// NOTE(fox): Pay attention to how the Y pitch differs between the unpacked
// bitmaps and the 4x4 packed bitmaps, since odd-sized bitmaps are padded.

// TODO(fox): I could write an AVX version of this function, but it may not be
// that much faster since we have to do a bit of uninterleaving.

// 0 - store in 4x4 chunks
// 1 - unpack to 1xwidth
internal void
Convert4x4Chunk(pixel_buffer *Buffer, uint8 Which)
{
    uint8 *Src = (uint8 *)Buffer->OriginalBuffer;
    uint8 *Temp = (uint8 *)Buffer->EffectBuffer;
    uint32 RemainderPixels = Buffer->Width % 4;
    for (uint32 Y = 0; Y < Buffer->Height; Y++) {
        uint32 X = 0;
        while (X < Buffer->Width - RemainderPixels) {
            uint32 XLookup = (X >> 2)*16 + (X % 4);
            uint32 YLookup = (Y >> 2)*(Buffer->FullWidth*4) + (Y % 4)*4;
            uint32 PixelToSeek = XLookup + YLookup;
            uint8 *DPixel, *Pixel;
            if (Which == 0) {
                DPixel = Temp + PixelToSeek*Buffer->BytesPerPixel;
                Pixel  = Src  + Y*Buffer->Width*4 + X*Buffer->BytesPerPixel;
            } else {
                Pixel  = Src  + PixelToSeek*Buffer->BytesPerPixel;
                DPixel = Temp + Y*Buffer->Width*4 + X*Buffer->BytesPerPixel;
            }

            if (InstructionMode == sse_enabled || InstructionMode == avx_enabled) {
                __m128i Row = _mm_loadu_si128((__m128i *)Pixel);
                _mm_storeu_si128((__m128i *)DPixel, Row);
                X+=4;
            } else {
                *(uint32 *)DPixel = *(uint32 *)Pixel;
                X++;
            }
        }
        while (X < Buffer->Width) {
            uint32 XLookup = (X >> 2)*16 + (X % 4);
            uint32 YLookup = (Y >> 2)*(Buffer->FullWidth*4) + (Y % 4)*4;
            uint32 PixelToSeek = XLookup + YLookup;
            uint8 *DPixel, *Pixel;
            if (Which == 0) {
                DPixel = Temp + PixelToSeek*Buffer->BytesPerPixel;
                Pixel  = Src  + Y*Buffer->Width*4 + X*Buffer->BytesPerPixel;
            } else {
                Pixel  = Src  + PixelToSeek*Buffer->BytesPerPixel;
                DPixel = Temp + Y*Buffer->Width*4 + X*Buffer->BytesPerPixel;
            }

            *(uint32 *)DPixel = *(uint32 *)Pixel;
            X++;
        }
    }
}

// TODO(fox): Replace this in the future.
internal void *
MoveImportToBitmap(memory *Memory, pixel_buffer *Raster, void *Input)
{
    uint8 *Row = ((uint8 *)Input);
    void *Output = AllocateMemory(Memory, Raster->FullWidth * Raster->FullHeight * Raster->BytesPerPixel, B_Scratch);
    uint8 *Row2 = ((uint8 *)Output);

    uint64 bytes = 0;
    uint16 ByteOffset = Raster->BytesPerPixel;
    if (InstructionMode == avx_enabled)
        ByteOffset = 8*Raster->BytesPerPixel;
    else if (InstructionMode == avx_enabled)
        ByteOffset = 4*Raster->BytesPerPixel;

    uint64 TotalBytes = Raster->Height*Raster->Width*Raster->BytesPerPixel;
    uint64 RemainderBytes = TotalBytes % ByteOffset;

    while (bytes <= TotalBytes - RemainderBytes) {
        uint8 *Pixel = (uint8 *)Row + bytes;
        uint8 *Pixel2 = (uint8 *)Row2 + bytes;
        if (InstructionMode == sse_enabled || InstructionMode == avx_enabled) {
            __m128i OutputPixel = _mm_loadu_si128((__m128i *)Pixel);
            _mm_storeu_si128((__m128i *)Pixel2, OutputPixel);
            bytes += 4*Raster->BytesPerPixel;
        } else {
            *(uint32 *)Pixel2 = *(uint32 *)Pixel;
            bytes += Raster->BytesPerPixel;
        }
    }
    while (bytes <= TotalBytes) {
        uint8 *Pixel = (uint8 *)Row + bytes;
        uint8 *Pixel2 = (uint8 *)Row2 + bytes;
        *(uint32 *)Pixel2 = *(uint32 *)Pixel;
        bytes += Raster->BytesPerPixel;
    }
    return Output;
}

internal void
ClearBuffer(pixel_buffer *Raster, void *Buffer)
{
    uint8 *Row = (uint8 *)Buffer;
    __m256i Zero8 = _mm256_setzero_si256();
    __m128i Zero = _mm_setzero_si128();

    uint64 bytes = 0;
    uint16 ByteOffset = Raster->BytesPerPixel;
    if (InstructionMode == avx_enabled)
        ByteOffset = 8*Raster->BytesPerPixel;
    else if (InstructionMode == avx_enabled)
        ByteOffset = 4*Raster->BytesPerPixel;

    uint64 TotalBytes = Raster->FullHeight*Raster->FullWidth*Raster->BytesPerPixel;

    while (bytes < TotalBytes) {
        uint8 *Pixel = Row + bytes;
        if (InstructionMode == avx_enabled) {
            _mm256_storeu_si256((__m256i *)Pixel, Zero8);
        } else if (InstructionMode == sse_enabled) {
            _mm_storeu_si128((__m128i *)Pixel, Zero);
        } else {
            *(uint32 *)Pixel = 0x00000000;
        }
        bytes += ByteOffset;
    }
}

