def step()

in envs/thor_beacons.py [0:0]

• 18 lines of code
• 11 McCabe index (conditional complexity)

    def step(self, action):

        if 'image_to_world' not in self.state.metadata:
            self.state.metadata['image_to_world'] = self.projector.image_to_world(self.agent_pose(self.state), self.state.depth_frame)

        prev_state = {'depth': np.array(self.state.depth_frame),
                      'P': self.state.metadata['image_to_world'].clone(),
                      'agent_y': self.state.metadata['agent']['position']['y']
                      }

        obs, rewards, done, info = super().step(action)

        # place beacon for interaction
        if info['action'] in self.interaction_set:
            beacon = self.get_beacon_position(prev_state, info)
            if beacon is not None:
                # place_beacon is called AFTER the action is executed. Place for previous time-step
                self.beacons.append({'t':self.t-1, 'action': info['action'], 'success': info['success'], **beacon})

        # get image_to_world for last timestep
        self.state.metadata['image_to_world'] = self.projector.image_to_world(self.agent_pose(self.state), self.state.depth_frame)

        hist = self.compile_history(self.state)
        self.history.append({'t':self.t-1, **hist})

        # if final timestep, then compute masks for the entire trajectory and add to info
        if self.t==self.max_t:
            info['traj_masks'] = self.compute_masks()

        return obs, rewards, done, info