def reset_puzzle(self, method):
self.total_iters = 0
self.puzzle_state = BlocksworldPuzzle(setup.generate_puzzle())
self.visited = set()
self.visited_directions = {}
if method == 'bfs':
self.queue = deque([self.puzzle_state])
elif method == 'dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'id_dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'a_star':
self.queue = PriorityQueue()
self.queue.put((0, self.puzzle_state))
self.heuristic = lambda heur: heur.total_manhattan_distance()
def reset_puzzle(self, method):
self.total_iters = 0
self.puzzle_state = BlocksworldPuzzle(setup.generate_puzzle())
self.visited = set()
self.visited_directions = {}
if method == 'bfs':
self.queue = deque([self.puzzle_state])
elif method == 'dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'id_dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'a_star':
self.queue = PriorityQueue()
self.queue.put((0, self.puzzle_state))
self.heuristic = lambda heur: heur.total_manhattan_distance()
class Solver:
def __init__(self):
self.total_iters = 0
self.queue = None
self.visited = None
self.visited_directions = None
self.puzzle_state = None
self.heuristic = None
self.method = None
self.methods = {
'bfs': self.bfs,
'dfs': self.dfs,
'id_dfs': self.id_dfs,
'a_star': self.a_star
}
def search(self, method):
self.reset_puzzle(method)
start = dt.datetime.now()
search_outcome = self.methods[method]()
end = dt.datetime.now()
if self.puzzle_state:
moves = self.puzzle_state.moves
directions = self.puzzle_state.directions if len(self.puzzle_state.directions) < 50 else []
else:
moves = None
directions = None
result = SearchResult(
search_outcome,
self.total_iters,
len(self.queue),
moves,
directions,
end - start
)
return result
def log_state(self):
if self.total_iters % 20000 == 0:
stdout.write(".")
stdout.flush()
def reset_puzzle(self, method):
self.total_iters = 0
self.puzzle_state = BlocksworldPuzzle(setup.generate_puzzle())
self.visited = set()
self.visited_directions = {}
if method == 'bfs':
self.queue = deque([self.puzzle_state])
elif method == 'dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'id_dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'a_star':
self.queue = PriorityQueue()
self.queue.put((0, self.puzzle_state))
self.heuristic = lambda heur: heur.total_manhattan_distance()
# Breadth-first search.
# FIFO queue -- Double Ended Queue(DEQUEUE)
# GET - left | ADD - right.
def bfs(self):
while len(self.queue) > 0:
self.total_iters += 1
self.log_state()
self.visited_directions = {}
# Get nearest(more shallow state.
self.puzzle_state = self.queue.popleft()
# Add current state to visited ones
self.visited.add(self.puzzle_state.puzzle_hash())
self.visited_directions[self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(state, visited):
return state.puzzle_hash() in visited
# Queue extend with not visited
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited):
valid_children.append(child_state)
self.queue.extend(valid_children)
# No result.
return False
# Depth First Search.
# Not optimal solution.
# It can go on infinitely.
# LIFO Queue --> python list()
def dfs(self):
self.queue = [self.puzzle_state]
# Visited : {state_hash:state_depth}
self.visited = {}
while len(self.queue) > 0:
# Stopping in 1000000 iterations in case of infinite loop
if self.total_iters > 1000000:
return False
self.total_iters += 1
self.log_state()
# Get the state that is deepest
self.puzzle_state = self.queue.pop() # Get deepest state.
# Add current state in visited.
self.visited[self.puzzle_state.puzzle_hash()] = self.puzzle_state.moves
self.visited_directions[self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(child, visited):
return child.puzzle_hash() in visited
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited) \
or self.visited[child_state.puzzle_hash()] > child_state.moves:
valid_children.append(child_state)
self.queue.extend(valid_children)
# No result
return False
# Iterative deepening depth-first search.
# The solution is more optimal than DFS.
# It is faster than BFS
# It can go on infinitely, like DFS.
# LIFO Queue --> python list()
def id_dfs(self):
start_state = self.puzzle_state
for depth in itertools.count():
self.queue = [start_state]
self.visited = {}
while len(self.queue) > 0:
self.total_iters += 1
self.log_state()
# Get the state that is the deepest.
self.puzzle_state = self.queue.pop()
# Add current state to visited ones.
self.visited[self.puzzle_state.puzzle_hash()] = self.puzzle_state.moves
self.visited_directions[self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(child, visited):
return child.puzzle_hash() in visited
if self.puzzle_state.moves < depth:
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited) \
or self.visited[child_state.puzzle_hash()] > child_state.moves:
valid_children.append(child_state)
self.queue.extend(valid_children)
