import streamlit as st
from llama_index import VectorStoreIndex, SimpleDirectoryReader
from langchain.embeddings.huggingface import HuggingFaceEmbeddings
from llama_index import LangchainEmbedding, ServiceContext
from llama_index import StorageContext, load_index_from_storage
from llama_index import LLMPredictor
#from transformers import HuggingFaceHub
from langchain import HuggingFaceHub
#from streamlit.components.v1 import html
from pathlib import Path
from time import sleep
import random
import string
import sys
import os
from dotenv import load_dotenv
load_dotenv()

st.set_page_config(page_title="Open AI Doc-Chat Assistant", layout="wide")
st.subheader("Open AI Doc-Chat Assistant: Life Enhancing with AI!")

css_file = "main.css"
with open(css_file) as f:
    st.markdown("<style>{}</style>".format(f.read()), unsafe_allow_html=True)
    
HUGGINGFACEHUB_API_TOKEN = os.getenv("HUGGINGFACEHUB_API_TOKEN")

# Load documents from a directory
documents = SimpleDirectoryReader('data').load_data()

embed_model = LangchainEmbedding(HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2'))

llm_predictor = LLMPredictor(HuggingFaceHub(repo_id="HuggingFaceH4/starchat-beta", model_kwargs={"min_length":100, "max_new_tokens":1024, "do_sample":True, "temperature":0.2,"top_k":50, "top_p":0.95, "eos_token_id":49155}))

service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, embed_model=embed_model)

def generate_random_string(length):
    letters = string.ascii_lowercase
    return ''.join(random.choice(letters) for i in range(length))  
random_string = generate_random_string(20)

new_index = VectorStoreIndex.from_documents(
    documents,
    service_context=service_context,
)

new_index.storage_context.persist("random_string")
storage_context = StorageContext.from_defaults(persist_dir="random_string")
loadedindex = load_index_from_storage(storage_context=storage_context, service_context=service_context)
query_engine = loadedindex.as_query_engine()

question = st.text_input("Enter your query here:")
display_output_text = st.checkbox("Check AI Repsonse", key="key_checkbox", help="Check me to get AI Response.")

with st.sidebar:    
    st.subheader("Valuation.pdf furnished background!")
    st.write("Disclaimer: This app is for information purpose only. NO liability could be claimed against whoever associated with this app in any manner. User should consult a qualified legal professional for legal advice.")
    st.sidebar.markdown("Contact: [aichat101@foxmail.com](mailto:aichat101@foxmail.com)")
    st.sidebar.markdown('WeChat: <span class="blue-underline">pat2win</span>, or scan the code below.', unsafe_allow_html=True)
    st.image(wechat_image)
    st.sidebar.markdown('<span class="blue-underline">Life Enhancing with AI.</span>', unsafe_allow_html=True)      
    st.subheader("Enjoy chatting!")

if question !="" and not question.strip().isspace() and not question == "" and not question.strip() == "" and not question.isspace():
    if display_output_text==True:
      with st.spinner("AI Thinking...Please wait a while to Cheers!"):
        initial_response = query_engine.query(question)
        temp_ai_response=str(initial_response)
        final_ai_response=temp_ai_response.partition('<|end|>')[0]
        st.write("AI Response:\n\n"+final_ai_response)
    else:
        print("Check the Checkbox to get AI Response.")
        sys.exit()          
else:
    print("Please enter your question first.")
    st.stop()