def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - timedelta(hours = 1)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property,))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." % (fill_state.state,))
currently_filled = currently_filled + timedelta(hours = 1)
while currently_filled <= fill_to_time:
logger.info("START %s %s %s" % (stat.property, stat.interval, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + timedelta(hours = 1)
logger.info("DONE %s %s (%dms)" % (stat.property, stat.interval, (end-start)*1000))
def test_process_stat(self):
# type: () -> None
# process new stat
current_time = installation_epoch() + self.HOUR
stat = self.make_dummy_count_stat(current_time)
property = stat.property
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# dirty stat
FillState.objects.filter(property=property).update(state=FillState.STARTED)
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# clean stat, no update
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# clean stat, with update
current_time = current_time + self.HOUR
stat = self.make_dummy_count_stat(current_time)
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 2)
def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - timedelta(hours = 1)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property,))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise ValueError("Unknown value for FillState.state: %s." % (fill_state.state,))
currently_filled = currently_filled + timedelta(hours = 1)
while currently_filled <= fill_to_time:
logger.info("START %s %s %s" % (stat.property, stat.interval, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + timedelta(hours = 1)
logger.info("DONE %s %s (%dms)" % (stat.property, stat.interval, (end-start)*1000))
def test_process_stat(self):
# type: () -> None
# process new stat
current_time = installation_epoch() + self.HOUR
stat = self.make_dummy_count_stat(current_time)
property = stat.property
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# dirty stat
FillState.objects.filter(property=property).update(state=FillState.STARTED)
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# clean stat, no update
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 1)
# clean stat, with update
current_time = current_time + self.HOUR
stat = self.make_dummy_count_stat(current_time)
process_count_stat(stat, current_time)
self.assertFillStateEquals(current_time)
self.assertEqual(InstallationCount.objects.filter(property=property).count(), 2)
def process_count_stat(stat: CountStat, fill_to_time: datetime,
realm: Optional[Realm]=None) -> None:
# TODO: The realm argument is not yet supported, in that we don't
# have a solution for how to update FillState if it is passed. It
# exists solely as partial plumbing for when we do fully implement
# doing single-realm analytics runs for use cases like data import.
#
# Also, note that for the realm argument to be properly supported,
# the CountStat object passed in needs to have come from
# E.g. get_count_stats(realm), i.e. have the realm_id already
# entered into the SQL query defined by the CountState object.
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency,))
verify_UTC(fill_to_time)
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" % (fill_to_time,))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s", stat.property, currently_filled)
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s", stat.property, fill_state.end_time)
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s", stat.property)
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." % (fill_state.state,))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning("DependentCountStat %s run before dependency %s.",
stat.property, dependency)
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s", stat.property, currently_filled)
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled, realm)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)", stat.property, (end-start)*1000)
def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency, ))
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" %
(fill_to_time, ))
if fill_to_time.tzinfo is None:
raise ValueError("fill_to_time must be timezone aware: %s" %
(fill_to_time, ))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property, ))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." %
(fill_state.state, ))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning(
"DependentCountStat %s run before dependency %s." %
(stat.property, dependency))
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s" % (stat.property, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)" % (stat.property, (end - start) * 1000))
def get_fill_state(self) -> Dict[str, Any]:
if not Realm.objects.exists():
return {"status": 0, "message": "No realms exist, so not checking FillState."}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = stat.last_successful_fill()
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimeZoneNotUTCException:
return {"status": 2, "message": f"FillState not in UTC for {property}"}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {
"status": 2,
