import os
#from langchain import PromptTemplate, HuggingFaceHub, LLMChain
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationChain
import langchain.globals
from transformers import AutoModelForCausalLM, AutoTokenizer
import streamlit as st
from langchain_community.llms import HuggingFaceHub
from transformers import pipeline

my_model_id = os.getenv('MODEL_REPO_ID', 'Default Value')
token = os.getenv('HUGGINGFACEHUB_API_TOKEN')

@st.cache_resource 
def load_model():
    tokenizer = AutoTokenizer.from_pretrained("KvrParaskevi/Hotel-Assistant-Attempt4-Llama-2-7b")
    model = AutoModelForCausalLM.from_pretrained("KvrParaskevi/Hotel-Assistant-Attempt4-Llama-2-7b")
    
    return tokenizer,model

def demo_miny_memory(model):
    # llm_data = get_Model(hugging_face_key)
    memory = ConversationBufferMemory(llm = model,max_token_limit = 512)
    return memory

def demo_chain(input_text, memory,model):
    # llm_data = get_Model(hugging_face_key)
    llm_conversation = ConversationChain(llm=model,memory=memory,verbose=langchain.globals.get_verbose())

    chat_reply = llm_conversation.predict(input=input_text)
    return chat_reply