app.py

import gradio as gr
from datetime import datetime
import json
import difflib  # For better product name matching
from utils import get_product_by_name, get_products_by_category, get_products_and_category

# Dummy moderation function (replace with actual API)
def openai_moderation_check(text):
    return {"flagged": False}  # Assume no flagging for now

# Dummy AI response function (replace with actual API call to generate responses)
def get_completion_from_messages(messages):
    return "This is a placeholder response from the model."  # Mock response for now

# Function to log conversation to a JSON file
def log_conversation(user_input, ai_response, metadata=None, history=None):
    log_entry = {
        "timestamp": str(datetime.now()),
        "user_input": user_input,
        "ai_response": ai_response,
        "metadata": metadata,
        "history": history
    }
    with open("conversation_log.json", "a") as log_file:
        log_file.write(json.dumps(log_entry) + "\n")

# Function to handle user queries using utils.py functions
def handle_user_query(user_input, all_messages=[], debug=False):
    # Step 1: Moderation check on user input
    moderation_result = openai_moderation_check(user_input)
    if moderation_result["flagged"]:
        return "Sorry, your request is non-compliant.", all_messages
    
    # Step 2: Extract products and categories from user input
    products_and_category = get_products_and_category()
    
    # Debugging: Print the structure of products_and_category
    if debug:
        print("Products and Categories Structure:", products_and_category)
    
    product_name = None
    category_name = None

    # Flatten the product list to check for partial matches
    all_products = []
    for category, products in products_and_category.items():
        all_products.extend(products)

    # Use difflib to find the closest match to the user input
    product_matches = difflib.get_close_matches(user_input, all_products, n=1, cutoff=0.4)
    if product_matches:
        product_name = product_matches[0]
    
    # Check for category match if no product match is found
    if not product_name:
        category_matches = difflib.get_close_matches(user_input, products_and_category.keys(), n=1, cutoff=0.4)
        if category_matches:
            category_name = category_matches[0]

    # Step 3: Generate a response
    if product_name:
        product_info = get_product_by_name(product_name)
        if product_info:
            response = f"Product: {product_info['name']}\n" \
                       f"Category: {product_info['category']}\n" \
                       f"Price: ${product_info['price']}\n" \
                       f"Rating: {product_info['rating']} stars\n" \
                       f"Features: {', '.join(product_info['features'])}\n" \
                       f"Description: {product_info['description']}"
            return response, all_messages
        else:
            return "Sorry, I couldn't find the product you're asking about.", all_messages
    
    elif category_name:
        products_in_category = get_products_by_category(category_name)
        if products_in_category:
            response = f"Category: {category_name}\n"
            for product in products_in_category:
                response += f"Product: {product['name']}\nPrice: ${product['price']}\n\n"
            return response.strip(), all_messages
        else:
            return "Sorry, I couldn't find products in that category.", all_messages
    
    else:
        return "Please provide the name of a product or category you'd like to know about.", all_messages

# Chatbot logic to handle conversation
def handle_chat(user_input, history):
    response, updated_history = handle_user_query(user_input, history)
    history.append((user_input, response))  # Append the latest user input and response to the history
    log_conversation(user_input, response)  # Log the interaction
    return response, history

# Gradio chatbot UI setup
def chatbot_ui():
    with gr.Blocks() as app:
        gr.Markdown("# Store Assistant Chatbot")
        chatbot = gr.Chatbot(label="Chat with Store Assistant")
        message_input = gr.Textbox(label="Ask about products!")
        clear_btn = gr.Button("Clear Chat")

        # Initialize conversation history as a stateful variable
        conversation_history = gr.State([])

        # Process user input and update the conversation
        def on_user_message(user_message, history):
            response, updated_history = handle_chat(user_message, history)
            return "", updated_history  # Removed the extra append

        # Clear chat history
        def clear_chat():
            return [], []

        # Link input to chatbot
        message_input.submit(on_user_message, inputs=[message_input, conversation_history], outputs=[message_input, chatbot])
        clear_btn.click(clear_chat, inputs=[], outputs=[chatbot, conversation_history])

        app.launch()

# Run the chatbot interface
if __name__ == "__main__":
    chatbot_ui()