Problems to read pdfs and get answers

Hi, I have a problem. In previous months I developed a script in python to read pdfs and get answers with Openai, langchain, gpt-index(llama-index), and fastapi libraries.
Currently, I notice that the chatbot functionality is deprecated by the davinci model 'model_not_found error_message=‘The model text-davinci-003 has been deprecated’. I changed the model by the following gpt-3.5-turbo-instruct, but started to show me an error with langchain library ‘Warning: Importing LLMs from langchain is deprecated. Importing from langchain will no longer be supported as of langchain==0.2.0.’.

I would like to know if it has a new way to implement this functionality without langchain library or if it is necessary to restructure all the scripts.

Thank you for your time

I would take a look a the OpenAI cookbooks to see if they have sample code similar to what you seek.