feat: 👽 update due to co_mas APIs change

smac_pettingzoo/env/smacv1.py

@@ -1779,11 +1779,11 @@ def get_avail_agent_actions(self, agent_id):
                     if dist <= shoot_range:
                         avail_actions[t_id + self.n_actions_no_attack] = 1
 
-            return avail_actions
+            return np.array(avail_actions)
 
         else:
             # only no-op allowed
-            return [1] + [0] * (self.n_actions - 1)
+            return np.array([1] + [0] * (self.n_actions - 1))
 
     def get_avail_actions(self):
         """Returns the available actions of all agents in a list."""

smac_pettingzoo/env/smacv2.py

@@ -2260,11 +2260,11 @@ def get_avail_agent_actions(self, agent_id):
                     if can_shoot:
                         avail_actions[t_id + self.n_actions_no_attack] = 1
 
-            return avail_actions
+            return np.array(avail_actions)
 
         else:
             # only no-op allowed
-            return [1] + [0] * (self.n_actions - 1)
+            return np.array([1] + [0] * (self.n_actions - 1))
 
     def get_avail_actions(self):
         """Returns the available actions of all agents in a list."""

smac_pettingzoo/smacv1_pettingzoo_v1.py

@@ -136,7 +136,8 @@ def step(self, actions: Dict) -> Tuple[Dict, Dict, Dict, Dict, Dict]:
         return observations, rewards, terminations, truncations, infos
 
     def close(self):
-        self._env.close()
+        if hasattr(self, "_env") and self._env is not None:
+            self._env.close()
 
 
 def parallel_env(

smac_pettingzoo/smacv2_pettingzoo_v1.py

@@ -223,7 +223,8 @@ def step(self, actions: Dict) -> Tuple[Dict, Dict, Dict, Dict, Dict]:
         return observations, rewards, terminations, truncations, infos
 
     def close(self):
-        self._env.close()
+        if hasattr(self, "_env") and self._env is not None:
+            self._env.close()
 
 
 def parallel_env(

tests/smacv1_pettingzoo_v1_test.py

@@ -1,4 +1,5 @@
-from co_mas.test import parallel_api_test, sample_action
+import numpy as np
+from co_mas.test.parallel_api import parallel_api_test, sample_action
 from loguru import logger
 
 from smac_pettingzoo import smacv1_pettingzoo_v1
@@ -10,7 +11,7 @@
     step = 0
     while True:
         obs, _, terminated, truncated, info = env.step(
-            {agent: sample_action(env, obs, agent, info) for agent in env.agents}
+            {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
         )
         step += 1
         if len(env.agents) <= 0:
@@ -28,11 +29,12 @@
 env1 = smacv1_pettingzoo_v1.parallel_env("8m", {})
 obs1_list = []
 obs1, info1 = env1.reset(seed=42)
+np.random.seed(42)
 obs1_list.append(obs1)
 
 while True:
     obs1, _, terminated1, _, info1 = env1.step(
-        {agent: sample_action(env1, obs1, agent, info1) for agent in env1.agents}
+        {agent: sample_action(agent, obs1[agent], info1[agent], env1.action_space(agent)) for agent in env1.agents}
     )
     obs1_list.append(obs1)
 
@@ -45,25 +47,26 @@
 
 obs2_list = []
 obs2, info2 = env2.reset(seed=42)
+np.random.seed(42)
 obs2_list.append(obs2)
 
 while True:
     obs2, _, terminated2, _, info2 = env2.step(
-        {agent: sample_action(env2, obs2, agent, info2) for agent in env2.agents}
+        {agent: sample_action(agent, obs2[agent], info2[agent], env2.action_space(agent)) for agent in env2.agents}
     )
     obs2_list.append(obs2)
 
     if any(terminated2.values()):
         break
 
+env2.close()
+
 for i, (obs1, obs2) in enumerate(zip(obs1_list, obs2_list)):
     assert all(
         (obs1[agent] == obs2[agent]).all() for agent in env1.agents
     ), f"Observations at step {i} differ:\n{obs1}\n{obs2}"
 
 logger.success("Seed Test Passed!")
-env1.close()
-env2.close()
 
