"""
Hugging Face Spaces deployment - Combined API and UI
"""
import os
os.environ['PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION'] = 'python'

import gradio as gr
import sys
import pandas as pd
from io import StringIO

# Add src to path for imports
from recommendation_engine import RecommendationEngine
import json

# Initialize engine
print("Loading recommendation engine...")
engine = RecommendationEngine(catalog_path='shl_catalogue.csv')
print("Engine loaded!")

def get_recommendations(query, top_k=10):
    """Get recommendations for a query"""
    try:
        if not query or len(query.strip()) < 10:
            return "❌ Query must be at least 10 characters long."
        
        # Get recommendations
        result = engine.recommend(query, top_k=int(top_k))
        
        # Format output
        output = f"## 🎯 Top {top_k} Recommendations\n\n"
        output += f"**Query:** {result['query']}\n\n"
        
        # Table header
        output += "| # | Assessment Name | Type | Score |\n"
        output += "|---|----------------|------|-------|\n"
        
        # Recommendations
        for i, rec in enumerate(result['recommendations'], 1):
            name = rec['assessment_name']
            test_type = rec['test_type_label']
            score = f"{rec['similarity_score']:.3f}"
            output += f"| {i} | [{name}]({rec['url']}) | {test_type} | {score} |\n"
        
        # LLM Explanation
        if result.get('explanation'):
            output += f"\n### 💡 AI Analysis\n\n{result['explanation']}\n"
        
        # Best recommendation
        if result.get('best_recommendation'):
            output += f"\n**🏆 Best Match:** {result['best_recommendation']}\n"
        
        return output
    
    except Exception as e:
        return f"❌ Error: {str(e)}"

def get_recommendations_with_csv(query, top_k=10):
    """Get recommendations and return both markdown and CSV file"""
    try:
        if not query or len(query.strip()) < 10:
            return "❌ Query must be at least 10 characters long.", None
        
        # Get recommendations
        result = engine.recommend(query, top_k=int(top_k))
        
        # Format markdown output
        output = f"## 🎯 Top {top_k} Recommendations\n\n"
        output += f"**Query:** {result['query']}\n\n"
        
        # Table header
        output += "| # | Assessment Name | Type | Score |\n"
        output += "|---|----------------|------|-------|\n"
        
        # Recommendations
        for i, rec in enumerate(result['recommendations'], 1):
            name = rec['assessment_name']
            test_type = rec['test_type_label']
            score = f"{rec['similarity_score']:.3f}"
            output += f"| {i} | [{name}]({rec['url']}) | {test_type} | {score} |\n"
        
        # LLM Explanation
        if result.get('explanation'):
            output += f"\n### 💡 AI Analysis\n\n{result['explanation']}\n"
        
        # Best recommendation
        if result.get('best_recommendation'):
            output += f"\n**🏆 Best Match:** {result['best_recommendation']}\n"
        
        # Create CSV file
        csv_data = []
        for i, rec in enumerate(result['recommendations'], 1):
            csv_data.append({
                'Rank': i,
                'Query': result['query'],
                'Assessment_Name': rec['assessment_name'],
                'Assessment_URL': rec['url'],
                'Test_Type': rec['test_type_label'],
                'Relevance_Score': round(rec['similarity_score'], 4)
            })
        
        df = pd.DataFrame(csv_data)
        csv_file = "recommendations.csv"
        df.to_csv(csv_file, index=False)
        
        return output, csv_file
    
    except Exception as e:
        return f"❌ Error: {str(e)}", None

def get_recommendations_json(query, top_k=10):
    """API endpoint - returns JSON"""
    try:
        if not query or len(query.strip()) < 10:
            return {"error": "Query must be at least 10 characters long."}
        
        result = engine.recommend(query, top_k=int(top_k))
        
        # Format as per API specification
        return {
            "query": result['query'],
            "recommendations": [
                {
                    "assessment_name": rec['assessment_name'],
                    "url": rec['url'],
                    "relevance_score": rec['similarity_score'],
                    "test_type": rec['test_type_label']
                }
                for rec in result['recommendations']
            ],
            "total_results": result['total_results'],
            "explanation": result.get('explanation'),
            "best_recommendation": result.get('best_recommendation')
        }
    
    except Exception as e:
        return {"error": str(e)}

# Create Gradio interface with tabs
with gr.Blocks(title="SHL Assessment Recommendation System", theme=gr.themes.Soft()) as demo:
    
    gr.Markdown("""
    # 🎯 SHL Assessment Recommendation System
    
    Intelligent AI-powered recommendation system for SHL assessments using RAG (Retrieval-Augmented Generation).
    
    **Features:**
    - 🔍 Semantic search with sentence embeddings
    - 🤖 AI explanations powered by Google Gemini
    - ⚖️ Smart balancing of technical & behavioral assessments
    - 📊 54 assessments from SHL catalog
    - 📥 Download recommendations as CSV
    """)
    
    with gr.Tabs():
        # Tab 1: Web UI
        with gr.Tab("🌐 Web Interface"):
            gr.Markdown("### Enter your job description or query:")
            
            with gr.Row():
                with gr.Column(scale=4):
                    query_input = gr.Textbox(
                        label="Job Query",
                        placeholder="e.g., 'Java developer with collaboration skills'",
                        lines=3
                    )
                with gr.Column(scale=1):
                    top_k_slider = gr.Slider(
                        minimum=1, 
                        maximum=10, 
                        value=10, 
                        step=1, 
                        label="Number of Recommendations"
                    )
            
            recommend_btn = gr.Button("🎯 Get Recommendations", variant="primary", size="lg")
            
            with gr.Row():
                with gr.Column(scale=3):
                    output_md = gr.Markdown()
                with gr.Column(scale=1):
                    csv_download = gr.File(label="📥 Download CSV", visible=True)
            
            recommend_btn.click(
                fn=get_recommendations_with_csv,
                inputs=[query_input, top_k_slider],
                outputs=[output_md, csv_download]
            )
            
            gr.Examples(
                examples=[
                    ["I am hiring for Java developers who can also collaborate effectively with my business teams.", 10],
                    ["Looking to hire mid-level professionals who are proficient in Python, SQL and JavaScript.", 10],
                    ["I am hiring for an analyst and want to screen using Cognitive and personality tests.", 8],
                    ["Entry-level sales professional needed for my team.", 5],
                ],
                inputs=[query_input, top_k_slider],
            )
        
        # Tab 2: API
        with gr.Tab("🔌 API"):
            gr.Markdown("""
            ### REST API Endpoint
            
            Use this interface to test the API JSON response format.
            
            **Endpoint:** `/recommend` (POST)
            
            **Request Format:**
            ```json
            {
                "query": "Your job description here",
                "top_k": 10
            }
            ```
            """)
            
            with gr.Row():
                with gr.Column():
                    api_query_input = gr.Textbox(
                        label="Query",
                        placeholder="Enter job description",
                        lines=3
                    )
                    api_top_k = gr.Number(label="top_k", value=10, precision=0)
                    api_btn = gr.Button("📡 Send API Request", variant="primary")
                
                with gr.Column():
                    api_output = gr.JSON(label="API Response")
            
            api_btn.click(
                fn=get_recommendations_json,
                inputs=[api_query_input, api_top_k],
                outputs=api_output
            )


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