split_map versions

[lopezvoliver]

There are three versions of split_map:

(1) as a method of leafmap.Map (ipyleaflet.Map)
(2) as function leafmap.split_map
(3) as a method of leafmap.foliumap.Map (folium.Map)

The leafmap.foliumap.split_map is not yet implemented (raises NotImplementedError)

The notebook 49_split_control.ipynb shows how to use (3) because ipyleaflet SplitControl does not support streamlit, and it shows how to do a splitmap using:

• basemaps
• Cloud-optimized geotiff (COG) urls
• folium.WmsTileLayer

The first two (basemap and COG urls) are easy to use, since you can pretty much just do m.split_map(left_layer, right_layer), while for the third example you need to first setup each layer by calling folium.WmsTileLayer. This is because all versions of split_map first check if they are strings (a basemap, a url, etc) or an ipyleaflet.TileLayer.

I am interested in making a split_map to compare between two local rasters, but I would like to know if there's room for improvement for streamlining the 3 versions of split_map and if there is already an effort to do so. I made some modifications to (1) that borrow code from add_raster (a call to array_to_image and get_local_tile_layer), and requires passing some keyword arguments to at least get_local_tile_layer, e.g., looks like this:

import leafmap # ipyleaflet
m = leafmap.Map(center=[37.6,-119], zoom=9)
m.split_map(dem_ds, dem_class,
            left_args={"layer_name":"DEM", "colormap": "terrain", 
                       "visible":True, 
                       "port":54605,
            },
            right_args={"layer_name":"Classified DEM","colormap": "coolwarm", 
                        "visible":True,
            },
           # left_array_args ={}   # not needed in this example but could be passed to array_to_image
           # right_array_args ={}   # not needed in this example but could be passed to array_to_image
            )
m

Similarly, I think the folium.WmsTileLayer example could be simplified by using left_args and right_args.

However, these changes would have to be repeated for all versions of split_map. Should we add these incrementally (e.g. I could make a PR for the changes to (1))?

[lopezvoliver]

I also wanted to check the reason for disabling dragging in (1) and (2)? could we add an option for this instead?

[giswqs]

These are great improvements. Please feel free to submit a PR.

Dragging is temporarily disabled due to the ipylealfet bug jupyter-widgets/ipyleaflet#1115. I don't think it has been fixed. You can give it a try by setting dragging to True

[lopezvoliver]

I didn't experience the bug, but since it's easy to set dragging=True after split_map, I didn't modify it in split_map for now.

Edit: I replaced the animation to the correct one I wanted to show.

[giswqs]

dragging is for dragging the map, not the slider. Try setting dragging=True and you may experience the bug.

[lopezvoliver]

I originally attached the wrong animation. I edited my comment above with the correct one.

However, for folium maps, I'm not sure what the expected behavior is? See below: the split slider stays in the center, and the whole map drags around it

Setting m.dragging=True makes no difference in this case.

[giswqs]

It seems the dragging issue has been resloved upstream. We can restore the dragging option now.

[giswqs]

I thought folium has fixed the issue, but it is happening again.
python-visualization/folium#1730

[lopezvoliver]

#681 Added support for numpy.ndarray and xarray.DataArray in split_map versions (1) and (3) mentioned above.