BUG: TypeError in index coercion

closes #12916
closes #12893

Author: Matt Wittmann

doc/source/whatsnew/v0.18.1.txt

@@ -305,6 +305,7 @@ Bug Fixes
 - Bug in ``Series.map`` raises ``TypeError`` if its dtype is ``category`` or tz-aware ``datetime`` (:issue:`12473`)
 
 
+- Bug in index coercion when falling back from ```RangeIndex``` construction (:issue:`12893`)
 
 - Bug in slicing subclassed ``DataFrame`` defined to return subclassed ``Series`` may return normal ``Series`` (:issue:`11559`)
 

pandas/indexes/multi.py

@@ -666,7 +666,7 @@ def get_level_values(self, level):
         filled = algos.take_1d(unique.values, labels,
                                fill_value=unique._na_value)
         _simple_new = unique._simple_new
-        values = _simple_new(filled, self.names[num],
+        values = _simple_new(filled, name=self.names[num],
                              freq=getattr(unique, 'freq', None),
                              tz=getattr(unique, 'tz', None))
         return values

pandas/tests/indexes/test_multi.py

@@ -2147,3 +2147,26 @@ def test_partial_string_timestamp_multiindex(self):
         # Slicing date on first level should break (of course)
         with assertRaises(KeyError):
             df_swap.loc['2016-01-01']
+
+    def test_rangeindex_fallback_coercion_bug(self):
+        # GH 12893
+        foo = pd.DataFrame(np.arange(100).reshape((10, 10)))
+        bar = pd.DataFrame(np.arange(100).reshape((10, 10)))
+        df = pd.concat({'foo': foo.stack(), 'bar': bar.stack()}, axis=1)
+        df.index.names = ['fizz', 'buzz']
+
+        str(df)
+        expected = pd.DataFrame({'bar': np.arange(100),
+                                 'foo': np.arange(100)},
+                                index=pd.MultiIndex.from_product(
+                                    [range(10), range(10)],
+                                    names=['fizz', 'buzz']))
+        tm.assert_frame_equal(df, expected, check_like=True)
+
+        result = df.index.get_level_values('fizz')
+        expected = pd.Int64Index(np.arange(10), name='fizz').repeat(10)
+        tm.assert_index_equal(result, expected)
+
+        result = df.index.get_level_values('buzz')
+        expected = pd.Int64Index(np.tile(np.arange(10), 10), name='buzz')
+        tm.assert_index_equal(result, expected)