diff -r eprints-2.3.4/CHANGELOG eprints-2.3.6/CHANGELOG
1941a1942,1944
> ======================================================================
> 06/07/2004 - EPRINTS-2.3.4
> ======================================================================
1942a1946,1988
> 2004-08-06 cjg
> 
>  - Added Hungarian chars to freetext index map
>  - New options to indexer --once, --notdaemon, --loglevel, --rollcount
>  - Indexer now creates log files
>  - Commented-out un-needed command in SearchExpression
>  - Fixed xmlns filtering so it only occurs on XHTML output, not XML
>  - Fixed a call in Config to use AnApache, so it'll work with v1.3 & 2
>  - Fixed a bug in MetaField/Set; it didn't show the currently 
>     selected value in any/all menus.
>  - documents_url option removed. It's hardwired to /archive/ in lots
>     of places.
>  - Added support for short URLs for records, so you can use
>     /33/ instead of /archive/00000033/
>  - added config option 'shorturl' to make the short version the
>     canonnical one. 
> 
> 2004-08-06 cjg
> 
>  - Added code to SQL prepare and do so that it tries to reconnect
>     if the mysql connection goes away (not yet for 'execute' however)
>  - Modified Fulltext field type to cache extracted index codes to
>     reduce CPU load.
> 
> 2004-08-06 cjg
> 
>  - Fixed spelling mistake in Fulltext.pm
>  - Made noxmlns actually cause/not cause xmlns to be stripped in
>     XML to_string (bug fix)
> 
> ======================================================================
> 06/08/2004 - EPRINTS-2.3.5
> ======================================================================
> 
> 2004-08-09 cjg
> 
>  - Fixed a bug introduced last version which failed to render buttons
>     which id's started with "_". (Such as those on the staff eprint
>     edit page.)
> 
> ======================================================================
> 09/08/2004 - EPRINTS-2.3.6
> ======================================================================
diff -r eprints-2.3.4/VERSION eprints-2.3.6/VERSION
1,2c1,2
< 2.3.4
< EPrints 2.3.4 (Chesed) [Born on 2004-07-06]
---
> 2.3.6
> EPrints 2.3.6 (Tiphareth) [Born on 2004-08-09]
diff -r eprints-2.3.4/bin/configure_archive eprints-2.3.6/bin/configure_archive
90c90
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/create_tables eprints-2.3.6/bin/create_tables
88c88
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/create_user eprints-2.3.6/bin/create_user
122c122
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/erase_archive eprints-2.3.6/bin/erase_archive
112c112
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/explain_sql_tables eprints-2.3.6/bin/explain_sql_tables
82c82
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/export_hashes eprints-2.3.6/bin/export_hashes
105c105
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/force_config_reload eprints-2.3.6/bin/force_config_reload
89c89
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/generate_abstracts eprints-2.3.6/bin/generate_abstracts
87c87
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/generate_apacheconf eprints-2.3.6/bin/generate_apacheconf
120c120
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/generate_static eprints-2.3.6/bin/generate_static
123c123
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/generate_views eprints-2.3.6/bin/generate_views
90c90
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
484c484
< 			print FILE EPrints::XML::to_string( $page );
---
> 			print FILE EPrints::XML::to_string( $page, undef, 1 );
577c577
< 		print FILE EPrints::XML::to_string( $page );
---
> 		print FILE EPrints::XML::to_string( $page, undef, 1 );
diff -r eprints-2.3.4/bin/import_subjects eprints-2.3.6/bin/import_subjects
42a43,44
> Without the B<--force> option, this script asks for confirmation before actually erasing anything.
> 
89a92,95
> =item B<--force>
> 
> Don't ask before making the changes.
> 
167c173
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/indexer eprints-2.3.6/bin/indexer
1c1
< #!/usr/bin/perl  -I/opt/ep2stable/perl_lib
---
> #!/usr/bin/perl  -I/opt/ep2stable/perl_lib -d:DProf
30c30
< B<indexer> - ???
---
> B<indexer> - Indexing daemon for EPrints
34c34
< B<indexer> start [B<options>] 
---
> B<indexer> start [B<options>]
36c36
< B<indexer> stop [B<options>] 
---
> B<indexer> stop 
38c38
< B<indexer> status [B<options>] 
---
> B<indexer> status
42c42
< ???
---
> This daemon runs in the background and creates index files for all eprints archives. It builds indexes for each archive in turn then starts again.
44c44
< =over 8
---
> Messages and errors are logged to /opt/eprints2/var/indexer.log unless you change the log options. If it appears to be having problems try raising the log level and examining the log.
46c46
< =item B<archiveid> 
---
> Once all the archives have been indexed, the indexer rolls the logs (up to logfile.5) and then starts again. See --rollcount for ways to customise this.
48c48
< The ID of the eprint archive to rehash the documents of.