"message": f"FillState not on {stat.frequency} boundary for {property}",
}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {"status": 0, "message": "FillState looks fine."}
if len(critical_unfilled_properties) == 0:
return {
"status": 1,
"message": "Missed filling {} once.".format(
", ".join(warning_unfilled_properties),
),
}
return {
"status": 2,
"message": "Missed filling {} once. Missed filling {} at least twice.".format(
", ".join(warning_unfilled_properties),
", ".join(critical_unfilled_properties),
),
}
def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency,))
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" % (fill_to_time,))
if fill_to_time.tzinfo is None:
raise ValueError("fill_to_time must be timezone aware: %s" % (fill_to_time,))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property,))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." % (fill_state.state,))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning("DependentCountStat %s run before dependency %s." %
(stat.property, dependency))
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s" % (stat.property, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)" % (stat.property, (end-start)*1000))
def get_fill_state(self):
# type: () -> Dict[str, Any]
if not Realm.objects.exists():
return {'status': 0, 'message': 'No realms exist, so not checking FillState.'}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = last_successful_fill(property)
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimezoneNotUTCException:
return {'status': 2, 'message': 'FillState not in UTC for %s' % (property,)}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {'status': 2, 'message': 'FillState not on %s boundary for %s' %
(stat.frequency, property)}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {'status': 0, 'message': 'FillState looks fine.'}
if len(critical_unfilled_properties) == 0:
return {'status': 1, 'message': 'Missed filling %s once.' %
(', '.join(warning_unfilled_properties),)}
return {'status': 2, 'message': 'Missed filling %s once. Missed filling %s at least twice.' %
(', '.join(warning_unfilled_properties), ', '.join(critical_unfilled_properties))}
def get_fill_state(self) -> Dict[str, Any]:
if not Realm.objects.exists():
return {'status': 0, 'message': 'No realms exist, so not checking FillState.'}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = last_successful_fill(property)
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimezoneNotUTCException:
return {'status': 2, 'message': 'FillState not in UTC for %s' % (property,)}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {'status': 2, 'message': 'FillState not on %s boundary for %s' %
(stat.frequency, property)}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {'status': 0, 'message': 'FillState looks fine.'}
if len(critical_unfilled_properties) == 0:
return {'status': 1, 'message': 'Missed filling %s once.' %
(', '.join(warning_unfilled_properties),)}
return {'status': 2, 'message': 'Missed filling %s once. Missed filling %s at least twice.' %
(', '.join(warning_unfilled_properties), ', '.join(critical_unfilled_properties))}
def get_chart_data(request: HttpRequest,
user_profile: UserProfile,
chart_name: str = REQ(),
min_length: Optional[int] = REQ(
converter=to_non_negative_int, default=None),
start: Optional[datetime] = REQ(converter=to_utc_datetime,
default=None),
end: Optional[datetime] = REQ(converter=to_utc_datetime,
default=None),
realm: Optional[Realm] = None,
for_installation: bool = False) -> HttpResponse:
aggregate_table = RealmCount
if for_installation:
aggregate_table = InstallationCount
if chart_name == 'number_of_humans':
stats = [
COUNT_STATS['1day_actives::day'],
COUNT_STATS['realm_active_humans::day'],
COUNT_STATS['active_users_audit:is_bot:day']
]
tables = [aggregate_table]
subgroup_to_label = {
stats[0]: {
None: '_1day'
},
stats[1]: {
None: '_15day'
},
stats[2]: {
'false': 'all_time'
}
} # type: Dict[CountStat, Dict[Optional[str], str]]
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == 'messages_sent_over_time':
stats = [COUNT_STATS['messages_sent:is_bot:hour']]
tables = [aggregate_table, UserCount]
subgroup_to_label = {stats[0]: {'false': 'human', 'true': 'bot'}}
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == 'messages_sent_by_message_type':
stats = [COUNT_STATS['messages_sent:message_type:day']]
tables = [aggregate_table, UserCount]
subgroup_to_label = {
stats[0]: {
'public_stream': _('Public streams'),
'private_stream': _('Private streams'),
'private_message': _('Private messages'),
'huddle_message': _('Group private messages')
}
}
labels_sort_function = lambda data: sort_by_totals(data['everyone'])
include_empty_subgroups = True
elif chart_name == 'messages_sent_by_client':
stats = [COUNT_STATS['messages_sent:client:day']]
tables = [aggregate_table, UserCount]
# Note that the labels are further re-written by client_label_map
subgroup_to_label = {
stats[0]: {
str(id): name
for id, name in Client.objects.values_list('id', 'name')
}
}
labels_sort_function = sort_client_labels
include_empty_subgroups = False
else:
raise JsonableError(_("Unknown chart name: %s") % (chart_name, ))
