Aggregates number of page views from individual URLS based on arguments in GA. Uses CSV.

ga_aggregate_report

import operator
import csv
import sys
import re

source = sys.argv[1]
output = sys.argv[2]
column = sys.argv[3]

reader = csv.reader(open(source, 'rb'), delimiter=',', quotechar='"')
index = 0

keys = {}
for row in reader:

    url = row[0]
    count = row[1]

    key = re.search('(%s=)(\w+)(&)' % column, url).group(2).lower()

    if key not in keys:
        keys[key] = int(count)
    else:
        keys[key] += int(count)

sorted_rows = sorted(keys.iteritems(), key=operator.itemgetter(1))
sorted_rows.reverse()

writer = csv.writer(open(output, "w"), delimiter=",", quotechar='"')
for row in sorted_rows:
    key = row[0]
    val = row[1]

    writer.writerow([key, val])

    print row