tPj:(FUdZddlmZddlmZddlmZmZmZddlm Z erddl m Z ddl mZe ezZ neZ dZeed <Gd d eZeDcic]}|j$|c}Zeed <d ddddZeed<dddZeed<dZeed<dedzdeefdZGddZycc}w)a{OTEL GenAI ``gen_ai_latest_experimental`` semantic conventions. Setting ``OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental`` switches the emitted traces to the experimental OTEL GenAI conventions (https://opentelemetry.io/docs/specs/semconv/gen-ai/). Concretely, versus the default LiteLLM output: Request span: - name is ``{operation} {model}`` (e.g. ``chat gpt-4``) instead of ``litellm_request``; span kind is ``CLIENT``. - ``gen_ai.operation.name`` is the actual operation (``chat`` / ``text_completion`` / ``embeddings``) instead of always ``chat``. - the provider is reported as ``gen_ai.provider.name``; the superseded ``gen_ai.system`` and the legacy ``llm.is_streaming`` are dropped. - adds ``gen_ai.request.{frequency_penalty,presence_penalty,top_k,seed}``, ``gen_ai.request.stop_sequences`` (a string array), ``gen_ai.request.stream`` (only when streaming), ``gen_ai.request.choice.count`` (only when n > 1), and ``gen_ai.usage.cache_{creation,read}.input_tokens``. - the non-standard ``raw_gen_ai_request`` child span is no longer created. Events: - the per-message ``gen_ai.content.prompt`` / per-choice ``gen_ai.content.completion`` log events are replaced by a single ``gen_ai.client.inference.operation.details`` log event carrying ``gen_ai.input.messages`` / ``gen_ai.output.messages`` (message content included only when content capture is enabled). )datetime)Enum) TYPE_CHECKINGAnyFinal) safe_dumps)Span)OpenTelemetryConfigOTEL_SEMCONV_STABILITY_OPT_IN!OTEL_SEMCONV_STABILITY_OPT_IN_ENVceZdZdZy)OTELSemconvCategorygen_ai_latest_experimentalN)__name__ __module__ __qualname__GEN_AI_LATEST_EXPERIMENTAL|/home/oem/CyberBrain26-unified/.venv/lib/python3.12/site-packages/litellm/integrations/opentelemetry_utils/gen_ai_semconv.pyrr5s!=rr_SEMCONV_CATEGORY_BY_VALUEz gen_ai.request.frequency_penaltyzgen_ai.request.presence_penaltyzgen_ai.request.top_kzgen_ai.request.seed)frequency_penaltypresence_penaltytop_kseed_SEMCONV_REQUEST_ATTRIBUTESz(gen_ai.usage.cache_creation.input_tokensz$gen_ai.usage.cache_read.input_tokens)cache_creation_input_tokenscache_read_input_tokens_SEMCONV_CACHE_TOKEN_ATTRIBUTESz)gen_ai.client.inference.operation.details_INFERENCE_DETAILS_EVENT_NAMErawNreturnc|s tSd|jdDDchc]}|tvr t|c}Scc}w)zParse the comma-separated OTEL_SEMCONV_STABILITY_OPT_IN value into the set of recognized categories. Unknown tokens are ignored per the spec.c3<K|]}|jywN)strip).0parts r z'parse_semconv_opt_in..Ws>tdjjl>s,)setsplitr)r!tokens rparse_semconv_opt_inr.PsK u ?syy~>   . . #5)  sAc 6eZdZUdZerIded<dedededdfd Z de fd Z d e e ezde e fd Zd e e de e fdZdedefdZdeeeffdZede fdZede defdZdede ddfdZdeddfdZde de dede eeffdZde de deddfdZy)OTELGenAISemconvMixinaXOTEL GenAI ``gen_ai_latest_experimental`` semantic-convention behavior. Mixed into ``OpenTelemetry`` (its only host). Every member is internal to the OTEL integration; the leading underscore marks "subsystem-internal", not "class-private" (the host lives in a sibling module). Members the host calls (the mixin -> host contract): - ``_gen_ai_semconv_latest_experimental`` -- opt-in gate; guards every semconv code path in ``opentelemetry.py``. - ``_gen_ai_operation_name`` -- LiteLLM ``call_type`` -> spec ``gen_ai.operation.name``. - ``_set_semconv_request_attributes`` / ``_set_semconv_cache_token_attributes`` -- add the ``gen_ai.request.