app.py

import streamlit as st
import os

from config import (
    OPENAI_API_KEY,
    GEMINI_API_KEY,
    DEFAULT_CHUNK_SIZE,
)
from models import configure_llms, openai_chat, gemini_chat
from pubmed_utils import (
    search_pubmed,
    fetch_pubmed_abstracts,
    chunk_and_summarize
)
from image_pipeline import load_image_model, analyze_image

###############################################################################
#                             STREAMLIT PAGE CONFIG                           #
###############################################################################
st.set_page_config(page_title="RAG + Image Captioning Demo", layout="wide")

###############################################################################
#                           INITIALIZE & LOAD MODELS                          #
###############################################################################
@st.cache_resource
def initialize_app():
    """
    Configure LLMs (OpenAI/Gemini) and load the image captioning model once.
    """
    configure_llms()
    model = load_image_model()
    return model

image_model = initialize_app()

###############################################################################
#                    HELPER: BUILD SYSTEM PROMPT WITH REFS                    #
###############################################################################
def build_system_prompt_with_refs(pmids, summaries):
    """
    Creates a system prompt for the LLM that includes references [Ref1], [Ref2], etc.
    """
    system_context = "You have access to the following summarized PubMed articles:\n\n"
    for idx, pmid in enumerate(pmids, start=1):
        ref_label = f"[Ref{idx}]"
        system_context += f"{ref_label} (PMID {pmid}): {summaries[pmid]}\n\n"
    system_context += "Use this info to answer the user's question, citing references if needed."
    return system_context

###############################################################################
#                                MAIN APP                                     #
###############################################################################
def main():
    st.title("RAG + Image Captioning: Production Demo")

    st.markdown("""
    This demonstration shows:
    1. **PubMed RAG**: Retrieve abstracts, summarize, and feed them into an LLM.
    2. **Image Captioning**: Upload an image for analysis using a known stable model.
    """)

    # Section A: Image Upload / Caption
    st.subheader("Image Captioning")
    uploaded_img = st.file_uploader("Upload an image (optional)", type=["png", "jpg", "jpeg"])
    if uploaded_img:
        with st.spinner("Analyzing image..."):
            caption = analyze_image(uploaded_img, image_model)
        st.image(uploaded_img, use_column_width=True)
        st.write("**Caption**:", caption)
        st.write("---")

    # Section B: PubMed-based RAG
    st.subheader("PubMed RAG Pipeline")
    user_query = st.text_input("Enter a medical question:", "What are the latest treatments for type 2 diabetes?")
    
    c1, c2, c3 = st.columns([2,1,1])
    with c1:
        st.markdown("**Parameters**:")
        max_papers = st.slider("Number of Articles", 1, 10, 3)
        chunk_size = st.slider("Chunk Size", 128, 1024, DEFAULT_CHUNK_SIZE)
    with c2:
        llm_choice = st.selectbox("Choose LLM", ["OpenAI: GPT-3.5", "Gemini: PaLM2"])
    with c3:
        temperature = st.slider("LLM Temperature", 0.0, 1.0, 0.3, step=0.1)

    if st.button("Run RAG Pipeline"):
        if not user_query.strip():
            st.warning("Please enter a query.")
            return
        
        with st.spinner("Searching PubMed..."):
            pmids = search_pubmed(user_query, max_papers)
        
        if not pmids:
            st.error("No PubMed results. Try a different query.")
            return
        
        with st.spinner("Fetching & Summarizing..."):
            abstracts_map = fetch_pubmed_abstracts(pmids)
            summarized_map = {}
            for pmid, text in abstracts_map.items():
                if text.startswith("Error:"):
                    summarized_map[pmid] = text
                else:
                    summarized_map[pmid] = chunk_and_summarize(text, chunk_size=chunk_size)

        st.subheader("Retrieved & Summarized PubMed Articles")
        for idx, pmid in enumerate(pmids, start=1):
            st.markdown(f"**[Ref{idx}] PMID {pmid}**")
            st.write(summarized_map[pmid])
            st.write("---")

        st.subheader("RAG-Enhanced Final Answer")
        system_prompt = build_system_prompt_with_refs(pmids, summarized_map)
        with st.spinner("Generating LLM response..."):
            if llm_choice == "OpenAI: GPT-3.5":
                answer = openai_chat(system_prompt, user_query, temperature=temperature)
            else:
                answer = gemini_chat(system_prompt, user_query, temperature=temperature)

        st.write(answer)
        st.success("Pipeline Complete.")

    st.markdown("---")
    st.markdown("""
    **Production Tips**:
    - Vector DB for advanced retrieval
    - Precise citation parsing
    - Rate limiting on PubMed
    - Multi-lingual expansions
    - Logging & monitoring
    - Security & privacy compliance
    """)

if __name__ == "__main__":
    main()