-
Notifications
You must be signed in to change notification settings - Fork 1
/
pyhop.py
325 lines (266 loc) · 12.1 KB
/
pyhop.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
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
"""
Pyhop, version 1.2.2 -- a simple SHOP-like planner written in Python.
Author: Dana S. Nau, 2013.05.31
Copyright 2013 Dana S. Nau - http://www.cs.umd.edu/~nau
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
Pyhop should work correctly in both Python 2.7 and Python 3.2.
For examples of how to use it, see the example files that come with Pyhop.
Pyhop provides the following classes and functions:
- foo = State('foo') tells Pyhop to create an empty state object named 'foo'.
To put variables and values into it, you should do assignments such as
foo.var1 = val1
- bar = Goal('bar') tells Pyhop to create an empty goal object named 'bar'.
To put variables and values into it, you should do assignments such as
bar.var1 = val1
- print_state(foo) will print the variables and values in the state foo.
- print_goal(foo) will print the variables and values in the goal foo.
- declare_operators(o1, o2, ..., ok) tells Pyhop that o1, o2, ..., ok
are all of the planning operators; this supersedes any previous call
to declare_operators.
- print_operators() will print out the list of available operators.
- declare_methods('foo', m1, m2, ..., mk) tells Pyhop that m1, m2, ..., mk
are all of the methods for tasks having 'foo' as their taskname; this
supersedes any previous call to declare_methods('foo', ...).
- print_methods() will print out a list of all declared methods.
- pyhop(state1,tasklist) tells Pyhop to find a plan for accomplishing tasklist
(a list of tasks), starting from an initial state state1, using whatever
methods and operators you declared previously.
- In the above call to pyhop, you can add an optional 3rd argument called
'verbose' that tells pyhop how much debugging printout it should provide:
- if verbose = 0 (the default), pyhop returns the solution but prints nothing;
- if verbose = 1, it prints the initial parameters and the answer;
- if verbose = 2, it also prints a message on each recursive call;
- if verbose = 3, it also prints info about what it's computing.
"""
# Pyhop's planning algorithm is very similar to the one in SHOP and JSHOP
# (see http://www.cs.umd.edu/projects/shop). Like SHOP and JSHOP, Pyhop uses
# HTN methods to decompose tasks into smaller and smaller subtasks, until it
# finds tasks that correspond directly to actions. But Pyhop differs from
# SHOP and JSHOP in several ways that should make it easier to use Pyhop
# as part of other programs:
#
# (1) In Pyhop, one writes methods and operators as ordinary Python functions
# (rather than using a special-purpose language, as in SHOP and JSHOP).
#
# (2) Instead of representing states as collections of logical assertions,
# Pyhop uses state-variable representation: a state is a Python object
# that contains variable bindings. For example, to define a state in
# which box b is located in room r1, you might write something like this:
# s = State()
# s.loc['b'] = 'r1'
#
# (3) You also can define goals as Python objects. For example, to specify
# that a goal of having box b in room r2, you might write this:
# g = Goal()
# g.loc['b'] = 'r2'
# Like most HTN planners, Pyhop will ignore g unless you explicitly
# tell it what to do with g. You can do that by referring to g in
# your methods and operators, and passing g to them as an argument.
# In the same fashion, you could tell Pyhop to achieve any one of
# several different goals, or to achieve them in some desired sequence.
#
# (4) Unlike SHOP and JSHOP, Pyhop doesn't include a Horn-clause inference
# engine for evaluating preconditions of operators and methods. So far,
# I've seen no need for it; I've found it easier to write precondition
# evaluations directly in Python. But I could consider adding such a
# feature if someone convinces me that it's really necessary.
#
# Accompanying this file are several files that give examples of how to use
# Pyhop. To run them, launch python and type "import blocks_world_examples"
# or "import simple_travel_example".
from __future__ import print_function
import copy, sys, pprint
############################################################
# States and goals
class State():
"""A state is just a collection of variable bindings."""
def __init__(self, name):
self.__name__ = name
class Goal():
"""A goal is just a collection of variable bindings."""
def __init__(self, name):
self.__name__ = name
### print_state and print_goal are identical except for the name
def print_state(state, indent=4):
"""Print each variable in state, indented by indent spaces."""
if state != False:
for (name, val) in vars(state).items():
if name != '__name__':
for x in range(indent): sys.stdout.write(' ')
sys.stdout.write(state.__name__ + '.' + name)
print(' =', val)
else:
print('False')
def print_goal(goal, indent=4):
"""Print each variable in goal, indented by indent spaces."""
if goal != False:
for (name, val) in vars(goal).items():
if name != '__name__':
for x in range(indent): sys.stdout.write(' ')
sys.stdout.write(goal.__name__ + '.' + name)
print(' =', val)
else:
print('False')
############################################################
# Helper functions that may be useful in domain models
def forall(seq, cond):
"""True if cond(x) holds for all x in seq, otherwise False."""
for x in seq:
if not cond(x): return False
return True
def find_if(cond, seq):
"""
Return the first x in seq such that cond(x) holds, if there is one.
Otherwise return None.
