adding support for gym to fix #379 and finish previous work

grid2op/gym_compat/gymenv.py

@@ -6,11 +6,14 @@
 # SPDX-License-Identifier: MPL-2.0
 # This file is part of Grid2Op, Grid2Op a testbed platform to model sequential decision making in power systems.
 
+from os import truncate
+import warnings
 import gym
 from grid2op.Chronics import Multifolder
 from grid2op.gym_compat.gym_obs_space import GymObservationSpace
 from grid2op.gym_compat.gym_act_space import GymActionSpace
-from grid2op.gym_compat.utils import check_gym_version
+from grid2op.gym_compat.utils import (check_gym_version, sample_seed,
+                                      _MAX_GYM_VERSION_RANDINT, GYM_VERSION)
 
 
 class GymEnv(gym.Env):
@@ -49,27 +52,54 @@ def __init__(self, env_init, shuffle_chronics=True):
         self.reward_range = self.init_env.reward_range
         self.metadata = self.init_env.metadata
         self._shuffle_chronics = shuffle_chronics
-
-    def step(self, gym_action):
+
+        if GYM_VERSION <= _MAX_GYM_VERSION_RANDINT:
+            self.seed = self._aux_seed
+            self.reset = self._aux_reset
+            self.step = self._aux_step
+        else:
+            self.reset = self._aux_reset_new
+            self.step = self._aux_step_new
+
+    def _aux_step(self, gym_action):
+        # used for gym < 0.26
         g2op_act = self.action_space.from_gym(gym_action)
         g2op_obs, reward, done, info = self.init_env.step(g2op_act)
         gym_obs = self.observation_space.to_gym(g2op_obs)
         return gym_obs, float(reward), done, info
+
+    def _aux_step_new(self, gym_action):
+        # used for gym >= 0.26
+        # TODO refacto with _aux_step
+        g2op_act = self.action_space.from_gym(gym_action)
+        g2op_obs, reward, done, info = self.init_env.step(g2op_act)
+        gym_obs = self.observation_space.to_gym(g2op_obs)
+        truncated = g2op_obs.current_step == g2op_obs.max_step
+        return gym_obs, float(reward), done, truncated, info
 
-    def reset(self, seed=None, return_info=False, options=None):
+    def _aux_reset(self, seed=None, return_info=None, options=None):
+        # used for gym < 0.26
         if self._shuffle_chronics and isinstance(
             self.init_env.chronics_handler.real_data, Multifolder
         ):
             self.init_env.chronics_handler.sample_next_chronics()
+
         if seed is not None:
-            self.init_env.seed(seed)
+            self._aux_seed(seed)
+
         g2op_obs = self.init_env.reset()
         gym_obs = self.observation_space.to_gym(g2op_obs)
+
         if return_info:
-            return gym_obs, {}
+            chron_id = self.init_env.chronics_handler.get_id()
+            return gym_obs, {"time serie id": chron_id}
         else:
             return gym_obs
 
+    def _aux_reset_new(self, seed=None, options=None):
+        # used for gym > 0.26
+        return self._aux_reset(seed, True, options)
+
     def render(self, mode="human"):
         """for compatibility with open ai gym render function"""
         super(GymEnv, self).render(mode=mode)
@@ -87,9 +117,15 @@ def close(self):
             self.observation_space.close()
         self.observation_space = None
 
-    def seed(self, seed=None):
-        self.init_env.seed(seed)
-        # TODO seed also env space and observation space
+    def _aux_seed(self, seed=None):
+        # deprecated in gym >=0.26
+        if seed is not None:
+            # seed the gym env
+            super().reset(seed=seed)
+            # then seed the underlying grid2op env
+            max_ = 2**32-1
+            next_seed = sample_seed(max_, self._np_random)
+            self.init_env.seed(next_seed)
 
     def __del__(self):
         # delete possible dangling reference

grid2op/tests/test_gym_compat.py

@@ -94,11 +94,11 @@ def test_convert_togym(self):
         ), f"Size should be {size_th} but is {dim_obs_space}"
 
         # test that i can do basic stuff there
-        obs = env_gym.reset()
+        obs, info = env_gym.reset()
         for k in env_gym.observation_space.spaces.keys():
             assert obs[k] in env_gym.observation_space[k], f"error for key: {k}"
         act = env_gym.action_space.sample()
-        obs2, reward2, done2, info2 = env_gym.step(act)
+        obs2, reward2, done2, truncated, info2 = env_gym.step(act)
         assert obs2 in env_gym.observation_space
 
         # test for the __str__ method
@@ -152,7 +152,7 @@ def test_scale_attr_converter(self):
             ScalerAttrConverter(substract=0.0, divide=self.env.gen_pmax),
         )
         env_gym.observation_space = ob_space
-        obs = env_gym.reset()
+        obs, info = env_gym.reset()
         assert key in env_gym.observation_space.spaces
         low = np.zeros(self.env.n_gen) - 1
         high = np.zeros(self.env.n_gen) + 1
@@ -181,7 +181,7 @@ def test_add_key(self):
         )
 
