[Question]: AzureChatStore: Ensuring Asynchronous Requests and Understanding _atable_service_client Usage

[gazambuja]

Question Validation

• [X] I have searched both the documentation and discord for an answer.

Question

I'm working on integrating AzureChatStore into my project and want to ensure that all requests to the store are made asynchronously to optimize performance. I've been reviewing the source code, but I'm having trouble understanding the following:

• Asynchronous Operations: How does AzureChatStore explicitly make asynchronous requests? Are there specific configurations or methods I need to be aware of to guarantee asynchronicity?
• _atable_service_client vs. _table_service_client: In the code, I see references to both _atable_service_client and _table_service_client. What's the difference between these two clients, and under what circumstances does AzureChatStore use one over the other?

Any insights or guidance on these questions would be greatly appreciated. Thanks in advance for your help!

chat_store = AzureChatStore.from_connection_string(
    connection_string=self.azure_connstr,
    chat_table_name="memory",
)

memory = ChatMemoryBuffer.from_defaults(
    token_limit=3900,
    chat_store=chat_store,
    chat_store_key=user_key
)

[dosubot[bot]]

Hey @gazambuja! I'm here to help you with any questions or issues you have regarding the AzureChatStore. Feel free to ask me anything!

AzureChatStore explicitly makes asynchronous requests using the _atable_service_client, which is an instance of AsyncTableServiceClient from the azure.data.tables.aio module. This client is designed to handle asynchronous operations.

The difference between _atable_service_client and _table_service_client is as follows:

• _table_service_client is an instance of TableServiceClient from the azure.data.tables module, which is used for synchronous operations.
• _atable_service_client is an instance of AsyncTableServiceClient from the azure.data.tables.aio module, which is used for asynchronous operations.

This distinction allows AzureChatStore to support both synchronous and asynchronous interactions with Azure Table Storage or Cosmos DB. The AzureKVStore class, for example, initializes these clients and provides methods like put and aput for synchronous and asynchronous operations, respectively ^[1][2].

For ensuring asynchronous operations, you should use methods provided by AsyncTableServiceClient, such as create_table_if_not_exists, upsert_entity, submit_transaction, get_entity, list_entities, delete_entity, and query_entities ^[1][2].

To continue talking to Dosu, mention @dosu.

[git-hamza]

Commenting here to follow this, Since I am also working on something similar.

Meanwhile If somebody can answer this or point me to the direction since I am unable to find anything in the documentation page.

I want to keep each user's chat history. For now I am been using SimpleChatStore but now I want to replace it with AzureChatStore.

Right now my functionality looks something like this; I load the chatstore from the disk with. chat_store = SimpleChatStore.from_persist_path(persist_path="chat_store.json")

It has all of the user's chat history and I can retrieve from it and displays it for the specific user when they log in to their session.

After they are done with their query, I run chat_store.persist(persist_path="chat_store.json") This stores the updated chat in the memory.

With AzureChatStore, LlamaIndex Demo: Azure Table Storage as a ChatStore shows loading of chat_store from AzureChatStore. My question is how to persist the chat history on the Azure after each query, Do we explicitly call add_message after processing of each query?