test_scaling.jl

using NCDatasets
using Test

filename = tempname()
# The mode "c" stands for creating a new file (clobber)
ds = NCDatasets.NCDataset(filename,"c")

# define the dimension "lon" and "lat" with the size 10 and 11 resp.
NCDatasets.defDim(ds,"lon",10)
NCDatasets.defDim(ds,"lat",11)

for T in [Int32,Float32]

    local v, data

    offset = 20.
    factor = 0.1

    v = NCDatasets.defVar(ds,"scaled_var_$(T)",T,("lon","lat"), attrib = [
        "add_offset" => offset,
        "scale_factor" => factor,
    ])

    data = [-12.3*i + 23.4*j for i = 1:10, j = 1:11]

    v[:,:] = data
    @test v[:,:] ≈ data atol=1e-4

    # load without transformation (offset/scaling)
    @test v.var[:,:] ≈ (data .- offset)/factor atol=1e-4

    # write/read without transformation (offset/scaling)
    v.var[:,:] = data
    if eltype(v.var) <: Integer
        @test v.var[:,:] == round.(Int,data)
    else
        @test v.var[:,:] ≈ data
    end
end

NCDatasets.close(ds)
rm(filename)