app.py

import streamlit as st
import requests
from PyPDF2 import PdfReader
from transformers import pipeline
from huggingface_hub import HfApi
import io
import os
from datetime import datetime

# --- Constants ---
COMMUNITY_BETA_MESSAGE = "This Streamlit app is part of a community in Beta. Please open discussions in the Community tab of the Community card."
DEFAULT_BADGE_IMAGE_URL = "https://img.shields.io/badge/Hugging%20Face-Space-blue"
COPYRIGHT_TEXT = f"© {datetime.now().year} Your Name/Organization. All rights reserved."

# --- CSS ---
st.markdown(
    """
    <style>
    .reportview-container {
        margin-top: -2em;
    }
    #MainMenu {visibility: hidden;}
    footer {visibility: hidden;}
    header {visibility: hidden;}
    .st-emotion-cache-z53if6 {
        padding-top: 10px;
    }
    </style>
    """,
    unsafe_allow_html=True,
)

# --- Sidebar Settings ---
st.sidebar.header("Settings")
arxiv_link = st.sidebar.text_input("arXiv Paper Link or ID", placeholder="e.g., https://arxiv.org/abs/2301.00001 or 2301.00001")
custom_space_name = st.sidebar.text_input("Custom Hugging Face Space Name (optional)")
badge_image_url = st.sidebar.text_input("Badge Image URL", DEFAULT_BADGE_IMAGE_URL)
copyright_text = st.sidebar.text_input("Copyright Text", COPYRIGHT_TEXT)

hf_token = st.sidebar.text_input("Hugging Face Token", type="password")

# --- Main App ---
st.title("arXiv Paper to Hugging Face Space")
st.info(COMMUNITY_BETA_MESSAGE)

if hf_token:
    try:
        api = HfApi(token=hf_token)
        user_info = api.whoami()
        hf_username = user_info['fullname'] if 'fullname' in user_info else user_info['name']
        st.sidebar.success(f"Logged in as: {hf_username}")
    except Exception as e:
        st.sidebar.error(f"Error with Hugging Face Token: {e}")

if arxiv_link:
    arxiv_id = None
    if arxiv_link.startswith("https://arxiv.org/abs/"):
        arxiv_id = arxiv_link.split("/")[-1]
    elif arxiv_link.isdigit():
        arxiv_id = arxiv_link
    elif arxiv_link.startswith("arxiv:"):
        arxiv_id = arxiv_link.split(":")[-1]

    if arxiv_id:
        pdf_url = f"https://arxiv.org/pdf/{arxiv_id}.pdf"
        try:
            response = requests.get(pdf_url)
            response.raise_for_status()
            pdf_content = response.content
            pdf_file = io.BytesIO(pdf_content)
            reader = PdfReader(pdf_file)
            text = ""
            for page in reader.pages:
                text += page.extract_text()

            st.subheader("Paper Content Preview:")
            st.markdown(f'<iframe src="{pdf_url}" width="700" height="600" type="application/pdf"></iframe>', unsafe_allow_html=True)

            if st.button("Convert to Hugging Face Space"):
                if not hf_token:
                    st.warning("Please enter your Hugging Face Token in the sidebar to create a Space.")
                else:
                    space_name_suffix = custom_space_name if custom_space_name else arxiv_id
                    space_name = f"arxiv-{space_name_suffix}"

                    try:
                        api = HfApi(token=hf_token)
                        repo_id = f"{hf_username}/{space_name}"
                        api.create_repo(repo_id=repo_id, space_sdk="static")

                        # Save the PDF to a temporary file
                        with open("paper.pdf", "wb") as f:
                            f.write(pdf_content)
                        api.upload_file(
                            path_or_fileobj="paper.pdf",
                            path_in_repo="paper.pdf",
                            repo_id=repo_id,
                            repo_type="space",
                        )
                        os.remove("paper.pdf")

                        # PDF Analysis
                        try:
                            st.info("Analyzing PDF content...")
                            pipe = pipeline("text2text-generation", model="deepseek-ai/DeepSeek-R1-Distill-Qwen-32B")
                            analysis_result = pipe(text[:4096], max_length=512)[0]['generated_text'] # Limit input for faster processing

                            analysis_pdf_content = f"""
                            # Analysis of arXiv Paper: {arxiv_id}

                            **Generated on:** {datetime.now().strftime("%Y-%m-%d %H:%M:%S")}

                            **Analysis:**
                            {analysis_result}

                            ---
                            {copyright_text}
                            """

                            # Save analysis to a temporary text file
                            with open("analysis.txt", "w") as f:
                                f.write(analysis_pdf_content)

                            api.upload_file(
                                path_or_fileobj="analysis.txt",
                                path_in_repo="analysis.txt",
                                repo_id=repo_id,
                                repo_type="space",
                            )
                            os.remove("analysis.txt")

                            st.success(f"Analysis saved to the Space as `analysis.txt`.")

                        except Exception as e_analysis:
                            st.error(f"Error during PDF analysis: {e_analysis}")

                        badge_html = f"""
                        <a href="https://huggingface.co/spaces/{repo_id}" target="_blank">
                            <img src="{badge_image_url}" alt="Hugging Face Space">
                        </a>
                        """
                        st.subheader("Hugging Face Space Created!")
                        st.markdown(f"Space URL: https://huggingface.co/spaces/{repo_id}")
                        st.markdown("Embed this badge in your README or website:")
                        st.code(badge_html, language="html")

                    except Exception as e_hf:
                        st.error(f"Error creating or updating Hugging Face Space: {e_hf}")

        except requests.exceptions.RequestException as e_http:
            st.error(f"Error fetching PDF from arXiv: {e_http}")
        except Exception as e_pdf:
            st.error(f"Error processing PDF: {e_pdf}")
    else:
        st.warning("Invalid arXiv link or ID format.")

st.markdown("---")
st.markdown(copyright_text)