tests/testthat/test-ml-feature-stop-words-remover.R

In rstudio/sparklyr: R Interface to Apache Spark

skip_connection("ml-feature-stop-words-remover")
skip_on_livy()
skip_on_arrow_devel()

skip_databricks_connect()
test_that("ft_stop_words_remover() default params", {
  test_requires_version("3.0.0")
  sc <- testthat_spark_connection()
  test_default_args(sc, ft_stop_words_remover)
})

test_that("ft_stop_words_remover() param setting", {
  test_requires_version("3.0.0")
  sc <- testthat_spark_connection()
  test_args <- list(
    input_col = "foo",
    output_col = "bar",
    case_sensitive = TRUE,
    stop_words = c("foo", "bar")
  )
  test_param_setting(sc, ft_stop_words_remover, test_args)
})



test_that("ft_stop_words_remover() works", {
  test_requires_version("2.0.0", "loadDefaultStopWords requires Spark 2.0+")
  sc <- testthat_spark_connection()
  df <- tibble(
    id = c(0, 1),
    raw = c("I saw the red balloon", "Mary had a little lamb")
  )
  df_tbl <- copy_to(sc, df, overwrite = TRUE)

  expect_identical(
    df_tbl %>%
      ft_tokenizer("raw", "words") %>%
      ft_stop_words_remover("words", "filtered") %>%
      pull(filtered) %>%
      lapply(as.list),
    list(list("saw", "red", "balloon"), list("mary", "little", "lamb"))
  )

  expect_identical(
    df_tbl %>%
      ft_tokenizer("raw", "words") %>%
      ft_stop_words_remover("words", "filtered", stop_words = list("I", "Mary", "lamb")) %>%
      pull(filtered) %>%
      lapply(as.list),
    list(list("saw", "the", "red", "balloon"), list("had", "a", "little"))
  )

  swr <- ft_stop_words_remover(
    sc, "input", "output",
    case_sensitive = TRUE,
    stop_words = as.list(letters), uid = "hello"
  )

  expect_equal(
    ml_params(swr, list(
      "input_col", "output_col", "case_sensitive", "stop_words"
    )),
    list(
      input_col = "input",
      output_col = "output",
      case_sensitive = TRUE,
      stop_words = as.list(letters)
    )
  )
})

test_that("ml_default_stop_words() defaults to English (#1280)", {
  test_requires_version("2.0.0", "loadDefaultStopWords requires Spark 2.0+")
  sc <- testthat_spark_connection()
  expect_identical(
    ml_default_stop_words(sc) %>%
      head(5),
    list("i", "me", "my", "myself", "we")
  )
})

test_clear_cache()