public static List <Position> FindPath(Grid grid, Position start, Position end, Offset[] movementPattern, int iterationLimit)
{
ClearStepList();
if (start == end)
{
return(new List <Position> {
start
});
}
var head = new MinHeapNode(start, ManhattanDistance(start, end));
var open = new MinHeap();
open.Push(head);
var costSoFar = new float[grid.DimX * grid.DimY];
var cameFrom = new Position[grid.DimX * grid.DimY];
while (open.HasNext() && iterationLimit > 0)
{
// Get the best candidate
var current = open.Pop().Position;
MessageCurrent(current, PartiallyReconstructPath(grid, start, current, cameFrom));
if (current == end)
{
return(ReconstructPath(grid, start, end, cameFrom));
}
Step(grid, open, cameFrom, costSoFar, movementPattern, current, end);
MessageClose(current);
--iterationLimit;
}
return(null);
}
private static void Step(
Grid grid,
Boundary boundary,
MinHeap open,
Position[] cameFrom,
float[] costSoFar,
Offset[] movementPattern,
AgentShape shape,
Position current,
Position end,
ref Position closestPosition,
ref float closestDistance)
{
// Get the cost associated with getting to the current position
var initialCost = costSoFar[grid.GetIndexUnchecked(current)];
// Get all directions we can move to according to the movement pattern and the dimensions of the grid
foreach (var option in GetMovementOptions(current, boundary, movementPattern))
{
var position = current + option;
var cellCost = grid.GetCellCostUnchecked(position, shape);
// Ignore this option if the cell is blocked
if (float.IsInfinity(cellCost))
{
continue;
}
var index = grid.GetIndexUnchecked(position);
// Compute how much it would cost to get to the new position via this path
var newCost = initialCost + cellCost * option.Cost;
// Compare it with the best cost we have so far, 0 means we don't have any path that gets here yet
var oldCost = costSoFar[index];
if (!(oldCost <= 0) && !(newCost < oldCost))
{
continue;
}
// Update the best path and the cost if this path is cheaper
costSoFar[index] = newCost;
cameFrom[index] = current;
// Use the heuristic to compute how much it will probably cost
// to get from here to the end, and store the node in the open list
var remainCost = ManhattanDistance(position, end);
var expectedCost = newCost + remainCost;
open.Push(new MinHeapNode(position, expectedCost));
// If current position is closest to the end then remember it,
// we may need it if no path is found.
if (remainCost < closestDistance)
{
closestPosition = position;
closestDistance = remainCost;
}
MessageOpen(position);
}
}
/// <summary>
/// Find path
/// </summary>
/// <param name="grid">Grid</param>
/// <param name="start">Start point</param>
/// <param name="end">End point</param>
/// <param name="movementPattern">Movement pattern</param>
/// <param name="shape">Shape of an agent</param>
/// <param name="path">Returns shortest path from start to the end if found. Can also return partial path if end is not reachable. In case of error returns empty array.</param>
/// <param name="iterationLimit">Maximum count of iterations</param>
/// <returns>Result of the path finding.</returns>
public static PathFindResult TryFindPath(Grid grid, Position start, Position end, Offset[] movementPattern, AgentShape shape, out List <Position> path, int iterationLimit = int.MaxValue)
{
ClearStepList();
// If is where should be, then there's no path to take
if (start == end)
{
path = new List <Position>();
return(PathFindResult.AlreadyAtTheEnd);
}
// To avoid lot of grid boundaries checking calculations during path finding, do the math here.
// So get the possible boundaries considering the shape of the agent.
var boundary = Boundary.FromSizeAndShape(grid.DimX, grid.DimY, shape);
// Make sure that agent is within grid. This means agent shape has to be considered.
if (!boundary.IsInside(start))
{
path = new List <Position>();
return(PathFindResult.StartOutsideBoundaries);
}
// Make sure that end position is within grid
if (!grid.IsInside(end))
{
path = new List <Position>();
return(PathFindResult.EndOutsideBoundaries);
}
var head = new MinHeapNode(start, ManhattanDistance(start, end));
var open = new MinHeap();
open.Push(head);
var costSoFar = new float[grid.DimX * grid.DimY];
var cameFrom = new Position[grid.DimX * grid.DimY];
// Keep record of closest point in case the end point is not reachable
Position closestPosition = start;
float closestDistance = float.PositiveInfinity;
// Try to find path until options run out or iterations limit is exceeded
while (open.HasNext() && iterationLimit > 0)
{
// Get the best candidate
var current = open.Pop().Position;
MessageCurrent(current, PartiallyReconstructPath(grid, start, current, cameFrom));
if (current == end)
{
path = ReconstructPath(grid, start, end, cameFrom);
return(PathFindResult.PathFound);
}
Step(grid, boundary, open, cameFrom, costSoFar, movementPattern, shape, current, end, ref closestPosition, ref closestDistance);
MessageClose(current);
--iterationLimit;
}
// Construct path to the closest position
if (start != closestPosition)
{
MessageCurrent(closestPosition, PartiallyReconstructPath(grid, start, closestPosition, cameFrom));
path = ReconstructPath(grid, start, closestPosition, cameFrom);
return(PathFindResult.PartialPathFound);
}
// Stuck
path = new List <Position>();
return(PathFindResult.Stuck);
}
