def count_active_users(frame, end_date, days_back):
"""
Tally active users according to externally-defined heuristics,
specifically client-reported subsession_start_date.
:frame DataFrame(): conforming to main_summary schema
:end_date DT.date(): the last day on which to count activity.
:days_back int: how far back to go; 0 for daily and 28 for monthly.
Note that return values are not stable over time as activity trickles
in with long submission latencies.
See https://bugzilla.mozilla.org/show_bug.cgi?id=1240849
"""
params = U.generate_filter_parameters(end_date, days_back)
filtered = filter_date_range(
frame,
frame.subsession_start_date,
params["min_activity_iso"],
params["max_activity_iso"],
frame.submission_date_s3,
params["min_submission_string"],
params["max_submission_string"],
)
return count_distinct_clientids(filtered)
def test_generate_filter_parameters():
"""
Check the two meaningful cases: DAU (0 days) and MAU(28 days).
"""
expected0 = {
'min_activity_iso': '2017-01-31',
'max_activity_iso': '2017-02-01',
'min_submission_string': '20170131',
'max_submission_string': '20170210'
}
actual0 = utils.generate_filter_parameters(DT.date(2017, 1, 31), 0)
assert expected0 == actual0, str(actual0)
expected28 = {
'min_activity_iso': '2017-01-03',
'max_activity_iso': '2017-02-01',
'min_submission_string': '20170103',
'max_submission_string': '20170210'
}
actual28 = utils.generate_filter_parameters(DT.date(2017, 1, 31), 28)
assert expected28 == actual28
def extract_month(first_day, frame):
"""
Pull a month's worth of activity out of frame according to the
heuristics implemented in moztelemetry.standards.
:first_day DT.date(Y, m, 1)
:frame DataFrame homologous with main_summary
"""
month = first_day.month
day_pointer = first_day
while day_pointer.month == month:
day_pointer += DT.timedelta(1)
last_day = day_pointer - DT.timedelta(1)
days_back = (last_day - first_day).days
params = generate_filter_parameters(last_day, days_back)
filtered = filter_date_range(frame, frame.subsession_start_date,
params['min_activity_iso'],
params['max_activity_iso'],
frame.submission_date_s3,
params['min_submission_string'],
params['max_submission_string'])
return filtered