BUG: from_dict ignored order of OrderedDict (#8425) (#26875)

Author: mazayo

doc/source/whatsnew/v0.25.0.rst

@@ -1152,6 +1152,7 @@ Reshaping
 - Bug in :func:`DataFrame.sort_index` where an error is thrown when a multi-indexed ``DataFrame`` is sorted on all levels with the initial level sorted last (:issue:`26053`)
 - Bug in :meth:`Series.nlargest` treats ``True`` as smaller than ``False`` (:issue:`26154`)
 - Bug in :func:`DataFrame.pivot_table` with a :class:`IntervalIndex` as pivot index would raise ``TypeError`` (:issue:`25814`)
+- Bug in which :meth:`DataFrame.from_dict` ignored order of ``OrderedDict`` when ``orient='index'`` (:issue:`8425`).
 - Bug in :meth:`DataFrame.transpose` where transposing a DataFrame with a timezone-aware datetime column would incorrectly raise ``ValueError`` (:issue:`26825`)
 - Bug in :func:`pivot_table` when pivoting a timezone aware column as the ``values`` would remove timezone information (:issue:`14948`)
 - Bug in :func:`merge_asof` when specifying multiple ``by`` columns where one is ``datetime64[ns, tz]`` dtype (:issue:`26649`)

pandas/core/internals/construction.py

@@ -9,6 +9,7 @@
 
 from pandas._libs import lib
 from pandas._libs.tslibs import IncompatibleFrequency, OutOfBoundsDatetime
+import pandas.compat as compat
 from pandas.compat import raise_with_traceback
 
 from pandas.core.dtypes.cast import (
@@ -338,13 +339,16 @@ 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
@@ -353,8 +357,10 @@ def extract_index(data):
         if not indexes and not raw_lengths:
             raise ValueError("If using all scalar values, you must pass" " an index")
 
-        if have_series or have_dicts:
+        if have_series:
             index = _union_indexes(indexes)
+        elif have_dicts:
+            index = _union_indexes(indexes, sort=not (compat.PY36 or have_ordered))
 
         if have_raw_arrays:
             lengths = list(set(raw_lengths))

pandas/tests/frame/test_constructors.py

@@ -517,7 +517,8 @@ def test_constructor_subclass_dict(self, float_frame):
             dct.update(v.to_dict())
             data[k] = dct
         frame = DataFrame(data)
-        tm.assert_frame_equal(float_frame.sort_index(), frame)
+        expected = frame.reindex(index=float_frame.index)
+        tm.assert_frame_equal(float_frame, expected)
 
     def test_constructor_dict_block(self):
         expected = np.array([[4.0, 3.0, 2.0, 1.0]])
@@ -1203,7 +1204,7 @@ 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 = [
@@ -1342,7 +1343,7 @@ def test_constructor_list_of_namedtuples(self):
     def test_constructor_orient(self, float_string_frame):
         data_dict = float_string_frame.T._series
         recons = DataFrame.from_dict(data_dict, orient="index")
-        expected = float_string_frame.sort_index()
+        expected = float_string_frame.reindex(index=recons.index)
         tm.assert_frame_equal(recons, expected)
 
         # dict of sequence
@@ -1351,6 +1352,19 @@ def test_constructor_orient(self, float_string_frame):
         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