BUG: DataFrame constructor fails when columns is set and data=[] (GH9948/9939)

Author: evanpw

doc/source/whatsnew/v0.16.1.txt

@@ -230,7 +230,7 @@ Bug Fixes
 - Bug in ``where`` when dtype is ``datetime64/timedelta64``, but dtype of other is not (:issue:`9804`)
 - Bug in ``MultiIndex.sortlevel()`` results in unicode level name breaks (:issue:`9856`)
 - Bug in which ``groupby.transform`` incorrectly enforced output dtypes to match input dtypes. (:issue:`9807`)
-
+- Bug in ``DataFrame`` constructor when ``columns`` parameter is set, and ``data`` is an empty list (:issue:`9939`)
 - Bug in bar plot with ``log=True`` raises ``TypeError`` if all values are less than 1 (:issue:`9905`)
 - Bug in horizontal bar plot ignores ``log=True`` (:issue:`9905`)
 
@@ -249,4 +249,6 @@ Bug Fixes
 - Bug causing an exception when setting an empty range using ``DataFrame.loc`` (:issue:`9596`)
 
 
+
+
 - Bug in hiding ticklabels with subplots and shared axes when adding a new plot to an existing grid of axes (:issue:`9158`)

pandas/core/frame.py

@@ -265,8 +265,7 @@ def __init__(self, data=None, index=None, columns=None, dtype=None,
                     mgr = self._init_ndarray(data, index, columns, dtype=dtype,
                                              copy=copy)
             else:
-                mgr = self._init_ndarray(data, index, columns, dtype=dtype,
-                                         copy=copy)
+                mgr = self._init_dict({}, index, columns, dtype=dtype)
         elif isinstance(data, collections.Iterator):
             raise TypeError("data argument can't be an iterator")
         else:

pandas/tests/test_frame.py

@@ -3165,6 +3165,19 @@ def test_constructor_empty_list(self):
         expected = DataFrame(index=[])
         assert_frame_equal(df, expected)
 
+        # GH 9939
+        df = DataFrame([], columns=['A', 'B'])
+        expected = DataFrame({}, columns=['A', 'B'])
+        assert_frame_equal(df, expected)
+
+        # Empty generator: list(empty_gen()) == []
+        def empty_gen():
+            return
+            yield
+
+        df = DataFrame(empty_gen(), columns=['A', 'B'])
+        assert_frame_equal(df, expected)
+
     def test_constructor_list_of_lists(self):
         # GH #484
         l = [[1, 'a'], [2, 'b']]