app.py

"""
Gradio UI wrapper for the Academic Recommendation API
"""
import gradio as gr
import requests
import json
import subprocess
import time
import threading

# Start FastAPI server in background
def start_fastapi():
    subprocess.Popen([
        "uvicorn", 
        "api_server:app", 
        "--host", "0.0.0.0", 
        "--port", "8000"
    ])

# Start server in background thread
threading.Thread(target=start_fastapi, daemon=True).start()
time.sleep(10)  # Wait for API to start

API_URL = "http://localhost:8000"

def get_recommendations(query, top_k=10):
    """Get recommendations from the API."""
    try:
        response = requests.post(
            f"{API_URL}/recommend",
            json={"query": query, "top_k": int(top_k)},
            timeout=30
        )
        
        if response.status_code == 200:
            data = response.json()
            
            # Format output nicely
            output = f"**Query:** {data['query']}\n\n"
            output += f"**Found {data['total_results']} papers** (in {data['execution_time_ms']:.2f}ms)\n\n"
            output += "---\n\n"
            
            for paper in data['recommendations']:
                output += f"### {paper['rank']}. {paper['title']}\n\n"
                output += f"**Authors:** {', '.join(paper['metadata']['authors'][:3])}\n\n"
                output += f"**Year:** {paper['metadata']['published'][:4]} | "
                output += f"**Citations:** {paper['metadata']['citationCount']} | "
                output += f"**Category:** {paper['metadata']['primary_category']}\n\n"
                output += f"**Scores:** Similarity: {paper['scores']['similarity']:.3f}, "
                output += f"Impact: {paper['scores']['impact_normalized']:.3f}, "
                output += f"Recency: {paper['scores']['recency']:.3f}\n\n"
                output += f"**Abstract:** {paper['abstract'][:300]}...\n\n"
                output += f"[View on arXiv]({paper['metadata']['url']})\n\n"
                output += "---\n\n"
            
            return output
        else:
            return f"❌ Error: {response.status_code} - {response.text}"
    
    except Exception as e:
        return f"❌ Error: {str(e)}"

def check_health():
    """Check API health."""
    try:
        response = requests.get(f"{API_URL}/health", timeout=5)
        if response.status_code == 200:
            data = response.json()
            return f"✅ **Status:** {data['status']}\n**Device:** {data['device']}\n**Corpus loaded:** {data['corpus_loaded']} items"
        else:
            return f"❌ API returned status {response.status_code}"
    except Exception as e:
        return f"❌ API not responding: {str(e)}"

# Create Gradio interface
with gr.Blocks(title="Academic Paper Recommender") as demo:
    gr.Markdown(
        """
        # 📚 Academic Paper Recommendation System
        
        LLM-powered recommendation system using SPECTER2 embeddings for semantic search.
        
        **Features:**
        - Semantic similarity matching using SPECTER2
        - Multi-signal ranking (relevance + citations + recency)
        - 6,436 physics papers from arXiv
        """
    )
    
    with gr.Tab("Search Papers"):
        with gr.Row():
            query_input = gr.Textbox(
                label="Search Query",
                placeholder="e.g., quantum entanglement, machine learning, gravitational waves",
                lines=2
            )
            top_k_input = gr.Slider(
                minimum=5,
                maximum=20,
                value=10,
                step=1,
                label="Number of Results"
            )
        
        search_button = gr.Button("🔍 Search", variant="primary")
        output = gr.Markdown(label="Results")
        
        search_button.click(
            fn=get_recommendations,
            inputs=[query_input, top_k_input],
            outputs=output
        )
        
        # Examples
        gr.Examples(
            examples=[
                ["quantum entanglement", 10],
                ["machine learning neural networks", 10],
                ["gravitational wave detection", 10],
                ["topological insulators", 5],
                ["quantum computing error correction", 10]
            ],
            inputs=[query_input, top_k_input]
        )
    
    with gr.Tab("API Health"):
        health_button = gr.Button("Check API Status")
        health_output = gr.Markdown()
        
        health_button.click(
            fn=check_health,
            outputs=health_output
        )

# Launch
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)