#! /usr/bin/gawk -f 
# Last edited on 2001-01-01 22:57:46 by stolfi

BEGIN {
  abort = -1;
  usage = ( \
    "compute-entropy < INFILE > OUTFILE" \
  );

  split("", ct);
  n = 0;
  totct = 0;
}

(abort >= 0) { exit abort; }

/./ {
  if (NF != 1) { data_error("bad field count"); }
  if ($1 !~ /^[+]*[.0-9]*[0-9][.0-9]*$/) { data_error("bad data format"); }
  ct[n] = $1; totct += $1; n++;
  next;
}

END {
  if (abort >= 0) { exit abort; }
  for (i = 0; i < n; i++) { ct[i] /= totct; }
  entropy = 0;
  for (i = 0; i < n; i++)
    { if (ct[i] > 0) 
        { entropy += -ct[i]*log(ct[i]); }
    }
  entropy /= log(2.0);
  printf "%8.4f\n", entropy;
}


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

function arg_error(msg)
{ 
  printf "%s\n", msg >> "/dev/stderr";
  abort = 1;
  exit 1;
}