test_semantics.py 文件源码

def test_describe_handling():
    env = gym.make('flashgames.DuskDrive-v0')
    env.configure(vnc_driver=FakeVNCSession, rewarder_driver=FakeRewarder, remotes='vnc://example.com:5900+15900')
    env.reset()

    reward_buffer = get_reward_buffer(env)
    rewarder_client = get_rewarder_client(env)

    rewarder_client._manual_recv('v0.env.describe', {'env_id': 'flashgames.DuskDrive-v0', 'env_state': 'resetting', 'fps': 60}, {'episode_id': '1'})

    assert reward_buffer._remote_episode_id == '1'
    assert reward_buffer._remote_env_state == 'resetting'
    assert reward_buffer._current_episode_id == None
    assert reward_buffer.reward_state(reward_buffer._current_episode_id)._env_state == None

    rewarder_client._manual_recv('v0.reply.env.reset', {}, {'episode_id': '1'})

    assert reward_buffer._remote_episode_id == '1'
    assert reward_buffer._remote_env_state == 'resetting'
    assert reward_buffer._current_episode_id == '1'
    assert reward_buffer.reward_state(reward_buffer._current_episode_id)._env_state == 'resetting'