Add function to save to geojson file

[mem48]

Unless I've missed it, why there is no easy way to save an sf object to disk as a geojson file?

I created a little test function

write_geojson <- function(x, path){
  x <- geojsonsf::sf_geojson(x)
  filecon <- file(path)
  writeLines(x, filecon)
  close(filecon)
}

And it was 7x faster than sf::st_write. The output wasn't identical as sf writes each feature to a new line in the file. But I was able to st_read it back in and it was identical to the original sf data frame.

I'm sure this function needs some refining, but it would be a really useful feature for geojsonsf

Thanks.

[dcooley]

why there is no easy way to save an sf object to disk as geojson file?

sf::st_write() is the easy way to save an sf object to disk as geojosn

---

If this was to become part of {geojosnsf} it would need to be implemented in C++, basically the opposite of read_geojson.cpp, using FileWriteStream

I don't have any time to implement this, but woud accept a PR if you're willing to write it.

[mem48]

My C++ is not great so I probably couldn't contribute in that way.

But in case anybody else comes across this issue; the problem with sf::st_write seems to be converting the sf object to geojson rather than the writing to disk. So sf_geojson provides a worthwhile speed boost on large datasets.

The best I could come up with is:

write_geojson <- function(x, path){
  headder <- paste0('{\n"type":"FeatureCollection",\n"name": "',deparse(substitute(x)),'",')
  if(sf::st_is_longlat(x)){
    headder <- paste0(headder,'"crs": { "type": "name", "properties": { "name": "urn:ogc:def:crs:OGC:1.3:CRS84" } },\n')
  }
  headder <- paste0(headder,'"features":[')
  footer <- ']\n}'
  commas <- c(rep(',', nrow(x) - 1),'')
  x <- c(headder, 
         paste0(geojsonsf::sf_geojson(x, atomise = TRUE),commas),
         footer)
  data.table::fwrite(list(x), path, quote  = FALSE)
}

This writes geojson that is identical to st_write at least for my test cases and saved hours on some large datasets I was working on. It uses data.table for writing to disk, but that could probably be replaced with a base R solution as the main benefit is from sf_geojson.