finishing the class to loop through episode data

CHANGELOG.rst

@@ -38,7 +38,8 @@ Change Log
 - [FIXED] issue https://github.com/rte-france/Grid2Op/issues/511
 - [FIXED] issue https://github.com/rte-france/Grid2Op/issues/508
 - [ADDED] some classes that can be used to reproduce exactly what happened in a previously run environment
-  see `grid2op.Chronics.FromOneEpisodeData` and `grid2op.Opponent.FromEpisodeDataOpponent`
+  see `grid2op.Chronics.FromOneEpisodeData` and `grid2op.Opponent.FromEpisodeDataOpponent` 
+  and `grid2op.Chronics.FromMultiEpisodeData`
 - [ADDED] An helper function to get the kwargs to disable the opponent (see `grid2op.Opponent.get_kwargs_no_opponent()`)
 - [IMPROVED] doc of `obs.to_dict` and `obs.to_json` (see https://github.com/rte-france/Grid2Op/issues/509)
 

grid2op/Agent/recoPowerLinePerArea.py

@@ -26,7 +26,7 @@ class RecoPowerlinePerArea(BaseAgent):
 
     You can use it like:
     
-    .. code-block::
+    .. code-block:: python
     
         import grid2op
         from grid2op.Agent import RecoPowerlinePerArea

grid2op/Chronics/__init__.py

@@ -12,7 +12,8 @@
     "FromNPY",
     "FromChronix2grid",
     "FromHandlers",
-    "FromOneEpisodeData"
+    "FromOneEpisodeData",
+    "FromMultiEpisodeData"
 ]
 
 from grid2op.Chronics.chronicsHandler import ChronicsHandler
@@ -31,4 +32,6 @@
 from grid2op.Chronics.fromNPY import FromNPY
 from grid2op.Chronics.fromChronix2grid import FromChronix2grid
 from grid2op.Chronics.time_series_from_handlers import FromHandlers
+
 from grid2op.Chronics.fromOneEpisodeData import FromOneEpisodeData
+from grid2op.Chronics.fromMultiEpisodeData import FromMultiEpisodeData

grid2op/Chronics/fromMultiEpisodeData.py

@@ -25,6 +25,82 @@
 
 
 class FromMultiEpisodeData(GridValue): 
+    """This class allows to redo some episode that have been previously run using a runner.
+    
+    It is an extension of the class :class:`FromOneEpisodeData` but with multiple episodes.
+    
+    .. seealso::
+        :class:`grid2op.Chronics.FromOneEpisodeData`if you want to use only one episode
+        
+    .. warning::
+        It has the same limitation as :class:`grid2op.Chronics.FromOneEpisodeData`, including:
+        
+        - forecasts are not saved so cannot be retrieved with this class. You can however
+          use `obs.simulate` and in this case it will lead perfect forecasts.
+        - to make sure you are running the exact same episode, you need to create the environment
+          with the :class:`grid2op.Opponent.FromEpisodeDataOpponent` opponent
+
+    Examples
+    ---------
+    You can use this class this way:
+    
+    First, you generate some data by running an episode with do nothing or reco powerline agent,
+    preferably episode that go until the end of your time series
+    
+    .. code-block:: python
+    
+        import grid2op
+        from grid2op.Runner import Runner
+        from grid2op.Agent import RecoPowerlineAgent
+        
+        path_agent = ....
+        nb_episode = ...
+        env_name = "l2rpn_case14_sandbox"  # or any other name
+        env = grid2op.make(env_name, etc.)
+        
+        # optional (change the parameters to allow the )
+        param = env.parameters
+        param.NO_OVERFLOW_DISCONNECTION = True
+        env.change_parameters(param)
+        env.reset()
+        # end optional
+        
+        runner = Runner(**env.get_params_for_runner(),
+                        agentClass=RecoPowerlineAgent)
+        runner.run(nb_episode=nb_episode,
+                   path_save=path_agent)
+    
+    And then you can load it back and run the exact same environment with the same
+    time series, the same attacks etc. with:
+    
+    .. code-block:: python
+    
+        import grid2op
+        from grid2op.Chronics import FromMultiEpisodeData
+        from grid2op.Opponent import FromEpisodeDataOpponent
+        from grid2op.Episode import EpisodeData
+        
+        path_agent = ....  # same as above
+        env_name = .... # same as above
+        
+        # path_agent is the path where data coming from a grid2op runner are stored
+        # NB it should come from a do nothing agent, or at least
+        # an agent that does not modify the injections (no redispatching, curtailment, storage)
+        li_episode = EpisodeData.list_episode(path_agent)
+        
+        env = grid2op.make(env_name,
+                           chronics_class=FromMultiEpisodeData,
+                           data_feeding_kwargs={"li_ep_data": li_episode},
+                           opponent_class=FromEpisodeDataOpponent,
+                           opponent_attack_cooldown=1,
+                      )
+        # li_ep_data in this case is a list of anything that is accepted by `FromOneEpisodeData`
+
+        obs = env.reset()
+    
+        # and now you can use "env" as any grid2op environment.
+        
+    """
     MULTI_CHRONICS = True
     def __init__(self,
                  path,  # can be None !
@@ -47,8 +123,9 @@ def __init__(self,
                                               start_datetime=start_datetime)
                            for el in li_ep_data
                            ]
-        self._prev_cache_id = 0
-        self.data = self.li_ep_data[0]
+        self._prev_cache_id = len(self.li_ep_data) - 1
+        self.data = self.li_ep_data[self._prev_cache_id]
+        self._episode_data = self.data._episode_data  # used by the fromEpisodeDataOpponent
 
