feat(audit): GetTraceEventsAsync returns the ScriptTrace chain for Replay

Flow Debugger (design 2026-09-05 §7.1): IAuditLogRepository.GetTraceEventsAsync walks any entry id to its root and down the Edges/Chain CTE exactly as GetExecutionTreeAsync, then returns the chain's ScriptTrace rows plus the ApiInbound.InboundRequest row via the canonical row mapper, ordered (OccurredAtUtc, EventId). Test doubles implement the new member with a NotSupported stub. Claude-Session: https://claude.ai/code/session_01GSLkkHKz9JrAtmJjmQK3cP
2026-09-05 23:59:08 -04:00
parent 981deed0fd
commit 9300bc1a4f
9 changed files with 246 additions and 0 deletions
@@ -296,6 +296,20 @@ public interface IAuditLogRepository
        Guid executionId,
        CancellationToken ct = default);

+    /// <summary>
+    /// Flow Debugger Replay (design 2026-09-05 §7.1): every row of the execution tree that
+    /// <paramref name="executionId"/> belongs to (walked to the root exactly as
+    /// <see cref="GetExecutionTreeAsync"/> does) whose channel is <c>ScriptTrace</c>, plus the
+    /// tree's <c>ApiInbound.InboundRequest</c> rows, ordered <c>(OccurredAtUtc, EventId)</c>.
+    /// Empty when nothing in the store carries the id.
+    /// </summary>
+    /// <param name="executionId">Any execution id in the chain.</param>
+    /// <param name="ct">Cancellation token.</param>
+    /// <returns>The canonical audit events that make up the flow, oldest first.</returns>
+    Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(
+        Guid executionId,
+        CancellationToken ct = default);
+
    /// <summary>
    /// Returns the distinct, non-null <c>SourceNode</c> values present in the
    /// <c>AuditLog</c> table, in ascending order. Backs the Audit Log page's
@@ -9,6 +9,7 @@ using ZB.MOM.WW.Audit;
 using ZB.MOM.WW.ScadaBridge.Commons.Interfaces.Repositories;
 using ZB.MOM.WW.ScadaBridge.Commons.Types;
 using ZB.MOM.WW.ScadaBridge.Commons.Types.Audit;
+using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
 using ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Entities;

 namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Repositories;
@@ -1125,6 +1126,131 @@ VALUES
        }
    }

