%GglUddlZddlmZddlmZmZmZmZmZm Z ddl m Z ddl m Z ddlZddlmZddlmZddlmZdd lmZdd lmZdd lmZdd lmZmZdd l m!Z!ddl"m#Z#ddl$m%Z%ddl&m'Z'ddl(m)Z)ddl*m+Z+ddl,m-Z- ddl.m/Z/m0Z0m1Z1m2Z2ddl3m4Z4ddl5m6Z7e0r<e0jrDcic]!}|d|de'jtjvf#c}niZe?e@e'jtffeAd<e/r#e/jrDcic]}|d| c}niZBe=e>e?e@e'jtffeAd<e/r4e/jrDchc]}|jddr|dc}neDZEeDe>eAd<e1r#e1jrDcic]}|d| c}niZFe=e>e?e@e'jtffeAd<e2r#e2jrDcic]}|d| c}niZGe=e>e?e@e'jtffeAd<Gdd eZHy#e8$rdZ0dZ/dZ4dZ1dZ2dZ7Y~wxYwcc}wcc}wcc}wcc}wcc}w)!N)tee)AnyIterableOptionalSequenceUnionget_args)deepcopy)Path) BaseModel) QdrantBase) common_types) GrpcToRest)INFERENCE_OBJECT_TYPES)InspectorEmbed) NumericVectorNumericVectorStruct)ModelSchemaParser) FieldPath) QueryResponse)models)reciprocal_rank_fusion)grpc) show_warning)SparseTextEmbedding TextEmbeddingLateInteractionTextEmbeddingImageEmbedding) OnnxProvider)ImagemodeldimSUPPORTED_EMBEDDING_MODELS!SUPPORTED_SPARSE_EMBEDDING_MODELS requires_idfIDF_EMBEDDING_MODELS"_LATE_INTERACTION_EMBEDDING_MODELS_IMAGE_EMBEDDING_MODELSc eZdZUdZiZeedfed<iZeedfed<iZ eedfed<iZ eedfed <e ed <d e d e ffd ZedefdZedeefdZ dVdedeedeedeedeedde deeede d e ddfdZ dWdeedeedeedeedde deeede d e ddfdZedXdZededeeej8ffdZe dYdedeedeedeedd e ddf d Ze dYdedeedeedeedd e ddf d!Ze dYdedeedeedeedd e ddf d"Z e dYdedeedeedeedd e ddf d#Z!ed$d%dfd&e"eded'ed(ed)eede"eeee#ff d*Z$ed$dfd&e"eded'ed)eede"e%jLf d+Z'defd,Z(deefd-Z)d.ee%jTdee+fd/Z, dZd0ee"ejZd1ee"eee fd2e"eeee#fd3ed4ee"e%jLde"ej\f d5Z/d6ej`ddfd7Z1 dYd8ee d9eejdd:eejfdeeejhffd;Z5 d[d8ee ded&e"ed1ee"eee fd0ee"ejZd'ed)eed e dee9eeffd?Z: d]d>ed@edAeejvdBed e dee+f dCZ< d]d>edDeedAeejvdBed e deee+f dEZ=edFe9e%j|ee#eee#e%jLe%j~e%jejejejdf deej~fdGZDdFejdejfdHZFdIeejdeejfdJZG d^dKeHdLeeeIdMe de9eHeJffdNZKeLdOejdejfdPZN d_dOejdMe deOfdQZPd_dRejdMe deJfdSZQdTejdeJfdUZRxZSS)`QdrantFastembedMixinzBAAI/bge-small-enrembedding_modelsrsparse_embedding_modelsr!late_interaction_embedding_modelsrimage_embedding_models_FASTEMBED_INSTALLEDparserkwargsc Rd|_d|_t||_ ddlm}m}t|jdkDsJt|jdkDsJd|j_ t|4di|y#t$rd|j_ Y,wxYw)N)r0r)rrTF)_embedding_model_name_sparse_embedding_model_namer_embed_inspector fastembedrrlenlist_supported_models __class__r/ ImportErrorsuper__init__)selfr0r1rrr:s {/home/kushmeetdev/apache_webroot/langgraph_flaskproject/venv/lib/python3.12/site-packages/qdrant_client/qdrant_fastembed.pyr=zQdrantFastembedMixin.__init__Zs48";?) .f = 8 D@*@@BCaG GG:}::<=A AA26DNN / "6" 827DNN / 8sAB B&%B&returncT|j|j|_|jSN)r4DEFAULT_EMBEDDING_MODELr>s r?embedding_model_namez)QdrantFastembedMixin.embedding_model_namejs(  % % -)-)E)ED &)))c|jSrB)r5rDs r?sparse_embedding_model_namez0QdrantFastembedMixin.sparse_embedding_model_nameps000rFNrE max_length cache_dirthreads providersrcuda device_ids lazy_loadc n|tdtd|jd|||||||d| ||_y)a\ Set embedding model to use for encoding documents and queries. Args: embedding_model_name: One of the supported embedding models. See `SUPPORTED_EMBEDDING_MODELS` for details. max_length (int, optional): Deprecated. Defaults to None. cache_dir (str, optional): The path to the cache directory. Can be set using the `FASTEMBED_CACHE_PATH` env variable. Defaults to `fastembed_cache` in the system's temp directory. threads (int, optional): The number of threads single onnxruntime session can use. Defaults to None. providers: The list of onnx providers (with or without options) to use. Defaults to None. Example configuration: https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#configuration-options cuda (bool, optional): Whether to use cuda for inference. Mutually exclusive with `providers` Defaults to False. device_ids (Optional[list[int]], optional): The list of device ids to use for data parallel processing in workers. Should be used with `cuda=True`, mutually exclusive with `providers`. Defaults to None. lazy_load (bool, optional): Whether to load the model during class initialization or on demand. Should be set to True when using multiple-gpu and parallel encoding. Defaults to False. Raises: ValueError: If embedding model is not supported. ImportError: If fastembed is not installed. Returns: None Nzhmax_length parameter is deprecated and will be removed in the future. It's not used by fastembed models.)messagecategory stacklevel model_namerJrKrLrMrNrOr3)rDeprecationWarning_get_or_init_modelr4) r>rErIrJrKrLrMrNrOr1s r? set_modelzQdrantFastembedMixin.set_modelts^N  ! 5+    +!  &:"rFc J||jd|||||||d|||_y)a Set sparse embedding model to use for hybrid search over documents in combination with dense embeddings. Args: embedding_model_name: One of the supported sparse embedding models. See `SUPPORTED_SPARSE_EMBEDDING_MODELS` for details. If None, sparse embeddings will not be used. cache_dir (str, optional): The path to the cache directory. Can be set using the `FASTEMBED_CACHE_PATH` env variable. Defaults to `fastembed_cache` in the system's temp directory. threads (int, optional): The number of threads single onnxruntime session can use. Defaults to None. providers: The list of onnx providers (with or without options) to use. Defaults to None. Example configuration: https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#configuration-options cuda (bool, optional): Whether to use cuda for inference. Mutually exclusive with `providers` Defaults to False. device_ids (Optional[list[int]], optional): The list of device ids to use for data parallel processing in workers. Should be used with `cuda=True`, mutually exclusive with `providers`. Defaults to None. lazy_load (bool, optional): Whether to load the model during class initialization or on demand. Should be set to True when using multiple-gpu and parallel encoding. Defaults to False. Raises: ValueError: If embedding model is not supported. ImportError: If fastembed is not installed. Returns: None NrUr3)_get_or_init_sparse_modelr5) r>rErJrKrLrMrNrOr1s r?set_sparse_modelz%QdrantFastembedMixin.set_sparse_modelsJJ + *D * * /##%#  -A)rFc2|jrytd)Nzjfastembed is not installed. Please install it to enable fast vector indexing with `pip install fastembed`.)r/r;)clss r?