Fix runModelingPipelineIntense()

R/run_ML.R

@@ -1429,16 +1429,16 @@ runModelingPipelineIntense <- function(parquet_dir,
   # -------------------------------
   # [1] Generate inputs
   # -------------------------------
- # safe_run(
-#    "[1] Generating ML feature matrices",
-#    generateMLInputs,
- #   parquet_dir = parquet_dir,
- #   out_path = out_root,
- #   n_fold = n_fold,
- #   split = split,
- #   min_n = min_n,
- #   verbosity = if (verbose) "minimal" else "debug"
- # )
+  safe_run(
+    "[1] Generating ML feature matrices",
+    generateMLInputs,
+    parquet_dir = parquet_dir,
+    out_path = out_root,
+    n_fold = n_fold,
+    split = split,
+    min_n = min_n,
+    verbosity = if (verbose) "minimal" else "debug"
+  )
 
   # -------------------------------
   # Seed loop

tests/testthat/test-run-modeling-pipeline-intense.R

@@ -0,0 +1,33 @@
+# Unit test for runModelingPipelineIntense()'s control flow. Mocks out the
+# heavy steps (matrix generation, model training) so this stays fast and only
+# checks that the wrapper actually calls them in order, rather than exercising
+# real matrix generation / model fitting (already covered elsewhere).
+
+test_that("runModelingPipelineIntense() generates matrices before training", {
+  tmp_root <- withr::local_tempdir()
+  parquet_dir <- file.path(tmp_root, "Test_species")
+  dir.create(parquet_dir)
+
+  generate_calls <- list()
+  local_mocked_bindings(
+    generateMLInputs = function(...) {
+      generate_calls[[length(generate_calls) + 1]] <<- list(...)
+      invisible(NULL)
+    },
+    runMLmodels = function(...) invisible(NULL),
+    runMDRmodels = function(...) invisible(NULL)
+  )
+
+  runModelingPipelineIntense(
+    parquet_dir = parquet_dir,
+    n_seeds = 1,
+    verbose = FALSE
+  )
+
+  # The matrix-generation step must actually run (regression test: this step
+  # was previously commented out, so every later step silently had nothing
+  # to work with).
+  expect_length(generate_calls, 1)
+  expect_equal(generate_calls[[1]]$parquet_dir, normalizePath(parquet_dir))
+  expect_equal(generate_calls[[1]]$out_path, normalizePath(tmp_root))
+})