app.py

import gradio as gr
from PyPDF2 import PdfReader
import tqdm
import os 
import openai
import time
import gradio as gr
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.text_splitter import CharacterTextSplitter, RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.docstore.document import Document
from langchain.prompts import PromptTemplate
from langchain.document_loaders import TextLoader
from langchain.chains.question_answering import load_qa_chain
from langchain.llms import AzureOpenAI
from chromadb.utils import embedding_functions
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain import VectorDBQA
from langchain.llms import AzureOpenAI
import openai


os.environ["OPENAI_API_TYPE"] = openai.api_type = "azure"
os.environ["OPENAI_API_VERSION"] = openai.api_version = "2022-12-01"
os.environ["OPENAI_API_BASE"] = openai.api_base = "https://openai-endpoint.openai.azure.com/"
os.environ["OPENAI_API_KEY"] = openai.api_key = "f056ead909e54ea0a2fb570e2febad2b"


embeddings = []


def pdf_to_text(file_obj, pdf_text, vectorstore, progress = gr.Progress(track_tqdm=True)):
    reader = PdfReader(file_obj)
    number_of_pages = len(reader.pages)
    pdf_text = ""
    for page_number in range(number_of_pages):
        page = reader.pages[page_number]
        pdf_text += page.extract_text()
    
    text_splitter = RecursiveCharacterTextSplitter(        
    chunk_size = 1000,
    chunk_overlap  = 200,
    length_function = len,)
    texts = text_splitter.split_text(pdf_text)


    

    for text in tqdm.tqdm(texts):

        try:
            response = openai.Embedding.create(
            input=text,
            engine="text-embedding-ada-002")
            emb = response['data'][0]['embedding']
            embeddings.append(emb)
        except Exception as e:
            print(e)
            time.sleep(5)
            response = openai.Embedding.create(
            input=text,
            engine="text-embedding-ada-002")
            emb = response['data'][0]['embedding']
            embeddings.append(emb)

    
    azure_embeddings = OpenAIEmbeddings(document_model_name="text-embedding-ada-002",query_model_name="text-embedding-ada-002")
    vectorstore = Chroma("collection", embedding_function=azure_embeddings)
    vectorstore._collection.add(
    ids= [f"doc_{i}" for i in range(len(texts))],
    documents=texts,
    embeddings=embeddings,
    metadatas=[{"source": "source"} for text in texts]
    )

    


    return pdf_text, vectorstore

def add_text(state, query, vectorstore):
    
    # state = state + [(text, text + "?")]
    qa = VectorDBQA.from_chain_type(llm= AzureOpenAI(deployment_name="davinci003", model_name="text-davinci-003"), chain_type="stuff", vectorstore=vectorstore)
    qa = qa.run(query)
    # chain.run(input_documents=docs, question=query)
    state = state + [(query, qa)]
    return state, state, vectorstore


with gr.Blocks(title="AOAI") as demo:
    pdf_text = gr.State([])
    vectorstore = gr.State([])
    text_box = gr.TextArea()
    upload_button = gr.UploadButton("Click to Upload a File", file_types=["pdf"])
    upload_button.upload(pdf_to_text, inputs=[upload_button, pdf_text, vectorstore], outputs=[pdf_text, vectorstore])

    with gr.Row():
        chatbot = gr.Chatbot()
        state = gr.State([])
    
 
        text = gr.Textbox(show_label=False, placeholder="Enter text and press enter").style(container=False)
            
        text.submit(add_text, [state, text, vectorstore], [chatbot, state, vectorstore])