#!/bin/sh
set -eu

# Print a comma-separated list of unique registrable root domains (eTLD+1)
# for non-suspended cPanel users based on /etc/userdomains.
#
# Tries to use `psl` (libpsl) if available for correct public suffix handling.
# Falls back to a heuristic for common multi-level TLDs if `psl` is unavailable.

USERDOMAINS_PATH="/etc/userdomains"
SUSPENDED_DIR="/var/cpanel/suspended"

error() {
  printf "Error: %s\n" "$*" 1>&2
}

have_psl() {
  command -v psl >/dev/null 2>&1
}

get_root_domain() {
  # Derive registrable domain (eTLD+1) from a hostname
  # 1) Prefer libpsl via `psl --print`
  # 2) Fallback heuristic for common multi-level TLDs
  hostname="$1"
  # strip trailing dot if any
  case "$hostname" in
    *.) hostname=${hostname%\.} ;;
  esac

  if have_psl; then
    printed=$(psl --print "$hostname" 2>/dev/null | tr -d '\r')
    if [ -n "$printed" ]; then
      printf "%s\n" "$printed"
      return 0
    fi
  fi

  # Fallback heuristic implemented in awk to avoid bashisms
  printf "%s\n" "$hostname" | awk '
    BEGIN { FS = "." }
    {
      n = NF
      if (n >= 3) {
        lastTwo = $(n-1) "." $n
        if (lastTwo ~ /^(co\.uk|org\.uk|ac\.uk|gov\.uk|ltd\.uk|plc\.uk|me\.uk|co\.jp|ne\.jp|or\.jp|go\.jp|ac\.jp|ed\.jp|gr\.jp|com\.au|net\.au|org\.au|edu\.au|gov\.au|id\.au|asn\.au|com\.br|com\.cn|net\.cn|org\.cn|com\.sg|com\.my|co\.nz)$/) {
          print $(n-2) "." lastTwo
          next
        }
      }
      if (n >= 2) {
        print $(n-1) "." $n
      } else {
        print $0
      }
    }'
}

main() {
  if [ ! -r "${USERDOMAINS_PATH}" ]; then
    error "Cannot read ${USERDOMAINS_PATH}"
    exit 1
  fi

  # Read, normalize whitespace removal like original, skip lines starting with '*'
  # Format is typically: domain: user
  # For each non-suspended user, emit root domain; then unique-sort and join with commas
  sed -e 's/[[:space:]]*//g' -e '/^\*/d' "${USERDOMAINS_PATH}" \
    | awk -F: 'NF>=2 { print $1 ":" $2 }' \
    | while IFS=":" read -r domain username; do
        if [ ! -f "${SUSPENDED_DIR}/${username}" ]; then
          get_root_domain "${domain}"
        fi
      done \
    | sort -u \
    | paste -sd, -
}

main "$@"