+    /// <inheritdoc />
+    public async Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(
+        Guid executionId,
+        CancellationToken ct = default)
+    {
+        // Flow Debugger Replay (design 2026-09-05 §7.1): same walk as GetExecutionTreeAsync
+        // — up to the root, then the Edges/Chain CTE down — but projecting the chain's
+        // execution ids, then the rows themselves filtered to the trace channel + the
+        // inbound row. Two statements keep the row read on EF's canonical mapper
+        // (RowToCanonical) instead of a second hand-mapped reader.
+        var conn = _context.Database.GetDbConnection();
+        var openedHere = false;
+        if (conn.State != System.Data.ConnectionState.Open)
+        {
+            await conn.OpenAsync(ct).ConfigureAwait(false);
+            openedHere = true;
+        }
+
+        var chainIds = new List<Guid>();
+        try
+        {
+            var rootExecutionId = executionId;
+            for (var depth = 0; depth < ExecutionChainMaxDepth; depth++)
+            {
+                Guid? parent;
+                await using (var upCmd = conn.CreateCommand())
+                {
+                    upCmd.CommandText =
+                        "SELECT TOP 1 ParentExecutionId FROM dbo.AuditLog " +
+                        "WHERE ExecutionId = @cur AND ParentExecutionId IS NOT NULL;";
+                    var pCur = upCmd.CreateParameter();
+                    pCur.ParameterName = "@cur";
+                    pCur.Value = rootExecutionId;
+                    upCmd.Parameters.Add(pCur);
+                    var result = await upCmd.ExecuteScalarAsync(ct).ConfigureAwait(false);
+                    parent = result is null or DBNull ? null : (Guid)result;
+                }
+                if (parent is null) break;
+                rootExecutionId = parent.Value;
+            }
+
+            DateTime? rootFirstOccurred;
+            await using (var winCmd = conn.CreateCommand())
+            {
+                winCmd.CommandText = "SELECT MIN(OccurredAtUtc) FROM dbo.AuditLog WHERE ExecutionId = @root;";
+                var pWinRoot = winCmd.CreateParameter();
+                pWinRoot.ParameterName = "@root";
+                pWinRoot.Value = rootExecutionId;
+                winCmd.Parameters.Add(pWinRoot);
+                var winResult = await winCmd.ExecuteScalarAsync(ct).ConfigureAwait(false);
+                rootFirstOccurred = winResult is null or DBNull ? null : (DateTime)winResult;
+            }
+            var edgesWindowPredicate = rootFirstOccurred is null
+                ? string.Empty
+                : "AND OccurredAtUtc >= @winStart AND OccurredAtUtc < @winEnd";
+
+            await using (var idsCmd = conn.CreateCommand())
+            {
+                idsCmd.CommandText = $@"
+                    WITH Edges AS (
+                        SELECT DISTINCT ExecutionId, ParentExecutionId
+                        FROM dbo.AuditLog
+                        WHERE ExecutionId IS NOT NULL
+                        {edgesWindowPredicate}
+                    ),
+                    Chain AS (
+                        SELECT CAST(@root AS uniqueidentifier) AS ExecutionId,
+                               CAST(NULL AS uniqueidentifier) AS ParentExecutionId
+                        UNION ALL
+                        SELECT e.ExecutionId, e.ParentExecutionId
+                        FROM Edges e
+                        INNER JOIN Chain c ON e.ParentExecutionId = c.ExecutionId
+                    )
+                    SELECT ExecutionId FROM Chain
+                    UNION
+                    SELECT ParentExecutionId FROM Chain WHERE ParentExecutionId IS NOT NULL
+                    OPTION (MAXRECURSION {ExecutionChainMaxDepth});";
+                var pRoot = idsCmd.CreateParameter();
+                pRoot.ParameterName = "@root";
+                pRoot.Value = rootExecutionId;
+                idsCmd.Parameters.Add(pRoot);
+                if (rootFirstOccurred is { } rootFirst)
+                {
+                    var pWinStart = idsCmd.CreateParameter();
+                    pWinStart.ParameterName = "@winStart";
+                    pWinStart.Value = rootFirst - ExecutionTreeWindowSlack;
+                    idsCmd.Parameters.Add(pWinStart);
+                    var pWinEnd = idsCmd.CreateParameter();
+                    pWinEnd.ParameterName = "@winEnd";
+                    pWinEnd.Value = rootFirst + ExecutionTreeMaxSpan;
+                    idsCmd.Parameters.Add(pWinEnd);
+                }
+                await using var reader = await idsCmd.ExecuteReaderAsync(ct).ConfigureAwait(false);
+                while (await reader.ReadAsync(ct).ConfigureAwait(false))
+                {
+                    if (!reader.IsDBNull(0)) chainIds.Add(reader.GetGuid(0));
+                }
+            }
+        }
+        finally
+        {
+            if (openedHere)
+            {
+                await conn.CloseAsync().ConfigureAwait(false);
+            }
+        }
+
+        if (chainIds.Count == 0)
+        {
+            return Array.Empty<AuditEvent>();
+        }
+
+        var rows = await _context.Set<AuditLogRow>()
+            .AsNoTracking()
+            .Where(r => r.ExecutionId != null && chainIds.Contains(r.ExecutionId.Value))
+            .Where(r => r.Channel == AuditChannel.ScriptTrace
+                        || (r.Channel == AuditChannel.ApiInbound && r.Kind == AuditKind.InboundRequest))
+            .OrderBy(r => r.OccurredAtUtc)
+            .ThenBy(r => r.EventId)
+            .ToListAsync(ct)
+            .ConfigureAwait(false);
+
+        return rows.Select(RowToCanonical).ToList();
+    }
+
    /// <inheritdoc />
    public async Task<IReadOnlyList<string>> GetDistinctSourceNodesAsync(CancellationToken ct = default)
    {
@@ -430,6 +430,9 @@ public class AuditLogIngestActorCombinedTelemetryTests : TestKit, IClassFixture<
    /// </summary>
    private sealed class BatchFailingAuditRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        private readonly IAuditLogRepository _inner;
        public BatchFailingAuditRepo(IAuditLogRepository inner) { _inner = inner; }

