AyCode.Core/AyCode.Core.Tests/TestModels/BenchmarkTestDataProvider.cs

using AyCode.Core.Serializers.Binaries;
using System.Collections;
using System.Reflection;
using System.Runtime.CompilerServices;

namespace AyCode.Core.Tests.TestModels;

/// <summary>
/// Charset suffix presets for the per-property string augmentation in
/// <c>BenchmarkStringSupport.ToLongString</c>. The benchmark applies the configured suffix to every
/// short (≤ <c>FixStrMaxLength</c>) string property across the test data graph (via reflection in
/// <c>BenchmarkStringSupport.EnsureAllStringsBypassFixStr</c>), producing long-string benchmark payloads
/// with a controlled UTF-8 content profile.
///
/// Switch by assigning to <see cref="BenchmarkTestDataProvider.LongStringSuffix"/> from the interactive
/// Settings → Charset submenu (or programmatically). The active charset is recorded in the .LLM
/// markdown output header so per-charset bench files are self-documenting.
/// </summary>
public static class CharsetSuffixes
{
    // ─────────────────────────────────────────────────────────────────────────
    // Consistent length across all charsets (UTF-16 char count, NOT UTF-8 byte count):
    //   *Short = 40 char (5-char base × 8 repetitions) → StringSmall / StringAscii tier
    //   *Long  = 280 char (Short × 7) → StringMedium / StringAscii tier
    //
    // Same length across charsets isolates the workload variable to UTF-8 byte content
    // (1-byte ASCII vs 2-byte Latin1 / Cyrillic vs 3-byte CJK vs mixed) — wire-size and
    // encode/decode cost differences are pure charset effects, not length effects.
    //
    // Const-concat for compile-time evaluation (usable as attribute / DataRow source).
    // ─────────────────────────────────────────────────────────────────────────

    /// <summary>Empty suffix — baseline string property values stay short, hitting the
    /// <c>FixStrAscii</c> / short-string fast-path. Stress-test for short-string code paths.</summary>
    public const string Latin1FixAscii = "";

    // ── Pure ASCII (every byte < 0x80) ──
    // Tier: StringAscii (167) — byte→char SIMD widening, zero UTF-8 decode.
    // UTF-8 byte count: 40 byte (Short), 280 byte (Long) — 1:1 char:byte.
    private const string AsciiBase = " quic";  // 5 char ASCII
    public const string AsciiShort = AsciiBase + AsciiBase + AsciiBase + AsciiBase
                                   + AsciiBase + AsciiBase + AsciiBase + AsciiBase;  // 40 char
    public const string AsciiLong = AsciiShort + AsciiShort + AsciiShort + AsciiShort
                                  + AsciiShort + AsciiShort + AsciiShort;            // 280 char

    // ── Latin1 (Hungarian proxy — ISO-8859-1 + Latin-2 ő/ű) ──
    // Tier: StringSmall (91) Short / StringMedium (94) Long.
    // UTF-8 byte count: ~72 byte Short (5 char base = 9 byte UTF-8: space+á+r+v+í), ~504 byte Long.
    private const string Latin1Base = " árví";  // 5 char (space + á + r + v + í) — multi-byte mix
    public const string Latin1Short = Latin1Base + Latin1Base + Latin1Base + Latin1Base
                                    + Latin1Base + Latin1Base + Latin1Base + Latin1Base;  // 40 char
    public const string Latin1Long = Latin1Short + Latin1Short + Latin1Short + Latin1Short
                                   + Latin1Short + Latin1Short + Latin1Short;              // 280 char

    // ── CJK BMP (Chinese / Japanese / Korean Basic Multilingual Plane) ──
    // Tier: StringSmall (91) Short / StringMedium (94) Long.
    // UTF-8 byte count: ~104 byte Short (5 char base = 13 byte UTF-8: 1 ASCII space + 4×3-byte CJK),
    // ~728 byte Long. Homogeneous 3-byte runs — primary win region for SIMD multi-byte transcoder.
    private const string CjkBmpBase = " 你好世界";  // 5 char (space + 4 Chinese)
    public const string CjkBmpShort = CjkBmpBase + CjkBmpBase + CjkBmpBase + CjkBmpBase
                                    + CjkBmpBase + CjkBmpBase + CjkBmpBase + CjkBmpBase;  // 40 char
    public const string CjkBmpLong = CjkBmpShort + CjkBmpShort + CjkBmpShort + CjkBmpShort
                                   + CjkBmpShort + CjkBmpShort + CjkBmpShort;              // 280 char

