import streamlit as st
import os
from langchain_groq import ChatGroq
from langchain.embeddings import HuggingFaceEmbeddings
from langchain_community.embeddings.yandex import YandexGPTEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain_core.prompts import ChatPromptTemplate
from langchain.chains import create_retrieval_chain
from langchain.vectorstores import FAISS
from langchain_community.document_loaders import PyPDFLoader

from dotenv import load_dotenv
load_dotenv()

#Load groq API Keys.
groq_api_key = os.getenv("groq_api_key")

st.title("ChatGroq with LLAMA3 Demo :sparkles:")

llm = ChatGroq(groq_api_key=groq_api_key ,model_name = "llama3-8b-8192")

prompt = ChatPromptTemplate.from_template(
"""
Answer the questions based on provided context only.
Please provide Accurate response based on question and explain it widely.
<context>
{context}
<context>
Question : {input}
"""
)


def vector_embeddings():
    if "vectors" not in st.session_state:
        st.session_state.embeddings = HuggingFaceEmbeddings()
        st.session_state.loader = PyPDFLoader("us_census.pdf") # Data Injection
        st.session_state.docs = st.session_state.loader.load() # Document Loading
        st.session_state.text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000,chunk_overlap=200) # Chunk Creation
        st.session_state.final_documents = st.session_state.text_splitter.split_documents(st.session_state.docs[:20]) # Document splitting
        st.session_state.vectors = FAISS.from_documents(st.session_state.final_documents,st.session_state.embeddings) # vector Huggingface Embedding

prompt1 = st.text_input("Enter the Question from your Mind : ")


if st.button("Document Embeddings"):
    vector_embeddings()
    st.write("Vector store DB is Ready.")

if prompt1:
    document_chain = create_stuff_documents_chain(llm,prompt)
    retriver = st.session_state.vectors.as_retriever()
    retrival_chain = create_retrieval_chain(retriver,document_chain)

    response = retrival_chain.invoke({'input':prompt1})
    st.write(response['answer'])