Merge pull request #193 from BDonnot/bd_dev

Fix notebook for backend example

CHANGELOG.rst

@@ -31,6 +31,13 @@ Change Log
 - [???] "asynch" multienv
 - [???] properly model interconnecting powerlines
 
+[1.10.3] - 2024-xx-yy
+-------------------------
+- TODO A number of max buses per sub
+- TODO Automatic "experimental_read_from_local_dir"
+- TODO Notebook for stable baselines
+- TODO in the reset options: datetime start and max number of steps
+
 [1.10.2] - 2024-05-27
 -------------------------
 - [BREAKING] the `runner.run_one_episode` now returns an extra first argument: 

Dockerfile

@@ -35,7 +35,7 @@ WORKDIR /Grid2Op
 RUN git pull
 RUN git remote update
 RUN git fetch --all --tags
-RUN git checkout "tags/v1.10.1" -b "v1.10.1-branch"
+RUN git checkout "tags/v1.10.2" -b "v1.10.2-branch"
 # Install Dependencies
 RUN pip3 install .[optional,challenge]
 WORKDIR /

docs/conf.py

@@ -22,7 +22,7 @@
 author = 'Benjamin Donnot'
 
 # The full version, including alpha/beta/rc tags
-release = '1.10.2.dev3'
+release = '1.10.3.dev0'
 version = '1.10'
 
 

examples/backend_integration/Step0_make_env.py

@@ -41,6 +41,28 @@
 from grid2op.Opponent import BaseOpponent
 
 
+class PandaPowerBackendNoShunt(PandaPowerBackend):
+    shunts_data_available = False
+
+
+def create_action(env, backend, action):
+    """this is done internally by grid2op.
+    
+    The idea is to generate a "backend action" (which again is provided by grid2op)
+    easily
+    """
+    # bk_act = env._backend_action_class()
+    # bk_act += action  # action for pandapower backend
+    # bk_act.reorder(env.backend._load_sr2tg,
+    #                env.backend._gen_sr2tg,
+    #                env.backend._topo_sr2tg,
+    #                env.backend._storage_sr2tg,
+    #                env.backend._shunt_sr2tg)
+    bk_act = type(backend).my_bk_act_class()
+    bk_act += action
+    return bk_act
+
+
 def make_env_for_backend(env_name, backend_class):
     # env_name: one of:
     # - rte_case5_example: the grid in the documentation (completely fake grid)
@@ -65,8 +87,9 @@ def make_env_for_backend(env_name, backend_class):
                            action_class=CompleteAction,  # we tell grid2op we will manipulate all type of actions
                            reward_class=ConstantReward,  # we don't have yet redispatching data, that might be use by the reward
                            opponent_class=BaseOpponent,  # we deactivate the opponents
+                        #    backend=backend_class() 
                            backend=BackendConverter(source_backend_class=backend_class,
-                                                    target_backend_class=PandaPowerBackend,
+                                                    target_backend_class=PandaPowerBackendNoShunt,
                                                     use_target_backend_name=True)
                           )
     obs = env.reset()

examples/backend_integration/Step1_loading.py

@@ -34,6 +34,7 @@
 
 
 class CustomBackend_Step1(Backend):
+    shunts_data_available = False
     def load_grid(self,
                   path : Union[os.PathLike, str],
                   filename : Optional[Union[os.PathLike, str]]=None) -> None:
@@ -170,8 +171,8 @@ def lines_ex_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
     # storage_pos_topo_vect
 
     # for example
-    print(type(backend).name_load)
-    print(type(backend).load_to_subid)
-    print(type(backend).load_to_sub_pos)
-    print(type(backend).load_pos_topo_vect)
+    print(f"Name of the loads, seen in grid2op: {type(backend).name_load}")
+    print(f"Id of substation, for each load: {type(backend).load_to_subid}")
+    print(f"Position in the substation topology vector, for each load: {type(backend).load_to_sub_pos}")
+    print(f"Position in the global topology vector, for each load: {type(backend).load_pos_topo_vect}")
 

examples/backend_integration/Step2_modify_load.py

@@ -69,7 +69,7 @@ def loads_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray]:
 if __name__ == "__main__":
     import grid2op
     import os
-    from Step0_make_env import make_env_for_backend
+    from Step0_make_env import make_env_for_backend, create_action
 
     path_grid2op = grid2op.__file__
     path_data_test = os.path.join(os.path.split(path_grid2op)[0], "data")
@@ -105,8 +105,7 @@ def loads_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray]:
     # have the proper size)
 
     # this is technical to grid2op (done internally)
-    bk_act = env._backend_action_class()
-    bk_act += action
+    bk_act = create_action(env, backend, action)
     #############
 