_import_fastembedz&QdrantFastembedMixin._import_fastembeds#  # #  ^  rFrVcn|j|tvrtd|dtt|S)NUnsupported embedding model: . Supported models: )r_r# ValueError)r^rVs r?_get_model_paramsz&QdrantFastembedMixin._get_model_paramssD  7 7/ |;OPjOkl **55rFc ||jvr|j|S|j|tvrtd|dtt d||||d||j|<|j|SNrarb)rVrJrKrLr3)r+r_r#rcrr^rVrJrKrLr1s r?rXz'QdrantFastembedMixin._get_or_init_models -- -'' 3 3  7 7/ |;OPjOkl ,9, ! ,  , Z(##J//rFc ||jvr|j|S|j|tvrtd|dtt d||||d||j|<|j|Srf)r,r_r$rcrrgs r?r[z.QdrantFastembedMixin._get_or_init_sparse_models 44 4..z: :  > >/ |;OPqOrs 3F3 ! 3  3 ##J/**:66rFc ||jvr|j|S|j|tvrtd|dtt d||||d||j|<|j|Srf)r-r_r'rcrrgs r?#_get_or_init_late_interaction_modelz8QdrantFastembedMixin._get_or_init_late_interaction_model/s >> >88D D  ? ?/ |;OPrOst =Y= ! =  = --j944Z@@rFc ||jvr|j|S|j|tvrtd|dtt d||||d||j|<|j|Srf)r.r_r(rcrrgs r?_get_or_init_image_modelz-QdrantFastembedMixin._get_or_init_image_modelKs 33 3--j9 9  4 4/ |;OPgOhi 2@2 ! 2  2 "":.))*55rF default documents batch_size embed_typeparallelc#F K|j| t|d\}}|dk(r j|||}n9|dk(r fd|D}n(|dk(r j|||}nt d|t ||D]\} } | | j fyw) NrVpassagerprrqueryc3ZK|]"}tj|d$yw)rxrN)list query_embed).0rxembedding_models r? z8QdrantFastembedMixin._embed_documents..vs.FK_00u0=>qAs(+rnzUnknown embed type: )rXr passage_embedembedrcziptolist) r>rorErprqrr documents_a documents_b vectors_itervectordocr~s @r?_embed_documentsz%QdrantFastembedMixin._embed_documentsgs11=Q1R#&y!#4 [  "*88 X9L7 "OZL9 $*00 X1L3J<@A A|[9 'KFCv}}& & 'sBB!c#K|j|}|j|||}|D]K}tj|jj |j j Myw)Nrtrwindicesvalues)r[rtypes SparseVectorrrr)r>rorErprrsparse_embedding_modelr sparse_vectors r?_sparse_embed_documentsz,QdrantFastembedMixin._sparse_embed_documentss"&!?!?K_!?!`-33 *x4 * M$$%--446$++224  sA8A:cd|jjddj}d|S) Returns name of the vector field in qdrant collection, used by current fastembed model. Returns: Name of the vector field. /zfast-)rEsplitlowerr>rVs r?get_vector_field_namez*QdrantFastembedMixin.get_vector_field_names5 ..44S9"=CCE zl##rFc~|j1|jjddj}d|Sy)rNrrz fast-sparse-)rHrrrs r?get_sparse_vector_field_namez1QdrantFastembedMixin.get_sparse_vector_field_namesD  + + 799??DRHNNPJ!*. .rF scored_pointsc g}|j}|j}|D]}t|jtr|jj |dnd}d}|8t|jtr|jj |dnd}|j t|j|||j|jj dd|j|S)Ndocument)id embeddingsparse_embeddingmetadatarscore) rr isinstancerdictgetappendrrpayloadr)r>rresponsevector_field_namesparse_vector_field_name scored_pointrrs r?!_scored_points_to_query_responsesz6QdrantFastembedMixin._scored_points_to_query_responsess 668#'#D#D#F ) Ll1148##''(94@  $ '3","5"5t<!''