DataFrame fail if columns is non-unique MultiIndex

[changhiskhan]

In [33]: cols2 = pd.MultiIndex.from_tuples(list(itertools.product(['A', 'B', 'C'], ['X', 'Y', 'Z']))*2)

In [34]: df2 = DataFrame(np.random.randn(3, len(cols2)), columns=cols2)

In [35]: df2
Out[35]: ---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
in ()
----> 1 df2

/Users/changshe/code/ipython/IPython/core/displayhook.pyc in call(self, result)
236 self.start_displayhook()
237 self.write_output_prompt()
--> 238 format_dict = self.compute_format_data(result)
239 self.write_format_data(format_dict)
240 self.update_user_ns(result)

...

/Users/changshe/code/pandas/pandas/core/format.pyc in _get_formatted_column_labels(self)
341 fmt_columns = self.columns.format(sparsify=False, adjoin=False)
342 fmt_columns = zip(fmt_columns)
--> 343 dtypes = self.frame.dtypes.values
344 need_leadsp = dict(zip(fmt_columns, map(is_numeric_dtype, dtypes)))
345 str_columns = zip([[' ' + y

/Users/changshe/code/pandas/pandas/core/frame.pyc in getattr(self, name)
1764 return self[name]
1765 raise AttributeError("'%s' object has no attribute '%s'" %
-> 1766 (type(self).name, name))
1767
1768 def setattr(self, name, value):

AttributeError: 'DataFrame' object has no attribute 'dtypes'

[changhiskhan]

Fixed underlying problem in BlockManager.iget