Google Group Members Evolution using Gource

group_evolution.py

'''
Python Google Group Members Evolution Analysis

Gource:
   http://code.google.com/p/gource


'''

import csv
import re
import time
import datetime

FILE_CSV =  'YOUR_INPUT.csv'
OUTPUT_FILE = 'OUTPUT_FILE.txt'


def create_customlog():
    group_history = []
    f = open(FILE_CSV, 'r')
    r = csv.reader(f)
    for line,row in enumerate(r):
        try:
            if row[2] != 'banned':
                timestamp = datetime.datetime(year=int(row[6]),month=int(row[7]),day=int(row[8]),
                                    hour=int(row[9]),minute=int(row[10]),second=int(row[11]))
                user = row[0]
                group_history.append((user,timestamp))
            line+=1
        except:
            print 'Error reading line %d. Skipped.' % line
    f.close()
    #Sort by the timestamp.
    group_history = sorted(group_history,key= lambda sh: sh[1])
    fhandle = open(OUTPUT_FILE, 'a')
    for sh in group_history:
        str_log = '%s|%s|%s|%s'
        #you can easily change the users in this section.   
        if sh[1].year == 2010 and  sh[1].month in [2]:
             user = 'USER_A'
        elif sh[1].year == 2010 and  sh[1].month in [3]:
             user = 'USER_B'
        elif sh[1].year == 2010 and  sh[1].month in [4]:
             user = 'USER_C'
        elif sh[1].year == 2010 and  sh[1].month in [5]:
             user = 'USER_D'
        else:
            user = 'USER_E'
        #Extract the logins of the user's emails.
        usr = re.match(r'(.*)@(.*)',sh[0]).group(1)
        #Creating the formatted string.
        str_log = str_log % (int(time.mktime(sh[1].timetuple())),user,'A',user+ '/'+ usr)
        #Write at the log.
        fhandle.write(str_log+'\n')

    fhandle.close()
    



if __name__ == '__main__':
    create_customlog()