feat: add production data retention

apps/api/app/cli.py

@@ -3,10 +3,14 @@ from __future__ import annotations
 import argparse
 import json
 import sys
+from dataclasses import asdict
 
+from .config import settings
 from .database import SessionLocal
 from .importer import import_records
 from .community_importer import stage_observations
+from .retention import RetentionPolicy, apply_retention
+from .storage import delete_screenshot
 
 
 def main() -> int:
@@ -19,12 +23,14 @@ def main() -> int:
     community = sub.add_parser("stage-community-json")
     community.add_argument("--input", default="-", help="JSON array path or - for stdin")
     community.add_argument("--limit", type=int, default=500)
+    cleanup = sub.add_parser("cleanup-retention")
+    cleanup.add_argument("--apply", action="store_true", help="apply changes; default is dry-run")
     args = parser.parse_args()
     with SessionLocal() as session:
         if args.command == "import-records":
             run = import_records(session, url=args.url, region=args.region, category=args.category)
             print(f"import {run.status.value}: seen={run.rows_seen} created={run.rows_created} updated={run.rows_updated}")
-        else:
+        elif args.command == "stage-community-json":
             if not 1 <= args.limit <= 5000:
                 parser.error("--limit must be between 1 and 5000")
             stream = sys.stdin if args.input == "-" else open(args.input, encoding="utf-8")
@@ -37,6 +43,17 @@ def main() -> int:
                 parser.error("input must be a JSON array")
             created, updated = stage_observations(session, payload[:args.limit])
             print(f"staged: created={created} updated={updated}")
+        else:
+            policy = RetentionPolicy(
+                submission_days=settings.retention_submission_days,
+                unreviewed_days=settings.retention_unreviewed_days,
+                approved_personal_days=settings.retention_approved_personal_days,
+                staging_days=settings.retention_staging_days,
+                audit_days=settings.retention_audit_days,
+                published_payload_days=settings.retention_published_payload_days,
+            )
+            counts = apply_retention(session, policy=policy, dry_run=not args.apply, delete_object=delete_screenshot)
+            print(json.dumps({"mode": "apply" if args.apply else "dry-run", "policy": asdict(policy), "counts": counts}, ensure_ascii=False))
     return 0