     # this is what the backend receive:

examples/backend_integration/Step3_modify_gen.py

@@ -67,7 +67,7 @@ def generators_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray]:
 if __name__ == "__main__":
     import grid2op
     import os
-    from Step0_make_env import make_env_for_backend
+    from Step0_make_env import make_env_for_backend, create_action
 
     path_grid2op = grid2op.__file__
     path_data_test = os.path.join(os.path.split(path_grid2op)[0], "data")
@@ -103,8 +103,7 @@ def generators_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray]:
     # have the proper size)
 
     # this is technical to grid2op (done internally)
-    bk_act = env._backend_action_class()
-    bk_act += action
+    bk_act = create_action(env, backend, action)
     #############
 
     # this is what the backend receive:

examples/backend_integration/Step4_modify_line_status.py

@@ -178,7 +178,7 @@ def lines_ex_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
 if __name__ == "__main__":
     import grid2op
     import os
-    from Step0_make_env import make_env_for_backend
+    from Step0_make_env import make_env_for_backend, create_action
 
     path_grid2op = grid2op.__file__
     path_data_test = os.path.join(os.path.split(path_grid2op)[0], "data")
@@ -205,8 +205,7 @@ def lines_ex_info(self)-> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
     action = env.action_space({"set_line_status": [(0, -1)]})
 
     # this is technical to grid2op
-    bk_act = env._backend_action_class()
-    bk_act += action
+    bk_act = create_action(env, backend, action)
     #############
 
     # this is what the backend receive:

examples/backend_integration/Step5_modify_topology.py

@@ -58,7 +58,7 @@ def _aux_change_bus_or_disconnect(self, new_bus, dt, key, el_id):
     # are either 1 or 2)
     def apply_action(self, backendAction: Union["grid2op.Action._backendAction._BackendAction", None]) -> None:
         # the following few lines are highly recommended
-        if action is None:
+        if backendAction is None:
             return
 
         # loads and generators are modified in the previous script
@@ -173,12 +173,12 @@ def get_topo_vect(self) -> np.ndarray:
 if __name__ == "__main__":
     import grid2op
     import os
-    from Step0_make_env import make_env_for_backend
+    from Step0_make_env import make_env_for_backend, create_action
 
     path_grid2op = grid2op.__file__
     path_data_test = os.path.join(os.path.split(path_grid2op)[0], "data")
 
-    env_name = "l2rpn_wcci_2022_dev"
+    env_name = "rte_case5_example"
     # one of:
     # - rte_case5_example: the grid in the documentation (completely fake grid)
     # - l2rpn_case14_sandbox: inspired from IEEE 14
@@ -206,6 +206,7 @@ def get_topo_vect(self) -> np.ndarray:
         sub_id = 1
         local_topo = (1, 2, 1, 2, 1, 2)
     elif env_name == "l2rpn_wcci_2022_dev":
+        raise RuntimeError("Storage units are not handled by the example backend, and there are some on the grid.")
         sub_id = 3
         local_topo = (1, 2, 1, 2, 1)
     else:
@@ -214,8 +215,7 @@ def get_topo_vect(self) -> np.ndarray:
     #############################    
 
     # this is technical to grid2op
-    bk_act = env._backend_action_class()
-    bk_act += action
+    bk_act = create_action(env, backend, action)
     ####################################
 
     # this is what the backend receive:

examples/backend_integration/Step6_integration.py

@@ -12,7 +12,7 @@
 interacts with it.
 
 """
-
+from tqdm import tqdm
 from Step5_modify_topology import CustomBackend_Minimal
 
 
@@ -60,42 +60,48 @@
     ########### First "test" perform nothing and see what it gives
     done = False
     nb_step = 0
-    while True:
-        obs, reward, done, info = env.step(env.action_space())
-        if done:
-            break
-        nb_step += 1
+    with tqdm() as pbar:
+        while True:
+            obs, reward, done, info = env.step(env.action_space())
+            if done:
+                break
+            nb_step += 1
+            pbar.update()
     print(f"{nb_step} steps have been made with your backend with do nothing")
 
     ########## Second "test" perform random actions every now and then
     env.seed(0)
     obs = env.reset()
     done = False
     nb_step = 0
-    while True:
-        if nb_step % 10 == 9:
-            # do a randome action sometime
-            act = env.action_space.sample()
-        else:
-            # do nothing most of the time
-            act = env.action_space()
-        obs, reward, done, info = env.step(act)
-        if done:
-            break
-        nb_step += 1
-    print(f"{nb_step} steps have been made with your backend with random actions")
+    with tqdm() as pbar:
+        while True:
+            if nb_step % 10 == 9:
+                # do a randome action sometime
+                act = env.action_space.sample()
+            else:
+                # do nothing most of the time
+                act = env.action_space()
+            obs, reward, done, info = env.step(act)
+            if done:
+                break
+            nb_step += 1
+            pbar.update()
+    print(f"{nb_step} steps have been made with your backend with some random actions")
 
