#!/usr/bin/perl

####################################
#
# TRfinder2txt -f html-file -o text-file
#
# Reads Tandem Repeats Finder output as .dat file and writes to text file as tab-delimited
# Authors: C. Grunau,  30 mars 2011
#####################################

use diagnostics;
use strict;
use Carp;
use FileHandle;
use File::Path;
use File::Basename;
use Getopt::Std;

my @inputlist         = ();
my @temp = ();
my $FILE              = FileHandle->new();
my $seqname = "";
my $i = 0;
my $j = 0;
my $current = "";
my $header  = "ID\tStart\tEnd\tPeriod_size\tCopy_number\tConsensus_size\tPercent_matches\tPercent_indels\tScore\tA\tC\tG\tT\tEntropy_\(0\-2\)\tTR_sequence\tComplete_sequence";

$| = 1;                          #print immediately

use vars qw( $opt_f $opt_o );
# Command line parsing 
#
getopts('f:o:');
#
# Read input file and split into fasta sequences on the fly
#
#replace Mac and PC CRLF by UNIX LF
system ("perl -pi -e 's/\r/\n/g' $opt_f");
open($FILE,$opt_f) || croak(sprintf("Cannot open file \"%s\"",$opt_f));
while ( <$FILE> ) 
     {
    chomp;
    if ( /^Sequence:\s+(\S+)/) 
         {
         $current=$1;
         } 
       elsif(/^\d+/)
         	{
         	push(@inputlist,$current." ".$_)
         	}
       }
unshift (@inputlist, $header);
close($FILE) || croak(sprintf("Cannot close open file \"%s\"",$opt_f));

print "printing to ".$opt_o."\n";

open($FILE,">".$opt_o) || croak(sprintf("Cannot open file \"%s\"",$opt_o));
for $i (0 .. $#inputlist)
         {
         #print [$i]."\n";
         @temp = ();
         @temp=split('\s+',$inputlist[$i]);
         my $j = 0;

         for $j (0 .. $#temp)
                {
                print $FILE $temp[$j]."\t";
                }
                print $FILE "\n";
         }
         
close ($FILE);