app.py

import pandas as pd
import streamlit as st
import google.generativeai as palm
import pandas as pd
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.embeddings import GooglePalmEmbeddings
from langchain.llms import GooglePalm
from langchain.document_loaders import PyPDFLoader,DirectoryLoader
# from langchain.llms import CTransformers
from langchain.vectorstores import FAISS
from langchain.chains import RetrievalQA,ConversationalRetrievalChain
from langchain.prompts import PromptTemplate
from langchain.cache import InMemoryCache
from langchain.llms import VLLM
from langchain.memory.buffer import ConversationBufferMemory
from langchain.chains.conversation.memory import ConversationSummaryBufferMemory
import gradio as gr
import requests
import os
from langchain.embeddings import HuggingFaceBgeEmbeddings

# models = [m for m in palm.list_models() if "generateText" in m.supported_generation_methods]
# model = models[0].name
# print('Imports Done')


db_path = './vectordb/db_faiss'

# print('Reading Document')
# os.mkdir('/home/Sparsh/data')
# url = 'https://ia803106.us.archive.org/13/items/Encyclopedia_Of_Agriculture_And_Food_Systems/Encyclopedia%20of%20Agriculture%20and%20Food%20Systems.pdf'
# response = requests.get(url)
# with open('/home/Sparsh/data/document.pdf', 'wb') as f:
#          f.write(response.content)
#
# print('Creating Chunks')


# loader = DirectoryLoader('C:/Users/HP/PycharmProjects/MLSCBot/venv/MLSCBot',glob = "*.pdf",loader_cls = PyPDFLoader)
# data = loader.load()
# splitter = RecursiveCharacterTextSplitter(chunk_size = 500,chunk_overlap = 100)
# chunks = splitter.split_documents(data)
#
# print('Mapping Embeddings')
model_name = "BAAI/bge-base-en"
encode_kwargs = {'normalize_embeddings': True} # set True to compute cosine similarity

model_norm = HuggingFaceBgeEmbeddings(model_name=model_name,
    encode_kwargs=encode_kwargs)
embeddings = model_norm
# # db = FAISS.from_documents(chunks,embeddings)
# # db.save_local(db_path)
db = FAISS.load_local(db_path,embeddings)
# print('Prompt Chain')


custom_prompt_template = """You are a helpful NGO bot working for Join The Dots foundation designd to answer the questions of the basic CS fundamentals, your task is to accurately answer the students and resolve their query using any information that you have if you think it is correct
Use the following context (delimited by <ctx></ctx>) and the chat history (delimited by <hs></hs>) to answer the question:
------
<ctx>
{context}
</ctx>
------
<hs>
{history}
</hs>
------
{question}

Only return the helpful answer and no extra information.Mention your name and what do now every message
Helpful answer:
"""

prompt = PromptTemplate(template=custom_prompt_template,
                            input_variables=['history','context', 'question'])

# memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)



# print('Creating LLM')

llm2 = GooglePalm(
           max_new_tokens=1024,
           top_k=10,
           top_p=0.5,
           temperature=0.5)

print(llm2("What is the capital of France ?"))
# qa_chain = ConversationalRetrievalChain.from_llm(llm2,retriever=db.as_retriever(search_kwargs={'k': 2}),
#                                        return_source_documents=False,
#                                        memory=memory)
qa_chain = RetrievalQA.from_chain_type(llm=llm2,
                                      chain_type='stuff',
                                      retriever=db.as_retriever(search_kwargs={'k': 1}),
                                      return_source_documents=False,
                                      chain_type_kwargs={'prompt': prompt
                                                         ,'memory' : ConversationBufferMemory(
                                              memory_key="history",
                                          input_key = "question")})
history_df = pd.DataFrame(columns = ['Question','Answer'])
def qa_bot(query):
  response = qa_chain({'query': query})
  print(response)
  return (response['result'])

st.markdown("<h1 style='text-align: center; color: white; font: Charcuterie'>JTDoubts</h1>", unsafe_allow_html=True)
hide_streamlit_style = """
            <style>
            #MainMenu {visibility: hidden;}
            header {visibility: hidden;}
            footer {visibility: hidden;}
            </style>
            """
st.markdown(hide_streamlit_style, unsafe_allow_html=True)
# st.image('./banner.png',use_column_width=True)
if "messages" not in st.session_state:
    st.session_state.messages = []

for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

if prompt := st.chat_input("Hello!How can I help you?"):
    st.session_state.messages.append({"role": "user", "content": prompt})
    with st.chat_message("user"):
        st.markdown(prompt)
    with st.chat_message("assistant"):
        message_placeholder = st.empty()
        full_response = qa_bot(prompt)
        message_placeholder.markdown(full_response + "▌")
        message_placeholder.markdown(full_response)
    st.session_state.messages.append({"role": "assistant", "content": full_response})
# with gr.Blocks(theme='upsatwal/mlsc_tiet') as demo:
#     title = gr.HTML("<h1>MLSCBot</h1>")
#     with gr.Row():
#       img = gr.Image('C:/Users/HP/Downlo0ads/banner.png',label = 'MLSC Logo',show_label = False,elem_id = 'image',height = 200)
#     input = gr.Textbox(label="How can I assist you?")  # Textbox for user input
#     output = gr.Textbox(label="Here you go:")  # Textbox for chatbot response
#     btn = gr.Button(value="Answer",elem_classes="button-chatbot",variant = "primary")  # Button to trigger the agent call
#     btn.click(fn=qa_bot, inputs=input,outputs=output)
# demo.launch(share=True, debug=True,show_api = False,show_error = False)