*`` / ``gen_ai.usage.cache_*`` span attributes. - ``_emit_inference_details_event`` -- emit the consolidated event. Helpers the host must provide (declared under ``TYPE_CHECKING`` below): ``config``, ``safe_set_attribute``, ``_capture_in_event``, ``_transform_messages_to_otel_semantic_conventions``, ``_transform_choices_to_otel_semantic_conventions``, ``_to_ns``, ``_otel_log_types``. r configspankeyvaluer"Ncyr%r)selfr2r3r4s rsafe_set_attributez(OTELGenAISemconvMixin.safe_set_attributexrcyr%rr6s r_capture_in_eventz'OTELGenAISemconvMixin._capture_in_eventzr8rmessagescyr%r)r6r<s r0_transform_messages_to_otel_semantic_conventionszFOTELGenAISemconvMixin._transform_messages_to_otel_semantic_conventions|r8rchoicescyr%r)r6r?s r/_transform_choices_to_otel_semantic_conventionszEOTELGenAISemconvMixin._transform_choices_to_otel_semantic_conventions~r8rdtcyr%r)r6rBs r_to_nszOTELGenAISemconvMixin._to_nsr8rcyr%rr:s r_otel_log_typesz%OTELGenAISemconvMixin._otel_log_typesr8rcNtj|jjvS)zWhether the ``gen_ai_latest_experimental`` opt-in is active. Every semconv behavior is gated on this; ``False`` => legacy output. )rrr1semconv_stability_opt_inr:s r#_gen_ai_semconv_latest_experimentalz9OTELGenAISemconvMixin._gen_ai_semconv_latest_experimentals #==AeAeeerkwargscP|jddxsd}|x}d|vry}d|vry y)zMap a LiteLLM ``call_type`` to spec ``gen_ai.operation.name``. Substring match (e.g. ``aembedding`` -> ``embeddings``); defaults to ``chat``. call_type embedding embeddingstext_completionchat)get)rJrLss r_gen_ai_operation_namez,OTELGenAISemconvMixin._gen_ai_operation_names?"::k26<" kQ&#'1,(roptional_paramsc tjD]-\}}|j|}||j|||/|jd}|>t |t r|n|g}|j d|Dcgc] }t|c}|jdr|j|dd|jd} t | tr| d kDr|j|d | yyycc}w) zAdd ``gen_ai.request.*`` span attributes from ``optional_params``. Covers the sampling params plus the conditionally-required ``stop_sequences`` / ``stream`` / ``choice.count`` per the spec. Nr2r3r4stopzgen_ai.request.stop_sequencesstreamzgen_ai.request.streamTnzgen_ai.request.choice.count) ritemsrRr7 isinstancelist set_attributestrint) r6r2rU source_key semconv_keyr4rX stop_listrSrZs r_set_semconv_request_attributesz5OTELGenAISemconvMixin._set_semconv_request_attributess  (C'H'H'J Q #J #'' 3E ''T{%'P Q &))&1  (2$'=tD6I   >QZ@[AQ@[ \   x (  # #3JRV # W #&&s+ a !a%  # #3PXY # Z#( A\s=C8 c|sy|jdxsijdxsi}tjD]-\}}|j|}|s|j|||/y)zAdd ``gen_ai.usage.cache_*.input_tokens`` from the usage object. No-op when the payload or the usage values are missing/zero. Nmetadata usage_objectrW)rRrr\r7)r6r2standard_logging_payloadusagerbrcr4s r#_set_semconv_cache_token_attributesz9OTELGenAISemconvMixin._set_semconv_cache_token_attributessu ( 044Z@FBKKN[a_a'F'L'L'N Q #J IIj)E''T{%'P Qr response_objproviderc&t||j|d}|js|S|j|j dxsg}|j |j dg}|rt ||d<|rt ||d<|S)zBuild the attribute payload for the inference-details event. Always includes provider/operation; input/output messages are added only when content capture is enabled and non-empty. Mixin-internal. ) event_namezgen_ai.provider.namezgen_ai.operation.namer<r?zgen_ai.input.messageszgen_ai.output.messages)r rTr;r>rRrAr)r6rJrlrmattrsinput_messagesoutput_messagess r_build_inference_details_attrsz4OTELGenAISemconvMixin._build_inference_details_attrss8$,%)%@%@%H(  %%'L $ U UV\V`V`akVlVrpr s!%!U!UVbVfVfgprtVu!v -7-GE) * .8.IE* + rc |j\}}||jtj|j|j |j |jdd|j|||}|j|y)zEmit the consolidated ``gen_ai.client.inference.operation.details`` log event, correlated to the request span via ``parent_ctx``. Replaces the legacy per-message / per-choice content events. INFON) timestamptrace_idspan_id trace_flagsseverity_number severity_textbody attributes) rFrDrnowrwrxryrursemit) r6rJrlrm otel_logger parent_ctx LogRecordSeverityNumber log_records r_emit_inference_details_eventz3OTELGenAISemconvMixin._emit_inference_details_events%)$8$8$:! >%kk(,,.1((&&"..*// ::6rArrarDtuplerFpropertyrI staticmethodrTrerkrsrrrrr0r0\sT0%%S4ScS#S$S0t0qTRVZZ]M]qbfgkblqi4PT:iZ^_cZdi2X2#29U38_9 fTff t   [D[4[TX[: Q Q[_ QTY\aefiknfnao,%%% % %rr0)rrenumrtypingrrr*litellm.litellm_core_utils.safe_json_dumpsropentelemetry.tracer _Span"litellm.integrations.opentelemetryr r rrr4rrrr r`r+r.r0)categorys0rrs>,,A1F 3;D D ,K!5J>$> Ob$b(X^^X%=$bEb <9 # ! &U$NE*(SuR cDj S1D-E X%X%E%cs B