MycLib/Src/Data/Myc.Data.Scalar.pas

unit Myc.Data.Scalar;

interface

uses
    System.SysUtils,
    Myc.Data.Decimal,
    Myc.Data.Series;

type
    // POD: Plain old data (...that fits into 64 bits)

    TScalarKind = (
        skInteger,
        skInt64,
        skUInt64,
        skSingle,
        skDouble,
        skDateTime,
        skTimestamp,
        skBoolean,
        skChar,
        skPChar,
        skString,
        skBytes,
        skDecimal
    );

    TScalarBytes = array[0..7] of Byte;
    TScalarPChar = array[0..3] of Char;
    TScalarString = String[7];

    TScalarValue = record
    public
        // Factory methods for creating a TScalarValue without a kind.
        class function FromInteger(AValue: Integer): TScalarValue; static; inline;
        class function FromInt64(AValue: Int64): TScalarValue; static; inline;
        class function FromUInt64(AValue: UInt64): TScalarValue; static; inline;
        class function FromSingle(AValue: Single): TScalarValue; static; inline;
        class function FromDouble(AValue: Double): TScalarValue; static; inline;
        class function FromDateTime(AValue: TDateTime): TScalarValue; static; inline;
        class function FromTimestamp(AValue: TTimestamp): TScalarValue; static; inline;
        class function FromBoolean(AValue: Boolean): TScalarValue; static; inline;
        class function FromDecimal(AValue: TDecimal): TScalarValue; static; inline;
        class function FromChar(AValue: Char): TScalarValue; static; inline;
        class function FromPChar(const AValue: String): TScalarValue; overload; static; inline;
        class function FromPChar(const AValue: TScalarPChar): TScalarValue; overload; static; inline;
        class function FromString(const AValue: String): TScalarValue; overload; static; inline;
        class function FromString(const AValue: TScalarString): TScalarValue; overload; static; inline;
        class function FromBytes(const AValue: TScalarBytes): TScalarValue; static; inline;

    // Direct field access for performance.
    case TScalarKind of
        skInteger: (AsInteger: Integer);
        skInt64: (AsInt64: Int64);
        skUInt64: (AsUInt64: UInt64);
        skSingle: (AsSingle: Single);
        skDouble: (AsDouble: Double);
        skDateTime: (AsDateTime: TDateTime);
        skTimestamp: (AsTimestamp: TTimestamp);
        skBoolean: (AsBoolean: Boolean);
        skChar: (AsChar: Char);
        skPChar: (AsPChar: TScalarPChar);
        skString: (AsString: String[7]);
        skBytes: (AsBytes: TScalarBytes);
        skDecimal: (AsDecimal: TDecimal);
    end;

    // A scalar value with a type identifier.
    TScalar = record
    public
        Kind: TScalarKind;
        Value: TScalarValue;

        // Creates a new scalar from a kind and a value.
        constructor Create(AKind: TScalarKind; const AValue: TScalarValue);

        // Factory methods for specific scalar types.
        class function FromInteger(AValue: Integer): TScalar; static; inline;
        class function FromInt64(AValue: Int64): TScalar; static; inline;
        class function FromUInt64(AValue: UInt64): TScalar; static; inline;
        class function FromSingle(AValue: Single): TScalar; static; inline;
        class function FromDouble(AValue: Double): TScalar; static; inline;
        class function FromDateTime(AValue: TDateTime): TScalar; static; inline;
        class function FromTimestamp(AValue: TTimestamp): TScalar; static; inline;
        class function FromBoolean(AValue: Boolean): TScalar; static; inline;
        class function FromChar(AValue: Char): TScalar; static; inline;
        class function FromPChar(const AValue: TScalarPChar): TScalar; static; inline;
        class function FromString(const AValue: String): TScalar; static; inline;
        class function FromBytes(const AValue: TScalarBytes): TScalar; static; inline;
        class function FromDecimal(const AValue: TDecimal): TScalar; static; inline;

        class function StringToKind(const AName: string): TScalarKind; static;

        function ToString: String;
    end;

