hH SrSSKrSSKJr SSKrSSKrSSKJrJ r SSK J r J r J r JrJr SS\S\4SjjrS \R$4S jrS \R$4S jrSS \R$4S jjrS \R$4S jrS \R$4SjrS \R$4SjrS\4SjrSS \R$S\S\S\4Sjjrg)aLA set of functions for checking an environment implementation. This file is originally from the Stable Baselines3 repository hosted on GitHub (https://github.com/DLR-RM/stable-baselines3/) Original Author: Antonin Raffin It also uses some warnings/assertions from the PettingZoo repository hosted on GitHub (https://github.com/PettingZoo-Team/PettingZoo) Original Author: J K Terry This was rewritten and split into "env_checker.py" and "passive_env_checker.py" for invasive and passive environment checking Original Author: Mark Towers These projects are covered by the MIT License. N)deepcopy)loggerspaces)check_action_spacecheck_observation_spaceenv_render_passive_checkerenv_reset_passive_checkerenv_step_passive_checkerexactreturnc2^^^[T5[T5Lag[T[5(aMTR5TR5:H=(a% [ UUU4SjTR555$[T[ [ 45(a=[T5[T5:H=(a [ U4Sj[TT555$[T[R5(aTRTR:XaTRTR:XajTR[:Xa[ U4Sj[TT555$T(a[R"TT:H5$[R"TTSSS9$gTT:H$)aAssert equality between data 1 and 2, i.e. observations, actions, info. Args: data_1: data structure 1 data_2: data structure 2 exact: whether to compare array exactly or not if false compares with absolute and relative tolerance of 1e-5 (for more information check [np.allclose](https://numpy.org/doc/stable/reference/generated/numpy.allclose.html)). Returns: If observation 1 and 2 are equivalent Fc3J># UHn[TUTUT5v M g7fNdata_equivalence).0kdata_1data_2r s U/home/james-whalen/.local/lib/python3.13/site-packages/gymnasium/utils/env_checker.py #data_equivalence../s)6 CPa VAYq 5 9 9=s #c3@># UHup[XT5v M g7frr)ro_1o_2r s rrr3s"2 >Q(# Su - ->Qc3@># UHup[XT5v M g7frr)rabr s rrr9s">Qda$Q511>Qrgh㈵>)rtolatol)type isinstancedictkeysalltuplelistlenzipnpndarrayshapedtypeobjectallclose)rrr s```rrr!s@ F|4<' FD ! !{{} - #6 CI;;=6 3   FUDM * *6{c&k) c2 >A&&>Q2 /   FBJJ ' ' <<6<< 'FLLFLL,H||v%>A&&>Q66&F"233;;vvDtLLenvc[R"UR5nSUR;dFSUR;GaURSR[R R LGaURSS9up#X R;dS5eURRcS5e[URR5nUR5upSXPR;dS5eURSS9upcX`R;dS5e[URR5nUR5upXR;dS5eURbURRS Lax[X&5(dS 5e[XX5(dS 5e[X&S S 9(d[R"S5 [XXS S 9(d[R"S5 UR R"UR R":XdS5eURSS9upXR;dS5eURRR R"UR R":wdS5eURR)S5n U b2U R*b$[R"SU R*35 ggg[,R.R1S5e![$an ['SU 35U eSn A ff=f)zCheck that the environment can be reset with a seed. Args: env: The environment to check Raises: AssertionError: The environment cannot be reset with a random seed, even though `seed` or `kwargs` appear in the signature. seedkwargs{r4zVThe observation returned by `env.reset(seed=123)` is not within the observation space.NzExpects the random number generator to have been generated given a seed was passed to reset. Most likely the environment reset function does not call `super().reset(seed=seed)`.zNThe observation returned by `env.reset()` is not within the observation space.FzXUsing `env.reset(seed=123)` is non-deterministic as the observations are not equivalent.zkUsing `env.reset(seed=123)` then `env.reset()` is non-deterministic as the observations are not equivalent.Tr zHUsing `env.reset(seed=123)` observations are not equal although similar.z[Using `env.reset(seed=123)` then `env.reset()` observations are not equal although similar.zMost likely the environment reset function does not call `super().reset(seed=seed)` as the random generates are not same when the same seeds are passed to `env.reset`.izVThe observation returned by `env.reset(seed=456)` is not within the observation space.zMost likely the environment reset function does not call `super().reset(seed=seed)` as the random number generators are not different when different seeds are passed to `env.reset`.zThe environment cannot be reset with a random seed, even though `seed` or `kwargs` appear in the signature. This should never happen, please report this issue. The error was: zThe default seed argument in reset should be `None`, otherwise the environment will by default always be deterministic. Actual default: zLThe `reset` method does not provide a `seed` or `**kwargs` keyword argument.)inspect signaturereset parameterskind Parameter VAR_KEYWORDobservation_space unwrapped _np_randomrspecnondeterministicrrwarn bit_generatorstate TypeErrorAssertionErrorgetdefaultgymerrorError) r2r:obs_1infoseed_123_rng_1obs_2obs_3seed_123_rng_3obs_4obs_5e seed_params rcheck_reset_seed_determinismrYGsX!!