#!/usr/bin/perl -w
#

if ($ARGV[0] eq "") { $tupel = 1; }
else { $tupel = $ARGV[0]; }

%hash = ();

warn "Getting text...\n";

while (<STDIN>) {
    $test .=  $_ . " ";
}

warn "Text stored.\n";
warn "Filtering real text.\n";

$test = lc($test);
$test =~ s/[.:,;#+*'!"§$%&\/\\()]/ /g;     
$test =~ s/[^a-zäöüß ]//g;     
$test =~ s/ {2,}/ /g;
$test =~ s/ /_/g;

warn "Filter complete.\n";
warn "Counting tuples.\n";

$all = 0;
$i = 0;

while ($i+$tupel <= length($test)) {
    $sub = substr($test, $i, $tupel);
    $base = substr($sub, 0, $tupel-1);
    $successor = substr($sub, $tupel-1, 1);
    $hash{ $base }{ "ALL" }++;
    $hash{ $base }{ $successor }++;
    $all++;
    $i++;
}

warn "Counting complete.\n";

@keys = sort keys %hash;

warn "Sorting complete.\n";

### printing results
print "$tupel $all \n";
foreach $k1 (@keys) {
    print "$k1 ALL $hash{$k1}{'ALL'}\n";
    foreach $k2 (sort keys %{$hash{$k1}} ) {
        print "$k1 $k2 $hash{$k1}{$k2}\n" if ($k2 ne 'ALL');
    }
}