++,DdK! OO#'%5)11)1155j"E&,,   0rFidsr encoded_docsids_accumulatorsparse_vectorsc#XK| tdd}| tdd}| tdd}|j}|j}t||||D]H\}} \} } } |j |d| i| } || i}|| | ||<t j || |Jyw)Nc>tjjSrB)uuiduuid4hexr3rFr?z7QdrantFastembedMixin._points_iterator..stzz|//rFciSrBr3r3rFr?rz7QdrantFastembedMixin._points_iterator..sBrFcyrBr3r3rFr?rz7QdrantFastembedMixin._points_iterator..srFTr)rrr)iterrrrrr PointStruct)r>rrrrr vector_namesparse_vector_nameidxmetarrrr point_vectors r?_points_iteratorz%QdrantFastembedMixin._points_iterators ;/6C  J-H  !!,5N002 !>>@7: <8  S 3C}Vm  " "3 '!3/$/G6A65JL!--2K3@ /0$$W\R R SsB(B*collection_infoc&|j|j\}}|j}t|jj j ts(Jd|jj j ||jj j vs+Jd|jj j d||jj j |}||jk(sJd|d|j||jk(sJd|d|j|j}|||jj jvs(Jd|jj j |jtvra|jj j|j}|tj j"k(sJ|jd|yyy)Nrtz,Collection have incompatible vector params: z , expected zEmbedding size mismatch: z != zDistance mismatch: z, requires modifier IDF, current modifier is )rdrErrconfigparamsvectorsrsizedistancerrrHr&modifierrModifierIDF)r>rembeddings_sizerr vector_paramsrrs r?_validate_collection_infoz.QdrantFastembedMixin._validate_collection_infos4$($:$:dF_F_$:$`! 668  " " ) ) 1 14  b 9/:P:P:W:W:_:_9` a b !7!7!>!>!F!F F @ 9/:P:P:W:W:_:_9``kl}k~  @ F(..55==>OP  }11 1 Q &&7tMT>T>[>[>c>c=de fX//3GG*1188GG,( 3 33o6677cdlcmno3 H 0rFon_diskquantization_config hnsw_configc|j}|j|j\}}|tj|||||iS)a Generates vector configuration, compatible with fastembed models. Args: on_disk: if True, vectors will be stored on disk. If None, default value will be used. quantization_config: Quantization configuration. If None, quantization will be disabled. hnsw_config: HNSW configuration. If None, default configuration will be used. Returns: Configuration for `vectors_config` argument in `create_collection` method. rt)rrrrr)rrdrEr VectorParams)r>rrrrrrs r?get_fastembed_vector_paramsz0QdrantFastembedMixin.get_fastembed_vector_paramssZ"!668$($:$:dF_F_$:$`! v22$!$7'    rFrc|j}|jtvr|tjj n|}|y|tj tj||iS)a Generates vector configuration, compatible with fastembed sparse models. Args: on_disk: if True, vectors will be stored on disk. If None, default value will be used. modifier: Sparse vector queries modifier. E.g. Modifier.IDF for idf-based rescoring. Default: None. Returns: Configuration for `vectors_config` argument in `create_collection` method. N)r)indexr)rrHr&rrrSparseVectorParamsSparseIndexParams)r>rrrs r?"get_fastembed_sparse_vector_paramsz7QdrantFastembedMixin.get_fastembed_sparse_vector_params+ss!==?  + +/C C.6.>v**HH  $ v88..#"    rFcollection_namec |j||j|d|}d} |j|j||j||} |j |} |j| g} |j|||| | } |jd || d|xsd |d || S#t $rE|j ||j|j|j |} YwxYw) a Adds text documents into qdrant collection. If collection does not exist, it will be created with default parameters. Metadata in combination with documents will be added as payload. Documents will be embedded using the specified embedding model. If you want to use your own vectors, use `upsert` method instead. Args: collection_name (str): Name of the