gym-csle-stopping-game 0.2.19__py3-none-any.whl → 0.2.20__py3-none-any.whl

gym_csle_stopping_game/__version__.py

@@ -1 +1 @@
-__version__ = '0.2.19'
+__version__ = '0.2.20'

gym_csle_stopping_game/dao/stopping_game_attacker_mdp_config.py

@@ -37,7 +37,6 @@ class StoppingGameAttackerMdpConfig(SimulationEnvInputConfig):
         :param d: the dict to convert
         :return: the created instance
         """
-        defender_strategy = None
         try:
             defender_strategy = MultiThresholdStoppingPolicy.from_dict(d["defender_strategy"])
         except Exception:
@@ -58,7 +57,7 @@ class StoppingGameAttackerMdpConfig(SimulationEnvInputConfig):
 
         :return: a dict representation of the object
         """
-        d = {}
+        d: Dict[str, Any] = {}
         d["stopping_game_config"] = self.stopping_game_config.to_dict()
         d["defender_strategy"] = self.defender_strategy.to_dict()
         d["stopping_game_name"] = self.stopping_game_name

gym_csle_stopping_game/dao/stopping_game_config.py

@@ -1,6 +1,7 @@
 from typing import Dict, Any
 import gymnasium as gym
 import numpy as np
+import numpy.typing as npt
 from csle_common.dao.simulation_config.simulation_env_input_config import SimulationEnvInputConfig
 
 
@@ -10,8 +11,10 @@ class StoppingGameConfig(SimulationEnvInputConfig):
     """
 
     def __init__(self, env_name: str,
-                 T: np.ndarray, O: np.ndarray, Z: np.ndarray, R: np.ndarray, S: np.ndarray, A1: np.ndarray,
-                 A2: np.ndarray, L: int, R_INT: int, R_COST: int, R_SLA: int, R_ST: int, b1: np.ndarray,
+                 T: npt.NDArray[Any], O: npt.NDArray[np.int_], Z: npt.NDArray[Any],
+                 R: npt.NDArray[Any], S: npt.NDArray[np.int_], A1: npt.NDArray[np.int_],
+                 A2: npt.NDArray[np.int_], L: int, R_INT: int, R_COST: int, R_SLA: int, R_ST: int,
+                 b1: npt.NDArray[np.float_],
                  save_dir: str, checkpoint_traces_freq: int, gamma: float = 1) -> None:
         """
         Initializes the DTO
@@ -59,7 +62,7 @@ class StoppingGameConfig(SimulationEnvInputConfig):
 
         :return: a dict representation of the object
         """
-        d = {}
+        d: Dict[str, Any] = {}
         d["T"] = list(self.T.tolist())
         d["O"] = list(self.O.tolist())
         d["Z"] = list(self.Z.tolist())

gym_csle_stopping_game/dao/stopping_game_defender_pomdp_config.py

@@ -38,7 +38,6 @@ class StoppingGameDefenderPomdpConfig(SimulationEnvInputConfig):
         :param d: the dict to convert
         :return: the created instance
         """
-        attacker_strategy = None
         try:
             attacker_strategy = MultiThresholdStoppingPolicy.from_dict(d["attacker_strategy"])
         except Exception:
@@ -63,7 +62,7 @@ class StoppingGameDefenderPomdpConfig(SimulationEnvInputConfig):
         
         :return: a dict representation of the object
         """
-        d = {}
+        d: Dict[str, Any] = {}
         d["stopping_game_config"] = self.stopping_game_config.to_dict()
         d["attacker_strategy"] = self.attacker_strategy.to_dict()
         d["stopping_game_name"] = self.stopping_game_name

gym_csle_stopping_game/dao/stopping_game_state.py

@@ -1,5 +1,6 @@
 from typing import Dict, Any
 import numpy as np
+import numpy.typing as npt
 from gym_csle_stopping_game.util.stopping_game_util import StoppingGameUtil
 from csle_base.json_serializable import JSONSerializable
 
@@ -9,7 +10,7 @@ class StoppingGameState(JSONSerializable):
     Represents the state of the optimal stopping game
     """
 
-    def __init__(self, b1: np.ndarray, L: int) -> None:
+    def __init__(self, b1: npt.NDArray[np.float_], L: int) -> None:
         """
         Intializes the state
 
@@ -34,13 +35,13 @@ class StoppingGameState(JSONSerializable):
         self.s = StoppingGameUtil.sample_initial_state(b1=self.b1)
         self.b = self.b1.copy()
 
-    def attacker_observation(self) -> np.ndarray:
+    def attacker_observation(self) -> npt.NDArray[Any]:
         """
         :return: the attacker's observation
         """
         return np.array([self.l, self.b[1], self.s])
 
-    def defender_observation(self) -> np.ndarray:
+    def defender_observation(self) -> npt.NDArray[Any]:
         """
         :return: the defender's observation
         """
@@ -73,7 +74,7 @@ class StoppingGameState(JSONSerializable):
         
         :return: a dict representation of the object
         """
-        d = {}
+        d: Dict[str, Any] = {}
         d["L"] = self.L
         d["b1"] = list(self.b1)
         d["b"] = list(self.b)

gym_csle_stopping_game/envs/stopping_game_env.py

@@ -1,5 +1,6 @@
-from typing import Tuple, Dict, Union, List, Any
+from typing import Tuple, Dict, List, Any
 import numpy as np
+import numpy.typing as npt
 import time
 import math
 import csle_common.constants.constants as constants
@@ -59,15 +60,15 @@ class StoppingGameEnv(BaseEnv):
         }
 
