test_that("hcut validates k and scaled inputs", {
  x <- iris[, 1:4]
  expect_error(hcut(x, k = 1), "k must be a single integer >= 2")
  expect_error(
    hcut(x, k = nrow(x) + 1),
    "elements of 'k' must be between 1 and"
  )

  x_const <- data.frame(a = 1:10, b = rep(1, 10))
  expect_error(
    hcut(x_const, k = 2, stand = TRUE),
    "Scaling produced NA values"
  )
})

test_that("hcut requires dist input when isdiss is TRUE", {
  x <- iris[, 1:4]
  expect_error(
    hcut(x, k = 2, isdiss = TRUE),
    "must be an object of class dist"
  )
})

test_that("hkmeans validates inputs and k bounds", {
  x <- iris[, 1:4]
  expect_error(hkmeans(x, k = 1), "k must be a single integer >= 2")
  expect_error(
    hkmeans(x, k = nrow(x) + 1),
    "elements of 'k' must be between 1 and"
  )
  expect_error(hkmeans(1:10, k = 2), "x must be a matrix or data.frame")
})

test_that("get_clust_tendency validates numeric data and n", {
  bad_x <- as.matrix(data.frame(a = letters[1:5], b = letters[1:5]))
  expect_error(get_clust_tendency(bad_x, n = 2, graph = FALSE), "numeric")
  expect_error(get_clust_tendency(iris[, 1:4], n = 0, graph = FALSE), "positive integer")
  expect_error(get_clust_tendency(iris[, 1:4], n = nrow(iris), graph = FALSE), "no larger")

  x_na <- matrix(c(1, 2, NA, NA), ncol = 2)
  expect_error(get_clust_tendency(x_na, n = 1, graph = FALSE), "at least two complete rows")
})