ޠj|(SrSSKJr SSKrSSKrSSKJrJrJrJ r J r SSK J r J r Jr SSKrSSKrSSKJr SSKJr SSKJrJrJr SS KJrJrJrJrJr SS KJ r SS K!J"r" \RF"\$5r%S r&SS jr'"SS\\5r(g)z#Base classes for OpenAI embeddings.) annotationsN) AwaitableCallableIterableMappingSequence)AnyLiteralcast) Embeddings)run_in_executor)from_envget_pydantic_field_namessecret_from_env) BaseModel ConfigDictField SecretStrmodel_validator)Self) _resolve_sync_and_async_api_keysic *[U5Vs/sHn/PM nn[U5Vs/sHn/PM nn[[U55HVnU(a[X(5S:XaMXcURX(5 XsUR[X55 MX /n [U5HnXhn [U 5S:XaU RS5 M)[U 5S:XaU RU S5 MN[Xx5n [ U SS06V s/sH#n [S[ XUSS955U - PM% n n [SU 55S-nU RU Vs/sHoU- PM sn5 M U $s snfs snfs sn fs snf) NrstrictFc3,# UH upX-v M g7fN).0valweights Z/var/www/html/gaurav/venv/lib/python3.13/site-packages/langchain_openai/embeddings/base.py 6_process_batched_chunked_embeddings..Fs#WKC #Ws)rc3(# UH oS-v M g7f)Nr)rrs r!r"r#Ps2'3Q'g?)rangelenappendsumzip) num_textstokensbatched_embeddingsindices skip_empty_resultsnum_tokens_in_batchi embeddings_result total_weight embeddingaverage magnituders r!#_process_batched_chunked_embeddingsr;s5:)4D'E4Dq4DG'E 9>i8H+I8H1B8H+I 3w<  #0349  ""#5#89AJ'..s69~> !,.J 9 %,Z w<1    d #  w<1    gaj )  .12 !'8%8  9 #&ya2HQV#W    9  2'22c9 g>gs?g>??B a(F ,J: ?s F F)F ' F c\rSrSr%Sr\"SSS9rS\S'\"SSS9rS\S'S r S \S 'Sr S \S '\ r S\S'\"\ "SSS9SS9r S\S'\"S\ "SSS9S9rS\S'\"\ "SSS9S9rS\S'\"\ "SSS9S9rS\S'SrS\S '\"S!\"S"SS9S9rS#\S$'\"S%\ "S&S'/SS9S9rS\S('SrS)\S*'SrS+\S,'S-rS\S.'S/rS\S0'\"SS1S29rS3\S4'SrS\S5'SrS6\S7'SrS\S8'S9rS6\S:'\"\S9rS;\S<'S9r S6\S='Sr!S>\S?'Sr"S@\SA'SBr#S\SC'SDr$S\SE'Sr%SF\SG'Sr&SF\SH'Sr'S6\SI'\("SJSSKSL9r)\*"SMSN9\+S]SOj55r,\*"SPSN9S^SQj5r-\.S_SRj5r/S`SSjr0SaSTjr1SSU.SbSVjjr2SSU.SbSWjjr3ScSdSXjjr4ScSdSYjjr5SeSZjr6SeS[jr7S\r8g)fOpenAIEmbeddingsVu OpenAI embedding model integration. Setup: Install `langchain_openai` and set environment variable `OPENAI_API_KEY`. ```bash pip install -U langchain_openai export OPENAI_API_KEY="your-api-key" ``` Key init args — embedding params: model: Name of OpenAI model to use. dimensions: The number of dimensions the resulting output embeddings should have. Only supported in `'text-embedding-3'` and later models. Key init args — client params: api_key: OpenAI API key. organization: OpenAI organization ID. If not passed in will be read from env var `OPENAI_ORG_ID`. max_retries: Maximum number of retries to make when generating. request_timeout: Timeout for requests to OpenAI completion API See full list of supported init args and their descriptions in the params section. Instantiate: ```python from langchain_openai import OpenAIEmbeddings embed = OpenAIEmbeddings( model="text-embedding-3-large" # With the `text-embedding-3` class # of models, you can specify the size # of the embeddings