get_spd(): get spd archived files

DESCRIPTION

@@ -22,6 +22,7 @@ Imports:
     magrittr,
     readr,
     rlang,
+    stringr,
     tibble
 Suggests: 
     testthat (>= 3.0.0)

R/find_latest_file.R

@@ -44,7 +44,7 @@ find_latest_file <- function(directory,
       path = directory,
       type = "file",
       regexp = regexp,
-      recurse = TRUE
+      recurse = FALSE
     ) |>
       dplyr::arrange(
         dplyr::desc(.data$path),
@@ -55,11 +55,15 @@ find_latest_file <- function(directory,
   }
 
   if (nrow(latest_file) == 1L) {
-    cli::cli_alert_info("Using {.val {fs::path_file(latest_file$path)}}.")
+    cli::cli_alert_info(
+      "Using the latest available version: {.val {fs::path_file(
+       fs::path_ext_remove(latest_file$path))}}.
+       If you require an older version specify an argument `version`."
+    )
   } else {
     cli::cli_abort(
       "There was no file in {.path {directory}} that matched the
-        regular expression {.val {regexp}}"
+       regular expression {.val {regexp}}"
     )
   }
 

R/get_spd.R

@@ -1,14 +1,20 @@
 #' Get the Scottish Postcode Directory
 #'
-#' @param version Default is "latest", otherwise supply a tag e.g. "2023_2"
-#' @inheritParams arrow::read_parquet
+#' Read a Scottish Postcode Directory (SPD) lookup file into a tibble.
+#' @param version A string defining a version to read in. The default value
+#'  is "latest" and the latest SPD file available on cl-out will be loaded.
+#'  Alternatively you can supply a tag, e.g. "2023_2", to load a specific file.
+#' @param col_select A character vector of column names to keep, as in
+#'  the "select" argument to data.table::fread(), or a tidy selection
+#'  specification of columns, as used in dplyr::select().
 #'
-#' @return a [tibble][tibble::tibble-package] of the Scottish Postcode Directory
+#' @return A [tibble][tibble::tibble-package] of the Scottish Postcode
+#'  Directory lookup file or its selected columns.
 #' @export
 #'
 #' @examples
 #' get_spd()
-#' get_spd(col_select = c("pc7", "latitude", "longitude"))
+#' get_spd(version = "2023_2", col_select = c("pc7", "latitude", "longitude"))
 get_spd <- function(
     version = "latest",
     col_select = NULL) {
@@ -21,9 +27,29 @@ get_spd <- function(
       selection_method = "file_name"
     )
   } else {
-    spd_path <- fs::path(
-      dir,
-      glue::glue("Scottish_Postcode_Directory_{version}.parquet")
+    if (!stringr::str_detect(version, "^20\\d{2}_[1-2]$")) {
+      cli::cli_abort(c(
+        "x" = "Invalid version name: {.val {version}}",
+        "i" = "It should follow pattern YYYY_1 or YYYY_2",
+        call = NULL
+      ))
+    }
+
+    name_ver_list <- paste0(
+      "Scottish_Postcode_Directory_", version,
+      c(".parquet", ".rds", ".csv")
+    )
+    path_ver_list <- fs::path(dir, "Archive", name_ver_list)
+    spd_path <- path_ver_list[file.exists(path_ver_list)][1]
+  }
+
+  if (is.na(spd_path)) {
+    cli::cli_abort(
+      c(
+        "x" = "SPD version {.val {version}} is NOT available",
+        "i" = "Contact phs.geography@phs.scot"
+      ),
+      call = NULL, rlang_backtrace_on_error = "none"
     )
   }
 

R/read_file.R

@@ -29,12 +29,21 @@ read_file <- function(path, col_select = NULL, ...) {
     ))
   }
 
+  if (!(file.exists(path))) {
+    cli::cli_abort(
+      c(
+        "x" = "File {.val {fs::path_file(fs::path_ext_remove(path))}}
+             is NOT available",
+        "i" = "Contact phs.geography@phs.scot"
+      ),
+      call = NULL, rlang_backtrace_on_error = "none"
+    )
+  }
+
   data <- switch(ext,
     "rds" = readr::read_rds(file = path),
     "csv" = readr::read_csv(
-      file = path,
-      ...,
-      col_select = {{ col_select }},
+      file = path, guess_max = 50000, ...,
       show_col_types = FALSE
     ),
     "parquet" = tibble::as_tibble(arrow::read_parquet(

tests/testthat/test-find_latest_file.R

@@ -15,13 +15,13 @@ test_that("By default will find the last created file", {
     find_latest_file(dir, "[a-z]\\d.txt"),
     fs::path(dir, "b0.txt")
   ) |>
-    expect_message("\"b0.txt\"")
+    expect_message("\"b0\"")
 })
 
 test_that("Can find latest file by file name", {
   expect_equal(
     find_latest_file(dir, "[a-z]\\d.txt", "file_name"),
     fs::path(dir, "z9.txt")
   ) |>
-    expect_message("\"z9.txt\"")
+    expect_message("\"z9\"")
 })