         # Setup traces
-        self.traces = []
+        self.traces: List[SimulationTrace] = []
         self.trace = SimulationTrace(simulation_env=self.config.env_name)
 
         # Reset
         self.reset()
         super().__init__()
 
-    def step(self, action_profile: Tuple[int, Tuple[np.ndarray, int]]) \
-            -> Tuple[Tuple[np.ndarray, np.ndarray], Tuple[int, int], bool, bool, dict]:
+    def step(self, action_profile: Tuple[int, Tuple[npt.NDArray[Any], int]]) \
+            -> Tuple[Tuple[npt.NDArray[Any], npt.NDArray[Any]], Tuple[int, int], bool, bool, Dict[str, Any]]:
         """
         Takes a step in the environment by executing the given action
 
@@ -81,7 +82,7 @@ class StoppingGameEnv(BaseEnv):
         assert pi2.shape[0] == len(self.config.S)
         assert pi2.shape[1] == len(self.config.A1)
         done = False
-        info = {}
+        info: Dict[str, Any] = {}
 
         # Compute r, s', b',o'
         r = self.config.R[self.state.l - 1][a1][a2][self.state.s]
@@ -134,8 +135,8 @@ class StoppingGameEnv(BaseEnv):
 
         return (defender_obs, attacker_obs), (r, -r), done, done, info
 
-    def step_test(self, action_profile: Tuple[int, Tuple[np.ndarray, int]], sample_Z) \
-            -> Tuple[Tuple[np.ndarray, np.ndarray], Tuple[int, int], bool, dict]:
+    def step_test(self, action_profile: Tuple[int, Tuple[npt.NDArray[Any], int]], sample_Z) \
+            -> Tuple[Tuple[npt.NDArray[Any], npt.NDArray[Any]], Tuple[int, int], bool, Dict[str, Any]]:
         """
         Takes a step in the environment by executing the given action
 
@@ -149,7 +150,7 @@ class StoppingGameEnv(BaseEnv):
         assert pi2.shape[0] == len(self.config.S)
         assert pi2.shape[1] == len(self.config.A1)
         done = False
-        info = {}
+        info: Dict[str, Any] = {}
 
         # Compute r, s', b',o'
         r = self.config.R[self.state.l - 1][a1][a2][self.state.s]
@@ -201,8 +202,8 @@ class StoppingGameEnv(BaseEnv):
 
         return (defender_obs, attacker_obs), (r, -r), done, info
 
-    def step_trace(self, trace: EmulationTrace, a1: int, pi2: np.ndarray) \
-            -> Tuple[Tuple[np.ndarray, np.ndarray], Tuple[int, int], bool, dict]:
+    def step_trace(self, trace: EmulationTrace, a1: int, pi2: npt.NDArray[Any]) \
+            -> Tuple[Tuple[npt.NDArray[Any], npt.NDArray[Any]], Tuple[int, int], bool, Dict[str, Any]]:
         """
         Utility function for stepping a given trace
 
@@ -212,7 +213,7 @@ class StoppingGameEnv(BaseEnv):
         :return: the result of the step
         """
         done = False
-        info = {}
+        info: Dict[str, Any] = {}
         if (self.state.t - 1) < len(trace.attacker_actions):
             a2_emulation_action = trace.attacker_actions[self.state.t - 1]
             a2 = 0
@@ -300,7 +301,7 @@ class StoppingGameEnv(BaseEnv):
         else:
             return 1 - (min(10, (first_stop - (intrusion_start + 1))) / 2) / 10
 
-    def _info(self, info) -> Dict[str, Union[float, int]]:
+    def _info(self, info: Dict[str, Any]) -> Dict[str, Any]:
         """
         Adds the cumulative reward and episode length to the info dict
 
@@ -363,7 +364,8 @@ class StoppingGameEnv(BaseEnv):
             defender_baseline_stop_on_first_alert_return
         return info
 
-    def reset(self, seed: int = 0, soft: bool = False) -> Tuple[Tuple[np.ndarray, np.ndarray], Dict[str, Any]]:
+    def reset(self, seed: int = 0, soft: bool = False) \
+            -> Tuple[Tuple[npt.NDArray[Any], npt.NDArray[Any]], Dict[str, Any]]:
         """
         Resets the environment state, this should be called whenever step() returns <done>
 
@@ -378,7 +380,7 @@ class StoppingGameEnv(BaseEnv):
         defender_obs = self.state.defender_observation()
         self.trace.attacker_observations.append(attacker_obs)
         self.trace.defender_observations.append(defender_obs)
-        info = {}
+        info: Dict[str, Any] = {}
         return (defender_obs, attacker_obs), info
 
