Use CodeAndValue when overlaps are detected in any columns

Fixes #24

Author: aleksanderbl29

R/dst_determine_overlaps.R

@@ -0,0 +1,26 @@
+#' Helper function to determine wether or not to include the id in a variable
+#' option
+#'
+#' @param meta_data Meta data object for the table of inquiry
+#' @noRd
+dst_determine_overlaps <- function(meta_data) {
+  # Get variable names
+  var_names <- get_vars(meta_data)
+
+  # Get options for all variable names
+  options <- get_var_options(meta_data, var_names)
+
+  # Index over all vars to determine if there is duplicates
+  dup <- list()
+
+  for (i in seq_along(var_names)) {
+    dup[i] <- length(
+      options[[var_names[i]]]
+    ) == length(
+      unique(options[[var_names[i]]])
+    )
+  }
+
+  # If any of the option/vars include duplicates, we should include the id
+  return(any(unlist(dup)))
+}

R/dst_get_data.R

@@ -47,6 +47,11 @@ dst_get_data <- function(table,
     }
   }
 
+  # If meta_data is NULL then get it automatically
+  if (is.null(meta_data)) {
+    meta_data <- dst_meta(table, lang = lang)
+  }
+
   # Force the names to be uppercase to match requirements from API
   names(query) <- toupper(names(query))
   dst_names <- names(query)
@@ -60,6 +65,11 @@ dst_get_data <- function(table,
     format = format
   )
 
+  # If overlaps in values are detected use CodeAndValue as presentation
+  if (dst_determine_overlaps(meta_data)) {
+    value_presentation <- "CodeAndValue"
+  }
+
   query$valuePresentation <- value_presentation
   query$lang <- lang
 
@@ -107,6 +117,11 @@ dst_get_data <- function(table,
   }
   names(dst_data) <- c(dst_names, "value")
 
+  # Remove the code
+  if (dst_determine_overlaps(meta_data)) {
+    dst_data$TID <- sapply(stringr::str_split(dst_data$TID, "\\s+"), `[`, 2)
+  }
+
   # Parse the dates if param is TRUE
   if (parse_dst_tid) {
     dst_data$TID <- dst_date_parse(dst_date = dst_data$TID)

tests/testthat/test-dst_determine_overlaps.R

@@ -0,0 +1,3 @@
+test_that("Overlap detection works", {
+  expect_true(dst_determine_overlaps(dst_meta("FV22TOTA")))
+})