     ########### Third "test" using an "agent" that "does smart actions" (greedy agent)
     done = False
     nb_step = 0
     obs = env.reset()
     reward = 0.
     agent = RecoPowerlineAgent(env.action_space)
-    while True:
-        act = agent.act(obs, reward)
-        obs, reward, done, info = env.step(act)
-        if done:
-            break
-        nb_step += 1
+    with tqdm() as pbar:
+        while True:
+            act = agent.act(obs, reward)
+            obs, reward, done, info = env.step(act)
+            if done:
+                break
+            nb_step += 1
+            pbar.update()
     print(f"{nb_step} steps have been made with the greedy agent")
 

examples/backend_integration/Step7_optional_make_test.py

@@ -0,0 +1,87 @@
+# Copyright (c) 2019-2020, RTE (https://www.rte-france.com)
+# See AUTHORS.txt
+# 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.
+
+"""
+This script provides a way to run the tests performed by grid2Op for the backend.
+
+These tests are not 100% complete (some things might not be tested and are tested somewhere else)
+but they cover a big part of what the backend is expected to do.
+
+YOU NEED TO INSTALL GRID2OP FROM THE GITHUB REPO FOR THIS TO WORK !
+To do that, simply:
+
+1) clone grid2op repo
+2) cd there 
+3) run `pip install -e .`
+
+(do this in a venv preferably)
+"""
+
+import unittest
+import warnings
+
+# first the backend class (for the example here)
+from Step5_modify_topology import CustomBackend_Minimal
+
+# then some required things
+from grid2op.tests.helper_path_test import PATH_DATA_TEST_PP, PATH_DATA_TEST
+from grid2op.tests.helper_path_test import HelperTests
+PATH_DATA_TEST_INIT = PATH_DATA_TEST
+PATH_DATA_TEST = PATH_DATA_TEST_PP
+
+# then all the tests that can be automatically performed
+from grid2op.tests.BaseBackendTest import BaseTestNames, BaseTestLoadingCase, BaseTestLoadingBackendFunc
+from grid2op.tests.BaseBackendTest import BaseTestTopoAction, BaseTestEnvPerformsCorrectCascadingFailures
+from grid2op.tests.BaseBackendTest import BaseTestChangeBusAffectRightBus, BaseTestShuntAction
+from grid2op.tests.BaseBackendTest import BaseTestResetEqualsLoadGrid, BaseTestVoltageOWhenDisco, BaseTestChangeBusSlack
+from grid2op.tests.BaseBackendTest import BaseIssuesTest, BaseStatusActions
+from grid2op.tests.test_Environment import (TestLoadingBackendPandaPower as BaseTestLoadingBackendPandaPower, 
+                                            TestResetOk as BaseTestResetOk)
+from grid2op.tests.test_Environment import (TestResetAfterCascadingFailure as TestResetAfterCascadingFailure,
+                                            TestCascadingFailure as BaseTestCascadingFailure)
+from grid2op.tests.BaseRedispTest import BaseTestRedispatch, BaseTestRedispatchChangeNothingEnvironment
+from grid2op.tests.BaseRedispTest import BaseTestRedispTooLowHigh, BaseTestDispatchRampingIllegalETC
+from grid2op.tests.BaseRedispTest import BaseTestLoadingAcceptAlmostZeroSumRedisp
+
+# then still some glue code, mainly for the names of the time series
+from grid2op.Converter import BackendConverter
+from grid2op.Backend import PandaPowerBackend
+
+# our backend does not read the names from the grid, so this test is not relevant
+# class TestNames(HelperTests, BaseTestNames):
+#     def make_backend(self, detailed_infos_for_cascading_failures=False):
+#         with warnings.catch_warnings():
+#             warnings.filterwarnings("ignore")
+#             bk = BackendConverter(source_backend_class=CustomBackend_Minimal,
+#                                   target_backend_class=PandaPowerBackend,
+#                                   use_target_backend_name=True,
+#                                   detailed_infos_for_cascading_failures=detailed_infos_for_cascading_failures)
+#         return bk
+
+#     def get_path(self):
+#         return PATH_DATA_TEST_INIT
+
+class TestLoadingCase(HelperTests, BaseTestLoadingCase):
+    def make_backend(self, detailed_infos_for_cascading_failures=False):
+        with warnings.catch_warnings():
+            warnings.filterwarnings("ignore")
+            bk = BackendConverter(source_backend_class=CustomBackend_Minimal,
+                                  target_backend_class=PandaPowerBackend,
+                                  use_target_backend_name=True,
+                                  detailed_infos_for_cascading_failures=detailed_infos_for_cascading_failures)
+        return bk
+
+    def get_path(self):
+        return PATH_DATA_TEST
+
+    def get_casefile(self):
+        return "test_case14.json"
+
+
+if __name__ == "__main__":
+    unittest.main()