    // ── Cyrillic (Russian / Ukrainian) ──
    // Tier: StringSmall (91) Short / StringMedium (94) Long.
    // UTF-8 byte count: ~72 byte Short (5 char base = 9 byte UTF-8: 1 ASCII + 4×2-byte Cyrillic),
    // ~504 byte Long. Homogeneous 2-byte runs — different shape than Latin1 interspersed.
    private const string CyrillicBase = " Прив";  // 5 char (space + 4 Cyrillic)
    public const string CyrillicShort = CyrillicBase + CyrillicBase + CyrillicBase + CyrillicBase
                                      + CyrillicBase + CyrillicBase + CyrillicBase + CyrillicBase;  // 40 char
    public const string CyrillicLong = CyrillicShort + CyrillicShort + CyrillicShort + CyrillicShort
                                     + CyrillicShort + CyrillicShort + CyrillicShort;                // 280 char

    // ── Mixed (multi-codepage in one payload) ──
    // Tier: StringSmall (91) Short / StringMedium (94) Long.
    // UTF-8 byte count: ~88 byte Short (5 char base = 11 byte UTF-8: 1 ASCII + 1×2-byte Hungarian
    // + 1×3-byte CJK + 2×2-byte Cyrillic), ~616 byte Long. No surrogate pairs (keeps UTF-16
    // length predictable); cross-tier transcoder coverage in one payload.
    private const string MixedBase = " á你Пй";  // 5 char (space + Hungarian + Chinese + 2× Cyrillic)
    public const string MixedShort = MixedBase + MixedBase + MixedBase + MixedBase
                                   + MixedBase + MixedBase + MixedBase + MixedBase;  // 40 char
    public const string MixedLong = MixedShort + MixedShort + MixedShort + MixedShort
                                  + MixedShort + MixedShort + MixedShort;            // 280 char
}

// ============================================================================================
// Cross-family shared state. The charset suffix is a global benchmark configuration — settable
// once via the interactive Menu, applied uniformly to every family's data construction. Lives in
// a non-generic helper so it ISN'T per-closed-generic (which would cause the Menu setter to affect
// only one family). The <see cref="BenchmarkTestDataProvider.LongStringSuffix"/> forwarding
// property preserves the existing Menu.cs API surface.
// ============================================================================================

internal static class BenchmarkStringSupport
{
    internal const int FixStrMaxLength = 31;

    internal static string LongStringSuffix = CharsetSuffixes.Latin1Long;

    private sealed class ReferenceComparer : IEqualityComparer<object>
    {
        public static readonly ReferenceComparer Instance = new();
        public new bool Equals(object? x, object? y) => ReferenceEquals(x, y);
        public int GetHashCode(object obj) => RuntimeHelpers.GetHashCode(obj);
    }

    internal static void EnsureAllStringsBypassFixStr(object? root)
    {
        if (root == null) return;

        var visited = new HashSet<object>(ReferenceComparer.Instance);
        var stack = new Stack<object>();
        stack.Push(root);

        while (stack.Count > 0)
        {
            var current = stack.Pop();
            if (!visited.Add(current)) continue;

            if (current is IEnumerable enumerable && current is not string)
            {
                foreach (var item in enumerable)
                {
                    if (item != null)
                        stack.Push(item);
                }
                continue;
            }

            var type = current.GetType();
            foreach (var property in type.GetProperties(BindingFlags.Instance | BindingFlags.Public))
            {
                if (!property.CanRead) continue;

                if (property.PropertyType == typeof(string))
                {
                    if (!property.CanWrite) continue;

                    var value = (string?)property.GetValue(current);
                    property.SetValue(current, ToLongString(value));
                    continue;
                }

