Improvements to GDPR/CCPA deletion handling

app/repositories/patcher_token_logs.py

@@ -0,0 +1,70 @@
+from datetime import datetime
+from typing import Any
+
+from pydantic import BaseModel
+
+import app.state
+
+
+class PatcherTokenLog(BaseModel):
+    id: str
+    score_id: int
+    client_hash: str
+    commentary: str | None
+    token_generated_at: datetime
+    created_at: datetime
+    updated_at: datetime
+
+
+READ_PARAMS = """\
+    id, score_id, client_hash, commentary, token_generated_at, created_at, updated_at
+"""
+
+
+async def delete_many_by_user_id_via_scores_tables(
+    user_id: int,
+    /,
+) -> list[PatcherTokenLog]:
+    query = f"""\
+        WITH score_ids AS (
+            SELECT id FROM scores WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_relax WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_ap WHERE user_id = :user_id
+        )
+        SELECT {READ_PARAMS}
+        FROM patcher_token_logs
+        WHERE score_id IN score_ids
+    """
+    params: dict[str, Any] = {"user_id": user_id}
+    recs = await app.state.database.fetch_all(query, params)
+    if not recs:
+        return []
+
+    query = """\
+        WITH score_ids AS (
+            SELECT id FROM scores WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_relax WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_ap WHERE user_id = :user_id
+        )
+        DELETE FROM patcher_token_logs
+        WHERE score_id IN score_ids
+    """
+    params = {"user_id": user_id}
+    await app.state.database.execute(query, params)
+
+    return [
+        PatcherTokenLog(
+            id=rec["id"],
+            score_id=rec["score_id"],
+            client_hash=rec["client_hash"],
+            commentary=rec["commentary"],
+            token_generated_at=rec["token_generated_at"],
+            created_at=rec["created_at"],
+            updated_at=rec["updated_at"],
+        )
+        for rec in recs
+    ]

app/repositories/score_submission_logs.py

@@ -0,0 +1,76 @@
+from datetime import datetime
+from datetime import time
+from typing import Any
+
+from pydantic import BaseModel
+
+import app.state
+
+
+class ScoreSubmissionLog(BaseModel):
+    id: str
+    score_id: int
+    uninstall_id_hash: str
+    disk_signature_hash: str
+    client_version: str
+    client_hash: str
+    score_time_elapsed: time
+    osu_auth_token: str | None
+    created_at: datetime
+
+
+READ_PARAMS = """\
+    id, score_id, uninstall_id_hash, disk_signature_hash, client_version,
+    client_hash, score_time_elapsed, osu_auth_token, created_at
+"""
+
+
+async def delete_many_by_user_id_via_scores_tables(
+    user_id: int,
+    /,
+) -> list[ScoreSubmissionLog]:
+    query = f"""\
+        WITH score_ids AS (
+            SELECT id FROM scores WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_relax WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_ap WHERE user_id = :user_id
+        )
+        SELECT {READ_PARAMS}
+        FROM score_submission_logs
+        WHERE score_id IN score_ids
+    """
+    params: dict[str, Any] = {"user_id": user_id}
+    recs = await app.state.database.fetch_all(query, params)
+    if not recs:
+        return []
+
+    query = """\
+        WITH score_ids AS (
+            SELECT id FROM scores WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_relax WHERE user_id = :user_id
+            UNION
+            SELECT id FROM scores_ap WHERE user_id = :user_id
+        )
+        DELETE FROM score_submission_logs
+        WHERE score_id IN score_ids
+    """
+    params = {"user_id": user_id}
+    await app.state.database.execute(query, params)
+
+    return [
+        ScoreSubmissionLog(
+            id=rec["id"],
+            score_id=rec["score_id"],
+            uninstall_id_hash=rec["uninstall_id_hash"],
+            disk_signature_hash=rec["disk_signature_hash"],
+            client_version=rec["client_version"],
+            client_hash=rec["client_hash"],
+            score_time_elapsed=rec["score_time_elapsed"],
+            osu_auth_token=rec["osu_auth_token"],
+            created_at=rec["created_at"],
+        )
+        for rec in recs
+    ]

