main.py

import os  
import shutil  
from flask import Flask, render_template, request, jsonify  
from llama_index.core import StorageContext, load_index_from_storage, VectorStoreIndex, SimpleDirectoryReader, ChatPromptTemplate, Settings  
from llama_index.llms.huggingface import HuggingFaceInferenceAPI  
from llama_index.embeddings.huggingface import HuggingFaceEmbedding  
from huggingface_hub import InferenceClient  
from transformers import AutoTokenizer, AutoModel


# Ensure HF_TOKEN is set  
HF_TOKEN = os.getenv("HF_TOKEN")  
if not HF_TOKEN:  
    raise ValueError("HF_TOKEN environment variable not set.")  

repo_id = "meta-llama/Meta-Llama-3-8B-Instruct"  
llm_client = InferenceClient(  
    model=repo_id,  
    token=HF_TOKEN,  
)  

# Configure Llama index settings  
Settings.llm = HuggingFaceInferenceAPI(  
    model_name=repo_id,  
    tokenizer_name=repo_id,  
    context_window=3000,  
    token=HF_TOKEN,  
    max_new_tokens=512,  
    generate_kwargs={"temperature": 0.1},  
)  
# Settings.embed_model = HuggingFaceEmbedding(  
#     model_name="BAAI/bge-small-en-v1.5"  
# )  
# Replace the embedding model with XLM-R
Settings.embed_model = HuggingFaceEmbedding(
    model_name="xlm-roberta-base"  # XLM-RoBERTa model for multilingual support
)

# Configure tokenizer and model if required
tokenizer = AutoTokenizer.from_pretrained("xlm-roberta-base")
model = AutoModel.from_pretrained("xlm-roberta-base")

PERSIST_DIR = "db"  
PDF_DIRECTORY = 'data'  

# Ensure directories exist  
os.makedirs(PDF_DIRECTORY, exist_ok=True)  
os.makedirs(PERSIST_DIR, exist_ok=True)  
chat_history = []  
current_chat_history = []  

def data_ingestion_from_directory():  
    # Clear previous data by removing the persist directory  
    if os.path.exists(PERSIST_DIR):  
        shutil.rmtree(PERSIST_DIR)  # Remove the persist directory and all its contents  
    
    # Recreate the persist directory after removal  
    os.makedirs(PERSIST_DIR, exist_ok=True)  
    
    # Load new documents from the directory  
    new_documents = SimpleDirectoryReader(PDF_DIRECTORY).load_data()  
    
    # Create a new index with the new documents  
    index = VectorStoreIndex.from_documents(new_documents)  
    
    # Persist the new index  
    index.storage_context.persist(persist_dir=PERSIST_DIR)  

def handle_query(query):  
    context_str = ""  
    
    # Build context from current chat history  
    for past_query, response in reversed(current_chat_history):  
        if past_query.strip():  
            context_str += f"User asked: '{past_query}'\nBot answered: '{response}'\n"  
    
    chat_text_qa_msgs = [
        (
            "user",
            """You are the Taj Hotel chatbot, Taj Hotel Helper.
    
            *Your Role:*
            - Respond accurately and concisely in the user's preferred language (English, Telugu, or Hindi).
            - Provide information about the hotel’s services, amenities, and policies.
    
            *Instructions:*
            - *Context:*  
              {context_str}
            - *User's Question:*  
              {query_str}
            
            *Response Guidelines:*
            1. *Language Adaptation:* Respond in the language of the question (English, Telugu, or Hindi).
            2. *Tone:* Maintain politeness, professionalism, and the luxury branding of the Taj Hotel.
            3. *Clarity:* Limit responses to 10-15 words for direct and clear communication.
            4. *Knowledge Boundaries:* If unsure of an answer, respond with:
               "I’m not sure. Please contact our staff for accurate information."
            5. *Actionable Help:* Offer suggestions or alternative steps to guide the user where applicable.
    
            *Response:* [Your concise response here]
            """.format(context_str=context_str, query_str=query)
        )
    ]


    
    text_qa_template = ChatPromptTemplate.from_messages(chat_text_qa_msgs)  
    
    storage_context = StorageContext.from_defaults(persist_dir=PERSIST_DIR)  
    index = load_index_from_storage(storage_context)  
    # context_str = ""  
    
    # # Build context from current chat history  
    # for past_query, response in reversed(current_chat_history):  
    #     if past_query.strip():  
    #         context_str += f"User asked: '{past_query}'\nBot answered: '{response}'\n"  

    query_engine = index.as_query_engine(text_qa_template=text_qa_template, context_str=context_str)  
    print(f"Querying: {query}")  
    answer = query_engine.query(query)  

    # Extracting the response  
    if hasattr(answer, 'response'):  
        response = answer.response  
    elif isinstance(answer, dict) and 'response' in answer:  
        response = answer['response']  
    else:  
        response = "I'm sorry, I couldn't find an answer to that."  

    # Append to chat history  
    current_chat_history.append((query, response))  
    return response

app = Flask(__name__)  

# Data ingestion  
data_ingestion_from_directory()  

# Generate Response  
def generate_response(query):  
    try:  
        # Call the handle_query function to get the response  
        bot_response = handle_query(query)  
        return bot_response  
    except Exception as e:  
        return f"Error fetching the response: {str(e)}"  

# Route for the homepage  
@app.route('/')  
def index():  
    return render_template('index.html')  

# Route to handle chatbot messages  
@app.route('/chat', methods=['POST'])  
def chat():  
    try:  
        user_message = request.json.get("message")  
        if not user_message:  
            return jsonify({"response": "Please say something!"})  

        bot_response = generate_response(user_message)  
        return jsonify({"response": bot_response})  
    except Exception as e:  
        return jsonify({"response": f"An error occurred: {str(e)}"})  

if __name__ == '__main__':  
    app.run(debug=True)