sum values of two csv files and results to a new file

Pandas has good support for reading and manipulating CSV files.

    import pandas as pd

    df = pd.concat([
        pd.read_csv('csv1.csv'),
        pd.read_csv('csv2.csv')
    ])

    result=df.groupby('link', as_index=False).sum()
    result['node'] = 'allnode'

    result.to_csv('result.csv')

Where csv1.csv is:

    node,link,rate-in,rate-out
    node1,link1,10,20
    node1,link2,30,50
    node1,link3,40,60

And csv2.csv is:

    node,link,rate-in,rate-out
    node2,link1,20,10
    node2,link2,50,70
    node2,link3,80,40

The result.csv file will contain:

    link,rate-in,rate-out,node
    link1,30,30,allnode
    link2,80,120,allnode
    link3,120,100,allnode

If you want to reorder the columns you can do that by providing an order list

    result[[
        'node','link','rate-in','rate-out'
    ]].to_csv('result.csv', index=False)