app.py

import openai
import gradio as gr
from langchain.chains import RetrievalQA
from langchain.llms import OpenAI
from langchain.document_loaders import PyPDFLoader
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import FAISS
from langchain.chat_models import ChatOpenAI
from PyPDF2 import PdfReader



#Function to load and process the PDF document
def load_pdf(file):
    #Load the PDF usign Langchain's PyPDFLoader
    loader=PyPDFLoader(file.name)
    documents=loader.load()
    return documents

# Summarization function using GPT-4
def summarize_pdf(file,openai_api_key):
    #set the openAI API key dynamically
    openai.api_key="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"

    # Load and process the PDF
    documents=load_pdf(file)


    # Create embeddings for the documents
    embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

    # Use Langchain's FAISS Vector Store to store and search the embeddings
    vector_store=FAISS.from_documents(documents,embeddings)

    # Create a RetrievalQA chain for summarization
    llm = ChatOpenAI(model='gpt-40', openai_api_key=openai_api_key)   #passing api key here
    qa_chain=RetrivalQA.from_chain_type(
        llm=llm,
        chain_type="stuff",
        retriever=vector_store.as_retriever()
    )


    # Query the model for a summary of the document
    response = qa_chain.run("Summarize the content of the research paper.")
    return response


#Function to handle user queries and provide answers from the document
def query_pdf(file,user_query,openai_api_key):
    #set the openai api key dynamically
    openai.api_key=openai_api_key

    #Load and process the PDF
    documents = load_pdf(file)

    # Create embeddings for the documents
    embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

    # Use Langchain's FAISS vector store to store and search the embeddings
    vector_store = FAISS.from_documents(documents, embeddings)

    # Create a RetrievalQA chain for querying the document
    llm=ChatOpenAI(model="gpt-40", openai_api_key=openai_api_key)  #passing api key here
    qa_chain=RetrivalQA.from_chain_type(
        llm=llm,
        chain_type="stuff",
        retriever=vector_store.as_retriever()
    )

    # Query the model for the user query
    response = qa_chain.run(user_query)
    return response

    # Define Gradio interface for the summarization
    def create_gradio_interface():
        with gr.Blocks() as demo:
            gr.Markdown("### ChatPDF and Research Paper Summarizer using GPT-4 and Langchain ")

            # Input field for API key
            with gr.Row():
                openai_api_key_input=gr.Textbox(label="Enter OpenAI API key",type ="password",placeholder="Enter your openai api key here")

                with gr.Tab("Summarize PDF"):
                    with gr.Row():
                        pdf_file = gr.file(label="Upload PDF Document")
                        summarize_btn=gr.Button("Summarize")
                        summary_output=gr.Textbox(label="Summary",interactive=False)
                        clear_btn_summary=gr.Button("Clear Response")


                        #Summarize Button Logic
                        summarize_btn.click(summarize_pdf,inputs=[pdf_file,openai_api_key_input],outputs=summary_output)

                        # Clear response Button Logic for summary Tab
                        clear_btn_summary.click(lambda:"",inputs=[],outputs=summary_output)

                with gr.Tab("Ask Questions"):
                    with gr.Row():
                        pdf_file_q = gr.File(label="Upload PDF Document")
                        user_input = gr.Textbox(label="Enter your question")
                        answer_output = gr.Textbox(label="Answer",interactive=False)
                        clear_btn_answer = gr.Button("clear Response")

                    # Submit Question Logic
                    user_input.submit(query_pdf,inputs=[pdf_file_q,user_input,openai_api_key_input],outputs=answer_output)


                    # Clear response button logic for anser tab
                    clear_btn_answer.click(lambda:"",inputs=[],outputs=answer_output)

                    user_input.submit(None,None,answer_output)
    return demo

# Run Gradio app
if __name__=="__main__":
    demo = create_gradio_interface()
    demo.launch(debug=True)