app.py

import os
import sqlite3
import requests
import openai
import gradio as gr

# Load API keys from environment
openai.api_key = os.getenv("OPENAI_API_KEY")

# --- Agents ---
def db_agent(query: str) -> str:
    try:
        conn = sqlite3.connect("shop.db")
        cur = conn.cursor()
        if "max revenue" in query.lower():
            cur.execute(
                """
                SELECT product, SUM(amount) AS revenue
                FROM transactions
                WHERE date = date('now')
                GROUP BY product
                ORDER BY revenue DESC
                LIMIT 1
                """
            )
            row = cur.fetchone()
            if row:
                return f"Top product today: {row[0]} with ₹{row[1]:,.2f}"
            return "No transactions found for today."
        return None
    except sqlite3.OperationalError as e:
        return f"Database error: {e}. Please initialize 'transactions' table in shop.db."


def web_search_agent(query: str) -> str:
    # Try fetching a snippet from SerpAPI, otherwise fallback to direct LLM
    try:
        resp = requests.get(
            "https://serpapi.com/search",
            params={"q": query, "api_key": os.getenv("SERPAPI_KEY")}  
        )
        data = resp.json()
        snippet = data.get("organic_results", [{}])[0].get("snippet", "").strip()
        if snippet:
            return llm_agent(f"Summarize: {snippet}")
    except Exception:
        pass
    # Fallback for no snippet or errors
    return llm_agent(query)


def llm_agent(prompt: str) -> str:
    # Updated for openai>=1.0.0 interface
    response = openai.chat.completions.create(
        model="gpt-4o-mini",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt},
        ],
        temperature=0.2,
    )
    return response.choices[0].message.content.strip()


def handle_query(query: str) -> str:
    q = query.lower()
    if any(k in q for k in ["max", "revenue", "today", "product"]):
        return db_agent(query)
    elif any(k in q for k in ["who", "what", "when", "where"]):
        return web_search_agent(query)
    else:
        return llm_agent(query)

# --- Gradio UI ---
with gr.Blocks() as demo:
    gr.Markdown("## Shop Voice-Box Assistant")
    user_input = gr.Textbox(placeholder="Type your question here...", lines=2)
    submit_btn = gr.Button("Submit")
    response_box = gr.Textbox(label="Answer")
    # Sample questions
    gr.Examples(
        examples=[
            ["What is the max revenue product today?"],
            ["Who invented the light bulb?"],
            ["Tell me a joke about cats."],
        ],
        inputs=user_input,
        outputs=response_box,
    )
    submit_btn.click(fn=handle_query, inputs=user_input, outputs=response_box)

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