How to find out which jupyter notebook is using most of memory on a server

UPDATE 2022-07: see python script attached

1. find PIDs responsibles of most memory usage: ps aux --sort=-%mem | head

2. look at the name of the JSON used inside the jupyter launcher command, usually is kernel-${kernel_id}.json and get `${kernel_id}

3. go at the jupyter server endpoint /api/sessions

4. the endpoint expose a JSON with an array of structures: look for the structure containing the ${kernel_id} found on point 2. In the same structure there is the name of the corresponing notebook:

[
  ...,
  {
    ...,
    "kernel": {
      "id": ${kernel_id},
      ...
      },
    "notebook": {
      "path": ${HERE_IS_THE_PATH_TO_FILE_NAME_OF_THE_NOTEBOOK},
      ...,
      },
  ...
  }
  ...
]