add database memory

bot/panda_bot.py

@@ -24,9 +24,14 @@ Patient Care Focus: As an assistant, Panda should regularly check in on the pati
 Patient Information:
 
 - Patient Name: '{username}'
-- Hemodialysis Schedule: Tuesday and Friday at 08:00 AM
+- Hemodialysis Schedule 1: {schedule_one}
+- Hemodialysis Schedule 2: {schedule_two}
+- Liquid Intake: {liquid_intake}
+- temperature: {temperature}
+- blood_pressure: {blood_pressure}
+- food_intake: {food_intake}
 
-Panda should always remember this schedule and provide reminders or updates when relevant.
+Panda should always remember this schedule and provide reminders or updates when relevant. Your current time is {current_time}.
 
 ---
 
@@ -71,6 +76,8 @@ Goal: Your goal is to ensure the patient feels well-informed, safe, and supporte
 
 ---
 
+
+
 Begin!
 
 Previous conversation history:
@@ -106,6 +113,13 @@ class CustomPromptTemplate(StringPromptTemplate):
     
     tools_getter: List[StructuredTool]
     username: str
+    schedule_one: str
+    schedule_two: str
+    liquid_intake: str
+    temperature: str
+    blood_pressure: str
+    food_intake: str
+    current_time: str
 
     def format(self, **kwargs) -> str:
         # Get the intermediate steps (AgentAction, Observation tuples)
@@ -125,6 +139,13 @@ class CustomPromptTemplate(StringPromptTemplate):
         # Create a list of tool names for the tools provided
         kwargs["tool_names"] = ", ".join([tool.name for tool in self.tools_getter])
         kwargs["username"] = self.username
+        kwargs["schedule_one"] = self.schedule_one
+        kwargs["schedule_two"] = self.schedule_two
+        kwargs["liquid_intake"] = self.liquid_intake
+        kwargs["temperature"] = self.temperature
+        kwargs["blood_pressure"] = self.blood_pressure
+        kwargs["food_intake"] = self.food_intake
+        kwargs["current_time"] = self.current_time
 
         return self.template.format(**kwargs)
     
@@ -163,26 +184,35 @@ class CustomOutputParser(AgentOutputParser):
 class PandaBot:
     def __init__(self, memory=None):
         # Initialize memory if provided, otherwise create a new one
-        self.memory = memory if memory else ConversationBufferMemory(memory_key="chat_history")
+        self.memory = None
         # Initialize LLM, prompt, output parser, and tool list
         self.llm = Mistral()
         self.tools_list = [send_emergency_message_to_medic, search_information_for_question]
         self.tool_names = [i.name for i in self.tools_list]
         self.output_parser = CustomOutputParser()
         os.environ["SERPER_API_KEY"] = 'f90fe84e78ef9d2d8e377ab5c6fe3a4a25f42ef0'
-        os.environ["LANGCHAIN_TRACING_V2"] = 'true'
-        os.environ["LANGCHAIN_ENDPOINT"] = 'https://api.smith.langchain.com'
-        os.environ["LANGCHAIN_API_KEY"] = 'ls__413a7563fa034592be6c6a241176932a'
-        os.environ["LANGCHAIN_PROJECT"] = 'LLM Patient Monitoring'
+        # os.environ["LANGCHAIN_TRACING_V2"] = 'true'
+        # os.environ["LANGCHAIN_ENDPOINT"] = 'https://api.smith.langchain.com'
+        # os.environ["LANGCHAIN_API_KEY"] = 'ls__413a7563fa034592be6c6a241176932a'
+        # os.environ["LANGCHAIN_PROJECT"] = 'LLM Patient Monitoring'
 
 
-    def query(self, input_text: str, username: str) -> str:
+    def query(self, input_text: str, username: str, memory: any, schedule_one:any, schedule_two:any, liquid_intake:any, temperature:any, blood_pressure:any, food_intake:any, current_time:any) -> str:
         prompt = CustomPromptTemplate(
             input_variables=["input", "intermediate_steps", "chat_history"], 
             template=template, 
             validate_template=False, 
             tools_getter=self.tools_list, 
-            username=username)
+            username=username,
+            schedule_one=schedule_one,
+            schedule_two=schedule_two,
+            liquid_intake=liquid_intake,
+            temperature=temperature,
+            blood_pressure=blood_pressure,
+            food_intake=food_intake,
+            current_time=current_time)
+        
+        self.memory = memory
 
         self.llm_chains = LLMChain(llm=self.llm, prompt=prompt)
 

