#!/usr/bin/env bash
# check-redirects.sh: test a Skewcheck redirect map against the live site.
# Usage: NEW_SITE=https://www.newsite.com ./check-redirects.sh redirect-map.csv > results.tsv
#
# Reads the columns named old_url, new_path and action (positions 2, 4 and 5
# on the redirect map template). One URL at a time: a GET for the first
# status (15s cap), then a GET that follows up to 10 redirects (30s cap).
set -u

usage="usage: NEW_SITE=https://www.newsite.com $0 redirect-map.csv"
map="${1:?$usage}"
new_site="${NEW_SITE:?set NEW_SITE to the live site, e.g. https://www.newsite.com}"
new_site="${new_site%/}"

if [ ! -f "$map" ]; then
  printf 'error: cannot read %s\n' "$map" >&2
  exit 1
fi

# nth comma-separated field, 1-based. Quoted commas are not supported.
field() {
  n=$1
  rest=$2
  i=1
  while [ "$i" -lt "$n" ]; do
    case "$rest" in
      *,*) rest="${rest#*,}" ;;
      *) rest="" ;;
    esac
    i=$((i + 1))
  done
  printf '%s' "${rest%%,*}"
}

trim() {
  s=$1
  s="${s#"${s%%[![:space:]]*}"}"
  s="${s%"${s##*[![:space:]]}"}"
  printf '%s' "$s"
}

old_i=2
new_i=4
action_i=5
first_row=1

printf 'old_url\texpected\tfirst_status\thops\tfinal_status\tfinal_url\tresult\n'

while IFS= read -r line || [ -n "${line:-}" ]; do
  line="${line%$'\r'}"
  [ -z "$line" ] && continue

  if [ "$first_row" -eq 1 ]; then
    first_row=0
    case "$line" in
      *old_url*)
        i=1
        rest=$line
        while [ -n "$rest" ]; do
          name=$(trim "${rest%%,*}")
          case "$name" in
            old_url) old_i=$i ;;
            new_path) new_i=$i ;;
            action) action_i=$i ;;
          esac
          case "$rest" in
            *,*) rest="${rest#*,}" ;;
            *) break ;;
          esac
          i=$((i + 1))
        done
        continue
        ;;
    esac
  fi

  old=$(trim "$(field "$old_i" "$line")")
  new_path=$(trim "$(field "$new_i" "$line")")
  action=$(trim "$(field "$action_i" "$line")")
  [ -z "$old" ] && continue

  case "$old" in
    http://*|https://*) ;;
    *)
      # A leading = + - or @ is a spreadsheet formula. A quote keeps the cell as text.
      shown=$old
      case "$shown" in
        [-+=@]*) shown="'${shown}" ;;
      esac
      printf '%s\t%s\t%s\t%s\t%s\t%s\t%s\n' \
        "$shown" "(skipped: not an http(s) URL)" "" "" "" "" "CHECK"
      continue
      ;;
  esac

  case "$action" in
    redirect|keep|retire-404|retire-410) ;;
    *)
      printf '%s\t%s\t%s\t%s\t%s\t%s\t%s\n' \
        "$old" "(action: ${action})" "" "" "" "" "CHECK"
      continue
      ;;
  esac

  # || true: curl exits non-zero on a dead host. This file only sets -u, but
  # one failure must not stop the rest of the file if the caller has -e on.
  # -g: a URL with [ ] or { } must be one request, not a curl glob.
  first=$(curl -g -sS -o /dev/null --proto '=http,https' --proto-redir '=http,https' --max-time 15 -w '%{http_code}' -- "$old") || true
  hops=""
  final_status=""
  final_url=""
  meta=$(curl -g -sS -o /dev/null --proto '=http,https' --proto-redir '=http,https' -L --max-redirs 10 --max-time 30 \
    -w '%{num_redirects}\t%{http_code}\t%{url_effective}' -- "$old") || true
  IFS=$'\t' read -r hops final_status final_url <<EOF
$meta
EOF
  hops=${hops:-}
  final_status=${final_status:-}
  final_url=${final_url:-}

  case "$action" in
    redirect)
      expected="${new_site}${new_path}"
      if [[ "$first" =~ ^30[18]$ && "$hops" == 1 && "$final_status" == 200 && "$final_url" == "$expected" ]]; then
        r=PASS
      else
        r=FAIL
      fi
      ;;
    keep)
      expected="${new_site}${new_path}"
      if [[ "$hops" =~ ^[0-9]+$ ]] && [ "$hops" -le 1 ] && [ "$final_status" = 200 ] && [ "$final_url" = "$expected" ]; then
        r=PASS
      else
        r=FAIL
      fi
      ;;
    retire-404|retire-410)
      code="${action#retire-}"
      expected="${code}, no redirect"
      if [ "$first" = "$code" ] && [ "$hops" = 0 ]; then
        r=PASS
      else
        r=FAIL
      fi
      ;;
    *)
      expected="(action: ${action})"
      r=CHECK
      ;;
  esac

  printf '%s\t%s\t%s\t%s\t%s\t%s\t%s\n' \
    "$old" "$expected" "$first" "$hops" "$final_status" "$final_url" "$r"
done < "$map"