Flip values from columns to rows in DataFrame

Question

I have a dataframe where I used pivot_table to get a occurrence of some values in one row in it. And I appended the series to dataframe like this

for dataframe in fordays: # List of DataFrames to process
    dups = dataframe.pivot_table(index=['DAIHYO_KISHU_SHOHIN_CATEGORY_NAME_SHORT'], aggfunc='size')
    dataframe = dataframe.append(dups, ignore_index=True)
    dataframe.drop_duplicates(inplace=True)

I get

0   COL1      COL2      COL3     val3.1     val3.2     val3.3
1   val1.1   val2.1    val3.1
2   val1.2   val2.2    val3.2
3   val1.3   val2.3    val3.3
4   val1.4   val2.4    val3.1
5   val1.5   val2.5    val3.3
                                   2           1          2

And I would like to have something like this

0   COL1      COL2      COL3     sum  
1   val1.1   val2.1    val3.1    val3.1 - 2
2   val1.2   val2.2    val3.2    val3.2 - 1
3   val1.3   val2.3    val3.3    val3.3 - 2
4   val1.4   val2.4    val3.1
5   val1.5   val2.5    val3.3

How could I achieved that?