Add support for ArcGIS services using {arcgislayers}

[elipousson]

I put together a couple of prototypes based on tarchetypes::tar_file_read() to support working with ArcGIS services using {arcgislayers}.

tar_arc_select() may be the better of the two since it could potentially use the service metadata (returned by arcgislayers::arc_open()) to determine when the target is stale and needs to be re-run. I'm new to developing targets functions so feedback on improvements are welcome!

tar_arc_select <- function(name,
                           url,
                           ...,
                           token = arcgisutils::arc_token(),
                           tidy_eval = targets::tar_option_get("tidy_eval"),
                           packages = targets::tar_option_get("packages"),
                           library = targets::tar_option_get("library"),
                           format = targets::tar_option_get("format"),
                           repository = targets::tar_option_get("repository"),
                           error = targets::tar_option_get("error"),
                           memory = targets::tar_option_get("memory"),
                           garbage_collection = targets::tar_option_get("garbage_collection"),
                           deployment = targets::tar_option_get("deployment"),
                           priority = targets::tar_option_get("priority"),
                           resources = targets::tar_option_get("resources"),
                           storage = targets::tar_option_get("storage"),
                           retrieval = targets::tar_option_get("retrieval"),
                           cue = targets::tar_option_get("cue")) {
  name <- targets::tar_deparse_language(substitute(name))

  envir <- targets::tar_option_get("envir")

  name_url <- paste0(name, "_url")

  name_layer <- paste0(name, "_layer")

  sym_url <- rlang::sym(name_url)

  url <- targets::tar_tidy_eval(
    expr = as.expression(substitute(url)),
    envir = targets::tar_option_get("envir"), tidy_eval = tidy_eval
  )

  layer <- targets::tar_tidy_eval(
    expr = as.expression(substitute(arcgislayers::arc_open(url = !!.x, token = token))),
    envir = list(.x = sym_url), tidy_eval = TRUE
  )

  select <- targets::tar_tidy_eval(
    expr = as.expression(substitute(arcgislayers::arc_select(x = !!.x, ..., token = token))),
    envir = list(.x = layer), tidy_eval = TRUE
  )

  list(
    targets::tar_target_raw(
      name = name_url,
      command = url,
      packages = packages, library = library, format = format,
      repository = repository, error = error, memory = memory,
      garbage_collection = garbage_collection, deployment = deployment,
      priority = priority, storage = storage, retrieval = retrieval,
      cue = cue
    ),
    targets::tar_target_raw(
      name = name_layer,
      command = layer,
      packages = packages, library = library, format = format,
      repository = repository, error = error, memory = memory,
      garbage_collection = garbage_collection, deployment = deployment,
      priority = priority, storage = storage, retrieval = retrieval,
      cue = cue
    ),
    targets::tar_target_raw(
      name = name,
      command = select,
      packages = packages, library = library,
      format = format, repository = repository, error = error,
      memory = memory, garbage_collection = garbage_collection,
      deployment = deployment, priority = priority, resources = resources,
      storage = storage, retrieval = retrieval, cue = cue
    )
  )
}

tar_arc_read <- function(name,
                         url,
                         ...,
                         token = arcgisutils::arc_token(),
                         tidy_eval = targets::tar_option_get("tidy_eval"),
                         packages = targets::tar_option_get("packages"),
                         library = targets::tar_option_get("library"),
                         format = targets::tar_option_get("format"),
                         repository = targets::tar_option_get("repository"),
                         error = targets::tar_option_get("error"),
                         memory = targets::tar_option_get("memory"),
                         garbage_collection = targets::tar_option_get("garbage_collection"),
                         deployment = targets::tar_option_get("deployment"),
                         priority = targets::tar_option_get("priority"),
                         resources = targets::tar_option_get("resources"),
                         storage = targets::tar_option_get("storage"),
                         retrieval = targets::tar_option_get("retrieval"),
                         cue = targets::tar_option_get("cue")) {
  name <- targets::tar_deparse_language(substitute(name))

  envir <- targets::tar_option_get("envir")

  name_url <- paste0(name, "_url")

  sym_url <- rlang::sym(name_url)

  url <- targets::tar_tidy_eval(
    expr = as.expression(substitute(url)),
    envir = targets::tar_option_get("envir"), tidy_eval = tidy_eval
  )

  res <- targets::tar_tidy_eval(
    expr = as.expression(substitute(arcgislayers::arc_read(url = !!.x, ..., token = token))),
    envir = list(.x = sym_url), tidy_eval = TRUE
  )

  list(
    targets::tar_target_raw(
      name = name_url,
      command = url,
      packages = packages, library = library, format = format,
      repository = repository, error = error, memory = memory,
      garbage_collection = garbage_collection, deployment = deployment,
      priority = priority, storage = storage, retrieval = retrieval,
      cue = cue
    ),
    targets::tar_target_raw(
      name = name,
      command = res,
      packages = packages, library = library,
      format = format, repository = repository, error = error,
      memory = memory, garbage_collection = garbage_collection,
      deployment = deployment, priority = priority, resources = resources,
      storage = storage, retrieval = retrieval, cue = cue
    )
  )
}

[Aariq]

This is a great idea and I think geotargets would be a nice place for this target factory. However, I don't currently have ArcGIS installed, so I suspect that's going to make it difficult to test and explore. I think I have access through my institution, so will definitely keep this in mind for the future.

[elipousson]

Great. Let me know if you have a chance to try it out and I'll share more here if I have more ideas for the implementation.

Since you mentioned needing to have ArcGIS installed, I wanted to clarify that you may be thinking of the arcgisbinding package that requires a local ArcGIS Pro installation. The arcgislayers package allows access to the ArcGIS REST API for both free and paid web services (mostly FeatureServers but also ImageServers, Tables, etc.).

[Aariq]

I was thinking of the arcgisbinding package. Thanks for the clarification!

[njtierney]

Just wanted to echo what @Aariq said - geotargets would be a nice place for this.

@elipousson do you happen to have any examples on hand to demonstrate the use of this target factory that don't require ArcGIS to be installed?

[elipousson]

I can put together a standalone repo with a reprex later this week but I'm using {arcgislayers} quite a bit in this basemap plotting project I put together earlier this year. The helper function doesn't take advantage of targets in the way that tar_arc_read() does but you can still see it here: https://github.com/elipousson/tigris-basemap/blob/main/R/functions.R#L36-L74