feat: add command for removing old user data

Command is supposed to be run in maintenance runs e.g cron jobs for
cleaning the old user data.

Refs KER-398

democracy/management/commands/remove_user_data.py

@@ -0,0 +1,53 @@
+from django.conf import settings
+from django.core.management.base import BaseCommand
+from django.db import transaction
+from django.utils import timezone
+
+from democracy.utils import user_data_remover
+
+
+class Command(BaseCommand):
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--remove-user-data-from-old-objects",
+            action="store_true",
+            help="Remove user reference from old objects.",
+        )
+        parser.add_argument(
+            "--delete-comment-version-history",
+            action="store_true",
+            help="Delete old comments version history.",
+        )
+        parser.add_argument(
+            "--delete-users",
+            action="store_true",
+            help="Delete users without activity created before threshold.",
+        )
+        parser.add_argument(
+            "--older-than-days",
+            default=settings.DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS,
+            type=int,
+            help=f"Specify the number of days for removal; "
+            f"defaults to {settings.DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS}."
+            f"Data as old or older than this will be removed.",
+        )
+
+    @transaction.atomic
+    def handle(self, *args, **options):
+        threshold_time = timezone.now() - timezone.timedelta(
+            days=options["older_than_days"]
+        )
+
+        if options["remove_user_data_from_old_objects"]:
+            user_data_remover.remove_old_objects_user_data(threshold_time)
+            user_data_remover.remove_user_from_old_comments(threshold_time)
+            user_data_remover.remove_user_votes_from_old_comments(threshold_time)
+            user_data_remover.remove_user_from_old_poll_answers(threshold_time)
+            user_data_remover.remove_user_from_old_hearings(threshold_time)
+            user_data_remover.remove_contact_persons_from_old_hearings(threshold_time)
+
+            if options["delete_comment_version_history"]:
+                user_data_remover.delete_old_comments_versions(threshold_time)
+
+        if options["delete_users"]:
+            user_data_remover.delete_old_users_without_activity(threshold_time)