     def next_chronics(self):
         self._prev_cache_id += 1
@@ -74,6 +151,7 @@ def initialize(
             order_backend_subs,
             names_chronics_to_backend=names_chronics_to_backend,
         )
+        self._episode_data = self.data._episode_data 
 
     def done(self):
         return self.data.done()
@@ -88,6 +166,7 @@ def forecasts(self):
         return self.data.forecasts()
 
     def tell_id(self, id_num, previous=False):
+        id_num = int(id_num)
         if not isinstance(id_num, (int, dt_int)):
             raise ChronicsError("FromMultiEpisodeData can only be used with `tell_id` being an integer "
                                 "at the moment. Feel free to write a feature request if you want more.")
@@ -99,8 +178,8 @@ def tell_id(self, id_num, previous=False):
             self._prev_cache_id -= 1
             self._prev_cache_id %= len(self.li_ep_data)
 
-    def get_id(self) -> int:
-        return self._prev_cache_id - 1  # TODO check
+    def get_id(self) -> str:
+        return f'{self._prev_cache_id }'
 
     def max_timestep(self):
         return self.data.max_timestep()

grid2op/Chronics/fromOneEpisodeData.py

@@ -62,6 +62,9 @@ class FromOneEpisodeData(GridValue):
         you to build this class with a complete episode (and not using an agent that games over after a 
         few steps), for example by using the "RecoPowerlineAgent" and the `NO_OVERFLOW_DISCONNECTION` 
         parameters (see example below)
+    
+    .. seealso::
+        :class:`grid2op.Chronics.FromMultiEpisodeData`if you want to use multiple episode data
         
     Examples
     ---------
@@ -128,7 +131,7 @@ class FromOneEpisodeData(GridValue):
     If you want to include perfect forecast (unfortunately you cannot retrieve the original forecasts)
     you can do:
     
-    .. code-block::
+    .. code-block:: python
     
         # same as above
         

grid2op/Environment/baseMultiProcessEnv.py

@@ -616,7 +616,7 @@ def simulate(self, actions):
 
         You can use this feature like:
 
-        .. code-block::
+        .. code-block:: python
 
             import grid2op
             from grid2op.Environment import BaseMultiProcessEnvironment

grid2op/Opponent/__init__.py

@@ -21,4 +21,4 @@
 from grid2op.Opponent.geometricOpponent import GeometricOpponent
 from grid2op.Opponent.geometricOpponentMultiArea import GeometricOpponentMultiArea
 from grid2op.Opponent.fromEpisodeDataOpponent import FromEpisodeDataOpponent
-from grid2op.Opponent.utils import get_kwargs_no_opponent
+from grid2op.Opponent.utils import get_kwargs_no_opponent

grid2op/Opponent/fromEpisodeDataOpponent.py

@@ -10,7 +10,7 @@
 import warnings
 
 from grid2op.Opponent.baseOpponent import BaseOpponent
-from grid2op.Chronics import FromOneEpisodeData
+from grid2op.Chronics import FromOneEpisodeData, FromMultiEpisodeData
 from grid2op.Exceptions import OpponentError
 
