app.py

# import os
# import base64
# import zipfile
# from pathlib import Path
# import streamlit as st
# from byaldi import RAGMultiModalModel
# from openai import OpenAI

# import os
# st.write("Current Working Directory:", os.getcwd())

# # Function to unzip a folder if it does not exist
# # def unzip_folder_if_not_exist(zip_path, extract_to):
# #     if not os.path.exists(extract_to):
# #         with zipfile.ZipFile(zip_path, 'r') as zip_ref:
# #             zip_ref.extractall(extract_to)

# # # Example usage
# # zip_path = 'medical_index.zip'
# # extract_to = 'medical_index'
# # unzip_folder_if_not_exist(zip_path, extract_to)

# # Preload the RAGMultiModalModel
# @st.cache_resource
# def load_model():
#     return RAGMultiModalModel.from_index("./medical_index")

# RAG = load_model()

# # OpenAI API key from environment
# api_key = os.getenv("OPENAI_API_KEY")
# client = OpenAI(api_key=api_key)

# # Streamlit UI
# st.title("Medical Diagnostic Assistant")
# st.write("Enter a medical query and get diagnostic recommendations along with visual references.")

# # User input
# query = st.text_input("Query", "What should be the appropriate diagnostic test for peptic ulcer?")

# if st.button("Submit"):
#     if query:
#         # Search using RAG model
#         with st.spinner('Retrieving information...'):
#             try:
#                 returned_page = RAG.search(query, k=1)[0].base64

#                 # Decode and display the retrieved image
#                 image_bytes = base64.b64decode(returned_page)
#                 filename = 'retrieved_image.jpg'
#                 with open(filename, 'wb') as f:
#                     f.write(image_bytes)

#                 # Display image in Streamlit
#                 st.image(filename, caption="Reference Image", use_column_width=True)

#                 # Get model response
#                 response = client.chat.completions.create(
#                     model="gpt-4o-mini-2024-07-18",
#                     messages=[
#                         {"role": "system", "content": "You are a helpful assistant. You only answer the question based on the provided image"},
#                         {
#                             "role": "user",
#                             "content": [
#                                 {"type": "text", "text": query},
#                                 {
#                                     "type": "image_url",
#                                     "image_url": {"url": f"data:image/jpeg;base64,{returned_page}"},
#                                 },
#                             ],
#                         },
#                     ],
#                     max_tokens=300,
#                 )
                
#                 # Display the response
#                 st.success("Model Response:")
#                 st.write(response.choices[0].message.content)
#             except Exception as e:
#                 st.error(f"An error occurred: {e}")
#     else:
#         st.warning("Please enter a query.")


import os
import base64
import zipfile
from pathlib import Path
import streamlit as st
from byaldi import RAGMultiModalModel
from openai import OpenAI

# Preload the RAGMultiModalModel
@st.cache_resource
def load_model():
    return RAGMultiModalModel.from_index("/home/user/app/medical_index")

RAG = load_model()

# OpenAI API key from environment
api_key = os.getenv("OPENAI_API_KEY")
client = OpenAI(api_key=api_key)

# Streamlit UI
st.title("Medical Diagnostic Assistant")
st.write("Enter a medical query and get diagnostic recommendations along with visual references.")

# User input for selecting the model
model_options = ["gpt-4o", "gpt-4o-mini", "o1-preview", "o1-mini"]
selected_model = st.selectbox("Choose a GPT model", model_options)

# User input for query
query = st.text_input("Query", "What should be the appropriate diagnostic test for peptic ulcer?")

if st.button("Submit"):
    if query:
        # Search using RAG model
        with st.spinner('Retrieving information...'):
            try:
                # Get top 10 images
                returned_pages = RAG.search(query, k=10)
                image_urls = []
                for i, page in enumerate(returned_pages):
                    image_bytes = base64.b64decode(page.base64)
                    filename = f'retrieved_image_{i}.jpg'
                    with open(filename, 'wb') as f:
                        f.write(image_bytes)
                    image_urls.append(f"data:image/jpeg;base64,{page.base64}")

                # Display the first image initially
                st.image(filename, caption="Reference Image", use_column_width=True)

                # Get model response
                response = client.chat.completions.create(
                    model=selected_model,
                    messages=[
                        {"role": "system", "content": "You are a helpful assistant. You only answer the question based on the provided image and select the right option. You will need to provide the exaplanation from the context as well. DO NOT answer from your previous knowledge ; only answer from the images provided."},
                        {
                            "role": "user",
                            "content": [
                                {"type": "text", "text": query},
                                *[{"type": "image_url", "image_url": {"url": url}} for url in image_urls],
                            ],
                        },
                    ],
                    max_tokens=300,
                )

                # Display the response
                st.success("Model Response:")
                st.write(response.choices[0].message.content)

                # Option to see all references
                if st.button("Show References"):
                    st.subheader("References")
                    for i, page in enumerate(returned_pages):
                        st.image(f'retrieved_image_{i}.jpg', caption=f"Reference Image {i+1}", use_column_width=True)
            except Exception as e:
                st.error(f"An error occurred: {e}")
    else:
        st.warning("Please enter a query.")