# Most likely someone using our API endpoint. The /stats page does not
# pass a start or end in its requests.
if start is not None:
start = convert_to_UTC(start)
if end is not None:
end = convert_to_UTC(end)
if start is not None and end is not None and start > end:
raise JsonableError(
_("Start time is later than end time. Start: %(start)s, End: %(end)s"
) % {
'start': start,
'end': end
})
if realm is None:
realm = user_profile.realm
if start is None:
if for_installation:
start = installation_epoch()
else:
start = realm.date_created
if end is None:
end = max(
last_successful_fill(stat.property) or datetime.min.replace(
tzinfo=timezone_utc) for stat in stats)
if end is None or start > end:
logging.warning(
"User from realm %s attempted to access /stats, but the computed "
"start time: %s (creation of realm or installation) is later than the computed "
"end time: %s (last successful analytics update). Is the "
"analytics cron job running?" % (realm.string_id, start, end))
raise JsonableError(
_("No analytics data available. Please contact your server administrator."
))
assert len(set([stat.frequency for stat in stats])) == 1
end_times = time_range(start, end, stats[0].frequency, min_length)
data = {
'end_times': end_times,
'frequency': stats[0].frequency
} # type: Dict[str, Any]
aggregation_level = {
InstallationCount: 'everyone',
RealmCount: 'everyone',
UserCount: 'user'
}
# -1 is a placeholder value, since there is no relevant filtering on InstallationCount
id_value = {
InstallationCount: -1,
RealmCount: realm.id,
UserCount: user_profile.id
}
for table in tables:
data[aggregation_level[table]] = {}
for stat in stats:
data[aggregation_level[table]].update(
get_time_series_by_subgroup(stat, table, id_value[table],
end_times, subgroup_to_label[stat],
include_empty_subgroups))
if labels_sort_function is not None:
data['display_order'] = labels_sort_function(data)
else:
data['display_order'] = None
return json_success(data=data)
def get_chart_data(request: HttpRequest, user_profile: UserProfile, chart_name: str=REQ(),
min_length: Optional[int]=REQ(converter=to_non_negative_int, default=None),
start: Optional[datetime]=REQ(converter=to_utc_datetime, default=None),
end: Optional[datetime]=REQ(converter=to_utc_datetime, default=None),
realm: Optional[Realm]=None, for_installation: bool=False) -> HttpResponse:
aggregate_table = RealmCount
if for_installation:
aggregate_table = InstallationCount
if chart_name == 'number_of_humans':
stats = [
COUNT_STATS['1day_actives::day'],
COUNT_STATS['realm_active_humans::day'],
COUNT_STATS['active_users_audit:is_bot:day']]
tables = [aggregate_table]
subgroup_to_label = {
stats[0]: {None: '_1day'},
stats[1]: {None: '_15day'},
stats[2]: {'false': 'all_time'}} # type: Dict[CountStat, Dict[Optional[str], str]]
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == 'messages_sent_over_time':
stats = [COUNT_STATS['messages_sent:is_bot:hour']]
tables = [aggregate_table, UserCount]
subgroup_to_label = {stats[0]: {'false': 'human', 'true': 'bot'}}
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == 'messages_sent_by_message_type':
stats = [COUNT_STATS['messages_sent:message_type:day']]
tables = [aggregate_table, UserCount]
subgroup_to_label = {stats[0]: {'public_stream': _('Public streams'),
'private_stream': _('Private streams'),
'private_message': _('Private messages'),
'huddle_message': _('Group private messages')}}
labels_sort_function = lambda data: sort_by_totals(data['everyone'])
include_empty_subgroups = True
elif chart_name == 'messages_sent_by_client':
stats = [COUNT_STATS['messages_sent:client:day']]
tables = [aggregate_table, UserCount]
# Note that the labels are further re-written by client_label_map
subgroup_to_label = {stats[0]:
{str(id): name for id, name in Client.objects.values_list('id', 'name')}}
labels_sort_function = sort_client_labels
include_empty_subgroups = False
else:
raise JsonableError(_("Unknown chart name: %s") % (chart_name,))