// 0 - original -> effect
// 1 - effect -> original
internal void
CopyToBuffer(pixel_buffer *Raster, uint16 Which)
{
    uint8 *Row, *Row2;
    if (Which == 0) {
        Row = ((uint8 *)Raster->OriginalBuffer);
        Row2 = ((uint8 *)Raster->EffectBuffer);
    } else {
        Row = ((uint8 *)Raster->EffectBuffer);
        Row2 = ((uint8 *)Raster->OriginalBuffer);
    }

    uint64 bytes = 0;
    uint16 ByteOffset = Raster->BytesPerPixel;
    if (InstructionMode == avx_enabled)
        ByteOffset = 8*Raster->BytesPerPixel;
    else if (InstructionMode == avx_enabled)
        ByteOffset = 4*Raster->BytesPerPixel;

    uint64 TotalBytes = Raster->FullHeight*Raster->FullWidth*Raster->BytesPerPixel;
    uint64 RemainderBytes = TotalBytes % ByteOffset;

    while (bytes <= TotalBytes - RemainderBytes) {
        uint8 *Pixel = (uint8 *)Row + bytes;
        uint8 *Pixel2 = (uint8 *)Row2 + bytes;
        if (InstructionMode == sse_enabled || InstructionMode == avx_enabled) {
            __m128i OutputPixel = _mm_loadu_si128((__m128i *)Pixel);
            _mm_storeu_si128((__m128i *)Pixel2, OutputPixel);
            bytes += 4*Raster->BytesPerPixel;
        } else {
            *(uint32 *)Pixel2 = *(uint32 *)Pixel;
            bytes += Raster->BytesPerPixel;
        }
    }
    while (bytes <= TotalBytes) {
        uint8 *Pixel = (uint8 *)Row + bytes;
        uint8 *Pixel2 = (uint8 *)Row2 + bytes;
        *(uint32 *)Pixel2 = *(uint32 *)Pixel;
        bytes += Raster->BytesPerPixel;
    }
}

internal void
BitmapPackRGB(pixel_buffer *Buffer) {
    Assert(Buffer->Pitch);
    Convert4x4Chunk(Buffer, 0);
    CopyToBuffer(Buffer, 1);
    ClearBuffer(Buffer, Buffer->EffectBuffer);
}

internal void
OutputToViewport(pixel_buffer *CompBuffer, project_state *State, GLuint textureID) {
    Convert4x4Chunk(CompBuffer, 1);
    EndRenderState(State);
    glBindTexture(GL_TEXTURE_2D, textureID);
    glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, CompBuffer->Width, CompBuffer->Height, GL_RGBA, GL_UNSIGNED_BYTE,
                           CompBuffer->EffectBuffer);
}

internal void
DebugFillSolid(pixel_buffer *Raster, v4 Color)
{
    uint32 ColS = ColToUint32(Color);
    __m256i Col8 = _mm256_set1_epi32(ColS);
    __m128i Col = _mm_set1_epi32(ColS);
    uint8 *Row = (uint8 *)Raster->OriginalBuffer;

    uint64 bytes = 0;
    uint16 ByteOffset = Raster->BytesPerPixel;
    if (InstructionMode == avx_enabled)
        ByteOffset = 8*Raster->BytesPerPixel;
    else if (InstructionMode == avx_enabled)
        ByteOffset = 4*Raster->BytesPerPixel;

    uint64 TotalBytes = Raster->FullHeight*Raster->FullWidth*Raster->BytesPerPixel;

    while (bytes < TotalBytes) {
        uint8 *Pixel = Row + bytes;
        if (InstructionMode == avx_enabled) {
            _mm256_storeu_si256((__m256i *)Pixel, Col8);
        } else if (InstructionMode == sse_enabled) {
            _mm_storeu_si128((__m128i *)Pixel, Col);
        } else {
            *(uint32 *)Pixel = ColS;
        }
        bytes += ByteOffset;
    }
}

internal void
DebugBitmap(pixel_buffer *Raster)
{
    uint8 asda = 0x0;
    uint8 *Row = ((uint8 *)Raster->OriginalBuffer);
    real32 XInc = 255.0f / Raster->Width;
    real32 YInc = 255.0f / Raster->Height;
    for (uint8 Y = 0; Y < Raster->Height; Y++) {
        for (uint8 X = 0; X < Raster->Width; X++) {
            uint8 *Pixel = (uint8 *)Row + Raster->FullWidth*Y*4 + X*4;
            // *(uint32 *)Pixel = 0xffffffff;
            if (Y > 3) { asda = 0xff; }
            *(uint32 *)Pixel = ((0xff << 24) |
                                (asda << 16) |
                                (RoundReal32ToInt32((YInc * Y)) << 8) |
                                (RoundReal32ToInt32((XInc * X))) );
        }
    }
}