tests/testthat/test-dictionary_based_cleaning.R

In cleanepi: Clean and Standardize Epidemiological Data

test_dictionary <- readRDS(system.file("extdata", "test_dict.RDS",
                                       package = "cleanepi"))
test_that("add_to_dictionary works when adding only one element", {
  test <- add_to_dictionary(
    dictionary = test_dictionary,
    option     = "ml",
    value      = "male",
    grp        = "gender",
    order      = NULL
  )
  expect_s3_class(test, "data.frame")
  expect_false(identical(test_dictionary, test))
  expect_identical(nrow(test), 7L)
  expect_true("ml" %in% test[["options"]])
})

test_that("add_to_dictionary works when adding multiple elements", {
  test <- add_to_dictionary(
    dictionary = test_dictionary,
    option     = c("homme", "femme"),
    value      = c("male", "female"),
    grp        = "gender",
    order      = NULL
  )
  expect_s3_class(test, "data.frame")
  expect_false(identical(test_dictionary, test))
  expect_identical(nrow(test), 8L)
  expect_true(all(c("homme", "femme") %in% test[["options"]]))

  test <- add_to_dictionary(
    dictionary = test_dictionary,
    option     = "homme",
    value      = "male",
    grp        = "gender",
    order      = NULL
  )
  expect_s3_class(test, "data.frame")
  expect_false(identical(test_dictionary, test))
  expect_identical(nrow(test), 7L)
  expect_true("homme" %in% test[["options"]])
})

test_that("add_to_dictionary works when order is not NULL", {
  test <- add_to_dictionary(
    dictionary = test_dictionary,
    option     = c("homme", "femme"),
    value      = c("male", "female"),
    grp        = "gender",
    order      = 7L:8L
  )
  expect_s3_class(test, "data.frame")
  expect_false(identical(test_dictionary, test))
  expect_identical(nrow(test), 8L)
  expect_true(all(c("homme", "femme") %in% test[["options"]]))
})

data <- readRDS(system.file("extdata", "messy_data.RDS", package = "cleanepi"))
# introduce some noise
data[["gender"]][sample(1L:nrow(data), 10L, replace = FALSE)] <- "homme"

# update the data dictionary
test_dictionary <- add_to_dictionary(
  dictionary = test_dictionary,
  option     = "homme",
  value      = "male",
  grp        = "gender",
  order      = NULL
)
test_that("clean_using_dictionary works", {
  cleaned_df     <- clean_using_dictionary(
    data       = data,
    dictionary = test_dictionary
  )
  expect_s3_class(cleaned_df, "data.frame")
  expect_identical(ncol(data), ncol(cleaned_df))
  expect_false("homme" %in% cleaned_df[["gender"]])
})

# introduce a new option
data[["gender"]][2L] <- "femme"
test_that("clean_using_dictionary works with misspelled values", {
  cleaned_df     <- clean_using_dictionary(
    data       = data,
    dictionary = test_dictionary
  )
  expect_s3_class(cleaned_df, "data.frame")
  expect_identical(ncol(data), ncol(cleaned_df))
  expect_false("homme" %in% cleaned_df[["gender"]])
  expect_true("femme" %in% cleaned_df[["gender"]])
  expect_message(clean_using_dictionary(data       = data,
                                        dictionary = test_dictionary),
                 "misspelled values at lines 2 of column 'gender'")
})

test_that("construct_misspelled_report works", {
  res <- construct_misspelled_report(
    misspelled_options = list(gender = 2L),
    data
  )
  expect_s3_class(res, "data.frame")
  expect_identical(ncol(res), 3L)
  expect_identical(nrow(res), 1L)
  expect_identical(res[["idx"]], 2L)
  expect_identical(res[["column"]], "gender")
  expect_identical(res[["value"]], "femme")
})


# testing internal functions
test_that("make_readcap_dictionary works as expected", {
  test_data <- readRDS(system.file("extdata", "test_readcap_dictionary.RDS",
                                   package = "cleanepi"))
  res <- make_readcap_dictionary(metadata     = test_data,
                                 field_column = "field_name",
                                 opt_column = "select_choices_or_calculations",
                                 field_type   = "field_type")
  expect_s3_class(res, "data.frame")
  expect_identical(ncol(res), 4L)
  expect_identical(nrow(res), 11L)
  expect_named(res, c("options", "values", "grp", "orders"))
})

test_that("make_readcap_dictionary fails when the column with the options does
          not exist", {
            expect_error(
              make_readcap_dictionary(metadata     = test_data,
                                      field_column = "field_name",
                                      opt_column = "fake_column_name",
                                      field_type   = "field_type"),
              regexp = cat("Unrecognised column name: 'fake_column_name'")
            )
})