################################################################################

context("CHECK_ARGS")

set.seed(SEED)

################################################################################

# defining some objects
X <- FBM(10, 10, type = "raw")
X.code <- add_code256(X, code = rep(0, 256))


test_that("checking of BM (desc) arguments", {
  # X
  expect_null( (function(X) check_args())(X))
  expect_null( (function(X) check_args())(X.code))
  # X.code
  expect_error((function(X.code) check_args())(X),
               "'X.code' is not of class 'FBM.code256'")
  expect_null( (function(X.code) check_args())(X.code))
})

################################################################################

# defining some objects
y <- rnorm(10)
y01 <- sample(0:1, size = 100, replace = TRUE)
ind1 <- 1:10
ind2 <- -c(1:10)
ind3 <- -10:10
ind4 <- seq(1, 10, by = 0.8)
ind5 <- ind1; ind5[10] <- NA

test_that("checking of vector arguments", {
  # y01.train
  expect_error((function(y01.train) check_args())(y),
               "'y01.train' should be composed of 0s and 1s")
  expect_null( (function(y01.train) check_args())(y01))
  # ind.train
  expect_null( (function(ind.train) check_args())(ind1))
  expect_error((function(ind.train) check_args())(ind2), "only positive")
  expect_error((function(ind.train) check_args())(ind3), "only positive")
  expect_error((function(ind.train) check_args())(ind4), "only integers")
  expect_error((function(ind.train) check_args())(ind5), "only positive")
  expect_error((function(ind.train) check_args())(c("a", "b")), "only integers")
  # ind.row
  expect_null( (function(ind.row) check_args())(ind1))
  expect_error((function(ind.row) check_args())(ind2), "only positive")
  expect_error((function(ind.row) check_args())(ind3), "only positive")
  expect_error((function(ind.row) check_args())(ind4), "only integers")
  expect_error((function(ind.row) check_args())(ind5), "only positive")
  expect_error((function(ind.row) check_args())(c("a", "b")), "only integers")
  # ind.col
  expect_null( (function(ind.col) check_args())(ind1))
  expect_error((function(ind.col) check_args())(ind2), "only positive")
  expect_error((function(ind.col) check_args())(ind3), "only positive")
  expect_error((function(ind.col) check_args())(ind4), "only integers")
  expect_error((function(ind.col) check_args())(ind5), "only positive")
  expect_error((function(ind.col) check_args())(c("a", "b")), "only integers")
})

################################################################################

# defining some objects
f1 <- FBM
f2 <- big_scale()
f3 <- FBM(10, 10)
f4 <- big_scale

test_that("checking of function arguments", {
  # fun.scaling
  expect_error((function(fun.scaling) check_args())(f1))
  expect_null( (function(fun.scaling) check_args())(f2))
  expect_error((function(fun.scaling) check_args())(f3))
  expect_error((function(fun.scaling) check_args())(f4))
})

################################################################################

# defining some objects
ncores <- getOption("bigstatsr.ncores.max")
covar1 <- NULL
covar2 <- matrix(0, 10, 2)
covar3 <- covar2; covar3[10] <- NA

test_that("checking of other arguments", {
  # ncores
  expect_null( (function(ncores) check_args())(ncores))
  expect_error((function(ncores) check_args())(ncores + 1))
  expect_error((function(ncores) check_args())(Inf))
  opt.save <- options(bigstatsr.ncores.max = Inf)
  expect_null( (function(ncores) check_args())(ncores + 1))
  expect_null( (function(ncores) check_args())(Inf))
  options(opt.save)
  # covar.train
  expect_null( (function(covar.train) check_args())(covar1))
  expect_null( (function(covar.train) check_args())(covar2))
  expect_error((function(covar.train) check_args())(covar3), "missing values")
  expect_error((function(covar.train) check_args())(ind1), "class 'matrix'")
  # covar.row
  expect_null( (function(covar.row) check_args())(covar1))
  expect_null( (function(covar.row) check_args())(covar2))
  expect_error((function(covar.row) check_args())(covar3), "missing values")
  expect_error((function(covar.row) check_args())(ind1), "class 'matrix'")
})

################################################################################

test_that("overwrite default checking", {
  # ind.train
  f1 <- function(ind.train) check_args(
    ind.train = "assert_int(ind.train); assert_pos(ind.train)")
  expect_error(f1(ind2))

  f2 <- function(ind.train) check_args(ind.train = "assert_int(ind.train)")
  expect_null(f2(ind2))
})

################################################################################

ind <- 1:10
y <- rnorm(10)
covar <- matrix(0, 10, 2)

test_that("checking lengths", {
  expect_error(assert_lengths(ind))
  expect_null( assert_lengths(ind, y))
  expect_error(assert_lengths(ind[-1], y))
  expect_error( assert_lengths(ind, y, covar))
  expect_null( assert_lengths(ind, y, rows_along(covar)))
})

################################################################################

X.code$code256 <- rep(NA_real_, 256)

test_that("checking missing values", {
  expect_error((function(X.code) check_args())(X.code))
  expect_null((function(X.code) check_args(
    X.code = "assert_class(X.code, 'FBM.code256')"))(X.code))
  expect_null(assert_noNA(FBM(1, 1, init = 0)))
  expect_null(assert_noNA(FBM(1, 1, init = 1)))
  expect_error(assert_noNA(FBM(1, 1, init = NA)), "You can't have missing values")
})

################################################################################