mirror of https://github.com/zulip/zulip.git
analytics: Remove unused analytics management commands.
Unlike stream_stats, I'm not aware of any of these having been used in the last few years, and it's basically just really bad subsets of the data in /activity, which also doesn't require shell access to use. These haven't had real work or usage, AFAIK, since 2013.
This commit is contained in:
Tim Abbott
Tim Abbott
parent
49d06b9d69
commit
bcab06509a
|
|
@ -1,82 +0,0 @@
|
|||
import datetime
|
||||
import logging
|
||||
import time
|
||||
from typing import Any, Dict
|
||||
|
||||
from django.core.management.base import BaseCommand, CommandParser
|
||||
|
||||
from zerver.lib.timestamp import timestamp_to_datetime
|
||||
from zerver.models import Message, Recipient
|
||||
|
||||
|
||||
def compute_stats(log_level: int) -> None:
|
||||
logger = logging.getLogger()
|
||||
logger.setLevel(log_level)
|
||||
|
||||
one_week_ago = timestamp_to_datetime(time.time()) - datetime.timedelta(weeks=1)
|
||||
mit_query = Message.objects.filter(sender__realm__string_id="zephyr",
|
||||
recipient__type=Recipient.STREAM,
|
||||
date_sent__gt=one_week_ago)
|
||||
for bot_sender_start in ["imap.", "rcmd.", "sys."]:
|
||||
mit_query = mit_query.exclude(sender__email__startswith=(bot_sender_start))
|
||||
# Filtering for "/" covers tabbott/extra@ and all the daemon/foo bots.
|
||||
mit_query = mit_query.exclude(sender__email__contains=("/"))
|
||||
mit_query = mit_query.exclude(sender__email__contains=("aim.com"))
|
||||
mit_query = mit_query.exclude(
|
||||
sender__email__in=["rss@mit.edu", "bash@mit.edu", "apache@mit.edu",
|
||||
"bitcoin@mit.edu", "lp@mit.edu", "clocks@mit.edu",
|
||||
"root@mit.edu", "nagios@mit.edu",
|
||||
"www-data|local-realm@mit.edu"])
|
||||
user_counts: Dict[str, Dict[str, int]] = {}
|
||||
for m in mit_query.select_related("sending_client", "sender"):
|
||||
email = m.sender.email
|
||||
user_counts.setdefault(email, {})
|
||||
user_counts[email].setdefault(m.sending_client.name, 0)
|
||||
user_counts[email][m.sending_client.name] += 1
|
||||
|
||||
total_counts: Dict[str, int] = {}
|
||||
total_user_counts: Dict[str, int] = {}
|
||||
for email, counts in user_counts.items():
|
||||
total_user_counts.setdefault(email, 0)
|
||||
for client_name, count in counts.items():
|
||||
total_counts.setdefault(client_name, 0)
|
||||
total_counts[client_name] += count
|
||||
total_user_counts[email] += count
|
||||
|
||||
logging.debug("%40s | %10s | %s", "User", "Messages", "Percentage Zulip")
|
||||
top_percents: Dict[int, float] = {}
|
||||
for size in [10, 25, 50, 100, 200, len(total_user_counts.keys())]:
|
||||
top_percents[size] = 0.0
|
||||
for i, email in enumerate(sorted(total_user_counts.keys(),
|
||||
key=lambda x: -total_user_counts[x])):
|
||||
percent_zulip = round(100 - (user_counts[email].get("zephyr_mirror", 0)) * 100. /
|
||||
total_user_counts[email], 1)
|
||||
for size in top_percents.keys():
|
||||
top_percents.setdefault(size, 0)
|
||||
if i < size:
|
||||
top_percents[size] += (percent_zulip * 1.0 / size)
|
||||
|
||||
logging.debug("%40s | %10s | %s%%", email, total_user_counts[email],
|
||||
percent_zulip)
|
||||
|
||||
logging.info("")
|
||||
for size in sorted(top_percents.keys()):
|
||||
logging.info("Top %6s | %s%%", size, round(top_percents[size], 1))
|
||||
|
||||
grand_total = sum(total_counts.values())
|
||||
print(grand_total)
|
||||
logging.info("%15s | %s", "Client", "Percentage")
|
||||
for client in total_counts.keys():
|
||||
logging.info("%15s | %s%%", client, round(100. * total_counts[client] / grand_total, 1))
|
||||
|
||||
class Command(BaseCommand):
|
||||
help = "Compute statistics on MIT Zephyr usage."