---
> =over 8
78a79,126
> =item B<--logfile> I<filename>
> 
> Log to I<filename> rather than default indexer log.
> 
> =item B<--loglevel> I<level>
> 
> Set the level of detail to log. Level may be 0-5.
> 
> =over 8
> 
> =item 0
> 
> Do not log anything.
> 
> =item 1
> 
> Report start, stop and errors.
> 
> =item 2 [default]
> 
> All above, plus report names of archives being indexed and warnings.
> 
> =item 3
> 
> All above, plus report processes starting and stopping and each dataset being indexed.
> 
> =item 4
> 
> All above, plus report details of index process.
> 
> =item 5
> 
> All above, plus report id of every item indexed.
> 
> =back
> 
> =item B<--rollcount> I<number>
> 
> Set the number of once-through logs that should be kept. If set to zero then indexer will never roll the logs but rather just keep writing to the main log.
> 
> =item B<--notdaemon>
> 
> Do not become a daemon, remain attached to the current terminal (errors still go to the log file).
> 
> =item B<--once>
> 
> Only index each archive once rather than loop for ever.
> 
91c139
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
142a191
> 
143a193,197
> my $logfile;
> my $loglevel = 2;
> my $rollcount = 5;
> my $notdaemon = 0;
> my $once = 0;
152c206,211
< 	'quiet' => \$quiet
---
> 	'quiet' => \$quiet,
> 	'notdaemon' => \$notdaemon,
> 	'once' => \$once,
> 	'rollcount=s' => \$rollcount,
> 	'logfile=s' => \$logfile,
> 	'loglevel=s' => \$loglevel
164c223,236
< my $pidfile = EPrints::Config::get("var_path")."/indexer.pid";
---
> my $p = {
> 	loglevel => $loglevel+0,
> 	rollcount => $rollcount+0,
> 	daemon => !$notdaemon,
> 	noise => $noise,
> 	once => $once,
> 	logfile => $logfile 
> };
> 
> $p->{pidfile} = EPrints::Config::get("var_path")."/indexer.pid";
> if( !defined $p->{logfile} ) 
> {
> 	$p->{logfile} = EPrints::Config::get("var_path")."/indexer.log";
> }
168c240
< 	if( !-e $pidfile )
---
> 	if( !-e $p->{pidfile} )
174,175c246
< 
< 	my $pid = get_pid( $pidfile, $noise );
---
> 	my $pid = get_pid( $p );
183c254
< 	if( !-e $pidfile )
---
> 	if( !-e $p->{pidfile} )
186c257
< $pidfile does not appear to exist.
---
> $p->{pidfile} does not appear to exist.
194c265
< 	my $pid = get_pid( $pidfile, $noise );
---
> 	my $pid = get_pid( $p );
198c269
< 		die "Could not find PID in $pidfile.  Weird. Better kill it by hand.";
---
> 		die "Could not find PID in $p->{pidfile}.  Weird. Better kill it by hand.";
202c273
< 	print "Sending TERM signal to $pid\n" if( $noise > 1 );
---
> 	print "Sending TERM signal to $pid\n" if( $p->{noise} > 1 );
210c281
< 		if( !-e $pidfile )
---
> 		if( !-e $p->{pidfile} )
212c283
< 			print "...Killed $pid\n" if( $noise > 1 );
---
> 			print "...Killed $pid\n" if( $p->{noise} > 1 );
216c287
< 		print "tick\n" if( $noise > 2 );
---
> 		print "tick\n" if( $p->{noise} > 2 );
222c293
< then removing $pidfile
---
> then removing $p->{pidfile}
242c313
< if( !$force && -e $pidfile )
---
> if( !$force && -e $p->{pidfile} )
244c315
< 	my $pid = get_pid( $pidfile, $noise );
---
> 	my $pid = get_pid( $p );
260c331
<  * delete the PID file: $pidfile 
---
>  * delete the PID file: $p->{pidfile} 
267,270d337
< 
< print "Becoming a daemon. Go me.\n" if( $noise > 1 );
< close STDIN;
< close STDOUT;
272,273c339,349
< exit if fork;
< exit if fork;
---
> if( $p->{loglevel} > 0 )
> {
> 	open( STDERR, ">>$p->{logfile}" ) || die "Error opening $p->{logfile}: $!";
> 	select( STDERR );
> 	$| = 1;
> 
> 	rolllogs( $p );
> 	indexlog();
> 	indexlog();
> 	indexlog( "**** Indexer starting..." );
> }
275a352,360
> if( $p->{daemon} )
> {
> 	indexlog( "**** Becoming Daemon" );
> 	close STDIN;
> 	close STDOUT;
> 	exit if fork;
> 	exit if fork;
> }
> 
277d361
< my $kid;
280,289c364,365
< 	print "Unlinking $pidfile\n" if( $noise > 1 );
< 	unlink( $pidfile ) || die( "Can't unlink $pidfile" );
