Merge/Join/Append two Pandas DataFrames with MultiIndex columns by both index and cols

Question

I have been banging my head against my desk over this one, can't figure out if there is a way, maybe I am trying something impossible.

I have two DataFrames with MultiIndex columns (three levels) and time Index (single level). The first is like this:

border           a-b                 c-d
from               a         b         c
to                 b         a         d
2009-03-01 -0.778346 -0.928997       NaN
2009-03-02 -1.352559  1.247335       NaN
2009-03-03 -0.967939  0.432638       NaN
2009-03-04  0.786094 -2.209559       NaN
2009-03-05 -0.001338  1.084152       NaN
2009-03-06  1.163334       NaN       NaN
2009-03-07 -0.587593       NaN       NaN
2009-03-08  0.118469       NaN       NaN
2009-03-09       NaN       NaN -1.272959
2009-03-10       NaN       NaN -1.207129
2009-03-11       NaN       NaN  0.244019

To this DF I want to add the following:

border           a-b
from               a
to                 b
2009-03-09  1.243296
2009-03-10 -0.049870
2009-03-11  1.599999

Taking into account both rows and columns indices. The result should be:

border           a-b                 c-d
from               a         b         c
to                 b         a         d
2009-03-01 -0.778346 -0.928997       NaN
2009-03-02 -1.352559  1.247335       NaN
2009-03-03 -0.967939  0.432638       NaN
2009-03-04  0.786094 -2.209559       NaN
2009-03-05 -0.001338  1.084152       NaN
2009-03-06  1.163334       NaN       NaN
2009-03-07 -0.587593       NaN       NaN
2009-03-08  0.118469       NaN       NaN
2009-03-09  1.243296       NaN -1.272959
2009-03-10 -0.049870       NaN -1.207129
2009-03-11  1.599999       NaN  0.244019

I have tried several ways, including with Merge and Join, but can't get it to work.

Any Ideas? Thanks in advance.

P.S. I could post the code I used to generate the two DFs above, if that is on any help, but it's a bit long. But in any case I am looking for a generic answer, the exact name of the columns or the index for rows is irrelevant (could even be a index of integers).

Answer 1

try pandas.DataFrame.update

DataFrame.update(other, join='left', overwrite=True,
                 filter_func=None, raise_conflict=False)

Modify DataFrame in place using non-NA values from passed DataFrame. Aligns on indices