app.py

"""
app.py – Gradio front-end for the Topic Modelling System.
Runs on HuggingFace Spaces and also accepts CLI:  python app.py data.csv
"""

import sys
import os
import tempfile
import logging

import pandas as pd
import gradio as gr

from agent import run_pipeline

logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
log = logging.getLogger(__name__)


# ─────────────────────────────────────────────────────────────────────────────
# Core processing wrapper for Gradio
# ─────────────────────────────────────────────────────────────────────────────

def process_csv(csv_file) -> tuple:
    """
    Gradio handler: receives an uploaded file object, runs the full pipeline,
    and returns display-ready outputs.

    Returns
    -------
    (
        status_msg    : str,
        review_df     : pd.DataFrame   – rendered in Gradio Dataframe,
        comparison_df : pd.DataFrame,
        gap_md        : str            – gap analysis as Markdown,
        narrative     : str,
        comp_file     : str            – path to comparison.csv for download,
        tax_file      : str            – path to taxonomy_map.json for download,
        narr_file     : str            – path to narrative.txt for download,
    )
    """
    if csv_file is None:
        empty = pd.DataFrame()
        return ("⚠️ Please upload a CSV file.", empty, empty, "", "", None, None, None)

    try:
        # csv_file.name is the temp-file path Gradio writes for us
        csv_path = csv_file.name if hasattr(csv_file, "name") else csv_file

        with tempfile.TemporaryDirectory() as tmpdir:
            result = run_pipeline(csv_path, output_dir=tmpdir)

            # Copy output files to a permanent temp location so Gradio can serve them
            import shutil
            out_dir = tempfile.mkdtemp()

            comp_dst = os.path.join(out_dir, "comparison.csv")
            tax_dst  = os.path.join(out_dir, "taxonomy_map.json")
            narr_dst = os.path.join(out_dir, "narrative.txt")

            shutil.copy(os.path.join(tmpdir, "comparison.csv"),    comp_dst)
            shutil.copy(os.path.join(tmpdir, "taxonomy_map.json"), tax_dst)
            shutil.copy(os.path.join(tmpdir, "narrative.txt"),     narr_dst)

        gap = result["gap"]
        gap_md = f"""### Gap Analysis Summary

| Metric | Value |
|--------|-------|
| **Total Topics Extracted** | {gap['total_topics']} |
| **MAPPED (in PAJAIS)** | {gap['mapped_count']} ({gap['mapped_percent']}%) |
| **NOVEL (emerging)** | {gap['novel_count']} ({gap['novel_percent']}%) |
| **Records Processed** | {result['record_count']} |

**Top MAPPED themes:** {', '.join(gap['top_mapped'])}

**Top NOVEL themes:** {', '.join(gap['top_novel'])}
"""

        status = (
            f"✅ Pipeline completed successfully!\n"
            f"   📄 {result['record_count']} records processed | "
            f"🏷️ {gap['total_topics']} topics extracted | "
            f"🗂️ {gap['mapped_count']} mapped | "
            f"✨ {gap['novel_count']} novel"
        )

        return (
            status,
            result["review_df"],
            result["comparison_df"],
            gap_md,
            result["narrative"],
            comp_dst,
            tax_dst,
            narr_dst,
        )

    except Exception as exc:
        log.exception("Pipeline failed")
        empty = pd.DataFrame()
        return (f"❌ Error: {exc}", empty, empty, "", "", None, None, None)


# ─────────────────────────────────────────────────────────────────────────────
# Gradio UI
# ─────────────────────────────────────────────────────────────────────────────

def build_ui() -> gr.Blocks:
    css = """
    .status-box textarea { font-size: 0.95rem; font-family: monospace; }
    .narrative-box textarea { font-size: 0.9rem; line-height: 1.6; }
    """

    with gr.Blocks(
        title="Topic Modelling System",
        theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="blue"),
        css=css,
    ) as demo:

        gr.Markdown(
            """
            # 📚 Topic Modelling System
            **Automated research-theme extraction, PAJAIS mapping, and gap analysis**

            Upload a CSV file containing `title` and `abstract` columns to begin.
            The system will extract ≥ 98 topics, compare title vs abstract themes,
            map topics against the PAJAIS taxonomy, and generate a 500-word academic narrative.
            """
        )

        with gr.Row():
            with gr.Column(scale=1):
                csv_input = gr.File(
                    label="📂 Upload CSV (title + abstract columns)",
                    file_types=[".csv"],
                    type="filepath",
                )
                run_btn = gr.Button("🚀 Run Analysis", variant="primary", size="lg")

            with gr.Column(scale=2):
                status_out = gr.Textbox(
                    label="Status",
                    interactive=False,
                    lines=3,
                    elem_classes=["status-box"],
                )

        gr.Markdown("---")

        with gr.Tabs():
            with gr.TabItem("🏷️ Extracted Topics"):
                review_table = gr.Dataframe(
                    label="Topic Review Table (topic_id | keyword | frequency)",
                    wrap=True,
                    interactive=False,
                )

            with gr.TabItem("🔄 Title vs Abstract Comparison"):
                comparison_table = gr.Dataframe(
                    label="Comparison Table",
                    wrap=True,
                    interactive=False,
                )

            with gr.TabItem("📊 Gap Analysis"):
                gap_md_out = gr.Markdown()

            with gr.TabItem("📝 Narrative (≈500 words)"):
                narrative_out = gr.Textbox(
                    label="Academic Narrative",
                    lines=28,
                    interactive=False,
                    elem_classes=["narrative-box"],
                )

        gr.Markdown("### 📥 Download Output Files")
        with gr.Row():
            dl_comparison = gr.File(label="comparison.csv",    interactive=False)
            dl_taxonomy   = gr.File(label="taxonomy_map.json", interactive=False)
            dl_narrative  = gr.File(label="narrative.txt",     interactive=False)

        run_btn.click(
            fn=process_csv,
            inputs=[csv_input],
            outputs=[
                status_out,
                review_table,
                comparison_table,
                gap_md_out,
                narrative_out,
                dl_comparison,
                dl_taxonomy,
                dl_narrative,
            ],
        )

        gr.Markdown(
            """
            ---
            *Topic Modelling System — powered by TF-IDF · LDA · NMF*
            """
        )

    return demo


# ─────────────────────────────────────────────────────────────────────────────
# Entry point
# ─────────────────────────────────────────────────────────────────────────────

if __name__ == "__main__":
    # CLI mode: python app.py data.csv
    if len(sys.argv) > 1:
        csv_path = sys.argv[1]
        if not os.path.isfile(csv_path):
            print(f"[ERROR] File not found: {csv_path}")
            sys.exit(1)

        print(f"[CLI] Running pipeline on: {csv_path}")

        result = run_pipeline(csv_path, output_dir=".")

        print("\n" + "=" * 60)
        print("PIPELINE COMPLETE")
        print("=" * 60)
        print(f"  Records processed : {result['record_count']}")
        print(f"  Topics extracted  : {result['gap']['total_topics']}")
        print(f"  MAPPED            : {result['gap']['mapped_count']} ({result['gap']['mapped_percent']}%)")
        print(f"  NOVEL             : {result['gap']['novel_count']} ({result['gap']['novel_percent']}%)")
        print(f"  Narrative words   : {len(result['narrative'].split())}")
        print("\nOutput files:")
        for f in result["output_files"]:
            print(f"  → {f}")
        print("=" * 60)

    else:
        # Gradio / HuggingFace Spaces mode
        demo = build_ui()
        demo.launch(server_name="0.0.0.0", server_port=7860)