Automate numberType in get_ncdf4_attributes

[maier-m]

Seems to me to be a possibly useful idea to automate the generation of eml defined numberType from netCDF files by adding

values <- ncdf4::ncvar_get(nc, attributes[i])
result[i, 'eml_numberType'] <- get_numberType(values)

between lines 41 and 42 in https://github.com/NCEAS/arcticdatautils/blob/master/R/attributes.R
Where:

get_numberType <- function(values){
  numberType=NA
  if (is.numeric(values)){
    if (any(values%%1!=0,na.rm = T)){numberType='real'}
    else{
      if (any(values<0,na.rm = T)){numberType='integer'}
      else{
        if (any(values==0,na.rm = T)){numberType='whole'}
        else{
          numberType='natural'
        }
      }
    }
  }
  return(numberType)
}

Small but possibly helpful especially when having to go through multiple netCDF files with multiple variables. (although as written would overwrite any variables in the netCDF file named 'eml_numberType')

[jeanetteclark]

Only issue I see here is that if the netCDF is large, this could be very, very slow. You might want to run some tests on this first

[maier-m]

Yes, jeanette. I noticed this problem, it was slow with how I was determining real numbers... much faster function below (along with other fixes)

get_numberType <- function(values){
  values <- unlist(values)
  numberType=NA
  if (is.numeric(values)){
    if (all(is.nan(values))){numberType='real'}
    else{
      if (any(round(values)!=values)){numberType='real'}
      else{
        if (any(values<0,na.rm = T)){numberType='integer'}
        else{
          if (any(values==0,na.rm = T)){numberType='whole'}
          else{
            numberType='natural'
          }
        }
      }
    }
  }
  return(numberType)
}