Implementation of the recursive backtracking algorithm for maze generation

recursive-backtracker.rb

# Recursive backtracking algorithm for maze generation. Requires that
# the entire maze be stored in memory, but is quite fast, easy to
# learn and implement, and (with a few tweaks) gives fairly good mazes.
# Can also be customized in a variety of ways.

DIRS     = (N, S, E, W = 1, 2, 4, 8)
DX       = { E => 1, W => -1, N =>  0, S => 0 }
DY       = { E => 0, W =>  0, N => -1, S => 1 }
OPPOSITE = { E => W, W =>  E, N =>  S, S => N }

width  = (ARGV[0] || 10).to_i
height = (ARGV[1] || width).to_i
seed   = (ARGV[2] || rand(0xFFFF_FFFF)).to_i

srand(seed)

cells = Array.new(height) { Array.new(width, 0) }
stack = [[0, 0, DIRS.sort_by{rand}]]

until stack.empty?
  x, y, directions = stack.last

  until directions.empty?
    direction = directions.pop
    nx, ny = x + DX[direction], y + DY[direction]

    if nx >= 0 && ny >= 0 && nx < width && ny < height && cells[ny][nx] == 0
      cells[y][x] |= direction
      cells[ny][nx] |= OPPOSITE[direction]
      stack.push((x, y, directions = [nx, ny, DIRS.sort_by{rand}]))
    end
  end

  stack.pop
end

puts " " + "_" * (width * 2 - 1)
height.times do |y|
  print "|"
  width.times do |x|
    print((cells[y][x] & S != 0) ? " " : "_")
    if cells[y][x] & E != 0
      print(((cells[y][x] | cells[y][x+1]) & S != 0) ? " " : "_")
    else
      print "|"
    end
  end
  puts
end

puts "#{$0} #{width} #{height} #{seed}"