    TScalarKindHelper = record helper for TScalarKind
    public
        function ToString: string;
    end;

    // Basic data structures using the scalar type (these are not POD of course)

    // An array of scalar values of the same kind.
    TScalarArray = record
    public
        Kind: TScalarKind;
        Items: TArray<TScalarValue>;

        // Creates a new scalar array.
        constructor Create(AKind: TScalarKind; const AItems: TArray<TScalarValue>);
    end;

    TScalarTuple = TArray<TScalar>;

    // A field definition for a scalar record.
    TScalarRecordField = record
        Name: String;
        Kind: TScalarKind;
        constructor Create(const AName: String; AKind: TScalarKind);
    end;

    TScalarRecordDefinition = record
    private
        FFields: TArray<TScalarRecordField>;
    public
        constructor Create(const AFields: TArray<TScalarRecordField>);
        function IndexOf(const Name: String): Integer;
        property Fields: TArray<TScalarRecordField> read FFields;
    end;

    // A record of scalar values, based on a definition.
    TScalarRecord = record
    public
        // Creates a new scalar record.
        constructor Create(const ADef: TScalarRecordDefinition; const AFields: TArray<TScalarValue>);

        function GetDef: TScalarRecordDefinition; inline;
        function GetFields: TArray<TScalarValue>; inline;
        function GetItems(const Name: String): TScalar;

        property Def: TScalarRecordDefinition read GetDef;
        property Fields: TArray<TScalarValue> read GetFields;
        property Items[const Name: String]: TScalar read GetItems; default;

    strict private
        FDef: TScalarRecordDefinition;
        FFields: TArray<TScalarValue>;
    end;

    // A time series of scalar values of the same kind.
    TScalarSeries = record
    private
        FKind: TScalarKind;
        FItems: TSeries<TScalarValue>;
    public
        // Creates a new scalar series.
        constructor Create(AKind: TScalarKind; const AItems: TSeries<TScalarValue>);

        property Kind: TScalarKind read FKind;
        property Items: TSeries<TScalarValue> read FItems;
    end;

    TScalarMemberSeries = record
    private
        FArray: TChunkArray<TScalarValue>;
        FElementIdx: Integer;
        FKind: TScalarKind;
        FElemCount: Integer;
        function GetCount: Int64;
        function GetItems(Idx: Integer): TScalar;
    public
        constructor Create(AKind: TScalarKind; const AArray: TChunkArray<TScalarValue>; AElementIdx, AElemCount: Integer);
        property Count: Int64 read GetCount;
        property Items[Idx: Integer]: TScalar read GetItems; default;
        property Kind: TScalarKind read FKind;
    end;

    // A time series of scalar tuples, optimized for memory and access speed.
    TScalarTupleSeries = record
    private
        FDef: TArray<TScalarKind>;
        FArray: TChunkArray<TScalarValue>;
        FTotalCount: Int64;
        function GetCount: Int64;
        function GetItems(Idx: Integer): TScalarTuple;
    public
        constructor Create(const ADef: TArray<TScalarKind>);
        procedure Add(const Item: TScalarTuple; Lookback: Int64 = -1);
        function CreateMemberSeries(Idx: Integer): TScalarMemberSeries;
        property Count: Int64 read GetCount;
        property Def: TArray<TScalarKind> read FDef;
        property Items[Idx: Integer]: TScalarTuple read GetItems; default;
        property TotalCount: Int64 read FTotalCount;
    end;

