How to create a valid `FeatureCollection` I can interface with `GeoJSON`?

Specific Domains Geo
1 
using GeoJSON

EPS = 5
MAD_BOUNDS = (40.2257307, 40.6072783, -3.9536568, -3.4525022)
grid = Dict("Lat" => range(MAD_BOUNDS[1], MAD_BOUNDS[2], EPS),
            "Lng" => range(MAD_BOUNDS[3], MAD_BOUNDS[4], EPS),
            "Eps" => EPS)

fc = Dict("type" => "FeatureCollection",
          "features" => [],
          "properties" => Dict{String, Any}())

tile_features = [Dict("type" => "Feature",
                      "properties" => Dict{String, Any}(),
                      "geometry" => Dict("type" => "Polygon",
                                         "coordinates" => [
                                            [
                                                [grid["Lng"][i], grid["Lat"][j]],
                                                [grid["Lng"][i+1], grid["Lat"][j+1]],
                                            ]
                                         ])
                      )
                 for i in 1:grid["Eps"]-1, j in 1:grid["Eps"]-1]

fc["features"] = tile_features

GeoJSON.parseFeature(fc["features"][1]) # works
geo_data = GeoJSON.parseFeatureCollection(fc)

Last line throws a MethodError on FeatureCollection(::Matrix{Feature})
I think it’s getting hung up on type of the coordinates
Any clues?

2

Well, it turns out I had a couple of mistakes. First and most importantly I wasn’t really defining a closed form polygon (make sure to follow the right-hand rule, most usually)


tile_features = [Dict("type" => "Feature",
                      "properties" => Dict{String, Any}(),
                      "geometry" => Dict("type" => "Polygon",
                                         "coordinates" => [
                                            [
                                                [grid["Lng"][i], grid["Lat"][j]],
                                                [grid["Lng"][i], grid["Lat"][j+1]],
                                                [grid["Lng"][i+1], grid["Lat"][j+1]],
                                                [grid["Lng"][i+1], grid["Lat"][j]],
                                                [grid["Lng"][i], grid["Lat"][j]]
                                            ]
                                         ])
                      )
                 for i in 1:grid["Eps"]-1, j in 1:grid["Eps"]-1]

And then I was able to turn it into a FeatureCollection object more properly by first

open("mad_grid.geojson", "w") do io
    JSON3.pretty(io, fc)
end

And reading it back on memory

jsonbytes = read("mad_grid.geojson")
fc = GeoJSON.read(jsonbytes)

Seems like a roundabout way of doing things, so won’t mark as a solution just in case someone has a more direct way of achieving this :smile:

3

What version of GeoJSON.jl are you using?

4

Hi Raf, I’m using v0.5.1

5

Ok. Much of this syntax has changed with v0.6, with GeoInterface.jl v1.0 and using JSON3.jl underneath rather than JSON.jl.

6

Seems like Turf.jl is restricting to me to v0.5.1
Everything is well for now, though. Will give a shout out if not :smile:

7

Ok seems you might be staying there! turf is 2 years without change.

It would be good to get Turf.jl moved to JuilaGeo so it gets updated with these changes, I didn’t even know it existed.

8

Oh I’m not married to it, but given that I don’t really know the ecosystem I chose it because it’s 100% Julia version a JS library by the same name (might as kill two birds with one stone type of thing… if I ever need the JS version :smile: ), its listed here: https://juliageo.org/, plus I thought it might be well tested.

I just need to do some spatial queries, like getting the area of the tiles above etc… I expect my FeatureCollection to get big so the above is not very efficient. What do you suggest?