Managing a CSV of users and permissions

I've got a CSV that contains users and permissions in the below format, where users can have as little as one or as many as eight different permissions:

USER,PERM1,PERM2,PERM3,PERM4,PERM5,PERM6,PERM7,PERM8
jdoe,perm1,perm2
tsmith,perm1,perm2,perm3,perm4,perm5,perm6,perm7,perm8


This is the desired format, with each unique user and permission pair on a new line:

USER,PERM
jdoe,perm1
jdoe,perm2
tsmith,perm1
tsmith,perm2
tsmith,perm3
tsmith,perm4
tsmith,perm5
tsmith,perm6
tsmith,perm7
tsmith,perm8


My script below accomplishes this, but it's ugly, repetitive and I know there's a more Pythonic way to do it. Even worse, I had to bring the output file into Excel afterwards to filter and delete the rows with blank PERM values. Any recommendations to shorten this code and cut down on repetition would be much appreciated.

import csv

def reformat_ul(original_ul, formated_ul):
    with open(original_ul) as user_list:
        dict_reader = csv.DictReader(user_list)
        ul = []
        for row in dict_reader:
            ul.append(row)

    with open(formated_ul, 'w') as output2:
        output2.write('USER,PERM\n')
        for uperm in ul:
            p1 = '{},{}\n'.format(uperm['USER'], uperm['PERM1'])
            p2 = '{},{}\n'.format(uperm['USER'], uperm['PERM2'])
            p3 = '{},{}\n'.format(uperm['USER'], uperm['PERM3'])
            p4 = '{},{}\n'.format(uperm['USER'], uperm['PERM4'])
            p5 = '{},{}\n'.format(uperm['USER'], uperm['PERM5'])
            p6 = '{},{}\n'.format(uperm['USER'], uperm['PERM6'])
            p7 = '{},{}\n'.format(uperm['USER'], uperm['PERM7'])
            p8 = '{},{}\n'.format(uperm['USER'], uperm['PERM8'])
            output2.write(p1)
            output2.write(p2)
            output2.write(p3)
            output2.write(p4)
            output2.write(p5)
            output2.write(p6)
            output2.write(p7)
            output2.write(p8)

reformat_ul('user_list.csv', 'output.txt')

This is shorter:

import csv

def reformat_ul(original_ul, formated_ul):
    with open(formated_ul, 'w') as output2:
        output2.write('USER,PERM\n')
        with open(original_ul) as user_list:
            dict_reader = csv.DictReader(user_list)
            for row in dict_reader:
                user = row['USER']
                for key in sorted(row)[:-1]:
                    if row[key]:
                        output2.write("%s,%s\n" % (user, row[key]))

reformat_ul('user_list.csv', 'output.txt')

The highlights:

• It doesn't waste memory by storing the rows in a temporary list. Using the two nested with there, it outputs while reading the output, piping properly

• In sorted(row)[:-1] I take advantage of the fact that I know the column names, and that USER will come after all the PERM1..PERM8. If you need a more flexible implementation then you can amend the deepest if there, for example if key.startswith('PERM') and row[key]

StackExchange Code Review Q#30473, answer score: 2