#!/usr/bin/perl -I/opt/eprints/perl_lib

######################################################################
#
# dump an eprints 1 archive.
#
# This script is not an official part of the EPrints release but
# is still distributed under the terms of the GNU Public License.
#
######################################################################


#use EPrints::Database;
#use EPrints::Deletion;
#use EPrints::EPrint;
#use EPrints::HTMLRender;
use EPrints::Session;
#use EPrints::Subject;
use EPrints::SearchExpression;
use EPrints::MetaInfo;
use XML::Writer;

use strict;
#
#
#
# New session
my $session = new EPrints::Session( 1 );

my $dir = $EPrintSite::SiteInfo::local_root. "/eprintsdump" ;
unless( -d $dir ) { `mkdir $dir`; }
print $dir."\n";

my $table;
foreach $table ( "subscriptions", "subjects", "users", "documents", 
	"deletions", "archive", "buffer", "inbox" )
{
	print "Dumping $table\n";	
 	my $output = new IO::File(">$dir/$table.xml");
	print $output '<?xml version="1.0" encoding="iso-8859-1" ?>'."\n";


	my $writer = new XML::Writer(
		OUTPUT => $output,
                DATA_MODE => 1,
                DATA_INDENT => 2 );

	my @fields;

	if( $table eq "subscriptions" )
	{
		@fields = EPrints::MetaInfo::get_subscription_fields();
	}
	elsif( $table eq "subjects" )
	{
		@fields = EPrints::MetaInfo::get_subject_fields();
	}
	elsif( $table eq "users" )
	{
		@fields = EPrints::MetaInfo::get_user_fields();
	}
	elsif( $table eq "documents" )
	{
		@fields = EPrints::MetaInfo::get_document_fields();
	}
	elsif( $table eq "deletions" )
	{
		@fields = EPrints::MetaInfo::get_deletion_fields();
	}
	elsif( $table eq "archive" || $table eq "buffer" || $table eq "inbox" )
	{
 		@fields = EPrints::MetaInfo::get_all_eprint_fields();
	}
	else
	{
		die "unknown table";
	}
	my @cols = ();
	foreach( @fields ) { push @cols, $_->{name}; }
	
	my $searchexp = new EPrints::SearchExpression( 
			$session,
			$table,
			1,
			1,
			[],
			undef,
			undef );
	
	$writer->startTag( "eprintsdata", "name"=>$table );
	
	my $results = $searchexp->do_raw_search(\@cols);
	my $row;
	foreach $row ( @{$results} )
	{
		$writer->startTag( "record" );
		foreach( @fields ) {
			my $value = shift @{$row};
			my @values = ( $value );
			if( $_->{type} eq "multiurl" )
			{
				@values = split /\s+/ , $value;
			}
			elsif( $_->{type} eq "name" || $_->{type} eq "set" ||
				$_->{type} eq "subjects" )
			{
				$value=~s/^://;
				$value=~s/:$//;
				@values = split /:/ , $value;
			}
			foreach $value ( @values )
			{
				if( $_->{type} eq "name" )
				{
					my( $family, $given ) = split /,/ , $value;
					$writer->startTag( 
						"field" , 
						"name"=>"$_->{name}" );
                        		$writer->dataElement( 
						"part" , 
						&cleanup($family) , 
						"name"=>"family" );
                        		$writer->dataElement( 
						"part" , 
						&cleanup($given) , 
						"name"=>"given" );
                        		$writer->endTag( "field" );
				}
				else
				{
	                		$writer->dataElement( 
						"field" , 
						&cleanup($value) , 
						"name"=>"$_->{name}" );
				}
			}
		}
		$writer->endTag( "record" );
	}

	$writer->endTag( "eprintsdata" );
        $writer->end();

        $output->close();
}




$session->terminate();


sub cleanup
{
	my( $text ) = @_;

	# lost all chars < 32 except LF and CR
	$text =~ s/[\x00-\x09\x0b\x0c\x0e-\x1f]//g;

	return $text;
}