Analyze Google Docs user contributions and activity

draftback.txt

Span	Duration	User	# Revisions
Sat, 9/30/2017, 12:56:47 PM - 1:00:18 PM	00:03:31s	Justin Shenk	372
Sat, 9/30/2017, 1:16:02 PM - 1:29:42 PM	00:13:40s	Justin Shenk	1136
Sat, 9/30/2017, 1:45:20 PM - 2:19:19 PM	00:33:59s	Justin Shenk	948
Sat, 9/30/2017, 4:28:13 PM - 4:29:14 PM	00:01:00s	06287517911256457317	38
Sat, 9/30/2017, 4:44:12 PM - 5:03:57 PM	00:19:44s	Justin Shenk	750
Sat, 9/30/2017, 8:34:08 PM - 8:40:02 PM	00:05:54s	Justin Shenk	42
Sat, 9/30/2017, 8:48:53 PM - 9:11:03 PM	00:22:10s	06287517911256457317	1204
Sat, 9/30/2017, 9:11:04 PM - 9:11:04 PM	00:00:00s	Justin Shenk	2
Sat, 9/30/2017, 9:11:04 PM - 9:11:07 PM	00:00:03s	06287517911256457317	6
Sat, 9/30/2017, 9:11:08 PM - 9:11:08 PM	00:00:00s	Justin Shenk	2
Sat, 9/30/2017, 9:11:08 PM - 9:11:08 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:11:08 PM - 9:11:08 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:11:08 PM - 9:11:09 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:11:09 PM - 9:11:09 PM	00:00:00s	Justin Shenk	2
Sat, 9/30/2017, 9:11:10 PM - 9:11:10 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:11:11 PM - 9:11:11 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:11:11 PM - 9:11:23 PM	00:00:11s	06287517911256457317	22
Sat, 9/30/2017, 9:11:27 PM - 9:11:37 PM	00:00:09s	Justin Shenk	12
Sat, 9/30/2017, 9:11:39 PM - 9:11:49 PM	00:00:10s	06287517911256457317	2
Sat, 9/30/2017, 9:11:49 PM - 9:11:49 PM	00:00:00s	Justin Shenk	2
Sat, 9/30/2017, 9:11:50 PM - 9:11:50 PM	00:00:00s	06287517911256457317	1
Sat, 9/30/2017, 9:11:50 PM - 9:11:50 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:11:50 PM - 9:11:51 PM	00:00:00s	06287517911256457317	3
Sat, 9/30/2017, 9:11:51 PM - 9:11:51 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:11:51 PM - 9:11:52 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:11:57 PM - 9:11:57 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:11:57 PM - 9:12:00 PM	00:00:02s	06287517911256457317	17
Sat, 9/30/2017, 9:12:00 PM - 9:12:00 PM	00:00:00s	Justin Shenk	1
Sat, 9/30/2017, 9:12:06 PM - 9:12:06 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:12:06 PM - 9:12:22 PM	00:00:15s	Justin Shenk	6
Sat, 9/30/2017, 9:12:23 PM - 9:12:23 PM	00:00:00s	06287517911256457317	1
Sat, 9/30/2017, 9:12:24 PM - 9:12:26 PM	00:00:01s	Justin Shenk	8
Sat, 9/30/2017, 9:12:26 PM - 9:12:26 PM	00:00:00s	06287517911256457317	1
Sat, 9/30/2017, 9:12:26 PM - 9:12:27 PM	00:00:01s	Justin Shenk	9
Sat, 9/30/2017, 9:12:27 PM - 9:12:27 PM	00:00:00s	06287517911256457317	1
Sat, 9/30/2017, 9:12:28 PM - 9:12:38 PM	00:00:10s	Justin Shenk	6
Sat, 9/30/2017, 9:12:38 PM - 9:12:38 PM	00:00:00s	06287517911256457317	2
Sat, 9/30/2017, 9:13:22 PM - 9:27:19 PM	00:13:56s	Justin Shenk	148
Sat, 9/30/2017, 9:37:21 PM - 9:57:34 PM	00:20:12s	Justin Shenk	776
Sat, 9/30/2017, 10:32:27 PM - 10:41:43 PM	00:09:16s	06287517911256457317	908
Sat, 9/30/2017, 11:27:38 PM - 11:31:03 PM	00:03:24s	Justin Shenk	72

draftback_contributions.py

import pandas as pd

# Click "document graphs and statistics" and save the data to `draftback.txt`
data = pd.read_csv('./draftback.txt',sep='\t')

# Get list of users
users = data['User'].unique()

for user in users:
    user_data = data[data['User'] == user]
    user_revisions = user_data['# Revisions'].sum()
    datetime.strptime(justin['Duration'][0], '%H:%M:%Ss')
    user_durations_raw = user_data['Duration'].values
    durations_split = [x.split(':') for x in user_durations_raw]
    _seconds_raw = [int(x[-1][:-1]) for x in durations_split]
    seconds = sum(_seconds_raw) % 60
    _minutes_raw = [int(x[-2]) for x in durations_split]
    minutes = (sum(_minutes_raw) + sum(_seconds_raw) // 60) % 60
    _hours_raw = [int(x[-3]) for x in durations_split]
    hours = sum(_hours_raw) + sum(_minutes_raw) // 60
    user_revisions = user_data['# Revisions'].values.sum()
    if user == '06287517911256457317':
        user = 'Anonymous'
    print("User {} was active for {} hours, {} minutes and {} seconds and contributed {} revisions".format(user, hours, minutes, seconds, user_revisions))