         # we highly recommend to "reset" the environment after setting up the observation space
-        obs_gym = env_gym.reset()
+        obs_gym, info = env_gym.reset()
         assert key in env_gym.observation_space.spaces
         assert obs_gym in env_gym.observation_space
 
@@ -756,7 +756,7 @@ def test_can_create(self):
                 )
             },
         )
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in self.env_gym.observation_space
         assert self.env_gym.observation_space._attr_to_keep == sorted(kept_attr)
         assert len(obs_gym) == 3583
@@ -766,7 +766,7 @@ def test_can_create_int(self):
         self.env_gym.observation_space = BoxGymObsSpace(
             self.env.observation_space, attr_to_keep=kept_attr
         )
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in self.env_gym.observation_space
         assert self.env_gym.observation_space._attr_to_keep == sorted(kept_attr)
         assert len(obs_gym) == 79
@@ -779,7 +779,7 @@ def test_scaling(self):
             self.env.observation_space, attr_to_keep=kept_attr
         )
         self.env_gym.observation_space = observation_space
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in observation_space
         assert observation_space._attr_to_keep == kept_attr
         assert len(obs_gym) == 17
@@ -792,7 +792,7 @@ def test_scaling(self):
             divide={"gen_p": self.env.gen_pmax, "load_p": self.obs_env.load_p},
         )
         self.env_gym.observation_space = observation_space
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in observation_space
         assert observation_space._attr_to_keep == kept_attr
         assert len(obs_gym) == 17
@@ -807,7 +807,7 @@ def test_scaling(self):
             subtract={"gen_p": 100.0, "load_p": 100.0},
         )
         self.env_gym.observation_space = observation_space
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in observation_space
         assert observation_space._attr_to_keep == kept_attr
         assert len(obs_gym) == 17
@@ -844,7 +844,7 @@ def test_functs(self):
                 )
             },
         )
-        obs_gym = self.env_gym.reset()
+        obs_gym, info = self.env_gym.reset()
         assert obs_gym in self.env_gym.observation_space
         assert self.env_gym.observation_space._attr_to_keep == sorted(kept_attr)
         assert len(obs_gym) == 3583
@@ -1846,7 +1846,7 @@ def test_all_attr_in_obs(self):
             env = grid2op.make("educ_case14_storage", test=True,
                                action_class=PlayableAction)
         gym_env = GymEnv(env)
-        obs = gym_env.reset()
+        obs, info = gym_env.reset()
         all_attrs = ["year",
                      "month",
                      "day",

grid2op/tests/test_issue_379.py

@@ -0,0 +1,45 @@
+# Copyright (c) 2019-2022, RTE (https://www.rte-france.com)
+# See AUTHORS.txt and https://github.com/rte-france/Grid2Op/pull/319
+# This Source Code Form is subject to the terms of the Mozilla Public License, version 2.0.
+# If a copy of the Mozilla Public License, version 2.0 was not distributed with this file,
+# you can obtain one at http://mozilla.org/MPL/2.0/.
+# SPDX-License-Identifier: MPL-2.0
+# This file is part of Grid2Op, Grid2Op a testbed platform to model sequential decision making in power systems.
+
+import grid2op
+import unittest
+import warnings
+
+from grid2op.gym_compat import GymEnv
+import grid2op
+from gym import Env
+from gym.utils.env_checker import check_env
+try:
+    from gym.utils.env_checker import check_reset_return_type, check_reset_options, check_reset_seed
+    CAN_TEST_ALL = True
+except ImportError:
+    CAN_TEST_ALL = False
+
+
+class Issue379Tester(unittest.TestCase):
+    def setUp(self) -> None:
+        with warnings.catch_warnings():
+            warnings.filterwarnings("ignore")
+            self.env = grid2op.make("l2rpn_case14_sandbox", test=True)
+            self.gym_env = GymEnv(self.env)
+
+    def tearDown(self) -> None:
+        self.env.close()
+        self.gym_env.close()
+        return super().tearDown()
+
+    def test_check_env(self):
+        if CAN_TEST_ALL:
+            check_reset_return_type(self.gym_env)
+            check_reset_seed(self.gym_env)
+            check_reset_options(self.gym_env)
+        check_env(self.gym_env)
+
+
+if __name__ == "__main__":
+    unittest.main()

setup.py

@@ -61,7 +61,8 @@ def my_test_suite():
         ],
         "plot": ["imageio"],
         "test": ["lightsim2grid",
-                 "numba"
+                 "numba",
+                 "gym>=0.26"
                  ],
         "chronix2grid": [
             "ChroniX2Grid>=1.1.0.post1"