#!/usr/local/bin/perl
use strict;
use Switch;

#
# downsample.pl by William Pu
#
# perl downsample.pl -i infile -o outfile [-p PROB -l LINES] 
# downsample a file by retaining a line with probability PROB (range 0-1); default = 1 (retain all);
# take the lines in blocks LINES. For example, fastq is in blocks of 2, bowtie is in blocks of 1; default = 1.

my $PROB = 1;
my $LINES = 1;
my $infile = "";
my $outfile = "";

		
# read parameters, assign output variables

if ($ARGV[0] eq "") {
	report_error("usage: downsample.pl -i infile -o outfile [-p PROB -l LINES]\n");
}

my $strip_str = "";
for (my $i=0;$i<@ARGV;$i++) {
	switch($ARGV[$i]) {
		case "-i"	{ $i++; $infile = $ARGV[$i]; }
		case "-o"	{ $i++; $outfile = $ARGV[$i]; }
		case "-p"	{ $i++; $PROB = $ARGV[$i]; }
		case "-l"	{ $i++; $LINES = $ARGV[$i]; }
		else { report_error("invalid parameter " . $ARGV[$i]) }
	}
}

if ($infile eq "" || $outfile eq "") {
	report_error("invalid input or output file\n");
}

print $infile." ".$outfile."\n";
open(INFILE, $infile) or die;
open(OUTFILE,">".$outfile);

my $j;
my $str=<INFILE>;

while ($str) {
	if (rand() < $PROB) {	# keep the block
		for ($j=0; $j<$LINES; $j++) {
			print OUTFILE $str;
			$str=<INFILE>;
		}
	} else {				# discard the block
		for ($j=0; $j<$LINES; $j++) {
			$str=<INFILE>;
		}
	}
}
# close files
close INFILE;
close OUTFILE;
exit;

sub report_error {
	print "@_\n";
	die;
}