forked from DLR-RM/stable-baselines3
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_monitor.py
99 lines (82 loc) · 3.21 KB
/
test_monitor.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
import json
import os
import uuid
import gym
import pandas
from stable_baselines3.common.monitor import Monitor, get_monitor_files, load_results
def test_monitor(tmp_path):
"""
Test the monitor wrapper
"""
env = gym.make("CartPole-v1")
env.seed(0)
monitor_file = os.path.join(str(tmp_path), "stable_baselines-test-{}.monitor.csv".format(uuid.uuid4()))
monitor_env = Monitor(env, monitor_file)
monitor_env.reset()
total_steps = 1000
ep_rewards = []
ep_lengths = []
ep_len, ep_reward = 0, 0
for _ in range(total_steps):
_, reward, done, _ = monitor_env.step(monitor_env.action_space.sample())
ep_len += 1
ep_reward += reward
if done:
ep_rewards.append(ep_reward)
ep_lengths.append(ep_len)
monitor_env.reset()
ep_len, ep_reward = 0, 0
monitor_env.close()
assert monitor_env.get_total_steps() == total_steps
assert sum(ep_lengths) == sum(monitor_env.get_episode_lengths())
assert sum(monitor_env.get_episode_rewards()) == sum(ep_rewards)
_ = monitor_env.get_episode_times()
with open(monitor_file, "rt") as file_handler:
first_line = file_handler.readline()
assert first_line.startswith("#")
metadata = json.loads(first_line[1:])
assert metadata["env_id"] == "CartPole-v1"
assert set(metadata.keys()) == {"env_id", "t_start"}, "Incorrect keys in monitor metadata"
last_logline = pandas.read_csv(file_handler, index_col=None)
assert set(last_logline.keys()) == {"l", "t", "r"}, "Incorrect keys in monitor logline"
os.remove(monitor_file)
def test_monitor_load_results(tmp_path):
"""
test load_results on log files produced by the monitor wrapper
"""
tmp_path = str(tmp_path)
env1 = gym.make("CartPole-v1")
env1.seed(0)
monitor_file1 = os.path.join(tmp_path, "stable_baselines-test-{}.monitor.csv".format(uuid.uuid4()))
monitor_env1 = Monitor(env1, monitor_file1)
monitor_files = get_monitor_files(tmp_path)
assert len(monitor_files) == 1
assert monitor_file1 in monitor_files
monitor_env1.reset()
episode_count1 = 0
for _ in range(1000):
_, _, done, _ = monitor_env1.step(monitor_env1.action_space.sample())
if done:
episode_count1 += 1
monitor_env1.reset()
results_size1 = len(load_results(os.path.join(tmp_path)).index)
assert results_size1 == episode_count1
env2 = gym.make("CartPole-v1")
env2.seed(0)
monitor_file2 = os.path.join(tmp_path, "stable_baselines-test-{}.monitor.csv".format(uuid.uuid4()))
monitor_env2 = Monitor(env2, monitor_file2)
monitor_files = get_monitor_files(tmp_path)
assert len(monitor_files) == 2
assert monitor_file1 in monitor_files
assert monitor_file2 in monitor_files
monitor_env2.reset()
episode_count2 = 0
for _ in range(1000):
_, _, done, _ = monitor_env2.step(monitor_env2.action_space.sample())
if done:
episode_count2 += 1
monitor_env2.reset()
results_size2 = len(load_results(os.path.join(tmp_path)).index)
assert results_size2 == (results_size1 + episode_count2)
os.remove(monitor_file1)
os.remove(monitor_file2)