< 	if( defined $kid )
< 	{
< 		print "Sending TERM signal to junior: $kid\n" if( $noise > 1 );
< 		kill 15, $kid;
< 	}
< 	
< 	print "indexer snr exiting $$\n" if( $noise > 1 );
< 	exit;
---
> 	indexlog( "*** TERM signal received" ) if( $p->{loglevel} > 1 );
> 	stopindexer( $p );
293c369,370
< open( PID, ">$pidfile" ) || die "Can't open $pidfile for writing";
---
> 
> open( PID, ">>$p->{pidfile}" ) || die "Can't open $p->{pidfile} for writing: $!";
303c380
< 
---
> indexlog( "** Indexer control process started with process ID: $$" ) if( $p->{loglevel} > 2 );
306a384,385
> 	indexlog() if( $p->{loglevel} > 0 );
> 	indexlog( "*** Starting indexing" ) if( $p->{loglevel} > 1 );
320a400
> 			indexlog( "*** Will not index '$arc_id': indexing disabled by config" ) if( $p->{loglevel} > 1 );
325,329c405,408
< 		print "\n.$$.. - START.outer\n" if( $noise > 1);
< 		$kid = fork();
< 		my $cmd = "/usr/bin/renice 8 $$";	
< 		`$cmd`;
< 		if( $kid == 0 )
---
> 		indexlog( "*** Staring index of archive '$arc_id'" ) if( $p->{loglevel} > 1 );
> 
> 		$p->{kid} = fork();
> 		if( $p->{kid} == 0 )
330a410,412
> 			indexlog( "** Worker process started: $$" ) if( $p->{loglevel} > 2 );
> 			my $cmd = "/usr/bin/renice 8 $$";	
> 			`$cmd`;
334c416
< 				print "indexer jnr exiting $$\n" if( $noise > 1 );
---
> 				indexlog( "** Worker process terminated: $$" ) if( $p->{loglevel} > 2 );
339,340c421
< 			print "\n.$$.. - START.inner\n" if( $noise > 1);
< 			$session = new EPrints::Session( 1 , $arc_id , $noise );
---
> 			$session = new EPrints::Session( 1 , $arc_id , $p->{noise} );
345c426
< 				print "dataset: $arc_id .. $ds_id\n" if( $noise > 1);
---
> 				indexlog( "** Indexing dataset '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 2 );
349c430
< 					print STDERR "Error opening session: $arc_id\n";
---
> 					indexlog( "**** Error starting eprints session on '$arc_id'" ) if( $p->{loglevel} > 0 );
351a433,448
> 				my $ilog;
> 				my $elog;
> 				if( $p->{loglevel} > 1 )
> 				{
> 					$elog = eval 'sub { my( $txt ) = @_; indexlog( "*** ".$txt ); }';
> 				}
> 				else
> 				{
> 					$elog = sub {;};
> 				}
> 
> 				if( $p->{loglevel} > 4 )
> 				{
> 					$ilog = \&indexlog;
> 				}
> 				
354c451,453
< 						$session->get_archive->get_dataset( $ds_id ) );
---
> 						$session->get_archive->get_dataset( $ds_id ),
> 						$ilog,
> 						$elog );
355a455
> 				indexlog( "* Creating temporary index tables for '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 3 );
356a457
> 				indexlog( "* Creating index for '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 3 );
357a459
> 				indexlog( "* Installing index for '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 3 );
358a461
> 				indexlog( "* Cleaning up after '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 3 );
360a464
> 				indexlog( "* Done indexing dataset '$ds_id' of '$arc_id'" ) if( $p->{loglevel} > 3 );
363a468
> 			indexlog( "** Sub process exiting: $$" ) if( $p->{loglevel} > 2 );
365d469
< 			print "\n.$$.. - END.inner\n" if( $noise > 1 );
369c473,475
< 		undef $kid;
---
> 		undef $p->{kid};
> 
> 		indexlog( "*** Done indexing archive '$arc_id'" ) if( $p->{loglevel} > 1 );
371d476
< 		print "\n.$$.. - END.outer\n" if( $noise > 1 );
372a478,508
> 	indexlog( "*** Done indexing all archives" ) if( $p->{loglevel} > 1 );
> 	indexlog() if( $p->{loglevel} > 0 );
> 
> 	stopindexer($p) if( $p->{once} );
> 
> 	rolllogs( $p );
> }
> 
> 
> ####################################################################################
> 
> sub rolllogs
> {
> 	my( $p ) = @_;
> 
> 	return if( $p->{loglevel} <= 0 );
> 	return if( $p->{rollcount} <= 0 );
> 
> 	indexlog( "** End of log. Closing and rolling." ) if( $p->{loglevel} > 2 );
> 	for( my $n = $p->{rollcount}; $n > 0; --$n )
