Allow the env to be closed many times (#68)

* expose env_ids and 3ple loop registration

* parametrize env test and double close check

* fix double close

* update version

* isort

* remove extra line in test

docs/conf.py

@@ -23,7 +23,7 @@
 author = "Quentin Gallouédec"
 
 # The full version, including alpha/beta/rc tags
-release = "v3.0.6"
+release = "v3.0.7"
 
 
 # -- General configuration ---------------------------------------------------

panda_gym/__init__.py

@@ -5,50 +5,20 @@
 with open(os.path.join(os.path.dirname(__file__), "version.txt"), "r") as file_handler:
     __version__ = file_handler.read().strip()
 
-for reward_type in ["sparse", "dense"]:
-    for control_type in ["ee", "joints"]:
-        reward_suffix = "Dense" if reward_type == "dense" else ""
-        control_suffix = "Joints" if control_type == "joints" else ""
-        kwargs = {"reward_type": reward_type, "control_type": control_type}
-
-        register(
-            id="PandaReach{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaReachEnv",
-            kwargs=kwargs,
-            max_episode_steps=50,
-        )
-
-        register(
-            id="PandaPush{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaPushEnv",
-            kwargs=kwargs,
-            max_episode_steps=50,
-        )
-
-        register(
-            id="PandaSlide{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaSlideEnv",
-            kwargs=kwargs,
-            max_episode_steps=50,
-        )
-
-        register(
-            id="PandaPickAndPlace{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaPickAndPlaceEnv",
-            kwargs=kwargs,
-            max_episode_steps=50,
-        )
-
-        register(
-            id="PandaStack{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaStackEnv",
-            kwargs=kwargs,
-            max_episode_steps=100,
-        )
-
-        register(
-            id="PandaFlip{}{}-v3".format(control_suffix, reward_suffix),
-            entry_point="panda_gym.envs:PandaFlipEnv",
-            kwargs=kwargs,
-            max_episode_steps=50,
-        )
+ENV_IDS = []
+
+for task in ["Reach", "Slide", "Push", "PickAndPlace", "Stack", "Flip"]:
+    for reward_type in ["sparse", "dense"]:
+        for control_type in ["ee", "joints"]:
+            reward_suffix = "Dense" if reward_type == "dense" else ""
+            control_suffix = "Joints" if control_type == "joints" else ""
+            env_id = f"Panda{task}{control_suffix}{reward_suffix}-v3"
+
+            register(
+                id=env_id,
+                entry_point=f"panda_gym.envs:Panda{task}Env",
+                kwargs={"reward_type": reward_type, "control_type": control_type},
+                max_episode_steps=100 if task == "Stack" else 50,
+            )
+
+            ENV_IDS.append(env_id)

panda_gym/pybullet.py

@@ -70,7 +70,8 @@ def step(self) -> None:
 
     def close(self) -> None:
         """Close the simulation."""
-        self.physics_client.disconnect()
+        if self.physics_client.isConnected():
+            self.physics_client.disconnect()
 
     def save_state(self) -> int:
         """Save the current simulation state.

panda_gym/version.txt

@@ -1 +1 @@
-3.0.6
+3.0.7

test/envs_test.py

@@ -1,4 +1,5 @@
 import gymnasium as gym
+import pytest
 
 import panda_gym
 
@@ -12,123 +13,12 @@ def run_env(env):
         if terminated or truncated:
             env.reset()
     env.close()
+    # check that it allows to be closed multiple times
+    env.close()
 
 
-def test_reach():
-    env = gym.make("PandaReach-v3")
-    run_env(env)
-
-
-def test_slide():
-    env = gym.make("PandaSlide-v3")
-    run_env(env)
-
-
-def test_push():
-    env = gym.make("PandaPush-v3")
-    run_env(env)
-
-
-def test_pickandplace():
-    env = gym.make("PandaPickAndPlace-v3")
-    run_env(env)
-
-
-def test_stack():
-    env = gym.make("PandaStack-v3")
-    run_env(env)
-
-
-def test_flip():
-    env = gym.make("PandaFlip-v3")
-    run_env(env)
-
-
-def test_dense_reach():
-    env = gym.make("PandaReachDense-v3")
-    run_env(env)
-
-
-def test_dense_slide():
-    env = gym.make("PandaSlideDense-v3")
-    run_env(env)
-
-
-def test_dense_push():
-    env = gym.make("PandaPushDense-v3")
-    run_env(env)
-
-
-def test_dense_pickandplace():
-    env = gym.make("PandaPickAndPlaceDense-v3")
-    run_env(env)
-
-
-def test_dense_stack():
-    env = gym.make("PandaStackDense-v3")
-    run_env(env)
-
-
-def test_dense_flip():
-    env = gym.make("PandaFlipDense-v3")
-    run_env(env)
-
-
-def test_reach_joints():
-    env = gym.make("PandaReachJoints-v3")
-    run_env(env)
-
-
-def test_slide_joints():
-    env = gym.make("PandaSlideJoints-v3")
-    run_env(env)
-
-
-def test_push_joints():
-    env = gym.make("PandaPushJoints-v3")
-    run_env(env)
-
-
-def test_pickandplace_joints():
-    env = gym.make("PandaPickAndPlaceJoints-v3")
-    run_env(env)
-
-
-def test_stack_joints():
-    env = gym.make("PandaStackJoints-v3")
-    run_env(env)
-
-
-def test_flip_joints():
-    env = gym.make("PandaFlipJoints-v3")
-    run_env(env)
-
-
-def test_dense_reach_joints():
-    env = gym.make("PandaReachJointsDense-v3")
-    run_env(env)
-
-
-def test_dense_slide_joints():
-    env = gym.make("PandaSlideDense-v3")
-    run_env(env)
-
-
-def test_dense_push_joints():
-    env = gym.make("PandaPushJointsDense-v3")
-    run_env(env)
-
-
-def test_dense_pickandplace_joints():
-    env = gym.make("PandaPickAndPlaceJointsDense-v3")
-    run_env(env)
-
-
-def test_dense_stack_joints():
-    env = gym.make("PandaStackJointsDense-v3")
-    run_env(env)
-
-
-def test_dense_flip_joints():
-    env = gym.make("PandaFlipJointsDense-v3")
+@pytest.mark.parametrize("env_id", panda_gym.ENV_IDS)
+def test_env(env_id):
+    """Tests running panda gym envs."""
+    env = gym.make(env_id)
     run_env(env)