app.py

import streamlit as st
from langchain_community.document_loaders import PyPDFLoader
from langchain_core.runnables import RunnableLambda, RunnablePassthrough
from langchain_openai import OpenAIEmbeddings, ChatOpenAI, OpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_community.vectorstores import FAISS
from langchain.callbacks import StreamingStdOutCallbackHandler
from dotenv import load_dotenv
import os
load_dotenv()
MYKEY=str(os.getenv('OPENAI'))
MYKEY = os.environ.get('OPENAI')

#MYKEY=str(os.getenv('OPENAI'))
# Set page title
st.title('PDF2BrainCells 📑➡️🧠')

# Upload PDF file
uploaded_file = st.sidebar.file_uploader('Upload a PDF file', type=['pdf'])
user_api_key = st.sidebar.text_input('Enter your OpenAI API key (Optional)',type='password')
if uploaded_file:
    # Perform text processing
    temp_file_path = 'temp.pdf'
    with open(temp_file_path, 'wb') as f:
        f.write(uploaded_file.getvalue())
    loader = PyPDFLoader(temp_file_path)
    pages= loader.load_and_split()
    if user_api_key is None:
        st.toast("For continued supprt, it is preferreable to use you won API key", icon="⚠️")
        exit
    else:

        faiss_index = FAISS.from_documents(pages, OpenAIEmbeddings(openai_api_key=user_api_key if user_api_key else st.secrets["OPENAI"]))
        retriever = faiss_index.as_retriever()
        template = """Answer the question based for an examination point of view only on the following context in Markdown format. :
            {context}

            Question: {question}
            """
        prompt = ChatPromptTemplate.from_template(template)
        streamingcall=StreamingStdOutCallbackHandler()
        model = ChatOpenAI(openai_api_key=user_api_key if user_api_key else st.secrets["OPENAI"],streaming=True,callbacks=[streamingcall],callback_manager=None)
        chain = (
                {"context": retriever, "question": RunnablePassthrough()}
                | prompt
                | model
                | StrOutputParser()
            )
        # Select operation
        option = st.selectbox('Select an operation', ['Generate Questions','Conversation on Text','Content Structure','Summarization'])

        if option == 'Summarization':
            # Perform summarization
            summary=chain.invoke("Provide an executive summary of the document")
            st.write(summary)
        elif option == 'Content Structure':
            # Perform structure extraction
            structure=chain.invoke("Provide a structure of contents for this document")
            print(structure)
            st.write(structure)
        elif option == 'Generate Questions':
            # Generate questions based on content
            questions = chain.invoke("Generate questions based on the content of the document")
            st.header('Generated Questions:')
            st.markdown(questions)
        elif option == 'Conversation on Text':
            def generate_response(input_text):
                # Create a placeholder for the response
                response_placeholder = st.empty()

                # Initialize an empty string to store the response
                response = ""

                # Use a context manager to stream the response
                with st.spinner('Generating response...'):
                    # Call your existing chain to get the response stream
                    stream = chain.stream(input_text + " \n Give a detailed answer in an examination point of view")
                    # Stream the response chunk by chunk
                    for chunk in stream:
                        # Append the current chunk to the response
                        response += chunk

                        # Update the placeholder with the current response
                        response_placeholder.markdown(response)
        # Add a horizontal line after the response is complete
                #st.markdown("---")

            with st.form('my_form'):
                text = st.text_area('Ask a Question:', 'What is this Document About?')
                submitted = st.form_submit_button('Submit')
                if not user_api_key.startswith('sk-'):
                    st.toast('It is preferrable to use your own OpenAI API key for continued working of the app', icon='❓')
                if submitted:
                    generate_response(text)
                # Add a button to the sidebar