1 # BEGIN BPS TAGGED BLOCK {{{
5 # This software is Copyright (c) 1996-2014 Best Practical Solutions, LLC
6 # <sales@bestpractical.com>
8 # (Except where explicitly superseded by other copyright notices)
13 # This work is made available to you under the terms of Version 2 of
14 # the GNU General Public License. A copy of that license should have
15 # been provided with this software, but in any event can be snarfed
18 # This work is distributed in the hope that it will be useful, but
19 # WITHOUT ANY WARRANTY; without even the implied warranty of
20 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
21 # General Public License for more details.
23 # You should have received a copy of the GNU General Public License
24 # along with this program; if not, write to the Free Software
25 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
26 # 02110-1301 or visit their web page on the internet at
27 # http://www.gnu.org/licenses/old-licenses/gpl-2.0.html.
30 # CONTRIBUTION SUBMISSION POLICY:
32 # (The following paragraph is not intended to limit the rights granted
33 # to you to modify and distribute this software under the terms of
34 # the GNU General Public License and is only of importance to you if
35 # you choose to contribute your changes and enhancements to the
36 # community by submitting them to Best Practical Solutions, LLC.)
38 # By intentionally submitting any modifications, corrections or
39 # derivatives to this work, or any other work intended for use with
40 # Request Tracker, to Best Practical Solutions, LLC, you confirm that
41 # you are the copyright holder for those contributions and you grant
42 # Best Practical Solutions, LLC a nonexclusive, worldwide, irrevocable,
43 # royalty-free, perpetual, license to use, copy, create derivative
44 # works based on those contributions, and sublicense and distribute
45 # those contributions and any derivatives thereof.
47 # END BPS TAGGED BLOCK }}}
58 RT::Shredder - Permanently wipeout data from RT
65 rt-shredder --force --plugin 'Tickets=query,Queue="General" and Status="deleted"'
69 RT::Shredder is extension to RT which allows you to permanently wipeout
70 data from the RT database. Shredder supports the wiping of almost
71 all RT objects (Tickets, Transactions, Attachments, Users...).
74 =head2 "Delete" vs "Wipeout"
76 RT uses the term "delete" to mean "deactivate". To avoid confusion,
77 RT::Shredder uses the term "Wipeout" to mean "permanently erase" (or
78 what most people would think of as "delete").
81 =head2 Why do you want this?
83 Normally in RT, "deleting" an item simply deactivates it and makes it
84 invisible from view. This is done to retain full history and
85 auditability of your tickets. For most RT users this is fine and they
86 have no need of RT::Shredder.
88 But in some large and heavily used RT instances the database can get
89 clogged up with junk, particularly spam. This can slow down searches
90 and bloat the size of the database. For these users, RT::Shredder
91 allows them to completely clear the database of this unwanted junk.
93 An additional use of Shredder is to obliterate sensitive information
94 (passwords, credit card numbers, ...) which might have made their way
98 =head2 Command line tools (CLI)
100 L<rt-shredder> is a program which allows you to wipe objects from
101 command line or with system tasks scheduler (cron, for example).
102 See also 'rt-shredder --help'.
105 =head2 Web based interface (WebUI)
107 Shredder's WebUI integrates into RT's WebUI. You can find it in the
108 Admin->Tools->Shredder tab. The interface is similar to the
109 CLI and gives you the same functionality. You can find 'Shredder' link
110 at the bottom of tickets search results, so you could wipeout tickets
111 in the way similar to the bulk update.
114 =head1 DATA STORAGE AND BACKUPS
116 Shredder allows you to store data you wiped in files as scripts with SQL
119 =head3 Restoring from backup
121 Should you wipeout something you did not intend to the objects can be
122 restored by using the storage files. These files are a simple set of
123 SQL commands to re-insert your objects into the RT database.
125 1) Locate the appropriate shredder SQL dump file. In the WebUI, when
126 you use shredder, the path to the dump file is displayed. It also
127 gives the option to download the dump file after each wipeout. Or
128 it can be found in your C<$ShredderStoragePath>.
