#! /usr/bin/gawk -f
# Last edited on 1999-01-06 06:42:58 by stolfi

# OBSOLETE?

BEGIN {
  abort = 0
  usage = (ARGV[0] " WORDFILE < TEXT")
  if (ARGC != 2) { error(("usage: " usage)); }
  wfile = ARGV[1];
  split("", dic)
  while((getline w < wfile) > 0) { dic[w] = 1 }
  if (ERRNO != "0") { error((wfile ": " ERRNO)); }
  close (wfile);
  ARGC = 1
  nRefs = 0
}

/#/ { 
  if (abort) exit;
  print; next
}

/./ {
  if (abort) exit;
  print
  if (substr($0,1,1) == "<") { skip = 19 } else { skip = 0 }
  if (skip >= length($0)) next;
  txt = substr($0,1+skip)
  for(w in dic)
    { i = index(txt, w);
      while (i != 0) 
        { nRefs++
          for(j=1;j<i+skip;j++) printf " ";
          printf "%s\n", w
          k = index(substr(txt, i+1), w);
          i = (k == 0 ? 0 : i + k)
        }
    }
  next
}

END {
  if(abort) exit;
  printf "found %d occurences\n", nRefs > "/dev/stderr"
}

function error(msg)
{ 
  printf "line %d: %s\n", NR, msg > "/dev/stderr"
  abort = 1
  exit
}