Adds management command to clear all deleted user data

bookwyrm/management/commands/erase_deleted_user_data.py

@@ -0,0 +1,40 @@
+""" Erase any data stored about deleted users """
+import sys
+from django.core.management.base import BaseCommand, CommandError
+from bookwyrm import models
+from bookwyrm.models.user import erase_user_data
+
+# pylint: disable=missing-function-docstring
+class Command(BaseCommand):
+    """command-line options"""
+
+    help = "Remove Two Factor Authorisation from user"
+
+    def add_arguments(self, parser):  # pylint: disable=no-self-use
+        parser.add_argument(
+            "--dryrun",
+            action="store_true",
+            help="Preview users to be cleared without altering the database",
+        )
+
+    def handle(self, *args, **options):  # pylint: disable=unused-argument
+
+        # Check for anything fishy
+        bad_state = models.User.objects.filter(is_deleted=True, is_active=True)
+        if bad_state.exists():
+            raise CommandError(
+                f"{bad_state.count()} user(s) marked as both active and deleted"
+            )
+
+        deleted_users = models.User.objects.filter(is_deleted=True)
+        self.stdout.write(f"Found {deleted_users.count()} deleted users")
+        if options["dryrun"]:
+            self.stdout.write("\n".join(u.username for u in deleted_users[:5]))
+            if deleted_users.count() > 5:
+                self.stdout.write("... and more")
+            sys.exit()
+
+        self.stdout.write("Erasing user data:")
+        for user_id in deleted_users.values_list("id", flat=True):
+            erase_user_data.delay(user_id)
+            self.stdout.write(".", ending="")

bookwyrm/models/user.py

@@ -523,6 +523,20 @@ class KeyPair(ActivitypubMixin, BookWyrmModel):
         return super().save(*args, **kwargs)
 
 
+@app.task(queue=MISC)
+def erase_user_data(user_id):
+    """Erase any custom data about this user asynchronously
+    This is for deleted historical user data that pre-dates data
+    being cleared automatically"""
+    user = User.objects.get(id=user_id)
+    user.erase_user_data()
+    user.save(
+        broadcast=False,
+        update_fields=["email", "avatar", "preview_image", "summary", "name"],
+    )
+    user.erase_user_statuses(broadcast=False)
+
+
 @app.task(queue=MISC)
 def set_remote_server(user_id, allow_external_connections=False):
     """figure out the user's remote server in the background"""

bw-dev

@@ -246,6 +246,9 @@ case "$CMD" in
     remove_remote_user_preview_images)
         runweb python manage.py remove_remote_user_preview_images
         ;;
+    erase_deleted_user_data)
+        runweb python manage.py erase_deleted_user_data "$@"
+        ;;
     copy_media_to_s3)
         awscommand "bookwyrm_media_volume:/images"\
             "s3 cp /images s3://${AWS_STORAGE_BUCKET_NAME}/images\
@@ -297,7 +300,7 @@ case "$CMD" in
         echo "Unrecognised command. Try:"
         echo "    setup"
         echo "    up [container]"
-        echo "    down"        
+        echo "    down"
         echo "    service_ports_web"
         echo "    initdb"
         echo "    resetdb"