Ngm0ddlmZddlZddlmZddlmZmZmZm Z m Z m Z m Z m Z mZddlmZddlmZer ddlmZddlmZGd d eeZe d eeef ZejeZd ZGddeZdS)) annotationsN)Enum) TYPE_CHECKINGAnyDict GeneratorIterableListOptionalTypeVarUnion)Document) VectorStore) Embeddings) Collectionc"eZdZdZdZ dZ dZdS)DocumentDBSimilarityTypez)DocumentDB Similarity Type as enumerator.cosine dotProduct euclideanN)__name__ __module__ __qualname____doc__COSDOTEUCg/var/www/html/ai-engine/env/lib/python3.11/site-packages/langchain_community/vectorstores/documentdb.pyrrs+33 C C CrrDocumentDBDocumentType)boundceZdZdZdddddGdZedHdZdIdZedJdZ dKdZ dLdZ de j ddfdMd%Z dNdOd,ZdPd/Ze dQdRd2ZdNdSd6ZdNdTd9Z dUdVdCZ dWd&dDdXdFZd&S)YDocumentDBVectorSearcha`Amazon DocumentDB (with MongoDB compatibility)` vector store. Please refer to the official Vector Search documentation for more details: https://docs.aws.amazon.com/documentdb/latest/developerguide/vector-search.html To use, you should have both: - the ``pymongo`` python package installed - a connection string and credentials associated with a DocumentDB cluster Example: . code-block:: python from langchain_community.vectorstores import DocumentDBVectorSearch from langchain_community.embeddings.openai import OpenAIEmbeddings from pymongo import MongoClient mongo_client = MongoClient("") collection = mongo_client[""][""] embeddings = OpenAIEmbeddings() vectorstore = DocumentDBVectorSearch(collection, embeddings) vectorSearchIndex textContent vectorContent) index_nametext_key embedding_key collection"Collection[DocumentDBDocumentType] embeddingrr)strr*r+cn||_||_||_||_||_t j|_dS)aConstructor for DocumentDBVectorSearch Args: collection: MongoDB collection to add the texts to. embedding: Text embedding model to use. index_name: Name of the Vector Search index. text_key: MongoDB field that will contain the text for each document. embedding_key: MongoDB field that will contain the embedding for each document. N) _collection _embedding _index_name _text_key_embedding_keyrr_similarity_type)selfr,r.r)r*r+s r __init__zDocumentDBVectorSearch.__init__Bs;(&#%!+ 8 <rreturnc|jSN)r2r7s r embeddingsz!DocumentDBVectorSearch.embeddings]s rc|jS)zUReturns the index name Returns: Returns the index name )r3r<s r get_index_namez%DocumentDBVectorSearch.get_index_nameas rconnection_string namespacekwargsrc  ddlm}n#t$rtdwxYw||}|d\}}|||} || |fi|S)aCreates an Instance of DocumentDBVectorSearch from a Connection String Args: connection_string: The DocumentDB cluster endpoint connection string namespace: The namespace (database.collection) embedding: The embedding utility **kwargs: Dynamic keyword arguments Returns: an instance of the vector store r) MongoClientzGCould not import pymongo, please install it with `pip install pymongo`..)pymongorD ImportErrorsplit) clsr@rAr.rBrDclientdb_namecollection_namer,s r from_connection_stringz-DocumentDBVectorSearch.from_connection_stringjs(  + + + + + + +   )   *k*;<<#,??3#7#7 G__5 s:y33F333s #boolc|j}|j}|D] }|d}||krdS!dS)zVerifies if the specified index name during instance construction exists on the collection Returns: Returns True on success and False if no such index exists on the collection nameTF)r1 list_indexesr3pop)r7cursorr)rescurrent_index_names r index_existsz#DocumentDBVectorSearch.index_existss]!..00%   C!$ !Z//tt0urNonecp|r!|j|jdSdS)zEDeletes the index specified during instance construction if it existsN)rVr1 drop_indexr3r<s r delete_indexz#DocumentDBVectorSearch.delete_indexs@      :   ' '(8 9 9 9 9 9 : :ri@ dimensionsint similarityrmef_constructiondict[str, Any]c ||_|jj|j|jdid||||ddgd}|jj}||}|S)aCreates an index using the index name specified at instance construction Args: dimensions: Number of dimensions for vector similarity. The maximum number of supported dimensions is 2000 similarity: Similarity algorithm to use with the HNSW index. Possible options are: - DocumentDBSimilarityType.COS (cosine distance), - DocumentDBSimilarityType.EUC (Euclidean distance), and - DocumentDBSimilarityType.DOT (dot product). m: Specifies the max number of connections for an HNSW index. Large impact on memory consumption. ef_construction: Specifies the size of the dynamic candidate list for constructing the graph for HNSW index. Higher values lead to more accurate results but slower indexing speed. Returns: An object describing the created index vectorhnsw)typer_r]r`efConstruction)rPkey vectorOptions) createIndexesindexes)r6r1rPr3r5databasecommand)r7r]r_r`racreate_index_commandscurrent_databasecreate_index_responsess r create_indexz#DocumentDBVectorSearch.create_indexs@!