test_that("hurdle execution", {
  skip_if_not_installed("pscl")
  skip_on_cran()

  data("bioChemists", package = "pscl", envir = rlang::current_env())

  hurdle_spec <- poisson_reg() |> set_engine("hurdle")
  ctrl <- control_parsnip(verbosity = 1, catch = FALSE)

  expect_no_error(
    fit(hurdle_spec, art ~ ., data = bioChemists, control = ctrl)
  )
  expect_no_error(
    res <- fit_xy(
      hurdle_spec,
      x = bioChemists[, 2:6],
      y = bioChemists$art,
      control = ctrl
    )
  )

  expect_false(has_multi_predict(res))
  expect_equal(multi_predict_args(res), NA_character_)
})

test_that("hurdle prediction", {
  skip_if_not_installed("pscl")
  skip_on_cran()

  data("bioChemists", package = "pscl", envir = rlang::current_env())

  hurdle_spec <- poisson_reg() |> set_engine("hurdle")
  quiet_ctrl <- control_parsnip(verbosity = 0, catch = TRUE)

  hurdle_pred <- c(
    2.00569689955261,
    1.29916133671851,
    1.30005204940495,
    1.43756150261801,
    2.37677697507562
  )

  res_xy <- fit_xy(
    hurdle_spec,
    x = bioChemists[, 2:6],
    y = bioChemists$art,
    control = quiet_ctrl
  )

  expect_equal(
    predict(res_xy, bioChemists[1:5, 2:6])$.pred,
    hurdle_pred,
    tolerance = 0.001
  )

  form_pred <- c(
    1.83402584880366,
    1.45332695575065,
    1.43412114470316,
    1.58891378718055,
    1.855682964733
  )

  res_form <- fit(
    hurdle_spec,
    art ~ . | 1,
    data = bioChemists,
    control = quiet_ctrl
  )

  expect_equal(
    predict(res_form, bioChemists[1:5, ])$.pred,
    form_pred,
    tolerance = 0.001
  )
})