> 	{
> 		my $src = $p->{logfile};	
> 		if( $n > 1 ) { $src.='.'.($n-1); }
> 		next unless( -f $src );
> 		my $tgt = $p->{logfile}.'.'.$n;
> 		rename( $src, $tgt ) || warn "Error renaming: $!";
> 	}
> 	close STDERR;
> 	open( STDERR, ">>$p->{logfile}" ) || warn "Error opening: $p->{logfile}: $!";
> 	select( STDERR );
> 	$| = 1;
378c514
< 	my( $pidfile, $noise ) = @_;
---
> 	my( $p ) = @_;
380,381c516,517
< 	print "Reading $pidfile\n" if( $noise > 1 );
< 	open( PID, $pidfile ) || die( "Could not open $pidfile" );
---
> 	print "Reading $p->{pidfile}\n" if( $p->{noise} > 1 );
> 	open( PID, $p->{pidfile} ) || die( "Could not open $p->{pidfile}: $!" );
394a531,562
> }
> 
> sub indexlog
> {
> 	my( $txt ) = @_;
> 
> 	if( !defined $txt )
> 	{
> 		print STDERR "\n";
> 		return;
> 	}
> 
> 	print STDERR "[".localtime()."] ".$txt."\n";
> }
> 
> sub stopindexer
> {
> 	my( $p ) = @_;
> 
> 	indexlog( "** Control process $$ stopping..." ) if( $p->{loglevel} > 2 );
> 	indexlog( "* Unlinking $p->{pidfile}" ) if( $p->{loglevel} > 3 );
> 	unlink( $p->{pidfile} ) || die( "Can't unlink $p->{pidfile}" );
> 	if( defined $p->{kid} )
> 	{
> 		indexlog( "* Sending TERM signal to worker process: $p->{kid}" ) if( $p->{loglevel} > 2 );
> 		kill 15, $p->{kid};
> 	}
> 
> 	indexlog( "** Control process $$ stopped", 1 ) if( $p->{loglevel} > 2 );
> 	indexlog( "**** Indexer stopped" ) if( $p->{loglevel} > 0 );
> 	indexlog() if( $p->{loglevel} > 0 );
> 	exit;
diff -r eprints-2.3.4/bin/rehash_documents eprints-2.3.6/bin/rehash_documents
93c93
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/reindex eprints-2.3.6/bin/reindex
105c105
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/send_subscriptions eprints-2.3.6/bin/send_subscriptions
99c99
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/bin/upgrade eprints-2.3.6/bin/upgrade
91c91
< EPrints Version: 2.3.4
---
> EPrints Version: 2.3.6
diff -r eprints-2.3.4/cgi/users/staff/edit_eprint eprints-2.3.6/cgi/users/staff/edit_eprint
327a328,329
> 	print STDERR "**". $eprint->get_dataset()->id()."**\n";
> 	print STDERR "**". join( '/', (@{$r1}, @{$r2}))."**\n";
diff -r eprints-2.3.4/defaultcfg/ArchiveConfig.pm eprints-2.3.6/defaultcfg/ArchiveConfig.pm
121,123d120
< # URL of document file hierarchy
< $c->{documents_url} = $c->{base_url}."/archive";
< 
135a133,137
> 
> # Use shorter URLs for records. 
> # Ie. use /23/ instead of /archive/00000023/
> $c->{use_short_urls} = 1;
> 
diff -r eprints-2.3.4/install.pl.in eprints-2.3.6/install.pl.in
512c512
< 		print "Unable to install directory: $dir. $!\n";
---
> 		print "Unable to open package directory: $dir. $!\n";
diff -r eprints-2.3.4/perl_lib/EPrints/Config.pm eprints-2.3.6/perl_lib/EPrints/Config.pm
130c130
< 			$r->send_http_header;
---
> 			EPrints::AnApache::send_http_header( $r );
diff -r eprints-2.3.4/perl_lib/EPrints/Database.pm eprints-2.3.6/perl_lib/EPrints/Database.pm
151,159c151
< 	# Connect to the database
< 	$self->{dbh} = DBI->connect( 
< 		build_connection_string( 
< 			dbhost => $session->get_archive()->get_conf("dbhost"),
< 			dbsock => $session->get_archive()->get_conf("dbsock"),
< 			dbport => $session->get_archive()->get_conf("dbport"),
< 			dbname => $session->get_archive()->get_conf("dbname") ),
< 	        $session->get_archive()->get_conf("dbuser"),
< 	        $session->get_archive()->get_conf("dbpass") );
---
> 	$self->connect;
161,166c153
< #	        { PrintError => 0, AutoCommit => 1 } );
< 
< 	if( !defined $self->{dbh} )
< 	{
< 		return( undef );
< 	}
---
> 	if( !defined $self->{dbh} ) { return( undef ); }
173,176d159
< 	if( $session->{noise} >= 4 )
< 	{
< 		$self->{dbh}->trace( 2 );
< 	}
181a165,196
> ######################################################################
> =pod
> 
> =item $foo = $db->connect
> 
> Connects to the database. 
