import FluidGridData;

struct Params
{
    // read-write
    FluidGridData<float> density;
    // read-only
    FluidGridData<float> density0;
    // read-only
    FluidGridData<float> velocityX;
    // read-only
    FluidGridData<float> velocityY;
    // read-only
    FluidGridData<float> velocityZ;
    float dt;
};
ParameterBlock<Params> params;

[shader("compute")]
[numthreads(32, 8, 1)]
void advect(uint3 dispatchThreadID : SV_DispatchThreadID)
{
    FluidGridData<float> density = params.density;
    FluidGridData<float> density0 = params.density0;
    FluidGridData<float> velocityX = params.velocityX;
    FluidGridData<float> velocityY = params.velocityY;
    FluidGridData<float> velocityZ = params.velocityZ;
    float dt = params.dt;

    int x = dispatchThreadID.x + 1;
    int y = dispatchThreadID.y + 1;
    int z = dispatchThreadID.z + 1;
    if(x >= gridSize.x - 1 || y >= gridSize.y - 1 || z >= gridSize.z - 1) return;

    float dt0x = dt * (gridSize.x - 2);
    float dt0y = dt * (gridSize.y - 2);
    float dt0z = dt * (gridSize.z - 2);

    float3 pos = float3(x - dt0x * velocityX[x, y, z], y - dt0y * velocityY[x, y, z], z - dt0z * velocityZ[x, y, z]);
    pos.x = clamp(pos.x, 0.5f, gridSize.x - 1.5f);
    pos.y = clamp(pos.y, 0.5f, gridSize.y - 1.5f);
    pos.z = clamp(pos.z, 0.5f, gridSize.z - 1.5f);

    int3 i0 = int3(pos);
    int3 i1 = i0 + int3(1, 1, 1);

    float3 s1 = pos - i0;
    float3 s0 = 1 - s1;

    density[x, y, z] = 
        s0.x * (s0.y * (s0.z * density0[i0.x, i0.y, i0.z] + s1.z * density0[i0.x, i0.y, i1.z]) + 
        s1.y * (s0.z * density0[i0.x, i1.y, i0.z] + s1.z * density0[i0.x, i1.y, i1.z])) + 
        s1.x * (s0.y * (s0.z * density0[i1.x, i0.y, i0.z] + s1.z * density0[i1.x, i0.y, i1.z]) + 
        s1.y * (s0.z * density0[i1.x, i1.y, i0.z] + s1.z * density0[i1.x, i1.y, i1.z]));
}