#!/usr/bin/perl

# this program figures tri-grams for a file

#  TODO:
#    1. input file on command line
#    2. checking for case where length of file % 3 == 1
#    3. output to file

use strict;

defined($ARGV[1]) or die "Usage: bigram.pl <input file> <output file>\n";

my ($buf, $first, $second, @answer, $kara,
    $length_of_file, $count, $max_length);

$max_length = 100000;

$length_of_file = 0;

open(O,">".$ARGV[1]) or die "Output file open failed: $!\n";

open(IN,"$ARGV[0]") || die "Sorry. Couldn't open: $!\n";

while (read (IN, $buf, 1)) { 
  $length_of_file++;
 }

close (IN);

open(IN,"$ARGV[0]") || die "Sorry. Couldn't open: $!\n";

print ("\nresults from : ");
$kara = <STDIN>;
chomp ($kara);


read(IN, $buf, 1);
$first = ord($buf);

read(IN, $buf, 1);
$second = ord($buf);

$answer[$first][$second]++;


for ($count = 2; $count <= $length_of_file; $count++) {  
  $first = $second;

  read(IN, $buf, 1);
  $second = ord($buf);
  
  if (($count == $length_of_file) && ($second == ord("\n"))) {
     $second = ord(" ");
  }
  
  while ($second == ord("\n") or $second == ord("\t")) {
  read(IN, $buf, 1);
  $second = ord($buf);
  }
  
  $answer[$first][$second]++;
}

$count = 0;

for ($first = ord(" "); $first <= ord("~"); $first++) {  
  for ($second = ord(" "); $second <= ord("~"); $second++) {
     
      if (($answer[$first][$second]) >= $kara)
      {
        $count++;
        print O (chr($first));
	print O (chr($second));
	print O ("\t");
        print O ($answer[$first][$second]);
	
	if ($count == 4) {
	  $count = 0;
	  print O ("\n");
	}
	
	else { print O ("\t");}
    }
  }
}

print("\n");

close(IN);
close(O);