 
@@ -75,7 +75,7 @@ def __init__(self, action_space):
         self._warning_cooldown_issued = False
 
     def init(self, partial_env, **kwargs):
-        if not isinstance(partial_env.chronics_handler.real_data, FromOneEpisodeData):
+        if not isinstance(partial_env.chronics_handler.real_data, (FromOneEpisodeData, FromMultiEpisodeData)):
             raise OpponentError("FromEpisodeDataOpponent can only be used with FromOneEpisodeData time series !")
         self._ptr_env = partial_env
         self._attacks = copy.deepcopy(self._ptr_env.chronics_handler.real_data._episode_data.attacks)

grid2op/Opponent/utils.py

@@ -20,7 +20,7 @@ def get_kwargs_no_opponent() -> Dict:
     
     You can use it like
     
-    .. code-block::
+    .. code-block:: python
 
         import grid2op
         env_name = "l2rpn_case14_sandbox"  # or any other name

grid2op/Reward/_alarmScore.py

@@ -39,7 +39,7 @@ class _AlarmScore(AlarmReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import AlarmReward

grid2op/Reward/alarmReward.py

@@ -29,7 +29,7 @@ class AlarmReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import AlarmReward

grid2op/Reward/alertReward.py

@@ -54,7 +54,7 @@ class AlertReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import AlertReward

grid2op/Reward/baseReward.py

@@ -41,7 +41,7 @@ class BaseReward(ABC):
     If you want the environment to compute a reward that is the sum of the flow (this is not a good reward, but
     we use it as an example on how to do it) you can achieve it with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import BaseReward

grid2op/Reward/bridgeReward.py

@@ -22,7 +22,7 @@ class BridgeReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import BridgeReward

grid2op/Reward/closeToOverflowReward.py

@@ -20,7 +20,7 @@ class CloseToOverflowReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import CloseToOverflowReward

grid2op/Reward/constantReward.py

@@ -22,7 +22,7 @@ class ConstantReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import ConstantReward

grid2op/Reward/distanceReward.py

@@ -20,7 +20,7 @@ class DistanceReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import DistanceReward

grid2op/Reward/economicReward.py

@@ -25,7 +25,7 @@ class EconomicReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import EconomicReward

grid2op/Reward/episodeDurationReward.py

@@ -20,7 +20,7 @@ class EpisodeDurationReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import EpisodeDurationReward

grid2op/Reward/flatReward.py

@@ -18,7 +18,7 @@ class FlatReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import FlatReward

grid2op/Reward/gameplayReward.py

@@ -21,7 +21,7 @@ class GameplayReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import GameplayReward

grid2op/Reward/increasingFlatReward.py

@@ -21,7 +21,7 @@ class IncreasingFlatReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import IncreasingFlatReward

grid2op/Reward/l2RPNReward.py

@@ -30,7 +30,7 @@ class L2RPNReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import L2RPNReward

grid2op/Reward/linesCapacityReward.py

@@ -25,7 +25,7 @@ class LinesCapacityReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import LinesCapacityReward

grid2op/Reward/linesReconnectedReward.py

@@ -22,7 +22,7 @@ class LinesReconnectedReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import LinesReconnectedReward

grid2op/Reward/redispReward.py

@@ -27,7 +27,7 @@ class RedispReward(BaseReward):
     ---------
     You can use this reward in any environment with:
 
-    .. code-block:
+    .. code-block:: python
 
         import grid2op
         from grid2op.Reward import RedispReward

grid2op/Rules/rulesByArea.py

@@ -36,7 +36,8 @@ class RulesByArea(BaseRules):
     ---------
     If you want the environment to take into account the rules by area, you can achieve it with:
 
-    .. code-block:
+    .. code-block:: python
+    
         import grid2op
         from grid2op.Rules.rulesByArea import RulesByArea