app.py

import pandas as pd
import spacy
from difflib import SequenceMatcher
import gradio as gr
# Load your knowledge base
kb_df = pd.read_csv("clean_basic_conversations.csv", delimiter='|')

# Load spaCy model safely
try:
    nlp = spacy.load("en_core_web_sm")
except:
    from spacy.cli import download
    download("en_core_web_sm")
    nlp = spacy.load("en_core_web_sm")

from chatbot_logic import get_response, find_best_match_with_thematic_roles

# Main chat function with explanation
def chat_with_explanation(user_input, history=[]):
    if user_input.strip().lower() == "bye":
        bot_response = "Goodbye!"
        explanation = "The bot detected a farewell keyword and responded accordingly."
    else:
        bot_response, explanation = get_response(user_input, kb_df)

    history.append((user_input, bot_response))
    return history, explanation, history


# Gradio Blocks UI
def build_interface():
    with gr.Blocks(title="Context-Aware Smart Chatbot") as demo:
        gr.Markdown("""
        # 🤖 Smart Thematic Chatbot
        This chatbot uses **Thematic Role Extraction** and **String Similarity** to understand and respond.
        
        - Understands subjects, objects, verbs
        - Matches queries from a predefined knowledge base
        - Shows explanation for response selection
        """)

        chatbot = gr.Chatbot()
        with gr.Row():
            msg = gr.Textbox(label="Enter your message")
            clear = gr.Button("Clear")
        explanation = gr.Textbox(label="Explanation", lines=4)

        state = gr.State([])
        msg.submit(chat_with_explanation, [msg, state], [chatbot, explanation, state])
        clear.click(lambda: ([], "", []), None, [chatbot, explanation, state])

    return demo

# Run the app
if __name__ == "__main__":
    build_interface().launch()