The MemoryStore records chat history in database. It should have methods below to adapt operations in chatbot:
MemoryStore
Methods:
add_history: insert chat history to database, given a list of dictionaries with keys of 'question' and 'answer', [{'question': 'xxx', 'answer': 'xxx'}]get_history: return chat history in a list of tuples, [('this is question', 'this is answer')]
By default, it allows any SQL database suppported by SQLAlchemy 2.0 to manage memory. You can modify config.py to configure your database.
To manage memory with some other database, you can import your own MemoryStore in init.py.
In the meantime, don't forget to modify config.py for a changed memorydb_config.
You can follow instructions below to customize the module:
from typing import List
from config import memorydb_config
class MemoryStore:
def __init__(self):
'''Initialize memory storage, eg. connect to your database'''
pass
def add_history(self, project: str, session_id: str, messages: List[dict]):
'''Insert messages to the project table with a field of session_id.'''
pass
def get_history(self, project: str, session_id: str) -> List[tuple]:
'''Get chat history from the project table for each session.'''
def drop(self, project):
'''Clear all memory saved in the table "project"'''
def check(self, project) -> bool:
'''Check if the project table exists in database.'''