/usr/share/games/angrydd/ai.py is in angrydd 1.0.1-11.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
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 | # ai.py -- AI and helper routines for Angry, Drunken Dwarves
# Copyright 2004 Joe Wreschnig <piman@sacredchao.net>
# Released under the terms of the GNU GPL v2.
__revision__ = "$Id: ai.py 286 2004-09-04 03:51:59Z piman $"
# Trick borrowed from Peter Norvig's Python IAQ:
# http://www.norvig.com/python-iaq.html
def _abstract():
import inspect
caller = inspect.getouterframes(inspect.currentframe())[1][3]
raise NotImplementedError(caller + ' must be implemented in subclass')
import random, math
import util
from game import BasicField
from boxes import TickBox, BreakBox, Box, Special, Diamond
def copy_box(box):
args = [box.color, [box.x, box.y]]
newbox = None
if isinstance(box, Box):
newbox = Box(*args)
elif isinstance(box, TickBox):
newbox = TickBox(*args)
newbox.time_left = box.time_left
elif isinstance(box, BreakBox):
newbox = BreakBox(*args)
elif isinstance(box, Special):
newbox = Special(box.color, [box.x, box.y], box.special)
elif isinstance(box, Diamond):
newbox = Diamond(box.color, [box.x, box.y])
return newbox
class AIField(BasicField):
def __init__(self, field):
BasicField.__init__(self, field.width, field.height)
for y, row in enumerate(field):
for x, box in enumerate(row):
if field[y][x] is not None:
# Only add a box if we're at its topleft, i.e.
# don't add merged boxes many times.
newbox = copy_box(box)
if isinstance(newbox, Box):
if box.x != x or box.y != y: continue
else: newbox.set_size(box.size, self)
if newbox:
self.add_box([newbox.x, newbox.y], newbox)
# Take a turn (move all pieces down, tick, merge, break, repeat)
# Doesn't do boning. Returns the value of gems broken.
def take_turn(self):
while self.fall(): pass
self.tick()
while self.merge(): pass
value, dead = self.breaking()
while len(dead) > 0:
while self.fall(): pass
while self.merge(): pass
newv, dead = self.breaking()
value += newv
return value
def can_drop1(self, col):
return (self._field[0][col] is None)
# Orientation is as described in FallingBoxes#_render.
def can_drop(self, col, orientation):
if col >= self.width: return False
elif orientation == 0: # 2 above 1
return (self._field[0][col] is None)
elif orientation == 1:
return ((col != 0) and self._field[0][col] is None and
self._field[0][col - 1] is None)
elif orientation == 2:
return (self._field[0][col] is None and
self._field[1][col] is None)
elif orientation == 3:
return (col != self.width - 1 and
self._field[0][col] is None and
self._field[0][col + 1] is None)
else: return False
# Drop box1 and box2 in the given orientation at the top of this
# field. Return True (and does it if it thinks such a thing is
# possible, False otherwise. Possibility is currently only
# decided by space available at the top, and not any intervening
# blocks that might block it (i.e. naively)
def drop(self, boxes, col, orientation):
box1, box2 = boxes
if not self.can_drop(col, orientation): return False
elif orientation == 0: # 2 above 1
if self._field[1][col] is not None:
self.add_box([col, 0], box1)
else:
self.add_box([col, 1], box1)
self.add_box([col, 0], box2)
elif orientation == 1: # 2 left of 1
self.add_box([col, 0], box1)
self.add_box([col - 1, 0], box2)
elif orientation == 2: # 2 below 1
self.add_box([col, 0], box1)
self.add_box([col, 1], box2)
elif orientation == 3: # 2 right of 1
self.add_box([col, 0], box1)
self.add_box([col + 1, 0], box2)
return True
def drop1(self, box, col):
if not self.can_drop1(col): return False
else:
self.add_box([col, 0], box)
return True
def get_height(self, col):
for y in range(self.height):
if self._field[y][col] != None: return self.height - y
else: return 0
def max_height(self):
return max([self.get_height(i) for i in range(self.width)])
def average_height(self):
s = 4 * self.get_height(3) # This is very bad to fill up
s += sum([self.get_height(x) for x in range(self.width)])
return s / self.width
def stddev_height(self):
avg = self.average_height()
std_dev = 0
for i in range(self.width):
v = avg - self.get_height(i)
std_dev += v * v
std_dev /= float(self.width)
return math.sqrt(std_dev)
# Expand given only one box; total states = field.width.
def expand1(self, box, rand = True):
columns = range(self.width)
if rand: random.shuffle(columns)
for col in columns:
if self.can_drop1(col):
f = AIField(self)
newbox = copy_box(box)
f.drop1(newbox, col)
yield 0, col, f, [newbox]
