#!/usr/bin/perl

# this program figures tri-grams for a file

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

use strict;

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

my ($buf, $first, $second, $third, @answer, $mod_length,
    $kara, $length_of_file, $openme, $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++;
 }


$mod_length = $length_of_file;

if (($length_of_file % 3) != 0) {
   $mod_length = $length_of_file + (3 - ($length_of_file % 3));
 }
 
close (IN);

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



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

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

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

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

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


for ($count = 3; $count <= $mod_length; $count++) {  
  $first = $second;

  $second = $third;

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


$count = 0;

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

print("\n");

close(IN);
close(O);