components/upload.py

import streamlit as st

from src.grobid import extract_metadata_grobid
from src.pdf_parser import extract_text_from_pdf

from src.rag_pipeline import (
    create_rag_database,
    analyze_paper
)


def upload_section():

    st.header("📤 Upload Research Papers")

    uploaded_files = st.file_uploader(
        "Upload PDF papers",
        type="pdf",
        accept_multiple_files=True
    )

    if not uploaded_files:
        return

    new_papers = []

    progress = st.progress(0)

    status = st.empty()

    total_files = len(uploaded_files)

    for idx, file in enumerate(uploaded_files):

        status.write(
            f"Processing **{file.name}**..."
        )

        # ----------------------------
        # GROBID Metadata
        # ----------------------------

        try:

            meta = extract_metadata_grobid(file)

        except Exception:

            meta = {

                "title": file.name.replace(".pdf", ""),

                "authors": ["Unknown"],

                "abstract": ""

            }

        # ----------------------------
        # Read Full Paper
        # ----------------------------

        file.seek(0)

        text = extract_text_from_pdf(file)

        meta["text"] = text

        if not meta.get("abstract"):

            meta["abstract"] = text[:3000]

        # ----------------------------
        # One AI Call
        # ----------------------------

        try:

            ai_result = analyze_paper(
                text=text,
                abstract=meta["abstract"]
            )

            meta.update(ai_result)

        except Exception:

            meta["summary"] = "Summary could not be generated."

            meta["abstract_summary"] = "Abstract summary unavailable."

            meta["limitations"] = "Limitations unavailable."

            meta["research_gaps"] = "Research gaps unavailable."

        new_papers.append(meta)

        progress.progress(
            (idx + 1) / total_files
        )

    st.session_state.papers.extend(
        new_papers
    )

    create_rag_database(
        st.session_state.papers
    )

    progress.empty()

    status.empty()

    st.success(
        f"Successfully processed {len(new_papers)} paper(s)."
    )