MycLib/Src/Myc.Data.Series.pas

unit Myc.Data.Series;

interface

type
    TChunkArray<T> = record
    private
        const
            ChunkSize = 1024;
        type
            TChunk = TArray<T>;
    private
        FChunks: TArray<TChunk>;
        FOffset: Integer;
        FCount: Integer;
        function GetItems(Idx: Integer): T; inline;

    public
        constructor Create(const AChunks: TArray<TChunk>; ACount: Integer; AOffset: Integer);
        class operator Initialize(out Dest: TChunkArray<T>);

        // Add a single item without creating a temporary array.
        procedure Add(const Data: T; MaxCount: Integer); overload;

        // Add items, but ensure that the result has MaxCount items.
        procedure Add(const Data: array of T; MaxCount: Integer); overload;

        class function CreateFromArray(const AData: TArray<T>; MaxCount: Integer): TChunkArray<T>; static;

        // Creates a deep copy of the array, optionally truncating it to the last MaxCount items.
        function Copy(MaxCount: Integer = -1): TChunkArray<T>;

        property Count: Integer read FCount;
        property Items[Idx: Integer]: T read GetItems; default;
    end;

    // A series where the last added item has index 0.
    TSeries<T> = record
    private
        FArray: TChunkArray<T>;
        FTotalCount: Int64;
        function GetCount: Integer;
        function GetItems(Idx: Integer): T;

    public
        constructor Create(const AArray: TChunkArray<T>; ATotalCount: Int64);
        class operator Initialize(out Dest: TSeries<T>);

        // Creates a deep copy of the series, optionally truncating it to the last Lookback items.
        function Copy(Lookback: Integer = -1): TSeries<T>;

        procedure Add(const Data: T; Lookback: Int64 = -1); overload;
        procedure Add(const Data: array of T; Lookback: Integer); overload;

        class function CreateFromArray(const AData: TArray<T>; Lookback: Integer): TSeries<T>; static;

        property Count: Integer read GetCount;
        property TotalCount: Int64 read FTotalCount;
        property Items[Idx: Integer]: T read GetItems; default;
    end;

implementation

uses
    System.Generics.Collections,
    System.Math;

{ TChunkArray<T> }

constructor TChunkArray<T>.Create(const AChunks: TArray<TChunk>; ACount, AOffset: Integer);
begin
    FChunks := AChunks;
    FCount := ACount;
    FOffset := AOffset;
    Assert(FOffset < ChunkSize);
end;

procedure TChunkArray<T>.Add(const Data: T; MaxCount: Integer);
var
    newCount, totalNewOffset: Integer;
    oldLogicalEnd, newLogicalEnd, oldStartChunk, newStartChunk, oldEndChunk, newEndChunk, chunksToRemove, chunksToAdd: Integer;
begin
    if MaxCount < 0 then
        MaxCount := MaxInt;

    if MaxCount <= 0 then
        exit;

    // 1. Calculate the new logical state (numToAdd is always 1)
    newCount := Min(FCount + 1, MaxCount);
    totalNewOffset := FOffset + FCount + 1 - newCount;

    // 2. Determine if the underlying FChunks array needs restructuring
    oldLogicalEnd := FOffset + FCount - 1;
    newLogicalEnd := totalNewOffset + newCount - 1;

    oldStartChunk := FOffset div ChunkSize;
    newStartChunk := totalNewOffset div ChunkSize;

    oldEndChunk := -1;
    if FCount > 0 then
        oldEndChunk := oldLogicalEnd div ChunkSize;

    newEndChunk := -1;
    if newCount > 0 then
        newEndChunk := newLogicalEnd div ChunkSize;

    chunksToRemove := newStartChunk - oldStartChunk;
    chunksToAdd := newEndChunk - oldEndChunk;

    // 3. Perform the add operation
    if (chunksToRemove > 0) or (chunksToAdd > 0) then
    begin
        // --- Rebuild Path: Chunks must be added or removed ---
        var newChunks: TArray<TChunk>;
        var oldChunkCount := Length(FChunks);
        var newChunkCount := oldChunkCount - chunksToRemove + chunksToAdd;
        SetLength(newChunks, newChunkCount);

        var numChunksToKeep := oldChunkCount - chunksToRemove;
        if numChunksToKeep > 0 then
            TArray.Copy<TChunk>(FChunks, newChunks, chunksToRemove, 0, numChunksToKeep);

        for var i := numChunksToKeep to newChunkCount - 1 do
            SetLength(newChunks[i], ChunkSize);

        FChunks := newChunks;

        // Copy the new data item into the rebuilt chunks
        var writeStartLogicalIdx := FOffset + FCount;
        var physicalWriteStartIdx := writeStartLogicalIdx - (chunksToRemove * ChunkSize);
        var chunkIdx := physicalWriteStartIdx div ChunkSize;
        var idxInChunk := physicalWriteStartIdx mod ChunkSize;

