`i] ddlZddlZddlmZddlmZddlmZddlm Z ddl m Z e rddlm Z m Z mZmZmZmZddlmZddlZdd lmZd Zd ZGd d ZejdgejddgejddgejddgiZiZe D]\Z!Z"e"D]Z#e!ee#<de$ddfdZ%de$de$fdZ& dFdZ' dFdZ( dFd Z) dFd!Z* dFd"Z+dGd%Z,dHd'd$d(e-dd$fd)Z. dHd*d+d,e$d-d$d(e-ddf d.Z/d/e$de$fd0Z0dId3Z1dJd5Z2d6dd7e3ddfd8Z4d1d9d:e3de3fd;Z5 dKd<Z6eefd1d9d:e3d=e3dd>fd?Z7efd1d@d*d$dAd$d=e3dd@f dBZ8efd1d@d*d$dAd$d:e3dd@f dCZ9dDe$ddfdEZ:dS)LN)deque)deepcopy) getsizeof) TYPE_CHECKING)BLOB_DATA_SUBSTITUTE)AnyCallableDictListOptionalTuple)Span)loggeri Ni'ceZdZdZdZdZdZdS)GEN_AI_ALLOWED_MESSAGE_ROLESsystemuser assistanttoolN)__name__ __module__ __qualname__SYSTEMUSER ASSISTANTTOOLg/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/sentry_sdk/ai/utils.pyrrs" F DI DDDrrrrhumanrair tool_callurlreturnzTuple[str, str]cd|vrtd|dd\}}|dr |dd}n|}|dd}||fS) u Parse a data URI and return (mime_type, content). Data URI format (RFC 2397): data:[][;base64], Examples: data:image/jpeg;base64,/9j/4AAQ... → ("image/jpeg", "/9j/4AAQ...") data:text/plain,Hello → ("text/plain", "Hello") data:;base64,SGVsbG8= → ("", "SGVsbG8=") Raises: ValueError: If the URL is not a valid data URI (missing comma separator) ,z)Invalid data URI: missing comma separatordata:N;r) ValueErrorsplit startswith)r#headercontent mime_part mime_types rparse_data_urir2+s #~~DEEEiiQ''OFG !!122J  $$Q'I g rr1c|sdS|}|drdS|drdS|drdS|ds|drd SdS) a Infer the content modality from a MIME type string. Args: mime_type: A MIME type string (e.g., "image/jpeg", "audio/mp3") Returns: One of: "image", "audio", "video", or "document" Defaults to "image" for unknown or empty MIME types. Examples: "image/jpeg" -> "image" "audio/mp3" -> "audio" "video/mp4" -> "video" "application/pdf" -> "document" "text/plain" -> "document" imagezimage/zaudio/audiozvideo/videoz application/ztext/document)lowerr-)r1 mime_lowers rget_modality_from_mime_typer:Ks$ w""JX&& w   x ( (w   x ( (w   ~ . .*2G2G2P2Pzwr content_partDict[str, Any]Optional[Dict[str, Any]]ct|tsdS|d}|dkrdS|d}t|tr|}n.t|tr|dd}ndS|sdS|dr> t |\}}dt |||dS#t$r d d d|d cYSwxYwd d d|d S) a Transform an OpenAI/LiteLLM content part to Sentry's standardized format. This handles the OpenAI image_url format used by OpenAI and LiteLLM SDKs. Input format: - {"type": "image_url", "image_url": {"url": "..."}} - {"type": "image_url", "image_url": "..."} (string shorthand) Output format (one of): - {"type": "blob", "modality": "image", "mime_type": "...", "content": "..."} - {"type": "uri", "modality": "image", "mime_type": "", "uri": "..."} Args: content_part: A dictionary representing a content part from OpenAI/LiteLLM Returns: A transformed dictionary in standardized format, or None if the format is not OpenAI image_url format or transformation fails. Ntype image_urlr#r(blobr?modalityr1r/urir4r?rDr1rE) isinstancedictgetstrr-r2r:r+)r; block_typeimage_url_datar#r1r/s rtransform_openai_content_partrMmsV. lD ) )t!!&))J[  t!%%k22N.#&& ND ) )  ++t t ~~g  !/!4!4 Iw7 BB&"      #         s*%CC$#C$ct|tsdS|d}|dvsd|vrdS|d}t|tsdS|d}|dd}|dkrdnt|}|dkrd |||d dd S|d krd |||d ddS|dkrd|||dddSdS)a Transform an Anthropic content part to Sentry's standardized format. This handles the Anthropic image and document formats with source dictionaries. Input format: - {"type": "image", "source": {"type": "base64", "media_type": "...", "data": "..."}} - {"type": "image", "source": {"type": "url", "media_type": "...", "url": "..."}} - {"type": "image", "source": {"type": "file", "media_type": "...", "file_id": "..."}} - {"type": "document", "source": {...}} (same source formats) Output format (one of): - {"type": "blob", "modality": "...", "mime_type": "...", "content": "..."} - {"type": "uri", "modality": "...", "mime_type": "...", "uri": "..."} - {"type": "file", "modality": "...", "mime_type": "...", "file_id": "..."