you want returned. # dimensions=1024 ) ``` Embed single text: ```python input_text = "The meaning of life is 42" vector = embeddings.embed_query("hello") print(vector[:3]) ``` ```python [-0.024603435769677162, -0.007543657906353474, 0.0039630369283258915] ``` Embed multiple texts: ```python vectors = embeddings.embed_documents(["hello", "goodbye"]) # Showing only the first 3 coordinates print(len(vectors)) print(vectors[0][:3]) ``` ```python 2 [-0.024603435769677162, -0.007543657906353474, 0.0039630369283258915] ``` Async: ```python await embed.aembed_query(input_text) print(vector[:3]) # multiple: # await embed.aembed_documents(input_texts) ``` ```python [-0.009100092574954033, 0.005071679595857859, -0.0029193938244134188] ``` !!! note "OpenAI-compatible APIs (e.g. OpenRouter, Ollama, vLLM)" When using a non-OpenAI provider, set `check_embedding_ctx_length=False` to send raw text instead of tokens (which many providers don't support), and optionally set `encoding_format` to `'float'` to avoid base64 encoding issues: ```python from langchain_openai import OpenAIEmbeddings embeddings = OpenAIEmbeddings( model="...", base_url="...", check_embedding_ctx_length=False, ) ``` NT)defaultexcluder client async_clientztext-embedding-ada-002strmodel int | None dimensionsz str | None deploymentOPENAI_API_VERSION)r? api_version)default_factoryaliasopenai_api_versionbase_urlOPENAI_API_BASE)rKrJopenai_api_baseOPENAI_API_TYPE)rJopenai_api_type OPENAI_PROXY openai_proxyiintembedding_ctx_lengthapi_keyOPENAI_API_KEYzCSecretStr | None | Callable[[], str] | Callable[[], Awaitable[str]]openai_api_key organization OPENAI_ORG_IDOPENAI_ORGANIZATIONopenai_organizationz Literal['all'] | set[str] | Noneallowed_specialz0Literal['all'] | set[str] | Sequence[str] | Nonedisallowed_speciali chunk_sizer% max_retriestimeout)r?rKz(float | tuple[float, float] | Any | Nonerequest_timeoutheadersbooltiktoken_enabledtiktoken_model_nameFshow_progress_bardict[str, Any] model_kwargsr0zMapping[str, str] | Nonedefault_headerszMapping[str, object] | None default_queryretry_min_secondsretry_max_secondsz Any | None http_clienthttp_async_clientcheck_embedding_ctx_lengthforbidr)extrapopulate_by_nameprotected_namespacesbefore)modec z[U5nURS05n[U5HSnXC;aSUS3n[U5eXB;dM [R "SUSUSUS35 UR U5X4'MU URUR55nU(aSUS 3n[U5eX1S'U$) z>Build extra kwargs from additional params that were passed in.rizFound z supplied twice.z WARNING! z/ is not default parameter. zJ was transferred to model_kwargs. Please confirm that z is what you intended.z Parameters za should be specified explicitly. Instead they were passed in as part of `model_kwargs` parameter.) rgetlist ValueErrorwarningswarnpop intersectionkeys)clsvaluesall_required_field_namesrt field_namemsginvalid_model_kwargss r! build_extraOpenAIEmbeddings.build_extra\s$2.v,J"zl*:; o%9 !