collection to add documents to. documents (Iterable[str]): List of documents to embed and add to the collection. metadata (Iterable[dict[str, Any]], optional): List of metadata dicts. Defaults to None. ids (Iterable[models.ExtendedPointId], optional): List of ids to assign to documents. If not specified, UUIDs will be generated. Defaults to None. batch_size (int, optional): How many documents to embed and upload in single request. Defaults to 32. parallel (Optional[int], optional): How many parallel workers to use for embedding. Defaults to None. If number is specified, data-parallel process will be used. Raises: ImportError: If fastembed is not installed. Returns: List of IDs of added documents. If no ids provided, UUIDs will be randomly generated on client side. rv)rorErprqrrN)rorErprr)r)rvectors_configsparse_vectors_config)rrrrrT)rpointswaitrrrpr3) rrErHrget_collection Exceptioncreate_collectionrrrr upload_points) r>rrorrrprrr1rencoded_sparse_docsr inserted_idsrs r?addzQdrantFastembedMixin.addIsOV,,!%!:!:!  - #  + + 7"&">">#%)%E%E%! #?#  S"11/1RO &&7 &&%(. '   +]!    ; S  " " /#??A&*&M&M&O #  #11/1RO  SsB((A C65C6 query_text query_filterlimitc |j|j}t|j|}|dj }|j J|j |jd |tj|j|||dd|S|j|j } t| j|d} tj| jj | jj } tjd tj|j|||dd|} tjd tj |j#| ||dd|} |j%|| | g \}}|j t'||g| S) aO Search for documents in a collection. This method automatically embeds the query text using the specified embedding model. If you want to use your own query vector, use `search` method instead. Args: collection_name: Collection to search in query_text: Text to search for. This text will be embedded using the specified embedding model. And then used as a query vector. query_filter: - Exclude vectors which doesn't fit given conditions. - If `None` - search among all vectors limit: How many results return **kwargs: Additional search parameters. See `qdrant_client.models.SearchRequest` for details. Returns: list[types.ScoredPoint]: List of scored points. rtrzrnamerT)r query_vectorrr with_payloadrrfilterrrrrequestsrr3)rXrEr{r|rrHrsearchr NamedVectorrr[rrr SearchRequestNamedSparseVectorr search_batchr)r>rrrrr1embedding_model_inst embeddingsrsparse_embedding_model_instrsparse_query_vector dense_requestsparse_requestdense_request_responsesparse_request_responses r?rxzQdrantFastembedMixin.querys : $66$B[B[6\.:::LM !!}++-  + + 399 $3!'!3!3!779,"".!%    '+&D&D77'E' #8DD:DVWXYZ $11!))002 ''..0  ,,  %%//1#      --  ++668*     ;?:K:K+}n6U;L; 7 755 "$:rrrrr1r query_vectorsrrrequest responsesrrrsparse_query_vectorsdense_responsessparse_responsesdense_responsesparse_responses r? query_batchz QdrantFastembedMixin.query_batchs8 $66$B[B[6\1==K=PQ # %F**))335fmmo$! G OOG $ %  + + 3)) /!*IV__D::8D_ _&*&D&D77'E' #"=!B!B[!B!Y     %--446$++224   2 %M** //::<($!  