target_market_share errors if input data has unexpected columns (#…

…268)

Co-authored-by: GitHub Actions <[email protected]>

NEWS.md

@@ -1,5 +1,8 @@
 # r2dii.analysis (development version)
 
+* `target_market_share()` now errors if input `data` has an unexpected column
+  (@georgeharris2deg #267).
+
 * `target_market_share()` now correctly outputs `technology_share` with
   multiple loans to the same company (@georgeharris2deg #262, @ab-bbva #265).
 

R/target_market_share.R

@@ -95,6 +95,38 @@ target_market_share <- function(data,
 
   data <- ungroup(warn_grouped(data, "Ungrouping input data."))
 
+  valid_columns <- c(
+    "id_loan",
+    "id_direct_loantaker",
+    "name_direct_loantaker",
+    "id_intermediate_parent_1",
+    "name_intermediate_parent_1",
+    "id_ultimate_parent",
+    "name_ultimate_parent",
+    "loan_size_outstanding",
+    "loan_size_outstanding_currency",
+    "loan_size_credit_limit",
+    "loan_size_credit_limit_currency",
+    "sector_classification_system",
+    "sector_classification_input_type",
+    "sector_classification_direct_loantaker",
+    "fi_type",
+    "flag_project_finance_loan",
+    "name_project",
+    "lei_direct_loantaker",
+    "isin_direct_loantaker",
+    "id_2dii",
+    "level",
+    "sector",
+    "sector_ald",
+    "name",
+    "name_ald",
+    "score",
+    "source",
+    "borderline"
+  )
+
+  check_valid_columns(data, valid_columns)
   data <- aggregate_by_loan_id(data)
 
   crucial_scenario <- c("scenario", "tmsr", "smsp")
@@ -406,3 +438,16 @@ aggregate_by_loan_id <- function(data) {
     ) %>%
     ungroup()
 }
+
+check_valid_columns <- function(data, valid_columns) {
+  invalid_columns <- setdiff(names(data), valid_columns)
+
+  if (length(invalid_columns) != 0) {
+    abort(
+      glue("Loanbook has unexpected names: `{invalid_columns}`."),
+      class = "invalid_columns"
+    )
+  }
+
+  invisible(data)
+}

tests/testthat/test-target_market_share.R

@@ -677,6 +677,21 @@ test_that("for one company with multiple loans of different size, unweighted
   expect_equal(projected$production, fake_ald()$production)
 })
 
+test_that("with bad column errors with informative message (#267)", {
+  bad_matched <- fake_matched(
+    bad_column = "bad"
+  )
+
+  expect_error(
+    class = "invalid_columns",
+    target_market_share(
+      bad_matched,
+      fake_ald(),
+      fake_scenario()
+    )
+  )
+})
+
 test_that("`technology_share` outputs consistently when multiple
           direct_loantakers match to a single company (#265)", {