*.L!))3 4JN %+JJz$:!' 8DDUZZ\R 234ST S/ !!&~ aftercURS;a Sn[U5eSnSnURb[UR5up#URUR UR URURURS.nUR(a_UR(dUR(a=URnURnURnSU<SU<SU<3n[U5eUR(dUcSUl O}UR(a6UR(d%SS KJn UR#URS 9Ul URUS .n [$R&"S 0UDU D6R(Ul UR*(d}UR(a6UR(d%SS KJn UR-URS 9Ul URUS .n [$R."S 0UDU D6R(UlU$![ an S n[!U5U eSn A ff=f![ an S n[!U5U eSn A ff=f)z?Validate that api key and python package exists in environment.)azureazure_adazureadzEIf you are using Azure, please use the `AzureOpenAIEmbeddings` class.N)rYrMrar`rjrkzwCannot specify 'openai_proxy' if one of 'http_client'/'http_async_client' is already specified. Received: openai_proxy=z http_client=z http_async_client=r)httpxzCould not import the httpx package used by the OpenAI SDK. It is normally installed alongside `openai`; reinstall with `pip install openai`.)proxy)rprVr)rQr|rXrr\rOrbr`rjrkrSrprqrAlangchain_openai._compatr ImportErrorClientopenaiOpenAIr5rB AsyncClient AsyncOpenAI) selfrsync_api_key_valueasync_api_key_value client_paramsrSrprqre sync_specificasync_specifics r!validate_environment%OpenAIEmbeddings.validate_environmentysf   #C CW S/ !>BIM    *7W##7 3  !44,,++++#33!//     $"2"2d6L6L,,L**K $ 6 6 !/K>1F4E3GI  S/ !{{!)# $$T-=-=6B(-||$:K:K|'LD$#'#3#31! %mmMmM}MXX     )?)?2>*/):):ARAR):)S&#55.N!' 2 2!! !j   C'6M *#.A5 6"#2I &c*1 2s0:H$I$ I.H<<I I!II!crSUR0UREnURbURUS'U$)NrDrF)rDrirF)rparamss r!_invocation_params#OpenAIEmbeddings._invocation_paramss8At/@/@A ?? &#'??F<  rc8URc Sn[U5eg)z8Check that sync client is available, raise error if not.NzSync client is not available. This happens when an async callable was provided for the API key. Use async methods (ainvoke, astream) instead, or provide a string or sync callable for the API key.)rAr|)rrs r!_ensure_sync_client_available.OpenAIEmbeddings._ensure_sync_client_availables( ;; Q  S/ ! rc /n/n/nUR=(d URnUR(dSSKJn URUS9n [U5HupU RU SS9n [S[U 5UR5Han U XUR-nU RU5nURU5 URU 5 UR[U55 Mc M GOK[R "U5nUR&UR(S.R+5VVs0sHunnUcM UU_M nnn[U5HupURR-S 5(aU R/S S 5n U(aUR"U 40UD6nOUR1U 5n[S[U5UR5H]n URUXUR-5 URU 5 UR[UXUR-55 M_ M UR2(a$SS KJn U"[S[U5U55nO[S[U5U5nUX4U4$![ a Sn[ U5ef=f!["a [R$"S5nGNf=fs snnf![ a [S[U5U5nNof=f) aTokenize and batch input texts. Splits texts based on `embedding_ctx_length` and groups them into batches of size `chunk_size`. Args: texts: The list of texts to tokenize. chunk_size: The maximum number of texts to include in a single batch. Returns: A tuple containing: 1. An iterable of starting indices in the token list for each batch. 2. A list of tokenized texts (token arrays for tiktoken, strings for HuggingFace). 