G OOG $ %%%+& $$6c+&67$S%5%7847HX3Y / #NO#DE R  R[[X66x@[[M`  8 \sH) AH..H3H9rxct|ttjst|tjr|St|tj rt j|St|tjr$t j|jSt|trt j|St|ttjrGt|tjrtj|n|}t j|St|trt j|S|yt!dt#|)acResolves query interface into a models.Query object Args: query: models.QueryInterface - query as a model or a plain structure like list[float] Returns: Optional[models.Query]: query as it was, models.Query(nearest=query) or None Raises: ValueError: if query is not of supported type )nearestNzUnsupported query type: )rr rQueryrrr NearestQuerynpndarrayrr{PointIdrconvert_point_idrrctype)r^rxs r?_resolve_queryz#QdrantFastembedMixin._resolve_queryTs6 eXekk2 3z%7TL eU// 0&&u5 5 eRZZ (&&u||~> > eT "&&u5 5 eXemm4 56@ 6U ++E2[` &&u5 5 e3 4&&u5 5 =3DK=ABBrFc\t|}|j|j|_|S)zResolve QueryRequest query field Args: query: models.QueryRequest - query request to resolve Returns: models.QueryRequest: A deepcopy of the query request with resolved query field )r rrx)r>rxs r?_resolve_query_requestz+QdrantFastembedMixin._resolve_query_requests())%++6  rFrcJ|Dcgc]}|j|c}Scc}w)aResolve query field for each query request in a batch Args: requests: Sequence[models.QueryRequest] - query requests to resolve Returns: Sequence[models.QueryRequest]: A list of deep copied query requests with resolved query fields )r)r>rrxs r?_resolve_query_batch_requestz1QdrantFastembedMixin._resolve_query_batch_requests%AIIu++E2IIIs r!pathsis_queryc h|It|tr|j||St|}|jj |}|D]}t|t s|gn|}|D]}t||jd}||jr|j||j|Ht|t }t|t s|gn|}|D cgc]} |j| |} } |rt||j| t||j| d|Scc} w)aEmbed model's fields requiring inference Args: model: Qdrant http model containing fields to embed paths: Path to fields to embed. E.g. [FieldPath(current="recommend", tail=[FieldPath(current="negative", tail=None)])] is_query: Flag to determine which embed method to use. Defaults to False. Returns: A deepcopy of the method with embedded fields Nrr) rr_embed_raw_datar r6inspectr{getattrcurrenttail _embed_modelssetattr) r>r!rrpath list_modelitem current_modelwas_listdatars r?r'z"QdrantFastembedMixin._embed_modelss= =%!78++EH+EEUOE))11%8E CD(25$(?%UJ" C 'dllD A  (99&&}dii(&S)->H/9-/NTa"S`"JN,,TH,E"J" dllJ?dllJqMB# C C( "sD/r.ct|tjs|S|j}|j}|j }|gt jtjtjvrtj|||S|tvrtj|||St|d)aResolve inference object into a model Args: data: models.VectorStruct - data to resolve, if it's an inference object, convert it to a proper type, otherwise - keep unchanged Returns: models.VectorStruct: resolved data )r!textoptions)r!imager1 is not among supported models)rrInferenceObjectr!objectr1r#keysr$r'Documentr(r rc)r.rVvaluer1s r?_resolve_inference_objectz.QdrantFastembedMixin._resolve_inference_objects$ 6 67KZZ  ,,  ' , , . . 3 3 5 0 4 4 6  ??