$max = 92;	#initialize maximum region sequence length
$min = 3;	#initialize  minimum region sequence length
$regioncounter = 18200; #initialize for total number of records in output.txt file.

if (open(InputOutputFile, 'output.txt')) {
	print "File output.txt opened OK \n";
}
else {
	print "Cannot open file output.txt\n";
	exit;
}

open (TempFile, ">OutputTemp.txt");
open(OutputRICOFile, ">OutputRICO.txt");
open(OutputC45File, ">OutputC45.txt");
@lengtharray = ();		#temp array to keep track of the number of rows for same length region sequences.

for ($loop1 = $min; $loop1 <= $max; $loop1 = $loop1+1) {	#loop for same length sequences
	open(InputOutputFile, 'output.txt');		#open again for each loop, go to beginning of file
	$rowcounter = 0;
	for ($loop2 = 1; $loop2 <= $regioncounter; $loop2=$loop2+1) {	#inner loop to loop through whole file
		$line = <InputOutputFile>;
		chomp ($line);
		if ((length $line) == $loop1) {
			$rowcounter = $rowcounter + 1;			#count number of rows for same length sequences
		    $line1 = join(',', split(//, $line));		#split the stupid thing and join them (to insert ,)
			print TempFile $line1."\n";
		}
	}
	
	$lengtharray[$loop1] = $rowcounter;
}

#temp array to keep track of the number of rows for same length region sequences.
# foreach $element (@lengtharray) {		
	# print $element, "\n";
# }

close(TempFile);
open(InputTempFile, 'OutputTemp.txt');

#write to file = RICO format
for ($loop1 = $min; $loop1 <= $max; $loop1 = $loop1+1) {
	if ($lengtharray[$loop1] > 0) {
		print OutputRICOFile $loop1."\n";		#number of attributes = length of sequence
		print OutputRICOFile $lengtharray[$loop1]."\n";		#number of rows of sequences of same length
		for ($loop2 = 1; $loop2 <= $loop1; $loop2 = $loop2+1) {
			print OutputRICOFile "a,c,d,e,f,g,h,i,k,l,m,n,p,q,r,s,t,v,w,y\n";
		}
		for ($loop2 = 1; $loop2 <= $lengtharray[$loop1]; $loop2 = $loop2+1) {
			$line = <InputTempFile>;
			print OutputRICOFile $line;
		}
		print OutputRICOFile "=======================================\n";
	}
}

close(TempFile);
open(InputTempFile, 'OutputTemp.txt');

#write to file = C45 format
for ($loop1 = $min; $loop1 <= $max; $loop1 = $loop1+1) {
	if ($lengtharray[$loop1] > 0) {
		for ($loop2 = 1; $loop2 <= $lengtharray[$loop1]; $loop2 = $loop2+1) {
			$line = <InputTempFile>;
			if (substr($line, ((length $line) -2), 1) eq 'a') {
				substr($line, ((length $line) -2), 1) = "+a";	
				#replace last nucleotide = length - /n - start at 0
			} else {
				substr($line, ((length $line) -2), 1) = "-a";
			}
			print OutputC45File $line;
		}
		print OutputC45File "=======================================\n";
	}
}

exit;