3. An iterable mapping each token array to the index of the original text. Same length as the token list. 4. A list of token counts for each tokenized text. r) AutoTokenizerzCould not import transformers python package. This is needed for OpenAIEmbeddings to work without `tiktoken`. Please install it with `pip install transformers`. )pretrained_model_name_or_pathF)add_special_tokens cl100k_base)r]r^001  )tqdm)rfrDre transformersrrr|from_pretrained enumerateencoder'r(rUdecoder)tiktokenencoding_for_modelKeyError get_encodingr]r^itemsendswithreplaceencode_ordinaryrg tqdm.autor)rtextsr_r-r/ token_counts model_namerr tokenizerr4text tokenizedj token_chunk chunk_textencodingkvencoder_kwargstokenr_iters r! _tokenizeOpenAIEmbeddings._tokenizes*)+"$ --; $$ &6&55.86I%U+'0'7'7QV'7'W q#i.$2K2KLA-6 9 99.K '0&6&6{&CJMM*-NN1% ''K(89M ,  @#66zB (,';';*.*A*A%'.DAq 1 .%U+::&&u-- <<c2D!$OODCNCE$44T:Eq#e*d.G.GHAMM%0I0I,I"JKNN1% ''E!$:S:S6S,T(UVI,$  ! ! :*"&uQF Z'H"I!S[*5Ef|33E &V !o%  &: @#00? @.> :aVj9 :s;J9J.9 KK"KJ+. KK K>=K>)r_c ^^^U=(d TRn0TREUEmTRX5upgp/n Sn U [U5:aSn U n [ U [ X-[U555H(nXnX-[ :a X:XaUS-n O X- n US-n M* X{U nTRR"SSU0TD6n[U[5(dUR5nU RSUS55 U n U [U5:aM[[U5XzUTR5nSmS UUU4SjjnUVs/sHnUbUOU"5PM sn$s snf) aGenerate length-safe embeddings for a list of texts. This method handles tokenization and embedding generation, respecting the `embedding_ctx_length` and `chunk_size`. Supports both `tiktoken` and HuggingFace `transformers` based on the `tiktoken_enabled` flag. Args: texts: The list of texts to embed. engine: The engine or model to use for embeddings. chunk_size: The size of chunks for processing embeddings. Returns: A list of embeddings for each input text. rrinputc3(# UH oSv M g7fr8Nrrrs r!r".s%O>N n>Nr&dataNc>TcNTRR"SSS0TD6n[U[5(dUR 5nUSSSmT$Nrrrr8r)rAcreate isinstancedict model_dumpaverage_embedded_cached_empty_embedding client_kwargsrs r!empty_embeddingBOpenAIEmbeddings._get_len_safe_embeddings..empty_embedding|s^&.#';;#5#5#PB#P-#P !"2D99'7'B'B'D$*:6*B1*Ek*R'* *rrreturn list[float])r_rrr(r'minMAX_TOKENS_PER_REQUESTrArrrrextendr;r0rrenginer_kwargs _chunk_sizerr-r/rr.r4batch_token_count batch_endr chunk_tokens batch_tokensresponser5rrrrs` @@r!_get_len_safe_embeddings)OpenAIEmbeddings._get_len_safe_embeddingsAsx,!3DOO =422=f= /3~~e/Q,w02 #f+o ! I1c!/3v;?@+ $36LL ~$%E !1!E A"I.L{{))N N NHh--#..0  % %%Ohv>N%O OA/#f+o29 JGT__ 7; + +DNN:aQ](99:NNNsEc ^^^# U=(d TRn0TREUEm[STRX5IShvN upgp/n Sn U [ U5:aSn U n [ U [ X-[ U555H(nXnX-[:a X:XaUS-n O X- n US-n M* X{U nTRR"SSU0TD6IShvN n[U[5(dUR5nU RSUS55 U n U [ U5:aM[[ U5XzUTR5nSmS UUU4SjjnUVs/sHnUbUOU"5IShvN PM sn$GN:NNs snf7f) aAsynchronously generate length-safe embeddings for a list of texts. This method handles tokenization and embedding generation, respecting the `embedding_ctx_length` and `chunk_size`. Supports both `tiktoken` and HuggingFace `transformers` based on the `tiktoken_enabled` flag. Args: texts: The list of texts to embed. engine: The engine or model to use for embeddings. chunk_size: The size of chunks for processing embeddings. Returns: A list of embeddings for each input text. Nrrrc3(# UH oSv M g7frrrs r!r"=OpenAIEmbeddings._aget_len_safe_embeddings..rr&rc># TcVTRR"SSS0TD6IShvN n[U[5(dUR 5nUSSSmT$N67fr)rBrrrrrs r!rCOpenAIEmbeddings._aget_len_safe_embeddings..empty_embeddingsw&.)-):):)A)A** -*$ ""2D99'7'B'B'D$*:6*B1*Ek*R'* * $s%A!A7A!rr)r_rr rr(r'rrrBrrrrrr;r0rs` @@r!_aget_len_safe_embeddings*OpenAIEmbeddings._aget_len_safe_embeddingss,!3DOO =422=f= 5D $..%6 0 ,w13 #f+o ! I1c!/3v;?@+ $36LL ~$%E !1!E A"I.L!..55"&3Hh--#..0  % %%Ohv>N%O OA3#f+o69 JGT__ 7; + +JTTAQ]o.?(??TTe0 22)@TsPAF E?BF FAF .1F F3F 4F<F F FF c UR5 U=(d URn0UREUEnUR(d/n[ S[ U5U5HenUR R"SSXXt-0UD6n[U[5(dUR5nURSUS55 Mg U$[[UR5n UR"U4XS.UD6$)aSCall OpenAI's embedding endpoint to embed search docs. Args: texts: The list of texts to embed. chunk_size: The chunk size of embeddings. If `None`, will use the chunk size specified by the class. kwargs: Additional keyword arguments to pass to the embedding API. Returns: List of embeddings, one for each text. rrc3(# UH oSv M g7frrrs r!r"3OpenAIEmbeddings.embed_documents..!K:JQK.:Jr&rrr_r)rr_rrrr'r(rArrrrrr rCrGr rrr_r chunk_size_rr5r4rrs r!embed_documents OpenAIEmbeddings.embed_documentss **, 3DOO =422=f= ..,.J1c%j+6;;--AO48E"(D11'224H!!!K(6:J!KK 7 c4??+,,    ;A  rc # U=(d URn0UREUEnUR(d/n[S[ U5U5HmnUR R "SSXXt-0UD6IShvN n[U[5(dUR5nURSUS55 Mo U$[[UR5n UR"U4XS.UD6IShvN $NN7f)abAsynchronously call OpenAI's embedding endpoint to embed search docs. Args: texts: The list of texts to embed. chunk_size: The chunk size of embeddings. If `None`, will use the chunk size specified by the class. kwargs: Additional keyword arguments to pass to the embedding API. Returns: List of embeddings, one for each text. rrNc3(# UH oSv M g7frrrs r!r"4OpenAIEmbeddings.aembed_documents..rr&rrr)r_rrrr'r(rBrrrrrr rCrGrrs r!aembed_documents!OpenAIEmbeddings.aembed_documentss!3DOO =422=f= ..,.J1c%j+6!%!2!2!9!9"AO4"8E""(D11'224H!!!K(6:J!KK 7 c4??+33    ;A    s%A7C?9C;:A L*!%#$8 9ISW)X K  ', 3 4d '  9=O5L.>VJI04O-415M.5 s7s7"K" %)z( (,+BL(#$6'"P#Pd "`4`4,/`4 J`4R"& DODO DO  DO  DO DOZ"& JUJU JU  JU  JU JUZ:>" " ,6" IL" " J:>! ! ,6! IL! ! F 9 rr=) r,rTr-zlist[list[int] | str]r.rr/z list[int]r0rdrzlist[list[float] | None]))r __future__rloggingr}collections.abcrrrrrtypingr r r rrlangchain_core.embeddingsr langchain_core.runnables.configr langchain_core.utilsrrrpydanticrrrrrtyping_extensionsr*langchain_openai.chat_models._client_utilsr getLoggerrloggerrr;r=rrr!r,s)"LL%% 0;TTMM"W   8 $199 !9*9 9  9  9x^ y*^ r