                if (property.PropertyType.IsValueType || property.PropertyType.IsEnum)
                    continue;

                var child = property.GetValue(current);
                if (child != null)
                    stack.Push(child);
            }
        }
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    private static string ToLongString(string? value)
    {
        if (string.IsNullOrEmpty(value))
            return "Benchmark_String_Value" + LongStringSuffix;

        if (value.Length > FixStrMaxLength)
            return value;

        return value + LongStringSuffix;
    }
}

// ============================================================================================
// Generic test-data provider. One closing-generic alias per family — see
// <see cref="BenchmarkTestDataProvider"/> (the <c>_All_True</c> family, MSTEST-compatible name) and
// <see cref="BenchmarkTestDataProvider_All_False"/> (the <c>_All_False</c> family, Phase 1 benchmark
// target). The five cell-creator methods + ClearDeepLevelRefs are written once on the generic base,
// using the constrained <c>TestDataFactory&lt;TOrder, ...&gt;</c> for per-family element creation.
// ============================================================================================

public abstract class BenchmarkTestDataProvider<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>
    where TOrder : TestOrderBase<TItem, TTag, TUser, TCategory, TMetadata, TPreferences>, new()
    where TItem : TestOrderItemBase<TPallet, TTag, TUser, TMetadata, TOrder, TPreferences>, new()
    where TPallet : TestPalletBase<TMeasurement, TTag, TUser, TCategory, TMetadata, TItem, TPreferences>, new()
    where TMeasurement : TestMeasurementBase<TPoint, TTag, TUser, TPallet, TPreferences>, new()
    where TPoint : TestMeasurementPointBase<TTag, TUser, TMeasurement, TPreferences>, new()
    where TTag : SharedTagBase, new()
    where TUser : SharedUserBase<TPreferences>, new()
    where TCategory : SharedCategoryBase, new()
    where TMetadata : MetadataInfoBase<TMetadata>, new()
    where TPreferences : UserPreferencesBase, new()
{
    /// <summary>
    /// Active long-string suffix appended to short string properties during benchmark data construction.
    /// Forwards to <see cref="BenchmarkStringSupport.LongStringSuffix"/> (a non-generic shared field) so
    /// the setter is family-agnostic — both <c>BenchmarkTestDataProvider.LongStringSuffix = …</c> and
    /// <c>BenchmarkTestDataProvider_All_False.LongStringSuffix = …</c> route to the same backing value.
    /// Without this forwarding, a per-closed-generic static field on the base would store the suffix
    /// independently per family — the Menu setter would only affect whichever alias it addressed.
    /// </summary>
    public static string LongStringSuffix
    {
        get => BenchmarkStringSupport.LongStringSuffix;
        set => BenchmarkStringSupport.LongStringSuffix = value;
    }

    // Shortcut alias for the matching factory closing-generic. Saves typing the 10-param cluster
    // on every Create* call inside this class.
    private static class Factory
    {
        public static void ResetIdCounter() =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.ResetIdCounter();
        public static TTag CreateTag(string? name = null) =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.CreateTag(name);
        public static TUser CreateUser(string? username = null) =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.CreateUser(username);
        public static TCategory CreateCategory(string? name = null) =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.CreateCategory(name);
        public static TMetadata CreateMetadata(string? key = null, bool withChild = false) =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.CreateMetadata(key, withChild);
        public static TOrder CreateOrder(
            int itemCount, int palletsPerItem, int measurementsPerPallet, int pointsPerMeasurement,
            TTag? sharedTag = null, TUser? sharedUser = null, TMetadata? sharedMetadata = null,
            TPreferences? sharedPreferences = null, TCategory? sharedCategory = null) =>
            TestDataFactory<TOrder, TItem, TPallet, TMeasurement, TPoint, TTag, TUser, TCategory, TMetadata, TPreferences>.CreateOrder(
                itemCount, palletsPerItem, measurementsPerPallet, pointsPerMeasurement,
                sharedTag, sharedUser, sharedMetadata, sharedPreferences, sharedCategory);
    }

