fix wrong normalisation of row/col names

Author: antaldaniel

R/coefficient_matrix_create.R

@@ -89,7 +89,7 @@ coefficient_matrix_create <- function(
     data_table <- data_table[, 1:last_column]
   }
 
-  key_column <- tolower(as.character(unlist(data_table[, 1])))
+  key_column <- as.character(unlist(data_table[, 1]))
 
   # find denominator row
   if (total %in% c("output", "p1", "output_bp")) {

R/input_coefficient_matrix_create.R

@@ -52,13 +52,15 @@ input_coefficient_matrix_create <- function(data_table,
     cpa_total = "total"
   )
   for (nm in names(alias_map)) {
-    if (nm %in% names(data_table) && !(alias_map[[nm]] %in% names(data_table))) {
+    if (nm %in% names(data_table) && 
+        !(alias_map[[nm]] %in% names(data_table))) {
       names(data_table)[names(data_table) == nm] <- alias_map[[nm]]
     }
   }
 
   # --- Handle missing total row (toy tables) ----------------------------------
-  if (!any(grepl("total|cpa_total|total output", tolower(data_table[[1]])))) {
+  if (!any(grepl("total|cpa_total|total output", 
+                 tolower(data_table[[1]])))) {
     return_part_arg <- NULL
     warning("No 'total' or 'cpa_total' row found; returning full matrix without subsetting.")
   } else {
@@ -73,9 +75,9 @@ input_coefficient_matrix_create <- function(data_table,
     digits = digits
   )
 
-  potential_total_names <- c("cpa_total", "total")
+  potential_total_names <- c("cpa_total", "total", "TOTAL", "CPA_TOTAL")
 
-  key_column <- tolower(as.character(unlist(cm[, 1])))
+  key_column <- as.character(unlist(cm[, 1]))
   remove_col <- which(names(cm) %in% potential_total_names)
   remove_row <- which(key_column %in% potential_total_names)
 

R/iotable_get_eurostat.R

@@ -348,5 +348,17 @@ iotable_get_eurostat <- function(
     version = 2
     )
   }
+  
+  # Normalise row + column names to uppercase to ensure symmetry 
+  # and joinability if not the custom iotables snake case names are 
+  # used
+  if (labelling != "iotables") {
+    names(out) <- toupper(names(out))
+    if ("prod_na" %in% names(out)) {
+      out$prod_na <- toupper(out$prod_na)
+    }
+  }
+  
   out
 }
+

tests/testthat/test-coefficient_matrix_create.R

@@ -89,3 +89,36 @@ test_that("Households are handled correctly in coefficient_matrix_create", {
   expect_true(all(vapply(cm_with_hh[-1], is.numeric, logical(1))))
   expect_true(all(vapply(cm_no_hh[-1], is.numeric, logical(1))))
 })
+
+
+test_that("coefficient_matrix_create preserves row/column name casing", {
+  
+  # A minimal symmetric toy table with precise uppercase CPA names
+  small_io <- data.frame(
+    prod_na = c("CPA_A01", "forestry", "CPA_A03", "TOTAL"),
+    CPA_A01 = c(1, 2, 3, 6),
+    forestry = c(4, 5, 6, 17),
+    CPA_A03 = c(7, 8, 9, 24),
+    TOTAL  = c(12, 15, 18, 47),
+    check.names = FALSE
+  )
+  
+  cm <- coefficient_matrix_create(small_io)
+  
+  # Check that coefficient matrix still has uppercase names
+  expect_identical(names(cm), names(small_io)[1:4])
+  
+  # Check that row names in column 1 are preserved in uppercase
+  expect_identical(
+    as.character(cm$prod_na),
+    as.character(small_io$prod_na)
+  )
+  
+  # Check that the set of row industry codes matches the set of column codes
+  # (this confirms that row/column name matching was not broken)
+  row_codes  <- as.character(cm$prod_na[cm$prod_na != "output"])
+  col_codes  <- names(cm)[-1]  # drop key column
+  
+  expect_identical(sort(row_codes[1:3]), sort(col_codes))
+})
+