> 
> =cut
> ######################################################################
> 
> sub connect
> {
> 	my( $self ) = @_;
> 
> 	# Connect to the database
> 	$self->{dbh} = DBI->connect( 
> 		build_connection_string( 
> 			dbhost => $self->{session}->get_archive()->get_conf("dbhost"),
> 			dbsock => $self->{session}->get_archive()->get_conf("dbsock"),
> 			dbport => $self->{session}->get_archive()->get_conf("dbport"),
> 			dbname => $self->{session}->get_archive()->get_conf("dbname") ),
> 	        $self->{session}->get_archive()->get_conf("dbuser"),
> 	        $self->{session}->get_archive()->get_conf("dbpass") );
> 
> 	return unless defined $self->{dbh};	
> 
> 	if( $self->{session}->{noise} >= 4 )
> 	{
> 		$self->{dbh}->trace( 2 );
> 	}
> }
> 
1835,1836c1850
< 
< 	if ( !$result ) 
---
> 	if( !$result )
1838a1853,1872
> 		$self->{session}->get_archive()->log( "SQL ERROR (do): ".$self->{dbh}->errstr.' (#'.$self->{dbh}->err.')' );
> 
> 		return undef unless( $self->{dbh}->err == 2006 );
> 
> 		my $ccount = 0;
> 		while( $ccount < 10 )
> 		{
> 			++$ccount;
> 			sleep 3;
> 			$self->{session}->get_archive()->log( "Attempting DB reconnect: $ccount" );
> 			$self->connect;
> 			if( defined $self->{dbh} )
> 			{
> 				$result = $self->{dbh}->do( $sql );
> 				return $result if( defined $result );
> 				$self->{session}->get_archive()->log( "SQL ERROR (do): ".$self->{dbh}->errstr );
> 			}
> 		}
> 		$self->{session}->get_archive()->log( "Giving up after 10 tries" );
> 		return undef;
1865,1866c1899,1900
< 
< 	if ( !$result ) 
---
> 	my $ccount = 0;
> 	if( !$result )
1868a1903,1922
> 		$self->{session}->get_archive()->log( "SQL ERROR (prepare): ".$self->{dbh}->errstr.' (#'.$self->{dbh}->err.')' );
> 
> 		return undef unless( $self->{dbh}->err == 2006 );
> 
> 		my $ccount = 0;
> 		while( $ccount < 10 )
> 		{
> 			++$ccount;
> 			sleep 3;
> 			$self->{session}->get_archive()->log( "Attempting DB reconnect: $ccount" );
> 			$self->connect;
> 			if( defined $self->{dbh} )
> 			{
> 				$result = $self->{dbh}->prepare( $sql );
> 				return $result if( defined $result );
> 				$self->{session}->get_archive()->log( "SQL ERROR (prepare): ".$self->{dbh}->errstr );
> 			}
> 		}
> 		$self->{session}->get_archive()->log( "Giving up after 10 tries" );
> 		return undef;
1896,1897c1950
< 
< 	if ( !$result ) 
---
> 	while( !$result )
1899a1953,1954
> 		$self->{session}->get_archive()->log( "SQL ERROR (execute): ".$self->{dbh}->errstr );
> 		return undef;
diff -r eprints-2.3.4/perl_lib/EPrints/Document.pm eprints-2.3.6/perl_lib/EPrints/Document.pm
603c603,607
< 	my $basepath;
---
> 	my $shorturl = $archive->get_conf( "use_short_urls" );
> 	$shorturl = 0 unless( defined $shorturl );
> 
> 	my $docpath = docid_to_path( $archive, $self->get_value( "docid" ) );
> 
606c610
< 		$basepath = $archive->get_conf( "documents_url" );
---
> 		return $eprint->url_stem.$docpath.'/';
608,614c612,617
< 	else
< 	{
< 		$basepath = $archive->get_conf( "secure_url" );
< 	}
< 	return $basepath . "/" . 
< 		sprintf( "%08d", $eprint->get_value( "eprintid" )) . "/" .
