read_excel drops columns when reading a one row file with a MultiIndex

[maxgrenderjones]

I'm seeing the weirdest error.

When I read an .xlsx excel file with only one row (example below)...

... and try to set the index_col, then pandas incorrectly reads the columns:

I.e. for a file with only one row, column headers A, B, C, D, E, F

pandas.read_excel(filename, 'Sheet1', index_col=['A', 'B', 'C'])

Yields

(I get an index with columns A, B, C, and a dataframe with columns B, D, F - what happened to E?)

Note that if I don't use index_col, I get a dataframe with columns A-F as expected. Likewise, if I add a row of actual data, the dataframe loads as expected.

Relevant versions:
pandas version 0.15.1 running on 2.7.7 |Anaconda 2.0.1 (64-bit)| (default, Jun 11 2014, 10:40:02) [MSC v.1500 64 bit (AMD64)]
xlrd.__VERSION__=0.9.3

[jorisvandenbossche]

I can confirm this with 0.15.2

[maxgrenderjones]

Would have to dig into it (unlikely to have time soon), but #10184 might suggest how this could be fixed, as the issues seem similar.