fix: prevent scoring data loss and fix fragile test in stale PR refresh (entrius#769 follow-up)

Author: wdeveloper16

gittensor/validator/storage/queries.py

@@ -100,6 +100,16 @@
     updated_at = NOW()
 """
 
+# Targeted state-only refresh for stale-closed PRs (avoids overwriting scored columns)
+REFRESH_STALE_PR_STATES = """
+UPDATE pull_requests
+   SET pr_state   = 'CLOSED',
+       updated_at = NOW()
+ WHERE number               = %s
+   AND repository_full_name = %s
+   AND pr_state            != 'CLOSED'
+"""
+
 # Issue Queries
 BULK_UPSERT_ISSUES = """
 INSERT INTO issues (

gittensor/validator/storage/repository.py

@@ -23,6 +23,7 @@
     CLEANUP_STALE_MINER_EVALUATIONS_BY_HOTKEY,
     CLEANUP_STALE_MINERS,
     CLEANUP_STALE_MINERS_BY_HOTKEY,
+    REFRESH_STALE_PR_STATES,
     SET_MINER,
 )
 
@@ -217,6 +218,28 @@ def store_pull_requests_bulk(self, pull_requests: List[PullRequest], commit: boo
             self.logger.error(f'Error in bulk pull request storage: {e}')
             return 0
 
+    def refresh_stale_pr_states(self, pull_requests: List[PullRequest], commit: bool = True) -> int:
+        """Update pr_state to CLOSED for stale PRs without touching scoring columns.
+
+        Uses a targeted UPDATE so previously-computed scores (earned_score, base_score,
+        credibility_multiplier, etc.) are preserved on rows that were already scored.
+        Only rows currently stored as non-CLOSED are affected.
+        """
+        if not pull_requests:
+            return 0
+        values = [(pr.number, pr.repository_full_name) for pr in pull_requests]
+        try:
+            with self.get_cursor() as cursor:
+                cursor.executemany(REFRESH_STALE_PR_STATES, values)
+                if commit:
+                    self.db.commit()
+                return len(values)
+        except Exception as e:
+            if commit:
+                self.db.rollback()
+            self.logger.error(f'Error refreshing stale PR states: {e}')
+            return 0
+
     def store_issues_bulk(self, issues: List[Issue], commit: bool = True) -> int:
         """
         Bulk insert/update issues with efficient SQL conflict resolution

gittensor/validator/utils/storage.py

@@ -71,7 +71,7 @@ def _adapt_mirror(scored_list):
                 result.stored_counts['closed_pull_requests'] = self.repo.store_pull_requests_bulk(
                     miner_eval.closed_pull_requests + _adapt_mirror(miner_eval.mirror_closed_prs), commit=False
                 )
-                result.stored_counts['stale_closed_pull_requests'] = self.repo.store_pull_requests_bulk(
+                result.stored_counts['stale_closed_pull_requests'] = self.repo.refresh_stale_pr_states(
                     miner_eval.stale_closed_pull_requests, commit=False
                 )
                 result.stored_counts['issues'] = self.repo.store_issues_bulk(miner_eval.get_all_issues(), commit=False)

tests/validator/utils/test_storage_mirror.py

@@ -86,6 +86,7 @@ def _make_storage_with_mock_repo():
             mock_repo = MagicMock()
             mock_repo.set_miner.return_value = 1
             mock_repo.store_pull_requests_bulk.return_value = 0  # actual count irrelevant
+            mock_repo.refresh_stale_pr_states.return_value = 0
             mock_repo.store_issues_bulk.return_value = 0
             mock_repo.store_file_changes_bulk.return_value = 0
             mock_repo.set_miner_evaluation.return_value = True
@@ -173,12 +174,17 @@ def test_stale_closed_prs_are_stored_separately(self):
 
         storage.store_evaluation(eval_)
 
-        stale_call = mock_repo.store_pull_requests_bulk.call_args_list[3]
-        stale_arg = stale_call.args[0]
+        # Stale PRs must go through refresh_stale_pr_states (targeted UPDATE), not the
+        # full-column UPSERT, so previously-scored rows are not overwritten with defaults.
+        mock_repo.refresh_stale_pr_states.assert_called_once()
+        stale_arg = mock_repo.refresh_stale_pr_states.call_args.args[0]
         assert len(stale_arg) == 1
         assert stale_arg[0].number == 7
         assert stale_arg[0].pr_state == PRState.CLOSED
         assert eval_.total_closed_prs == 0
+        # Verify the stale PR did not leak into store_pull_requests_bulk
+        for call in mock_repo.store_pull_requests_bulk.call_args_list:
+            assert not any(pr.number == 7 for pr in call.args[0])
 
 
 def test_cleanup_stale_called_with_commit_false():