mirror of
https://github.com/zulip/zulip.git
synced 2025-11-07 15:33:30 +00:00
72 lines
2.3 KiB
Python
72 lines
2.3 KiB
Python
"""
|
|
Implements the per-domain data retention policy.
|
|
|
|
The goal is to have a single place where the policy is defined. This is
|
|
complicated by needing to apply this policy both to the database and to log
|
|
files. Additionally, we want to use an efficient query for the database,
|
|
rather than iterating through messages one by one.
|
|
|
|
The code in this module does not actually remove anything; it just identifies
|
|
which items should be kept or removed.
|
|
"""
|
|
|
|
import sys
|
|
import operator
|
|
|
|
from django.utils import timezone
|
|
from django.db.models import Q
|
|
from datetime import datetime, timedelta
|
|
from zephyr.models import Realm, UserMessage
|
|
|
|
# Each domain has a maximum age for retained messages.
|
|
#
|
|
# FIXME: Move this into the database.
|
|
max_age = {
|
|
'customer1.invalid': timedelta(days=31),
|
|
}
|
|
|
|
def should_expunge_from_log(msg, now):
|
|
"""Should a particular log entry be expunged?
|
|
|
|
msg: a log entry dict
|
|
now: current time for purposes of determining log entry age"""
|
|
|
|
# This function will be called many times, but we want to compare all
|
|
# entries against a consistent "current time". So the caller passes
|
|
# that time as a parameter.
|
|
|
|
if msg.get('type') not in ('stream', 'huddle', 'personal'):
|
|
# Keep all metadata changes like realm_created, subscription_added,
|
|
# etc.
|
|
return False
|
|
|
|
# FIXME: Yet another place where we compute the domain manually.
|
|
# See #260.
|
|
domain = msg['sender_email'].split('@', 1)[1]
|
|
|
|
if domain not in max_age:
|
|
# Keep forever.
|
|
return False
|
|
|
|
age = now - datetime.fromtimestamp(msg['timestamp'])
|
|
return age > max_age[domain]
|
|
|
|
def get_UserMessages_to_expunge():
|
|
"""Fetch all UserMessages which should be expunged from the database.
|
|
|
|
After deleting these, you may also want to call
|
|
Message.remove_unreachable()."""
|
|
# Unlike retain_in_log, this handles all messages at once, so we
|
|
# use the actual current time.
|
|
now = timezone.now()
|
|
queries = [Q(user_profile__realm = realm,
|
|
message__pub_date__lt = now - max_age[realm.domain])
|
|
for realm in Realm.objects.all()
|
|
if realm.domain in max_age]
|
|
|
|
if not queries:
|
|
return UserMessage.objects.none()
|
|
|
|
# Return all objects matching any of the queries in 'queries'.
|
|
return UserMessage.objects.filter(reduce(operator.or_, queries))
|