Implement workspace history reconciliation

core/management/commands/reconcile_workspace_metric_history.py

@@ -14,6 +14,7 @@ from core.models import (
     WorkspaceMetricSnapshot,
 )
 from core.views.workspace import _conversation_for_person
+from core.workspace import compact_snapshot_rows
 
 
 def _score_from_lag(lag_ms, target_hours=4):
@@ -219,6 +220,7 @@ class Command(BaseCommand):
         parser.add_argument("--limit", type=int, default=200000)
         parser.add_argument("--dry-run", action="store_true", default=False)
         parser.add_argument("--no-reset", action="store_true", default=False)
+        parser.add_argument("--skip-compact", action="store_true", default=False)
 
     def handle(self, *args, **options):
         days = max(1, int(options.get("days") or 365))
@@ -229,6 +231,7 @@ class Command(BaseCommand):
         limit = max(1, int(options.get("limit") or 200000))
         dry_run = bool(options.get("dry_run"))
         reset = not bool(options.get("no_reset"))
+        compact_enabled = not bool(options.get("skip_compact"))
         today_start = dj_timezone.now().astimezone(timezone.utc).replace(
             hour=0,
             minute=0,
@@ -250,6 +253,7 @@ class Command(BaseCommand):
         deleted = 0
         snapshots_created = 0
         checkpoints_total = 0
+        compacted_deleted = 0
 
         for person in people:
             identifiers_qs = PersonIdentifier.objects.filter(user=person.user, person=person)
@@ -410,6 +414,28 @@ class Command(BaseCommand):
                         "participant_feedback",
                     ]
                 )
+                if compact_enabled:
+                    snapshot_rows = list(
+                        WorkspaceMetricSnapshot.objects.filter(conversation=conversation)
+                        .order_by("computed_at", "id")
+                        .values("id", "computed_at", "source_event_ts")
+                    )
+                    now_ts_ms = int(dj_timezone.now().timestamp() * 1000)
+                    keep_ids = compact_snapshot_rows(
+                        snapshot_rows=snapshot_rows,
+                        now_ts_ms=now_ts_ms,
+                        cutoff_ts_ms=cutoff_ts,
+                    )
+                    if keep_ids:
+                        compacted_deleted += (
+                            WorkspaceMetricSnapshot.objects.filter(conversation=conversation)
+                            .exclude(id__in=list(keep_ids))
+                            .delete()[0]
+                        )
+                    else:
+                        compacted_deleted += WorkspaceMetricSnapshot.objects.filter(
+                            conversation=conversation
+                        ).delete()[0]
 
         self.stdout.write(
             self.style.SUCCESS(
@@ -418,6 +444,8 @@ class Command(BaseCommand):
                 f"checkpoints={checkpoints_total} "
                 f"created={snapshots_created} "
                 f"deleted={deleted} "
+                f"compacted_deleted={compacted_deleted} "
+                f"compact_enabled={compact_enabled} "
                 f"reset={reset} dry_run={dry_run} "
                 f"days={days} step_messages={step_messages} limit={limit}"
             )