    // A time series of scalar records, optimized for memory and access speed.
    TScalarRecordSeries = record
    private
        FDef: TScalarRecordDefinition;
        FArray: TChunkArray<TScalarValue>;
        FTotalCount: Int64;
        function GetCount: Int64;
        function GetDef: TScalarRecordDefinition; inline;
        function GetItems(Idx: Integer): TScalarRecord;
    public
        constructor Create(const ADef: TScalarRecordDefinition);
        procedure Add(const Item: TScalarRecord; Lookback: Int64 = -1);
        function AsTupleSeries: TScalarTupleSeries;
        function CreateMemberSeries(const Field: String): TScalarMemberSeries;
        property Count: Int64 read GetCount;
        property Def: TScalarRecordDefinition read GetDef;
        property Items[Idx: Integer]: TScalarRecord read GetItems; default;
        property TotalCount: Int64 read FTotalCount;
    end;

implementation

{ TScalarValue }

class function TScalarValue.FromBoolean(AValue: Boolean): TScalarValue;
begin
    Result.AsBoolean := AValue;
end;

class function TScalarValue.FromBytes(const AValue: TScalarBytes): TScalarValue;
begin
    Result.AsBytes := AValue;
end;

class function TScalarValue.FromChar(AValue: Char): TScalarValue;
begin
    Result.AsChar := AValue;
end;

class function TScalarValue.FromDateTime(AValue: TDateTime): TScalarValue;
begin
    Result.AsDateTime := AValue;
end;

class function TScalarValue.FromDecimal(AValue: TDecimal): TScalarValue;
begin
    Result.AsDecimal := AValue;
end;

class function TScalarValue.FromDouble(AValue: Double): TScalarValue;
begin
    Result.AsDouble := AValue;
end;

class function TScalarValue.FromInt64(AValue: Int64): TScalarValue;
begin
    Result.AsInt64 := AValue;
end;

class function TScalarValue.FromInteger(AValue: Integer): TScalarValue;
begin
    Result.AsInteger := AValue;
end;

class function TScalarValue.FromPChar(const AValue: String): TScalarValue;
var
    len, i: Integer;
begin
    FillChar(Result.AsPChar, SizeOf(Result.AsPChar), #0);
    len := Length(AValue);
    if (len > 0) then
    begin
        if (len > Length(Result.AsPChar)) then
            len := Length(Result.AsPChar);
        for i := 0 to len - 1 do
            Result.AsPChar[i] := AValue[i + 1];
    end;
end;

class function TScalarValue.FromPChar(const AValue: TScalarPChar): TScalarValue;
begin
    Result.AsPChar := AValue;
end;

class function TScalarValue.FromSingle(AValue: Single): TScalarValue;
begin
    Result.AsSingle := AValue;
end;

class function TScalarValue.FromString(const AValue: String): TScalarValue;
begin
    Result.AsString := ShortString(AValue);
end;

class function TScalarValue.FromString(const AValue: TScalarString): TScalarValue;
begin
    Result.AsString := AValue;
end;

class function TScalarValue.FromTimestamp(AValue: TTimestamp): TScalarValue;
begin
    Result.AsTimestamp := AValue;
end;

class function TScalarValue.FromUInt64(AValue: UInt64): TScalarValue;
begin
    Result.AsUInt64 := AValue;
end;

{ TScalar }

constructor TScalar.Create(AKind: TScalarKind; const AValue: TScalarValue);
begin
    Kind := AKind;
    Value := AValue;
end;

class function TScalar.FromBoolean(AValue: Boolean): TScalar;
begin
    Result.Kind := skBoolean;
    Result.Value := TScalarValue.FromBoolean(AValue);
end;

class function TScalar.FromBytes(const AValue: TScalarBytes): TScalar;
begin
    Result.Kind := skBytes;
    Result.Value := TScalarValue.FromBytes(AValue);
end;

class function TScalar.FromChar(AValue: Char): TScalar;
begin
    Result.Kind := skChar;
    Result.Value := TScalarValue.FromChar(AValue);
end;

class function TScalar.FromDateTime(AValue: TDateTime): TScalar;
begin
    Result.Kind := skDateTime;
    Result.Value := TScalarValue.FromDateTime(AValue);
end;

class function TScalar.FromDecimal(const AValue: TDecimal): TScalar;
begin
    Result.Kind := skDecimal;
    Result.Value := TScalarValue.FromDecimal(AValue);
end;

class function TScalar.FromDouble(AValue: Double): TScalar;
begin
    Result.Kind := skDouble;
    Result.Value := TScalarValue.FromDouble(AValue);
end;

