#!/usr/bin/perl -w

###############################################
# pe-sync-2-files.pl # this is a stripped-down version of pe-sync-directory.pl
# John Garbe
# February 2012
#
# Paired-end read file synchronization script: checks if left and right 
# reads are in the same order.
# This script checks the two files specificd on the command line. 
# The "quick" option will not report the percent of out-of-sync reads
# for many failed files, but will run much faster.
#
# This script should run fine on linux systems or solaris 
# (which is what the project space fileserver runs)
#
# This script can handle Casava 1.8.0 style read IDs, and pre 1.8.0 style ids.
# Other types of read ID formats will cause a terminal error. 
#
###############################################

die "Usage: pe-sync-directory.pl [quick] left_fastqfile right_fastqfile\n" unless ($#ARGV > 0);

# determine if we're running on solaris in order to change system calls
$output = `perl -v`;
$solaris = 0;
$solaris = 1 if ($output =~ /solaris/);
if ($solaris) {
#    print STDERR "Running on Solaris\n";
}

# handle "quick" option
$quick = 0; # 1 for quickcheck, 0 for thorough check
if (defined($ARGV[0])) { 
    if ($ARGV[0] eq "quick") {
	$quick = 1;
	shift @ARGV;
    }
}
print STDERR "QUICK CHECK enabled\n" if ($quick);

$filename_r1 = $ARGV[0];
$filename_r2 = $ARGV[1];

# identify read id format type
if ($solaris) {
    $readid = `head -1 $filename_r1`;
} else {
    $readid = `head -n 1 $filename_r1`;
}
chomp $readid;
if ($readid =~ /^@\S+\/[12]$/) { # @ at the start of the line followed by non-whitespace, a /, a 1 or 2, the end of the line
    $id_type = 1;
    print STDERR "Casava 1.7 read id style\n"; # TESTING
} elsif ($readid =~ /^@\S+\W[12]\S+$/) { # @ at the start of the line followed by non-whitspace, a space, a 1 or 2, non-whitespace
    $id_type = 2;
    print STDERR "Casava 1.8 read id style\n"; # TESTING
} else {
    print STDERR "Cannot determine read id style\n";
    print STDOUT "Unknwon id style: $readid\n";
    exit 1;
}

chomp $filename_r1;
chomp $filename_r2;

if ($quick) {
    $failure = &quickcheck($filename_r1, $filename_r2); # quickie check
    print STDERR "FAILED quickcheck\n" if ($failure);
}
# If we aren't doing the quick check, or the quick check passed, do the slow check
if (!($quick) || !($failure)) {
    $failure = &percent($filename_r1, $filename_r2); # slow thorough check
}

exit;

######################## subs #########################
# do a quick check to see if the files are bad (many bad files will pass this quick check)
sub quickcheck {
    my ($file1, $file2) = @_;
    if ($solaris) {
	$result1 = `tail -10000l $file1 | head -1`; # grab the line 10,000 lines from the EOF
	$result2 = `tail -10000l $file2 | head -1`;
    } else {
	$result1 = `tail -n 10000 $file1 | head -n 1`;
	$result2 = `tail -n 10000 $file2 | head -n 1`;
    }
    # process the ids depending on id style
    if ($id_type == 1) { # 1 or 2 at end of id
	chomp $result1;
	chomp $result2;
	chop $result1;
	chop $result2;
    } else { # first word is a unique read id, second word is left/right-specific
	($result1) = split ' ', $result1;
	($result2) = split ' ', $result2;
    }
    if ($result1 eq $result2) {
#	print "pass\n";
	return 0;
    } else {
#	print "fail\n";
	return 1;
    }
}


# compare every read id from each file and determine what percent of reads are
# out of sync
sub percent {
    my($file1, $file2) = @_;
    
    open R1, '<', $file1 or die "Couldn't open $file1 for reading:$!";
    open R2, '<', $file2 or die "Couldn't open $file2 for reading:$!";
    
    my $readCount = 0;
    my $failcount = 0;
    while (!eof(R1) and !eof(R2)) {
	$id1 = <R1>;
	$id2 = <R2>;
	next unless ($. % 4 == 1); # check every sequence (4th line in the file)
	# process the ids depending on id style
	if ($id_type == 1) { # 1 or 2 at end of id
	    chomp $id1;  # newline
	    chop $id1;   # last char of the id (should be the "1" or "2")
	    chomp $id2;
	    chop $id2;
	} else {
	    ($id1) = split ' ', $id1;
	    ($id2) = split ' ', $id2;
	}
	$failcount++ unless ($id1 eq $id2);
	$readCount++;
    }
    my $percent = $failcount / $readCount;
    $prettypercent = $percent * 100; # make it percent
    $prettypercent = sprintf("%.2f", $prettypercent);
    if ($failcount > 0) {
	print STDERR "FAILED full check\t$prettypercent% of reads are out-of-sync\n";
	$exitStatus=1;
    } else {
	print STDERR "PASSED full check\n";
	$exitStatus=0;
    }

    return $exitStatus;
}