%Q Q 0 0<r.rkeyr8s r?r"z$QdrantFastembedMixin._embed_raw_datas--d3 dFOO ,''x'@ @ fll +$$T* * d #VZV`V`VbHRUT))%()CC d # 47E2 PTUuD(((BU U Vs C9C?rc|j}|j}|jxsi}|tvrs|jdd|i|}|s/t |j |gdj}|St |j|dj}|S|tvr|jdd|i|}|s t |j |gd} nt |j|d} tj| jj| jjS|tvrs|j dd|i|} |s/t | j |gdj}|St | j|dj}|St#|d)aOEmbed a document using the specified embedding model Args: document: Document to embed is_query: Flag to determine which embed method to use. Defaults to False. Returns: NumericVector: Document's embedding Raises: ValueError: If model is not supported rVrrrzrr3r3)r!r0r1r#rXr{rrr|r$r[rrrrr'rjrc) r>rrrVr0r1rrrrli_embedding_model_insts r?r;z$QdrantFastembedMixin._embed_document s^^ }}""(b 3 3#:4#:#:#\j#\T[#\  !5!;!;tf!;!MNqQXXZ  !!5!A!A!A!MNqQXXZ   < <*H$*H*H+%+)0+ '#'(C(I(IUYTZ(I([#\]^#_ #'(C(O(OVZ(O([#\]^#_ &&(00779BRBYBYB`B`Bb = =&Nd&N&N'%')0' # !8!>!>$!>!PQRST[[]  !!8!D!D4!D!PQRST[[]   |+IJK KrFr2c|j}|tvr|jdd|i|jxsi}t |j t ttjfs.tdt|j d|j t|j|j gdj}|St|d)zEmbed an image using the specified embedding model Args: image: Image to embed Returns: NumericVector: Image's embedding Raises: ValueError: If model is not supported rVzUnsupported image type: z . Image: )imagesrr3r3)r!r(rlr1rr2strr PilImager rcrr{rr)r>r2rVrrs r?r<z!QdrantFastembedMixin._embed_image8s[[ 0 0#@4#@#@$%$*/--*=2$ ekkCx~~+FG!.tEKK/@.A5;;-X177 }7MNqQXXZI J<'EFGGrF)NNNNFNF)NNNFNF)r@N)NNNrB)NN)NNrmN)N )NF)F)T__name__ __module__ __qualname__rCr+rrD__annotations__r,r-r.boolrrr=propertyrErrHintrr{rYr\ classmethodr_tuplerDistancerdrXr[rjrlrr>rrrrrr ScoredPointrrExtendedPointIdrrCollectionInforQuantizationConfigHnswConfigDiffrrrrrrrFilterrxrrr NumpyArrayr7r r4r QueryRequestrrr rrr' staticmethod VectorStructr9rr"r;r< __classcell__)r:s@r?r*r*Qsr 135d3/05@BT#'<"<=BSU%tC1O,O'PU:<D&6!67<#0#C# *c** 1Xc]11 %)#'!%8<*.9:!9:SM9:C= 9: # 9: H^45 9:9:T#Y'9:9:9: 9:|$(!%8<*.0A&sm0AC=0A# 0A H^45 0A  0AT#Y'0A0A0A 0Ad  6365foo9M3N66$(!%8< 00C=0# 0 H^45 0  0 006$(!%8< 77C=7# 7 H^45 7  7 776$(!%8< AAC=A# A H^45 A  A (AA6$(!%8< 66C=6# 6 H^45 6  6 66<%<#"& 'C='"' '  ' 3- ' %T%[() *'>%<"& C="  3-  %$$ % &$s$ hsm  E--.  m  PBF S hv556 7S8DcN34SuS$u+%567 S  S !%*<* S &$$ %S<"o9N9N"oSW"oL#'CG7;  $ &f&?&?@ f334  c6&&& '  >#'.2 $ 6??+  $sF5556 7  D8<:>"&\\C=\8DcN34 \ hv556 7 \  \3-\\ eCHo \D15 R R R v}}- R  R  R  m R p15 W\W\#YW\v}}- W\  W\  W\ d=! "W\r1C MM K e     KK    OO LL  " "    1C &,, 1C1Cf F,?,? FDWDW  J !4!45 J &%% & J ,0 ))Y() ) y-' ( )VH(;(;H@S@SHH>!!  >-L-L4-LTa-L^H&,,H=HrFr*)Ir itertoolsrtypingrrrrrr copyr pathlibr numpyrpydanticr qdrant_client.client_baser qdrant_client.conversionsrr$qdrant_client.conversions.conversionrqdrant_client.embed.commonr#qdrant_client.embed.embed_inspectorrqdrant_client.embed.modelsrr!qdrant_client.embed.schema_parserrqdrant_client.embed.utilsrqdrant_client.fastembed_commonrqdrant_client.httprqdrant_client.hybrid.fusionr qdrant_clientr$qdrant_client.common.client_warningsrr7rrrrfastembed.commonrPILr rEr;r9rPCOSINEr#rrDrOrMrJr$rsetr&r'r(r*)r! model_configs00r?rtsp EE0;;=>I?/8%>= .%9]88:  guv'='=>> DeC,@&A!AB)R(;(Q(Q(STuU7^UT "4U33G-H(H#IF/EEG    ND 1 W c#h$)[(D(Z(Z(\]uU7^U] #DeC4H.I)I$J)M(L(L(NOuU7^UO c5foo)=#>>?AH:AH_ML#' NH U ^ Ps0H0&H H" H' H, H1HH