@@ -382,6 +382,9 @@ public class AuditLogIngestActorTests : TestKit, IClassFixture<MsSqlMigrationFix
    /// </summary>
    private sealed class TokenRecordingRepository : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public CancellationToken? BatchToken { get; private set; }

        public List<CancellationToken> RowTokens { get; } = new();
@@ -438,6 +441,9 @@ public class AuditLogIngestActorTests : TestKit, IClassFixture<MsSqlMigrationFix
    /// </summary>
    private sealed class ThrowingRepository : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        private readonly IAuditLogRepository _inner;
        private readonly Guid _poisonId;

@@ -45,6 +45,9 @@ public class AuditLogPurgeActorTests : TestKit, IClassFixture<MsSqlMigrationFixt
    /// </summary>
    private sealed class RecordingRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public List<DateTime> ThresholdQueries { get; } = new();
        public List<DateTime> SwitchedBoundaries { get; } = new();
        public Func<DateTime, long> RowsPerBoundary { get; set; } = _ => 0L;
@@ -37,6 +37,9 @@ public class CentralAuditWriteFailuresTests : TestKit
    /// </summary>
    private sealed class ThrowingRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public Task InsertIfNotExistsAsync(AuditEvent evt, CancellationToken ct = default) =>
            throw new InvalidOperationException("simulated repo failure");
        public Task<IReadOnlyList<AuditEvent>> QueryAsync(
@@ -68,6 +68,9 @@ public class SiteAuditReconciliationActorTests : TestKit, IClassFixture<MsSqlMig
    /// </summary>
    private sealed class RecordingRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public List<AuditEvent> Inserted { get; } = new();
        private readonly HashSet<Guid> _seen = new();
        public int InsertCallCount { get; private set; }
@@ -121,6 +124,9 @@ public class SiteAuditReconciliationActorTests : TestKit, IClassFixture<MsSqlMig
    /// </summary>
    private sealed class PoisonInsertRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        private readonly Guid _poisonId;
        private readonly Exception _fault;
        public List<AuditEvent> Inserted { get; } = new();
@@ -192,6 +198,9 @@ public class SiteAuditReconciliationActorTests : TestKit, IClassFixture<MsSqlMig
    /// </summary>
    private sealed class AlwaysFailUnlessSyntheticRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public List<AuditEvent> Inserted { get; } = new();
        private readonly HashSet<Guid> _seen = new();

@@ -603,6 +612,9 @@ public class SiteAuditReconciliationActorTests : TestKit, IClassFixture<MsSqlMig
    /// <summary>Repository whose every insert throws, so the retry hold-back path is taken.</summary>
    private sealed class AlwaysThrowingRepo : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        public Task InsertIfNotExistsAsync(AuditEvent evt, CancellationToken ct = default) =>
            throw new InvalidOperationException("central insert failed");

