sort=True):
     indexes : list of Index or list objects
     sort : bool, default True
         Whether the result index should come out sorted or not.
+    ordered : bool, default False
+        Whether ordered indexes, such as keys of OrderedDict, is passed
 
     Returns
     -------
@@ -171,7 +173,8 @@ def conv(i):
             return i
 
         return Index(
-            lib.fast_unique_multiple_list([conv(i) for i in inds], sort=sort))
+            lib.fast_unique_multiple_list([conv(i) for i in inds],
+                                          sort=sort, ordered=ordered))
 
     if kind == 'special':
         result = indexes[0]

pandas/core/internals/construction.py

@@ -285,13 +285,16 @@ def extract_index(data):
         have_raw_arrays = False
         have_series = False
         have_dicts = False
+        have_ordered = False
 
         for val in data:
             if isinstance(val, ABCSeries):
                 have_series = True
                 indexes.append(val.index)
             elif isinstance(val, dict):
                 have_dicts = True
+                if isinstance(val, OrderedDict):
+                    have_ordered = True
                 indexes.append(list(val.keys()))
             elif is_list_like(val) and getattr(val, 'ndim', 1) == 1:
                 have_raw_arrays = True
@@ -302,7 +305,7 @@ def extract_index(data):
                              ' an index')
 
         if have_series or have_dicts:
-            index = _union_indexes(indexes)
+            index = _union_indexes(indexes, ordered=have_ordered)
 
         if have_raw_arrays:
             lengths = list(set(raw_lengths))

pandas/tests/frame/test_constructors.py

@@ -1153,7 +1153,7 @@ def test_constructor_list_of_series(self):
 
         sdict = OrderedDict(zip(['x', 'Unnamed 0'], data))
         expected = DataFrame.from_dict(sdict, orient='index')
-        tm.assert_frame_equal(result.sort_index(), expected)
+        tm.assert_frame_equal(result, expected)
 
         # none named
         data = [OrderedDict([['a', 1.5], ['b', 3], ['c', 4], ['d', 6]]),
@@ -1288,7 +1288,7 @@ def test_constructor_list_of_namedtuples(self):
     def test_constructor_orient(self):
         data_dict = self.mixed_frame.T._series
         recons = DataFrame.from_dict(data_dict, orient='index')
-        expected = self.mixed_frame.sort_index()
+        expected = self.mixed_frame.reindex(recons.index)
         tm.assert_frame_equal(recons, expected)
 
         # dict of sequence
@@ -1298,6 +1298,16 @@ def test_constructor_orient(self):
         xp = DataFrame.from_dict(a).T.reindex(list(a.keys()))
         tm.assert_frame_equal(rs, xp)
 
+    def test_constructor_from_ordered_dict(self):
+        # GH8425
+        a = OrderedDict([
+            ('one', OrderedDict([('col_a', 'foo1'), ('col_b', 'bar1')])),
+            ('two', OrderedDict([('col_a', 'foo2'), ('col_b', 'bar2')])),
+            ('three', OrderedDict([('col_a', 'foo3'), ('col_b', 'bar3')]))])
+        expected = DataFrame.from_dict(a, orient='columns').T
+        result = DataFrame.from_dict(a, orient='index')
+        tm.assert_frame_equal(result, expected)
+
     def test_from_dict_columns_parameter(self):
         # GH 18529
         # Test new columns parameter for from_dict that was added to make

pandas/tests/test_lib.py

@@ -203,3 +203,20 @@ def test_get_reverse_indexer(self):
 def test_cache_readonly_preserve_docstrings():
     # GH18197
     assert Index.hasnans.__doc__ is not None
+
+
+def test_fast_unique_multiple_list_with_sort():
+    keys = [['p', 'a'], ['n', 'd'], ['a', 's']]
+
+    expected = np.array(['a', 'd', 'n', 'p', 's'])
+    result = lib.fast_unique_multiple_list(keys, sort=True)
+    tm.assert_numpy_array_equal(np.array(result), expected)
+
+
+def test_fast_unique_multiple_list_with_ordered():
+    # GH8425
+    keys = [['p', 'a'], ['n', 'd'], ['a', 's']]
+
+    expected = np.array(['p', 'a', 'n', 'd', 's'])
+    result = lib.fast_unique_multiple_list(keys, sort=True, ordered=True)
+    tm.assert_numpy_array_equal(np.array(result), expected)