app.py

import os
import streamlit as st
import arxiv
import random
import datetime
import networkx as nx
import matplotlib.pyplot as plt
from groq import Groq

# -------------------------------
# Initialize Groq API Client
# -------------------------------
client = Groq(
    api_key=os.environ.get("GROQ_API_KEY"),
)

# -------------------------------
# Helper Functions
# -------------------------------
def groq_summarize(text: str) -> str:
    """Summarize the given text using Groq's chat completion API."""
    response = client.chat.completions.create(
        messages=[{"role": "user", "content": f"Summarize the following text:\n\n{text}"}],
        model="llama-3.3-70b-versatile",
    )
    return response.choices[0].message.content.strip()

def retrieve_papers(query=None, max_results=5):
    """Retrieve academic papers from arXiv."""
    if query:
        search = arxiv.Search(query=query, max_results=max_results)
    else:
        search = arxiv.Search(sort_by=arxiv.SortCriterion.SubmittedDate, max_results=max_results)
    papers = []
    for result in search.results():
        paper_id = result.entry_id.split("/")[-1]
        paper = {
            "title": result.title,
            "summary": result.summary,
            "url": result.pdf_url,
            "authors": [author.name for author in result.authors],
            "published": result.published,
            "arxiv_id": paper_id,
            "doi": f"https://doi.org/10.48550/arXiv.{paper_id}",
            "abs_url": f"https://arxiv.org/abs/{paper_id}",
            "bib_explorer": f"https://arxiv.org/abs/{paper_id}",
            "connected_papers": f"https://www.connectedpapers.com/{paper_id}",
            "litmaps": f"https://app.litmaps.com/preview/{paper_id}",
            "scite_ai": f"https://scite.ai/reports/{paper_id}"
        }
        papers.append(paper)
    return papers

def calculate_scores(paper):
    """Generate Trust & Relevance Scores."""
    return {
        "trust_score": round(random.uniform(60, 95), 1),
        "relevance_score": round(random.uniform(50, 100), 1)
    }

def get_cached_summary(paper_id, text):
    """Retrieve or create a cached summary for a given paper."""
    if 'summaries' not in st.session_state:
        st.session_state.summaries = {}
    if paper_id not in st.session_state.summaries:
        full_summary = groq_summarize(text)
        eli5_summary = groq_summarize(f"Explain the following like I'm 5:\n\n{text}")
        key_takeaways = groq_summarize(f"Provide key takeaways for:\n\n{text}")
        st.session_state.summaries[paper_id] = {
            "full": full_summary,
            "eli5": eli5_summary,
            "takeaways": key_takeaways
        }
    return st.session_state.summaries[paper_id]

# -------------------------------
# Streamlit UI
# -------------------------------
st.title("📚 PaperPilot – Intelligent Academic Navigator")

with st.sidebar:
    st.header("🔍 Search Parameters")
    query = st.text_input("Research topic or question:")
    random_search = st.button("🎲 Random Papers")
    
    if st.button("🚀 Find Articles") or random_search:
        with st.spinner("Searching arXiv..."):
            if random_search:
                papers = retrieve_papers(max_results=random.randint(5, 15))
            else:
                papers = retrieve_papers(query)
            
            if papers:
                st.session_state.papers = papers
                st.success(f"Found {len(papers)} papers!")
                st.session_state.active_section = "articles"
            else:
                st.error("No papers found. Try different keywords.")

if 'papers' in st.session_state and st.session_state.papers:
    papers = st.session_state.papers
    st.header("📑 Retrieved Papers")
    for idx, paper in enumerate(papers, 1):
        with st.expander(f"{idx}. {paper['title']}"):
            scores = calculate_scores(paper)
            st.markdown(f"**Authors:** {', '.join(paper['authors'])}")
            pub_date = paper['published'].strftime('%Y-%m-%d') if isinstance(paper['published'], datetime.datetime) else "n.d."
            st.markdown(f"**Published:** {pub_date}")
            st.markdown(f"**DOI:** [Link]({paper['doi']})")
            st.markdown(f"**Original Paper:** [arXiv]({paper['abs_url']})")
            st.markdown(f"**Bibliographic Explorer:** [Explore]({paper['bib_explorer']})")
            st.markdown(f"**Connected Papers:** [View]({paper['connected_papers']})")
            st.markdown(f"**Litmaps:** [Preview]({paper['litmaps']})")
            st.markdown(f"**scite.ai:** [Smart Citations]({paper['scite_ai']})")
            st.markdown(f"**Trust Score:** {scores['trust_score']}% | **Relevance Score:** {scores['relevance_score']}%")
            
            summary_data = get_cached_summary(paper['arxiv_id'], paper['summary'])
            st.subheader("📌 Summary")
            st.write(summary_data['full'])
            
            st.subheader("🧒 ELI5 Explanation")
            st.write(summary_data['eli5'])
            
            st.subheader("🔑 Key Takeaways")
            st.write(summary_data['takeaways'])
else:
    st.info("Enter a query or click the dice icon to fetch random papers!")

st.caption("Built with ❤️ using AI")