# Most likely someone using our API endpoint. The /stats page does not
# pass a start or end in its requests.
if start is not None:
start = convert_to_UTC(start)
if end is not None:
end = convert_to_UTC(end)
if start is not None and end is not None and start > end:
raise JsonableError(_("Start time is later than end time. Start: %(start)s, End: %(end)s") %
{'start': start, 'end': end})
if realm is None:
realm = user_profile.realm
if start is None:
if for_installation:
start = installation_epoch()
else:
start = realm.date_created
if end is None:
end = max(last_successful_fill(stat.property) or
datetime.min.replace(tzinfo=timezone_utc) for stat in stats)
if end is None or start > end:
logging.warning("User from realm %s attempted to access /stats, but the computed "
"start time: %s (creation of realm or installation) is later than the computed "
"end time: %s (last successful analytics update). Is the "
"analytics cron job running?" % (realm.string_id, start, end))
raise JsonableError(_("No analytics data available. Please contact your server administrator."))
assert len(set([stat.frequency for stat in stats])) == 1
end_times = time_range(start, end, stats[0].frequency, min_length)
data = {'end_times': end_times, 'frequency': stats[0].frequency} # type: Dict[str, Any]
aggregation_level = {InstallationCount: 'everyone', RealmCount: 'everyone', UserCount: 'user'}
# -1 is a placeholder value, since there is no relevant filtering on InstallationCount
id_value = {InstallationCount: -1, RealmCount: realm.id, UserCount: user_profile.id}
for table in tables:
data[aggregation_level[table]] = {}
for stat in stats:
data[aggregation_level[table]].update(get_time_series_by_subgroup(
stat, table, id_value[table], end_times, subgroup_to_label[stat], include_empty_subgroups))
if labels_sort_function is not None:
data['display_order'] = labels_sort_function(data)
else:
data['display_order'] = None
return json_success(data=data)
def get_chart_data(
request: HttpRequest,
user_profile: UserProfile,
chart_name: str = REQ(),
min_length: Optional[int] = REQ(converter=to_non_negative_int,
default=None),
start: Optional[datetime] = REQ(converter=to_utc_datetime, default=None),
end: Optional[datetime] = REQ(converter=to_utc_datetime, default=None),
realm: Optional[Realm] = None,
for_installation: bool = False,
remote: bool = False,
remote_realm_id: Optional[int] = None,
server: Optional["RemoteZulipServer"] = None,
) -> HttpResponse:
TableType = Union[Type["RemoteInstallationCount"], Type[InstallationCount],
Type["RemoteRealmCount"], Type[RealmCount], ]
if for_installation:
if remote:
assert settings.ZILENCER_ENABLED
aggregate_table: TableType = RemoteInstallationCount
assert server is not None
else:
aggregate_table = InstallationCount
else:
if remote:
assert settings.ZILENCER_ENABLED
aggregate_table = RemoteRealmCount
assert server is not None
assert remote_realm_id is not None
else:
aggregate_table = RealmCount
tables: Union[Tuple[TableType], Tuple[TableType, Type[UserCount]]]
if chart_name == "number_of_humans":
stats = [
COUNT_STATS["1day_actives::day"],
COUNT_STATS["realm_active_humans::day"],
COUNT_STATS["active_users_audit:is_bot:day"],
]
tables = (aggregate_table, )
subgroup_to_label: Dict[CountStat, Dict[Optional[str], str]] = {
stats[0]: {
None: "_1day"
},
stats[1]: {
None: "_15day"
},
stats[2]: {
"false": "all_time"
},
}
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == "messages_sent_over_time":
stats = [COUNT_STATS["messages_sent:is_bot:hour"]]
tables = (aggregate_table, UserCount)
subgroup_to_label = {stats[0]: {"false": "human", "true": "bot"}}
labels_sort_function = None
include_empty_subgroups = True
elif chart_name == "messages_sent_by_message_type":
stats = [COUNT_STATS["messages_sent:message_type:day"]]
tables = (aggregate_table, UserCount)
subgroup_to_label = {
stats[0]: {
"public_stream": _("Public streams"),
"private_stream": _("Private streams"),
"private_message": _("Private messages"),
"huddle_message": _("Group private messages"),
}
}
labels_sort_function = lambda data: sort_by_totals(data["everyone"])
include_empty_subgroups = True
elif chart_name == "messages_sent_by_client":
stats = [COUNT_STATS["messages_sent:client:day"]]
tables = (aggregate_table, UserCount)
# Note that the labels are further re-written by client_label_map
subgroup_to_label = {
stats[0]: {
str(id): name
for id, name in Client.objects.values_list("id", "name")
}
}
labels_sort_function = sort_client_labels
include_empty_subgroups = False
elif chart_name == "messages_read_over_time":
stats = [COUNT_STATS["messages_read::hour"]]
tables = (aggregate_table, UserCount)
subgroup_to_label = {stats[0]: {None: "read"}}
labels_sort_function = None
include_empty_subgroups = True
else:
raise JsonableError(_("Unknown chart name: {}").format(chart_name))
