#!/usr/bin/perl -w

# save_server_dir -server T08 -target T0403
#	looks for a prediction for servers@predictioncenter org for
#	in the log file for the server, and rsyncs the directory to 
#	/projects/compbio/experiments/protein-predict/casp9/decoys/SAM_T08

# Options:
#	-server T08	T02, T06, or T08 (default T08)
#	-target T0403	default (name of current directory)
#	-email servers@predictioncenter.org
#	-dir /projects/compbio/experiments/predict-protein/casp9

use strict;
use Getopt::Long;

my $server="T08";
my $here= `pwd`;
chomp($here);
my @path = split('/', $here);
my $target=$path[scalar(@path)-1];
my $email = "servers\@predictioncenter.org";
my $root_dir = "/projects/compbio/experiments/protein-predict/casp9";

GetOptions(
	  "server=s"	=> \$server
	, "target=s" 	=> \$target
	, "email=s"	=> \$email
	, "dir=s"	=> \$root_dir
	);


# subscript in logfile for the name of the a2m file
my %field_in_log =
	( "T02" => 7
	, "T06" => 8
	, "T08" => 8
	);

my $email_expression = qr/^${email}/;
my $target_expression = qr/${target}[.]a2m/;

my $logfile = "/projects/compbio/.html/SAM_$server/log/logfile";

open(LOGFILE, "<$logfile") || die "ERROR: can't read $logfile\n";


while (<LOGFILE>)
{
    next if ! /${email_expression}/;
    
    # found a line for the casp server
    my @fields = split /\s+/;
    my $a2m = $fields[$field_in_log{$server}];
    next if $a2m !~ /${target_expression}/;
    
    # found a line for this target for the casp server
    
    $a2m =~ /^(.+)${target_expression}/;
    my $from_dir = $1;
    my $to_dir = "$root_dir/$target/decoys/SAM_$server";
    
    print STDERR "Rsyncing $from_dir to $to_dir\n";
    system("ssh silo \"rsync -axu $from_dir $to_dir\" < /dev/null");
}

close LOGFILE;