    public static List<TestDataSet> CreateTestDataSets(bool resetId = true)
    {
        return new List<TestDataSet>
        {
            CreateSmallTestData(resetId),
            CreateMediumTestData(resetId),
            CreateLargeTestData(resetId),
            CreateRepeatedStringsTestData(resetId),
            CreateDeepNestedTestData(resetId)
        };
    }

    private static TestDataSet<TOrder> CreateSmallTestData(bool resetId = true)
    {
        if (resetId) Factory.ResetIdCounter();

        var sharedTag = Factory.CreateTag("SharedTag");
        var sharedUser = Factory.CreateUser("shareduser");

        var order = Factory.CreateOrder(
            itemCount: 2,
            palletsPerItem: 2,
            measurementsPerPallet: 2,
            pointsPerMeasurement: 2,
            sharedTag: sharedTag,
            sharedUser: sharedUser);

        BenchmarkStringSupport.EnsureAllStringsBypassFixStr(order);

        ClearDeepLevelRefs(order);

        return new TestDataSet<TOrder>("Small (2x2x2x2)", order, iidRefPercent: 20);
    }

    private static TestDataSet<TOrder> CreateMediumTestData(bool resetId = true)
    {
        if (resetId) Factory.ResetIdCounter();

        var sharedTag = Factory.CreateTag("SharedTag");
        var sharedUser = Factory.CreateUser("shareduser");
        var sharedMeta = Factory.CreateMetadata("shared", withChild: true);

        var sharedPreferences = new TPreferences
        {
            Theme = "dark",
            Language = "hungarian",
            NotificationsEnabled = true,
            EmailDigestFrequency = "weekly"
        };
        sharedUser.Preferences = sharedPreferences;

        var order = Factory.CreateOrder(
            itemCount: 3,
            palletsPerItem: 3,
            measurementsPerPallet: 3,
            pointsPerMeasurement: 4,
            sharedTag: sharedTag,
            sharedUser: sharedUser,
            sharedMetadata: sharedMeta,
            sharedPreferences: sharedPreferences);

        BenchmarkStringSupport.EnsureAllStringsBypassFixStr(order);

        ClearDeepLevelRefs(order);

        return new TestDataSet<TOrder>("Medium (3x3x3x4)", order, iidRefPercent: 20);
    }

    private static TestDataSet<TOrder> CreateLargeTestData(bool resetId = true)
    {
        if (resetId) Factory.ResetIdCounter();

        var sharedTag = Factory.CreateTag("SharedTag");
        var sharedUser = Factory.CreateUser("shareduser");

        var sharedPreferences = new TPreferences
        {
            Theme = "light",
            Language = "german",
            NotificationsEnabled = false,
            EmailDigestFrequency = "daily"
        };
        sharedUser.Preferences = sharedPreferences;

        var order = Factory.CreateOrder(
            itemCount: 5,
            palletsPerItem: 5,
            measurementsPerPallet: 5,
            pointsPerMeasurement: 10,
            sharedTag: sharedTag,
            sharedUser: sharedUser,
            sharedPreferences: sharedPreferences);

        BenchmarkStringSupport.EnsureAllStringsBypassFixStr(order);

        ClearDeepLevelRefs(order);

        return new TestDataSet<TOrder>("Large (5x5x5x10)", order, iidRefPercent: 20);
    }

    private static TestDataSet<TOrder> CreateRepeatedStringsTestData(bool resetId = true)
    {
        if (resetId) Factory.ResetIdCounter();

        var sharedTag = Factory.CreateTag("RepeatedTag");
        var sharedUser = Factory.CreateUser("repeateduser");

        var sharedPreferences = new TPreferences
        {
            Theme = "dark",
            Language = "hungarian",
            NotificationsEnabled = true,
            EmailDigestFrequency = "weekly"
        };
        sharedUser.Preferences = sharedPreferences;

        var order = Factory.CreateOrder(
            itemCount: 10,
            palletsPerItem: 2,
            measurementsPerPallet: 2,
            pointsPerMeasurement: 2,
            sharedTag: sharedTag,
            sharedUser: sharedUser,
            sharedPreferences: sharedPreferences);