< 		docid_to_path( $archive, $self->get_value( "docid" ) ) . "/";
---
> 
> 	my $url = $archive->get_conf( "secure_url" ).'/';
> 	$url .= sprintf( "%08d", $eprint->get_value( "eprintid" ) );
> 	$url .= '/'.$docpath.'/';
> 
> 	return $url;
diff -r eprints-2.3.4/perl_lib/EPrints/EPrint.pm eprints-2.3.6/perl_lib/EPrints/EPrint.pm
1264,1269c1264,1283
< 	return( 
< 		sprintf( 
< 			"%s/%08d/", 
< 			$self->{session}->get_archive()->get_conf( 
< 							"documents_url" ), 
< 			$self->{data}->{eprintid} ) );
---
> 	my $archive = $self->{session}->get_archive;
> 
> 	my $shorturl = $archive->get_conf( "use_short_urls" );
> 	$shorturl = 0 unless( defined $shorturl );
> 
> 	my $url;
> 	$url = $archive->get_conf( "base_url" );
> 	$url .= '/archive' unless( $shorturl );
> 	$url .= '/';
> 	if( $shorturl )
> 	{
> 		$url .= $self->get_value( "eprintid" )+0;
> 	}
> 	else
> 	{
> 		$url .= sprintf( "%08d", $self->get_value( "eprintid" ) );
> 	}
> 	$url .= '/';
> 
> 	return $url;
diff -r eprints-2.3.4/perl_lib/EPrints/Index.pm eprints-2.3.6/perl_lib/EPrints/Index.pm
71c71
< 	my( $class, $session, $dataset ) = @_;
---
> 	my( $class, $session, $dataset, $logfn, $elogfn ) = @_;
79c79,81
< 		grep_table_tmp => $dataset->get_sql_index_table_name."_grep_tmp" 
---
> 		grep_table_tmp => $dataset->get_sql_index_table_name."_grep_tmp",
> 		logfn => $logfn,
> 		elogfn => $elogfn
89a92,134
> =item $success = $index->log( $message );
> 
> undocumented
> 
> =cut
> ######################################################################
> 
> sub log
> {
> 	my( $self, $message ) = @_;
> 
> 	if( defined $self->{logfn} )
> 	{
> 		&{$self->{logfn}}( $message );
> 	}
> }
> 
> ######################################################################
> =pod
> 
> =item $success = $index->logerror( $message );
> 
> undocumented
> 
> =cut
> ######################################################################
> 
> sub logerror
> {
> 	my( $self, $message ) = @_;
> 
> 	if( defined $self->{elogfn} )
> 	{
> 		&{$self->{elogfn}}( $message );
> 		return;
> 	}
> 
> 	$self->{session}->get_archive->log( $message );
> }
> 
> ######################################################################
> =pod
> 
116a162
> 		$self->logerror( "Table $table still exists. Dropping it now." );
157c203,204
< 		print STDERR "$self->{index_table_tmp} already exists. Indexer exited abnormally or still running?\nDropping table: $self->{index_table_tmp}\n";
---
> 		$self->logerror( "$self->{index_table_tmp} already exists. Indexer exited abnormally or still running?" );
> 		$self->log( "Dropping table: $self->{index_table_tmp}" );
162a210
> 	$self->log( "Creating table: $self->{index_table_tmp}" );
183c231,232
< 		print STDERR "$self->{grep_table_tmp} already exists. Indexer exited abnormally or still running?\nDropping table: $self->{grep_table_tmp}\n";
---
> 		$self->logerror( "$self->{grep_table_tmp} already exists. Indexer exited abnormally or still running?" );
> 		$self->log( "Dropping table: $self->{grep_table_tmp}" );
189a239
> 	$self->log( "Creating table: $self->{grep_table_tmp}" );
220c270,271
< 			print STDERR "$order_table_tmp already exists. Indexer exited abnormally or still running?\nDropping table: $order_table_tmp\n";
---
> 			$self->logerror( "$order_table_tmp already exists. Indexer exited abnormally or still running?" );
> 			$self->log( "Dropping table: $order_table_tmp" );
225a277
> 		$self->log( "Creating table: $order_table_tmp" );
273a326
> 	$info->{indexer}->log( "Indexing: ".$dataset->get_archive->get_id.'.'.$dataset->id.".".$id );
431a485
> 		$self->log( "Installing table: ".$self->{index_table_tmp} );
438c492
< 		$self->{session}->get_archive->log( "Table does not exist to install: ".$self->{index_table_tmp} );
---
> 		$self->logerror( "Table does not exist to install: ".$self->{index_table_tmp} );
443a498
> 		$self->log( "Installing table: ".$self->{grep_table_tmp} );
450c505
< 		$self->{session}->get_archive->log( "Table does not exist to install: ".$self->{grep_table_tmp} );
---
> 		$self->logerror( "Table does not exist to install: ".$self->{grep_table_tmp} );
461c516
< 			$self->{session}->get_archive->log( "Table does not exist to install: ".$order_table_tmp );
---
> 			$self->logerror( "Table does not exist to install: ".$order_table_tmp );
464a520
> 		$self->log( "Installing table: ".$order_table_tmp );
474c530
< 		$self->{session}->get_archive->log( "EPrints::Index::install failed to open\n$statusfile\nfor writing." );
---
> 		$self->errorlog( "EPrints::Index::install failed to open $statusfile for writing." );
651,652c707,708
< # This map is used to convert ASCII characters over
< # 127 to characters below 127, in the word index.