 # Wrapper Tests
 from co_mas.wrappers import AutoResetParallelEnvWrapper, OrderForcingParallelEnvWrapper
@@ -75,12 +78,16 @@
 obs, info = env.reset(seed=42)
 
 while True:
-    obs, _, terminated, _, info = env.step({agent: sample_action(env, obs, agent, info) for agent in env.agents})
+    obs, _, terminated, _, info = env.step(
+        {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
+    )
 
     if all(terminated.values()):
         break
 
-obs, _, terminated, _, info = env.step({agent: sample_action(env, obs, agent, info) for agent in env.agents})
+obs, _, terminated, _, info = env.step(
+    {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
+)
 
 assert terminated != {agent: True for agent in env.agents}
 

tests/smacv2_pettingzoo_v1_test.py

@@ -1,4 +1,5 @@
-from co_mas.test import parallel_api_test, sample_action
+import numpy as np
+from co_mas.test.parallel_api import parallel_api_test, sample_action
 from loguru import logger
 
 from smac_pettingzoo import smacv2_pettingzoo_v1
@@ -10,7 +11,7 @@
     step = 0
     while True:
         obs, _, terminated, truncated, info = env.step(
-            {agent: sample_action(env, obs, agent, info) for agent in env.agents}
+            {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
         )
         step += 1
         if len(env.agents) <= 0:
@@ -19,24 +20,25 @@
     ep_i += 1
     if any(truncated.values()):
         break
+
 env.close()
 
-parallel_api_test(env, 200)
+parallel_api_test(env, 400)
 
 # Seed Tests
 env1 = smacv2_pettingzoo_v1.parallel_env("10gen_terran_10_vs_10")
 obs1_list = []
 obs1, info1 = env1.reset(seed=42)
+np.random.seed(42)
 obs1_list.append(obs1)
 
 while True:
-    obs1, _, terminated1, truncated1, info1 = env1.step(
-        {agent: sample_action(env1, obs1, agent, info1) for agent in env1.agents}
+    obs1, _, terminated1, _, info1 = env1.step(
+        {agent: sample_action(agent, obs1[agent], info1[agent], env1.action_space(agent)) for agent in env1.agents}
     )
     obs1_list.append(obs1)
-    logger.trace(f"{env1.agents}")
 
-    if len(env1.agents) <= 0:
+    if any(terminated1.values()):
         break
 
 env1.close()
@@ -45,25 +47,26 @@
 
 obs2_list = []
 obs2, info2 = env2.reset(seed=42)
+np.random.seed(42)
 obs2_list.append(obs2)
 
 while True:
     obs2, _, terminated2, _, info2 = env2.step(
-        {agent: sample_action(env2, obs2, agent, info2) for agent in env2.agents}
+        {agent: sample_action(agent, obs2[agent], info2[agent], env2.action_space(agent)) for agent in env2.agents}
     )
     obs2_list.append(obs2)
 
-    if len(env2.agents) <= 0:
+    if any(terminated2.values()):
         break
 
+env2.close()
+
 for i, (obs1, obs2) in enumerate(zip(obs1_list, obs2_list)):
     assert all(
         (obs1[agent] == obs2[agent]).all() for agent in env1.agents
     ), f"Observations at step {i} differ:\n{obs1}\n{obs2}"
 
 logger.success("Seed Test Passed!")
-env1.close()
-env2.close()
 
 # Wrapper Tests
 from co_mas.wrappers import AutoResetParallelEnvWrapper, OrderForcingParallelEnvWrapper
@@ -75,12 +78,16 @@
 obs, info = env.reset(seed=42)
 
 while True:
-    obs, _, terminated, _, info = env.step({agent: sample_action(env, obs, agent, info) for agent in env.agents})
+    obs, _, terminated, _, info = env.step(
+        {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
+    )
 
-    if len(env.agents) <= 0:
+    if all(terminated.values()):
         break
 
-obs, _, terminated, _, info = env.step({agent: sample_action(env, obs, agent, info) for agent in env.agents})
+obs, _, terminated, _, info = env.step(
+    {agent: sample_action(agent, obs[agent], info[agent], env.action_space(agent)) for agent in env.agents}
+)
 
 assert terminated != {agent: True for agent in env.agents}