democracy/tests/integrationtest/test_remove_user_data_command.py

@@ -0,0 +1,216 @@
+import freezegun
+import pytest
+import reversion
+from django.conf import settings
+from django.core.management import call_command
+from django.utils import timezone
+from reversion.models import Version
+
+from democracy.factories.hearing import (
+    MinimalHearingFactory,
+    SectionCommentFactory,
+    SectionFactory,
+)
+from democracy.factories.poll import SectionPollFactory, SectionPollOptionFactory
+from democracy.models import ContactPerson, SectionComment, SectionPollAnswer
+from kerrokantasi.models import User
+from kerrokantasi.tests.factories import UserFactory
+
+
+def run_remove_user_data_command(*args):
+    call_command("remove_user_data", *args)
+
+
+@pytest.mark.django_db
+class TestRemoveUserDataCommand:
+    @pytest.fixture(autouse=True)
+    def init_test_data(self):
+        with freezegun.freeze_time(
+            timezone.now()
+            - timezone.timedelta(
+                days=settings.DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS + 1
+            )
+        ):
+            sec = SectionFactory(post=True)
+            self.old_user = UserFactory(username="old_user", date_joined=timezone.now())
+            self.old_user_without_activity = UserFactory(
+                username="tobe_deleted_user", date_joined=timezone.now()
+            )
+            with reversion.create_revision():
+                self.old_section_comment = SectionCommentFactory(
+                    created_by=self.old_user, section=sec, post=True
+                )
+                self.old_section_comment.title = "Old Title"
+                self.old_section_comment.save()
+            self.old_section_comment.voters.add(self.old_user)
+            self.old_section_comment.recache_n_votes()
+            self.old_hearing = MinimalHearingFactory(created_by=self.old_user)
+            SectionComment.objects.filter(section__hearing=self.old_hearing).delete()
+            self.old_contact_person = ContactPerson.objects.create(
+                name="Old Contact Person", created_by=self.old_user
+            )
+            self.old_hearing.contact_persons.add(self.old_contact_person)
+            poll = SectionPollFactory(section=sec)
+            option = SectionPollOptionFactory(poll=poll)
+            self.old_poll_answer = SectionPollAnswer.objects.create(
+                created_by=self.old_user,
+                option=option,
+                comment=self.old_section_comment,
+            )
+
+        self.new_user = UserFactory(username="newer_user", date_joined=timezone.now())
+        self.new_section_comment = SectionCommentFactory(
+            created_by=self.new_user, section=sec, post=True
+        )
+        self.new_hearing = MinimalHearingFactory(
+            created_by=self.new_user, close_at=timezone.now()
+        )
+        self.new_contact_person = ContactPerson.objects.create(
+            name="New Contact Person", created_by=self.new_user
+        )
+        self.new_hearing.contact_persons.add(self.new_contact_person)
+        self.new_poll_answer = SectionPollAnswer.objects.create(
+            created_by=self.new_user, option=option, comment=self.new_section_comment
+        )
+
+    old_objects = [
+        "old_section_comment",
+        "old_poll_answer",
+        "old_hearing",
+    ]
+    new_objects = [
+        "new_section_comment",
+        "new_hearing",
+        "new_contact_person",
+        "new_poll_answer",
+    ]
+
+    def test_delete_user(self):
+        self.old_user.delete()
+        self.old_section_comment.refresh_from_db()
+        self.old_hearing.refresh_from_db()
+        self.old_poll_answer.refresh_from_db()
+        self.old_contact_person.refresh_from_db()
+        assert self.old_section_comment.created_by is None
+        assert self.old_hearing.created_by is None
+        assert self.old_poll_answer.created_by is None
+        assert self.old_contact_person.created_by is None
+        assert self.old_section_comment.content is not None
+        assert self.old_section_comment.content != ""
+        assert self.old_section_comment.id > 0
+
+    def assert_old_objects_created_by_matches(self, exclude=()):
+        for model in [model for model in self.old_objects if model not in exclude]:
+            obj = getattr(self, model)
+            obj.refresh_from_db()
+            assert obj.created_by == self.old_user
+
+    def assert_old_objects_created_by_none(self, exclude=()):
+        for model in [model for model in self.old_objects if model not in exclude]:
+            obj = getattr(self, model)
+            obj.refresh_from_db()
+            assert obj.created_by is None
+
+    def assert_new_objects_created_by_matches(self):
+        for model in self.new_objects:
+            obj = getattr(self, model)
+            obj.refresh_from_db()
+            assert obj.created_by == self.new_user
+
+    def test_all_options(self):
+        """Test remove_user_data command with all options."""
+        args = [
+            "--remove-user-data-from-old-objects",
+            "--delete-comment-version-history",
+            "--delete-users",
+            "--older-than-days",
+            str(settings.DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS),
+        ]
+        run_remove_user_data_command(*args)
+
+        self.old_section_comment.refresh_from_db()
+        self.old_poll_answer.refresh_from_db()
+        self.old_hearing.refresh_from_db()
+        old_user = User.objects.filter(id=self.old_user.id).first()
+        user_without_activity = User.objects.filter(
+            username=self.old_user_without_activity.username
+        ).first()
+        self.new_user.refresh_from_db()
+
+        self.assert_old_objects_created_by_none()
+        assert self.old_hearing.contact_persons.count() == 0
+        assert self.old_section_comment.n_unregistered_votes == 1
+        assert self.old_section_comment.voters.count() == 0
+        assert self.old_section_comment.n_votes == 1
+        self.assert_new_objects_created_by_matches()
+        assert Version.objects.get_for_object(self.old_section_comment).count() == 0
+
+        assert user_without_activity is None
+        assert old_user is None
+        assert self.new_user is not None
+
+    def test_remove_only_user_data_from_old_objects(self):
+        """Test remove_user_data command with remove_user_data_from_old_objects option."""
+        args = ["--remove-user-data-from-old-objects"]
+        run_remove_user_data_command(*args)
+
+        self.old_section_comment.refresh_from_db()
+        self.old_poll_answer.refresh_from_db()
+        self.old_hearing.refresh_from_db()
+
+        self.assert_old_objects_created_by_none()
+        assert self.old_section_comment.n_unregistered_votes == 1
+        assert self.old_section_comment.voters.count() == 0
+        assert self.old_section_comment.n_votes == 1
+        assert self.old_hearing.contact_persons.count() == 0
+
+        self.assert_new_objects_created_by_matches()
+
+    def test_remove_user_data_from_old_objects_with_delete_version_option(self):
+        """Test remove_user_data command with remove_user_data_from_old_objects option."""
+        args = [
+            "--remove-user-data-from-old-objects",
+            "--delete-comment-version-history",
+        ]
+
+        assert Version.objects.get_for_object(self.old_section_comment).count() > 0
+        run_remove_user_data_command(*args)
+
+        self.old_section_comment.refresh_from_db()
+        self.old_poll_answer.refresh_from_db()
+        self.old_hearing.refresh_from_db()
+
+        self.assert_old_objects_created_by_none()
+        assert self.old_section_comment.n_unregistered_votes == 1
+        assert self.old_section_comment.voters.count() == 0
+        assert self.old_section_comment.n_votes == 1
+        assert self.old_hearing.contact_persons.count() == 0
+
+        self.assert_new_objects_created_by_matches()
+
+        assert Version.objects.get_for_object(self.old_section_comment).count() == 0
+
+    def test_only_delete_inactive_users(self):
+        """Test remove_user_data command with delete_users option."""
+        args = ["--delete-users"]
+        run_remove_user_data_command(*args)
+
+        self.old_section_comment.refresh_from_db()
+        self.old_poll_answer.refresh_from_db()
+        self.old_hearing.refresh_from_db()
+
+        self.assert_old_objects_created_by_matches()
+        assert self.old_section_comment.n_unregistered_votes == 0
+        assert self.old_section_comment.voters.count() == 1
+        assert self.old_section_comment.n_votes == 1
+        assert self.old_hearing.contact_persons.count() == 1
+
+        self.assert_new_objects_created_by_matches()
+        old_user = User.objects.filter(id=self.old_user.id).first()
+        user_without_activity = User.objects.filter(
+            username=self.old_user_without_activity.username
+        ).first()
+        self.new_user.refresh_from_db()
+        assert user_without_activity is None
+        assert old_user is not None
+        assert self.new_user is not None

kerrokantasi/settings/base.py

@@ -92,6 +92,7 @@ def get_git_revision_hash():
     GDPR_API_DELETE_SCOPE=(str, "gdprdelete"),
     # Audit logging
     AUDIT_LOG_ENABLED=(bool, False),
+    DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS=(int, 365 * 5),  # Five years.
 )
 
 # Build paths inside the project like this: os.path.join(BASE_DIR, ...)
@@ -433,3 +434,7 @@ def get_git_revision_hash():
     "ENABLED": env("AUDIT_LOG_ENABLED"),
     "ORIGIN": "kerrokantasi",
 }
+
+DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS = env(
+    "DEFAULT_USER_DATA_REMOVAL_THRESHOLD_DAYS"
+)