     @staticmethod
@@ -408,7 +410,7 @@ class StoppingGameEnv(BaseEnv):
             done = False
             defender_obs_space = simulation_env_config.joint_observation_space_config.observation_spaces[0]
             b = env.state.b1
-            o = env.reset()
+            o, _ = env.reset()
             (d_obs, a_obs) = o
             t = 0
             s.reset()
@@ -419,7 +421,7 @@ class StoppingGameEnv(BaseEnv):
             while not done:
                 a1 = defender_policy.action(d_obs)
                 a2 = attacker_policy.action(a_obs)
-                o, r, done, info = env.step((a1, a2))
+                o, r, done, info, _ = env.step((a1, a2))
                 (d_obs, a_obs) = o
                 r_1, r_2 = r
                 logger.debug(f"a1:{a1}, a2:{a2}, d_obs:{d_obs}, a_obs:{a_obs}, r:{r}, done:{done}, info: {info}")
@@ -448,12 +450,12 @@ class StoppingGameEnv(BaseEnv):
                              f"{defender_obs_space.observation_id_to_observation_vector_inv}")
                 logger.debug(f"observation_id_to_observation_vector_inv:"
                              f"{o_components_str in defender_obs_space.observation_id_to_observation_vector_inv}")
+                emulation_o = 0
                 if o_components_str in defender_obs_space.observation_id_to_observation_vector_inv:
-                    o = defender_obs_space.observation_id_to_observation_vector_inv[o_components_str]
-                else:
-                    o = 0
-                logger.debug(f"o:{o}")
-                b = StoppingGameUtil.next_belief(o=o, a1=a1, b=b, pi2=a2, config=env.config, l=env.state.l, a2=a2)
+                    emulation_o = defender_obs_space.observation_id_to_observation_vector_inv[o_components_str]
+                logger.debug(f"o:{emulation_o}")
+                b = StoppingGameUtil.next_belief(o=emulation_o, a1=a1, b=b, pi2=a2, config=env.config,
+                                                 l=env.state.l, a2=a2)
                 d_obs[1] = b[1]
                 a_obs[1] = b[1]
                 logger.debug(f"b:{b}")
@@ -464,7 +466,7 @@ class StoppingGameEnv(BaseEnv):
                 simulation_trace.infos.append(info)
                 simulation_trace.states.append(s)
                 simulation_trace.beliefs.append(b[1])
-                simulation_trace.infrastructure_metrics.append(o)
+                simulation_trace.infrastructure_metrics.append(emulation_o)
 
             em_sim_trace = EmulationSimulationTrace(emulation_trace=emulation_trace, simulation_trace=simulation_trace)
             MetastoreFacade.save_emulation_simulation_trace(em_sim_trace)
@@ -556,10 +558,10 @@ class StoppingGameEnv(BaseEnv):
                 stage_policy = []
                 for s in self.config.S:
                     if s != 2:
-                        dist = [0, 0]
-                        dist[a2] = 1
+                        dist = [0.0, 0.0]
+                        dist[a2] = 1.0
                         stage_policy.append(dist)
                     else:
                         stage_policy.append([0.5, 0.5])
-                stage_policy = np.array(stage_policy)
-                _, _, done, _ = self.step(action_profile=(a1, (stage_policy, a2)))
+                pi2 = np.array(stage_policy)
+                _, _, done, _, _ = self.step(action_profile=(a1, (pi2, a2)))

gym_csle_stopping_game/envs/stopping_game_mdp_attacker_env.py

@@ -1,6 +1,6 @@
 from typing import Tuple, List, Union, Dict, Any
-import gymnasium as gym
 import numpy as np
+import numpy.typing as npt
 import torch
 import math
 from csle_common.dao.simulation_config.base_env import BaseEnv
@@ -9,6 +9,7 @@ from gym_csle_stopping_game.dao.stopping_game_attacker_mdp_config import Stoppin
 from csle_common.dao.simulation_config.simulation_trace import SimulationTrace
 from gym_csle_stopping_game.util.stopping_game_util import StoppingGameUtil
 import gym_csle_stopping_game.constants.constants as env_constants
+from gym_csle_stopping_game.envs.stopping_game_env import StoppingGameEnv
 
 
 class StoppingGameMdpAttackerEnv(BaseEnv):
@@ -23,7 +24,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         :param config: the configuration of the environment
         """
         self.config = config
-        self.stopping_game_env = gym.make(self.config.stopping_game_name, config=self.config.stopping_game_config)
+        self.stopping_game_env: StoppingGameEnv = StoppingGameEnv(config=self.config.stopping_game_config)
 
         # Setup spaces
         self.observation_space = self.config.stopping_game_config.attacker_observation_space()
@@ -33,41 +34,47 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         self.static_defender_strategy = self.config.defender_strategy
 
         # Setup Config
-        self.viewer = None
+        self.viewer: Union[None, Any] = None
         self.metadata = {
             'render.modes': ['human', 'rgb_array'],
             'video.frames_per_second': 50  # Video rendering speed
         }
 
-        self.latest_defender_obs = None
-        self.latest_attacker_obs = None
-        self.model = None
+        self.latest_defender_obs: Union[None, List[Any], npt.NDArray[Any]] = None
+        self.latest_attacker_obs: Union[None, List[Any], npt.NDArray[Any]] = None
+        self.model: Union[None, Any] = None
 
         # Reset
         self.reset()
         super().__init__()
 
-    def step(self, pi2: Union[List[List[float]], int, float, np.int64, float, np.float64]) \
-            -> Tuple[np.ndarray, int, bool, bool, dict]:
+    def step(self, pi2: Union[npt.NDArray[Any], int, float, np.int_, np.float_]) \
+            -> Tuple[npt.NDArray[Any], int, bool, bool, Dict[str, Any]]:
         """
         Takes a step in the environment by executing the given action
 
