html to plain text kotlin port of https://stackoverflow.com/a/50363077

HtmlToPlainText.kt

/**
 * It depends on JSoup.
 * Kotlin port of https://stackoverflow.com/a/50363077
 */
object Utils {
  private val block = "address|article|aside|blockquote|canvas|dd|div|dl|dt|" +
    "fieldset|figcaption|figure|footer|form|h\\d|header|hr|li|main|nav|" +
    "noscript|ol|output|p|pre|section|table|tfoot|ul|video"
  private val rxlongWSpaces  = Regex("""\s{2,}""")
  private val rxNestedBlock = Regex("""(\s*?</?(${block})[^>]*?>)+\s*""", RegexOption.IGNORE_CASE)
  private val rxBrToNewLine = Regex("""<br[^>]*>""", RegexOption.IGNORE_CASE)
  private val removeAllTags = Regex("""<[^>]*(>|$)""", RegexOption.MULTILINE)

  fun htmlToPlainText(html: String): String {
    var buf = html
    buf = rxlongWSpaces.replace(buf, " ")
    buf = rxNestedBlock.replace(buf, "\n").trim()
    buf = rxBrToNewLine.replace(buf, "\n")
    buf = removeAllTags.replace(buf, "")
    buf = Parser.unescapeEntities(buf, false)
    return buf;
  }
}