[ENG-11874] Create a periodic task to clean up recipients

api_tests/notifications/test_campaign_recipient_cleanup.py

@@ -0,0 +1,115 @@
+from datetime import timedelta
+
+import pytest
+from django.utils import timezone
+
+from website import settings
+from osf.models import (
+    NotificationCampaign,
+    NotificationCampaignRecipient,
+    NotificationTypeEnum,
+)
+from osf_tests.factories import AuthUserFactory
+from notifications.tasks import delete_notification_campaign_recipients
+
+
+@pytest.mark.django_db
+class TestDeleteNotificationCampaignRecipients:
+
+    @pytest.fixture(autouse=True)
+    def setup(self):
+        self.notification_type = NotificationTypeEnum.BLANK.instance
+
+    def test_deletes_recipients_for_old_campaigns(self):
+        now = timezone.now()
+        cutoff = now - settings.NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE
+
+        campaign = NotificationCampaign.objects.create(
+            name='Old campaign',
+            notification_type=self.notification_type,
+            completed_at=cutoff - timedelta(seconds=1),
+        )
+
+        recipients = NotificationCampaignRecipient.objects.bulk_create([
+            NotificationCampaignRecipient(
+                campaign=campaign,
+                user_id=AuthUserFactory().id,
+            ),
+            NotificationCampaignRecipient(
+                campaign=campaign,
+                user_id=AuthUserFactory().id,
+            ),
+        ])
+
+        delete_notification_campaign_recipients()
+
+        assert not NotificationCampaignRecipient.objects.filter(
+            id__in=[recipient.id for recipient in recipients],
+        ).exists()
+
+    def test_does_not_delete_recipients_for_recent_campaigns(self):
+        now = timezone.now()
+        cutoff = now - settings.NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE
+
+        campaign = NotificationCampaign.objects.create(
+            name='Recent campaign',
+            notification_type=self.notification_type,
+            completed_at=cutoff + timedelta(seconds=1),
+        )
+
+        recipient = NotificationCampaignRecipient.objects.create(
+            campaign=campaign,
+            user_id=AuthUserFactory().id,
+        )
+
+        delete_notification_campaign_recipients()
+
+        assert NotificationCampaignRecipient.objects.filter(
+            id=recipient.id,
+        ).exists()
+
+    def test_does_not_delete_recipients_for_incomplete_campaigns(self):
+        campaign = NotificationCampaign.objects.create(
+            name='Incomplete campaign',
+            notification_type=self.notification_type,
+            completed_at=None,
+        )
+
+        recipient = NotificationCampaignRecipient.objects.create(
+            campaign=campaign,
+            user_id=AuthUserFactory().id,
+        )
+
+        delete_notification_campaign_recipients()
+
+        assert NotificationCampaignRecipient.objects.filter(
+            id=recipient.id,
+        ).exists()
+
+    def test_deletes_recipients_in_batches(self):
+        now = timezone.now()
+        cutoff = now - settings.NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE
+
+        campaign = NotificationCampaign.objects.create(
+            name='Large campaign',
+            notification_type=self.notification_type,
+            completed_at=cutoff - timedelta(seconds=1),
+        )
+
+        NotificationCampaignRecipient.objects.bulk_create([
+            NotificationCampaignRecipient(
+                campaign=campaign,
+                user_id=AuthUserFactory().id,
+            )
+            for _ in range(10)
+        ])
+
+        assert NotificationCampaignRecipient.objects.filter(
+            campaign=campaign,
+        ).count() == 10
+
+        delete_notification_campaign_recipients()
+
+        assert not NotificationCampaignRecipient.objects.filter(
+            campaign=campaign,
+        ).exists()