|
||||
|
||||
def add_arguments(self, parser: CommandParser) -> None:
|
||||
parser.add_argument('--verbose', default=False, action='store_true')
|
||||
|
||||
def handle(self, *args: Any, **options: Any) -> None:
|
||||
level = logging.INFO
|
||||
if options["verbose"]:
|
||||
level = logging.DEBUG
|
||||
compute_stats(level)
|
||||
|
|
@ -1,56 +0,0 @@
|
|||
import datetime
|
||||
from typing import Any, Dict
|
||||
|
||||
from django.core.management.base import BaseCommand, CommandParser
|
||||
|
||||
from zerver.lib.statistics import seconds_usage_between
|
||||
from zerver.models import UserProfile
|
||||
|
||||
|
||||
def analyze_activity(options: Dict[str, Any]) -> None:
|
||||
day_start = datetime.datetime.strptime(options["date"], "%Y-%m-%d").replace(tzinfo=datetime.timezone.utc)
|
||||
day_end = day_start + datetime.timedelta(days=options["duration"])
|
||||
|
||||
user_profile_query = UserProfile.objects.all()
|
||||
if options["realm"]:
|
||||
user_profile_query = user_profile_query.filter(realm__string_id=options["realm"])
|
||||
|
||||
print("Per-user online duration:\n")
|
||||
total_duration = datetime.timedelta(0)
|
||||
for user_profile in user_profile_query:
|
||||
duration = seconds_usage_between(user_profile, day_start, day_end)
|
||||
|
||||
if duration == datetime.timedelta(0):
|
||||
continue
|
||||
|
||||
total_duration += duration
|
||||
print(f"{user_profile.email:<37}{duration}")
|
||||
|
||||
print(f"\nTotal Duration: {total_duration}")
|
||||
print(f"\nTotal Duration in minutes: {total_duration.total_seconds() / 60.}")
|
||||
print(f"Total Duration amortized to a month: {total_duration.total_seconds() * 30. / 60.}")
|
||||
|
||||
class Command(BaseCommand):
|
||||
help = """Report analytics of user activity on a per-user and realm basis.
|
||||
|
||||
This command aggregates user activity data that is collected by each user using Zulip. It attempts
|
||||
to approximate how much each user has been using Zulip per day, measured by recording each 15 minute
|
||||
period where some activity has occurred (mouse move or keyboard activity).
|
||||
|
||||
It will correctly not count server-initiated reloads in the activity statistics.
|
||||
|
||||
The duration flag can be used to control how many days to show usage duration for
|
||||
|
||||
Usage: ./manage.py analyze_user_activity [--realm=zulip] [--date=2013-09-10] [--duration=1]
|
||||
|
||||
By default, if no date is selected 2013-09-10 is used. If no realm is provided, information
|
||||
is shown for all realms"""
|
||||
|
||||
def add_arguments(self, parser: CommandParser) -> None:
|
||||
parser.add_argument('--realm', action='store')
|
||||
parser.add_argument('--date', action='store', default="2013-09-06")
|
||||
parser.add_argument('--duration', action='store', default=1, type=int,
|
||||
help="How many days to show usage information for")
|
||||
|
||||
def handle(self, *args: Any, **options: Any) -> None:
|
||||
analyze_activity(options)
|
||||
|
|
@ -1,74 +0,0 @@
|
|||
import datetime
|
||||
from argparse import ArgumentParser
|
||||
from typing import Any, Optional
|
||||
|
||||
from django.db.models import Count, QuerySet
|
||||
from django.utils.timezone import now as timezone_now
|
||||
|
||||
from zerver.lib.management import ZulipBaseCommand
|
||||
from zerver.models import UserActivity
|
||||
|
||||
|
||||
class Command(ZulipBaseCommand):
|
||||
help = """Report rough client activity globally, for a realm, or for a user
|
||||
|
||||
Usage examples:
|
||||
|
||||
./manage.py client_activity --target server
|
||||
./manage.py client_activity --target realm --realm zulip
|
||||
./manage.py client_activity --target user --user hamlet@zulip.com --realm zulip"""
|
||||
|
||||
def add_arguments(self, parser: ArgumentParser) -> None:
|
||||
parser.add_argument('--target', dest='target', required=True, type=str,
|
||||
help="'server' will calculate client activity of the entire server. "
|
||||
"'realm' will calculate client activity of realm. "
|
||||
"'user' will calculate client activity of the user.")