         :param pi2: attacker stage policy
         :return: (obs, reward, terminated, truncated, info)
         """
-        if type(pi2) is int or type(pi2) is float or type(pi2) is np.int64 or type(pi2) is float \
-                or type(pi2) is np.float64:
+        if type(pi2) is int or type(pi2) is float or type(pi2) is np.int64 or type(pi2) is np.float64:
             a2 = pi2
-            pi2 = self.calculate_stage_policy(o=self.latest_attacker_obs, a2=a2)
+            if self.latest_attacker_obs is None:
+                raise ValueError("Attacker observation is None")
+            pi2 = self.calculate_stage_policy(o=list(self.latest_attacker_obs), a2=int(a2))
         else:
             if self.model is not None:
-                pi2 = self.calculate_stage_policy(o=self.latest_attacker_obs)
+                if self.latest_attacker_obs is None:
+                    raise ValueError("Attacker observation is None")
+                pi2 = self.calculate_stage_policy(o=list(self.latest_attacker_obs))
                 a2 = StoppingGameUtil.sample_attacker_action(pi2=pi2, s=self.stopping_game_env.state.s)
             else:
                 pi2 = np.array(pi2)
-                if (not pi2.shape[0] == len(self.config.stopping_game_config.S)
-                        or pi2.shape[1] != len(self.config.stopping_game_config.A1)) and self.model is not None:
-                    pi2 = self.calculate_stage_policy(o=self.latest_attacker_obs)
+                try:
+                    if self.latest_attacker_obs is None:
+                        raise ValueError("Attacker observation is None")
+                    pi2 = self.calculate_stage_policy(o=list(self.latest_attacker_obs))
+                except Exception:
+                    pass
                 a2 = StoppingGameUtil.sample_attacker_action(pi2=pi2, s=self.stopping_game_env.state.s)
 
         # a2 = pi2
@@ -83,7 +90,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         a1 = self.static_defender_strategy.action(o=self.latest_defender_obs)
 
         # Step the game
-        o, r, d, _, info = self.stopping_game_env.step((a1, (pi2, a2)))
+        o, r, d, _, info = self.stopping_game_env.step((int(a1), (pi2, int(a2))))
         self.latest_defender_obs = o[0]
         self.latest_attacker_obs = o[1]
         attacker_obs = o[1]
@@ -94,7 +101,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
 
         return attacker_obs, r[1], d, d, info
 
-    def reset(self, seed: int = 0, soft: bool = False) -> Tuple[np.ndarray, Dict[str, Any]]:
+    def reset(self, seed: int = 0, soft: bool = False) -> Tuple[npt.NDArray[Any], Dict[str, Any]]:
         """
         Resets the environment state, this should be called whenever step() returns <done>
 
@@ -104,7 +111,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         self.latest_defender_obs = o[0]
         self.latest_attacker_obs = o[1]
         attacker_obs = o[1]
-        info = {}
+        info: Dict[str, Any] = {}
         return attacker_obs, info
 
     def set_model(self, model) -> None:
@@ -116,7 +123,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         """
         self.model = model
 
-    def calculate_stage_policy(self, o: List, a2: int = 0) -> np.ndarray:
+    def calculate_stage_policy(self, o: List[Any], a2: int = 0) -> npt.NDArray[Any]:
         """
         Calculates the stage policy of a given model and observation
 
@@ -127,15 +134,14 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
             stage_policy = []
             for s in self.config.stopping_game_config.S:
                 if s != 2:
-                    dist = [0, 0]
-                    dist[a2] = 1
+                    dist = [0.0, 0.0]
+                    dist[a2] = 1.0
                     stage_policy.append(dist)
                 else:
                     stage_policy.append([0.5, 0.5])
             return np.array(stage_policy)
         if isinstance(self.model, MixedMultiThresholdStoppingPolicy):
-            stage_policy = np.array(self.model.stage_policy(o=o))
-            return stage_policy
+            return np.array(self.model.stage_policy(o=o))
         else:
             b1 = o[1]
             l = int(o[0])
@@ -146,18 +152,19 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
                     stage_policy.append(self._get_attacker_dist(obs=o))
                 else:
                     stage_policy.append([0.5, 0.5])
-            stage_policy = np.array(stage_policy)
-            return stage_policy
+            return np.array(stage_policy)
 
-    def _get_attacker_dist(self, obs: List) -> List:
+    def _get_attacker_dist(self, obs: List[Any]) -> List[float]:
         """
         Utility function for getting the attacker's action distribution based on a given observation
 
         :param obs: the given observation
         :return:  the action distribution
         """
-        obs = np.array([obs])
-        actions, values, log_prob = self.model.policy.forward(obs=torch.tensor(obs).to(self.model.device))
+        np_obs = np.array([obs])
+        if self.model is None:
+            raise ValueError("Model is None")
+        actions, values, log_prob = self.model.policy.forward(obs=torch.tensor(np_obs).to(self.model.device))
         action = actions[0]
         if action == 1:
             stop_prob = math.exp(log_prob)
@@ -211,7 +218,7 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
         Closes the viewer (cleanup)
         :return: None
         """
-        if self.viewer:
+        if self.viewer is not None:
             self.viewer.close()
             self.viewer = None
 
