Evaluate compare_ard() keys and columns against both ARDs

R/compare_ard_helpers.R

@@ -8,8 +8,13 @@
 #' @keywords internal
 #' @noRd
 .process_keys_arg <- function(x, y, keys) {
-  keys_x <- cards_select({{ keys }}, data = x)
-  keys_y <- cards_select({{ keys }}, data = y)
+  # defuse the selection once, so that it is evaluated against each ARD in turn.
+  # evaluating `{{ keys }}` twice resolves the selection against `x` and then
+  # re-uses those column positions on `y`.
+  keys <- enquo(keys)
+
+  keys_x <- cards_select(expr = keys, data = x, arg_name = "keys")
+  keys_y <- cards_select(expr = keys, data = y, arg_name = "keys")
 
   .check_not_empty(keys_x)
 
@@ -36,8 +41,11 @@
 #' @keywords internal
 #' @noRd
 .process_compare_arg <- function(x, y, columns) {
-  columns_x <- cards_select({{ columns }}, data = x)
-  columns_y <- cards_select({{ columns }}, data = y)
+  # defused for the same reason as `keys` above
+  columns <- enquo(columns)
+
+  columns_x <- cards_select(expr = columns, data = x, arg_name = "columns")
+  columns_y <- cards_select(expr = columns, data = y, arg_name = "columns")
 
   .check_not_empty(columns_x)
   if (!setequal(columns_x, columns_y)) {

tests/testthat/test-compare_ard.R

@@ -463,3 +463,37 @@ test_that("compare_ard detects differences in complex stat values from ard_ident
   result <- compare_ard(ard_base, ard_modified)
   expect_false(is_ard_equal(result))
 })
+
+test_that("compare_ard() resolves keys and columns against each ARD", {
+  ard <- ard_tabulate(ADSL, variables = SEX)
+
+  # the same ARD with its columns in a different order, as it may be returned
+  # from a source other than cards
+  ard_reordered <- dplyr::relocate(ard, "stat_label", .before = 1L)
+
+  expect_silent(result <- compare_ard(ard, ard_reordered))
+
+  expect_equal(result$keys, c("variable", "variable_level", "stat_name"))
+  expect_equal(result$columns, c("stat_label", "stat"))
+  expect_true(is_ard_equal(result))
+})
+
+test_that("compare_ard() accepts selectors combined with column names", {
+  # the same summary under two names for the treatment variable
+  ard_arm <- ard_summary(ADSL, by = ARM, variables = AGE)
+  ard_trt01a <- ard_summary(ADSL, by = TRT01A, variables = AGE)
+
+  expect_silent(
+    result <-
+      compare_ard(
+        ard_arm,
+        ard_trt01a,
+        keys = c(all_ard_groups("levels"), all_ard_variables(), "stat_name"),
+        columns = c(any_of("stat_label"), "stat")
+      )
+  )
+
+  expect_equal(result$keys, c("group1_level", "variable", "stat_name"))
+  expect_equal(result$columns, c("stat_label", "stat"))
+  expect_true(is_ard_equal(result))
+})