130 2) Load the shredder SQL dump into your RT database. The details will
131 be different for each database and RT configuration, consult your
132 database manual and RT config. For example, in MySQL...
134 mysql -u your_rt_user -p your_rt_database < /path/to/rt/var/data/shredder/dump.sql
136 That's it.i This will restore everything you'd deleted during a
137 shredding session when the file had been created.
141 =head2 $DependenciesLimit
143 Shredder stops with an error if the object has more than
144 C<$DependenciesLimit> dependencies. For example: a ticket has 1000
145 transactions or a transaction has 1000 attachments. This is protection
146 from bugs in shredder from wiping out your whole database, but
147 sometimes when you have big mail loops you may hit it.
149 Defaults to 1000. To change this (for example, to 10000) add the
150 following to your F<RT_SiteConfig.pm>:
152 Set( $DependenciesLimit, 10_000 );>
155 =head2 $ShredderStoragePath
157 Directory containing Shredder backup dumps; defaults to
158 F</opt/rt4/var/data/RT-Shredder> (assuming an /opt/rt4 installation).
160 To change this (for example, to /some/backup/path) add the following to
161 your F<RT_SiteConfig.pm>:
163 Set( $ShredderStoragePath, "/some/backup/path" );>
165 Be sure to specify an absolute path.
167 =head1 Database Indexes
169 We have found that the following indexes significantly speed up
170 shredding on most databases.
172 CREATE INDEX SHREDDER_CGM1 ON CachedGroupMembers(MemberId, GroupId, Disabled);
173 CREATE INDEX SHREDDER_CGM2 ON CachedGroupMembers(ImmediateParentId,MemberId);
174 CREATE INDEX SHREDDER_CGM3 on CachedGroupMembers (Via, Id);
176 CREATE UNIQUE INDEX SHREDDER_GM1 ON GroupMembers(MemberId, GroupId);
178 CREATE INDEX SHREDDER_TXN1 ON Transactions(ReferenceType, OldReference);
179 CREATE INDEX SHREDDER_TXN2 ON Transactions(ReferenceType, NewReference);
180 CREATE INDEX SHREDDER_TXN3 ON Transactions(Type, OldValue);
181 CREATE INDEX SHREDDER_TXN4 ON Transactions(Type, NewValue)
183 =head1 INFORMATION FOR DEVELOPERS
187 L<RT::Shredder> is an extension to RT which adds shredder methods to
188 RT objects and classes. The API is not well documented yet, but you
189 can find usage examples in L<rt-shredder> and the
190 F<lib/t/regression/shredder/*.t> test files.
192 However, here is a small example that do the same action as in CLI
193 example from L</SYNOPSIS>:
196 RT::Shredder::Init( force => 1 );
197 my $deleted = RT::Tickets->new( RT->SystemUser );
198 $deleted->{'allow_deleted_search'} = 1;
199 $deleted->LimitQueue( VALUE => 'general' );
200 $deleted->LimitStatus( VALUE => 'deleted' );
201 while( my $t = $deleted->Next ) {
206 =head2 RT::Shredder class' API
208 L<RT::Shredder> implements interfaces to objects cache, actions on the
209 objects in the cache and backups storage.
217 # I can't use 'use lib' here since it breakes tests
218 # because test suite uses old RT::Shredder setup from
221 ### after: push @INC, qw(@RT_LIB_PATH@);
222 use RT::Shredder::Constants;
223 use RT::Shredder::Exceptions;
227 require RT::Shredder::Record;
229 require RT::Shredder::ACE;
230 require RT::Shredder::Attachment;
231 require RT::Shredder::CachedGroupMember;
232 require RT::Shredder::CustomField;
233 require RT::Shredder::CustomFieldValue;
234 require RT::Shredder::GroupMember;
235 require RT::Shredder::Group;
236 require RT::Shredder::Link;
237 require RT::Shredder::Principal;
238 require RT::Shredder::Queue;
239 require RT::Shredder::Scrip;
240 require RT::Shredder::ScripAction;
241 require RT::Shredder::ScripCondition;
242 require RT::Shredder::Template;
243 require RT::Shredder::ObjectCustomFieldValue;
244 require RT::Shredder::Ticket;
245 require RT::Shredder::Transaction;
246 require RT::Shredder::User;
249 our @SUPPORTED_OBJECTS = qw(
264 ObjectCustomFieldValue
274 RT::Shredder::Init( %default_options );
276 C<RT::Shredder::Init()> should be called before creating an
277 RT::Shredder object. It iniitalizes RT and loads the RT
280 %default_options are passed to every C<<RT::Shredder->new>> call.
