using Akka.Actor;
using Akka.Cluster;
using Akka.Configuration;
using Shouldly;
using Xunit;
using ZB.MOM.WW.OtOpcUa.Commons.Messages.Redundancy;
using ZB.MOM.WW.OtOpcUa.ControlPlane.Redundancy;

namespace ZB.MOM.WW.OtOpcUa.ControlPlane.Tests;

/// <summary>
/// Pins <b>which</b> node <see cref="RedundancyStateActor"/> elects Primary, using a real two-node
/// cluster deliberately built so that the oldest member is <i>not</i> the lowest-addressed one.
/// </summary>
/// <remarks>
///     <para>
///         The distinction is the whole point. Akka offers two different "first" members and they
///         disagree after any restart:
///     </para>
///     <list type="bullet">
///         <item>
///             <description>
///                 <c>ClusterState.RoleLeader(role)</c> — the <b>lowest-addressed</b> Up member with
///                 that role. Address order is host, then port; it has nothing to do with time.
///             </description>
///         </item>
///         <item>
///             <description>
///                 <b>Oldest</b> — the Up member with the lowest up-number, i.e. the one that has been
///                 in the cluster longest. This is what <c>ClusterSingletonManager</c> uses to place
///                 singletons, and (under <c>keep-oldest</c>) what the split-brain resolver keys on.
///             </description>
///         </item>
///     </list>
///     <para>
///         They coincide on a freshly-formed cluster, which is why single-node and
///         happy-path tests never caught the difference. They diverge the moment a node restarts: the
///         restarted node becomes the <i>youngest</i> while keeping its address, so if it happens to
///         hold the lower address it becomes role leader — and the actor would have named it Primary
///         while the cluster singletons, and every piece of work they own, stayed on the other node.
///         The Primary-gated data plane (writes, alarm acks, alerts emit, alarm-history drain) would
///         then be enabled on the node that is not hosting the work.
///     </para>
///     <para>
///         This test reproduces that divergence directly rather than waiting for a restart: node A
///         binds the <b>higher</b> port and joins first (so it is oldest), node B binds the
///         <b>lower</b> port and joins second (so it is role leader). The correct answer is A.
///     </para>
/// </remarks>
public sealed class RedundancyPrimaryElectionTests : IAsyncLifetime
{
    // Fixed, unusual ports: the test needs a deterministic address ordering, which port 0 cannot give.
    private const int OldestPort = 19_531;   // joins FIRST  -> oldest, but the HIGHER address
    private const int YoungestPort = 19_530; // joins SECOND -> role leader, the LOWER address

    private ActorSystem? _oldest;
    private ActorSystem? _youngest;

    private static Config NodeConfig(int port) => ConfigurationFactory.ParseString($$"""
        akka {
            loglevel = "WARNING"
            actor.provider = "Akka.Cluster.ClusterActorRefProvider, Akka.Cluster"
            remote.dot-netty.tcp {
                hostname = "127.0.0.1"
                port = {{port}}
            }
            cluster {
                seed-nodes = ["akka.tcp://redundancy-election@127.0.0.1:{{OldestPort}}"]
                roles = ["admin", "driver"]
                min-nr-of-members = 1
                run-coordinated-shutdown-when-down = off
                downing-provider-class = ""
            }
        }
        """);

    public async Task InitializeAsync()
    {
        // Order matters: the seed forms the cluster and is therefore the oldest member.
        _oldest = ActorSystem.Create("redundancy-election", NodeConfig(OldestPort));
        await WaitForUpAsync(_oldest, expectedMembers: 1);