# Expand the state space based on the boxes given. 4 orientations,
# width of 6 = 22 expansions (after illegal ones removed).
def expand(self, boxes, rand = True):
orientations = range(4)
columns = range(self.width)
if rand:
random.shuffle(orientations)
random.shuffle(columns)
columns.remove(3); columns.append(3)
for orientation in orientations:
for col in columns:
if self.can_drop(col, orientation):
f = AIField(self)
newboxes = [copy_box(b) for b in boxes]
f.drop(newboxes, col, orientation)
yield orientation, col, f, newboxes
class AI(object):
def __init__(self, player):
self.player = player
self.insane = False
self.drops = 70
self.delta = 120
# Expand and find the highest value for the given callback; it
# defaults to the value of the gems broken if no callback is
# passed in.
def find_high(self, h = (lambda *args: args[1])):
best_score = None
best_move = (0, 0)
for orient, col, field, boxes in self.field.expand(self.falling):
fscore = field.take_turn()
score = h(orient, col, field, boxes, fscore)
if best_score is None or score > best_score:
best_score = score
best_move = (orient, col)
yield None
yield best_move
return
# Expand and find the lowest value for the given callback; if
# the value is 0, it returns it and doesn't expand anymore.
def find_low(self, h, halt_on_zero = True):
best_score = None
best_move = (0, 0)
for orient, col, field, boxes in self.field.expand(self.falling):
fscore = field.take_turn()
score = h(orient, col, field, boxes, fscore)
if best_score is None or score < best_score:
best_score = score
best_move = (orient, col)
if best_score == 0 and halt_on_zero: break
yield None
yield best_move
return
def incoming(self, field, falling):
self.field = AIField(field)
self.falling = [falling._box1, falling._box2]
def __iter__(self):
# Override this function to change how the AI moves. It should
# return an iterator (or generator) that returns either None
# when it's still thinking, or a tuple (orientation, column)
# to put the piece in when it's done. This function is called
# when a new piece is put into play, and after a move is made
# is not called again until another piece is put in. (i.e.
# you can't change your mind.)
# Note that once your moves are committed, each left/right/rotate
# motion will take a (nontrivial) fraction of a second! Around
# 1/12th of a second per move. So, thinking fast is important.
# Look at the RandomAI for a trivial example.
_abstract()
class SlowAI(AI):
def __init__(self, *args):
AI.__init__(self, *args)
self.drops = 0
self.delta = 500
class MediumAI(AI):
def __init__(self, *args):
AI.__init__(self, *args)
self.drops = 100
self.delta = 150
class FastAI(AI):
def __init__(self, *args):
AI.__init__(self, *args)
self.drops = 80
self.delta = 100
class SuperFastAI(AI):
def __init__(self, *args):
AI.__init__(self, *args)
self.drops = 25
self.delta = 25
# A test AI that plays totally randomly.
class RandomAI(AI):
def __iter__(self):
yield None
yield (random.randrange(0, 4),
random.randrange(0, self.player.field.width))
# Another test AI, just tries to keep average field height low
class KeepLowAI(AI):
def __iter__(self):
return self.find_low(self._total_height)
def _total_height(self, orient, col, field, boxes, score):
return sum([field.get_height(x) for x in range(field.width)])
class GoodAI(AI):
def __iter__(self):
if self.field.get_height(3) > 8:
return self.find_low(self._emergency)
elif self.field.max_height() > 9:
return self.find_high(self._lower_max)
elif self.field.average_height() < 7:
return self.find_high(self._form_boxes)
else:
return self.find_high()
def _emergency(self, orient, col, field, boxes, score):
return (score > 25, field.get_height(3), score)
def _lower_max(self, orient, col, field, boxes, score):
return (
score > 25,
-field.max_height(),
score,
self._gemsize(field))
def _form_boxes(self, orient, col, field, boxes, score):
return (
score > 25,
field.max_height() < 10,
self._gemsize(field),
self._adjacencies(field, boxes))
def _adjacencies(self, field, boxes):
adj = 0
for box in filter(None, boxes):
if not box.crashed:
for adjbox in filter(None, box.adjacent(field)):
if adjbox.color == box.color: adj += 3
else: adj -= 1
return adj
def _gemsize(self, field):
return max([1] + [box.size[0] * box.size[1]
for box in util.flatten(field)
if box])
class StupidAI(RandomAI, SlowAI): pass
class VeryEasyAI(KeepLowAI, SlowAI): pass
class EasyAI(KeepLowAI, MediumAI): pass
class NormalAI(GoodAI, MediumAI): pass
class HardAI(GoodAI, FastAI): pass
class InsaneAI(GoodAI, SuperFastAI): pass
|