295 my $shredder = RT::Shredder->new(%options);
297 Construct a new RT::Shredder object.
299 There currently are no %options.
306 my $self = bless( {}, ref $proto || $proto );
314 $self->{'opt'} = { %opt, @_ };
315 $self->{'cache'} = {};
316 $self->{'resolver'} = {};
317 $self->{'dump_plugins'} = [];
320 =head4 CastObjectsToRecords( Objects => undef )
322 Cast objects to the C<RT::Record> objects or its ancesstors.
323 Objects can be passed as SCALAR (format C<< <class>-<id> >>),
324 ARRAY, C<RT::Record> ancesstors or C<RT::SearchBuilder> ancesstor.
326 Most methods that takes C<Objects> argument use this method to
327 cast argument value to list of records.
329 Returns an array of records.
333 my @objs = $shredder->CastObjectsToRecords(
334 Objects => [ # ARRAY reference
335 'RT::Attachment-10', # SCALAR or SCALAR reference
336 $tickets, # RT::Tickets object (isa RT::SearchBuilder)
337 $user, # RT::User object (isa RT::Record)
343 sub CastObjectsToRecords
346 my %args = ( Objects => undef, @_ );
349 my $targets = delete $args{'Objects'};
351 RT::Shredder::Exception->throw( "Undefined Objects argument" );
354 if( UNIVERSAL::isa( $targets, 'RT::SearchBuilder' ) ) {
355 #XXX: try to use ->_DoSearch + ->ItemsArrayRef in feature
356 # like we do in Record with links, but change only when
357 # more tests would be available
358 while( my $tmp = $targets->Next ) { push @res, $tmp };
359 } elsif ( UNIVERSAL::isa( $targets, 'RT::Record' ) ) {
361 } elsif ( UNIVERSAL::isa( $targets, 'ARRAY' ) ) {
362 foreach( @$targets ) {
363 push @res, $self->CastObjectsToRecords( Objects => $_ );
365 } elsif ( UNIVERSAL::isa( $targets, 'SCALAR' ) || !ref $targets ) {
366 $targets = $$targets if ref $targets;
367 my ($class, $id) = split /-/, $targets;
368 RT::Shredder::Exception->throw( "Unsupported class $class" )
369 unless $class =~ /^\w+(::\w+)*$/;
370 $class = 'RT::'. $class unless $class =~ /^RTx?::/i;
371 eval "require $class";
372 die "Couldn't load '$class' module" if $@;
373 my $obj = $class->new( RT->SystemUser );
374 die "Couldn't construct new '$class' object" unless $obj;
376 unless ( $obj->id ) {
377 $RT::Logger->error( "Couldn't load '$class' object with id '$id'" );
378 RT::Shredder::Exception::Info->throw( 'CouldntLoadObject' );
380 die "Loaded object has different id" unless( $id eq $obj->id );
383 RT::Shredder::Exception->throw( "Unsupported type ". ref $targets );
390 =head4 PutObjects( Objects => undef )
392 Puts objects into cache.
394 Returns array of the cache entries.
396 See C<CastObjectsToRecords> method for supported types of the C<Objects>
404 my %args = ( Objects => undef, @_ );
407 for( $self->CastObjectsToRecords( Objects => delete $args{'Objects'} ) ) {
408 push @res, $self->PutObject( %args, Object => $_ )
414 =head4 PutObject( Object => undef )
416 Puts record object into cache and returns its cache entry.
418 B<NOTE> that this method support B<only C<RT::Record> object or its ancesstor
419 objects>, if you want put mutliple objects or objects represented by different
420 classes then use C<PutObjects> method instead.