} Args: content_part: A dictionary representing a content part from Anthropic Returns: A transformed dictionary in standardized format, or None if the format is not Anthropic format or transformation fails. Nr?)r4r7source media_typerAr7base64rBdatarCr#rErFfilefile_idr?rDr1rTrGrHrIr:)r;rKrO source_typerPrDs r transform_anthropic_content_partrXsd4 lD ) )t!!&))J...(,2N2Nt   h ' 'F fd # #t**V$$KL"--J  # #  ( 4 4 h #zz&"--       #::eR((       #zz)R00     4rct|tsdSd|vrj|d}t|tr>|dd}dt|||dddSdSd|vrj|d}t|tr>|dd}d t|||d dd SdSdS) a Transform a Google GenAI content part to Sentry's standardized format. This handles the Google GenAI inline_data and file_data formats. Input format: - {"inline_data": {"mime_type": "...", "data": "..."}} - {"file_data": {"mime_type": "...", "file_uri": "..."}} Output format (one of): - {"type": "blob", "modality": "...", "mime_type": "...", "content": "..."} - {"type": "uri", "modality": "...", "mime_type": "...", "uri": "..."} Args: content_part: A dictionary representing a content part from Google GenAI Returns: A transformed dictionary in standardized format, or None if the format is not Google format or transformation fails. N inline_datar1rArBrRrC file_datarEfile_urirFrV)r;rZr1r[s rtransform_google_content_partr]s. lD ) )t $$"&&}55 k4 ( ( # R88I7 BB&&??6266   tl"" $$[11 i & & ! k266I7 BB& }}Z44   t 4rcpt|tsdS|d}|dvrdSd|vrdS|dd}|dkr|nd}d |vrd |||d dd Sd |vrd |||d ddSd|vrd|||dddSdS)a Transform a generic/LangChain-style content part to Sentry's standardized format. This handles generic formats where the type indicates the modality and the data is provided via direct base64, url, or file_id fields. Input format: - {"type": "image", "base64": "...", "mime_type": "..."} - {"type": "audio", "url": "...", "mime_type": "..."} - {"type": "video", "base64": "...", "mime_type": "..."} - {"type": "file", "file_id": "...", "mime_type": "..."} Output format (one of): - {"type": "blob", "modality": "...", "mime_type": "...", "content": "..."} - {"type": "uri", "modality": "...", "mime_type": "...", "uri": "..."} - {"type": "file", "modality": "...", "mime_type": "...", "file_id": "..."} Args: content_part: A dictionary representing a content part in generic format Returns: A transformed dictionary in standardized format, or None if the format is not generic format or transformation fails. Nr?)r4r5r6rSrOr1rArSr7rQrBrCr#rErFrTrU)rGrHrI)r;rKr1rDs rtransform_generic_content_partr_1s(6 lD ) )t!!&))J<<<t<t  b11I'611zzzH< "#''"55     ,   "##E2..     l " " "#'' 266     4rct|tsdSt|}||St|}||St |}||St |}||SdS)a Transform a content part from various AI SDK formats to Sentry's standardized format. This is a heuristic dispatcher that detects the format and delegates to the appropriate SDK-specific transformer. For direct SDK integration, prefer using the specific transformers directly: - transform_openai_content_part() for OpenAI/LiteLLM - transform_anthropic_content_part() for Anthropic - transform_google_content_part() for Google GenAI - transform_generic_content_part() for LangChain and other generic formats Detection order: 1. OpenAI: type == "image_url" 2. Google: "inline_data" or "file_data" keys present 3. Anthropic: type in ("image", "document") with "source" key 4. Generic: type in ("image", "audio", "video", "file") with base64/url/file_id Output format (one of): - {"type": "blob", "modality": "...", "mime_type": "...", "content": "..."} - {"type": "uri", "modality": "...", "mime_type": "...", "uri": "..."} - {"type": "file", "modality": "...", "mime_type": "...", "file_id": "..."