@@ -0,0 +1,76 @@
+using Microsoft.EntityFrameworkCore;
+using ZB.MOM.WW.ScadaBridge.Commons.Types.Audit;
+using ZB.MOM.WW.ScadaBridge.Commons.Types.Debugging;
+using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
+using ZB.MOM.WW.ScadaBridge.ConfigurationDatabase;
+using ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Repositories;
+using ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests.Migrations;
+using Xunit;
+
+namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests.Repositories;
+
+/// <summary>
+/// Flow Debugger Replay (design 2026-09-05 §7.1): <see cref="AuditLogRepository.GetTraceEventsAsync"/>
+/// returns the whole chain's ScriptTrace rows plus the inbound row, from ANY entry id, ordered,
+/// and excludes unrelated and non-trace rows. Runs against the shared MSSQL fixture.
+/// </summary>
+public class AuditLogRepositoryTraceEventsTests : IClassFixture<MsSqlMigrationFixture>
+{
+    private readonly MsSqlMigrationFixture _fixture;
+    public AuditLogRepositoryTraceEventsTests(MsSqlMigrationFixture fixture) => _fixture = fixture;
+
+    private ScadaBridgeDbContext CreateContext() => new(
+        new DbContextOptionsBuilder<ScadaBridgeDbContext>().UseSqlServer(_fixture.ConnectionString).Options);
+
+    [SkippableFact]
+    public async Task ReturnsInboundPlusChainTraceRows_Ordered_FromAnyEntry()
+    {
+        Skip.IfNot(_fixture.Available, _fixture.SkipReason);
+        var siteId = "test-trace-" + Guid.NewGuid().ToString("N")[..8];
+        var t0 = new DateTime(2026, 9, 5, 12, 0, 0, DateTimeKind.Utc);
+        var root = Guid.NewGuid();
+        var siteRun = Guid.NewGuid();
+        var cascade = Guid.NewGuid();
+
+        var inbound = ScadaBridgeAuditEventFactory.Create(AuditChannel.ApiInbound, AuditKind.InboundRequest, AuditStatus.Delivered,
+            occurredAtUtc: t0, actor: "key", target: "M", executionId: root, sourceSiteId: siteId, responseSummary: "{}");
+        var route = ScadaBridgeAuditEventFactory.Create(AuditChannel.ScriptTrace, AuditKind.RouteCall, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(10), target: "I.S", executionId: root, sourceSiteId: siteId,
+            extra: new ScriptTraceExtra { ChildExecutionId = siteRun }.ToJson());
+        var run = ScadaBridgeAuditEventFactory.Create(AuditChannel.ScriptTrace, AuditKind.ScriptRun, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(20), target: "S", executionId: siteRun, parentExecutionId: root, sourceSiteId: siteId);
+        var write = ScadaBridgeAuditEventFactory.Create(AuditChannel.ScriptTrace, AuditKind.AttributeWrite, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(30), target: "A", executionId: siteRun, parentExecutionId: root, sourceSiteId: siteId);
+        var cascadeRun = ScadaBridgeAuditEventFactory.Create(AuditChannel.ScriptTrace, AuditKind.ScriptRun, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(40), target: "C", executionId: cascade, parentExecutionId: siteRun, sourceSiteId: siteId);
+        // Same chain but NOT a trace row: an outbound API call the site script made — excluded.
+        var outbound = ScadaBridgeAuditEventFactory.Create(AuditChannel.ApiOutbound, AuditKind.ApiCall, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(25), target: "ERP.x", executionId: siteRun, parentExecutionId: root, sourceSiteId: siteId);
+        // Unrelated chain.
+        var other = ScadaBridgeAuditEventFactory.Create(AuditChannel.ScriptTrace, AuditKind.ScriptRun, AuditStatus.Delivered,
+            occurredAtUtc: t0.AddMilliseconds(5), target: "Z", executionId: Guid.NewGuid(), sourceSiteId: siteId);
+
+        await using (var ctx = CreateContext())
+        {
+            var repo = new AuditLogRepository(ctx);
+            foreach (var e in new[] { cascadeRun, other, write, outbound, inbound, run, route })
+                await repo.InsertIfNotExistsAsync(e);
+        }
+
+        await using (var ctx = CreateContext())
+        {
+            var repo = new AuditLogRepository(ctx);
+            foreach (var entry in new[] { root, siteRun, cascade })
+            {
+                var events = await repo.GetTraceEventsAsync(entry);
+                Assert.Equal(
+                    new[] { inbound.EventId, route.EventId, run.EventId, write.EventId, cascadeRun.EventId },
+                    events.Select(e => e.EventId).ToArray());
+                Assert.Equal(inbound, events[0]);
+                Assert.Equal(cascadeRun, events[4]);
+            }
+
+            Assert.Empty(await repo.GetTraceEventsAsync(Guid.NewGuid()));
+        }
+    }
+}
@@ -89,6 +89,9 @@ public class SiteAuditPushFlowTests : TestKit
    /// </summary>
    private sealed class InMemoryAuditLogRepository : IAuditLogRepository
    {
+        public Task<IReadOnlyList<AuditEvent>> GetTraceEventsAsync(Guid executionId, CancellationToken ct = default)
+            => throw new NotSupportedException("GetTraceEventsAsync is not exercised by this test double.");
+
        private readonly ConcurrentDictionary<Guid, AuditEvent> _rows = new();

        public IReadOnlyCollection<AuditEvent> Rows => _rows.Values.ToList();