|
||||
parser.add_argument('--user', dest='user', type=str,
|
||||
help="The email address of the user you want to calculate activity.")
|
||||
self.add_realm_args(parser)
|
||||
|
||||
def compute_activity(self, user_activity_objects: QuerySet) -> None:
|
||||
# Report data from the past week.
|
||||
#
|
||||
# This is a rough report of client activity because we inconsistently
|
||||
# register activity from various clients; think of it as telling you
|
||||
# approximately how many people from a group have used a particular
|
||||
# client recently. For example, this might be useful to get a sense of
|
||||
# how popular different versions of a desktop client are.
|
||||
#
|
||||
# Importantly, this does NOT tell you anything about the relative
|
||||
# volumes of requests from clients.
|
||||
threshold = timezone_now() - datetime.timedelta(days=7)
|
||||
client_counts = user_activity_objects.filter(
|
||||
last_visit__gt=threshold).values("client__name").annotate(
|
||||
count=Count('client__name'))
|
||||
|
||||
total = 0
|
||||
counts = []
|
||||
for client_type in client_counts:
|
||||
count = client_type["count"]
|
||||
client = client_type["client__name"]
|
||||
total += count
|
||||
counts.append((count, client))
|
||||
|
||||
counts.sort()
|
||||
|
||||
for count in counts:
|
||||
print(f"{count[1]:>25} {count[0]:15}")
|
||||
print("Total:", total)
|
||||
|
||||
def handle(self, *args: Any, **options: Optional[str]) -> None:
|
||||
realm = self.get_realm(options)
|
||||
if options["user"] is None:
|
||||
if options["target"] == "server" and realm is None:
|
||||
# Report global activity.
|
||||
self.compute_activity(UserActivity.objects.all())
|
||||
elif options["target"] == "realm" and realm is not None:
|
||||
self.compute_activity(UserActivity.objects.filter(user_profile__realm=realm))
|
||||
else:
|
||||
self.print_help("./manage.py", "client_activity")
|
||||
elif options["target"] == "user":
|
||||
user_profile = self.get_user(options["user"], realm)
|
||||
self.compute_activity(UserActivity.objects.filter(user_profile=user_profile))
|
||||
else:
|
||||
self.print_help("./manage.py", "client_activity")
|
||||
|
|
@ -1,160 +0,0 @@
|
|||
import datetime
|
||||
from argparse import ArgumentParser
|
||||
from typing import Any, List
|
||||
|
||||
from django.core.management.base import BaseCommand, CommandError
|
||||
from django.db.models import Count
|
||||
from django.utils.timezone import now as timezone_now
|
||||
|
||||
from zerver.models import (
|
||||
Message,
|
||||
Realm,
|
||||
Recipient,
|
||||
Stream,
|
||||
Subscription,
|
||||
UserActivity,
|
||||
UserMessage,
|
||||
UserProfile,
|
||||
get_realm,
|
||||
)
|
||||
|
||||
MOBILE_CLIENT_LIST = ["Android", "ios"]
|
||||
HUMAN_CLIENT_LIST = MOBILE_CLIENT_LIST + ["website"]
|
||||
|
||||
human_messages = Message.objects.filter(sending_client__name__in=HUMAN_CLIENT_LIST)
|
||||
|
||||
class Command(BaseCommand):
|
||||
help = "Generate statistics on realm activity."