notifications/tasks.py

@@ -11,7 +11,7 @@
 
 from framework.postcommit_tasks.handlers import run_postcommit
 from osf.models import OSFUser, Notification, NotificationTypeEnum, EmailTask, RegistrationProvider, \
-    CollectionProvider, AbstractProvider
+    CollectionProvider, AbstractProvider, NotificationCampaign, NotificationCampaignRecipient
 from framework.sentry import log_message
 from osf.registrations.utils import get_registration_provider_submissions_url
 from osf.utils.permissions import ADMIN
@@ -544,3 +544,37 @@ def delete_batch(
     logger.info(f'Deleted {deleted} rows from {model_name}')
 
     delete_batch.delay(app_label, model_name, filters, order_field, batch_size)
+
+
+@celery_app.task(
+    bind=True,
+    name='notifications.tasks.delete_notification_campaign_recipients',
+)
+def delete_notification_campaign_recipients(self):
+    """Delete recipients for old notification campaigns."""
+
+    cutoff = timezone.now() - settings.NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE
+
+    campaigns = NotificationCampaign.objects.filter(
+        completed_at__lt=cutoff,
+    ).values_list('id', flat=True)
+
+    for campaign_id in campaigns.iterator():
+        total_deleted = 0
+        while True:
+            recipient_ids = list(
+                NotificationCampaignRecipient.objects
+                .filter(campaign_id=campaign_id)
+                .values_list('id', flat=True)[:settings.NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_BATCH_SIZE]
+            )
+
+            if not recipient_ids:
+                break
+
+            deleted, _ = NotificationCampaignRecipient.objects.filter(
+                id__in=recipient_ids,
+            ).delete()
+
+            total_deleted += deleted
+
+        logger.info(f'Deleted {total_deleted} recipients for campaign {campaign_id}')

website/settings/defaults.py

@@ -190,6 +190,8 @@ def parent_dir(path):
 NO_LOGIN_OSF4M_WAIT_TIME = timedelta(weeks=52)  # 1 year for "We miss you at OSF" email to users created from OSF4M
 NOTIFICATIONS_CLEANUP_AGE = timedelta(weeks=12)  # 3 months to clean up old notifications and email tasks
 NOTIFICATIONS_CLEANUP_BATCH_SIZE = 10000  # Batch size for notifications and email tasks cleanup
+NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE = timedelta(weeks=12)  # 3 months to clean up old notification campaign recipients
+NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_BATCH_SIZE = 5000  # Batch size for notification campaign recipients cleanup
 
 # Notification campaign execution defaults (overridable per campaign in admin metadata)
 DEFAULT_CAMPAIGN_ACTIVITY_THRESHOLD = 3  # Users at/above this activity total are scheduled in the high-activity phase
@@ -712,6 +714,10 @@ class CeleryConfig:
             'schedule': crontab(minute=0, hour=7),  # Daily 2 a.m
             'kwargs': {'dry_run': False},
         },
+        'delete_notification_campaign_recipients': {
+            'task': 'notifications.tasks.delete_notification_campaign_recipients',
+            'schedule': crontab(minute=0, hour=3, day_of_month=1),
+        },
         'clear_expired_sessions': {
             'task': 'osf.management.commands.clear_expired_sessions',
             'schedule': crontab(minute=0, hour=5),  # Daily 12 a.m

website/settings/local-ci.py

@@ -67,6 +67,8 @@ class CeleryConfig(defaults.CeleryConfig):
 NO_ADDON_WAIT_TIME = timedelta(weeks=8)
 NO_LOGIN_WAIT_TIME = timedelta(weeks=4)
 NO_LOGIN_OSF4M_WAIT_TIME = timedelta(weeks=6)
+NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_AGE = timedelta(weeks=12)  # 3 months to clean up old notification campaign recipients
+NOTIFICATION_CAMPAIGN_RECIPIENTS_CLEANUP_BATCH_SIZE = 10  # Batch size for notification campaign recipients cleanup
 
 # Configuration for "We miss you at OSF" email (`NotificationTypeEnum.USER_NO_LOGIN`)
 MAX_DAILY_NO_LOGIN_EMAILS = None