ScadaBridge/tests/ZB.MOM.WW.ScadaBridge.SiteRuntime.Tests/Actors/InstanceActorOverrideRepublishTests.cs

using System.Text.Json;
using Akka.Actor;
using Akka.TestKit.Xunit2;
using Microsoft.Extensions.Logging.Abstractions;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Instance;
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
using ZB.MOM.WW.ScadaBridge.Commons.Types.Flattening;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
using ZB.MOM.WW.ScadaBridge.TestSupport;

namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Tests.Actors;

/// <summary>
/// Review 2026-10-21 01/SR-U1: persisted static overrides were applied to the Instance Actor's
/// own <c>_attributes</c> but never republished. The Script/Alarm children are created in
/// <c>PreStart</c> BEFORE the async override load lands, so their snapshots were seeded with the
/// deploy default — and static attributes never re-emit a change — so after a restart, failover,
/// enable or sweep start every trigger evaluated against the default, indefinitely.
/// </summary>
public class InstanceActorOverrideRepublishTests : TestKit, IDisposable
{
    private readonly SiteStorageService _storage;
    private readonly ScriptCompilationService _compilationService;
    private readonly SharedScriptLibrary _sharedScriptLibrary;
    private readonly TestLocalDb _localDb;

    /// <summary>Creates a temporary LocalDb-backed site store per test.</summary>
    public InstanceActorOverrideRepublishTests()
    {
        _localDb = TestLocalDb.CreateTemp("instance-override-republish");
        _storage = new SiteStorageService(_localDb.Db, NullLogger<SiteStorageService>.Instance);
        _storage.InitializeAsync().GetAwaiter().GetResult();
        _compilationService = new ScriptCompilationService(NullLogger<ScriptCompilationService>.Instance);
        _sharedScriptLibrary = new SharedScriptLibrary(
            _compilationService, NullLogger<SharedScriptLibrary>.Instance);
    }

    /// <inheritdoc />
    void IDisposable.Dispose()
    {
        Shutdown();
        var path = _localDb.Path;
        _localDb.Dispose();
        TestLocalDb.DeleteFiles(path);
    }

    /// <summary>
    /// <c>Mode</c> defaults to <c>Manual</c>; the runtime override is <c>Auto</c>. An Expression
    /// alarm on <c>Mode == "Auto" &amp;&amp; Temp &gt; 100</c>.
    /// </summary>
    private static FlattenedConfiguration ModeAlarmConfig(string instanceName) => new()
    {
        InstanceUniqueName = instanceName,
        Attributes =
        [
            new ResolvedAttribute { CanonicalName = "Mode", Value = "Manual", DataType = "String" },
            new ResolvedAttribute { CanonicalName = "Temp", Value = "50", DataType = "Int32" }
        ],
        Alarms =
        [
            new ResolvedAlarm
            {
                CanonicalName = "AutoOverTemp",
                TriggerType = "Expression",
                TriggerConfiguration = JsonSerializer.Serialize(new
                {
                    expression = "\"Auto\".Equals(Attributes[\"Mode\"]) && int.Parse(Attributes[\"Temp\"].ToString()) > 100"
                }),
                PriorityLevel = 500
            }
        ]
    };

    private IActorRef StartInstance(string instanceName, FlattenedConfiguration config) =>
        ActorOf(Props.Create(() => new InstanceActor(
            instanceName, JsonSerializer.Serialize(config), _storage, _compilationService,
            _sharedScriptLibrary, null, new SiteRuntimeOptions(), NullLogger<InstanceActor>.Instance)));

    private AlarmState AlarmStateOf(IActorRef instance, string instanceName, string alarm)
    {
        instance.Tell(new GetAlarmSnapshotRequest("snap", instanceName, DateTimeOffset.UtcNow));
        var snapshot = ExpectMsg<GetAlarmSnapshotResponse>(TimeSpan.FromSeconds(5));
        return snapshot.Alarms.Single(a => a.AlarmName == alarm).State;
    }

    /// <summary>
    /// The finding's regression: deploy → override → restart → the Expression alarm sees the
    /// override once another attribute changes. Before the fix the child snapshot still held
    /// <c>Manual</c>, so the alarm never raised.
    /// </summary>
    /// <returns>A task that represents the asynchronous operation.</returns>
    [Fact]
    public async Task PersistedOverride_ReachesExpressionAlarm_AfterActorRestart()
    {
        const string name = "PumpU1";
        await _storage.SetStaticOverrideAsync(name, "Mode", "Auto");

        var instance = StartInstance(name, ModeAlarmConfig(name));

