tests/testthat/test-utf8.R

In ropensci/testdat: A suite for running automated tests for data

context("UTF8")

# Sample data objects
data(iris)
utf8df <- structure(
  list(
    x = 'The\U3e32393cs quick brown fox jumps over the \U3e32393cs the lazy dog',
    y = 'Lorem\U3e32393cs ipsum \U3e32393cs dolor',
    dat = 43.5
  ),
  .Names = c('x', 'y', 'dat'), row.names = c(NA, -1L), class = 'data.frame'
)

# This CSV file is UTF-8
utf8csv <- read.csv(system.file("extdata", "km1314-waypoints.csv",
                                package = "testdat"))


test_that("test_utf8 errors on non-data frames", {
  expect_error(test_utf8(1:8))
  expect_error(test_utf8(list(1,2,3)))
})


test_that("test_utf8 correctly detects UTF-8", {
  expect_false(test_utf8(iris))
  # THESE TWO NEED FIXING
  # expect_true(test_utf8(utf8df))
  # expect_true(test_utf8(utf8csv))

  # Testing latin1 encoded strings
  txt <- "fa\xE7ile"
  Encoding(txt) <- "latin1"
  df <- data.frame(txt)
  # THIS ALSO FAILS
  # expect_true(test_utf8(df))
  # FIXME: test_utf8 returns TRUE for non-UTF-8 encoded strings.
  # Maybe it should be called test_nonascii?
})


test_that("sanitize_text cleans non-ASCII chars", {
  expect_identical(
    sanitize_text("This is some bad text \U3e32393cs that contains utf-8 characters"),
    "This is some bad text s that contains utf-8 characters"
  )

  # Should work on vectors, even with mixed encodings.
  # This vector has "UTF-8" and "unknown".
  expect_identical(
    sanitize_text(c("xx\U3e32393cxx", "yy\xE7yy", "zzzz")),
    c("xxxx", "yyyy", "zzzz")
  )
})


test_that("sanitize_text errors on non-character vectors", {
  expect_error(sanitize_text(iris))
  expect_error(sanitize_text(1:10))
})