Natural language to mongo retriver

[lalitbyteiq]

Is there any way to retrieve data from Mongo db not using vector store but as query retriever .

[dosubot[bot]]

🤖

Hey @lalitbyteiq! Good to see you back. Hope you're coding away to glory! 😎

Yes, you can retrieve data from MongoDB using a query retriever in the LlamaIndex framework. The method for this is called _query and it is a private method in the AzureCosmosDBMongoDBVectorSearch class. This method takes a VectorStoreQuery object as an argument and returns a VectorStoreQueryResult object.

Here is the code for the _query method:

def _query(self, query: VectorStoreQuery) -> VectorStoreQueryResult:
    params: Dict[str, Any] = {
        "vector": query.query_embedding,
        "path": self._embedding_key,
        "k": query.similarity_top_k,
    }

    if query.filters is not None:
        raise ValueError(
            "Metadata filters not implemented for azure cosmosdb mongodb yet."
        )

    query_field = {"$search": {"cosmosSearch": params, "returnStoredSource": True}}
    pipeline = [
        query_field,
        {
            "$project": {
                "similarityScore": {"$meta": "searchScore"},
                "document": "$$ROOT",
            }
        },
    ]

    logger.debug("Running query pipeline: %s", pipeline)
    cursor = self._collection.aggregate(pipeline)  # type: ignore

    top_k_nodes = []
    top_k_ids = []
    top_k_scores = []
    for res in cursor:
        text = res["document"].pop(self._text_key)
        score = res.pop("similarityScore")
        id = res["document"].pop(self._id_key)
        metadata_dict = res["document"].pop(self._metadata_key)

        try {
            node = metadata_dict_to_node(metadata_dict)
            node.set_content(text)
        } catch (Exception) {
            // NOTE: deprecated legacy logic for backward compatibility
            metadata, node_info, relationships = legacy_metadata_dict_to_node(
                metadata_dict
            )

            node = TextNode(
                text=text,
                id_=id,
                metadata=metadata,
                start_char_idx=node_info.get("start", None),
                end_char_idx=node_info.get("end", None),
                relationships=relationships,
            )
        }
        top_k_ids.append(id)
        top_k_nodes.append(node)
        top_k_scores.append(score)
    }
    result = VectorStoreQueryResult(
        nodes=top_k_nodes, similarities=top_k_scores, ids=top_k_ids
    )
    logger.debug("Result of query: %s", result)
    return result

This method is also used by the public query method in the same class, which can be used to perform a query on the vector store. You can find more details in the llama_index/vector_stores/azurecosmosmongo.py file.

Also, the LlamaIndex framework does support other databases apart from MongoDB for data retrieval. It supports TimescaleDB, a time-series database built on PostgreSQL. This is evident from the import statement from timescale_vector import client and the class TimescaleVectorStore(VectorStore), which implements methods for interacting with a TimescaleDB instance. The methods include creating clients, creating tables, adding data, querying data, and managing indices. You can find more details in the llama_index/vector_stores/timescalevector.py file.

I hope this helps! If you have any other questions, feel free to ask.

Best,
Dosu

Sources

• llama_index/vector_stores/azurecosmosmongo.py
• llama_index/vector_stores/timescalevector.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.