#!/usr/local/bin/perl

use strict;

my $usage = "$0 psl_input_file [gff_output_file]
Psl files are generated by BLAT and have the following format (tab-separated):

n_match n_mismatch rep_match Ncount QgapCount QgapBases TgapCount TgapBases strand Qname Qsize Qstart Qend Tname Tsize Tstart Tend blockCount blockSizes qStarts tStarts

Last 3 fields are comma-separated lists\n";

my $arg=0;
my $input_file = $ARGV[$arg++] or die $usage;
my $output_file = $ARGV[$arg++];

print "Executing: $0\n";
my %hashChr;
my $chr;
$output_file =~ s/\.gff$//i;
open (INFO, $input_file) or die "file $input_file not found";
while(my $line = <INFO>){
	my ($n_match,$n_mismatch,$rep_match,$Ncount,$QgapCount,$QgapBases,$TgapCount,$TgapBases,$strand,$Qname,$Qsize,$Qstart,$Qend,$Tname,$Tsize,$Tstart,$Tend,$blockCount,$blockSizes,$qStarts,$tStarts)=split(/\t/, $line);
	if(!$chr){
		$chr = $Tname;
		my $gff_file = "$chr.gff";
		if($output_file){
			$gff_file = $output_file . "_" . $gff_file;
		}
		open (OUTPUT, ">$gff_file") or die "cannot create output file";
		print "Writing output to $gff_file\n";
	}
	elsif($chr ne $Tname && !$hashChr{$Tname}){
		$hashChr{$Tname}=1;
		next;
	}
	if($chr ne $Tname){ next; }
	my @blockSize = split(/,/,$blockSizes);
	my @tStart = split(/,/, $tStarts);
	for(my $i=0; $i<@tStart; ++$i){
		my $start = $tStart[$i]+1;
		my $end = $tStart[$i]+$blockSize[$i];
		print OUTPUT "$Qname\tNIA_Mouse_4.0\texon\t$start\t$end\t.\t$strand\t.\n";
	}
}
close INFO;
close OUTPUT;
foreach my $chr (sort keys %hashChr){
	my $gff_file = "$chr.gff";
	if($output_file){
		$gff_file = $output_file . "_" . $gff_file;
	}
	open (OUTPUT, ">$gff_file") or die "cannot create output file";
	print "Writing output to $gff_file\n";
	open (INFO, $input_file) or die "file $input_file not found";
	while(my $line = <INFO>){
		my ($n_match,$n_mismatch,$rep_match,$Ncount,$QgapCount,$QgapBases,$TgapCount,$TgapBases,$strand,$Qname,$Qsize,$Qstart,$Qend,$Tname,$Tsize,$Tstart,$Tend,$blockCount,$blockSizes,$qStarts,$tStarts)=split(/\t/, $line);
		if($chr ne $Tname){ next; }
		my @blockSize = split(/,/,$blockSizes);
		my @tStart = split(/,/, $tStarts);
		for(my $i=0; $i<@tStart; ++$i){
			my $start = $tStart[$i]+1;
			my $end = $tStart[$i]+$blockSize[$i];
			print OUTPUT "$Qname\tNIA_Mouse_4.0\texon\t$start\t$end\t.\t$strand\t.\n";
		}
	}
	close INFO;
	close OUTPUT;
}
exit(0);