test-clean_html.R

context("test-clean_html")

test_that("html cleansing works", {
  filename_raw <- "article-4-raw.html"
  filename_cleansed <- "article-4-cleansed.txt"
  html <- readChar(filename_raw, file.info(filename_raw)$size)

  expected <- gsub("\\s", "", readChar(filename_cleansed, file.info(filename_cleansed)$size))
  actual <- gsub("\\s", "", clean_html(html))

  expect_equal(expected, actual)
})