        // Repeated string fields — ProductName on items + PalletCode on pallets. Both are common
        // across the hierarchy, exercising string-interning deduplication on the Default preset
        // (which has UseStringInterning = All). Targeting ~20% repeated-string share overall.
        // Baselines are short ASCII (≤ FixStrMaxLength) so EnsureAllStringsBypassFixStr appends the
        // active CharsetSuffix — the resulting payload's UTF-8 content profile is governed entirely
        // by the selected charset (not contaminated by hard-coded Hungarian baseline values).
        foreach (var item in order.Items)
        {
            item.Status = TestStatus.Processing;
            item.ProductName = "ProductName";

            foreach (var pallet in item.Pallets)
            {
                pallet.PalletCode = "PalletCode";
            }
        }

        BenchmarkStringSupport.EnsureAllStringsBypassFixStr(order);

        ClearDeepLevelRefs(order);

        return new TestDataSet<TOrder>("Repeated Strings (10 items)", order, iidRefPercent: 20);
    }

    private static TestDataSet<TOrder> CreateDeepNestedTestData(bool resetId = true)
    {
        if (resetId) Factory.ResetIdCounter();

        var sharedTag = Factory.CreateTag("DeepTag");
        var sharedUser = Factory.CreateUser("deepuser");
        var sharedCategory = Factory.CreateCategory("DeepCategory");

        var sharedPreferences = new TPreferences
        {
            Theme = "light",
            Language = "french",
            NotificationsEnabled = false,
            EmailDigestFrequency = "monthly"
        };
        sharedUser.Preferences = sharedPreferences;

        var order = Factory.CreateOrder(
            itemCount: 2,
            palletsPerItem: 4,
            measurementsPerPallet: 4,
            pointsPerMeasurement: 8,
            sharedTag: sharedTag,
            sharedUser: sharedUser,
            sharedPreferences: sharedPreferences,
            sharedCategory: sharedCategory);

        BenchmarkStringSupport.EnsureAllStringsBypassFixStr(order);

        ClearDeepLevelRefs(order);

        return new TestDataSet<TOrder>("Deep Nested (2x4x4x8)", order, iidRefPercent: 20);
    }

    private static void ClearDeepLevelRefs(TOrder order)
    {
        // Keep shared IId refs at the pallet level (Tag + Inspector) — these contribute the bulk of
        // the ~20% IId-ref share that the test data targets. Only Category is cleared at this level
        // (one-of-three clears keep the share moderate). The deeper measurement / point levels are
        // cleared entirely so deep-tree ref noise does not skew the share upward beyond ~20%.
        foreach (var item in order.Items)
        {
            foreach (var pallet in item.Pallets)
            {
                // pallet.Tag = null;       // KEEP for ~20% IId-ref share (was cleared)
                // pallet.Inspector = null; // KEEP for ~20% IId-ref share (was cleared)
                pallet.Category = null;

                foreach (var measurement in pallet.Measurements)
                {
                    measurement.Tag = null;
                    measurement.Operator = null;

                    foreach (var point in measurement.Points)
                    {
                        point.Tag = null;
                        point.Verifier = null;
                    }
                }
            }
        }
    }
}

// ============================================================================================
// Closing-generic aliases for the provider. Same pattern as the factory: a bare-name class for
// MSTEST backward compatibility (kept on _All_True), and a _All_False suffix variant for the
// Phase 1 benchmark target. The static <c>LongStringSuffix</c> forwarding property lives on the
// generic base above — accessible identically through either alias (<c>BenchmarkTestDataProvider.LongStringSuffix</c>
// or <c>BenchmarkTestDataProvider_All_False.LongStringSuffix</c>), both routing to the same
// <see cref="BenchmarkStringSupport.LongStringSuffix"/> shared field. Symmetric API surface across
// families — no per-alias asymmetry.
// ============================================================================================

