o Zhm0@sddlmZddlZddlmZddlmZmZmZm Z m Z m Z m Z m Z mZddlmZddlmZer@ddlmZddlmZGd d d eeZe d eeefd ZeeZd ZGdddeZdS)) annotationsN)Enum) TYPE_CHECKINGAnyDict GeneratorIterableListOptionalTypeVarUnion)Document) VectorStore) Embeddings) Collectionc@s eZdZdZdZ dZ dZdS)DocumentDBSimilarityTypez)DocumentDB Similarity Type as enumerator.ZcosineZ dotProductZ euclideanN)__name__ __module__ __qualname____doc__COSDOTZEUCrrb/var/www/html/lang_env/lib/python3.10/site-packages/langchain_community/vectorstores/documentdb.pyrsrDocumentDBDocumentType)boundc@seZdZdZdddddUddZedVddZdWddZedXddZ dYddZ dZdd Z d!e j d"d#fd[d+d,Z -d\d]d3d4Zd^d7d8Ze - -d_d`d;d<Zd\dad@dAZd\dbdDdEZ F G -dcdddOdPZ F Gded-dQdfdSdTZd-S)gDocumentDBVectorSearcha`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) ZvectorSearchIndexZ textContentZ vectorContent) index_nametext_key embedding_key collection"Collection[DocumentDBDocumentType] embeddingrrstrrr cCs*||_||_||_||_||_tj|_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#rrr rrr__init__Bs  zDocumentDBVectorSearch.__init__returncC|jSN)r&r+rrr embeddings]sz!DocumentDBVectorSearch.embeddingscCr.)zUReturns the index name Returns: Returns the index name )r'r0rrrget_index_nameasz%DocumentDBVectorSearch.get_index_nameconnection_string namespacekwargsrc Ks\zddlm}Wn tytdw||}|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`..)Zpymongor6 ImportErrorsplit) clsr3r4r#r5r6clientZdb_nameZcollection_namer!rrrfrom_connection_stringjs  z-DocumentDBVectorSearch.from_connection_stringboolcCs6|j}|j}|D]}|d}||krdSq 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)r%Z list_indexesr'pop)r+cursorrresZcurrent_index_namerrr index_existss  z#DocumentDBVectorSearch.index_existsNonecCs|r |j|jdSdS)zEDeletes the index specified during instance construction if it existsN)rBr%Z drop_indexr'r0rrr delete_indexsz#DocumentDBVectorSearch.delete_indexi@ dimensionsint similarityrmef_constructiondict[str, Any]c CsH||_|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 vectorZhnsw)typerIrGrJZefConstruction)r>keyZ vectorOptions)Z createIndexesZindexes)r*r%r>r'r)Zdatabasecommand)r+rGrIrJrKZcreate_index_commandsZcurrent_databaseZcreate_index_responsesrrr create_indexs$ z#DocumentDBVectorSearch.create_indexNtexts Iterable[str] metadatasOptional[List[Dict[str, Any]]]r c Ks|dt}|pdd|D}g}g}g}tt||D]%\} \} } || || | d|dkrA||||g}g}q|rM|||||S)N batch_sizecss|]}iVqdSr/r).0_rrr sz3DocumentDBVectorSearch.add_texts..r)getDEFAULT_INSERT_BATCH_SIZE enumeratezipappendextend _insert_texts) r+rRrTr5rVZ _metadatasZ texts_batchZmetadatas_batchZ result_idsitextmetadatarrr add_textss   z DocumentDBVectorSearch.add_texts List[str]List[Dict[str, Any]]cs@|sgSj|}fddt|||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: cs&g|]\}}}j|j|i|qSr)r(r))rWtrJr#r0rr sz8DocumentDBVectorSearch._insert_texts..)r&Zembed_documentsr^r%Z insert_manyZ inserted_ids)r+rRrTr1Z to_insertZ insert_resultrr0rras     z$DocumentDBVectorSearch._insert_textsOptional[List[dict]],Optional[Collection[DocumentDBDocumentType]]cKs4|durtd|||fi|}|j||d|S)Nz*Must provide 'collection' named parameter.)rT) ValueErrorre)r:rRr#rTr!r5Z vectorstorerrr from_textss  z!DocumentDBVectorSearch.from_textsidsOptional[List[str]]Optional[bool]cKs(|durtd|D]}||q dS)Nz#No document ids provided to delete.T)rldelete_document_by_id)r+rnr5 document_idrrrdeletes  zDocumentDBVectorSearch.deleterr Optional[str]c Cs\zddlm}Wnty}ztd|d}~ww|dur"td|jd||idS)zjRemoves a Specific Document by Id Args: document_id: The document identifier r)ObjectIdz>Unable to import bson, please install with `pip install bson`.Nz"No document id provided to delete.Z_id)Z bson.objectidrur8rlr%Z delete_one)r+rrruerrrrq&sz,DocumentDBVectorSearch.delete_document_by_id(r1 List[float]k ef_searchfilterOptional[Dict[str, Any]]List[Document]c Csh|si}d|idd||j|j||diig}|j|}g}|D]}||j} |t| |dq |S)aReturns a list of documents. Args: embeddings: The query vector k: the number of documents to return ef_search: Specifies the size of the dynamic candidate list that HNSW index uses during search. A higher value of efSearch provides better recall at cost of speed. filter (Optional[Dict[str, str]]): Filter by metadata. Defaults to None. Returns: A list of documents closest to the query vector z$matchz$searchZ vectorSearch)rMpathrIrzZefSearch)Z page_contentrd)r)r*r%Z aggregater?r(r_r ) r+r1rzr{r|Zpipeliner@docsrArcrrr _similarity_search_without_score7s(  z7DocumentDBVectorSearch._similarity_search_without_score)r|querycKs,|j|}|j||||d}dd|DS)N)r1rzr{r|cSsg|]}|qSrr)rWdocrrrritsz.)r&Z embed_queryr)r+rrzr{r|r5r1rrrrsimilarity_searchgs z(DocumentDBVectorSearch.similarity_search) r!r"r#rrr$rr$r r$)r-r)r-r$) r3r$r4r$r#rr5rr-r)r-r=)r-rC) rGrHrIrrJrHrKrHr-rLr/)rRrSrTrUr5rr-r )rRrfrTrgr-r )NN) rRrfr#rrTrjr!rkr5rr-r)rnror5rr-rp)rrrtr-rC)rwrxN) r1ryrzrHr{rHr|r}r-r~)rwrx) rr$rzrHr{rHr|r}r5rr-r~)rrrrr,propertyr1r2 classmethodr<rBrDrrrQrerarmrsrqrrrrrrr,sH       A    3r) __future__rloggingenumrtypingrrrrrr r r r Zlangchain_core.documentsr Zlangchain_core.vectorstoresrZlangchain_core.embeddingsrZpymongo.collectionrr$rr getLoggerrloggerr\rrrrrs  ,