"""
for x in seq:
if cond(x): return x
return None
############################################################
# Commands to tell Pyhop what the operators and methods are
operators = {}
methods = {}
def declare_operators(*op_list):
"""
Call this after defining the operators, to tell Pyhop what they are.
op_list must be a list of functions, not strings.
"""
operators.update({op.__name__: op for op in op_list})
return operators
def declare_methods(task_name, *method_list):
"""
Call this once for each task, to tell Pyhop what the methods are.
task_name must be a string.
method_list must be a list of functions, not strings.
"""
methods.update({task_name: list(method_list)})
return methods[task_name]
############################################################
# Commands to find out what the operators and methods are
def print_operators(olist=operators):
"""Print out the names of the operators"""
print('OPERATORS:', ', '.join(olist))
def print_methods(mlist=methods):
"""Print out a table of what the methods are for each task"""
print('{:<14}{}'.format('TASK:', '\t\t\t\tMETHODS:'))
for task in mlist:
print('{:<14}\t'.format(task) + ', '.join([f.__name__ for f in mlist[task]]))
############################################################
# The actual planner
def pyhop(state, tasks, verbose=0, all_plans=False, sort_asc=True):
"""
Try to find a plan that accomplishes tasks in state.
If successful, return the plan. Otherwise return False.
"""
if verbose > 0: print(
'** pyhop, verbose={}: **\n state = {}\n tasks = {}'.format(verbose, state.__name__, tasks))
results = seek_plans(state, tasks, [], 0, verbose, all_plans)
if verbose > 0:
if all_plans:
if isinstance(results, list):
if sort_asc:
# results.sort(key=len)
results = sorted(results)
for i in range(len(results)):
print('** result {} = {}'.format(i, results[i]))
else:
print('** result = {}'.format(results))
else:
print('** result = {}'.format(results))
return results
def seek_plans(state, tasks, plan, depth, verbose=0, all_plans=False):
"""
Workhorse for pyhop. state and tasks are as in pyhop.
- plan is the current partial plan.
- depth is the recursion depth, for use in debugging
- verbose is whether to print debugging messages
"""
plans = []
if verbose > 1: print('depth {} tasks {}'.format(depth, tasks))
if tasks == []:
if verbose > 2: print('depth {} returns plan {}'.format(depth, plan))
if all_plans:
plans.append(plan)
else:
return plan
task1 = tasks[0]
if task1[0] in operators:
if verbose > 2: print('depth {} action {}'.format(depth, task1))
operator = operators[task1[0]]
newstate = operator(copy.deepcopy(state), *task1[1:])
if verbose > 2:
print('depth {} new state:'.format(depth))
print_state(newstate)
if newstate:
solution = seek_plans(newstate, tasks[1:], plan + [task1], depth + 1, verbose)
if solution != False:
if all_plans:
plans.append(solution)
else:
return solution
if task1[0] in methods:
if verbose > 2: print('depth {} method instance {}'.format(depth, task1))
relevant = methods[task1[0]]
for method in relevant:
subtasks = method(state, *task1[1:])
# Can't just say "if subtasks:", because that's wrong if subtasks == []
if verbose > 2:
print('depth {} new tasks: {}'.format(depth, subtasks))
if subtasks != False:
solution = seek_plans(state, subtasks + tasks[1:], plan, depth + 1, verbose)
if solution != False:
if all_plans:
plans.append(solution)
else:
return solution
if all_plans:
if len(plans) > 0:
return plans
if verbose > 2: print('depth {} returns failure'.format(depth))
return False
# TODO: return all possible plans
# def seek_plan(state, tasks, plan, depth, verbose=0):
# """
# Workhorse for pyhop. state and tasks are as in pyhop.
# - plan is the current partial plan.
# - depth is the recursion depth, for use in debugging
# - verbose is whether to print debugging messages
# """
# if verbose > 1: print('depth {} tasks {}'.format(depth, tasks))
# if tasks == []:
# if verbose > 2: print('depth {} returns plan {}'.format(depth, plan))
# return plan
# task1 = tasks[0]
# if task1[0] in operators:
# if verbose > 2: print('depth {} action {}'.format(depth, task1))
# operator = operators[task1[0]]
# newstate = operator(copy.deepcopy(state), *task1[1:])
# if verbose > 2:
# print('depth {} new state:'.format(depth))
# print_state(newstate)
# if newstate:
# solution = seek_plan(newstate, tasks[1:], plan + [task1], depth + 1, verbose)
# if solution != False:
# return solution
# if task1[0] in methods:
# if verbose > 2: print('depth {} method instance {}'.format(depth, task1))
# relevant = methods[task1[0]]
# for method in relevant:
# subtasks = method(state, *task1[1:])
# # Can't just say "if subtasks:", because that's wrong if subtasks == []
# if verbose > 2:
# print('depth {} new tasks: {}'.format(depth, subtasks))
# if subtasks != False:
# solution = seek_plan(state, subtasks + tasks[1:], plan, depth + 1, verbose)
# if solution != False:
# return solution
# if verbose > 2: print('depth {} returns failure'.format(depth))
# return False