merge two dataframes and add column level with names

Question

Hi I've been digging through the concat, join, and merge methods for pandas and can't seem to find what I want.

Lets assume I have two dataframes

A = pd.DataFrame("A",index=[0,1,2,3,4],columns=['Col 1','Col 2','Col 3'])
B = pd.DataFrame("B",index=[0,1,2,3,4],columns=['Col 1','Col 2','Col 3'])
>>> A
  Col 1 Col 2 Col 3
0     A     A     A
1     A     A     A
2     A     A     A
3     A     A     A
4     A     A     A
>>> B
  Col 1 Col 2 Col 3
0     B     B     B
1     B     B     B
2     B     B     B
3     B     B     B
4     B     B     B

Now I want to make a new dataframe with the columns merged, I think its easiest to explain if I make a multi index for how I want the columns

index = pd.MultiIndex.from_product([A.columns.values,['A','B']])
>>> index
MultiIndex(levels=[['Col 1', 'Col 2', 'Col 3'], ['A', 'B']],
           labels=[[0, 0, 1, 1, 2, 2], [0, 1, 0, 1, 0, 1]])

Now if I make an empty dataframe with this multi index for the columns

empty_df = pd.DataFrame('-',index=A.index,columns=index)
>>> empty_df
  Col 1    Col 2    Col 3
      A  B     A  B     A  B
0     -  -     -  -     -  -
1     -  -     -  -     -  -
2     -  -     -  -     -  -
3     -  -     -  -     -  -
4     -  -     -  -     -  -

My question is, what merge, concat, or join do I use to obtain that? I've tried multiple things for concat...inner,outer etc. I can't seem to find what I want. Only thing I can think of is making the empty dataframe and then back filling.

Edit: After trying out Jezrael's response, it is close but not it exactly. What I want is like nested columns of sort? For example

empty_df['Col 1']
>>> empty_df['Col 1']
   A  B
0  -  -
1  -  -
2  -  -
3  -  -
4  -  -

Or

>>> empty_df['Col 1']['A']
0    -
1    -
2    -
3    -
4    -
Name: A, dtype: object

So this is a solution I've come up with but its from iterating over the columns.

row_idx = A.index.union(B.index)
col_idx = pd.MultiIndex.from_product([A.columns.values,['A','B']])
new_df = pd.DataFrame('-',index=row_idx,columns=col_idx)
for column in A.columns:
   new_df.loc[:,(column,'A')] = A[column]
   new_df.loc[:,(column,'B')] = B[column]
>>> new_df
  Col 1    Col 2    Col 3
      A  B     A  B     A  B
0     A  B     A  B     A  B
1     A  B     A  B     A  B
2     A  B     A  B     A  B
3     A  B     A  B     A  B
4     A  B     A  B     A  B
>>> new_df['Col 1']
   A  B
0  A  B
1  A  B
2  A  B
3  A  B
4  A  B
>>> new_df['Col 1']['A']
0    A
1    A
2    A
3    A
4    A
Name: A, dtype: object

Answer 1

I think you need concat with keys parameter and axis=1, last change order of levels by DataFrame.swaplevel and sorting by first level by DataFrame.sort_index:

df1 = (pd.concat([A, B], axis=1, keys=('A','B'))
         .swaplevel(0,1, axis=1)
         .sort_index(axis=1, level=0))
print (df1)
  Col 1    Col 2    Col 3   
      A  B     A  B     A  B
0     A  B     A  B     A  B
1     A  B     A  B     A  B
2     A  B     A  B     A  B
3     A  B     A  B     A  B
4     A  B     A  B     A  B

For working with MultiIndex is possible use DataFrame.xs:

print (df1.xs('Col 1', axis=1, level=0))
   A  B
0  A  B
1  A  B
2  A  B
3  A  B
4  A  B

If want select MultiIndex column use tuple:

print (df1[('Col 1', 'A')])
0    A
1    A
2    A
3    A
4    A
Name: (Col 1, A), dtype: object

If want select by index and by column use loc:

print (df1.loc[4, ('Col 1', 'A')])
A