# The words behind a SMOG or Gunning fog score: each one of three or more
# syllables. The scores say a paragraph reads hard; this says which words
# made it so. Syllables are counted by vowel groups, with the silent e and
# the -ed and -es endings taken off, which agrees with the scores' count on
# nearly every word and errs on the long side for the rest.
#
# A suggestion, since a paragraph can carry a few. Turn it on while bringing
# a score down, and off again after. See issue 8.
extends: script
message: "'%s' has three or more syllables. It counts toward the SMOG and Gunning fog scores."
link: https://en.wikipedia.org/wiki/SMOG
level: suggestion
scope: paragraph
script: |
  text := import("text")

  matches := []
  for m in text.re_find(`[A-Za-z]+(?:'[A-Za-z]+)?`, scope, -1) {
    w := text.to_lower(m[0].text)
    if len(w) < 5 {
      continue
    }
    groups := text.re_find(`[aeiouy]+`, w, -1)
    n := groups == undefined ? 0 : len(groups)
    if text.has_suffix(w, "e") && !text.has_suffix(w, "le") && !text.has_suffix(w, "ee") {
      n--
    }
    if text.has_suffix(w, "ed") && !text.re_match(`[td]ed$`, w) {
      n--
    }
    if text.has_suffix(w, "es") && !text.re_match(`[sxz]es$|[cs]hes$`, w) {
      n--
    }
    if n >= 3 {
      matches = append(matches, {begin: m[0].begin, end: m[0].end})
    }
  }