Add bulk mail scrubbing without user log entries

Author: lukasjuhrich

pycroft/lib/user_deletion.py

@@ -13,13 +13,14 @@
 from sqlalchemy import CTE, ScalarResult, func, and_, not_
 from sqlalchemy.future import select
 from sqlalchemy.orm import joinedload, Session
-from sqlalchemy.sql import Select
+from sqlalchemy.sql import Select, insert, literal_column, update
 
 from pycroft.helpers.i18n.deferred import deferred_gettext
 from pycroft.lib.logging import log_user_event
 from pycroft.model.host import Host
 from pycroft.model.property import CurrentProperty
 from pycroft.model.scrubbing import ScrubLog
+from pycroft.model.session import current_timestamp, utcnow
 from pycroft.model.user import RoomHistoryEntry, User
 from pycroft.lib.membership import select_user_and_last_mem
 
@@ -165,8 +166,22 @@ def scrub_mail(session: Session, user: User, author: User):
 
 
 def scrub_all_mails(session: Session, author: User):
-    stmt, _ = scrubbable_mails(session)
-    # TODO implement
+    year = utcnow().year
+    users_with_mail = scrubbable_mails_stmt(year)
+    ids_subq = users_with_mail.with_only_columns(User.id).scalar_subquery()
+    # TODO report affected cols in return value
+    _ = session.execute(update(User).where(User.id.in_(ids_subq)).values(email=None))
+    # TODO insert user_log_entry
+    _ = session.execute(
+        insert(ScrubLog).from_select(
+            ("executed_at", "scrubber", "info"),
+            users_with_mail.with_only_columns(
+                current_timestamp().label("executed_at"),
+                literal_column("'mail'").label("scrubber"),
+                func.jsonb_build_object("user_id", User.id).label("info"),
+            ),
+        )
+    )
 
 
 def scrubbable_hosts_stmt(year: int) -> Select[tuple[Host]]:

tests/lib/scrubbing/conftest.py

@@ -32,23 +32,43 @@ def user_archivable(module_session: Session, config: Config) -> User:
     )
 
 
+@pytest.fixture(scope="module")
+def users_archivable(module_session: Session, config: Config) -> list[User]:
+    # TODO: user wtih membership
+    return f.UserFactory.create_batch(
+        10,
+        registered_at=datetime(2020, 7, 1),
+        with_membership=True,
+        membership__active_during=closedopen(datetime(2020, 7, 1), datetime(2021, 11, 25)),
+        membership__group=config.member_group,
+        without_room=True,
+    )
+
+
 @pytest.fixture(scope="module", autouse=True)
-def user_do_not_archive(
+def users_do_not_archive(
     module_session: Session, config: Config, group_do_not_archive: PropertyGroup
-) -> User:
+) -> list[User]:
     """
     Create a user with a membership in a group that has the do-not-archive
     property.
     """
-    # TODO _very old_ membership, but also mem in a group
-    user = f.UserFactory.create(
-        registered_at=datetime(2020, 7, 1),
+    # TODO wayyy better example set! instead of making all `do-not-archive`,
+    # we would want them to have different membership ranges, also additional memberships;
+    # also edge-cases like upcoming memberships or on-new-years membership terminations
+    beginning = datetime(2020, 7, 1)
+    users = f.UserFactory.create_batch(
+        10,
+        registered_at=beginning,
         with_membership=True,
-        membership__active_during=closedopen(datetime(2020, 7, 1), datetime(2021, 11, 25)),
+        membership__active_during=closedopen(beginning, datetime(2021, 11, 25)),
         membership__group=config.member_group,
         without_room=True,
     )
-    f.MembershipFactory.create(
-        user=user, group=group_do_not_archive, active_during=closedopen(datetime(2020, 7, 1), None)
-    )
-    return user
+    for user in users:
+        f.MembershipFactory.create(
+            user=user,
+            group=group_do_not_archive,
+            active_during=closedopen(beginning, None),
+        )
+    return users

tests/lib/scrubbing/test_mail_scrubbing.py

@@ -1,3 +1,4 @@
+from itertools import chain
 import pytest
 from sqlalchemy import select
 from sqlalchemy.orm import Session
@@ -35,6 +36,17 @@ def test_scrubbing_scrubs_mail(user_archivable: User, processor: User, session:
 
 
 def test_bulk_scrubbing_scrubs_mail(
-    users_archivable: list[User], users_notarchivable: list[User], processor: User, session: Session
+    users_archivable: list[User],
+    users_do_not_archive: list[User],
+    processor: User,
+    session: Session,
 ):
     scrub_all_mails(session, author=processor)
+    for u in chain(users_archivable, users_do_not_archive):
+        session.refresh(u)
+    assert all(
+        u.email is None for u in users_archivable
+    ), "mails should have been None for prepared users"
+    assert all(
+        u.email is not None for u in users_do_not_archive
+    ), "mails should not have been scrubbed for do-not-archive users"