feat: verify required variables in dataset

.Rbuildignore

@@ -2,6 +2,7 @@
 ^\.Rproj\.user$
 ^LICENSE\.md$
 ^\.github$
+^data-raw$
 ^dev$
 ^CODE_OF_CONDUCT\.md$
 ^_pkgdown\.yml$

DESCRIPTION

@@ -23,20 +23,25 @@ LazyData: true
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.3.1
 Imports: 
+    checkmate,
     data.table,
+    dplyr,
     fs,
     haven,
     here,
     lifecycle,
     lubridate,
-    purrr
+    purrr,
+    rlang
 Suggests:
     furrr,
     DiagrammeR,
     knitr,
     rmarkdown,
-    spelling,
-    testthat (>= 3.0.0)
+    testthat (>= 3.0.0),
+    spelling
+Depends: 
+    R (>= 2.10)
 VignetteBuilder: knitr
 Language: en-US
 Config/testthat/edition: 3

R/get-variables.R

@@ -0,0 +1,11 @@
+
+#' Get a list of the registers' abbreviations.
+#'
+#' @return A character string.
+#' @export
+#'
+#' @examples
+#' get_register_abbrev()
+get_register_abbrev <- function() {
+  unique(required_variables$register_abbrev)
+}

R/osdc-package.R

@@ -5,3 +5,8 @@
 #' @importFrom lifecycle deprecated
 ## usethis namespace: end
 NULL
+
+# Allows for using tidyverse functionality without triggering CRAN NOTES,
+# since CRAN doesn't know that packages like dplyr use NSE.
+# For more details, see https://rlang.r-lib.org/reference/dot-data.html#where-does-data-live
+utils::globalVariables(".data")

R/verify-variables.R

@@ -0,0 +1,27 @@
+#' Verify that the dataset has the required variables for the algorithm.
+#'
+#' @param data The dataset to check.
+#' @param register The abbreviation of the register name. See list of
+#'   abbreviations in [get_register_abbrev()].
+#'
+#' @return Either TRUE if the verification passes, or a character string if
+#'   there is an error.
+#' @keywords internal
+#'
+#' @examples
+#' library(tibble)
+#' library(dplyr)
+#' # TODO: Replace with simulated data.
+#' example_bef_data <- tibble(pnr = 1, koen = 1, foed_dato = 1)
+#' verify_required_variables(example_bef_data, "bef")
+verify_required_variables <- function(data, register) {
+  checkmate::assert_choice(register, get_register_abbrev())
+  expected_variables <- required_variables |>
+    dplyr::filter(.data$register_abbrev == register) |>
+    dplyr::pull(.data$variable_name)
+  actual_variables <- names(data)
+  checkmate::check_names(
+    x = actual_variables,
+    must.include = expected_variables
+  )
+}

data-raw/variable-description.R

@@ -0,0 +1,8 @@
+## code to prepare `variable-description` dataset goes here
+
+library(tidyverse)
+
+required_variables <- read_csv(here::here("data-raw/variable_description.csv")) |>
+  select(register_abbrev = raw_register_filename, variable_name)
+
+usethis::use_data(required_variables, overwrite = TRUE, internal = TRUE)

tests/testthat/test-verify-variables.R

@@ -0,0 +1,26 @@
+library(tibble)
+
+test_that("the correct abbreviation for the register is used", {
+  bef_complete <- tibble(pnr = 1, koen = 1, foed_dato = 1)
+
+  # When incorrect register abbreviation is given
+  expect_error(verify_required_variables(bef_complete, "bef1"))
+  # When correct abbreviation is given
+  expect_true(verify_required_variables(bef_complete, "bef"))
+})
+
+test_that("the required variables are present in the dataset", {
+  # Expected
+  bef_complete <- tibble(pnr = 1, koen = 1, foed_dato = 1)
+  bef_complete_extra <- tibble(pnr = 1, koen = 1, foed_dato = 1, something = 1)
+  bef_incomplete <- tibble(pnr = 1, koen = 1)
+
+  # When all variables are the required variables
+  expect_true(verify_required_variables(bef_complete, "bef"))
+
+  # When some of the variables are the required variables
+  expect_true(verify_required_variables(bef_complete_extra, "bef"))
+
+  # When it is a character output, it is a fail
+  expect_character(verify_required_variables(bef_incomplete, "bef"))
+})