app.py

import streamlit as st
from langchain_experimental.data_anonymizer import PresidioAnonymizer, PresidioReversibleAnonymizer
from langchain_groq import ChatGroq
# from langchain.chat_models import ChatGroq 
from dotenv import load_dotenv
import os
# Load environment variables
load_dotenv()

GROQ_API_KEY = os.getenv("GROQ_API_KEY")

# Initialize anonymizer
anonymizer = PresidioReversibleAnonymizer()
llm = ChatGroq(model_name="llama-3.3-70b-versatile")  # Choose an available model

st.title("Call on Doc Data Anonymization")

# User Input
user_input = st.text_area("Enter your text:", "My name is John Doe and my phone number is 123-456-7890.")

if st.button("Process"):
    # Anonymization
    anonymized_text = anonymizer.anonymize(user_input)
    st.subheader("1. Original Text:")
    st.write(user_input)
    
    st.subheader("2. Anonymized Text:")
    st.write(anonymized_text)
    
    # Get LLM response
    response = llm.predict(anonymized_text)
    st.subheader("3. LLM Response:")
    st.write(response)
    
    # De-anonymization
    deanonymized_response = anonymizer.deanonymize(response)
    st.subheader("4. De-anonymized Response:")
    st.write(deanonymized_response)