427 my %args = ( Object => undef, @_ );
429 my $obj = $args{'Object'};
430 unless( UNIVERSAL::isa( $obj, 'RT::Record' ) ) {
431 RT::Shredder::Exception->throw( "Unsupported type '". (ref $obj || $obj || '(undef)')."'" );
434 my $str = $obj->_AsString;
435 return ($self->{'cache'}->{ $str } ||= { State => ON_STACK, Object => $obj } );
438 =head4 GetObject, GetState, GetRecord( String => ''| Object => '' )
440 Returns record object from cache, cache entry state or cache entry accordingly.
442 All three methods takes C<String> (format C<< <class>-<id> >>) or C<Object> argument.
443 C<String> argument has more priority than C<Object> so if it's not empty then methods
444 leave C<Object> argument unchecked.
446 You can read about possible states and their meanings in L<RT::Shredder::Constants> docs.
458 if( $args{'String'} && $args{'Object'} ) {
460 Carp::croak( "both String and Object args passed" );
462 return $args{'String'} if $args{'String'};
463 return $args{'Object'}->_AsString if UNIVERSAL::can($args{'Object'}, '_AsString' );
467 sub GetObject { return (shift)->GetRecord( @_ )->{'Object'} }
468 sub GetState { return (shift)->GetRecord( @_ )->{'State'} }
472 my $str = $self->_ParseRefStrArgs( @_ );
473 return $self->{'cache'}->{ $str };
476 =head3 Dependencies resolvers
478 =head4 PutResolver, GetResolvers and ApplyResolvers
480 TODO: These methods have no documentation.
493 unless( UNIVERSAL::isa( $args{'Code'} => 'CODE' ) ) {
494 die "Resolver '$args{Code}' is not code reference";
499 $self->{'resolver'}->{ $args{'BaseClass'} } ||= {}
500 )->{ $args{'TargetClass'} || '' } ||= []
502 unshift @$resolvers, $args{'Code'};
516 if( $args{'TargetClass'} && exists $self->{'resolver'}->{ $args{'BaseClass'} }->{ $args{'TargetClass'} } ) {
517 push @res, @{ $self->{'resolver'}->{ $args{'BaseClass'} }->{ $args{'TargetClass'} || '' } };
519 if( exists $self->{'resolver'}->{ $args{'BaseClass'} }->{ '' } ) {
520 push @res, @{ $self->{'resolver'}->{ $args{'BaseClass'} }->{''} };
529 my %args = ( Dependency => undef, @_ );
530 my $dep = $args{'Dependency'};
532 my @resolvers = $self->GetResolvers(
533 BaseClass => $dep->BaseClass,
534 TargetClass => $dep->TargetClass,
537 unless( @resolvers ) {
538 RT::Shredder::Exception::Info->throw(
540 error => "Couldn't find resolver for dependency '". $dep->AsString ."'",
545 BaseObject => $dep->BaseObject,
546 TargetObject => $dep->TargetObject,
547 ) foreach @resolvers;
556 foreach my $cache_val ( values %{ $self->{'cache'} } ) {
557 next if $cache_val->{'State'} & (WIPED | IN_WIPING);
558 $self->Wipeout( Object => $cache_val->{'Object'} );
567 die "Couldn't begin transaction" unless $RT::Handle->BeginTransaction;
568 $mark = $self->PushDumpMark or die "Couldn't get dump mark";
569 $self->_Wipeout( @_ );
570 $self->PopDumpMark( Mark => $mark );
571 die "Couldn't commit transaction" unless $RT::Handle->Commit;
575 $RT::Handle->Rollback('force');
576 $self->RollbackDumpTo( Mark => $mark ) if $mark;
577 die $error if RT::Shredder::Exception::Info->caught;
578 die "Couldn't wipeout object: $error";
585 my %args = ( CacheRecord => undef, Object => undef, @_ );
587 my $record = $args{'CacheRecord'};
588 $record = $self->PutObject( Object => $args{'Object'} ) unless $record;
589 return if $record->{'State'} & (WIPED | IN_WIPING);
591 $record->{'State'} |= IN_WIPING;
592 my $object = $record->{'Object'};
594 $self->DumpObject( Object => $object, State => 'before any action' );
