ChatBotGUI.py

import gradio as gr
import os
import json
from chatbot import ChatBot
from question_answer import EvalMetrics, RelevanceScore

# Define model names
EMBEDDING_MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"
LLM_MODEL_NAME = "mistral"

# Initialize chatbot
print("[INFO] Initializing ChatBot...")
chatbot = ChatBot(LLM_MODEL_NAME, EMBEDDING_MODEL_NAME)
top_k = 10
relevance = RelevanceScore(EMBEDDING_MODEL_NAME, top_k)
metric = EvalMetrics()


# Load queries and expected answers
def load_query_dict(query_file_path):
    if not os.path.exists(query_file_path):
        print(f"[WARNING] Query file not found at {query_file_path}")
        return {}

    with open(query_file_path, "r") as file:
        data = json.load(file)
        return {
            item['question'].lower(): item['answer']
            for item in data.get('queries', [])
        }

current_workdir = os.getcwd()
query_dict = load_query_dict(os.path.join(current_workdir, "query.json"))

# Chat history storage
chat_history = []

def chat_with_bot(user_input):
    user_input_lower = user_input.lower()
    predicted, sources = chatbot.response(user_input_lower, top_k=6)

    expected = query_dict.get(user_input_lower)
    validation_feedback = ""

    if expected:
        match_result = chatbot.validate_answers(predicted, expected)
        validation_feedback = (
            f"\n\n📖 Expected: {expected}\n"
            f"🎯 Match: {'✅' if 'yes' in match_result.lower() else '❌'}"
        )
    
    source_names = [ source['source'] for source in sources ]
    source_content = [ source['page_content'] for source in sources ]

    relevant_flags = [ relevance.get_relevancy(chunk, expected) for chunk in source_content ]
    precision_at_k = relevance.precision_at_k(relevant_flags)
    ndcg_at_k = relevance.ndcg_at_k(relevant_flags)

    em = metric.compute_exact_match(predicted, expected)
    f1 = metric.compute_f1(predicted, expected)

    scores = f" Exact Match = {em}\n F1 Score = {f1}"
    relevancy = f" Precision@K = {precision_at_k}\n nDCG@K = {ndcg_at_k}"

    sources_str = ", ".join(source_names)
    response_text = f"{predicted}{validation_feedback}\n\n Relevancy:\n{relevancy} \n\n Eval Metrics:\n{scores} \n\n 📚 Sources: {sources_str}"

    chat_history.append((user_input, response_text))
    return "", chat_history

# Create Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("## 💬 Live Chatbot")

    chatbot_ui = gr.Chatbot(height=600)

    with gr.Row():
        msg_box = gr.Textbox(
            placeholder="Type your question here...", 
            show_label=False
        )
        send_button = gr.Button("Send", variant="primary")

    send_button.click(fn=chat_with_bot, inputs=[msg_box], outputs=[msg_box, chatbot_ui])
    msg_box.submit(fn=chat_with_bot, inputs=[msg_box], outputs=[msg_box, chatbot_ui])

if __name__ == "__main__":
    demo.launch(share=True)  # Set to False for private interface