app.py

import os
import streamlit as st
import arxiv
import random
import datetime

# -------------------------------
# Groq API Client
# -------------------------------
from groq import Groq

client = Groq(
    api_key=os.environ.get("GROQ_API_KEY"),
)

# -------------------------------
# Helper Functions (Groq-based)
# -------------------------------
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 this paper in one sentence and provide 3 key takeaways:\n\n{text}"
            }
        ],
        model="llama-3.3-70b-versatile",
    )
    return response.choices[0].message.content.strip()


def groq_eli5(text: str) -> str:
    """
    Explain the paper like I'm 5 years old.
    """
    response = client.chat.completions.create(
        messages=[
            {
                "role": "user",
                "content": f"Explain this paper as if I were 5 years old in one sentence:\n\n{text}"
            }
        ],
        model="llama-3.3-70b-versatile",
    )
    return response.choices[0].message.content.strip()


def calculate_scores(paper):
    """
    Generate trust and relevance scores for a paper.
    """
    trust_score = random.randint(5, 10)  # Placeholder, can be improved with citations data
    relevance_score = random.randint(5, 10)  # Placeholder, can use NLP topic matching
    return trust_score, relevance_score


def retrieve_papers(query=None, max_results=10, random_mode=False):
    """
    Retrieve academic papers from arXiv, either based on search or randomly.
    """
    if random_mode:
        query = ""  # Empty query fetches random results
    
    search = arxiv.Search(query=query, max_results=max_results)
    papers = []
    
    for result in search.results():
        trust_score, relevance_score = calculate_scores(result)
        paper = {
            "title": result.title,
            "summary": result.summary,
            "url": result.pdf_url,
            "authors": [author.name for author in result.authors],
            "published": result.published.strftime('%Y-%m-%d') if isinstance(result.published, datetime.datetime) else "n.d.",
            "doi": f"https://doi.org/10.48550/arXiv.{result.entry_id.split('/')[-1]}",
            "bib_explorer": f"https://arxiv.org/abs/{result.entry_id.split('/')[-1]}",
            "litmaps": f"https://app.litmaps.com/preview/{result.entry_id.split('/')[-1]}",
            "connected_papers": f"https://www.connectedpapers.com/{result.entry_id.split('/')[-1]}",
            "trust_score": trust_score,
            "relevance_score": relevance_score
        }
        papers.append(paper)
    return papers

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

with st.sidebar:
    st.header("🔍 Search Parameters")
    query = st.text_input("Research topic or question:")
    
    col1, col2 = st.columns([4, 1])
    with col1:
        search_button = st.button("🚀 Find Articles")
    with col2:
        random_button = st.button("🎲 Random Papers")
    
    if search_button:
        if query.strip():
            with st.spinner("Searching arXiv..."):
                st.session_state.papers = retrieve_papers(query=query, max_results=10)
                st.success(f"Found {len(st.session_state.papers)} papers!")
        else:
            st.warning("Please enter a search query")
    
    if random_button:
        with st.spinner("Fetching random papers..."):
            st.session_state.papers = retrieve_papers(max_results=random.randint(5, 15), random_mode=True)
            st.success(f"Fetched {len(st.session_state.papers)} random papers!")

if 'papers' in st.session_state and st.session_state.papers:
    st.header("📑 Retrieved Papers")
    for idx, paper in enumerate(st.session_state.papers, 1):
        with st.expander(f"{idx}. {paper['title']}"):
            st.markdown(f"**Authors:** {', '.join(paper['authors'])}")
            st.markdown(f"**Published:** {paper['published']}")
            st.markdown(f"**[PDF Link]({paper['url']})** | **[DOI]({paper['doi']})** | **[Bib Explorer]({paper['bib_explorer']})** | **[Litmaps]({paper['litmaps']})** | **[Connected Papers]({paper['connected_papers']})**")
            
            with st.spinner("Generating summaries..."):
                summary = groq_summarize(paper['summary'])
                eli5_summary = groq_eli5(paper['summary'])
            
            st.markdown(f"**Summary:** {summary}")
            st.markdown(f"**ELI5:** {eli5_summary}")
            
            st.markdown(f"**Trust Score:** {paper['trust_score']} / 10  ⭐ | **Relevance Score:** {paper['relevance_score']} / 10 🔥")
else:
    st.info("Enter a query or click 🎲 Random Papers to get started.")