        _youngest = ActorSystem.Create("redundancy-election", NodeConfig(YoungestPort));
        await WaitForUpAsync(_oldest, expectedMembers: 2);
        await WaitForUpAsync(_youngest, expectedMembers: 2);
    }

    public async Task DisposeAsync()
    {
        if (_youngest is not null) await _youngest.Terminate();
        if (_oldest is not null) await _oldest.Terminate();
    }

    private static async Task WaitForUpAsync(ActorSystem system, int expectedMembers)
    {
        var cluster = Akka.Cluster.Cluster.Get(system);
        var deadline = DateTime.UtcNow + TimeSpan.FromSeconds(30);
        while (DateTime.UtcNow < deadline)
        {
            var up = cluster.State.Members.Count(m => m.Status == MemberStatus.Up);
            if (up >= expectedMembers) return;
            await Task.Delay(100);
        }

        throw new TimeoutException(
            $"cluster on port {cluster.SelfAddress.Port} never saw {expectedMembers} Up members "
            + $"(saw {cluster.State.Members.Count(m => m.Status == MemberStatus.Up)})");
    }

    /// <summary>
    /// Sanity check on the fixture itself: if the two orderings did not actually diverge, the real
    /// assertion below would pass for the wrong reason. This asserts the setup produced the
    /// divergence it claims to — role leader is the younger node, not the oldest one.
    /// </summary>
    [Fact]
    public void Fixture_actually_produces_divergent_age_and_address_ordering()
    {
        var state = Akka.Cluster.Cluster.Get(_oldest!).State;

        state.RoleLeader("driver")!.Port.ShouldBe(
            YoungestPort,
            "the fixture is only meaningful if the lowest-addressed node is the YOUNGER one");
    }

    /// <summary>
    /// The elected Primary is the oldest driver member — the node that hosts the cluster singletons —
    /// not the lowest-addressed one.
    /// </summary>
    [Fact]
    public async Task Primary_is_the_oldest_driver_member_not_the_role_leader()
    {
        var snapshot = await CaptureSnapshotAsync();

        var primaries = snapshot.Where(n => n.Role == RedundancyRole.Primary).ToList();
        primaries.Count.ShouldBe(1, "exactly one node may be Primary");
        primaries[0].NodeId.Value.ShouldBe(
            $"127.0.0.1:{OldestPort}",
            "Primary must follow cluster-singleton placement (oldest Up member), or the Primary-gated data plane is enabled on a node that is not hosting the work");
    }

    /// <summary>The younger node is Secondary, not a second Primary and not Detached.</summary>
    [Fact]
    public async Task Younger_driver_member_is_secondary()
    {
        var snapshot = await CaptureSnapshotAsync();

        snapshot.Single(n => n.NodeId.Value == $"127.0.0.1:{YoungestPort}")
            .Role.ShouldBe(RedundancyRole.Secondary);
    }

    /// <summary>
    /// <c>IsDriverPrimary</c> agrees with <c>Role</c>. They feed different consumers — the OPC UA
    /// ServiceLevel calculation reads the flag while the data-plane gates read the role — so a
    /// disagreement would advertise one node as authoritative while gating writes on the other.
    /// </summary>
    [Fact]
    public async Task IsDriverPrimary_flag_agrees_with_the_role()
    {
        var snapshot = await CaptureSnapshotAsync();

        foreach (var node in snapshot)
        {
            node.IsDriverPrimary.ShouldBe(
                node.Role == RedundancyRole.Primary,
                $"{node.NodeId.Value} disagrees between Role and IsDriverPrimary");
        }
    }

    private async Task<IReadOnlyList<NodeRedundancyState>> CaptureSnapshotAsync()
    {
        var tcs = new TaskCompletionSource<RedundancyStateChanged>(
            TaskCreationOptions.RunContinuationsAsynchronously);

        _oldest!.ActorOf(
            RedundancyStateActor.Props(broadcast: msg =>
            {
                if (msg is RedundancyStateChanged changed) tcs.TrySetResult(changed);
            }),
            $"redundancy-{Guid.NewGuid():N}");

        var published = await tcs.Task.WaitAsync(TimeSpan.FromSeconds(15));
        return published.Nodes;
    }
}