retrieve-pv-from-ga.py

#!/usr/bin/env python
"""
This script can retrieve pageviews count by month and export to a CSV file

Usage:

>> python pv.py 2013 <input.csv> <output-csv>

"""
import sys
import csv
import calendar

from gdata.analytics.client import AnalyticsClient
import gdata


def ga_pageview(client, view_id, path, start_date, end_date):
    data_query = gdata.analytics.client.DataFeedQuery({
        'ids': view_id,
        'dimensions': '', # ga:source,ga:medium
        'metrics': 'ga:pageviews',
        'filters': 'ga:pagePath%s' % path,
        'start-date': start_date,
        'end-date': end_date,
    })
 
    feed = client.GetDataFeed(data_query)
    
    try:
        pageview = feed.entry[0].metric[0].value
    except IndexError:
        pageview = 0

    return pageview




SOURCE_APP_NAME = '<your-google-app-name>'
GA_VIEW_ID = 'ga:XXXXXXX' # example: ga:1111111

client = AnalyticsClient(source=SOURCE_APP_NAME)

client.client_login(
    '<your-google-username>',
    '<your-google-password>',
    source=SOURCE_APP_NAME,
    service=client.auth_service,
    account_type = 'GOOGLE',
)
 

data = csv.reader(open(input_csv, "rb"))
output = csv.writer(open(output_csv, "wb"))

for name, username in data:
    username = username.strip()
    print name, username

    pageviews = []
    for month in range(1, 13):
        last_day = calendar.monthrange(year, month)[1]
        
        start_date = "%d-%02d-%02d" % (year, month, 1)
        end_date = "%d-%02d-%02d" % (year, month, last_day)

        pageview = ga_pageview(client, GA_VIEW_ID, '=~/.*/%s/' % username, start_date, end_date)

        pageviews.append(pageview)

    pageviews.insert(0, year)

    label = "%s, %s" % (name, username)
    output.writerow([label])
    output.writerow(pageviews)