        Assert(chunkIdx < Length(FChunks), 'Chunk index out of bounds during rebuild');
        FChunks[chunkIdx][idxInChunk] := Data;
    end
    else
    begin
        // --- Fast Path: Window slides within existing chunk allocation ---
        var writeStartLogicalIdx := FOffset + FCount;
        var chunkIdx := writeStartLogicalIdx div ChunkSize;
        var idxInChunk := writeStartLogicalIdx mod ChunkSize;

        Assert(chunkIdx < Length(FChunks), 'Chunk index out of bounds on fast path');
        if Length(FChunks[chunkIdx]) = 0 then // Safeguard
            SetLength(FChunks[chunkIdx], ChunkSize);

        FChunks[chunkIdx][idxInChunk] := Data;
    end;

    // 4. Finalize the new state
    FOffset := totalNewOffset mod ChunkSize;
    FCount := newCount;
end;

class function TChunkArray<T>.CreateFromArray(const AData: TArray<T>; MaxCount: Integer): TChunkArray<T>;
begin
    Result.Add(AData, MaxCount);
end;

procedure TChunkArray<T>.Add(const Data: array of T; MaxCount: Integer);
var
    dataLen, dataReadOffset, numToAdd, newCount, totalNewOffset: Integer;
    oldLogicalEnd, newLogicalEnd, oldStartChunk, newStartChunk, oldEndChunk, newEndChunk, chunksToRemove, chunksToAdd: Integer;
begin
    if MaxCount < 0 then
        MaxCount := MaxInt;

    dataLen := Length(Data);
    if (dataLen = 0) or (MaxCount <= 0) then
        exit;

    // 1. Trim input data if it's larger than MaxCount
    dataReadOffset := 0;
    if dataLen > MaxCount then
        dataReadOffset := dataLen - MaxCount;
    numToAdd := dataLen - dataReadOffset;

    // 2. Calculate the new logical state (total offset and count)
    newCount := Min(FCount + numToAdd, MaxCount);
    // Total logical offset of the new window's start
    totalNewOffset := FOffset + FCount + numToAdd - newCount;

    // 3. Determine if the underlying FChunks array needs restructuring
    oldLogicalEnd := FOffset + FCount - 1;
    newLogicalEnd := totalNewOffset + newCount - 1;

    oldStartChunk := FOffset div ChunkSize;
    newStartChunk := totalNewOffset div ChunkSize;

    oldEndChunk := -1;
    if FCount > 0 then
        oldEndChunk := oldLogicalEnd div ChunkSize;

    newEndChunk := -1;
    if newCount > 0 then
        newEndChunk := newLogicalEnd div ChunkSize;

    chunksToRemove := newStartChunk - oldStartChunk;
    chunksToAdd := newEndChunk - oldEndChunk;

    // 4. Perform the add operation
    if (chunksToRemove > 0) or (chunksToAdd > 0) then
    begin
        // --- Rebuild Path: Chunks must be added or removed ---
        var newChunks: TArray<TChunk>;
        var oldChunkCount := Length(FChunks);
        var newChunkCount := oldChunkCount - chunksToRemove + chunksToAdd;
        SetLength(newChunks, newChunkCount);

        // Copy references to the chunks that are kept
        var numChunksToKeep := oldChunkCount - chunksToRemove;
        if numChunksToKeep > 0 then
            TArray.Copy<TChunk>(FChunks, newChunks, chunksToRemove, 0, numChunksToKeep);

        // Allocate new chunks at the end
        for var i := numChunksToKeep to newChunkCount - 1 do
            SetLength(newChunks[i], ChunkSize);

        FChunks := newChunks;

        // Copy the new data into the rebuilt chunks
        var writeStartLogicalIdx := FOffset + FCount;
        var physicalWriteStartIdx := writeStartLogicalIdx - (chunksToRemove * ChunkSize);
        var remainingToAdd := numToAdd;
        var currentDataOffset := dataReadOffset;

        while remainingToAdd > 0 do
        begin
            var chunkIdx := physicalWriteStartIdx div ChunkSize;
            var idxInChunk := physicalWriteStartIdx mod ChunkSize;
            var spaceInChunk := ChunkSize - idxInChunk;
            var countToCopy := Min(remainingToAdd, spaceInChunk);

            Assert(chunkIdx < Length(FChunks), 'Chunk index out of bounds during rebuild');
            TArray.Copy<T>(Data, FChunks[chunkIdx], currentDataOffset, idxInChunk, countToCopy);

            inc(physicalWriteStartIdx, countToCopy);
            inc(currentDataOffset, countToCopy);
            dec(remainingToAdd, countToCopy);
        end;
    end
    else
    begin
        // --- Fast Path: Window slides within existing chunk allocation ---
        var writeStartLogicalIdx := FOffset + FCount;
        var remainingToAdd := numToAdd;
        var currentDataOffset := dataReadOffset;

        while remainingToAdd > 0 do
        begin
            var chunkIdx := writeStartLogicalIdx div ChunkSize;
            var idxInChunk := writeStartLogicalIdx mod ChunkSize;
            var spaceInChunk := ChunkSize - idxInChunk;
            var countToCopy := Min(remainingToAdd, spaceInChunk);

