#! /bin/bash -ue
# Last edited on 2026-07-17 22:17:27 by stolfi

smps=( "$@" )

# Prints number of items (including punct, sybols, etc) and tokens
# from all "dat/{smp}/org/main.wds" files

wdsFiles=( `echo ${smps[@]} | tr ' ' '\012' | sed -e "s:^:dat/:g" -e "s:\$:/org/main.wds:g" ` )
echo "wdsFiles = ${wdsFiles[@]}" 1>&2
printf "    %6s %6s %-s\n" "TOKENS" "OTHER"  "FILE" 1>&2
printf "    %6s %6s %-s\n" "------" "------" "----------------------------------------" 1>&2
for file in ${wdsFiles[@]}; do
  if [[ ! ( -s ${file} ) ]]; then
    echo "** file ${file} not found" 1>&2
  else
    cts=( $( cat ${file} | gawk 'BEGIN{ nt=0; no=0; } /^[b-z]/{ no++ } /^a/ { nt++ } END { print nt, no }' ) )
    printf "    %6d %6d %s\n" ${cts[0]} ${cts[1]} ${file} 1>&2
  fi
done