class function TScalar.FromInt64(AValue: Int64): TScalar;
begin
    Result.Kind := skInt64;
    Result.Value := TScalarValue.FromInt64(AValue);
end;

class function TScalar.FromInteger(AValue: Integer): TScalar;
begin
    Result.Kind := skInteger;
    Result.Value := TScalarValue.FromInteger(AValue);
end;

class function TScalar.FromPChar(const AValue: TScalarPChar): TScalar;
begin
    Result.Kind := skPChar;
    Result.Value := TScalarValue.FromPChar(AValue);
end;

class function TScalar.FromSingle(AValue: Single): TScalar;
begin
    Result.Kind := skSingle;
    Result.Value := TScalarValue.FromSingle(AValue);
end;

class function TScalar.FromString(const AValue: String): TScalar;
begin
    Result.Kind := skString;
    Result.Value := TScalarValue.FromString(AValue);
end;

class function TScalar.FromTimestamp(AValue: TTimestamp): TScalar;
begin
    Result.Kind := skTimestamp;
    Result.Value := TScalarValue.FromTimestamp(AValue);
end;

class function TScalar.FromUInt64(AValue: UInt64): TScalar;
begin
    Result.Kind := skUInt64;
    Result.Value := TScalarValue.FromUInt64(AValue);
end;

class function TScalar.StringToKind(const AName: string): TScalarKind;
begin
    if SameText(AName, 'integer') then
        Result := skInteger
    else if SameText(AName, 'int64') then
        Result := skInt64
    else if SameText(AName, 'uint64') then
        Result := skUInt64
    else if SameText(AName, 'single') then
        Result := skSingle
    else if SameText(AName, 'double') then
        Result := skDouble
    else if SameText(AName, 'datetime') then
        Result := skDateTime
    else if SameText(AName, 'timestamp') then
        Result := skTimestamp
    else if SameText(AName, 'boolean') then
        Result := skBoolean
    else if SameText(AName, 'char') then
        Result := skChar
    else if SameText(AName, 'pchar') then
        Result := skPChar
    else if SameText(AName, 'string') then
        Result := skString
    else if SameText(AName, 'bytes') then
        Result := skBytes
    else if SameText(AName, 'decimal') then
        Result := skDecimal
    else
        raise EArgumentException.CreateFmt('Unknown scalar type name: "%s"', [AName]);
end;

function TScalar.ToString: String;
begin
    case Kind of
        skInteger: Result := IntToStr(Value.AsInteger);
        skInt64: Result := IntToStr(Value.AsInt64);
        skUInt64: Result := UIntToStr(Value.AsUInt64);
        skSingle: Result := FloatToStr(Value.AsSingle);
        skDouble: Result := FloatToStr(Value.AsDouble);
        skDateTime: Result := DateTimeToStr(Value.AsDateTime);
        skTimestamp: Result := FormatDateTime('yyyy-mm-dd hh:nn:ss.zzz', TimeStampToDateTime(Value.AsTimestamp));
        skBoolean: Result := BoolToStr(Value.AsBoolean, True);
        skChar: Result := '''' + Value.AsChar + '''';
        skPChar: Result := PChar(@Value.AsPChar); // Needs explicit cast/pointer
        skString: Result := '''' + string(Value.AsString) + '''';
        skBytes: Result := '(Bytes)';
        skDecimal: Result := FloatToStr(Double(Value.AsDecimal));
    else
        Result := '[Unknown Scalar]';
    end;
end;

{ TScalarArray }

constructor TScalarArray.Create(AKind: TScalarKind; const AItems: TArray<TScalarValue>);
begin
    Kind := AKind;
    Items := AItems;
end;

{ TScalarRecordField }

constructor TScalarRecordField.Create(const AName: String; AKind: TScalarKind);
begin
    Name := AName;
    Kind := AKind;
end;

