BUG? Parser adds empty MultiIndex level names

[chris-b1]

In [179]: midx = MultiIndex.from_tuples([('A', 1, 2), ('A', 1, 2), ('B', 1, 2)])
    ...:  df = DataFrame(np.random.randn(3, 3), index=midx,
    ...:                 columns=['x', 'y', 'z'])

In [180]: df
Out[180]: 
            x         y         z
A 1 2  1.236013  2.239172 -1.513059
    2 -0.159620  0.222250 -1.635894
B 1 2  0.211036 -0.441065 -0.661737

In [181]: df.to_csv('test.csv')

In [182]: result = pd.read_csv('test.csv', index_col=[0, 1, 2])

In [183]: result
Out[183]: 
                                x         y         z
Unnamed: 1 Unnamed: 2                              
A 1          2           1.236013  2.239172 -1.513059
             2          -0.159620  0.222250 -1.635894
B 1          2           0.211036 -0.441065 -0.661737

[jreback]

use read_csv

from_csv has certain defaults to make it a mirror of to_csv exactly

[chris-b1]

Updated example, same behavior. Original block was from a test here with a TODO:
https://github.com/pydata/pandas/blob/master/pandas/tests/test_frame.py#L6340

[jreback]

look for an open issue about this
iirc this was noted when I originally wrote this

[chris-b1]

Hmm, not have any luck finding one, although I might just have the wrong search terms. Here's the pr where that line was added (changing default of assert_frame_equal to include names) #2964

[jreback]

xref #6618

[mroeschke]

Looks like names are not added anymore. Could use a test.

In [145]: pd.read_csv('test.csv', index_col=[0, 1, 2])
Out[145]:
              x         y         z
A 1 2  0.593604 -0.836343 -0.786727
    2 -0.077489 -0.829363 -0.453257
B 1 2  0.131251 -0.622905  0.672914

In [146]: pd.__version__
Out[146]: '0.26.0.dev0+576.gde67bb72e'