Prompt Refined, changed schema

agent.py

@@ -1,20 +1,14 @@
 import os
 from dotenv import load_dotenv
 from langchain_community.document_loaders import TextLoader, DirectoryLoader, UnstructuredPDFLoader, UnstructuredWordDocumentLoader
-from langchain_google_genai import ChatGoogleGenerativeAI, GoogleGenerativeAIEmbeddings
+from langchain_google_genai import ChatGoogleGenerativeAI
 from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain_community.vectorstores import PGVector
 from langchain.chains import RetrievalQA
 from langchain.prompts import PromptTemplate
 import json
-import time
-# import google.generativeai as genai
-from google import genai
 from google.oauth2 import service_account
 load_dotenv()
-# credential_file = os.getenv("GOOGLE_APPLICATION_CREDENTIAL")
-# # Load environment variables
-# os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = credential_file
+
 
 GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
 if GEMINI_API_KEY is None:
@@ -23,11 +17,7 @@ if GEMINI_API_KEY is None:
 conf = os.environ.get('GOOGLE_APPLICATION_CREDENTIALS')
 
 service_account_info = json.loads(conf)
-print(service_account_info)
-print(type(service_account_info))
 service_account_info = eval(service_account_info)
-print(service_account_info)
-print(type(service_account_info))
 
 credentials = service_account.Credentials.from_service_account_info(service_account_info)
 
@@ -66,7 +56,6 @@ def load_documents(directory):
 import os
 from dotenv import load_dotenv
 from langchain_community.llms import HuggingFacePipeline
-# from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_huggingface import HuggingFaceEmbeddings
 from langchain_community.document_loaders import TextLoader, DirectoryLoader, UnstructuredPDFLoader, UnstructuredWordDocumentLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
@@ -176,16 +165,27 @@ def create_health_agent(vector_store):
     Use Previous_Conversation to maintain consistency in the conversation.
     These are Previous_Conversation between you and user.
     Previous_Conversation: \n{previous_conversation}
+
+    These are info about the person.
+    User_Data: \n{user_data}
     Thoroughly analyze the Context, and also use context to answer the questions, aside of your knowledge.
-    Keep the answer concise.
-        
-    Context: {context}
+
+    Points to Adhere:
+
+    1. Only tell the schemes if user specifically asked, otherwise don't share schemes information.
+    2. If the user asks about schemes, Ask about what state they belong to first.
+    2. You can act as a mental-health counselor if needed. 
+    3. Give precautions and natural-remedies for the diseases, if user asked or it's needed, only for Common diseases include the common cold, flu etc.
+    4. Also Use Information from Context to answer the questions.
+    6. Ask the preferred language of the user, In the starting of the conversation.
+    
+    Context: {context}\n
     Question: {question}
     Answer:"""
     
     PROMPT = PromptTemplate(
         template=prompt_template, 
-        input_variables=["context", "question", "previous_conversation"]
+        input_variables=["context", "question", "previous_conversation", "user_data"]
     )
 
     if llm is None:

fetch.py

@@ -5,7 +5,7 @@ import json
 Contains the example code to retrieve response from the server in python-requests"""
 
 ## without previous_state
-url = "http://localhost:8000/retrieve"
+url = "https://arpit-bansal-healthbridge.hf.space/"
 headers = {
     "accept": "application/json",
     "Content-Type": "application/json"

main.py

@@ -22,11 +22,24 @@ agent = agent_with_db()
 @app.post("/retrieve", status_code=200)
 async def retrieve(request:request):
     prev_conv = request.previous_state
+    user_data = request.user_data
+    user_info = {}
+    if user_data is None:
+        user_info = "No user data available"
+
+    if user_data["state"] is None:
+        user_info["state_user_belongs_to"] = "No state information available"
+    
+    if user_data["gender"] is None:
+        user_info["sex_of_user"] = "Information not available"
+    
+    user_info["state_user_belongs_to"] = user_data["state"]
+    user_info["sex_of_user"] = user_data["gender"]
     print(prev_conv)
     if prev_conv is None:
         prev_conv = "No previous conversation available, first time"
     query = request.query
     prev_conv = str(prev_conv)
-    response = agent({"query": query, "previous_conversation": prev_conv})
+    response = agent({"query": query, "previous_conversation": prev_conv, "user_data": user_info})
 
     return {"response": response["result"]}  

schemas.py

@@ -3,4 +3,5 @@ from typing import Optional, Literal, List, Dict, Any
 
 class request(BaseModel):
     previous_state: Optional[List[Dict]]=None
-    query: str
+    query: str
+    user_data: Optional[Dict]=None