analytics: Add 15day_actives CountStat.

This commit is contained in:
Rishi Gupta
2017-03-15 23:58:23 -07:00
committed by Tim Abbott
parent 9b661ca91f
commit 2f74ccabf9
3 changed files with 68 additions and 0 deletions

View File

@@ -375,6 +375,24 @@ count_message_by_stream_query = """
"""
zerver_count_message_by_stream = ZerverCountQuery(Message, StreamCount, count_message_by_stream_query)
check_useractivityinterval_by_user_query = """
INSERT INTO analytics_usercount
(user_id, realm_id, value, property, subgroup, end_time)
SELECT
zerver_userprofile.id, zerver_userprofile.realm_id, 1, '%(property)s', %(subgroup)s, %%(time_end)s
FROM zerver_userprofile
JOIN zerver_useractivityinterval
ON
zerver_userprofile.id = zerver_useractivityinterval.user_profile_id
WHERE
zerver_useractivityinterval.end >= %%(time_start)s AND
zerver_useractivityinterval.start < %%(time_end)s
%(join_args)s
GROUP BY zerver_userprofile.id %(group_by_clause)s
"""
zerver_check_useractivityinterval_by_user = ZerverCountQuery(
UserActivityInterval, UserCount, check_useractivityinterval_by_user_query)
def do_pull_minutes_active(stat, start_time, end_time):
# type: (CountStat, datetime, datetime) -> None
timer_start = time.time()
@@ -410,6 +428,10 @@ count_stats_ = [
(Message, 'sending_client_id'), CountStat.DAY),
CountStat('messages_in_stream:is_bot:day', zerver_count_message_by_stream, {},
(UserProfile, 'is_bot'), CountStat.DAY),
# The minutes=15 part is due to the 15 minutes added in
# zerver.lib.actions.do_update_user_activity_interval.
CountStat('15day_actives::day', zerver_check_useractivityinterval_by_user, {},
None, CountStat.DAY, interval=timedelta(days=15)-timedelta(minutes=15)),
LoggingCountStat('active_users_log:is_bot:day', RealmCount, CountStat.DAY),
CustomPullCountStat('minutes_active::day', UserCount, CountStat.DAY, do_pull_minutes_active)
]

View File

@@ -489,6 +489,50 @@ class TestCountStats(AnalyticsTestCase):
user_profile=user, start=self.TIME_ZERO-start_offset,
end=self.TIME_ZERO-end_offset)
def test_15day_actives(self):
# type: () -> None
stat = COUNT_STATS['15day_actives::day']
self.current_property = stat.property
_15day = 15*self.DAY - 15*self.MINUTE
# Outside time range, should not appear. Also tests upper boundary.
user1 = self.create_user()
self.create_interval(user1, _15day + self.DAY, _15day + timedelta(seconds=1))
self.create_interval(user1, timedelta(0), -self.HOUR)
# On lower boundary, should appear
user2 = self.create_user()
self.create_interval(user2, _15day + self.DAY, _15day)
# Multiple intervals, including one outside boundary
user3 = self.create_user()
self.create_interval(user3, 20*self.DAY, 19*self.DAY)
self.create_interval(user3, 20*self.HOUR, 19*self.HOUR)
self.create_interval(user3, 20*self.MINUTE, 19*self.MINUTE)
# Intervals crossing boundary
user4 = self.create_user()
self.create_interval(user4, 20*self.DAY, 10*self.DAY)
user5 = self.create_user()
self.create_interval(user5, self.MINUTE, -self.MINUTE)
# Interval subsuming time range
user6 = self.create_user()
self.create_interval(user6, 20*self.DAY, -2*self.DAY)
# Second realm
user7 = self.create_user(realm=self.second_realm)
self.create_interval(user7, 20*self.MINUTE, 19*self.MINUTE)
do_fill_count_stat_at_hour(stat, self.TIME_ZERO)
self.assertTableState(UserCount, ['value', 'user'],
[[1, user2], [1, user3], [1, user4], [1, user5], [1, user6], [1, user7]])
self.assertTableState(RealmCount, ['value', 'realm'],
[[5, self.default_realm], [1, self.second_realm]])
self.assertTableState(InstallationCount, ['value'], [[6]])
self.assertTableState(StreamCount, [], [])
def test_minutes_active(self):
# type: () -> None
stat = COUNT_STATS['minutes_active::day']

View File

@@ -2301,6 +2301,8 @@ def streams_to_dicts_sorted(streams):
def do_update_user_activity_interval(user_profile, log_time):
# type: (UserProfile, datetime.datetime) -> None
# Update any stats in analytics.lib.counts.count_stats_ that rely on
# this if the 15 minutes is changed to something else.
effective_end = log_time + datetime.timedelta(minutes=15)
# This code isn't perfect, because with various races we might end
# up creating two overlapping intervals, but that shouldn't happen