File size: 13,020 Bytes
8311f95 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 |
import pickle
import random
import math
import os
from rooms import NameGenerator
class MazeLoader:
"""
Loads a maze that can be plotted and used to generate a natural language problem.
"""
def __init__(
self,
filename,
shuffle_description=True,
hide_coordinates=False,
removed_key_count=None,
solvable=True,
):
self.solvable = solvable
self.filename = filename
self.data = pickle.load(open(filename, "rb"))
self.shuffle_description = shuffle_description
self.hide_coordinates = hide_coordinates
self.removed_key_count = removed_key_count
if self.removed_key_count is None:
self.removed_key_count = 0
if removed_key_count not in self.data["sub_maze_configurations"].keys():
raise ValueError(
f"Removed key count {removed_key_count} is not in the sub_maze_configurations keys"
)
self.room_name = self.data["room_name"]
self.maze = self.data["sub_maze_configurations"][self.removed_key_count]["maze"]
self.doors = self.data["sub_maze_configurations"][self.removed_key_count][
"doors"
]
self.keys_locations = self.data["sub_maze_configurations"][
self.removed_key_count
]["keys_locations"]
self.connected_cells = self.data["connected_cells"]
self.num_locked_doors = (
self.data["world_parameters"]["N_locked_doors"] - self.removed_key_count
)
self.N = self.data["world_parameters"]["N"]
self.M = self.data["world_parameters"]["M"]
name_generator = NameGenerator(self.N, self.M)
self.R = lambda x: name_generator.get_name(x)
if not self.solvable:
print("Maze is not solvable")
# get keys to first and last locked doors
first_key, last_key = self.get_first_and_last_key()
self.swap_key_locations(first_key, last_key)
self.solution = None
else:
self.solution = self.data["standardized_problem_solution"][
self.removed_key_count
][::-1]
self.solution_with_room_names = [
(item[0], self.room_name[item[1]])
if item[0].split("_")[-1] in ["to", "start"]
else item
for item in self.solution
]
self.solution_with_friendly_room_names = [
(item[0], self.R(item[1]))
if item[0].split("_")[-1] in ["to", "start"]
else item
for item in self.solution
]
# we keep the definition of supporting facts in the original solvable problem even if the maze is not solvable as the same path still needs to be used to make the conclusion
self.used_keys = [item[1] for item in self.solution if item[0] == "use_key"]
self.used_path = (
[self.data["start_room"]]
+ [item[1] for item in self.solution if item[0] == "move_to"]
+ [self.data["end_room"]]
)
self.used_connections = list(zip(self.used_path[:-1], self.used_path[1:]))
def compile_description(self, args, mode, friendly=False):
if mode == "door":
cell1, cell2 = args
status = self.doors[(cell1, cell2)][0]
if friendly:
room1, room2 = self.R(cell1), self.R(cell2)
else:
room1, room2 = self.room_name[cell1], self.room_name[cell2]
return f"Room {room1} has a door to room {room2}. "
elif mode == "key_door_relation":
cell1, cell2 = args[:2]
if friendly:
room1, room2 = self.R(cell1), self.R(cell2)
else:
room1, room2 = self.room_name[cell1], self.room_name[cell2]
return (
f"""The locked door between {room1} and {room2} requires key {self.doors[(cell1, cell2)][1]}. """
if self.doors[(cell1, cell2)][0] == "closed and locked"
else ""
)
elif mode == "connected_rooms":
cell1, cell2, door_status = args
if not self.hide_coordinates:
location_description1 = " at " + str(cell1)
location_description2 = " at " + str(cell2)
else:
location_description1 = ""
location_description2 = ""
if friendly:
roomA, roomB = self.R(cell1), self.R(cell2)
else:
roomA, roomB = self.room_name[cell1], self.room_name[cell2]
return (
f"""Room {roomA}{location_description1}"""
+ f""" and {roomB}{location_description2} are connected by a{'n' if door_status == 'open' else ''} {door_status} door. """
)
elif mode == "key_location":
key_id, room = args
if friendly:
room1 = self.R(room)
else:
room1 = self.room_name[room]
return f"""Key {key_id} is in room {room1}. """
elif mode == "rescue_agent_location":
room = args
if friendly:
room1 = self.R(room)
else:
room1 = self.room_name[room]
return f"{self.rescue_agent} is in room {room1}. "
elif mode == "victim_location":
room = args
if friendly:
room1 = self.R(room)
else:
room1 = self.room_name[room]
return f"{self.victim} is in room {room1}. "
def encode_problem_into_nlp(self, shuffle_ratio=0.5, noise_ratio=0.5):
"""
Encodes the problem into a natural language problem.
