/
Planner_env.py
198 lines (191 loc) · 4.86 KB
/
Planner_env.py
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
from __future__ import print_function
from pyddl import Domain, Problem, Action, neg, planner
#import sqlite3
#import timeit
#import time
from Grounder import Grounder
from sklearn.preprocessing import OneHotEncoder
import numpy as np
from pyddl.planner import monotone_heuristic
class Planner_env():
#define problem & domain in init
def __init__(self):
#self.con = sqlite3.connect("story_demo.db")
self.domain = Domain((
Action(
'go',
parameters=(
('character', 'p'),
('location', 'l1'),
('location', 'l2'),
),
preconditions=(
('at', 'p', 'l1'),
),
effects = (
('at', 'p', 'l2'),
neg(('at', 'p', 'l1')),
),
),
Action(
'purchase',
parameters=(
('character', 'p'),
('product', 'pr'),
('location', 'l'),
),
preconditions=(
('at', 'p', 'l'),
('instock','pr','l'),
('want','p','pr'),
('need','p','pr'),
),
effects=(
('purchased', 'p', 'pr'),
('have', 'p', 'pr'),
neg(('instock', 'pr', 'l')),
('outstock', 'pr', 'l'),
neg(('need','p','pr')),
),
),
Action(
'fail_buy',
parameters=(
('character', 'p'),
('product', 'pr'),
('location', 'l'),
),
preconditions=(
('at', 'p', 'l'),
('outstock', 'pr', 'l'),
('want','p','pr'),
('need','p','pr'),
),
effects=(
('failedbuy', 'p', 'pr'),
),
),
Action(
'refund',
parameters=(
('character', 'p'),
('product', 'pr'),
('location', 'l'),
),
preconditions=(
('at', 'p', 'l'),
('have','p','pr'),
('outstock', 'pr', 'l'),
),
effects=(
('instock', 'pr', 'l'),
neg(('have','p','pr')),
neg(('outstock', 'pr', 'l')),
neg(('purchased','p','pr')),
),
),
Action(
'change_mind',
parameters=(
('character', 'p'),
('product', 'pr'),
),
preconditions=(
('have','p','pr'),
('want', 'p', 'pr'),
),
effects=(
neg(('want', 'p', 'pr')),
),
)
))
self.problem = Problem(
self.domain,
{
'location': ('Ahome','Shop','Bhome'),
'character': ('A', 'B'),
'product': ('D'),
},
init=(
('at', 'A', 'Ahome'),
('at', 'B', 'Bhome'),
('want','A','D'),
('want','B','D'),
('need','A','D'),
('need','B','D'),
('instock', 'D', 'Shop'),
),
goal=(
('failedbuy', 'A', 'D'),
('at', 'B', 'Shop'),
('instock', 'D', 'Shop'),
('at', 'A', 'Ahome'),
('need','A','D'),
),
)
#self.actions={}
#self.ground_actions()
grounder=Grounder(self)
self.predicate_comb=grounder.get_predicate_combination(self.problem,self.domain)
self.action_comb=[str(action) for action in self.problem.grounded_actions]
self.num_states=len(self.predicate_comb)
self.num_actions=len(self.action_comb)
self.init_state=self.problem.initial_state
print(self.predicate_comb)
#print(self.action_comb)
#print(self.actions)
#self.state_binarizer = OneHotEncoder(categories=self.predicate_comb)
#self.state_binarizer.fit(self.predicate_comb)
self.goal = (self.problem.goals, self.problem.num_goals)
self.goal_set=set(self.problem.goals)
#print('goals',self.problem.goals,type(self.problem.goals))
def ground_actions(self):
#self.initialize_problem()
#print("problem initiated")
self.actions={}
for action in self.problem.grounded_actions:
#print("name",action.name,"prec",action.preconditions)
if(action.name not in self.actions):
self.actions[action.name]=[]
self.actions[action.name].append(action)
return
def reset(self):
#return to inital state
return self.init_state
def state_to_onehot(self,state):
#print('state',state.predicates)
onehot=np.zeros(self.num_states)
for i in range(self.num_states):
if self.predicate_comb[i] in state.predicates:
onehot[i]=1
return onehot
def apply_action(self,node,action_idx):
applied=False
goal_reached=False
target_action=self.action_comb[action_idx]
action_name=target_action[:target_action.find('(')]
action_params=target_action[target_action.find('(')+1:-1]
#print(action_name)
#print(action_params)
#print(tuple((str(action_name)+", "+str(action_params)).split(', ')))
for action in self.actions[action_name]:
#print(action.sig)
if action.sig==tuple((str(action_name)+", "+str(action_params)).split(', ')):
#print("found action:",action.sig)
#print("my action",(str(row[1])+","+str(row[2])),end=",")
if node.is_true(action.preconditions,action.num_preconditions):
node=node.apply(action)
applied=True
#print('Applied')
#check goal!!!!
if node.is_true(*self.goal):
goal_reached=True
return node,applied,goal_reached
else:
#print("Cannot apply action")
goal_reached=True
break
return node,applied,goal_reached
def simple_heuristic(self,node):
inter=self.goal_set.intersection(node.predicates)
return len(inter)