dh}SSKJr SSKrSSKrSSKrSSKrSSKJrJrJ r J r J r J r J r JrJrJr SSKrSSKrSSKJrJr SSKJrJr SSKJr SSKJr SSKJr SS KJ r SS KJ!r! SS K$J%r% SS K&J'r' SS K(J)r) SSK*J+r+ Sr,Sr-SSjr."SS5r/SSjr0"SS5r1"SS\2\Rf5r4\4Rjr6"SS\+5r7g!\"a SS K#J!r! N|f=f)) annotationsN) AnyCallableDict GeneratorIterableListOptionalTupleTypeUnion)deletefunc)JSONUUID)ProgrammingError)Session) quoted_name)maximal_marginal_relevance)declarative_base)Document) Embeddings)get_from_dict_or_env) VectorStorei langchainc:UVVs/sHupUPM snn$s snnf)z!Return docs from docs and scores.)docs_and_scoresdoc_s `/var/www/html/shao/venv/lib/python3.13/site-packages/langchain_community/vectorstores/lantern.py_results_to_docsr",s- .oFCCo .. .s c\rSrSrSrSrg)BaseEmbeddingStore1z+Base class for the Lantern embedding store.rN)__name__ __module__ __qualname____firstlineno____doc____static_attributes__rr!r$r$1s5r,r$DistanceStrategyc^^SmU[R:Xa[RmO[Rm[ [ 5S9n"UU4SjSU[5nU$)zGet the embedding store class.N)class_registryc8>\rSrSrYr\R "\"SS9S\RS9rSS0r \R "\RSS9r \R "\ SS9r\R "\RSS9r\R "\R "Y55rSrg) +get_embedding_store..EmbeddingStoreCT)as_uuid) primary_keydefaultextend_existing)nullablerN)r&r'r(r) __tablename__ sqlalchemyColumnruuiduuid4__table_args__Stringdocumentr cmetadata custom_idARRAY embeddingr+)collection_nameembedding_typesr!EmbeddingStorer1Cs'    D$** ,T2$$Z%6%6F%%dT: %%j&7&7$G %%j&6&6~&FG r,rF)r-HAMMINGr9INTEGERREALrdictr$)distance_strategyrD DynamicBaserFrEs ` @r!get_embedding_storerM5sX N,444#++#"$&9K H H&8 H r,c.\rSrSr%SrS\S'S\S'Srg) QueryResultRzResult from a query.r$rFfloatdistancerN)r&r'r(r)r*__annotations__r+rr,r!rOrORs&&Or,rOc$\rSrSrSrSrSrSrSrg)r-Yz&Enumerator of the Distance strategies.l2sqcosinehammingrN) r&r'r(r)r* EUCLIDEANCOSINErGr+rr,r!r-r-Ys0I FGr,c\rSrSrSr\\SSSS4S1SjjrS2Sjr\ S3Sj5r \ S4Sj5r \ S5S j5r S6S jr\ S7S j5rS2S jrS2S jrS2SjrS2SjrS8SjrS9SjrS:SjrS:SjrS;Sjr\ S:Sj5r\SSS4SS?Sjjr"\ SS\\S4S@Sjj5r#SASjr$SBSCS jjr%SDS!jr&SESFS"jjr'SESGS#jjr(SESHS$jjr)SESIS%jjr*SESJS&jjr+\ S\\SS4SKS'jj5r,\ S\SS\4SLS(jj5r-\ \S\4SMS)jj5r.\ SNS*j5r/\ \\SS4SOS+jj5r0SPSQS,jjr1SPSRS-jjr2SPSSS.jjr3SPSTS/jjr4S0r5g)ULanternda`Postgres` with the `lantern` extension as a vector store. lantern uses sequential scan by default. but you can create a HNSW index using the create_hnsw_index method. - `connection_string` is a postgres connection string. - `embedding_function` any embedding function implementing `langchain.embeddings.base.Embeddings` interface. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. NFc XlX lX@lXPlX0lX`lU=(d [ R"[5Ul Xl [URU5Ul UR5 gN)connection_stringembedding_functionrDcollection_metadata_distance_strategypre_delete_collectionlogging getLoggerr&loggeroverride_relevance_score_fnrMrKrF __post_init__) selfr`rarKrDrbrdrgrelevance_score_fns r!__init__Lantern.__init__ysk"3"4.#6 "3%:"; 1 1( ; +=(1  " "O  r,cnUR5UlUR5 UR5 gr_)connect_conncreate_hnsw_extensioncreate_collectionrjs r!riLantern.__post_init__s)\\^  ""$  r,c B[UR[5(a UR$UR[RR:Xa[R$UR[R R:Xa[R $UR[R R:Xa[R $[SURSSR[Vs/sHoRPM sn5S35es snf)Nz#Got unexpected value for distance: z. Should be one of z, .) isinstancercr-rYvaluerZrG ValueErrorjoin)rjdss r!rKLantern.distance_strategys d--/? @ @** *  " "&6&@&@&F&F F#-- -  $ $(8(?(?