+"-2!, /: &&0&0*9 &&   ! ! $ +42B1I1I !2 2 &%rNtexts Iterable[str] metadatasOptional[List[Dict[str, Any]]]r c |dt}|p d|D}g}g}g}tt||D]k\} \} } || || | dz|zdkr-||||g}g}l|r)|||||S)N batch_sizec3K|]}iVdSr;r).0_s r z3DocumentDBVectorSearch.add_texts..s":M:M!2:M:M:M:M:M:Mrr)getDEFAULT_INSERT_BATCH_SIZE enumeratezipappendextend _insert_texts) r7rrrtrBrw _metadatas texts_batchmetadatas_batch result_idsitextmetadatas r add_textsz DocumentDBVectorSearch.add_textss  ZZ .GHH -6-M:M:Mu:M:M:M   #,S -C-C#D#D % % Ah   t $ $ $  " "8 , , ,A#q((!!$"4"4[/"R"RSSS "$  P   d00oNN O O Or List[str]List[Dict[str, Any]]c|sgSj|}fdt|||D}j|}|jS)zUsed to Load Documents into the collection Args: texts: The list of documents strings to load metadatas: The list of metadata objects associated with each document Returns: c<g|]\}}}j|j|i|Sr)r4r5)rytr`r.r7s r z8DocumentDBVectorSearch._insert_texts..sB   1i^Q 3Y D! D   r)r2embed_documentsrr1 insert_many inserted_ids)r7rrrtr= to_insert insert_results` r rz$DocumentDBVectorSearch._insert_textss~ I_44U;;     #&ui#D#D   (44Y?? ))rOptional[List[dict]],Optional[Collection[DocumentDBDocumentType]]c j|td|||fi|}||||S)Nz*Must provide 'collection' named parameter.)rt) ValueErrorr)rIrrr.rtr,rB vectorstores r from_textsz!DocumentDBVectorSearch.from_textssQ  IJJ Jc*i::6:: ey999ridsOptional[List[str]]Optional[bool]c \|td|D]}||dS)Nz#No document ids provided to delete.T)rdelete_document_by_id)r7rrB document_ids r deletezDocumentDBVectorSearch.deletesA ;BCC C 4 4K  & &{ 3 3 3 3trr Optional[str]c ddlm}n"#t$r}td|d}~wwxYw|td|jd||idS)zjRemoves a Specific Document by Id Args: document_id: The document identifier r)ObjectIdz>Unable to import bson, please install with `pip install bson`.Nz"No document id provided to delete._id) bson.objectidrrGrr1 delete_one)r7rres r rz,DocumentDBVectorSearch.delete_document_by_id&s   . . . . . . .   P    ABB B ##UHH[,A,A$BCCCCCs (#((r= List[float]k ef_searchfilterOptional[Dict[str, Any]]List[Document]c |si}d|idd||j|j||diig}|j|}g}|D]@}||j} |t| |A|S)aReturns a list of documents. Args: embeddings: The query vector k: the number of documents to return ef_search: Specifies the size of the dynamic candidate list that HNSW index uses during search. A higher value of efSearch provides better recall at cost of speed. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: A list of documents closest to the query vector z$matchz$search vectorSearch)rdpathr_refSearch) page_contentr)r5r6r1 aggregaterRr4rr) r7r=rrrpipelinerSdocsrTrs r _similarity_search_without_scorez7DocumentDBVectorSearch._similarity_search_without_score7s* F v "", $ 3&*&;$- %%  * !++H55 C CC774>**D KKdSAAA B B B B r)rqueryc |j|}|||||}d|DS)N)r=rrrcg|]}|Srr)rydocs r rz.ts$$$$$$r)r2 embed_queryr)r7rrrrrBr=rs r similarity_searchz(DocumentDBVectorSearch.similarity_searchgsR_0077 44!Q)F5  %$t$$$$r) r,r-r.rr)r/r*r/r+r/)r9r)r9r/) r@r/rAr/r.rrBrr9r%)r9rN)r9rW) r]r^r_rr`r^rar^r9rbr;)rrrsrtrurBrr9r )rrrrtrr9r )NN) rrrr.rrtrr,rrBrr9r%)rrrBrr9r)rrr9rW)rrN) r=rrr^rr^rrr9r)rr) rr/rr^rr^rrrBrr9r)rrrrr8propertyr=r? classmethodrMrVrZrrrqrrrrrrrrrr r%r%,s4.%,======6X    444[4>$::::/G/K! <&<&<&<&<&B59,****0 +/CG    [ DDDDD(+/ .....f % ,0 % % % % % % % %rr%) __future__rloggingenumrtypingrrrrr r r r r langchain_core.documentsrlangchain_core.vectorstoresrlangchain_core.embeddingsrpymongo.collectionrr/rr! getLoggerrloggerr~r%rrr rs""""""                      .-----333333.444444------sD!!9c3hPPP  8 $ $H%H%H%H%H%[H%H%H%H%H%r