database/__init__.py

@@ -0,0 +1,40 @@
+from supabase import create_client
+
+# Supabase Database URL & API Key (from Supabase dashboard)
+SUPABASE_URL = "https://ixugilhodhqmdedbzoaq.supabase.co"
+SUPABASE_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6Iml4dWdpbGhvZGhxbWRlZGJ6b2FxIiwicm9sZSI6ImFub24iLCJpYXQiOjE3NDIyMTg0MzIsImV4cCI6MjA1Nzc5NDQzMn0.YHD5VBaUKLHieIG7KsI3ZHBCeSBli27F2lOdNrQCfck"
+
+# Connect to Supabase
+supabase = create_client(SUPABASE_URL, SUPABASE_KEY)
+
+# PostgreSQL connection string for LangChain
+DB_URI = f"postgresql://postgres.ixugilhodhqmdedbzoaq:[email protected]:5432/postgres"
+
+
+# # Function to get user-specific chat memory stored in Supabase
+# def get_user_memory(user_id):
+#     return ConversationBufferMemory(
+#         chat_memory=SQLChatMessageHistory(
+#             connection_string=DB_URI,
+#             session_id=user_id
+#         ),
+#         return_messages=True  # FIX: Prevent duplicate storage
+#     )
+
+# # Example Usage
+# if __name__ == "__main__":
+#     user_id = "user_meh12312313123"
+
+#     #Create new user memory if not exists
+
+#     memory = ConversationBufferMemory(        chat_memory=SQLChatMessageHistory(
+#             connection_string=DB_URI,
+#             session_id=user_id
+#         ),
+#         return_messages=False )
+
+#     memory.save_context({"input": "Hello, Supabase!"}, {"output": "Hello! How can I assist you?"})
+
+#     memory = get_user_memory('user_meh12312313123')
+
+#     print(memory.load_memory_variables({}))

database/database_util.py

@@ -0,0 +1,13 @@
+from langchain.memory import ConversationBufferMemory
+from langchain.memory.chat_message_histories import SQLChatMessageHistory
+from database import DB_URI
+
+def get_user_memory(user_id):
+    # Check if the user exists in the database, if not, create new memory
+    return ConversationBufferMemory(
+        chat_memory=SQLChatMessageHistory(
+            connection_string=DB_URI,
+            session_id=user_id,
+        ),memory_key="chat_history",
+        return_messages=True  # FIX: Prevent duplicate storage
+    )

server.py

@@ -4,7 +4,8 @@ from flask import Flask, request, jsonify
 # from langchain_core.messages import HumanMessage, AIMessage
 # from langchain.prompts import ChatPromptTemplate
 from deep_translator import GoogleTranslator
-
+from database.database_util import get_user_memory
+from datetime import datetime
 
 app = Flask(__name__)
 
@@ -42,26 +43,36 @@ Output:
 #     llm.memory.chat_memory.add_ai_message(AIMessage(response))
 #     return jsonify({"response": translate})
 
-def simulate_llm_query(user_input, username):
+def simulate_llm_query(user_input, username, schedule_one, schedule_two, liquid_intake, temperature, blood_pressure, food_intake):
     """
     Simulates querying a language model.
     Replace this function's logic with actual LLM querying.
     """
     # Placeholder response logic, replace with actual LLM integration
-    return llm.query(user_input, username)
+    memory = get_user_memory(username)
+
+    now = datetime.now()
+    current_time = now.strftime("%A, %I:%M %p")
+    return llm.query(user_input, username, memory, schedule_one, schedule_two, liquid_intake, temperature, blood_pressure, food_intake, current_time)
 
 @app.route('/query', methods=['POST'])
 def query_llm():
     data = request.json
     user_input = data.get('input')
     username = data.get('username')
+    schedule_one = str(data.get('schedule_one'))
+    schedule_two = str(data.get('schedule_two'))
+    liquid_intake = str(data.get('liquid_intake'))
+    temperature = str(data.get('temperature'))
+    blood_pressure = str(data.get('blood_pressure'))
+    food_intake = str(data.get('food_intake'))
 
     input_translate = GoogleTranslator(source='id', target='en').translate(user_input)
     
     if not user_input:
         return jsonify({"error": "No input provided"}), 400
     
-    response = simulate_llm_query(input_translate, username)
+    response = simulate_llm_query(input_translate, username, schedule_one, schedule_two, liquid_intake, temperature, blood_pressure, food_intake)
 
     output_translate = GoogleTranslator(source='en', target='id').translate(response)