📡 Remote Jupyter#

localtileserver is usable in remote Jupyter environments such JupyterHub on services like MyBinder. Further, you may be running Jupyter in a Docker container or other host and accessing through a browser on an arbitrary client. In order to retrieve tiles into the ipyleaflet or folium Jupyter widgets client-side in the browser, we must make sure the port on which localtileserver is serving tiles is accessible to your browser.

To make this easy, we can levarage jupyter-server-proxy to expose the port on the Jupyter server through a proxy URL.

Steps to use localtileserver in remote Jupyter environments:

  1. Install jupyter-server-proxy for JupyterLab >= 3

pip install jupyter-server-proxy

  1. Set LOCALTILESERVER_CLIENT_PREFIX in your environment to 'proxy/{port}' (stop here in most cases, continue to 3. if using JupyterHub):

export LOCALTILESERVER_CLIENT_PREFIX='proxy/{port}'

  1. If using JupyterHub, you may need to alter LOCALTILESERVER_CLIENT_PREFIX such that it includes your users ID. For example, on MyBinder, we are required to do:

# Set host forwarding for MyBinder
import os
os.environ['LOCALTILESERVER_CLIENT_PREFIX'] = f"{os.environ['JUPYTERHUB_SERVICE_PREFIX']}/proxy/{{port}}"

Note

For more context, check out 📓 Jupyter in Docker