autogen.agentchat.contrib.rag.MongoDBQueryEngine

MongoDBQueryEngine

MongoDBQueryEngine(
    connection_string: str,
    llm: ForwardRef('LLM') | None = None,
    database_name: str | None = None,
    embedding_function: ForwardRef('BaseEmbedding') | Callable[..., Any] | None = None,
    embedding_model: ForwardRef('BaseEmbedding') | str | None = None,
    collection_name: str | None = None
)

A query engine backed by MongoDB Atlas that supports document insertion and querying.
This engine initializes a vector database, builds an index from input documents, and allows querying using the chat engine interface.
Attributes:
vector_db (MongoDBAtlasVectorDB): The MongoDB vector database instance.
vector_search_engine (MongoDBAtlasVectorSearch): The vector search engine.
storage_context (StorageContext): The storage context for the vector store.
index (Optional[VectorStoreIndex]): The index built from the documents.
Initializes a MongoDBQueryEngine instance.

Parameters:

Name	Description
connection_string	Type: str
llm	Type: ForwardRef(‘LLM’) | None Default: None
database_name	Type: str | None Default: None
embedding_function	Type: ForwardRef(‘BaseEmbedding’) | Callable[…, Any] | None Default: None
embedding_model	Type: ForwardRef(‘BaseEmbedding’) | str | None Default: None
collection_name	Type: str | None Default: None

Instance Methods

add_docs

add_docs(
    self,
    new_doc_dir: Path | str | None = None,
    new_doc_paths_or_urls: Sequence[Path | str] | None = None,
    *args: Any,
    **kwargs: Any
) -> None

Adds new documents to the existing vector store index.
This method validates that the index exists, loads documents from the specified directory or file paths, and inserts them into the vector store index.

Parameters:

Name	Description
new_doc_dir	Directory containing new documents. Type: pathlib.Path | str | None Default: None
new_doc_paths_or_urls	List of file paths or URLs for new documents. Type: Sequence[pathlib.Path | str] | None Default: None
*args	Additional positional arguments. Type: Any
**kwargs	Additional keyword arguments. Type: Any

connect_db

connect_db(
    self,
    *args: Any,
    **kwargs: Any
) -> bool

Connects to the MongoDB database and initializes the query index from the existing collection.
This method verifies the existence of the collection, sets up the database connection, builds the vector store index, and pings the MongoDB server.
Returns:
bool: True if connection is successful; False otherwise.

Parameters:

Name	Description
*args	Type: Any
**kwargs	Type: Any

Returns:

Type	Description
bool	bool: True if connection is successful; False otherwise.

get_collection_name

get_collection_name(self) -> str

Retrieves the name of the MongoDB collection.
Returns:
str: The collection name.
Raises:
ValueError: If the collection name is not set.

Returns:

Type	Description
str	str: The collection name.

init_db

init_db(
    self,
    new_doc_dir: Path | str | None = None,
    new_doc_paths_or_urls: Sequence[Path | str] | None = None,
    *args: Any,
    **kwargs: Any
) -> bool

Initializes the MongoDB database by creating or overwriting the collection and indexing documents.
This method loads documents from a directory or provided file paths, sets up the database (optionally overwriting any existing collection), builds the vector store index, and inserts the documents.

Parameters:

Name	Description
new_doc_dir	Directory containing documents to be indexed. Type: pathlib.Path | str | None Default: None
new_doc_paths_or_urls	List of file paths or URLs for documents. Type: Sequence[pathlib.Path | str] | None Default: None
*args	Additional positional arguments. Type: Any
**kwargs	Additional keyword arguments. Type: Any

Returns:

Type	Description
bool	bool: True if the database is successfully initialized; False otherwise.

query

query(
    self,
    question: str,
    *args: Any,
    **kwargs: Any
) -> Any

Queries the indexed documents using the provided question.
This method validates that the query index is initialized, creates a query engine from the vector store index, and executes the query. If the response is empty, a default reply is returned.

Parameters:

Name	Description
question	The query question. Type: str
*args	Type: Any
**kwargs	Type: Any

Returns:

Type	Description
Any	Any: The query response as a string, or a default reply if no results are found.