# ------------------------------------------------------------------------------
# Generated by 'pre-generate/generate-steps.R': do not edit by hand.
# ------------------------------------------------------------------------------

dist_train <- data.frame(
  pins = "cities",
  dist = I(list(eurodist))
)
dist_test <- data.frame(
  pins = "cities",
  dist = I(list(UScitiesD))
)
dist_rec <- recipe(~ ., data = dist_train) |> 
  step_pd_point_cloud(dist)
scale_seq <- seq(0, 5000, 100)

test_that("`step_vpd_complex_polynomial()` agrees with raw function", {
  
  vpd_rec <- dist_rec |> 
    step_vpd_complex_polynomial(
      dist,
      hom_degree = 0
    )
  
  vpd_prep <- prep(vpd_rec, training = dist_train)
  
  vpd_pred <- bake(vpd_prep, new_data = dist_test) |> 
    select(contains("_cp_")) |> 
    unlist() |> unname()
  
  vpd_exp <- dist_test$dist[[1L]] |> 
    ripserr::vietoris_rips() |> as.matrix() |> 
    TDAvec::computeComplexPolynomial(
      homDim = 0
    ) |> 
    as.vector()
  
  expect_equal(vpd_pred, vpd_exp)
})

test_that("`tunable()` returns standard names", {
  
  vpd_rec <- dist_rec |> 
    step_vpd_complex_polynomial(dist)
  tun <- tunable(vpd_rec$steps[[2]])
  
  expect_equal(
    names(tun),
    c("name", "call_info", "source", "component", "component_id")
  )
  expect_equal(
    tun$name,
    c("hom_degree", "num_coef", "poly_type")
  )
  expect_equal(unique(tun$source), "recipe")
  expect_true(is.list(tun$call_info))
})

# infrastructure tests
# prompted by issue #14

test_that("recipe and preparation printing is consistent", {
  vpd_rec <- dist_rec |>
    step_vpd_complex_polynomial(
      dist,
      hom_degree = 0
    )
  
  expect_snapshot(print(vpd_rec))
  expect_snapshot(prep(vpd_rec))
})

test_that("data with 0 or 1 rows works with `bake()` method", {
  
  vpd_prep <- dist_rec |> 
    step_vpd_complex_polynomial(
      dist,
      hom_degree = 0
    ) |> 
    prep()
  
  expect_identical(
    nrow(bake(vpd_prep, slice(dist_train, 1L))),
    1L
  )
  expect_identical(
    nrow(bake(vpd_prep, slice(dist_train, 0L))),
    0L
  )
})

test_that("`bake()` method errs needed non-standard role columns are missing", {
  
  vpd_rec <- dist_rec |> 
    step_vpd_complex_polynomial(
      dist,
      hom_degree = 0
    ) |>
    update_role(
      dist,
      new_role = "yam"
    ) |>
    update_role_requirements(role = "yam", bake = FALSE)
  
  vpd_prep <- prep(vpd_rec, training = dist_train)
  
  expect_snapshot(
    error = TRUE,
    bake(vpd_prep, new_data = subset(dist_test, select = -c(dist)))
  )
})

test_that("recipe successfully prints with empty predictors", {
  
  vpd_rec <- dist_rec |> 
    step_vpd_complex_polynomial(
      hom_degree = 0
    )
  
  expect_snapshot(vpd_rec)
  
  vpd_prep <- prep(vpd_rec, training = dist_train)
  
  expect_snapshot(vpd_prep)
})

test_that("recipe with empty selection incurs no `prep()` or `bake()` change", {
  
  vpd_rec1 <- dist_rec
  vpd_rec2 <- step_vpd_complex_polynomial(
    vpd_rec1,
    hom_degree = 0
  )
  
  vpd_prep1 <- prep(vpd_rec1, dist_train)
  vpd_prep2 <- prep(vpd_rec2, dist_train)
  
  vpd_bake1 <- bake(vpd_prep1, dist_train)
  vpd_bake2 <- bake(vpd_prep2, dist_train)
  
  expect_identical(vpd_bake1, vpd_bake2)
})

test_that("tidy method for empty selection works", {
  
  vpd_rec <- step_vpd_complex_polynomial(
    dist_rec,
    hom_degree = 0
  )
  
  expect <- tibble(terms = character(), value = double(), id = character())
  
  expect_identical(tidy(vpd_rec, number = 2L), expect)
  
  vpd_prep <- prep(vpd_rec, dist_train)
  
  expect_identical(tidy(vpd_prep, number = 2L), expect)
})