(E(E E#** *  $ $(8(@(@(F(F F#++ +5d6M6M5NO$$(IIBR.SBRBxxBR.S$T#UUVX .Ss:DcUR$r_)rarss r! embeddingsLantern.embeddingss&&&r,c *SUSUSUSUSUSU3 $)z2Return connection string from database parameters.z postgresql+z://:@/r)clsdriverhostportdatabaseuserpasswords r! connection_string_from_db_params(Lantern.connection_string_from_db_paramss/VHCvQxj$qazRRr,cf[R"UR5nUR5nU$r_)r9 create_enginer`ro)rjengineconns r!roLantern.connects)))$*@*@A~~ r,cUR[R:XagUR[R:XagUR[R:Xagg)N l2sq_distcos_dist hamming_dist)rKr-rYrZrGrss r!distance_functionLantern.distance_functionsP  ! !%5%?%? ?  # #'7'>'> >  # #'7'?'? ?!@r,c6[UR5n[R"S5nUR U5 UR 5 SSS5 g!,(df  g=f![ a%nURRU5 SnAgSnAff=f)Nz&CREATE EXTENSION IF NOT EXISTS lantern) rrpr9textexecutecommit Exceptionrg exception)rjsession statementes r!rqLantern.create_hnsw_extensionsl %$&OO,TU  * %$$ % KK ! !! $ $ %s4A)8AA) A&"A)&A)) B3BBcFUR5 g![a gf=fr_)rrrrss r!create_tables_if_not_exists#Lantern.create_tables_if_not_existss%   " " $   s   cURRRURR5 g![ a gf=fr_)rF __table__droprprrrss r! drop_tableLantern.drop_tables=     ) ) . .tzz/@/@ A   s9< A A c$UR5 gr_)rrss r! drop_tablesLantern.drop_tabless  r,cU$r_r)rjrRs r!_hamming_relevance_score_fn#Lantern._hamming_relevance_score_fnsr,cbURb UR$UR[R:Xa UR$UR[R :Xa UR $UR[R:Xa UR$[SURS35e)a The 'correct' relevance function may differ depending on a few things, including: - the distance / similarity metric used by the VectorStore - the scale of your embeddings (OpenAI's are unit normed. Many others are not!) - embedding dimensionality - etc. z=No supported normalization function for distance_strategy of z>.Consider providing relevance_score_fn to Lantern constructor.) rhrKr-rZ_cosine_relevance_score_fnrY_euclidean_relevance_score_fnrGrryrcrss r!_select_relevance_score_fn"Lantern._select_relevance_score_fns  + + 733 3  ! !%5%<%< <22 2  # #'7'A'A A55 5  # #'7'?'? ?33 3--1-D-D,EFPP r,cUR[R:XagUR[R:XagUR[R:Xag[ SUR S35e)N dist_cos_ops dist_l2sq_opsdist_hamming_opsz5No supported operator class for distance_strategy of rvrKr-rZrYrGryrcrss r! _get_op_classLantern._get_op_classsp  ! !%5%<%< <!  # #'7'A'A A"  # #'7'?'? ?%--1-D-D,EQH r,cUR[R:XagUR[R:XagUR[R:Xag[ SUR S35e)Nz<=>z<->z<+>z/No supported operator for distance_strategy of rvrrss r! _get_operatorLantern._get_operator sp  ! !%5%<%< <  # #'7'A'A A  # #'7'?'? ?--1-D-D,EQH r,cnUR[R:Xa[[ SU55$U$)Nc[U5$r_)int)xs r!1Lantern._typed_arg_for_distance..sc!fr,)rKr-rGlistmap)rjrCs r!_typed_arg_for_distanceLantern._typed_arg_for_distances2  ! !%5%=%= =,i89 9r,c"SURS3$)N langchain__idx)rDrss r! _index_nameLantern._index_namesD00166r,@c [R"SR[URS5[UR S5UR 555n[UR5nURUUUUUS.5 UR5 SSS5 URRS5 g!,(df  N*=f)aCreate HNSW index on collection. Optional Keyword Args for HNSW Index: engine: "nmslib", "faiss", "lucene"; default: "nmslib" ef: Size of the dynamic list used during k-NN searches. Higher values lead to more accurate but slower searches; default: 64 ef_construction: Size of the dynamic list used during k-NN graph creation. Higher values lead to more accurate graph but slower indexing speed; default: 64 m: Number of bidirectional links created for each new element. Large impact on memory consumption. Between 2 and 100; default: 16 dims: Dimensions of the vectors in collection. default: 1536 zCREATE INDEX IF NOT EXISTS {} ON {} USING hnsw (embedding {}) WITH (dim = :dim, m = :m, ef_construction = :ef_construction, ef = :ef);T)dimmef_constructionefNz.HNSW extension and index created successfully.) r9rformatrrrDrrrprrrginfo)rjdimsrr ef_search_kwargscreate_index_queryrs r!create_hnsw_indexLantern.create_hnsw_index#s2(__ D,,d3D00$7""$  TZZ G OO"'6#   NN ! IJ! s 3(B?? C c [UR5nUR[R"SR [ URS5555 UR5 SSS5 