From a6753d092fd3705f6d36f7922969f6339f051f47 Mon Sep 17 00:00:00 2001
From: CC Worker <cc-worker@local>
Date: Mon, 8 Jun 2026 00:26:24 +0000
Subject: [PATCH] fix(reset): fold --user-subset cleanup into scope=all and
 scope=exam-corpus

t_d1600327 added a standalone scope=user-subset, but a full reset (scope=all)
and scope=exam-corpus still left the --user-subset cc.users storage objects
orphaned (files rows are wiped by the table clear, but the Storage API objects
are not). Call the same _clear_user_subset_files() helper in both paths so the
finding-#2 gap is fully closed: storage removed before rows, idempotent.

Closes overwatch review finding #2 (user-subset not cleaned by reset).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 run/initialization/reset_environment.py | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/run/initialization/reset_environment.py b/run/initialization/reset_environment.py
index 70d0c84..b050c22 100644
--- a/run/initialization/reset_environment.py
+++ b/run/initialization/reset_environment.py
@@ -298,11 +298,12 @@ def reset(scope: str = "all") -> Dict[str, Any]:
     """Destructive reset. scope ∈ {all, exam-corpus, timetable, user-subset}.
 
     - all          : full wipe (Neo4j + Supabase data + auth users) AND the entire
-                     exam-marker subsystem listed below.
+                     exam-marker subsystem listed below, including --user-subset copies.
     - exam-corpus  : ONLY the entire exam-marker subsystem, not just public papers:
                      public corpus/eb_* data, cc.examboards storage objects, exam
                      templates, template layouts, questions, boundaries, response
-                     areas, marking batches, student submissions, and mark entries.
+                     areas, marking batches, student submissions, mark entries, and
+                     --user-subset cc.users copies.
     - timetable    : ONLY timetable/calendar materialization tables.
     - user-subset  : ONLY files rows and cc.users storage objects created by
                      seed_exam_corpus.py --user-subset.
@@ -314,10 +315,11 @@ def reset(scope: str = "all") -> Dict[str, Any]:
     _assert_reset_allowed(url, scope)
 
     if scope == "exam-corpus":
-        logger.info("RESET (scope=exam-corpus) — entire exam-marker subsystem: public corpus/eb_* data, cc.examboards storage, templates/layout/questions/boundaries/response areas, marking batches, submissions, mark entries")
+        logger.info("RESET (scope=exam-corpus) — entire exam-marker subsystem: public corpus/eb_* data, cc.examboards storage, templates/layout/questions/boundaries/response areas, marking batches, submissions, mark entries, and --user-subset copies")
+        user_subset = _clear_user_subset_files()
         storage = _clear_exam_storage()
         cleared, failed = _clear_tables(url, headers, EXAM_CORPUS_TABLES)
-        return {"scope": scope, "exam_storage": storage, "tables_cleared": cleared, "tables_failed": failed}
+        return {"scope": scope, "user_subset": user_subset, "exam_storage": storage, "tables_cleared": cleared, "tables_failed": failed}
 
     if scope == "timetable":
         logger.info("RESET (scope=timetable) — timetable/calendar tables")
@@ -341,6 +343,9 @@ def reset(scope: str = "all") -> Dict[str, Any]:
     results["neo4j"] = {"dropped": dropped}
 
     # ── 2. Supabase: clear all data tables (GAIS preserved) ──────────────────
+    # First remove --user-subset cc.users storage objects (+ their files rows) via the
+    # Storage API, so the generic files-table clear below doesn't strand orphaned objects.
+    results["user_subset"] = _clear_user_subset_files()
     logger.info("\n[Supabase] Clearing data tables (preserving gais_*)...")
     url, headers = _sb_headers()
     cleared, failed = [], []