add_raster doesn't work with ssh remote environment

[zxdawn]

As discussed here, the localtileserver needs some special settings to make add_raster function in leafmap work with SSH remote environment. I will update the answer when we get the best answer.

[zxdawn]

If you are running leafmap remotely, you need to install jupyter-server-proxy and add export LOCALTILESERVER_CLIENT_PREFIX='proxy/{port}' to your remote environment, then add_raster will work well.

[giswqs]

Thank you for sharing the solution.

[zxdawn]

@giswqs You're welcome. Actually I realize this method works in Remote Jupyter Notebook, but not for Remote Streamlit ...

import os
os.environ['LOCALTILESERVER_CLIENT_PREFIX'] = 'proxy/{port}/'

import leafmap.foliumap as leafmap
import streamlit as st

url = 'https://github.com/opengeos/data/raw/main/raster/srtm90.tif'

image = 'srtm90.tif'
if not os.path.exists(image):
    leafmap.download_file(url, image)

filename = 'srtm90.tif'

# plot the basemap
m = leafmap.Map()
m.add_basemap('Esri.WorldImagery')

m.add_raster(filename, band=1, palette='plasma', layer_name='test')

m.to_streamlit()

[zxdawn]

This line os.environ['LOCALTILESERVER_CLIENT_PREFIX'] = 'proxy/{port}/' will cause the issue even with local streamlit. I suppose the port of streamlit is special.

[giswqs]

Streamlit does not support localtileserver. Don't wait your time on this. See streamlit/streamlit#4677

The workaround to to make the raster available through HTTP, then use m.add_cog_layer() to display it.

[giswqs]

It will only work if you use it locally. If you deploy it to the Streamlt Cloud, it will not work.

[zxdawn]

Got it. Actually, I only want to run it locally on my Linux machine and connect to it using SSH from laptop (because the data is private and large). It seems tricky ...

[giswqs]

Try Solara: https://giswqs-solara-geospatial.hf.space/leafmap

[zxdawn]

Thanks. I will try that instead of streamlit!

[rh-afk]

Hi @giswqs , I just try your solara geospatial, unfortunately there was an error:

I have restart the kernel, shutdown the kernel or even with shutdown server in jupyter hub and up again the server. However there still have error. did you have any suggestion.
If run individual leaflet(without solara) there is no issue