I want to extract the graph and the description of the graph from the PDF of the paper as a set.

A program that calls Scala's program from Python has been published.
It's an experimental code for the paper itself, and it says it's not practical, but it might be helpful.

allenai/deepfigures-open

I don't know if I can pull out the others as a set, but PDFminer and PyPDF2 seem to be famous.

• PDFminer
  How to extract text from PDF in Python/How to extract jpeg images from PDF in Python
  [PDFMiner] Extracting text from PDF /Reporting the results of validation of text data from PDF (pdfminer.six)/Python

• PyPDF2
  Memory to extract images from PDFs with python
  Extracting PDF Metadata and Text With Python
  How to robustly extract author names from pdf papers?

PDFminer How to extract text from PDF in Python/How to extract jpeg images from PDF in Python
[PDFMiner] Extracting text from PDF /Reporting the results of validation of text data from PDF (pdfminer.six)/Python

PyPDF2
Memory to extract images from PDFs with python
Extracting PDF Metadata and Text With Python
How to robustly extract author names from pdf papers?

Featured Python Tool.
Working with PDFs in Python:Reading and Splitting Pages

There was a Q&A article that introduced not only Python but also various tools.
Extracting information from PDFs of search papers closed

There is also a set of tools written in C.
XpdfReader