#)),I %%%I(((   * / /73D3D3P3P P; )))-KE... hg h. ((4 CC C4%cmm&>&>?N))+KE... `_ `.)))-KE... hg h.%cmm&>&>?N))+KE... `_ `.xx#(A(AU(J'nmn(AAA(DAKKb(DAKKu ,,22n6R6R6X6XX yy yX)))-KE... hg h. ((66<<!//556 GG G6))--f5  !j&8&8&D KK##-#5#5"68 'E ! iioo Z    VVWUXZ  s8H=L&& M0L??Mc[R"UR5nSUR;dDSUR;aEURSR[R R LaUR0S9 g[RRS5e![an[SU35UeSnAff=f)zCheck that the environment can be reset with options. Args: env: The environment to check Raises: AssertionError: The environment cannot be reset with options, even though `options` or `kwargs` appear in the signature. optionsr5)r[zThe environment cannot be reset with options, even though `options` or `**kwargs` appear in the signature. This should never happen, please report this issue. The error was: NzPThe `reset` method does not provide an `options` or `**kwargs` keyword argument.) r9r:r;r<r=r>r?rHrIrLrMrN)r2r:rWs rcheck_reset_optionsr\s!!#)),II(((I(((   * / /73D3D3P3P P  IIbI !iioo ^    VVWUXZ  s6B%% C/B>>CcURbURRSLagURRU5 URR 5nUR US9 UR U5up4pVn[URR5nUR US9 UR U5uppn URRRRURR:XdS5e[X95(dS5e[X9SS9(d[R"S5 [XJ5(dS5e[XJSS9(d[R"S 5 [X[SS9(dS 5eUS LaU S LdS 5e[UU 5(dS 5e[X}SS9(d[R"S5 gg)aCheck that the environment steps deterministically after reset. Note: This check assumes that seeded `reset()` is deterministic (it must have passed `check_reset_seed`) and that `step()` returns valid values (passed `env_step_passive_checker`). Note: A single step should be enough to assert that the state transition function is deterministic (at least for most environments). Raises: AssertionError: The environment cannot be step deterministically after resetting with a random seed, or it truncates after 1 step. NTr7z9The `.np_random` is not properly been updated after step.zODeterministic step observations are not equivalent for the same seed and actionr8zOStep observations are not equal although similar given the same seed and actionzJDeterministic step rewards are not equivalent for the same seed and actionzJStep rewards are not equal although similar given the same seed and actionzNDeterministic step termination are not equivalent for the same seed and actionFzBEnvironment truncates after 1 step, something has gone very wrong.zGDeterministic step info are not equivalent for the same seed and actionzGStep info are not equal although similar given the same seed and action)rCrD action_spacer4sampler;steprrArBrFrGrrrE)r2r4actionobs_0rew_0term_0trunc_0info_0 seeded_rngrOrew_1term_1trunc_1info_1s rcheck_step_determinismrls xx 9 9T A$    $ $ &FII4I,/HHV,<)E&6&.s}}/G/G&HJII4I,/HHV,<)E&6   ..44  # # ) ) *CCC *    YXY  E 5 ]     TST  E 5 X  d XWX  5W-LKL-   QQQ  F$ 7 U  8r1c[R"UR5nSUR;a[R "S5 gg)zMakes sure support for deprecated `return_info` argument is dropped. Args: env: The environment to check Raises: UserWarning return_infoz`return_info` is deprecated as an optional argument to `reset`. `reset`should now always return `obs, info` where `obs` is an observation, and `info` is a dictionarycontaining additional information.N)r9r:r;r<rrE)r2r:s r#check_reset_return_info_deprecationros<!!#)),I ,,,  1 -r1cl[USS5n[U5(a[R"S5 gg)zMakes sure support for deprecated function `seed` is dropped. Args: env: The environment to check Raises: UserWarning r4NzOfficial support for the `seed` function is dropped. Standard practice is to reset gymnasium environments using `env.reset(seed=)`)getattrcallablerrE)r2seed_fns rcheck_seed_deprecationrt s4c64(G  j r1c>UR5n[U[5(dS[U5S35e[ U5S:XdS[ U535eUup#X R ;dS5e[U[ 5(dS[U535eg)zChecks that :meth:`reset` correctly returns a tuple of the form `(obs , info)`. Args: env: The environment to check Raises: AssertionError depending on spec violation zThe result returned by `env.reset()` was not a tuple of the form `(obs, info)`, where `obs` is a observation and `info` is a dictionary containing additional information. Actual type: ``zBCalling the reset method did not return a 2-tuple, actual length: zPThe first element returned