#!/usr/bin/perl -w

#============================================================
# correct input list according to PSI2 original tier information
#===============================================================

use Getopt::Long;

				# default options
$opt_help = '';
$opt_debug = 0;
$opt_verbose = 0;

$name_scr = $0;
$name_scr =~ s/.*\///g;

$file_psi = 'psi.list';
@tiers = qw(tier1 tier2 tier3);

$Lok = GetOptions ('debug!' => \$opt_debug,
		   'i=s' => \$file_in,
		   'p=s' => \$file_psi,
		   'o=s' => \$file_out,
		   'help'  => \$opt_help,
		   'verbose!' => \$opt_verbose,
		   );

if ( ! $Lok ) {
    print STDERR "*** ERROR: Invalid arguments found, -h for help\n";
    &usage();
    exit(1);
}


if ( $opt_help ) {
    &usage();
    exit(1);
}

if ( ! $file_in ) {
    print STDERR
	"*** ERROR: input file not specified\n";
    &usage();
    exit(1);
}

if ( ! -s $file_in ) {
    print STDERR
	"*** ERROR: input file '$file_in' not found, exiting..\n";
    &usage();
    exit(1);
}

				# end of option/sanity check

$fh_out = 'STDOUT';
if ( $file_out ) {
    $fh_out = 'OUT';
    open ( $fh_out, ">$file_out") or die "cannot write to $file_out:$!";
}

if ( $opt_verbose ) {
    print $fh_out
	"# input: $file_in\n",
	"# generated by: $name_scr\n";
}

($list_psi{tier1},$list_psi{tier2},$list_psi{tier3}) = 
    &read_input($file_psi);
foreach $tier ( @tiers ) {
    foreach $id ( @{$list_psi{$tier}} ) {
        $in_tier{$id} = $tier;
    }
}

($list_in{tier1},$list_in{tier2},$list_in{tier3}) = 
    &read_input($file_in);

foreach $tier_in ( @tiers ) {
    foreach $id ( @{$list_in{$tier_in}} ) {
	$tier_out = $in_tier{$id};
	if ( ! $tier_out ) {
	    print STDERR "*** tier info for $id not found, skip\n";
	    next;
	}
	if ( $tier_out eq $tier_in ) {
	    push @{$list_out{$tier_in}},$id;
	} else {
	    push @{$in_out{$tier_in}{$tier_out}},$id;
	    print STDERR "$id belongs to $tier_out instead of $tier_in\n";
	}
    }
}

push @{$list_out{tier1}}, @{$in_out{tier2}{tier1}} if $in_out{tier2}{tier1};
push @{$list_out{tier1}}, @{$in_out{tier3}{tier1}} if $in_out{tier3}{tier1};

unshift @{$list_out{tier2}}, @{$in_out{tier1}{tier2}} if $in_out{tier1}{tier2};
push @{$list_out{tier2}}, @{$in_out{tier3}{tier2}} if $in_out{tier3}{tier2};

unshift @{$list_out{tier3}}, @{$in_out{tier2}{tier3}} if $in_out{tier2}{tier3};
unshift @{$list_out{tier3}}, @{$in_out{tier1}{tier3}} if $in_out{tier1}{tier3};

foreach $tier ( @tiers ) {
    print $fh_out "# $tier\n";
    foreach $id ( @{$list_out{$tier}} ) {
	print $fh_out $id,"\n";
    }
}
close $fh_out;


exit;


sub read_input {
    my ( $file ) = @_;
    my ( $fh,$line,$ct_tier );

    my %list = ();
    $ct_tier = 0;
    open ($fh,$file) or die "cannot open $file:$!";
    while ( $line=<$fh> ) {
	$line =~ s/\s+//g;
	next if ( $line =~ /^$/ );
	if ( $line =~ /^\#/ ) {
	    $ct_tier++;
	    last if ( $ct_tier > 3 );
	} elsif ( $line =~ /\w+/ ) {
	    next if ( $ct_tier == 0 );
	    $line =~ s/\s+.*//g;
	    push @{$list{"tier$ct_tier"}},$line;
	}
    }
    close $fh;

    return ($list{tier1},$list{tier2},$list{tier3});
}


sub usage {

    print STDERR
	"$name_scr: correct input list according to PSI2 original tier information \n",
	"Usage: $name_scr [options] -i in_file \n",
	"  Opt:  -h            print this help\n",
	"        -i <file>     input file (REQUIRED)\n",
	"        -p <file>     input PSI2 list file (REQUIRED, default=$file_psi)\n",
	"        -o <file>     output file (default STDOUT)\n",
	"        --(no)debug   print debug info(default=nodebug)\n";
}