def __init__()

in robosumo/envs/sumo.py [0:0]

• 56 lines of code
• 13 McCabe index (conditional complexity)

    def __init__(self, agent_names,
                 xml_path=None,
                 init_pos_noise=.1,
                 init_vel_noise=.1,
                 agent_kwargs=None,
                 frame_skip=5,
                 tatami_size=2.0,
                 timestep_limit=500,
                 **kwargs):
        EzPickle.__init__(self)
        self._tatami_size = tatami_size + 0.1
        self._timestep_limit = timestep_limit
        self._init_pos_noise = init_pos_noise
        self._init_vel_noise = init_vel_noise
        self._n_agents = len(agent_names)
        self._mujoco_init = False
        self._num_steps = None
        self._spec = None

        # Resolve agent scopes
        agent_scopes = [
            "%s%d" % (name, i)
            for i, name in enumerate(agent_names)
        ]

        # Consturct scene XML
        scene_xml_path = os.path.join(os.path.dirname(__file__),
                                      "assets", "tatami.xml")
        agent_xml_paths = [_AGENTS[name] for name in agent_names]
        scene = construct_scene(scene_xml_path, agent_xml_paths,
                                agent_scopes=agent_scopes,
                                tatami_size=tatami_size,
                                **kwargs)

        # Init MuJoCo
        if xml_path is None:
            with tempfile.TemporaryDirectory() as tmpdir_name:
                scene_filepath = os.path.join(tmpdir_name, "scene.xml")
                scene.write(scene_filepath)
                MujocoEnv.__init__(self, scene_filepath, frame_skip)
        else:
            with open(xml_path, 'w') as fp:
                scene.write(fp.name)
            MujocoEnv.__init__(self, fp.name, frame_skip)
        self._mujoco_init = True

        # Construct agents
        agent_kwargs = agent_kwargs or {}
        self.agents = [
            agents.get(name, env=self, scope=agent_scopes[i], **agent_kwargs)
            for i, name in enumerate(agent_names)
        ]

        # Set opponents
        for i, agent in enumerate(self.agents):
            agent.set_opponents([
                agent for j, agent in enumerate(self.agents) if j != i
            ])

        # Setup agents
        for i, agent in enumerate(self.agents):
            agent.setup_spaces()

        # Set observation and action spaces
        self.observation_space = Tuple([
            agent.observation_space for agent in self.agents
        ])
        self.action_space = Tuple([
            agent.action_space for agent in self.agents
        ])