o `۷iK@sDddlZddlmZddlmZddlmZeGdddeZdS)N)Optional) EnvConfigDict) DeveloperAPIcseZdZdZ    ddedededed eed ef d d Z      ddeedeedeedeed eed eed dfddZ de d dfddZ fddZ Z S) EnvContextaKWraps env configurations to include extra rllib metadata. These attributes can be used to parameterize environments per process. For example, one might use `worker_index` to control which data file an environment reads in on initialization. RLlib auto-sets these attributes when constructing registered envs. rFN env_config worker_index vector_indexremote num_workersrecreated_workercCs.t||||_||_||_||_||_dS)aInitializes an EnvContext instance. Args: env_config: The env's configuration defined under the "env_config" key in the Algorithm's config. worker_index: When there are multiple workers created, this uniquely identifies the worker the env is created in. 0 for local worker, >0 for remote workers. vector_index: When there are multiple envs per worker, this uniquely identifies the env index within the worker. Starts from 0. remote: Whether individual sub-environments (in a vectorized env) should be @ray.remote actors or not. num_workers: The total number of (remote) workers in the set. 0 if only a local worker exists. recreated_worker: Whether the worker that holds this env is a recreated one. This means that it replaced a previous (failed) worker when `restart_failed_env_runners=True` in the Algorithm's config. N)dict__init__rrr r r selfrrrr r r rO/home/ubuntu/vllm_env/lib/python3.10/site-packages/ray/rllib/env/env_context.pyr s  zEnvContext.__init__returncCslt|dur t|n||dur|n|j|dur|n|j|dur!|n|j|dur)|n|j|dur2|S|jS)a]Returns a copy of this EnvContext with some attributes overridden. Args: env_config: Optional env config to use. None for not overriding the one from the source (self). worker_index: Optional worker index to use. None for not overriding the one from the source (self). vector_index: Optional vector index to use. None for not overriding the one from the source (self). remote: Optional remote setting to use. None for not overriding the one from the source (self). num_workers: Optional num_workers to use. None for not overriding the one from the source (self). recreated_worker: Optional flag, indicating, whether the worker that holds the env is a recreated one. This means that it replaced a previous (failed) worker when `restart_failed_env_runners=True` in the Algorithm's config. Returns: A new EnvContext object as a copy of self plus the provided overrides. N)rcopydeepcopyrrr r r rrrrcopy_with_overrides9s zEnvContext.copy_with_overridesdefaultscCs&|D] \}}||vr|||<qdS)aSets missing keys of self to the values given in `defaults`. If `defaults` contains keys that already exist in self, don't override the values with these defaults. Args: defaults: The key/value pairs to add to self, but only for those keys in `defaults` that don't exist yet in self. .. testcode:: :skipif: True from ray.rllib.env.env_context import EnvContext env_ctx = EnvContext({"a": 1, "b": 2}, worker_index=0) env_ctx.set_defaults({"a": -42, "c": 3}) print(env_ctx) .. testoutput:: {"a": 1, "b": 2, "c": 3} N)items)rrkeyvaluerrr set_defaultsas zEnvContext.set_defaultsc s:tddd|jd|jd|jd|jdS)Nz , worker=/z , vector_idx=z , remote=})super__str__rr rr )r __class__rrr{szEnvContext.__str__)rFNF)NNNNNN)__name__ __module__ __qualname____doc__rintboolrr rr rr __classcell__rrr rrsT  ( (r) rtypingrray.rllib.utils.typingrray.util.annotationsrr rrrrrs