zulip/analytics/management/commands/populate_analytics_db.py

136 lines
7.1 KiB
Python

from __future__ import absolute_import, print_function
from argparse import ArgumentParser
from django.core.management.base import BaseCommand
from django.utils import timezone
from analytics.lib.counts import COUNT_STATS, CountStat, do_drop_all_analytics_tables
from analytics.lib.fixtures import generate_time_series_data
from analytics.lib.time_utils import time_range
from analytics.models import BaseCount, InstallationCount, RealmCount, \
UserCount, StreamCount, FillState
from zerver.lib.timestamp import floor_to_day
from zerver.models import Realm, UserProfile, Stream, Message, Client
from datetime import datetime, timedelta
from six.moves import zip
from typing import Any, Dict, List, Optional, Text, Type, Union
class Command(BaseCommand):
help = """Populates analytics tables with randomly generated data."""
DAYS_OF_DATA = 100
random_seed = 26
def create_user(self, email, full_name, is_staff, date_joined, realm):
# type: (Text, Text, Text, bool, datetime, Realm) -> UserProfile
return UserProfile.objects.create(
email=email, full_name=full_name, is_staff=is_staff,
realm=realm, short_name=full_name, pointer=-1, last_pointer_updater='none',
api_key='42', date_joined=date_joined)
def generate_fixture_data(self, stat, business_hours_base, non_business_hours_base,
growth, autocorrelation, spikiness, holiday_rate=0):
# type: (CountStat, float, float, float, float, float, float) -> List[int]
self.random_seed += 1
return generate_time_series_data(
days=self.DAYS_OF_DATA, business_hours_base=business_hours_base,
non_business_hours_base=non_business_hours_base, growth=growth,
autocorrelation=autocorrelation, spikiness=spikiness, holiday_rate=holiday_rate,
frequency=stat.frequency, is_gauge=(stat.interval == CountStat.GAUGE),
random_seed=self.random_seed)
def handle(self, *args, **options):
# type: (*Any, **Any) -> None
do_drop_all_analytics_tables()
# I believe this also deletes any objects with this realm as a foreign key
Realm.objects.filter(string_id='analytics').delete()
Client.objects.filter(name__startswith='_').delete()
installation_time = timezone.now() - timedelta(days=self.DAYS_OF_DATA)
last_end_time = floor_to_day(timezone.now())
realm = Realm.objects.create(
string_id='analytics', name='Analytics', domain='analytics.ds',
date_created=installation_time)
shylock = self.create_user('shylock@analytics.ds', 'Shylock', True, installation_time, realm)
def insert_fixture_data(stat, fixture_data, table):
# type: (CountStat, Dict[Optional[str], List[int]], Type[BaseCount]) -> None
end_times = time_range(last_end_time, last_end_time, stat.frequency,
len(list(fixture_data.values())[0]))
if table == RealmCount:
id_args = {'realm': realm}
if table == UserCount:
id_args = {'realm': realm, 'user': shylock}
for subgroup, values in fixture_data.items():
table.objects.bulk_create([
table(property=stat.property, subgroup=subgroup, end_time=end_time,
value=value, **id_args)
for end_time, value in zip(end_times, values) if value != 0])
stat = COUNT_STATS['active_users:is_bot:day']
realm_data = {
'false': self.generate_fixture_data(stat, .1, .03, 3, .5, 3),
'true': self.generate_fixture_data(stat, .01, 0, 1, 0, 1)
} # type: Dict[Optional[str], List[int]]
insert_fixture_data(stat, realm_data, RealmCount)
FillState.objects.create(property=stat.property, end_time=last_end_time,
state=FillState.DONE)
stat = COUNT_STATS['messages_sent:is_bot:hour']
user_data = {'false': self.generate_fixture_data(stat, 2, 1, 1.5, .6, 8, holiday_rate=.1)}
insert_fixture_data(stat, user_data, UserCount)
realm_data = {'false': self.generate_fixture_data(stat, 35, 15, 6, .6, 4),
'true': self.generate_fixture_data(stat, 15, 15, 3, .4, 2)}
insert_fixture_data(stat, realm_data, RealmCount)
FillState.objects.create(property=stat.property, end_time=last_end_time,
state=FillState.DONE)
stat = COUNT_STATS['messages_sent:message_type:day']
user_data = {
'public_stream': self.generate_fixture_data(stat, 1.5, 1, 3, .6, 8),
'private_message': self.generate_fixture_data(stat, .5, .3, 1, .6, 8)}
insert_fixture_data(stat, user_data, UserCount)
realm_data = {
'public_stream': self.generate_fixture_data(stat, 30, 8, 5, .6, 4),
'private_stream': self.generate_fixture_data(stat, 7, 7, 5, .6, 4),
'private_message': self.generate_fixture_data(stat, 13, 5, 5, .6, 4)}
insert_fixture_data(stat, realm_data, RealmCount)
FillState.objects.create(property=stat.property, end_time=last_end_time,
state=FillState.DONE)
# We use client names that start with _ to make it easy to
# drop just these data manually without affecting other data.
website = Client.objects.create(name='_Website')
old_desktop = Client.objects.create(name='_Old desktop app')
android = Client.objects.create(name='_Android app')
iOS = Client.objects.create(name='_Old iOS app')
react_native = Client.objects.create(name='_New iOS app')
API = Client.objects.create(name='_Python API')
barnowl = Client.objects.create(name='_Barnowl')
unused = Client.objects.create(name='_Unused webhook')
long_webhook = Client.objects.create(name='_Webhook with loooooooong name')
stat = COUNT_STATS['messages_sent:client:day']
user_data = {
website.id: self.generate_fixture_data(stat, 2, 1, 1.5, .6, 8),
barnowl.id: self.generate_fixture_data(stat, 0, .3, 1.5, .6, 8)}
insert_fixture_data(stat, user_data, UserCount)
realm_data = {
website.id: self.generate_fixture_data(stat, 30, 20, 5, .6, 3),
old_desktop.id: self.generate_fixture_data(stat, 5, 3, 8, .6, 3),
android.id: self.generate_fixture_data(stat, 5, 5, 2, .6, 3),
iOS.id: self.generate_fixture_data(stat, 5, 5, 2, .6, 3),
react_native.id: self.generate_fixture_data(stat, 5, 5, 10, .6, 3),
API.id: self.generate_fixture_data(stat, 5, 5, 5, .6, 3),
barnowl.id: self.generate_fixture_data(stat, 1, 1, 3, .6, 3),
unused.id: self.generate_fixture_data(stat, 0, 0, 0, 0, 0),
long_webhook.id: self.generate_fixture_data(stat, 5, 5, 2, .6, 3)}
insert_fixture_data(stat, realm_data, RealmCount)
FillState.objects.create(property=stat.property, end_time=last_end_time,
state=FillState.DONE)
# TODO: messages_sent_to_stream:is_bot