g!,(df  g=f)NzDROP INDEX IF EXISTS {}T) rrprr9rrrrr)rjrs r! drop_indexLantern.drop_indexYs^ TZZ G OO-44#D$4$4d;  NN ! s AA88 BcUR(a UR5 UR5 URR 5 UR R RURR5 SSS5 g![anURS:XaSnAN(UeSnAff=f!,(df  g=f)Nf405) rddelete_collectionrrpbeginrFrcreaterrcode)rjrs r!rrLantern.create_collectiones  % %  " " $ OO  ZZ    ##--44TZZ5F5FG $ 66V#G   s6 B99B B6B1*B9/B11B66B99 CcZURRS5 UR5 g)NzTrying to delete collection)rgdebugrrss r!rLantern.delete_collectionts 78 r,c#:# [UR5v g7f)z?Create a context manager for the session, bind to _conn string.N)rrprss r! _make_sessionLantern._make_sessionxsdjj!!sc n[UR5nUbsURRS5 [ UR 5R UR RRU55nURU5 UR5 SSS5 g!,(df  g=f)zGDelete vectors by ids or uuids. Args: ids: List of ids to delete. NzUTrying to delete vectors by ids (represented by the model using the custom ids field)) rrprgrrrFwhererAin_rr)rjidskwargsrstmts r!rLantern.delete}sTZZ G !!2d11288''1155c:% NN ! s BB&& B4c  .Uc-UV s/sH!n [[R"55PM# nn U(dUV s/sHn 0PM nn URU 5n U"U UUUUS9n U R"SXXES.U D6 U R "S0U D6 U $s sn fs sn f)a Order of elements for lists `ids`, `embeddings`, `texts`, `metadatas` should match, so each row will be associated with correct values. Postgres connection string is required "Either pass it as `connection_string` parameter or set the LANTERN_CONNECTION_STRING environment variable. - `texts` texts to insert into collection. - `embeddings` an Embeddings to insert into collection - `embedding` is :class:`Embeddings` that will be used for embedding the text sent. If none is sent, then the multilingual Tensorflow Universal Sentence Encoder will be used. - `metadatas` row metadata to insert into collection. - `ids` row ids to insert into collection. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. r`rDrardrK)textsr~ metadatasrr)strr;r<_Lantern__get_connection_stringadd_embeddingsr) rrr~rCrrrDrKrdrr r`stores r!_initialize_from_embeddings#Lantern._initialize_from_embeddingssP ;.34e3tzz|$eC4%*+UUI+77?/+("7/    ) PV  )&) +5,s 'B  Bc [UR5n[XX$5H)upxpURU UUU S9n UR U 5 M+ UR 5 SSS5 g!,(df  g=fN)rCr?r@rA)rrpziprFaddr) rjrr~rrrrrmetadatarCidembedding_stores r!rLantern.add_embeddingsspTZZ G14Uz1W- "&"5"5'!& #6#  O,2X NN ! s A A)) A7c Uc-UVs/sH!n[[R"55PM# nnURR [ U55nU(dUVs/sHn0PM nn[ UR5n[XXc5H)uppURU UU U S9n URU 5 M+ UR5 SSS5 U$s snfs snf!,(df  U$=fr) rr;r<raembed_documentsrrrprrFrr) rjrrrrr r~rrrrCrrs r! add_textsLantern.add_textss ;.34e3tzz|$eC4,,<++77U7C //0  r,c\URRU5nURXBUS9nU$Nr)rar&similarity_search_with_score_by_vector)rjrrrrCrs r!similarity_search_with_score$Lantern.similarity_search_with_scores; ++77> ::V;  r,cDURXUS9nURU5$r)_Lantern__query_collectionr)rjrCrrr s r!r.Lantern.similarity_search_with_score_by_vector*s* ))I6)R//88r,c([UR5n[R"S5n[R"S5nUR U5 UR USU05 SnUGb/nUR 5HupSn [ U [5(aU [[RU 5;ayU R 5VV s0sHup,UR5U _M n nn URRU RRX5nURU5 MURRU R[U 5:HnURU5 M [R "U6nUR#U5nUR%UR['[(UR*5"URR,U5R/S55nUbUR1U5nUR3URR,R5UR755"U55R9W5R;5nSSS5 U$s sn nf!,(df  W$=f)NzSET enable_seqscan = offzSET hnsw.init_k = :krinrR)rrpr9rritemsrwrJrrlowerrFr@astextrappendand_rrgetattrrrrClabelrorder_byoprlimitall)rjrCrrrset_enable_seqscan_stmt set_init_k filter_byfilter_clauseskeyrxINvvalue_case_insensitivefilter_by_metadatarr s r!__query_collectionLantern.__query_collection4s/ TZZ G&0oo6P&Q ##)?@J OO3 4 OOJa 1I!!#"(,,.JCB!%..2SYY9N3N5:[[]25BTQAGGIqL]/2.2-@-@-J-J. &%;%?!@+'--.@A-1-@-@-J-J. &CJ./*'--.@A#1 'OO^< 44Y?IMM##d445''119% # E$ Y/''1144T5G5G5IJ9Uq M!