@@ -244,4 +251,4 @@ class StoppingGameMdpAttackerEnv(BaseEnv):
                 self.reset()
             else:
                 action_idx = int(raw_input)
-                _, _, done, _ = self.step(pi2=action_idx)
+                _, _, done, _, _ = self.step(pi2=action_idx)

gym_csle_stopping_game/envs/stopping_game_pomdp_defender_env.py

@@ -1,6 +1,6 @@
-from typing import Tuple, List, Dict, Any
-import gymnasium as gym
+from typing import Tuple, List, Dict, Any, Union
 import numpy as np
+import numpy.typing as npt
 from csle_common.dao.simulation_config.base_env import BaseEnv
 from gym_csle_stopping_game.dao.stopping_game_defender_pomdp_config import StoppingGameDefenderPomdpConfig
 from csle_common.dao.simulation_config.simulation_trace import SimulationTrace
@@ -27,7 +27,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
         :param attacker_strategy: the strategy of the static attacker
         """
         self.config = config
-        self.stopping_game_env = gym.make(self.config.stopping_game_name, config=self.config.stopping_game_config)
+        self.stopping_game_env = StoppingGameEnv(config=self.config.stopping_game_config)
 
         # Setup spaces
         self.observation_space = self.config.stopping_game_config.defender_observation_space()
@@ -37,18 +37,18 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
         self.static_attacker_strategy = self.config.attacker_strategy
 
         # Setup Config
-        self.viewer = None
+        self.viewer: Union[None, Any] = None
         self.metadata = {
             'render.modes': ['human', 'rgb_array'],
             'video.frames_per_second': 50  # Video rendering speed
         }
 
-        self.latest_attacker_obs = None
+        self.latest_attacker_obs: Union[None, npt.NDArray[Any]] = None
         # Reset
         self.reset()
         super().__init__()
 
-    def step(self, a1: int) -> Tuple[np.ndarray, int, bool, bool, dict]:
+    def step(self, a1: int) -> Tuple[npt.NDArray[Any], int, bool, bool, Dict[str, Any]]:
         """
         Takes a step in the environment by executing the given action
 
@@ -66,7 +66,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
 
         return defender_obs, r[0], d, d, info
 
-    def step_test(self, a1: int, sample_Z) -> Tuple[np.ndarray, int, bool, dict]:
+    def step_test(self, a1: int, sample_Z) -> Tuple[npt.NDArray[Any], int, bool, Dict[str, Any]]:
         """
         Takes a step in the environment by executing the given action
 
@@ -84,7 +84,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
 
         return defender_obs, r[0], d, info
 
-    def reset(self, seed: int = 0, soft: bool = False) -> Tuple[np.ndarray, Dict[str, Any]]:
+    def reset(self, seed: int = 0, soft: bool = False) -> Tuple[npt.NDArray[Any], Dict[str, Any]]:
         """
         Resets the environment state, this should be called whenever step() returns <done>
 
@@ -93,7 +93,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
         o, _ = self.stopping_game_env.reset()
         self.latest_attacker_obs = o[1]
         defender_obs = o[0]
-        dict = {}
+        dict: Dict[str, Any] = {}
         return defender_obs, dict
 
     def render(self, mode: str = 'human'):
@@ -105,7 +105,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
         """
         raise NotImplementedError("Rendering is not implemented for this environment")
 
-    def step_trace(self, trace: EmulationTrace, a1: int) -> Tuple[np.ndarray, int, bool, dict]:
+    def step_trace(self, trace: EmulationTrace, a1: int) -> Tuple[npt.NDArray[Any], int, bool, Dict[str, Any]]:
         """
         Utility method for stopping a pre-recorded trace
 
@@ -178,7 +178,7 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
         Closes the viewer (cleanup)
         :return: None
         """
-        if self.viewer:
+        if self.viewer is not None:
             self.viewer.close()
             self.viewer = None
 
@@ -211,4 +211,4 @@ class StoppingGamePomdpDefenderEnv(BaseEnv):
                 self.reset()
             else:
                 action_idx = int(raw_input)
-                _, _, done, _ = self.step(pi2=action_idx)
+                _, _, done, _, _ = self.step(a1=action_idx)

gym_csle_stopping_game/util/stopping_game_util.py

@@ -1,5 +1,6 @@
-from typing import List, Dict, Tuple
+from typing import List, Dict, Tuple, Any
 import numpy as np
+import numpy.typing as npt
 from scipy.stats import betabinom
 from csle_common.dao.system_identification.emulation_statistics import EmulationStatistics
 from csle_common.dao.simulation_config.observation_space_config import ObservationSpaceConfig
@@ -14,7 +15,7 @@ class StoppingGameUtil:
     """
 
     @staticmethod
-    def b1() -> np.ndarray:
+    def b1() -> npt.NDArray[np.int_]:
         """
         Gets the initial belief
 
@@ -32,7 +33,7 @@ class StoppingGameUtil:
         return np.array([0, 1, 2])
 
     @staticmethod
-    def defender_actions() -> np.ndarray:
+    def defender_actions() -> npt.NDArray[np.int_]:
         """
         Gets the action space of the defender
 
@@ -41,7 +42,7 @@ class StoppingGameUtil:
         return np.array([0, 1])
 
     @staticmethod
-    def attacker_actions() -> np.ndarray:
+    def attacker_actions() -> npt.NDArray[np.int_]:
         """
         Gets the action space of the attacker
 
@@ -60,7 +61,7 @@ class StoppingGameUtil:
         return np.array(list(range(n + 1)))
 
     @staticmethod
-    def reward_tensor(R_SLA: int, R_INT: int, R_COST: int, L: int, R_ST: int) -> np.ndarray:
+    def reward_tensor(R_SLA: int, R_INT: int, R_COST: int, L: int, R_ST: int) -> npt.NDArray[Any]:
         """
         Gets the reward tensor
 