            Assert(chunkIdx < Length(FChunks), 'Chunk index out of bounds on fast path');
            if Length(FChunks[chunkIdx]) = 0 then // Should not happen on fast path, but as a safeguard
                SetLength(FChunks[chunkIdx], ChunkSize);

            TArray.Copy<T>(Data, FChunks[chunkIdx], currentDataOffset, idxInChunk, countToCopy);

            inc(writeStartLogicalIdx, countToCopy);
            inc(currentDataOffset, countToCopy);
            dec(remainingToAdd, countToCopy);
        end;
    end;

    // 5. Finalize the new state
    FOffset := totalNewOffset mod ChunkSize;
    FCount := newCount;
end;

function TChunkArray<T>.GetItems(Idx: Integer): T;
begin
    // Convert logical index to physical index inside chunks
    var physicalIdx := Idx + FOffset;
    Result := FChunks[physicalIdx div ChunkSize][physicalIdx mod ChunkSize];
end;

function TChunkArray<T>.Copy(MaxCount: Integer): TChunkArray<T>;
var
    effectiveCount: Integer;
    logicalStartIndex: Integer;
    physicalStartIndex, physicalEndIndex: Integer;
    startChunkIdx, endChunkIdx: Integer;
    newOffset: Integer;
    newChunkCount: Integer;
    i: Integer;
begin
    // 1. Determine the number of items to copy.
    effectiveCount := FCount;
    if (MaxCount >= 0) and (MaxCount < FCount) then
        effectiveCount := MaxCount;

    if effectiveCount = 0 then
    begin
        // Return an empty array.
        Result := Default(TChunkArray<T>);
        exit;
    end;

    // 2. Calculate physical location of the data to be copied.
    logicalStartIndex := FCount - effectiveCount;
    physicalStartIndex := FOffset + logicalStartIndex;
    physicalEndIndex := physicalStartIndex + effectiveCount - 1;
    startChunkIdx := physicalStartIndex div ChunkSize;
    endChunkIdx := physicalEndIndex div ChunkSize;
    newOffset := physicalStartIndex mod ChunkSize;

    // 3. Create the new chunk array for the result.
    newChunkCount := endChunkIdx - startChunkIdx + 1;
    SetLength(Result.FChunks, newChunkCount);

    // 4. Copy chunks. Boundary chunks are deep-copied, internal chunks are shared.
    if newChunkCount = 1 then
    begin
        // All data is in a single chunk, which is both a start and end boundary.
        // It must always be a deep copy.
        Result.FChunks[0] := System.Copy(FChunks[startChunkIdx]);
    end
    else
    begin
        // First chunk (start boundary) must be a deep copy.
        Result.FChunks[0] := System.Copy(FChunks[startChunkIdx]);

        // Middle, immutable chunks can be shared by reference.
        // This loop only runs if there are 3 or more chunks in the copy.
        for i := 1 to newChunkCount - 2 do
        begin
            Result.FChunks[i] := FChunks[startChunkIdx + i];
        end;

        // Last chunk (end boundary) must be a deep copy.
        Result.FChunks[newChunkCount - 1] := System.Copy(FChunks[endChunkIdx]);
    end;

    // 5. Finalize the new TChunkArray state.
    Result.FOffset := newOffset;
    Result.FCount := effectiveCount;
end;

class operator TChunkArray<T>.Initialize(out Dest: TChunkArray<T>);
begin
    Dest.FOffset := 0;
    Dest.FCount := 0;
end;

{ TSeries<T> }

procedure TSeries<T>.Add(const Data: T; Lookback: Int64);
begin
    FArray.Add(Data, Lookback);
    inc(FTotalCount);
end;

procedure TSeries<T>.Add(const Data: array of T; Lookback: Integer);
begin
    FArray.Add(Data, Lookback);
    inc(FTotalCount, Length(Data));
end;

constructor TSeries<T>.Create(const AArray: TChunkArray<T>; ATotalCount: Int64);
begin
    FArray := AArray;
    FTotalCount := ATotalCount;
end;

function TSeries<T>.Copy(Lookback: Integer): TSeries<T>;
begin
    // Create a copy of the underlying chunk array, optionally truncating it.
    Result.FArray := FArray.Copy(Lookback);
    // The new series' total count is the number of items it actually contains.
    Result.FTotalCount := Result.FArray.Count;
end;

class function TSeries<T>.CreateFromArray(const AData: TArray<T>; Lookback: Integer): TSeries<T>;
begin
    Result.FArray.Add(AData, Lookback);
    Result.FTotalCount := Length(AData);
end;

function TSeries<T>.GetCount: Integer;
begin
    Result := FArray.Count;
end;

function TSeries<T>.GetItems(Idx: Integer): T;
begin
    // Access is reversed: series index 0 is the last element in the underlying array
    Assert((Idx >= 0) and (Idx < FArray.Count), 'Index out of bounds');
    Result := FArray.Items[FArray.Count - 1 - Idx];
end;

class operator TSeries<T>.Initialize(out Dest: TSeries<T>);
begin
    Dest.FTotalCount := 0;
end;

end.