} Args: content_part: A dictionary representing a content part from an AI SDK Returns: A transformed dictionary in standardized format, or None if the format is unrecognized or transformation fails. N)rGrHrMr]rXr_)r;results rtransform_content_partrbws@ lD ) )t+< 8 8F  +< 8 8F  .l ; ;F  ,L 9 9F   4rr/rc$t|tr|St|ttfr\g}|D]U}t|tr)t |}|||n|@||V|S|S)a Transform message content, handling both string content and list of content blocks. For list content, each item is transformed using transform_content_part(). Items that cannot be transformed (return None) are kept as-is. Args: content: Message content - can be a string, list of content blocks, or other Returns: - String content: returned as-is - List content: list with each transformable item converted to standardized format - Other: returned as-is )rGrJlisttuplerHrbappend)r/ transformeditemras rtransform_message_contentris'3'D%=))   ) )D$%% )/55""V-?66TJJJJ""4(((( NrTrRunpackct|drtj|r d|jdS t |S#t $rX}tjd|t|ttttfr|nt|cYd}~Sd}~wwxYwt|trGr*t|dkrt |dStfd|DSt|t r fd |DSt|ttttfr|nt|S) N model_dumpz rjz+Could not convert pydantic data to JSON: %sr'rc3:K|]}t|VdS)rnN_normalize_data).0xrjs r z"_normalize_data..s0DD!OAf555DDDDDDrc:i|]\}}|t|S)rnrp)rrkvrjs r z#_normalize_data..s,PPP!Q?1V444PPPr)hasattrinspectisclassrrqrl ExceptionrwarningrGintfloatboolrJrdlenrHitems)rRrjes ` rrqrqst\"" T ?4  32$-222 2 T"4??#4#4VDDD D T T T NH! L L L%dS%s,CDDS44#d)) S S S S S S T$E  ;c$ii1nn"476::: :DDDDtDDDDDD$QPPPP4::<<PPPPdS%s$;<< K44#d))Ks"A B7A B2,B72B7spanrkeyvaluect||}t|tttt fr|||dS||tj|dS)Nrn) rqrGr~rrrJset_datajsondumps)rrrrj normalizeds rset_data_normalizedrso!v666J*sE45663 c:&&&&& c4:j1122222rrolec8t||S)z Normalize a message role to one of the 4 allowed gen_ai role values. Maps "ai" -> "assistant" and keeps other standard roles unchanged. )GEN_AI_MESSAGE_ROLE_MAPPINGrI)rs rnormalize_message_rolers ' * *4 6 66rmessageslist[dict[str, Any]]cg}|D]r}t|ts||-|}d|vrt |d|d<||s|S)z Normalize roles in a list of messages to use standard gen_ai role values. Creates a deep copy to avoid modifying the original messages. r)rGrHrfcopyr)rnormalized_messagesmessagenormalized_messages rnormalize_message_rolesrs 77'4((   & &w / / / $\\^^ W  )?)P)P v &""#56666 rCallable[..., Any]cvtj}|duo|jdu}|r tjn tjS)N) sentry_sdkget_current_spancontaining_transaction start_spanstart_transaction) current_spantransaction_existss rget_start_span_functionrsC.00LD T\%HPT%T%7 X: J.>&.H.H $IE   E  X&&M!;;'4((  ++i(( gt $ $ ; ; ;dD));dhhv.>.>&.H.H&:DO rmax_single_message_charsz Tuple[List[Dict[str, Any]], int]ctj|d}t|d}||kr|dfSt ||}|t|kr ||d}nt|dz }|dd}t|dkr't t |d||d<||fS) a Returns a truncated messages list, consisting of - the last message, with its content truncated to `max_single_message_chars` characters, if the last message's size exceeds `max_bytes` bytes; otherwise, - the maximum number of messages, starting from the end of the `messages` list, whose total serialized size does not exceed `max_bytes` bytes. In the single message case, the serialized message size may exceed `max_bytes`, because truncation is based only on character count in that case. rrrrNr'rr)rrrrrrr)rrrserialized_json current_sizetruncation_indextruncated_messagess rtruncate_messages_by_sizersjjAAAO--g6677Ly  {-h BB#h--''%&6&7&78x==1,%bcc] !## K '* + +7O! ! ! 1 / //rzOptional[List[Dict[str, Any]]]scopec|sdSt|}tt|d|}t|dkrt||j|j<|gS)Nrrr')rrrr_gen_ai_original_message_countspan_id)rrrrtruncated_messages rtruncate_and_annotate_messagesrsv t(22HC"*B 8}}q=@]],T\:  rc|sdSt|}t||\}}|dkrt||j|j<|S)Nr)rrrrr)rrrrr removed_counts r&truncate_and_annotate_embedding_inputsrsY t(22H(A(I(V(V% q=@]],T\: rconversation_idcVtj}||dS)z/ Set the conversation_id in the scope. N)rget_current_scopeset_conversation_id)rrs rrrs,  ( * *E o.....r)r;r<r$r=)r/rr$r)T)rrr$r)r$r)rrr$r);rzr collectionsrrrsysrtypingrsentry_sdk._typesrrr r r r r sentry_sdk.tracingrrsentry_sdk.utilsrMAX_GEN_AI_MESSAGE_BYTES MAX_SINGLE_MESSAGE_CONTENT_CHARSrrrrr#GEN_AI_MESSAGE_ROLE_REVERSE_MAPPINGrr target_role source_roles source_rolerJr2r:rMrXr]r_rbrirrqrrrrr~rrrrrrrrrrrs  222222(AAAAAAAAAAAAAAAA''''''######!#) !'( %'8 *[$,? % '< '#!!D!J!J!L!L??K#?? 3>#K00? 1@33DC "C C C C C LD"DDDDDN4"44444nC"CCCCCL8"88888v@LL%LLLLLL4:>33 33#(3263 3333777777$YYYY *-$  %;    PS     L$LLLLLb.$D!0!0$!0!0"!0( !0!0!0!0P%E .  "  & 0. .    & $////////r