596 unless( $object->BeforeWipeout ) {
597 RT::Shredder::Exception->throw( "BeforeWipeout check returned error" );
600 my $deps = $object->Dependencies( Shredder => $self );
602 WithFlags => DEPENDS_ON | VARIABLE,
603 Callback => sub { $self->ApplyResolvers( Dependency => $_[0] ) },
605 $self->DumpObject( Object => $object, State => 'after resolvers' );
608 WithFlags => DEPENDS_ON,
609 WithoutFlags => WIPE_AFTER | VARIABLE,
610 Callback => sub { $self->_Wipeout( Object => $_[0]->TargetObject ) },
612 $self->DumpObject( Object => $object, State => 'after wiping dependencies' );
615 $record->{'State'} |= WIPED; delete $record->{'Object'};
616 $self->DumpObject( Object => $object, State => 'after wipeout' );
619 WithFlags => DEPENDS_ON | WIPE_AFTER,
620 WithoutFlags => VARIABLE,
621 Callback => sub { $self->_Wipeout( Object => $_[0]->TargetObject ) },
623 $self->DumpObject( Object => $object, State => 'after late dependencies' );
628 =head3 Data storage and backups
630 =head4 GetFileName( FileName => '<ISO DATETIME>-XXXX.sql', FromStorage => 1 )
632 Takes desired C<FileName> and flag C<FromStorage> then translate file name to absolute
635 * Default value of the C<FileName> option is C<< <ISO DATETIME>-XXXX.sql >>;
637 * if C<FileName> has C<XXXX> (exactly four uppercase C<X> letters) then it would be changed with digits from 0000 to 9999 range, with first one free value;
639 * if C<FileName> has C<%T> then it would be replaced with the current date and time in the C<YYYY-MM-DDTHH:MM:SS> format. Note that using C<%t> may still generate not unique names, using C<XXXX> recomended.
641 * if C<FromStorage> argument is true (default behaviour) then result path would always be relative to C<StoragePath>;
643 * if C<FromStorage> argument is false then result would be relative to the current dir unless it's already absolute path.
645 Returns an absolute path of the file.
648 # file from storage with default name format
649 my $fname = $shredder->GetFileName;
651 # file from storage with custom name format
652 my $fname = $shredder->GetFileName( FileName => 'shredder-XXXX.backup' );
654 # file with path relative to the current dir
655 my $fname = $shredder->GetFileName(
657 FileName => 'backups/shredder.sql',
660 # file with absolute path
661 my $fname = $shredder->GetFileName(
663 FileName => '/var/backups/shredder-XXXX.sql'
671 my %args = ( FileName => '', FromStorage => 1, @_ );
674 my $file = $args{'FileName'} || '%t-XXXX.sql';
675 if( $file =~ /\%t/i ) {
677 my $date_time = POSIX::strftime( "%Y%m%dT%H%M%S", gmtime );
678 $file =~ s/\%t/$date_time/gi;
681 # convert to absolute path
682 if( $args{'FromStorage'} ) {
683 $file = File::Spec->catfile( $self->StoragePath, $file );
684 } elsif( !File::Spec->file_name_is_absolute( $file ) ) {
685 $file = File::Spec->rel2abs( $file );
689 if( $file =~ /XXXX[^\/\\]*$/ ) {
690 my( $tmp, $i ) = ( $file, 0 );
694 $tmp =~ s/XXXX([^\/\\]*)$/sprintf("%04d", $i).$1/e;
695 } while( -e $tmp && $i < 9999 );
701 die "File '$file' exists, but is read-only";
704 unless( File::Spec->file_name_is_absolute( $file ) ) {
705 $file = File::Spec->rel2abs( $file );
709 my $dir = File::Spec->join( (File::Spec->splitpath( $file ))[0,1] );
710 unless( -e $dir && -d _) {
711 die "Base directory '$dir' for file '$file' doesn't exist";
714 die "Base directory '$dir' is not writable";
717 die "'$file' is not regular file";