# Most likely someone using our API endpoint. The /stats page does not
# pass a start or end in its requests.
if start is not None:
start = convert_to_UTC(start)
if end is not None:
end = convert_to_UTC(end)
if start is not None and end is not None and start > end:
raise JsonableError(
_("Start time is later than end time. Start: {start}, End: {end}").
format(
start=start,
end=end,
))
if realm is None:
# Note that this value is invalid for Remote tables; be
# careful not to access it in those code paths.
realm = user_profile.realm
if remote:
# For remote servers, we don't have fillstate data, and thus
# should simply use the first and last data points for the
# table.
assert server is not None
assert aggregate_table is RemoteInstallationCount or aggregate_table is RemoteRealmCount
aggregate_table_remote = cast(
Union[Type[RemoteInstallationCount],
Type[RemoteRealmCount]], aggregate_table
) # https://stackoverflow.com/questions/68540528/mypy-assertions-on-the-types-of-types
if not aggregate_table_remote.objects.filter(server=server).exists():
raise JsonableError(
_("No analytics data available. Please contact your server administrator."
))
if start is None:
first = aggregate_table_remote.objects.filter(
server=server).first()
assert first is not None
start = first.end_time
if end is None:
last = aggregate_table_remote.objects.filter(server=server).last()
assert last is not None
end = last.end_time
else:
# Otherwise, we can use tables on the current server to
# determine a nice range, and some additional validation.
if start is None:
if for_installation:
start = installation_epoch()
else:
start = realm.date_created
if end is None:
end = max(stat.last_successful_fill() or datetime.min.replace(
tzinfo=timezone.utc) for stat in stats)
if start > end and (timezone_now() - start >
MAX_TIME_FOR_FULL_ANALYTICS_GENERATION):
logging.warning(
"User from realm %s attempted to access /stats, but the computed "
"start time: %s (creation of realm or installation) is later than the computed "
"end time: %s (last successful analytics update). Is the "
"analytics cron job running?",
realm.string_id,
start,
end,
)
raise JsonableError(
_("No analytics data available. Please contact your server administrator."
))
assert len({stat.frequency for stat in stats}) == 1
end_times = time_range(start, end, stats[0].frequency, min_length)
data: Dict[str, Any] = {
"end_times": [int(end_time.timestamp()) for end_time in end_times],
"frequency": stats[0].frequency,
}
aggregation_level = {
InstallationCount: "everyone",
RealmCount: "everyone",
UserCount: "user",
}
if settings.ZILENCER_ENABLED:
aggregation_level[RemoteInstallationCount] = "everyone"
aggregation_level[RemoteRealmCount] = "everyone"
# -1 is a placeholder value, since there is no relevant filtering on InstallationCount
id_value = {
InstallationCount: -1,
RealmCount: realm.id,
UserCount: user_profile.id,
}
if settings.ZILENCER_ENABLED:
if server is not None:
id_value[RemoteInstallationCount] = server.id
# TODO: RemoteRealmCount logic doesn't correctly handle
# filtering by server_id as well.
if remote_realm_id is not None:
id_value[RemoteRealmCount] = remote_realm_id
for table in tables:
data[aggregation_level[table]] = {}
for stat in stats:
data[aggregation_level[table]].update(
get_time_series_by_subgroup(
stat,
table,
id_value[table],
end_times,
subgroup_to_label[stat],
include_empty_subgroups,
))
if labels_sort_function is not None:
data["display_order"] = labels_sort_function(data)
else:
data["display_order"] = None
return json_success(request, data=data)