|
||||
|
||||
def add_arguments(self, parser: ArgumentParser) -> None:
|
||||
parser.add_argument('realms', metavar='<realm>', type=str, nargs='*',
|
||||
help="realm to generate statistics for")
|
||||
|
||||
def active_users(self, realm: Realm) -> List[UserProfile]:
|
||||
# Has been active (on the website, for now) in the last 7 days.
|
||||
activity_cutoff = timezone_now() - datetime.timedelta(days=7)
|
||||
return [activity.user_profile for activity in (
|
||||
UserActivity.objects.filter(user_profile__realm=realm,
|
||||
user_profile__is_active=True,
|
||||
last_visit__gt=activity_cutoff,
|
||||
query="/json/users/me/pointer",
|
||||
client__name="website"))]
|
||||
|
||||
def messages_sent_by(self, user: UserProfile, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return human_messages.filter(sender=user, date_sent__gt=sent_time_cutoff).count()
|
||||
|
||||
def total_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return Message.objects.filter(sender__realm=realm, date_sent__gt=sent_time_cutoff).count()
|
||||
|
||||
def human_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return human_messages.filter(sender__realm=realm, date_sent__gt=sent_time_cutoff).count()
|
||||
|
||||
def api_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
return (self.total_messages(realm, days_ago) - self.human_messages(realm, days_ago))
|
||||
|
||||
def stream_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return human_messages.filter(sender__realm=realm, date_sent__gt=sent_time_cutoff,
|
||||
recipient__type=Recipient.STREAM).count()
|
||||
|
||||
def private_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return human_messages.filter(sender__realm=realm, date_sent__gt=sent_time_cutoff).exclude(
|
||||
recipient__type=Recipient.STREAM).exclude(recipient__type=Recipient.HUDDLE).count()
|
||||
|
||||
def group_private_messages(self, realm: Realm, days_ago: int) -> int:
|
||||
sent_time_cutoff = timezone_now() - datetime.timedelta(days=days_ago)
|
||||
return human_messages.filter(sender__realm=realm, date_sent__gt=sent_time_cutoff).exclude(
|
||||
recipient__type=Recipient.STREAM).exclude(recipient__type=Recipient.PERSONAL).count()
|
||||
|
||||
def report_percentage(self, numerator: float, denominator: float, text: str) -> None:
|
||||
if not denominator:
|
||||
fraction = 0.0
|
||||
else:
|
||||
fraction = numerator / float(denominator)
|
||||
print(f"{fraction * 100:.2f}% of", text)
|
||||
|
||||
def handle(self, *args: Any, **options: Any) -> None:
|
||||
if options['realms']:
|
||||
try:
|
||||
realms = [get_realm(string_id) for string_id in options['realms']]
|
||||
except Realm.DoesNotExist as e:
|
||||
raise CommandError(e)
|
||||
else:
|
||||
realms = Realm.objects.all()
|
||||
|
||||
for realm in realms:
|
||||
print(realm.string_id)
|
||||
|
||||
user_profiles = UserProfile.objects.filter(realm=realm, is_active=True)
|
||||
active_users = self.active_users(realm)
|
||||
num_active = len(active_users)
|
||||
|
||||
print(f"{num_active} active users ({len(user_profiles)} total)")
|
||||
streams = Stream.objects.filter(realm=realm).extra(
|
||||
tables=['zerver_subscription', 'zerver_recipient'],
|
||||
where=['zerver_subscription.recipient_id = zerver_recipient.id',
|
||||
'zerver_recipient.type = 2',
|
||||
'zerver_recipient.type_id = zerver_stream.id',
|
||||
'zerver_subscription.active = true']).annotate(count=Count("name"))
|
||||
print(f"{streams.count()} streams")
|
||||
|
||||
for days_ago in (1, 7, 30):
|
||||
print(f"In last {days_ago} days, users sent:")
|
||||
sender_quantities = [self.messages_sent_by(user, days_ago) for user in user_profiles]
|
||||
for quantity in sorted(sender_quantities, reverse=True):
|
||||
print(quantity, end=' ')
|
||||
print("")
|
||||
|
||||
print(f"{self.stream_messages(realm, days_ago)} stream messages")
|
||||
print(f"{self.private_messages(realm, days_ago)} one-on-one private messages")
|
||||
print(f"{self.api_messages(realm, days_ago)} messages sent via the API")
|
||||
