the new segment intersection algorithm

[BigZebra101010]

I have updated to the latest code and encountered an issue. When I use the RectClip method, there are extra vertices in the resulting graphics. After debugging, it seems the problem lies in the GetIntersectPoint method.

I read the Wikipedia page referenced in the comments：https://en.wikipedia.org/wiki/Line%E2%80%93line_intersection

I don't fully understand the current implementation.

I made some modifications following the description on Wikipedia，problem was solved.

  inline bool GetIntersectPoint(const Point64& ln1a, const Point64& ln1b,
    const Point64& ln2a, const Point64& ln2b, Point64& ip)
  {  
    // https://en.wikipedia.org/wiki/Line%E2%80%93line_intersection
    double dx1 = static_cast<double>(ln1b.x - ln1a.x);
    double dy1 = static_cast<double>(ln1b.y - ln1a.y);
    double dx2 = static_cast<double>(ln2b.x - ln2a.x);
    double dy2 = static_cast<double>(ln2b.y - ln2a.y);

    double det = dy1 * dx2 - dy2 * dx1;
    if (det == 0.0) return false;

  #if 0
    double t = ((ln1a.x - ln2a.x) * dy2 - (ln1a.y - ln2a.y) * dx2) / det;
    if (t <= 0.0) ip = ln1a;        // ?? check further (see also #568)
    else if (t >= 1.0) ip = ln1b;   // ?? check further
    else
    {
      ip.x = static_cast<int64_t>(ln1a.x + t * dx1);
      ip.y = static_cast<int64_t>(ln1a.y + t * dy1);
    }
  #else
    double t = ((ln1a.x - ln2a.x) * dy2 - (ln1a.y - ln2a.y) * dx2) / det;
    if (t < 0.0 || t > 1.0)
    {
      return false;
    }

    double u = ((ln1a.x - ln2a.x) * dy1 - (ln1a.y - ln2a.y) * dx1) / det;
    if (u < 0.0 || u > 1.0)
    {
      return false;
    }

    ip.x = static_cast<int64_t>(ln1a.x + t * dx1);
    ip.y = static_cast<int64_t>(ln1a.y + t * dy1);
  #endif
    return true;
  }

I didn't do a lot of testing, so I'm not sure if there are other issues.

[AngusJohnson]

I have updated to the latest code and encountered an issue. When I use the RectClip method, there are extra vertices in the resulting graphics.

Please provide an example.

I don't fully understand the current implementation.
I made some modifications following the description on Wikipedia，problem was solved.

I'm afraid your proposed solution would be less satisfactory with the other clipping routines.
The intersection point will only be significantly outside the intersecting segments (ie t < 0.0 or t > 1.0) when the edges are almost parallel. And in these circumstances, in my testing, the best solution is to return the segment vertices that are closest to these (outside) intersections.

[BigZebra101010]

Thanks very much for your reply, I'm trying to debug and provide an example.

[BigZebra101010]

Very awkward, somehow there was an error when I updated the code. #568