LlamaIndex Chat Store Integration: MongoDB Chat Store

Installation

pip install llama-index-storage-chat-store-mongodb

Usage

Using MongoChatStore from llama_index.storage.chat_store.mongo you can store chat history in MongoDB.

from llama_index.storage.chat_store.mongo import MongoChatStore

# Initialize the MongoDB chat store with URI and database name and collection name
chat_store = MongoChatStore(
    mongodb_uri="mongodb://localhost:27017/",
    db_name="llama_index",
    collection_name="chat_sessions",
)

You can also initialize the chat store with a MongoClient or AsyncMongoClient and a database name and collection name.

from pymongo import MongoClient, AsyncMongoClient

client = MongoClient("mongodb://localhost:27017/")
async_client = AsyncMongoClient("mongodb://localhost:27017/")

chat_store = MongoChatStore(
    client=client,
    amongo_client=async_client,
    db_name="llama_index",
    collection_name="chat_sessions",
)

You can also initialize the chat store with a Collection or AsyncCollection.

from pymongo import Collection
from pymongo.asynchronous import AsyncCollection

client = MongoClient("mongodb://localhost:27017/")
async_client = AsyncMongoClient("mongodb://localhost:27017/")

collection = client["llama_index"]["chat_sessions"]
async_collection = async_client["llama_index"]["chat_sessions"]

chat_store = MongoChatStore(
    collection=collection, async_collection=async_collection
)

Usage with LlamaIndex

from llama_index.core.chat_engine.types import ChatMessage

chat_memory = ChatMemoryBuffer.from_defaults(
    token_limit=3000,
    chat_store=chat_store,
    chat_store_key="user1",
)