testing round trip of action -> order and order -> action

integration_tests/test_env.py

@@ -21,12 +21,19 @@ def embed_battle(self, battle):
         return np.array([0])
 
 
-def play_function(env, n_battles):
+def play_function(env: CIEnv, n_battles: int):
     for _ in range(n_battles):
         done = False
         env.reset()
         while not done:
+            assert env.battle1 is not None
+            assert env.battle2 is not None
             actions = {name: env.action_space(name).sample() for name in env.agents}
+            [a1, a2] = list(actions.values())
+            o1 = CIEnv.action_to_order(a1, env.battle1)
+            o2 = CIEnv.action_to_order(a2, env.battle2)
+            assert a1 == CIEnv.order_to_action(o1, env.battle1)
+            assert a2 == CIEnv.order_to_action(o2, env.battle2)
             _, _, terminated, truncated, _ = env.step(actions)
             done = any(terminated.values()) or any(truncated.values())
 

src/poke_env/player/env.py

@@ -360,10 +360,10 @@ def close(self, purge: bool = True):
         )
         closing_task.result()
 
-    def observation_space(self, agent: str) -> Space:
+    def observation_space(self, agent: str) -> Space[ObsType]:
         return self.observation_spaces[agent]
 
-    def action_space(self, agent: str) -> Space:
+    def action_space(self, agent: str) -> Space[ActionType]:
         return self.action_spaces[agent]
 
     ###################################################################################