---
> # This map is used to convert Unicode characters
> # to ASCII characters below 127, in the word index.
657a714,715
> 
> 	# Basic latin1 mappings
705c763,770
< 	latin1("�") => "y",	latin1("'") => "" };
---
> 	latin1("�") => "y",	latin1("'") => "",
> 
> 	# Hungarian characters. 
> 	'�' => "o",	
> 	'Ő' => "o",  
> 	'ű' => "u",  
> 	'Ű' => "u",
>  };
diff -r eprints-2.3.4/perl_lib/EPrints/MetaField/Fulltext.pm eprints-2.3.6/perl_lib/EPrints/MetaField/Fulltext.pm
47d46
< 
52,53c51
< 		my $text = $doc->get_text;
< 		push @{$r}, $text;
---
> 		push @{$r}, "_FULLTEXT_:".$doc->get_id;
57a56,115
> sub get_index_codes_basic
> {
> 	my( $self, $session, $value ) = @_;
> 
> 	if( $value !~ s/^_FULLTEXT_:// )
> 	{
> 		return $self->SUPER::get_index_codes_basic( $session, $value );
> 	}
> 	my $doc = EPrints::Document->new( $session, $value );
> 
> 
> 	my $eprint =  $doc->get_eprint;
> 	return( [], [], [] ) unless( defined $eprint );
> 
> 	my $words_file = $eprint->local_path."/". $doc->get_value( "docid" ).".words";
> 	my $indexcodes_file = $eprint->local_path."/". $doc->get_value( "docid" ).".indexcodes";
> 	my @s1 = stat( $words_file );
> 	my @s2 = stat( $indexcodes_file );
> 
> 	if( !defined $s1[9] ) 
> 	{ 
> 		# no words file!
> 		return( [], [], [] );
> 	}
> 
>     if( defined $s2[9] && $s2[9] > $s1[9] )
> 	{
> 		my $codes = [];
> 		unless( open( CODELOG, $indexcodes_file ) )
> 		{
> 			$session->get_archive->log( "Failed to open $indexcodes_file: $!" );
> 		}
> 		else
> 		{
> 			foreach( <CODELOG> ) { chomp; push @$codes, $_; }
> 			close CODELOG;
> 		}
> 		return( $codes, [], [] );
> 	}
> 
> 	$value = $doc->get_text;
> 	my( $codes, $badwords ) = ( [], [] );
> 	if( EPrints::Utils::is_set( $value ) )
> 	{
> 		( $codes, $badwords ) = EPrints::MetaField::Text::_extract_words( $session, $value );
> 	}
> 	
> 	unless( open( CODELOG, ">".$indexcodes_file ) )
> 	{
> 		$session->get_archive->log( "Failed to write to $indexcodes_file: $!" );
> 	}
> 	else
> 	{
> 		print CODELOG join( "\n", @$codes );
> 		close CODELOG;
> 	}
> 		
> 	# does not return badwords
> 	return( $codes, [], [] );
> }
diff -r eprints-2.3.4/perl_lib/EPrints/MetaField/Set.pm eprints-2.3.6/perl_lib/EPrints/MetaField/Set.pm
244c244
< 				value=>$searchfield->get_merge,
---
> 				default=>$searchfield->get_merge,
diff -r eprints-2.3.4/perl_lib/EPrints/Rewrite.pm eprints-2.3.6/perl_lib/EPrints/Rewrite.pm
44c44,46
< 	if( $uri =~ m#^/archive/([0-9]{1,7})($|/[^0-9].*)# )
---
> 	
> 	# shorturl does not (yet) effect secure docs.