        // The override load is async; GetAttribute is the authoritative view. Generous timeouts:
        // a cold Roslyn compile of the trigger expression runs inside PreStart.
        AwaitAssert(() =>
        {
            instance.Tell(new GetAttributeRequest("g", name, "Mode", DateTimeOffset.UtcNow));
            Assert.Equal("Auto", ExpectMsg<GetAttributeResponse>(TimeSpan.FromSeconds(15)).Value?.ToString());
        }, TimeSpan.FromSeconds(30));

        instance.Tell(new SetStaticAttributeCommand("w", name, "Temp", "150", DateTimeOffset.UtcNow));
        Assert.True(ExpectMsg<SetStaticAttributeResponse>().Success);

        AwaitAssert(() => Assert.Equal(AlarmState.Active, AlarmStateOf(instance, name, "AutoOverTemp")),
            TimeSpan.FromSeconds(10), TimeSpan.FromMilliseconds(100));
    }

    /// <summary>
    /// A static write that lands BEFORE the async override load is newer than the stored
    /// override, so the load must not overwrite (or republish over) it.
    /// </summary>
    /// <returns>A task that represents the asynchronous operation.</returns>
    [Fact]
    public async Task WriteBeforeOverrideLoad_IsNotOverwrittenByTheOlderStoredValue()
    {
        const string name = "PumpU1Race";
        await _storage.SetStaticOverrideAsync(name, "Mode", "Auto");
        var gate = new TaskCompletionSource();

        // A long flush window keeps the write dirty in memory, so the store still holds the
        // OLDER "Auto" when the (gated) load finally reads it.
        var flushWindow = TimeSpan.FromMinutes(1);
        var configJson = JsonSerializer.Serialize(ModeAlarmConfig(name));
        var instance = ActorOf(Props.Create(() => new InstanceActor(
            name, configJson, _storage, _compilationService,
            _sharedScriptLibrary, null, new SiteRuntimeOptions(), NullLogger<InstanceActor>.Instance,
            null, null, null, flushWindow, gate.Task)));

        instance.Tell(new SetStaticAttributeCommand("w", name, "Mode", "Maintenance", DateTimeOffset.UtcNow));
        Assert.True(ExpectMsg<SetStaticAttributeResponse>(TimeSpan.FromSeconds(15)).Success);

        gate.SetResult();
        await Task.Delay(500);

        instance.Tell(new GetAttributeRequest("g", name, "Mode", DateTimeOffset.UtcNow));
        Assert.Equal("Maintenance", ExpectMsg<GetAttributeResponse>().Value?.ToString());
    }

    /// <summary>
    /// Review 2026-10-21 01/SR-M3: stopping the instance answers every pending
    /// <c>WaitAsync</c> at once ("instance stopped") instead of leaving the script to burn its
    /// whole wait timeout plus the Ask timeout, and leaves no timeout self-message to dead-letter.
    /// </summary>
    [Fact]
    public void Stop_AnswersPendingAttributeWaiters_Immediately()
    {
        const string name = "PumpM3";
        var instance = StartInstance(name, ModeAlarmConfig(name));
        var waiter = CreateTestProbe();

        instance.Tell(new WaitForAttributeRequest(
            "wait-1", name, "Mode", "\"NeverReached\"", null, TimeSpan.FromMinutes(5), DateTimeOffset.UtcNow),
            waiter.Ref);
        // A round trip proves the waiter is registered before the stop.
        instance.Tell(new GetAttributeRequest("g", name, "Mode", DateTimeOffset.UtcNow));
        ExpectMsg<GetAttributeResponse>(TimeSpan.FromSeconds(15));

        Sys.Stop(instance);

        var response = waiter.ExpectMsg<WaitForAttributeResponse>(TimeSpan.FromSeconds(5));
        Assert.Equal("wait-1", response.CorrelationId);
        Assert.False(response.Matched);
        Assert.False(response.TimedOut);
        Assert.Contains("instance stopped", response.ErrorMessage);
    }
}