#!/usr/bin/perl -w
#
# words.	--crb3 02nov01
# list and count words used in a document.
# case sensitive.
# --crb3 03jun02: -i for case-insensitive treatment
#
#
$chatty=0;
$sep=",";
$srt="b";
$icase=0;

die "usage: words [-i] [-a -n] infile\n  -i: case-insensitive, -a: alpha-sort, -n: count-sort\n" unless defined($ARGV[0]);

while(defined($arg=shift(@ARGV))){
  if(substr($arg,0,1) eq '-'){
    substr($arg,0,1)="";
    if( ($val=substr($arg,0,1)) eq 'i'){
      $icase=1;
    }else{
      $srt=$val;
    }
  }else{
    push(@infiles,$arg);
  }
}

$wordscount=0;
while(defined($infile=shift(@infiles))){
  open(IFIL,"<$infile") or die "can't open infile $infile\n";
  print "$infile...\n" if $chatty;
  while(defined($inline=<IFIL>)){
    chomp $inline;
    next if $inline =~ /^\s*$/;

    foreach $word (split(' ',$inline)){
      $word =~ s/\W+$//;
      $word =~ s/^\W+//;
      next if $word eq "";
      $word = lc($word) if $icase;	# locase em all for case-insensitive
      if(exists $words{$word}){
        $words{$word}++;    
      }else{
        $words{$word}=1;
        $wordscount++;
      }
    }

  }
  close IFIL;
}

if($srt eq "a"){
  @uniquewords = sort { lc($a) cmp lc($b) } (keys %words);
}elsif($srt eq "n"){
  @uniquewords = sort { $words{$b} <=> $words{$a} } (keys %words);
}else{
  @uniquewords = sort { $words{$b} <=> $words{$a} } (sort(keys %words));
}

foreach $word (@uniquewords){
  print "$word$sep$words{$word}\n"; 
}
$in = ($icase ? "in" : "");
print "$wordscount unique words, case-".$in."sensitive.\n";


__END__