squish.py

# download /srv/reportupdater/output/metrics/sessions in the working folder as sessions.old, then run:
# python squish.py

import csv

from path import glob
from collections import OrderedDict, defaultdict
from datetime import datetime, timedelta


DATE_FORMAT = '%Y-%m-%d'


def getDate(dtString):
    return datetime.strptime(dtString, DATE_FORMAT)


def toString(dt):
    return dt.strftime(DATE_FORMAT)


def lastSunday(dt):
    return dt - timedelta(days=(dt.isoweekday() % 7))


for path in glob.glob('sessions.old/*/*.tsv'):
    output = OrderedDict()

    with open(path) as f:
        currentWeek = defaultdict(lambda: 0)
        previousDateKey = None
        dateKey = None
        header = True

        for row in csv.reader(f, delimiter='\t'):
            if header:
                output[row[0]] = {row[1]: row[2]}
                header = False
                continue

            dateKey = lastSunday(getDate(row[0]))
            actionsKey = row[1]
            actionCount = int(row[2] or 0)

            if dateKey != previousDateKey:
                output[previousDateKey] = currentWeek
                previousDateKey = dateKey
                currentWeek = defaultdict(lambda: 0)

            currentWeek[actionsKey] += actionCount

    with open(path.replace('sessions.old', 'sessions'), 'w') as o:
        writer = csv.writer(o, delimiter='\t')
        writer.writerows((
            [d, k, v]
            for d, a in output.items()
            for k, v in a.items()
        ))