print(f"{self.group_private_messages(realm, days_ago)} group private messages")
|
||||
|
||||
num_notifications_enabled = len([x for x in active_users if x.enable_desktop_notifications])
|
||||
self.report_percentage(num_notifications_enabled, num_active,
|
||||
"active users have desktop notifications enabled")
|
||||
|
||||
num_enter_sends = len([x for x in active_users if x.enter_sends])
|
||||
self.report_percentage(num_enter_sends, num_active,
|
||||
"active users have enter-sends")
|
||||
|
||||
all_message_count = human_messages.filter(sender__realm=realm).count()
|
||||
multi_paragraph_message_count = human_messages.filter(
|
||||
sender__realm=realm, content__contains="\n\n").count()
|
||||
self.report_percentage(multi_paragraph_message_count, all_message_count,
|
||||
"all messages are multi-paragraph")
|
||||
|
||||
# Starred messages
|
||||
starrers = UserMessage.objects.filter(user_profile__in=user_profiles,
|
||||
flags=UserMessage.flags.starred).values(
|
||||
"user_profile").annotate(count=Count("user_profile"))
|
||||
print("{} users have starred {} messages".format(
|
||||
len(starrers), sum([elt["count"] for elt in starrers])))
|
||||
|
||||
active_user_subs = Subscription.objects.filter(
|
||||
user_profile__in=user_profiles, active=True)
|
||||
|
||||
# Streams not in home view
|
||||
non_home_view = active_user_subs.filter(is_muted=True).values(
|
||||
"user_profile").annotate(count=Count("user_profile"))
|
||||
print("{} users have {} streams not in home view".format(
|
||||
len(non_home_view), sum([elt["count"] for elt in non_home_view])))
|
||||
|
||||
# Code block markup
|
||||
markup_messages = human_messages.filter(
|
||||
sender__realm=realm, content__contains="~~~").values(
|
||||
"sender").annotate(count=Count("sender"))
|
||||
print("{} users have used code block markup on {} messages".format(
|
||||
len(markup_messages), sum([elt["count"] for elt in markup_messages])))
|
||||
|
||||
# Notifications for stream messages
|
||||
notifications = active_user_subs.filter(desktop_notifications=True).values(
|
||||
"user_profile").annotate(count=Count("user_profile"))
|
||||
print("{} users receive desktop notifications for {} streams".format(
|
||||
len(notifications), sum([elt["count"] for elt in notifications])))
|
||||
|
||||
print("")
|
||||
|
|
@ -1,42 +0,0 @@
|
|||
import datetime
|
||||
from argparse import ArgumentParser
|
||||
from typing import Any
|
||||
|
||||
from django.core.management.base import BaseCommand, CommandError
|
||||
from django.utils.timezone import now as timezone_now
|
||||
|
||||
from zerver.models import Message, Realm, Stream, UserProfile, get_realm
|
||||
|
||||
|
||||
class Command(BaseCommand):
|
||||
help = "Generate statistics on user activity."
|
||||
|
||||
def add_arguments(self, parser: ArgumentParser) -> None:
|
||||
parser.add_argument('realms', metavar='<realm>', type=str, nargs='*',
|
||||
help="realm to generate statistics for")
|
||||
|
||||
def messages_sent_by(self, user: UserProfile, week: int) -> int:
|
||||
start = timezone_now() - datetime.timedelta(days=(week + 1)*7)
|
||||
end = timezone_now() - datetime.timedelta(days=week*7)
|
||||
return Message.objects.filter(sender=user, date_sent__gt=start, date_sent__lte=end).count()
|
||||
|
||||
def handle(self, *args: Any, **options: Any) -> None:
|
||||
if options['realms']:
|
||||
try:
|
||||
realms = [get_realm(string_id) for string_id in options['realms']]
|
||||
except Realm.DoesNotExist as e:
|
||||
raise CommandError(e)
|
||||
else:
|
||||
realms = Realm.objects.all()
|
||||
|
||||
for realm in realms:
|
||||
print(realm.string_id)
|
||||
user_profiles = UserProfile.objects.filter(realm=realm, is_active=True)
|
||||
print(f"{len(user_profiles)} users")
|
||||
print(f"{len(Stream.objects.filter(realm=realm))} streams")
|
||||
|
||||
for user_profile in user_profiles:
|
||||
print(f"{user_profile.email:>35}", end=' ')
|
||||
for week in range(10):
|
||||
print(f"{self.messages_sent_by(user_profile, week):5}", end=' ')
|
||||
print("")
|
||||
Loading…
Reference in New Issue