import gradio as gr
import pandas as pd
import requests
import os
import datetime

# --- CONFIGURATION ---
# This pulls the URL you saved in your Hugging Face "Secrets"
N8N_WEBHOOK_URL = os.getenv("N8N_WEBHOOK_URL", "")
LOG_FILE = "hotel_analysis_history.csv"

# --- HELPER FUNCTIONS ---
def log_locally(hotel, lead_time, result):
    """Saves a local backup of the analysis for the 'Internal Database' tab."""
    now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M")
    new_entry = pd.DataFrame([[now, f"{hotel} (LT: {lead_time})", result]], 
                             columns=["Time", "Input Details", "n8n Outcome"])
    
    if not os.path.exists(LOG_FILE):
        new_entry.to_csv(LOG_FILE, index=False)
    else:
        new_entry.to_csv(LOG_FILE, mode='a', header=False, index=False)
    
    return pd.read_csv(LOG_FILE).sort_index(ascending=False)

def run_n8n_logic(hotel, lead_time):
    """Calls the n8n workflow to process the local CSV data."""
    if not N8N_WEBHOOK_URL:
        return "⚠️ Error: N8N_WEBHOOK_URL secret is not set in Hugging Face Space settings.", None

    # This is the 'Prompt' sent to the n8n Webhook node
    payload = {
        "hotel": hotel,
        "lead_time": int(lead_time)
    }

    try:
        # 1. Send data to n8n
        response = requests.post(N8N_WEBHOOK_URL, json=payload, timeout=15)
        
        # 2. Parse the response from the 'Respond to Webhook' node
        if response.status_code == 200:
            data = response.json()
            risk = data.get("risk_score", "Unknown")
            advice = data.get("recommendation", "No advice returned.")
            full_result = f"Risk Score: {risk}\nRecommendation: {advice}"
        else:
            full_result = f"Error: n8n returned status code {response.status_code}"
            
    except Exception as e:
        full_result = f"Connection Error: {str(e)}"

    # 3. Update the internal log for the UI
    updated_history = log_locally(hotel, lead_time, full_result)
    
    return full_result, updated_history

# --- GRADIO UI ---
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("# 🏨 LuxeRate AI: n8n Data Engine")
    gr.Markdown("This app sends your prompt to a private n8n workflow that analyzes local hotel records.")

    with gr.Tabs():
        # TAB 1: The Analysis Interface
        with gr.TabItem("🎯 Risk Analysis"):
            with gr.Row():
                with gr.Column():
                    hotel_input = gr.Dropdown(
                        choices=["City Hotel", "Resort Hotel"], 
                        label="Select Hotel Type", 
                        value="City Hotel"
                    )
                    lead_input = gr.Slider(
                        minimum=0, 
                        maximum=365, 
                        value=30, 
                        label="Lead Time (Days before arrival)"
                    )
                    analyze_btn = gr.Button("Analyze via n8n", variant="primary")
                
                with gr.Column():
                    output_display = gr.Textbox(
                        label="n8n Engine Output", 
                        lines=8, 
                        placeholder="Results will appear here..."
                    )

        # TAB 2: The Internal Database (CSV Log)
        with gr.TabItem("📊 Internal Database"):
            gr.Markdown("### Real-time Activity Log")
            gr.Markdown("Every analysis triggered above is recorded here for project auditing.")
            history_df = gr.DataFrame(
                value=pd.read_csv(LOG_FILE) if os.path.exists(LOG_FILE) else None,
                label="Historical Records",
                interactive=False
            )
            refresh_btn = gr.Button("Refresh Database View")

    # --- EVENT HANDLING ---
    analyze_btn.click(
        fn=run_n8n_logic, 
        inputs=[hotel_input, lead_input], 
        outputs=[output_display, history_df]
    )

    refresh_btn.click(
        fn=lambda: pd.read_csv(LOG_FILE).sort_index(ascending=False) if os.path.exists(LOG_FILE) else None,
        outputs=history_df
    )

# --- LAUNCH ---
if __name__ == "__main__":
    demo.launch()