The facts can be of the following types:
connected_rooms:
1. Room A and B are connected by an open door. (for regular connections)
2. Room A and B are connected by a closed and locked door. (for locked doors)
key_door_relation:
3. The locked door between Room A and Room B requires key 5. (for key door relations)
key_location:
4. Key 5 is in Room C. (for key locations)
rescue_agent_location:
5. The rescue agent is in Room A. (for rescue agent location)
victim_location:
6. The victim is in Room A. (for victim location)
"""
# I. connected_rooms
self.nlp_problem = []
covered_cells = set()
for cell in sorted(self.connected_cells.keys()):
for neighbor in sorted(self.connected_cells[cell]):
if (cell, neighbor) in covered_cells or cell == neighbor:
continue
if (cell, neighbor) in self.used_connections or (
neighbor,
cell,
) in self.used_connections:
supporting = True
else:
supporting = False
door_status = self.doors[(cell, neighbor)][0]
key_id = self.doors[(cell, neighbor)][1]
self.nlp_problem.append(
((cell, neighbor, door_status), "connected_rooms", supporting)
)
# II. key_door_relation
if door_status == "closed and locked" and supporting:
supporting1 = True
else:
supporting1 = False
if door_status == "closed and locked":
self.nlp_problem.append(
((cell, neighbor, key_id), "key_door_relation", supporting1)
)
# III. key_location
if key_id in self.used_keys:
supporting2 = True
else:
supporting2 = False
if door_status == "closed and locked":
self.nlp_problem.append(
(
(key_id, self.keys_locations[key_id]),
"key_location",
supporting2,
)
)
covered_cells.add((cell, neighbor))
covered_cells.add((neighbor, cell))
self.victim = self.data["world_parameters"]["victim"]
self.rescue_agent = self.data["world_parameters"]["rescue_agent"]
# ordering the facts by supporting facts first
self.nlp_problem = [
item[1]
for item in sorted(
[(i, it) for i, it in enumerate(self.nlp_problem[::-1])],
key=lambda x: (x[1][2], x[0]),
reverse=True,
)
]
# number of distracting facts
N_minus = len([item for item in self.nlp_problem if item[2] == False])
# number of supporting facts
N_plus = len(self.nlp_problem) - N_minus
# number of distracting facts to remove
N_minus_x = int(N_minus * (1 - noise_ratio)) - 1
if N_minus_x > 0:
self.nlp_problem = self.nlp_problem[:-N_minus_x]
x = int((N_plus) * (1 - shuffle_ratio))
ordered_part = self.nlp_problem[:x]
shuffle_part = self.nlp_problem[x:]
random.shuffle(shuffle_part)
self.nlp_problem = ordered_part + shuffle_part
self.nlp_problem.append(
(self.data["start_room"], "rescue_agent_location", True)
)
self.nlp_problem.append((self.data["end_room"], "victim_location", True))
natural_language_problem = []
natural_language_problem_friendly = []
for item in self.nlp_problem:
natural_language_problem += [self.compile_description(item[0], item[1])]
natural_language_problem_friendly += [
self.compile_description(item[0], item[1], friendly=True)
]
N_minus = len([item for item in self.nlp_problem if item[2] == False])
N_plus = len(self.nlp_problem) - N_minus
self.support_weight = round(N_plus / float(N_plus + N_minus), 2)
self.shuffle_entropy = self.measure_distraction_entropy()
return (
self.nlp_problem,
"".join(natural_language_problem),
"".join(natural_language_problem_friendly),
self.support_weight,
self.shuffle_entropy,
)
def measure_distraction_entropy(self):
entropy = 0
count = 0
for i in range(0, len(self.nlp_problem), 2):
first_fact = self.nlp_problem[i][-1]
second_fact = (
self.nlp_problem[i + 1][-1] if i + 1 < len(self.nlp_problem) else None
)
if second_fact is None:
continue
count += 1
if first_fact != second_fact and (
first_fact == True or second_fact == True
):
entropy -= math.log(0.5)
return entropy / float(count)
def get_first_and_last_key(self):
pass
def swap_key_locations(self, first_key, last_key):
pass
def single_file_load(
folder_name="maze_5_5_3_0.5_0.0_True",
file_name=None,
removed_key_count=0,
hide_coordinates=True,
):
solvable = True
folder = f"generated_data/{folder_name}/"
if file_name is None:
for file in os.listdir(folder):
maze_loader = MazeLoader(
folder + file,
removed_key_count=removed_key_count,
solvable=solvable,
hide_coordinates=hide_coordinates,
)
break
else:
maze_loader = MazeLoader(
folder + file_name,
removed_key_count=removed_key_count,
solvable=solvable,
hide_coordinates=hide_coordinates,
)
nlp = maze_loader.encode_problem_into_nlp(shuffle_ratio=0.2, noise_ratio=0.0)
facts = nlp[1]
solution = maze_loader.solution_with_room_names
return facts, solution, maze_loader, file_name if file_name is not None else file
if __name__ == "__main__":
removed_key_count = 4
solvable = True
hide_coordinates = True
folder = "generated_data/maze_7_7_7_True/"
for file in os.listdir(folder):
if file not in ["874841.pkl"]:
continue
maze_loader = MazeLoader(
folder + file,
removed_key_count=removed_key_count,
solvable=solvable,
hide_coordinates=hide_coordinates,
)
break
from plot import pretty_plot_maze
# pretty_plot_maze(maze_loader)
nlp = maze_loader.encode_problem_into_nlp(shuffle_ratio=0.2, noise_ratio=1.0)
# colored print
print(f"\033[91mProvided Facts:\033[0m {nlp[2]}")
print(f"\033[92mSolution:\033[0m {maze_loader.solution_with_room_names}")
# print("Problem Prompt NLP: Compact List format", nlp[0])
# print("Distraction Weight: ", nlp[2])
# print("Shuffle Entropy: ", nlp[3])
|