namegen

#!/bin/sh
usage="Usage: ${0##*/} [OPTIONS] [COUNT|DICT...]"
help() { cat <</help
Generate a block of random names, without listing real words.
$usage
  -1      Show one name per line; \`namegen -1 1\` gives exactly one random name
  -c NUM  Output width in characters, default is \$COLUMNS, -c${COLUMNS:-80}
  -m NUM  Minimum letters in name (down to two), default is -m4
  -n NUM  Number of lines of these names to display, default is -n10
  -w      Generate words only, arguments are dictionary files/dirs
          or else $dict
  -x NUM  Maximum letters in name, default is -x6, output width grows for this
  COUNT   Same as -n NUM, requires a lack of -w
  DICT    File(s) or directori(es) to sample, each line is a word (requires -w)

When generating potential domain names, even 5 is a rough minimum.

/help
version $1
}

version() {
  echo "A part of misc-scripts, https://github.com/adamhotep/misc-scripts"
  echo "namegen 0.6.20200827.1 copyright 2005+ by Adam Katz, GPLv3+"
  exit $1
}

die() { [ -n "$1" ] && printf '%s\n' "$@" >&2; exit 2; }
if ! command -v utf2ascii >/dev/null 2>&1; then
  utf2ascii() { sed 's/[^ -~]/-/g' "$@"; }
fi

min=4
max=6
dict="/usr/share/dict"
lines= single= words=
[ -z "$COLUMNS" ] && COLUMNS="$(tput cols)" # get window width
numcol=${COLUMNS:-80} # default to window width or else 80 if still unknown

isnum() {
  if [ "$1" -gt 0 ] 2>/dev/null; then
    echo "$1"
  else
    local n="$1"
    shift
    die "Invalid number of ${*:-letters} '$n'"
  fi
}

while getopts 1c:hm:n:wvVx:-: OPT; do
  if [ "$OPT" = - ]; then # long opt https://stackoverflow.com/a/28466267/519360
    OPT="${OPTARG%%=*}" OPTARG="${OPTARGS#"$OPT"}" OPTARG="${OPTARG#=}"
  fi
  case "$OPT" in
    ( 1 | single )	SINGLE=true ;;
    ( c | char* )	numcol="$(isnum "$OPTARG" chars for width)" || die ;;
    ( h | help )	help ;;
    ( m | min* )	min="$(isnum "$OPTARG" minimum letters)" || die ;;
    ( n | lin* | num* )	lines="$(isnum "$OPTARG" lines)" || die ;;
    ( [vV] | ver* )	version ;;
    ( w | word* )	words=true ;;
    ( x | max* )	max="$(isnum "$OPTARG" maxiumum letters)" || die ;;
    ( \? )		die ;;
    ( * )		die "Illegal option --$OPT" ;;
  esac
done
shift $((OPTIND-1))

if [ -z "$words$lines" -a -n "$1" ]; then
  lines="$(isnum "$1" lines)" || die
elif [ -z "$lines" ]; then
  lines=10
fi

# sanitize min and max
[ "$min" -lt 2 ] && min=2
[ "$max" -lt 2 ] && max=2
[ "$min" -gt "$max" ] && foo="$min" && min="$max" && max="$foo"

type apg >/dev/null 2>&1 || apg() {
  echo "You'd get better results with APG (Automated Password Generator)..." >&2
  head -c"$((lines*COLUMNS*2000))" /dev/urandom |strings |tr '[A-Z]' '[a-z]' \
    |tr -cd "[a-z'-]" \
    |awk -v min=$min -v max=$max -v apos="'" -v lines=$lines -v cols=$COLUMNS '
      BEGIN {
        for (i=97; i<=122; i++) ord[sprintf("%c", i)] = i
        ord[apos] = 123  # not ascii!
        ord["-"]  = 124  # not ascii!
      }
      {
        # TODO: this is broken
        while (length($0) > 1 && count < lines * cols * 20) {
          srand(srand() + ord[substr($0, 1, 1)]);	# steal a char for seed
          i = (ord[substr($0, 2, 1)] - 97) / (max-min);	# another for rand len
          i = sprintf("%.0f", rand() * i + min);	# set random length
          print substr($0, 3, i);			# get substr
          $0 = substr($0, 4+i)
          count++
        }
      }'
}

col=$(($max+1))


# if 'aspell list' will list mispelled words (and not real words), use it
if [ "$(echo vgqvq word |aspell list 2>/dev/null)"  = vgqvq ]; then
  spell="aspell list"
fi

# that list of consonants is by frequency of top >2% of English
for name in $(
  if [ "$words" = true ]; then
    find -L "${@:-$dict}" -maxdepth 1 -not -type d -not -name 'README*' -print0 \
      |xargs -0 awk -v min="$min" -v max="$max" "
        !/'s$/ { L = length(\$0); if (min <= L && L <= max) print }" \
      |shuf -n "$((lines*COLUMNS/3))" \
      |utf2ascii \
      |awk '!seen[tolower($0)]++'
  else
    apg -M l -m$min -x$max -n"$((lines*COLUMNS*20))" \
      |grep "[srntlcdgmph]" |grep "[aeiou]" \
      |egrep -v "cr?$|c[^ekhr]|a[eo]|e[io]|i[eu]|oe|u[aio]|[a-df-kmnpqt-z]{2}" \
      |${spell:-cat}
  fi
); do
   if [ -n "$SINGLE" ]; then
     echo "$name"
   else
     while [ ${#name} -lt $col ]; do name=" $name"; done # spacing for columns
     i="$((${i:-1} + 1))"   				# count words per line
     printf "$name"
     [ $i -gt $(($numcol/$col)) ] && echo "" && i=1 # line full -> wrap
   fi
   [ $? = 0 ] && lines="$(($lines - 1))" && [ 0 = "$lines" ] && exit 0 # done?
done	# done when $lines has shrunk to zero (as above) or we're out of names