> 	if( $uri =~ s#^/secure/([0-9]+)([0-9][0-9])([0-9][0-9])([0-9][0-9])#/secure/$1/$2/$3/$4# )
45a48,59
> 		$r->filename( $archive->get_conf( "htdocs_path" )."/".$uri );
> 		return OK;
> 	}
> 
> 
> 	my $shorturl = $archive->get_conf( "use_short_urls" );
> 	$shorturl = 0 unless( defined $shorturl );
> 
> 	#$uri =~ s#^/archive/([0-9]+)([0-9][0-9])([0-9][0-9])([0-9][0-9])#/archive/$1/$2/$3/$4#;
> 	if( $uri =~ m#^/archive/([0-9]+)(.*)$# )
> 	{
> 		# is a long record url
48c62,73
< 		if( $tail eq "" )
---
> 		my $redir =0;
> 		if( $tail eq "" ) { $tail = "/"; $redir=1 }
> 		
> 		if( $shorturl )
> 		{
> 			# redirect to short form
> 			return redir( $r, sprintf( "%s/%d%s",$urlpath, $n, $tail ) );
> 		}
> 
> 		my $s8 = sprintf('%08d',$n);
> 		$s8 =~ m/(..)(..)(..)(..)/;	
> 		if( length $n < 8 || $redir)
50c75,76
< 			$tail = "/";
---
> 			# not enough zeros, redirect to correct version
> 			return redir( $r, sprintf( "%s/archive/%08d%s",$urlpath, $n, $tail ) );
52,56c78
< 		my $url = sprintf( "%s/archive/%08d%s",$urlpath, $n, $tail );
< 		$r->status_line( "302 Close but no Cigar" );
< 		EPrints::AnApache::header_out( $r, "Location", $url );
< 		$r->send_http_header;
< 		return DONE;
---
> 		$uri = "/archive/$1/$2/$3/$4$tail";
59c81,82
< 	if( $uri =~ s#^/secure/([0-9]+)([0-9][0-9])([0-9][0-9])([0-9][0-9])#/secure/$1/$2/$3/$4# )
---
> 	
> 	if( $uri =~ m#^/([0-9]+)(.*)$# )
61,63c84,93
< 		$r->filename( $archive->get_conf( "htdocs_path" )."/".$uri );
< 		return OK;
< 	}
---
> 		# is a shorturl record url
> 		my $n = $1;
> 		my $tail = $2;
> 		my $redir = 0;
> 		if( $tail eq "" ) { $tail = "/"; $redir = 1; }
> 		if( !$shorturl )
> 		{
> 			# redir to long form
> 			return redir( $r, sprintf( "%s/archive/%08d%s",$urlpath, $n, $tail ));
> 		} 
65c95,103
< 	$uri =~ s#^/archive/([0-9]+)([0-9][0-9])([0-9][0-9])([0-9][0-9])#/archive/$1/$2/$3/$4#;
---
> 		if( ($n + 0) ne $n || $redir)
> 		{
> 			# leading zeros
> 			return redir( $r, sprintf( "%s/%d%s",$urlpath, $n, $tail ) );
> 		}
> 		my $s8 = sprintf('%08d',$n);
> 		$s8 =~ m/(..)(..)(..)(..)/;	
> 		$uri = "/archive/$1/$2/$3/$4$tail";
> 	}
80a119,127
> sub redir
> {
> 	my( $r, $url ) = @_;
> 
> 	$r->status_line( "302 Close but no Cigar" );
> 	EPrints::AnApache::header_out( $r, "Location", $url );
> 	$r->send_http_header;
> 	return DONE;
> } 
diff -r eprints-2.3.4/perl_lib/EPrints/SearchExpression.pm eprints-2.3.6/perl_lib/EPrints/SearchExpression.pm
1453c1453
< 	my $conditions = $self->get_conditions;
---
> 	#my $conditions = $self->get_conditions;
diff -r eprints-2.3.4/perl_lib/EPrints/Session.pm eprints-2.3.6/perl_lib/EPrints/Session.pm
137a138
> 	$mode = 0 unless defined( $mode );
1246c1247,1248
< 		next if( $_ =~ m/^_/ );
---
> 		next if( $button_id eq '_class' );
> 		next if( $button_id eq '_order' );
1513c1515
< 	foreach( %INPUT_FORM_DEFAULTS )
---
> 	foreach( keys %INPUT_FORM_DEFAULTS )
1846c1848
< 	print EPrints::XML::to_string( $self->{page} );
---
> 	print EPrints::XML::to_string( $self->{page}, undef, 1 );
diff -r eprints-2.3.4/perl_lib/EPrints/XML.pm eprints-2.3.6/perl_lib/EPrints/XML.pm
434c434
< =item $string = EPrints::XML::to_string( $node, [$enc] )
---
> =item $string = EPrints::XML::to_string( $node, [$enc], [$noxmlns] )
439a440,442
> If $stripxmlns is true then all xmlns attributes are removed. Handy
> for making legal XHTML.
> 
450c453
< 	my( $node, $enc ) = @_;
---
> 	my( $node, $enc, $noxmlns ) = @_;
470c473
< 			next if( $name =~ m/^xmlns/ );
---
> 			next if( $noxmlns && $name =~ m/^xmlns/ );
501c504
< 				push @n, to_string( $kid );
---
> 				push @n, to_string( $kid, $enc, $noxmlns );
510c513
< 			push @n, to_string( $kid );
---
> 			push @n, to_string( $kid, $enc, $noxmlns );
517c520
< 		#push @n, $docType->toString, "\n";, to_string( $elem );
---
> 		#push @n, $docType->toString, "\n";, to_string( $elem , $enc, $noxmlns);
682c685
< 		print XMLFILE EPrints::XML::to_string( $node, "utf-8" );
---
> 		print XMLFILE EPrints::XML::to_string( $node, "utf-8", 1 );