{ TScalarRecord }

constructor TScalarRecord.Create(const ADef: TScalarRecordDefinition; const AFields: TArray<TScalarValue>);
begin
    Assert(Length(ADef.Fields) = Length(AFields), 'Field definition and value count must match.');
    FDef := ADef;
    FFields := AFields;
end;

function TScalarRecord.GetDef: TScalarRecordDefinition;
begin
    Result := FDef;
end;

function TScalarRecord.GetFields: TArray<TScalarValue>;
begin
    Result := FFields;
end;

function TScalarRecord.GetItems(const Name: String): TScalar;
var
    idx: Integer;
begin
    idx := FDef.IndexOf(Name);
    if idx < 0 then
        raise EArgumentException.CreateFmt('Field "%s" not found in record definition.', [Name]);

    Result.Create(FDef.Fields[idx].Kind, FFields[idx]);
end;

{ TScalarSeries }

constructor TScalarSeries.Create(AKind: TScalarKind; const AItems: TSeries<TScalarValue>);
begin
    FKind := AKind;
    FItems := AItems;
end;

{ TScalarTupleSeries }

constructor TScalarTupleSeries.Create(const ADef: TArray<TScalarKind>);
begin
    // Implements a time series of scalar tuples using a chunk array for efficient storage.
    Assert(Length(ADef) > 0, 'Tuple definition cannot be empty.');
    FDef := ADef;
    FTotalCount := 0;
end;

procedure TScalarTupleSeries.Add(const Item: TScalarTuple; Lookback: Int64 = -1);
var
    values: TArray<TScalarValue>;
    i: Integer;
    tupleSize: Integer;
    maxCount: Integer;
begin
    tupleSize := Length(FDef);
    Assert(Length(Item) = tupleSize, 'Tuple does not match series definition.');

    // Extract TScalarValue from TScalarTuple.
    SetLength(values, tupleSize);
    for i := 0 to tupleSize - 1 do
    begin
        Assert(Item[i].Kind = FDef[i], 'Tuple element kind mismatch.');
        values[i] := Item[i].Value;
    end;

    if Lookback < 0 then
        maxCount := -1
    else
        maxCount := tupleSize * Lookback;

    FArray.Add(values, maxCount);
    inc(FTotalCount);
end;

function TScalarTupleSeries.CreateMemberSeries(Idx: Integer): TScalarMemberSeries;
var
    tupleSize: Integer;
begin
    tupleSize := Length(FDef);
    if (Idx < 0) or (Idx >= tupleSize) then
        raise EArgumentException.CreateFmt('Index %d is out of bounds for a tuple of size %d.', [Idx, tupleSize]);

    // Creates a series view for a specific member of the tuple.
    Result := TScalarMemberSeries.Create(FDef[Idx], FArray, Idx, tupleSize);
end;

function TScalarTupleSeries.GetCount: Int64;
begin
    if Length(FDef) = 0 then
        exit(0);
    Result := FArray.Count div Length(FDef);
end;

function TScalarTupleSeries.GetItems(Idx: Integer): TScalarTuple;
var
    values: TArray<TScalarValue>;
    tupleSize, i: Integer;
begin
    tupleSize := Length(FDef);
    Assert(tupleSize > 0);
    Assert((Idx >= 0) and (Idx < (FArray.Count div tupleSize)));

    // Create a temporary array to hold the values for one tuple.
    SetLength(values, tupleSize);
    // Copy the tuple data from the chunk array (latest item is at the end).
    Move(FArray.ItemRef[(FArray.Count - tupleSize) - (Idx * tupleSize)]^, values[0], sizeof(TScalarValue) * tupleSize);

    // Reconstruct the TScalarTuple from the values and the definition.
    SetLength(Result, tupleSize);
    for i := 0 to tupleSize - 1 do
        Result[i] := TScalar.Create(FDef[i], values[i]);
end;

{ TScalarMemberSeries }

constructor TScalarMemberSeries.Create(AKind: TScalarKind; const AArray: TChunkArray<TScalarValue>; AElementIdx, AElemCount: Integer);
begin
    Assert(AArray.Count mod AElemCount = 0);
    Assert(AElementIdx >= 0);
    Assert(AElementIdx < AElemCount);