/// <summary>
/// <c>_All_True</c> family provider — preserves the bare-name API surface
/// (<c>BenchmarkTestDataProvider.CreateTestDataSets()</c>) that the SGen-vs-runtime compatibility
/// test depends on. <c>LongStringSuffix</c> is inherited from the generic base.
/// </summary>
public sealed class BenchmarkTestDataProvider : BenchmarkTestDataProvider<
    TestOrder_All_True, TestOrderItem_All_True, TestPallet_All_True, TestMeasurement_All_True, TestMeasurementPoint_All_True,
    SharedTag_All_True, SharedUser_All_True, SharedCategory_All_True, MetadataInfo_All_True, UserPreferences_All_True>
{
}

/// <summary>
/// <c>_All_False</c> family provider — Phase 1 benchmark target. Inherits the generic cell-creator
/// methods unchanged; the closed-generic <c>new TOrder()</c> calls inside the cell methods construct
/// <c>TestOrder_All_False</c> graphs.
/// </summary>
public sealed class BenchmarkTestDataProvider_All_False : BenchmarkTestDataProvider<
    TestOrder_All_False, TestOrderItem_All_False, TestPallet_All_False, TestMeasurement_All_False, TestMeasurementPoint_All_False,
    SharedTag_All_False, SharedUser_All_False, SharedCategory_All_False, MetadataInfo_All_False, UserPreferences_All_False>
{
}

// ============================================================================================
// TestDataSet — abstract metadata base + generic-ordered concrete. Orchestration code iterates
// over the base type (Name/DisplayName/TypeName/IIdRefPercent only); concrete consumers
// (CreateSerializers, Output binary-output dump) downcast to TestDataSet<TOrder> to access the
// typed Order.
// ============================================================================================

public abstract class TestDataSet
{
    public string Name { get; }

    /// <summary>
    /// Percentage of IId shared references in the data (0-100).
    /// Higher values mean more deduplication benefit for Default mode.
    /// </summary>
    public int IIdRefPercent { get; }

    // Type-keyed variant registry. Phase 2 multi-variant dispatch: AcBinary's options preset
    // decides which variant graph it serializes (FastMode → _All_False, Default → _All_True),
    // while MemPack/MsgPack canonically use one (typically _All_True). The cells build all
    // known variants upfront and register them here so CreateSerializers can hand each benchmark
    // its matching graph instance.
    private readonly Dictionary<Type, object> _variants = new();

    protected TestDataSet(string name, int iidRefPercent)
    {
        Name = name;
        IIdRefPercent = iidRefPercent;
    }

    public abstract string TypeName { get; }

    /// <summary>
    /// Gets display name including IId ref percentage if set.
    /// </summary>
    public string DisplayName => IIdRefPercent > 0
        ? $"{Name} [{IIdRefPercent}% IId refs]"
        : Name;

    /// <summary>
    /// Register a variant graph for this cell. Called by builders. Idempotent on the same type
    /// (last-write-wins, no error) so an alias's primary registration is harmless even if
    /// cross-registration adds the same variant later.
    /// </summary>
    public void RegisterVariant<T>(T variant) where T : class => _variants[typeof(T)] = variant;

    /// <summary>
    /// Get a registered variant by type. Throws <see cref="InvalidOperationException"/> if not
    /// registered — fail-fast surfaces a mismatch between the variant a benchmark expects and
    /// what the cell-builder populated.
    /// </summary>
    public T GetOrder<T>() where T : class
    {
        if (_variants.TryGetValue(typeof(T), out var v)) return (T)v;
        throw new InvalidOperationException($"Variant '{typeof(T).Name}' not registered for cell '{Name}' (registered: {string.Join(", ", _variants.Keys.Select(k => k.Name))})");
    }

    /// <summary>
    /// Check whether a variant is registered. Use to gate optional benchmarks that may not have
    /// their variant prepared in every cell.
    /// </summary>
    public bool HasOrder<T>() where T : class => _variants.ContainsKey(typeof(T));
}

public sealed class TestDataSet<TOrder> : TestDataSet
    where TOrder : class
{
    public TOrder Order { get; }

    public TestDataSet(string name, TOrder order, int iidRefPercent = 0)
        : base(name, iidRefPercent)
    {
        Order = order;
        RegisterVariant(order); // primary registers itself
    }

    public override string TypeName => Order.GetType().Name;
}