@@ -89,11 +90,10 @@ class StoppingGameUtil:
                 ]
             ]
             R_l.append(R)
-        R = np.array(R_l)
-        return R
+        return np.array(R_l)
 
     @staticmethod
-    def transition_tensor(L: int, p: float) -> np.ndarray:
+    def transition_tensor(L: int, p: float) -> npt.NDArray[Any]:
         """
         Gets the transition tensor
 
@@ -171,15 +171,14 @@ class StoppingGameUtil:
                     ]
                 ]
             T_l.append(T)
-        T = np.array(T_l)
-        return T
+        return np.array(T_l)
 
     @staticmethod
     def observation_tensor_from_emulation_statistics(emulation_statistic: EmulationStatistics,
                                                      observation_space_defender: ObservationSpaceConfig,
                                                      joint_action_space: JointActionSpaceConfig,
                                                      state_space: StateSpaceConfig) \
-            -> Tuple[np.ndarray, Dict[str, List]]:
+            -> Tuple[npt.NDArray[Any], Dict[str, List[Any]]]:
         """
         Returns an observation tensor based on measured emulation statistics
 
@@ -189,9 +188,9 @@ class StoppingGameUtil:
         :param state_space: the state space
         :return: a |A1|x|A2|x|S|x|O| tensor
         """
-        intrusion_severe_alerts_probabilities = []
-        intrusion_warning_alerts_probabilities = []
-        intrusion_login_attempts_probabilities = []
+        intrusion_severe_alerts_probabilities: List[float] = []
+        intrusion_warning_alerts_probabilities: List[float] = []
+        intrusion_login_attempts_probabilities: List[float] = []
         norm = sum(emulation_statistic.conditionals_counts["intrusion"]["severe_alerts"].values())
         for severe_alert_obs in observation_space_defender.component_observations["severe_alerts"]:
             count = emulation_statistic.conditionals_counts["intrusion"]["severe_alerts"][severe_alert_obs.id]
@@ -229,14 +228,14 @@ class StoppingGameUtil:
             login_attempts_a1_a2_s_o_dist = []
             for a2 in range(len(joint_action_space.action_spaces[1].actions)):
                 a2_s_o_dist = []
-                severe_alerts_a2_s_o_dist = []
-                warning_alerts_a2_s_o_dist = []
-                login_attempts_a2_s_o_dist = []
+                severe_alerts_a2_s_o_dist: List[List[float]] = []
+                warning_alerts_a2_s_o_dist: List[List[float]] = []
+                login_attempts_a2_s_o_dist: List[List[float]] = []
                 for s in range(len(state_space.states)):
                     s_o_dist = []
-                    severe_alerts_s_o_dist = []
-                    warning_alerts_s_o_dist = []
-                    login_attempts_s_o_dist = []
+                    severe_alerts_s_o_dist: List[float] = []
+                    warning_alerts_s_o_dist: List[float] = []
+                    login_attempts_s_o_dist: List[float] = []
                     for o in range(len(observation_space_defender.observations)):
                         obs_vector = observation_space_defender.observation_id_to_observation_id_vector[o]
                         if s == 0:
@@ -256,8 +255,8 @@ class StoppingGameUtil:
                         s_o_dist.append(p)
                     a2_s_o_dist.append(s_o_dist)
                     severe_alerts_a2_s_o_dist.append(severe_alerts_s_o_dist)
-                    warning_alerts_a2_s_o_dist.append(warning_alerts_a2_s_o_dist)
-                    login_attempts_a2_s_o_dist.append(login_attempts_a2_s_o_dist)
+                    warning_alerts_a2_s_o_dist.append(warning_alerts_s_o_dist)
+                    login_attempts_a2_s_o_dist.append(login_attempts_s_o_dist)
                 a1_a2_s_o_dist.append(a2_s_o_dist)
                 severe_alerts_a1_a2_s_o_dist.append(severe_alerts_a2_s_o_dist)
                 warning_alerts_a1_a2_s_o_dist.append(warning_alerts_a2_s_o_dist)
@@ -316,7 +315,7 @@ class StoppingGameUtil:
         return Z
 
     @staticmethod
-    def sample_next_state(T: np.ndarray, l: int, s: int, a1: int, a2: int, S: np.ndarray) -> int:
+    def sample_next_state(T: npt.NDArray[Any], l: int, s: int, a1: int, a2: int, S: npt.NDArray[np.int_]) -> int:
         """
         Samples the next state
 
@@ -331,22 +330,20 @@ class StoppingGameUtil:
         state_probs = []
         for s_prime in S:
             state_probs.append(T[l - 1][a1][a2][s][s_prime])
-        s_prime = np.random.choice(np.arange(0, len(S)), p=state_probs)
-        return s_prime
+        return int(np.random.choice(np.arange(0, len(S)), p=state_probs))
 
     @staticmethod
-    def sample_initial_state(b1: np.ndarray) -> int:
+    def sample_initial_state(b1: npt.NDArray[np.float_]) -> int:
         """
         Samples the initial state
 
         :param b1: the initial belief
         :return: s1
         """
-        s1 = np.random.choice(np.arange(0, len(b1)), p=b1)
-        return s1
+        return int(np.random.choice(np.arange(0, len(b1)), p=b1))
 