\E2! \sB7J I< )F J<J Jc 8URXUS9n[U5$r)rr")rjrCrrrrs r!r#Lantern.similarity_search_by_vectorjs,EEVF  00r,c lUR[U55n UR"UU U4UUUUUS.UD6$)a Initialize Lantern vectorstore from list of texts. The embeddings will be generated using `embedding` class provided. Order of elements for lists `ids`, `texts`, `metadatas` should match, so each row will be associated with correct values. Postgres connection string is required "Either pass it as `connection_string` parameter or set the LANTERN_CONNECTION_STRING environment variable. - `connection_string` is fully populated connection string for postgres database - `texts` texts to insert into collection. - `embedding` is :class:`Embeddings` that will be used for embedding the text sent. If none is sent, then the multilingual Tensorflow Universal Sentence Encoder will be used. - `metadatas` row metadata to insert into collection. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. - `ids` row ids to insert into collection. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. rrrDrdrK)rrr) rrrCrrDrKrrdrr~s r! from_textsLantern.from_textsvsUT..tE{; ..      +"7/    r,c UV s/sHoSPM n n UV s/sHoSPM n n UR"U U U4UUUUUS.UD6$s sn fs sn f)aConstruct Lantern wrapper from raw documents and pre- generated embeddings. Postgres connection string is required "Either pass it as `connection_string` parameter or set the LANTERN_CONNECTION_STRING environment variable. Order of elements for lists `ids`, `text_embeddings`, `metadatas` should match, so each row will be associated with correct values. - `connection_string` is fully populated connection string for postgres database - `text_embeddings` is array with tuples (text, embedding) to insert into collection. - `embedding` is :class:`Embeddings` that will be used for embedding the text sent. If none is sent, then the multilingual Tensorflow Universal Sentence Encoder will be used. - `metadatas` row metadata to insert into collection. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `ids` row ids to insert into collection. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. rr=)r) rtext_embeddingsrCrrDrrdrKrtrr~s r!from_embeddingsLantern.from_embeddingssuT //!1/$34OqdO 4..      +"7/    04s A Ac <URU5nU"UUUUUS9nU$)a Get instance of an existing Lantern store.This method will return the instance of the store without inserting any new embeddings Postgres connection string is required "Either pass it as `connection_string` parameter or set the LANTERN_CONNECTION_STRING environment variable. - `connection_string` is a postgres connection string. - `embedding` is :class:`Embeddings` that will be used for embedding the text sent. If none is sent, then the multilingual Tensorflow Universal Sentence Encoder will be used. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `ids` row ids to insert into collection. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. r)r)rrCrDrdrKrr`rs r!from_existing_indexLantern.from_existing_indexs6F 77?/+("7/   r,c@[USSS9nU(d [S5eU$)Nr`LANTERN_CONNECTION_STRING)datar3env_keyzPostgres connection string is requiredEither pass it as `connection_string` parameteror set the LANTERN_CONNECTION_STRING variable.)rry)rrr`s r!__get_connection_stringLantern.