#!/usr/bin/perl -w -I/opt/eprints2/perl_lib

######################################################################
#
# Attempt to import eprints 1 data into eprints 2.
#
# 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::Session;
use EPrints::ImportXML;

my $session = new EPrints::Session( 1, "cogprints", 2 );
exit( 1 ) unless defined $session;

# The location of the XML files dumped from the v1 system.
local $xmlbase = "/opt/eprints2/local/cogprints/eprintsdump";

# The directory containing disk0/ and any other document dirs 
# from the old system.
local $datapath = "/opt/eprints2/local/cogprints/";

# user to use if the correct user is not found:
local $adminid = 'cjg'; 

import_subjects( $session );

import_users( $session );

import_eprints( $session );

import_subscriptions( $session );

import_documents( $session );



$session->terminate();
exit;






sub import_subjects
{
	my( $session ) = @_;

	my $subds = $session->get_archive()->get_dataset( "subject" );
	my $filename = $xmlbase."/subjects.xml";
	print "Importing from $filename...\n";
	my $count = 0;
	my $info =  { count=>0 }; 
	EPrints::ImportXML::import_file( $session , $filename , \&dealsubject, $subds, $info );
	$count = $info->{count};
	print "Done importing $count subjects from $filename\n";
	print "Adding top level subject.\n";
	EPrints::Subject::create( 
			$session, 
			"subjects",
			{ en=>"Subject Areas" },
			[ "ROOT" ],
			0 );
	print "Reindexing subject dataset to set ancestor data\n";
	$subds->reindex( $session );
	print "Done reindexing\n";
}

sub dealsubject
{
	my( $session , $dataset , $subject, $info ) = @_;

	print "Importing: ".$subject->get_value( "subjectid" )."\n";

	my $parent = $subject->get_data->{"parent"};
	$parent = "subjects" if( $parent eq "ROOT" );

	$subject = EPrints::Subject::create( 
		$session, 
		$subject->get_value( "subjectid" ),
		$subject->get_value( "name" ),
		[ $parent ],
		$subject->get_value( "depositable" ) eq "TRUE" );

	$info->{count}+=1;
}

##############################


sub import_users
{
	my( $session ) = @_;

	$notes->{usermap} = {};
	my $ds = $session->get_archive()->get_dataset( "user" );
	my $filename = $xmlbase."/users.xml";
	print "Importing from $filename...\n";
	my $count = 0;
	my $info =  { count=>0 }; 
	EPrints::ImportXML::import_file( $session , $filename , \&dealuser, $ds, $info );
	print "done users.\n";
}

sub dealuser
{
	my( $session , $dataset , $item, $info ) = @_;

	print "Importing: ".$item->get_value( "username" )."\n";

	my $group = $item->{data}->{groups};
	my $type = "user";
	if( $group eq "Staff" )
	{
		$type = "admin";
	}
	my $user = EPrints::User::create_user( $session, $type );

	my $data = $item->get_data;

	my %osmap = ( 
		"Unspecified"=> "",
		"unspec" => "",
		"Windows/DOS"=> "win",
		"win9x" => "win",
		"UNIX"=> "unix",
		"VMS"=> "vms",
		"Macintosh"=> "mac",
		"Other" => "other");
	$data->{os} = "" if !defined $data->{os};
	if( defined $osmap{$data->{os}} )
	{
		$data->{os} = $osmap{$data->{os}};
	}

	$data->{password} = EPrints::Utils::crypt_password( $data->{passwd} );
 
	$notes->{usermap}->{$data->{username}} = $user->{data}->{userid};

	foreach( keys %{$data} )
	{
		$user->{data}->{$_} = $data->{$_};
	}
	$user->commit;

	$info->{count}+=1;
}


#########################################################

sub import_eprints
{
	my( $session ) = @_;

	my $info =  { maxid=>0 }; 
	foreach( "archive", "inbox", "buffer", "deletion" )
	{
		my $ds = $session->get_archive()->get_dataset( $_ );
		my $file = $_;
		$file = "deletions" if( $_ eq "deletion" );
		my $filename = $xmlbase."/$file.xml";
		print "Importing from $filename...\n";
		EPrints::ImportXML::import_file( $session , $filename , \&dealeprint, $ds, $info );
		print "done $_.\n";
	}

	my $sql = "update counters set counter=".($info->{maxid}+1)." where countername='eprintid'";
	$session->get_db->do( $sql );

	print "done eprints.\n";
}

sub dealeprint
{
	my( $session , $dataset , $item, $info ) = @_;

	print "Importing: ".$item->get_value( "eprintid" )."\n";

	# don't want to mangle the origional data.
	my $data = $item->get_data;
	foreach( "eprintid", "succeeds", "commentary", "replacement" )
	{
		next unless EPrints::Utils::is_set( $data->{$_} );
		$data->{$_} =~ m/(\d+)$/;
		$data->{$_} = $1+0;
	}