     @staticmethod
-    def sample_next_observation(Z: np.ndarray, s_prime: int, O: np.ndarray) -> int:
+    def sample_next_observation(Z: npt.NDArray[Any], s_prime: int, O: npt.NDArray[np.int_]) -> int:
         """
         Samples the next observation
 
@@ -366,7 +363,7 @@ class StoppingGameUtil:
         return int(o)
 
     @staticmethod
-    def bayes_filter(s_prime: int, o: int, a1: int, b: np.ndarray, pi2: np.ndarray, l: int,
+    def bayes_filter(s_prime: int, o: int, a1: int, b: npt.NDArray[np.float_], pi2: npt.NDArray[Any], l: int,
                      config: StoppingGameConfig) -> float:
         """
         A Bayesian filter to compute the belief of player 1
@@ -404,7 +401,7 @@ class StoppingGameUtil:
         return b_prime_s_prime
 
     @staticmethod
-    def p_o_given_b_a1_a2(o: int, b: List, a1: int, a2: int, config: StoppingGameConfig) -> float:
+    def p_o_given_b_a1_a2(o: int, b: List[float], a1: int, a2: int, config: StoppingGameConfig) -> float:
         """
         Computes P[o|a,b]
 
@@ -423,8 +420,8 @@ class StoppingGameUtil:
         return prob
 
     @staticmethod
-    def next_belief(o: int, a1: int, b: np.ndarray, pi2: np.ndarray, config: StoppingGameConfig, l: int,
-                    a2: int = 0, s: int = 0) -> np.ndarray:
+    def next_belief(o: int, a1: int, b: npt.NDArray[np.float_], pi2: npt.NDArray[Any],
+                    config: StoppingGameConfig, l: int, a2: int = 0, s: int = 0) -> npt.NDArray[np.float_]:
         """
         Computes the next belief using a Bayesian filter
 
@@ -449,7 +446,7 @@ class StoppingGameUtil:
         return b_prime
 
     @staticmethod
-    def sample_attacker_action(pi2: np.ndarray, s: int) -> int:
+    def sample_attacker_action(pi2: npt.NDArray[Any], s: int) -> int:
         """
         Samples the attacker action
 
@@ -457,5 +454,4 @@ class StoppingGameUtil:
         :param s: the game state
         :return: a2 (the attacker action
         """
-        a2 = np.random.choice(np.arange(0, len(pi2[s])), p=pi2[s])
-        return a2
+        return int(np.random.choice(np.arange(0, len(pi2[s])), p=pi2[s]))

{gym_csle_stopping_game-0.2.19.dist-info → gym_csle_stopping_game-0.2.20.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gym-csle-stopping-game
-Version: 0.2.19
+Version: 0.2.20
 Summary: OpenAI gym reinforcement learning environment of a Dynkin (Optimal stopping) game in CSLE
 Author: Kim Hammar
 Author-email: hammar.kim@gmail.com
@@ -15,10 +15,10 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Intended Audience :: Science/Research
 Requires-Python: >=3.8
 Requires-Dist: gymnasium >=0.27.1
-Requires-Dist: csle-base >=0.2.19
-Requires-Dist: csle-common >=0.2.19
-Requires-Dist: csle-attacker >=0.2.19
-Requires-Dist: csle-defender >=0.2.19
+Requires-Dist: csle-base >=0.2.20
+Requires-Dist: csle-common >=0.2.20
+Requires-Dist: csle-attacker >=0.2.20
+Requires-Dist: csle-defender >=0.2.20
 Requires-Dist: csle-collector >=0.2.9
 Provides-Extra: testing
 Requires-Dist: pytest >=6.0 ; extra == 'testing'
@@ -26,6 +26,9 @@ Requires-Dist: pytest-cov >=2.0 ; extra == 'testing'
 Requires-Dist: pytest-mock >=3.6.0 ; extra == 'testing'
 Requires-Dist: pytest-grpc >=0.8.0 ; extra == 'testing'
 Requires-Dist: mypy >=1.4.1 ; extra == 'testing'
+Requires-Dist: mypy-extensions >=1.0.0 ; extra == 'testing'
+Requires-Dist: mypy-protobuf >=3.5.0 ; extra == 'testing'
+Requires-Dist: types-PyYAML >=6.0.12.11 ; extra == 'testing'
 Requires-Dist: types-paramiko >=3.2.0.0 ; extra == 'testing'
 Requires-Dist: types-protobuf >=4.23.0.3 ; extra == 'testing'
 Requires-Dist: types-requests >=2.31.0.1 ; extra == 'testing'

