test+chore: real-SQLite query coverage, batch detect writes, finish date centralization

Review follow-ups:
- (HIGH) Add real-SQLite round-trip tests for the new query methods so the load-bearing
  lexical O-format date ordering is verified, not just mocked: Anomaly
  ListByDateRange/CountSince, Snapshot CountSince/ListByEvents grouping, Event Query/GetMany.
- (MED) DetectAnomaliesUseCase: one SaveChanges per event instead of per anomaly.
- (LOW) Route PlacedBetRepository + ExcelExporter date bounds through SqliteDateText.
- (LOW) Backtest: reject a one-sided date range (was silently ignored).
- (LOW) Refresh stale comments after the detector fan-out.
This commit is contained in:
2026-05-29 01:25:25 +03:00
parent c9eee9f907
commit b67030ae7f
5 changed files with 107 additions and 11 deletions
@@ -14,7 +14,7 @@ namespace Marathon.Application.UseCases;
/// <item>Loads all tracked events.</item>
/// <item>For each event, fetches its last-24-hour live snapshots.</item>
/// <item>Runs <see cref="AnomalyDetector"/> over the snapshot timeline.</item>
/// <item>Persists any new anomalies that have not already been stored (dedup by EventId + DetectedAt minute-window).</item>
/// <item>Persists any new anomalies that have not already been stored (dedup by EventId + Kind + DetectedAt minute-window).</item>
/// </list>
/// </summary>
/// <remarks>
@@ -138,8 +138,8 @@ public sealed class DetectAnomaliesUseCase
List<Anomaly> existingForEvent,
CancellationToken ct)
{
// Fan out over every detector kind; dedup below keys on EventId + Kind so the
// flip and steam signals for one event persist independently.
// Fan out over every detector; dedup below keys on EventId + Kind so the flip,
// steam, and freeze signals for one event persist independently.
var detected = detectors
.SelectMany(d => d.Detect(ev.Id, snapshots))
.ToList();
@@ -154,11 +154,15 @@ public sealed class DetectAnomaliesUseCase
continue;
await _anomalyRepo.AddAsync(anomaly, ct);
await _anomalyRepo.SaveChangesAsync(ct);
existingForEvent.Add(anomaly); // Keep local list in sync so the same cycle doesn't re-add.
persisted++;
}
// One write per event rather than per anomaly — with three detectors an event
// can yield several new anomalies in a single cycle.
if (persisted > 0)
await _anomalyRepo.SaveChangesAsync(ct);
return persisted;
}