app/usecases/users.py

@@ -12,6 +12,8 @@
 from app.repositories import clans
 from app.repositories import lastfm_flags
 from app.repositories import password_recovery
+from app.repositories import patcher_token_logs
+from app.repositories import score_submission_logs
 from app.repositories import user_badges
 from app.repositories import user_hwid_associations
 from app.repositories import user_ip_associations
@@ -212,26 +214,26 @@ async def delete_one_by_user_id(user_id: int, /) -> None | Error:
     # - [leave as-is] users_stats
     # - [leave as-is] rx_stats
     # - [leave as-is] ap_stats
-    # - [TODO/AC] ip_user
-    # - [TODO/AC] hw_user
+    # - [delete for now; TODO anonymize] ip_user
+    # - [delete for now; TODO anonymize] hw_user
     # - [leave as-is] user_badges
     # - [leave as-is] user_tourmnt_badges
     # - [leave as-is] user_achievements
     # - [transfer perms if owner & kick] clans
     # - [leave as-is] identity_tokens
     # - [leave as-is] irc_tokens
-    # - [TODO/AC] lastfm_flags
+    # - [delete] lastfm_flags
     # - [leave as-is] beatmaps_rating
     # - [leave as-is] clan_requests (empty?)
     # - [leave as-is] comments
     # - [leave as-is] matches
     # - [leave as-is] match_events
     # - [leave as-is] match_games
     # - [leave as-is] match_game_scores
-    # - [TODO/financial] notifications
+    # - [leave-as-is (FINANCE)] notifications
     # - [delete; key'd by username??] password_recovery
-    # - [TODO/AC] patcher_detections
-    # - [TODO/AC] patcher_token_logs
+    # - [delete] patcher_detections
+    # - [delete] patcher_token_logs
     # - [TODO] profile_backgrounds (and filesystem data)
     # - [TODO] rap_logs
     # - [leave as-is] remember
@@ -244,7 +246,7 @@ async def delete_one_by_user_id(user_id: int, /) -> None | Error:
     # - [leave as-is] scores_ap
     # - [leave as-is] scores_relax
     # - [leave as-is] scores_first
-    # - [TODO/AC] score_submission_logs
+    # - [delete] score_submission_logs
     # - [leave as-is] tokens
     # - [leave as-is] user_relationships
     # - [leave as-is] user_beatmaps
@@ -256,6 +258,8 @@ async def delete_one_by_user_id(user_id: int, /) -> None | Error:
     # misc.
     # - [anonymize] replay data for all scores
     # - [TODO] youtube uploads
+    # - [TODO] static content (screenshots, profile bgs, etc.)
+    # - [TODO] database/fs backups older than 50 days
 
     # PII to focus on:
     # - username / username aka
@@ -302,7 +306,10 @@ async def delete_one_by_user_id(user_id: int, /) -> None | Error:
         await user_ip_associations.delete_many_by_user_id(user_id)
         await user_hwid_associations.delete_many_by_user_id(user_id)
         await lastfm_flags.delete_many_by_user_id(user_id)
-        # TODO: patcher_detections & patcher_token_logs
+
+        await score_submission_logs.delete_many_by_user_id_via_scores_tables(user_id)
+        await patcher_token_logs.delete_many_by_user_id_via_scores_tables(user_id)
+        # TODO: delete user records from `patcher_detections` table as well
 
         # TODO: wipe or anonymize all replay data.
         #       probably a good idea to call scores-service
@@ -315,7 +322,7 @@ async def delete_one_by_user_id(user_id: int, /) -> None | Error:
         #       at the usecase layer
         await users.anonymize_one_by_user_id(user_id)
 
-        # TODO: (technically required) anonymize data in data backups
+        # TODO: don't store backups older than 50 days via sql-backup-job
 
         # inform other systems of the user's deletion (or "ban")
         await app.state.redis.publish("peppy:ban", str(user_id))