gym_csle_stopping_game-0.2.20.dist-info/RECORD

@@ -0,0 +1,19 @@
+gym_csle_stopping_game/__init__.py,sha256=ooy6TjxvBi1sZMEX3_mVlvfskqI5GqwITWzI882tfk0,657
+gym_csle_stopping_game/__version__.py,sha256=qMaQjrsi_F2wUsxjrQ7pmHBHD2rS1ibAIXrI1iCRn08,23
+gym_csle_stopping_game/constants/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+gym_csle_stopping_game/constants/constants.py,sha256=eIoD9eXifZ73kP-lSlvG-IXCpe4n6D-_aDygx0zOr5U,1030
+gym_csle_stopping_game/dao/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+gym_csle_stopping_game/dao/stopping_game_attacker_mdp_config.py,sha256=kmtrVa2CCVbi5xfd6uPWqMvhGmP8ccrtn1e_VmVvH7k,3494
+gym_csle_stopping_game/dao/stopping_game_config.py,sha256=5jFMvSWkJ0_PqlVZlAf2pzJFttHeUdUv_G4GeXIrdm0,5595
+gym_csle_stopping_game/dao/stopping_game_defender_pomdp_config.py,sha256=3FfNi2-R6n1LqjA644EVq-v7wtp6sqyEkEdBN90-2n0,3753
+gym_csle_stopping_game/dao/stopping_game_state.py,sha256=Fyl19vIMeShYnHaixeJ-OHM4LHpJAswhBIp_5ytN3bM,2768
+gym_csle_stopping_game/envs/__init__.py,sha256=SQHaqXI0_2HYsC8i9swXEHDFcXKEYpb8GRP9l_S0Sw8,74
+gym_csle_stopping_game/envs/stopping_game_env.py,sha256=SY3vYUhtS42XIkzJfIhiJE-dKjSTRH7iiLpEQ7Id9P8,26178
+gym_csle_stopping_game/envs/stopping_game_mdp_attacker_env.py,sha256=JGKp2B3s7Hf6z_nfOmDtv9LZo26i9HayTcVBq6gW6sk,9570
+gym_csle_stopping_game/envs/stopping_game_pomdp_defender_env.py,sha256=JD7hGrBMYrNXh6A3JXFLvz4Op1ZWOZqlFEcDU4jS49k,8754
+gym_csle_stopping_game/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+gym_csle_stopping_game/util/stopping_game_util.py,sha256=Kgt6o8nobpdlx0zoPxfoA8sH24KxMxQ5gh4Txt9wWgs,19160
+gym_csle_stopping_game-0.2.20.dist-info/METADATA,sha256=Ez_SGYA2X1QFWfE4arGr2LLHKQIliBQJ8RIdP804FyU,2055
+gym_csle_stopping_game-0.2.20.dist-info/WHEEL,sha256=5sUXSg9e4bi7lTLOHcm6QEYwO5TIF1TNbTSVFVjcJcc,92
+gym_csle_stopping_game-0.2.20.dist-info/top_level.txt,sha256=3DBHkAEI00nq0aXZlJUkXJrLiwkcJCfaFoYcaOzEZUU,23
+gym_csle_stopping_game-0.2.20.dist-info/RECORD,,

gym_csle_stopping_game-0.2.19.dist-info/RECORD

@@ -1,19 +0,0 @@
-gym_csle_stopping_game/__init__.py,sha256=ooy6TjxvBi1sZMEX3_mVlvfskqI5GqwITWzI882tfk0,657
-gym_csle_stopping_game/__version__.py,sha256=8swO86HIyP6FWtTckB79yaNlkfWbV8msnhazFt9EOKQ,23
-gym_csle_stopping_game/constants/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-gym_csle_stopping_game/constants/constants.py,sha256=eIoD9eXifZ73kP-lSlvG-IXCpe4n6D-_aDygx0zOr5U,1030
-gym_csle_stopping_game/dao/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-gym_csle_stopping_game/dao/stopping_game_attacker_mdp_config.py,sha256=e8KwNZpZ_VswM57ZmjaGDeFTFButVCB8WUAHEctAWJk,3511
-gym_csle_stopping_game/dao/stopping_game_config.py,sha256=qWcfGLf7X6ymp8R_SLkXxNgH1AngIsGrhFXsB5ZnBxw,5448
-gym_csle_stopping_game/dao/stopping_game_defender_pomdp_config.py,sha256=niK4cK_YoyFW6Wq-rFK_5hbLsDtlQ-UivcVf5BtGUrA,3770
-gym_csle_stopping_game/dao/stopping_game_state.py,sha256=tqlFJ9sjlNzkgKJPBbLKhi_HFEnuTJfFcZzs-idrf4w,2701
-gym_csle_stopping_game/envs/__init__.py,sha256=SQHaqXI0_2HYsC8i9swXEHDFcXKEYpb8GRP9l_S0Sw8,74
-gym_csle_stopping_game/envs/stopping_game_env.py,sha256=J0fL4z6cNhmXiwSXtaU9wbrAvVBeDuW2bJQ0YWAqGMs,25889
-gym_csle_stopping_game/envs/stopping_game_mdp_attacker_env.py,sha256=wsOZRDgktz5ENvmhQI8DLCLIoN2JhmY4eoLyX0X0zsA,9060
-gym_csle_stopping_game/envs/stopping_game_pomdp_defender_env.py,sha256=IvY4gwg3Mz2hNBTuTToCbPLA8Zp1KRWc_P9H5KXVvug,8636
-gym_csle_stopping_game/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-gym_csle_stopping_game/util/stopping_game_util.py,sha256=RaR-onJtmuTk15B2m4KUYTbXNElKBw2iEkoP6m0n1b0,18912
-gym_csle_stopping_game-0.2.19.dist-info/METADATA,sha256=NJe-ly9n_tOtOTMS0AHj7OwBAsGuVL2Qxu5bAl08wUA,1876
-gym_csle_stopping_game-0.2.19.dist-info/WHEEL,sha256=5sUXSg9e4bi7lTLOHcm6QEYwO5TIF1TNbTSVFVjcJcc,92
-gym_csle_stopping_game-0.2.19.dist-info/top_level.txt,sha256=3DBHkAEI00nq0aXZlJUkXJrLiwkcJCfaFoYcaOzEZUU,23
-gym_csle_stopping_game-0.2.19.dist-info/RECORD,,