    FKind := AKind;
    FArray := AArray;
    FElementIdx := AElementIdx;
    FElemCount := AElemCount;
end;

function TScalarMemberSeries.GetCount: Int64;
begin
    if FElemCount = 0 then
        exit(0);
    Result := FArray.Count div FElemCount;
end;

function TScalarMemberSeries.GetItems(Idx: Integer): TScalar;
begin
    Result.Create(FKind, FArray.Items[(FArray.Count - FElemCount) - (Idx * FElemCount) + FElementIdx]);
end;

{ TScalarRecordDefinition }

constructor TScalarRecordDefinition.Create(const AFields: TArray<TScalarRecordField>);
begin
    FFields := AFields;
end;

function TScalarRecordDefinition.IndexOf(const Name: String): Integer;
var
    i: Integer;
begin
    for i := 0 to High(FFields) do
    begin
        if (CompareText(FFields[i].Name, Name) = 0) then
            exit(i);
    end;
    Result := -1;
end;

{ TScalarRecordSeries }

// Implements a time series of scalar records using a chunk array for efficient storage.
// Each record's fields are stored sequentially in the TChunkArray.

constructor TScalarRecordSeries.Create(const ADef: TScalarRecordDefinition);
begin
    Assert(Length(ADef.Fields) > 0);
    FDef := ADef;
    FTotalCount := 0;
end;

procedure TScalarRecordSeries.Add(const Item: TScalarRecord; Lookback: Int64 = -1);
begin
    FArray.Add(Item.Fields, Length(FDef.Fields) * Lookback);
    inc(FTotalCount);
end;

function TScalarRecordSeries.AsTupleSeries: TScalarTupleSeries;
var
    tupleDef: TArray<TScalarKind>;
    i: Integer;
begin
    // Extract the kinds from the record definition to create a tuple definition.
    SetLength(tupleDef, Length(FDef.Fields));
    for i := 0 to High(FDef.Fields) do
        tupleDef[i] := FDef.Fields[i].Kind;

    // Create the tuple series with the new definition.
    Result := TScalarTupleSeries.Create(tupleDef);

    // The underlying data is compatible, so we can share it.
    Result.FArray := FArray;
    Result.FTotalCount := FTotalCount;
end;

function TScalarRecordSeries.CreateMemberSeries(const Field: String): TScalarMemberSeries;
begin
    var elem := FDef.IndexOf(Field);
    if elem < 0 then
        raise EArgumentException.CreateFmt('Field "%s" not found in record definition.', [Field]);

    Result := TScalarMemberSeries.Create(FDef.Fields[elem].Kind, FArray, elem, Length(FDef.Fields));
end;

function TScalarRecordSeries.GetCount: Int64;
begin
    Result := FArray.Count div Length(FDef.Fields);
end;

function TScalarRecordSeries.GetDef: TScalarRecordDefinition;
begin
    Result := FDef;
end;

function TScalarRecordSeries.GetItems(Idx: Integer): TScalarRecord;
var
    values: TArray<TScalarValue>;
    fieldCount: Integer;
begin
    fieldCount := Length(FDef.Fields);
    SetLength(values, fieldCount);
    Move(FArray.ItemRef[(FArray.Count - fieldCount) - (Idx * fieldCount)]^, values[0], sizeof(TScalarValue) * fieldCount);
    Result.Create(FDef, values);
end;

function TScalarKindHelper.ToString: string;
begin
    case Self of
        skInteger: Result := 'integer';
        skInt64: Result := 'int64';
        skUInt64: Result := 'uint64';
        skSingle: Result := 'single';
        skDouble: Result := 'double';
        skDateTime: Result := 'datetime';
        skTimestamp: Result := 'timestamp';
        skBoolean: Result := 'boolean';
        skChar: Result := 'char';
        skPChar: Result := 'pchar';
        skString: Result := 'string';
        skBytes: Result := 'bytes';
        skDecimal: Result := 'decimal';
    else
        Result := 'unknown';
    end;
end;

initialization
    Assert(sizeof(TScalarValue) = 8);

end.