root/trunk/midgard/tools/yamp/sort_by_guid.pl

#!/usr/bin/perl -w

# $Header$

#-----------------------------------------------------------------------------------------------------------
#       File: sort_by_guid.pl
#
#               Parser for xml files exported by repligard.  Sorts the data by GUID number.
#
#       Originally by Tim Higgs (tim@cwa.co.nz). 
#       Also Torben (torben@nehmer.net) and Martin Langhoff (martin@nzl.com.ar)
#
#-----------------------------------------------------------------------------------------------------------

use strict;
use utf8;
use XML::Parser::Expat;

# Get the file name from the command line
my $file = $ARGV[0] || die( &usage );

my %hosts;

# ID of the current record
my $id = "";

# Content for the current record
my $content = "";

my $start = "";
my $end = "";

my $current_element_name = "";

my @element_names = ( "HOST", "PAGE", "PERSON" );

my $parser = new XML::Parser::Expat;

# Set up the handlers for the parser
$parser->setHandlers(   'Start'         => \&start_handler,
                                                'End'           => \&end_handler,
                                                'Default'       => \&default_handler);

open(FILE, $file) || die "Couldn't open file $!\n";

$parser->parse(*FILE);

close(FILE);

open (OUTFILE, ">$file.sorted") || die "Couldn't open $file.sorted $!\n";

print OUTFILE "$start\n"; # if keys( %hosts );
foreach my $key ( sort( keys( %hosts ) ) ){
        print OUTFILE $hosts{$key}."\n";
} 

print OUTFILE "\n$end";
print OUTFILE "\n";


close OUTFILE;

# Only replace the file if we happened to have done something
if (keys %hosts ){
        unlink $file || die "could not delete $file : $!";
        rename ("$file.sorted", "$file") or die "could not rename $file.sorted : $!";
} else {
        # we did nothing, leave the original alone, delete the tmp
        unlink  "$file.sort" || die "could not delete $file.sort : $!";
}

#-----------------------------------------------------------------------------------------------------------
#       Sub routines
#-----------------------------------------------------------------------------------------------------------

sub start_handler {
        my ($p, $element, %attributes) = @_;

        if( uc( $element ) eq "DATABASE" ){
                $start = $content.$p->recognized_string();
        }
        elsif( defined( $attributes{"id"} ) ){
                # Get it's ID
                $id = $attributes{"id"};
                $content = $p->recognized_string();
                $current_element_name = $element;
        }
        else{
                $content .= $p->recognized_string();
        }
}


sub end_handler {
        my ($p, $element) = @_;

        if( uc( $element ) eq "DATABASE" ){
                $end = $content.$p->recognized_string();
        }
        elsif( uc( $element ) eq uc( $current_element_name ) ){
                $hosts{$id} = $content.$p->recognized_string();
                $content = "";
                $id = "";
                $current_element_name = "";
        }
        else{
                $content .= $p->recognized_string();
        }
} 

sub default_handler {
        my ($p, $string) = @_;
        $content .= $string;
} 


sub usage {
        print<<EOP;
sort_by_guid.pl <filename>
EOP
}

#-----------------------------------------------------------------------------------------------------------