#! /bin/bash
# Last edited on 2024-03-30 18:15:02 by stolfi

# Usage: $0 <inprefix> <outprefix>
# Collects all ".good", ".bad", and ".dunno" word lists 
#   whose names start with <inprefix>* into the consolidated files 
#   <outprefix>.good, <outprefix>.dunno, <outprefix>.bad, <outprefix>.bdg
#   Also checks disjointness of these output lists.

if [[ $# -ne 2 ]]; then
  echo "** usage: $0 <inprefix> <outprefix>" 1>&2
  exit 1
fi

inprefix="$1"
outprefix="$2"

for f in good bad dunno; do
  echo '==============' "$f" '==============' 1>&2
  wc ${inprefix}*.$f 1>&2
  echo ' ' 1>&2
  cat ${inprefix}*.$f | sort -T . | uniq > ${outprefix}.$f
done

cat ${outprefix}.{good,bad,dunno} | sort | uniq > ${outprefix}.bdg

echo ' ' 1>&2
wc ${outprefix}.{bad,dunno,good,bdg} 1>&2

echo ' ' 1>&2

dicio_list_dups_in_bad_dunno_good ${outprefix}
