#!/usr/bin/perl

open(FILE, "$ARGV[0]")|| die "Can not open $ARGV[0]!!\n";
#open(OUT, ">>$ARGV[1]")|| die "Can not open $ARGV[1]!!\n";
%id = ();
$org = lc($ARGV[2]); 
#print "Organism: $org"; 

while(<FILE>){
	chomp($_);
	my @field = split(" ", $_);
	if($org eq "grape"){
		$id{$field[0]}++;
		#if($id{$field[0]}){print "$field[0]\t",$id{$field[0]},"\n";}#else{$id{$field[0]}++;}
	}
	elsif($org eq "arabidopsis"){
		$id{$field[1]}++;
		#if($id{$field[1]}){print "$field[1]\t",$id{$field[1]},"\n";}#else{$id{$field[1]}++;}
	}
	elsif($org eq "moss"){
		$id{$field[2]}++;
		#if($id{$field[2]}){print "$field[2]\t",$id{$field[2]},"\n";}else{$id{$field[2]}++;}
	}
	elsif($org eq "populus"){ 
		#if($id{$field[3]}){print "$field[3]\t",$id{$field[3]},"\n";}#else{$id{$field[3]}++;}
		$id{$field[3]}++;
	}
	elsif($org eq "rice"){
		#if($id{$field[4]}){print "$field[4]\t",$id{$field[4]},"\n";}else{$id{$field[4]}++;}
		$id{$field[4]}++;
	}
}

#foreach $key (sort{ $a <=> $b }(keys(%id))) {
#	print "$key\t$id{$key}\n";
#}


open(FILE2, "$ARGV[1]")|| die "Can not open $ARGV[1]!!\n";
$org = $ARGV[2];
$/ = "\n>";
while(<FILE2>){
	chomp($_);
	my @line = split("\n", $_);
	if($line[0] =~ /(INTRON\_\d+)\s*(\d+[a-z]*\_N.{1}\_\d+)/ig){
		my $a = $1."-".$2;
		if($id{$a}){
			$count++;
			print "> $1 $2 $org\n$line[1]\n\n";
		}
	}
}
#$enum = keys(%id);
#print "$enum\t$count\n";