by `env.reset()` is not within the observation space.zPThe second element returned by `env.reset()` was not a dictionary, actual type: N)r;r#r'r"r)r@r$)r2resultobsrPs rcheck_reset_return_typerzsYY[F   S CDHIODPCQQRSS  F qZ KCPVK=YZIC $$$ZYZ$  d  g YZ^_cZdYefg r1 space_typec [U[R5(Ga[R"[R "UR [R*55(a[R"SUS35 [R"[R "UR[R55(a[R"SUS35 US:XGa[UR5S:XGa[R"[R"UR [R"UR 5:g[R"UR 5[R"UR5:g55(dP[R"UR S:5(d([R"URS:5(a[R"S5 gggg[U[R 5(aURHn[#X!5 M g[U[R$5(a#UR'5Hn[#X!5 M gg) z]Check the space limit for only the Box space as a test that only runs as part of `check_env`.zA Box z< space minimum value is -infinity. This is probably too low.z< space maximum value is infinity. This is probably too high.razFor Box action spaces, we recommend using a symmetric and normalized space (range=[-1, 1] or [0, 1]). See https://stable-baselines3.readthedocs.io/en/master/guide/rl_tips.html for more information.N)r#rBoxr+anyequallowinfrrEhighr)r- logical_and zeros_likeabsTuplecheck_space_limitDictvalues)spacer{subspaces rrr3s%$$ 66"((599rvvg. / / KK $`a  66"((5::rvv. / / KK $`a   !5;;1$FF!IIuyy)AAFF599- 1CC vveii"n--vvejj1n--KKz.% "" E6<< ( ( H h 3% E6;; ' ' H h 3' (r1rEskip_render_checkskip_close_checkcUb[R"S5 [U[R5(di[ UR R5S:Xd#[ UR R5S:Xa [S5e[S[U5S35eURULa'[R"SUS URS 35 URRS S 5(a [RRU5nO@URRS S 5(a[RRU5n[!US5(d [#S5e[%UR&5 [)UR&S5 [!US5(d [#S5e[+UR,5 [)UR,S5 [/U5 [1U5 [3U5 [5U5 [7U5 [9U5 [;XR&R=55 [?U5 U(dUR@b [CU5 URDb[URSHGnURDRGUS9nURI5 [CU5 URK5 MI O[R"S5 U(dJURDb<URDRG5nURK5 URK5 ggg![La#n[R"SU35 SnAgSnAff=f)aZCheck that an environment follows Gymnasium's API. .. py:currentmodule:: gymnasium.Env To ensure that an environment is implemented "correctly", ``check_env`` checks that the :attr:`observation_space` and :attr:`action_space` are correct. Furthermore, the function will call the :meth:`reset`, :meth:`step` and :meth:`render` functions with a variety of values. We highly recommend users call this function after an environment is constructed and within a project's continuous integration to keep an environment update with Gymnasium's API. Args: env: The Gym environment that will be checked warn: Ignored, previously silenced particular warnings skip_render_check: Whether to skip the checks for the render method. False by default (useful for the CI) skip_close_check: Whether to skip the checks for the close method. False by default Nz/`check_env(warn=...)` parameter is now ignored.zzzGym is incompatible with Gymnasium, please update the environment class to `gymnasium.Env`. See https://gymnasium.farama.org/introduction/create_custom_env/ for more info.zIThe environment must inherit from the gymnasium.Env class, actual class: zQ. See https://gymnasium.farama.org/introduction/create_custom_env/ for more info.zThe environment (z+) is different from the unwrapped version (z). This could effect the environment checker as the environment most likely has a wrapper applied to it. We recommend using the raw environment for `check_env` using `env.unwrapped`.jaxFtorchr^z}The environment must specify an action space. See https://gymnasium.farama.org/introduction/create_custom_env/ for more info.rar@zThe environment must specify an observation space. See https://gymnasium.farama.org/introduction/create_custom_env/ for more info. observation render_modes) render_modezNot able to test alternative render modes due to the environment not having a spec. Try instantiating the environment through `gymnasium.make`z_Calling `env.close()` on the closed environment should be allowed, but it raised an exception: )'rrEr#rLEnvstr __class____base__rHr"rAmetadatarJwrappers JaxToNumpy TorchToNumpyhasattrAttributeErrorrr^rrr@rtrorzrYr\r r r_rlrrrCmaker;close Exception)r2rErrrnew_envrWs r check_envrYs*  EF c377 # #  && '+C C3==))*.JJb  [\`ad\e[fgbb  }}C u$OPSP]P]_U V  ||u%%ll%%c*   '5 ) )ll'', 3 ' ' L  s''(c&&1 3+ , , Q  C112c++];3',C  %c"S"2"2"9"9";<3  ?? & &s + 88 "||N; ((--K-@ *73  < KKa   4((--/   MMO !5    KKqrsqtu   s3M M3M..M3)F)r6)NFF)__doc__r9copyrnumpyr+ gymnasiumrLrr#gymnasium.utils.passive_env_checkerrrrr r boolrrrYr\rlrortrzrrrr1rrs $# D# T# LV cggV r SWW 8: : z SWW "    gg2#4#4P#" e e ee er1