return False
# A* search.
# Most optimal solution time-wise.
# QUEUE: PRIORITY QUEUE(when _get() is used, it gets the item with the minimum value first)
# ORDER: Heuristic(Total Manhattan Distance + number of state moves)
def a_star(self):
# Total_moves = total so far + total to be made
def total_moves(puzzle_state, heuristic):
return puzzle_state.moves + heuristic(self.puzzle_state)
def state_cost_tuple(puzzle_state, heuristic):
return total_moves(puzzle_state, heuristic), puzzle_state
def valid_children(puzzle_state, visited):
children = []
for child_state in puzzle_state.get_children():
if child_state.puzzle_hash() not in visited:
children.append(child_state)
return children
while not self.queue.empty():
self.total_iters += 1
self.log_state()
# Get state with lowest value(moves left)
self.puzzle_state = self.queue.get()[1]
# Add state to visited
self.visited.add(self.puzzle_state.puzzle_hash())
self.visited_directions[self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
self.queue = self.queue.queue
return True
for valid_child in valid_children(self.puzzle_state, self.visited):
self.queue.put(state_cost_tuple(valid_child, self.heuristic))
# No result
self.queue = self.queue.queue
return False
class Solver:
def __init__(self):
self.total_iters = 0
self.queue = None
self.visited = None
self.visited_directions = None
self.puzzle_state = None
self.heuristic = None
self.method = None
self.methods = {
'bfs': self.bfs,
'dfs': self.dfs,
'id_dfs': self.id_dfs,
'a_star': self.a_star
}
def search(self, method):
self.reset_puzzle(method)
start = dt.datetime.now()
search_outcome = self.methods[method]()
end = dt.datetime.now()
if self.puzzle_state:
moves = self.puzzle_state.moves
directions = self.puzzle_state.directions if len(
self.puzzle_state.directions) < 50 else []
else:
moves = None
directions = None
result = SearchResult(search_outcome, self.total_iters,
len(self.queue), moves, directions, end - start)
return result
def log_state(self):
if self.total_iters % 20000 == 0:
stdout.write(".")
stdout.flush()
def reset_puzzle(self, method):
self.total_iters = 0
self.puzzle_state = BlocksworldPuzzle(setup.generate_puzzle())
self.visited = set()
self.visited_directions = {}
if method == 'bfs':
self.queue = deque([self.puzzle_state])
elif method == 'dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'id_dfs':
self.queue = [self.puzzle_state]
self.visited = {}
elif method == 'a_star':
self.queue = PriorityQueue()
self.queue.put((0, self.puzzle_state))
self.heuristic = lambda heur: heur.total_manhattan_distance()
# Breadth-first search.
# FIFO queue -- Double Ended Queue(DEQUEUE)
# GET - left | ADD - right.
def bfs(self):
while len(self.queue) > 0:
self.total_iters += 1
self.log_state()
self.visited_directions = {}
# Get nearest(more shallow state.
self.puzzle_state = self.queue.popleft()
# Add current state to visited ones
self.visited.add(self.puzzle_state.puzzle_hash())
self.visited_directions[
self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(state, visited):
return state.puzzle_hash() in visited
# Queue extend with not visited
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited):
valid_children.append(child_state)
self.queue.extend(valid_children)
# No result.
return False
# Depth First Search.
# Not optimal solution.
# It can go on infinitely.
# LIFO Queue --> python list()
def dfs(self):
self.queue = [self.puzzle_state]
# Visited : {state_hash:state_depth}
self.visited = {}
while len(self.queue) > 0:
# Stopping in 1000000 iterations in case of infinite loop
if self.total_iters > 1000000:
return False
self.total_iters += 1
self.log_state()
# Get the state that is deepest
self.puzzle_state = self.queue.pop() # Get deepest state.
# Add current state in visited.
self.visited[
self.puzzle_state.puzzle_hash()] = self.puzzle_state.moves
self.visited_directions[
self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(child, visited):
return child.puzzle_hash() in visited
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited) \
or self.visited[child_state.puzzle_hash()] > child_state.moves:
valid_children.append(child_state)
self.queue.extend(valid_children)
# No result
return False
# Iterative deepening depth-first search.
# The solution is more optimal than DFS.
# It is faster than BFS
# It can go on infinitely, like DFS.
# LIFO Queue --> python list()
def id_dfs(self):
start_state = self.puzzle_state
for depth in itertools.count():
self.queue = [start_state]
self.visited = {}
while len(self.queue) > 0:
self.total_iters += 1
self.log_state()
# Get the state that is the deepest.
self.puzzle_state = self.queue.pop()
# Add current state to visited ones.
self.visited[
self.puzzle_state.puzzle_hash()] = self.puzzle_state.moves
self.visited_directions[self.puzzle_state.puzzle_hash(
)] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
return True
def is_visited(child, visited):
return child.puzzle_hash() in visited
if self.puzzle_state.moves < depth:
valid_children = []
for child_state in self.puzzle_state.get_children():
if not is_visited(child_state, self.visited) \
or self.visited[child_state.puzzle_hash()] > child_state.moves:
valid_children.append(child_state)
self.queue.extend(valid_children)
return False
# A* search.
# Most optimal solution time-wise.
# QUEUE: PRIORITY QUEUE(when _get() is used, it gets the item with the minimum value first)
# ORDER: Heuristic(Total Manhattan Distance + number of state moves)
def a_star(self):
# Total_moves = total so far + total to be made
def total_moves(puzzle_state, heuristic):
return puzzle_state.moves + heuristic(self.puzzle_state)
def state_cost_tuple(puzzle_state, heuristic):
return total_moves(puzzle_state, heuristic), puzzle_state
def valid_children(puzzle_state, visited):
children = []
for child_state in puzzle_state.get_children():
if child_state.puzzle_hash() not in visited:
children.append(child_state)
return children
while not self.queue.empty():
self.total_iters += 1
self.log_state()
# Get state with lowest value(moves left)
self.puzzle_state = self.queue.get()[1]
# Add state to visited
self.visited.add(self.puzzle_state.puzzle_hash())
self.visited_directions[
self.puzzle_state.puzzle_hash()] = self.puzzle_state.directions
if self.puzzle_state.is_goal_state():
self.queue = self.queue.queue
return True
for valid_child in valid_children(self.puzzle_state, self.visited):
self.queue.put(state_cost_tuple(valid_child, self.heuristic))
# No result
self.queue = self.queue.queue
return False