725 Returns an absolute path to the storage dir. See
726 L</$ShredderStoragePath>.
728 See also description of the L</GetFileName> method.
734 return scalar( RT->Config->Get('ShredderStoragePath') )
735 || File::Spec->catdir( $RT::VarPath, qw(data RT-Shredder) );
738 my %active_dump_state = ();
741 my %args = ( Object => undef, Name => 'SQLDump', Arguments => undef, @_ );
743 my $plugin = $args{'Object'};
745 require RT::Shredder::Plugin;
746 $plugin = RT::Shredder::Plugin->new;
747 my( $status, $msg ) = $plugin->LoadByName( $args{'Name'} );
748 die "Couldn't load dump plugin: $msg\n" unless $status;
750 die "Plugin is not of correct type" unless lc $plugin->Type eq 'dump';
752 if ( my $pargs = $args{'Arguments'} ) {
753 my ($status, $msg) = $plugin->TestArgs( %$pargs );
754 die "Couldn't set plugin args: $msg\n" unless $status;
757 my @applies_to = $plugin->AppliesToStates;
758 die "Plugin doesn't apply to any state" unless @applies_to;
759 $active_dump_state{ lc $_ } = 1 foreach @applies_to;
761 push @{ $self->{'dump_plugins'} }, $plugin;
768 my %args = (Object => undef, State => undef, @_);
769 die "No state passed" unless $args{'State'};
770 return unless $active_dump_state{ lc $args{'State'} };
772 foreach (@{ $self->{'dump_plugins'} }) {
773 next unless grep lc $args{'State'} eq lc $_, $_->AppliesToStates;
774 my ($state, $msg) = $_->Run( %args );
775 die "Couldn't run plugin: $msg" unless $state;
779 { my $mark = 1; # XXX: integer overflows?
783 foreach (@{ $self->{'dump_plugins'} }) {
784 my ($state, $msg) = $_->PushMark( Mark => $mark );
785 die "Couldn't push mark: $msg" unless $state;
791 foreach (@{ $self->{'dump_plugins'} }) {
792 my ($state, $msg) = $_->PushMark( @_ );
793 die "Couldn't pop mark: $msg" unless $state;
798 foreach (@{ $self->{'dump_plugins'} }) {
799 my ($state, $msg) = $_->RollbackTo( @_ );
800 die "Couldn't rollback to mark: $msg" unless $state;
810 =head2 Database transactions support
812 Since 0.03_01 RT::Shredder uses database transactions and should be
813 much safer to run on production servers.
817 Mainstream RT doesn't use FKs, but at least I posted DDL script that creates them
818 in mysql DB, note that if you use FKs then this two valid keys don't allow delete
819 Tickets because of bug in MySQL:
821 ALTER TABLE Tickets ADD FOREIGN KEY (EffectiveId) REFERENCES Tickets(id);
822 ALTER TABLE CachedGroupMembers ADD FOREIGN KEY (Via) REFERENCES CachedGroupMembers(id);
824 L<http://bugs.mysql.com/bug.php?id=4042>
826 =head1 BUGS AND HOW TO CONTRIBUTE
828 We need your feedback in all cases: if you use it or not,
829 is it works for you or not.
833 Don't skip C<make test> step while install and send me reports if it's fails.
834 Add your own tests, it's easy enough if you've writen at list one perl script
835 that works with RT. Read more about testing in F<t/utils.pl>.
839 Send reports to L</AUTHOR> or to the RT mailing lists.
843 Many bugs in the docs: insanity, spelling, gramar and so on.
844 Patches are wellcome.
848 Please, see Todo file, it has some technical notes
849 about what I plan to do, when I'll do it, also it
850 describes some problems code has.
854 Since RT-3.7 shredder is a part of the RT distribution.
855 Versions of the RTx::Shredder extension could
856 be downloaded from the CPAN. Those work with older
857 RT versions or you can find repository at
858 L<https://opensvn.csie.org/rtx_shredder>
862 Ruslan U. Zakirov <Ruslan.Zakirov@gmail.com>
866 This program is free software; you can redistribute
867 it and/or modify it under the same terms as Perl itself.
869 The full text of the license can be found in the
874 L<rt-shredder>, L<rt-validator>