stateful-chat-client 0.1.0

No project description provided

stateful-chat-client

Conversation Logger for MongoDB

A Python package to facilitate recording and retrieving AI conversations from a MongoDB database.

Features

• Record conversations to MongoDB in real-time.
• Retrieve past conversations from the database.
• Organize conversations based on unique thread IDs.

Installation

To install the package, use pip:

pip install stateful-chat-client

Quickstart

1. Setting up the logger

   from stateful-chat-client import ConversationLogger
   
   logger = ConversationLogger(host='localhost', port=27017,    database_name='conversation_database', collection_name='conversations')
   

2. Recording a conversation

   from stateful-chat-client.conversation_logger import record_conversation
   from langchain.llms import OpenAI
   from langchain.chains import ConversationChain
   from langchain.chains.conversation.memory import ConversationBufferMemory
   
   conversation = ConversationChain(
       llm=OpenAI(),
       memory=ConversationBufferMemory(),
       verbose=True,
   )
   conversation.predict(input="Hello!")
   thread_id = str(uuid.uuid4())
   messages = conversation.memory.buffer_as_messages
   record_conversation(messages, thread_id, logger=logger)
   

3. Retrieve past conversations and save to memory

   from stateful-chat-client.conversation_logger import retrieve_and_save_to_memory
   
   new_conversation = ConversationChain(
       llm=OpenAI(),
       memory=ConversationBufferMemory(),
       verbose=True,
   )
   new_conversation = retrieve_and_save_to_memory(thread_id, new_conversation_instance, logger=logger)
   new_conversation.predict(input="Hello! What did I say first?")
   

Requirements

• MongoDB server (local or remote)
• Python 3.8 or newer