__get_connection_strings6!5#/"  !A  ! r,c UVs/sHoRPM n nUVs/sHoRPM n nURU5n XS'UR"SU UUU UUUS.UD6$s snfs snf)a Initialize a vector store with a set of documents. Postgres connection string is required "Either pass it as `connection_string` parameter or set the LANTERN_CONNECTION_STRING environment variable. - `connection_string` is a postgres connection string. - `documents` is list of :class:`Document` to initialize the vector store with - `embedding` is :class:`Embeddings` that will be used for embedding the text sent. If none is sent, then the multilingual Tensorflow Universal Sentence Encoder will be used. - `collection_name` is the name of the collection to use. (default: langchain) - NOTE: This is the name of the table in which embedding data will be stored The table will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables. - `distance_strategy` is the distance strategy to use. (default: EUCLIDEAN) - `EUCLIDEAN` is the euclidean distance. - `COSINE` is the cosine distance. - `HAMMING` is the hamming distance. - `ids` row ids to insert into collection. - `pre_delete_collection` if True, will delete the collection if it exists. (default: False) - Useful for testing. r`)rrdrCrrrDrKr)r rrr>) r documentsrCrDrKrrdrdrrr`s r!from_documentsLantern.from_documents'sH*33A3)23AZZ 377?&7"#~~  "7+/    43s A$A)c HURXUS9nUVs/sHoRRPM n n[[R "U[R S9U UUS9n URU5n [U 5V V s/sHupX;dM U PM sn n $s snfs sn n f)aZReturn docs selected using the maximal marginal relevance with score to embedding vector. Maximal marginal relevance optimizes for similarity to query AND diversity among selected documents. Args: embedding: Embedding to look up documents similar to. k (int): Number of Documents to return. Defaults to 4. fetch_k (int): Number of Documents to fetch to pass to MMR algorithm. Defaults to 20. lambda_mult (float): Number between 0 and 1 that determines the degree of diversity among the results with 0 corresponding to maximum diversity and 1 to minimum diversity. Defaults to 0.5. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: List[Tuple[Document, float]]: List of Documents selected by maximal marginal relevance to the query and score for each. r)dtype)r lambda_mult) r rFrCrnparrayfloat32r enumerate)rjrCrfetch_krVrrr r embedding_list mmr_selected candidatesirs r!2max_marginal_relevance_search_with_score_by_vector:Lantern.max_marginal_relevance_search_with_score_by_vector\s<))IQW)XHOPf//99P1 HHYbjj 1 #  55g> ' 3I3daq7H3IIQJsB BBc hURRU5nUR"U4UUUUS.UD6$)aReturn docs selected using the maximal marginal relevance. Maximal marginal relevance optimizes for similarity to query AND diversity among selected documents. Args: query (str): Text to look up documents similar to. k (int): Number of Documents to return. Defaults to 4. fetch_k (int): Number of Documents to fetch to pass to MMR algorithm. Defaults to 20. lambda_mult (float): Number between 0 and 1 that determines the degree of diversity among the results with 0 corresponding to maximum diversity and 1 to minimum diversity. Defaults to 0.5. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: List[Document]: List of Documents selected by maximal marginal relevance. rr[rVr)rar'max_marginal_relevance_search_by_vector)rjrrr[rVrrrCs r!max_marginal_relevance_search%Lantern.max_marginal_relevance_searchsK8++77> ;;  #     r,c lURRU5nUR"SUUUUUS.UD6nU$)a?Return docs selected using the maximal marginal relevance with score. Maximal marginal relevance optimizes for similarity to query AND diversity among selected documents. Args: query (str): Text to look up documents similar to. k (int): Number of Documents to return. Defaults to 4. fetch_k (int): Number of Documents to fetch to pass to MMR algorithm. Defaults to 20. lambda_mult (float): Number between 0 and 1 that determines the degree of diversity among the results with 0 corresponding to maximum diversity and 1 to minimum diversity. Defaults to 0.5. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: List[Tuple[Document, float]]: List of Documents selected by maximal marginal relevance to the query and score for each. )rCrr[rVrr)rarra) rjrrr[rVrrrCrs r!(max_marginal_relevance_search_with_score0Lantern.max_marginal_relevance_search_with_scoresO:++77> FF #     r,c HUR"U4UUUUS.UD6n[U5$)aReturn docs selected using the maximal marginal relevance to embedding vector. Maximal marginal relevance optimizes for similarity to query AND diversity among selected documents. Args: embedding (str): Text to look up documents similar to. k (int): Number of Documents to return. Defaults to 4. fetch_k (int): Number of Documents to fetch to pass to MMR algorithm. Defaults to 20. lambda_mult (float): Number between 0 and 1 that determines the degree of diversity among the results with 0 corresponding to maximum diversity and 1 to minimum diversity. Defaults to 0.5. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: List[Document]: List of Documents selected by maximal marginal relevance. rd)rar")rjrCrr[rVrrrs r!re/Lantern.max_marginal_relevance_search_by_vectors?:QQ  #     00r,) rFrprcrbrDr`rargrhrd)r`rrarrKr-rDrrbOptional[dict]rdboolrgzOptional[logging.Logger]rkz"Optional[Callable[[float], float]]returnNone)rorp)ror-)ror)rrrrrrrrrrrrror)rozsqlalchemy.engine.Connection)ror)rRrQrorQ)rozCallable[[float], float])ror)rCList[Union[float, int]]rorq) rrrrrrrrrrrorp)rozGenerator[Session, None, None]r_)rOptional[List[str]]rrrorp)r List[str]r~List[List[float]]rCrrOptional[List[dict]]rrrrDrrKr-rdrnrrror\) rrsr~rtrz List[dict]rrsrrrorp)NN) rz Iterable[str]rrurrrrrrors)r rroList[Tuple[Document, float]])N) rrrrrrmrrroList[Document])rrrrrrmrorv)rC List[float]rrrrmrorv)rCryrrrrmroz List[Any]) rCryrrrrmrrrorx)r Type[Lantern]rrsrCrrrurDrrKr-rrrrdrnrrror\)rBzList[Tuple[str, List[float]]]rCrrrurDrrrrrdrnrKr-rrror\)rrzrCrrDrrdrnrKr-rrror\)rzDict[str, Any]ror)rrzrPrxrCrrDrrKr-rrrrdrnrrror\)rwg?N)rCryrrr[rrVrQrOptional[Dict[str, str]]rrrorv)rrrrr[rrVrQrr|rrrorx)rrrrr[rrVrQrrmrrrorv)rCryrrr[rrVrQrr|rrrorx)6r&r'r(r)r*DEFAULT_DISTANCE_STRATEGY"_LANGCHAIN_DEFAULT_COLLECTION_NAMErlripropertyrKr~ classmethodrrorrqrrrrrrrrrADA_TOKEN_COUNTrrrrr contextlibcontextmanagerrrrrrrrrrr rr>rDrGrrRrarfrirer+rr,r!r\r\ds0/HA.2&++/AE',    ,  $)? 0! ! '' S S S S  S  S S  S S ""%  6  0 77 $! 4K4K 4K 4K  4K  4K 4Kl  "" $(   , +/#'A.G&+==&= = ( = ! ==,= $== ==~&      ,+/#' (!     8 "!%             "!%      & !% 99 9 9 & 9!% 44 4 4  4r!% 1 1  1 1  1  1 +/A.G#'&+5 5 5 5 ( 5  5 , 5 !5  $5 5  5 5 n +/A#'&+.G6 66 6 ( 6  6 ! 6  $6 ,6 6  6 6 p B&+.G , ,,, $ , , ,  , ,,\!!  B.G#'&+2 2 !2 2  2 , 2 ! 2  $2 2  2 2 n +/ *J*J *J *J  *J ) *J*J &*J^ +/ $ $  $  $  $ ) $ $  $ R !% && & &  &  && &&V +/ &1&1 &1 &1  &1 ) &1&1 &1&1r,r\)rrrorx)rKr-rDrror)8 __future__rrenumrer;typingrrrrrr r r r r numpyrWr9rrsqlalchemy.dialects.postgresqlrrsqlalchemy.excrsqlalchemy.ormrsqlalchemy.sqlr&langchain_community.vectorstores.utilsrr ImportErrorsqlalchemy.ext.declarativelangchain_core.documentsrlangchain_core.embeddingsrlangchain_core.utilsrlangchain_core.vectorstoresrrr~r"r$rMrOrEnumr-rZr}r\rr,r!rs"     #5+"&M</.053%0"/ 66':=:sDII-33X1kX1I<;