	$data->{month} = '' if( defined $data->{month} && $data->{month} eq "unspec" );
	$data->{username} = $adminid unless( defined $data->{username} );
	$data->{userid} = $notes->{usermap}->{$data->{username}};
	if( defined $data->{deletiondate} )
	{
		$data->{datestamp} = $data->{deletiondate};
	}
	$data->{replacedby} = $data->{replacement};
	
	my $dir = EPrints::EPrint::_create_directory( $session, $data->{eprintid} );
	$data->{dir} = $dir;

	my $success = $session->get_db()->add_record( $dataset, $data );

	if( $data->{eprintid} > $info->{maxid} )
	{
		$info->{maxid} = $data->{eprintid};
		print "NEW MAX!\n";
	}
}

##########################

sub import_subscriptions
{
	my( $session ) = @_;

	my $ds = $session->get_archive()->get_dataset( "subscription" );
	my $filename = $xmlbase."/subscriptions.xml";
	print "Importing from $filename...\n";
	my $info =  {};
	EPrints::ImportXML::import_file( $session , $filename , \&dealsubscriptions, $ds, $info );
	print "done subscriptions.\n";
}

sub dealsubscriptions
{
	my( $session , $dataset , $item, $info ) = @_;

	my $data = $item->get_data;
	$data->{username} = $adminid unless( defined $data->{username} );
	$data->{userid} = $notes->{usermap}->{$data->{username}};

	my $ordermap = {
		order0 => "byname",
		order1 => "bytitle",
		order2 => "byyear",
		order3 => "byyearoldest"
	};

	if( EPrints::Utils::is_set( $data->{spec} ) )
	{

       # push @parts, $self->{allow_blank}?1:0;
       # push @parts, $self->{satisfy_all}?1:0;
       # push @parts, $self->{order};
       # push @parts, $self->{dataset}->id();
	# 0|1|bytitle|archive|-|ispublished:ANY:EQ:unpub inpress pub
	# [ALL][order0][subjects][comp-sci-speech:brain-img:neuro-ling:psy-ling:ANY][refereed][TRUE][ispublished][]'

		my @bits = split( '\]\[', substr( $data->{spec}, 1, length( $data->{spec} )-2 ) );
		my $sersearch = '1|';
		$sersearch .= (( shift @bits eq "ALL" ) ? "1" : "0")."|";
		$sersearch .= $ordermap->{shift @bits}."|";
		$sersearch .= "archive|-";
		while( scalar @bits )
		{
			my $field = shift @bits;
			my $v = shift @bits;
			next unless EPrints::Utils::is_set( $v );
			my @p = split /:/, $v;
			my $any = "ANY";
			if( $field eq "subjects" )
			{
				$any = pop @p;
			}
			$sersearch .= "|$field:$any:EQ:".join( " ", @p );
		}
	
		$data->{spec} = $sersearch;	
	}

	$data->{subid} = $session->get_db()->counter_next( "subscriptionid" );
	print "$data->{subid}\n";

	# Add the subscription to the database
	my $x = $session->get_db()->add_record( $dataset, $data );
		
}

###################################

sub import_documents
{
	my( $session ) = @_;

	my $ds = $session->get_archive()->get_dataset( "document" );
	my $filename = $xmlbase."/documents.xml";
	print "Importing from $filename...\n";
	my $info =  {};
	EPrints::ImportXML::import_file( $session , $filename , \&dealdocuments, $ds, $info );
	print "done documents.\n";
}

sub dealdocuments
{
	my( $session , $dataset , $item, $info ) = @_;

	my $data = $item->get_data;
	$data->{eprintid} =~ m/(\d+)$/;
	$data->{eprintid} = $1+0;
	my $olddocid = $data->{docid};
	$data->{docid} =~ m/(\d+)-(\d\d)$/;
	my $docpath=$2;
	$data->{docid} = ($1+0)."-".$2;
	$data->{format} = "\L$data->{format}";
	my $eprint = EPrints::EPrint->new( $session, $data->{eprintid} );
	
	my $oldpath = $datapath.'/'.$eprint->get_value( "dir" ).'/'.$olddocid;

	$session->get_db()->add_record(
		$dataset,
		$data );
	my $doc = EPrints::Document->new( $session, $data->{docid} );
	EPrints::Document::_create_directory( $data->{docid}, $eprint );

	my $cmd = 'cp -R '.$oldpath.'/* '.$doc->local_path.'/';
	print "$cmd\n";
	`$cmd`;

	# Make secure area symlink
	my $linkdir = EPrints::Document::_secure_symlink_path( $eprint );
	$doc->create_symlink( $eprint, $linkdir );
}