Upgrade to 4.2.2
[usit-rt.git] / sbin / rt-fulltext-indexer
CommitLineData
84fb5b46
MKG
1#!/usr/bin/perl
2# BEGIN BPS TAGGED BLOCK {{{
3#
4# COPYRIGHT:
5#
320f0092 6# This software is Copyright (c) 1996-2014 Best Practical Solutions, LLC
84fb5b46
MKG
7# <sales@bestpractical.com>
8#
9# (Except where explicitly superseded by other copyright notices)
10#
11#
12# LICENSE:
13#
14# This work is made available to you under the terms of Version 2 of
15# the GNU General Public License. A copy of that license should have
16# been provided with this software, but in any event can be snarfed
17# from www.gnu.org.
18#
19# This work is distributed in the hope that it will be useful, but
20# WITHOUT ANY WARRANTY; without even the implied warranty of
21# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22# General Public License for more details.
23#
24# You should have received a copy of the GNU General Public License
25# along with this program; if not, write to the Free Software
26# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
27# 02110-1301 or visit their web page on the internet at
28# http://www.gnu.org/licenses/old-licenses/gpl-2.0.html.
29#
30#
31# CONTRIBUTION SUBMISSION POLICY:
32#
33# (The following paragraph is not intended to limit the rights granted
34# to you to modify and distribute this software under the terms of
35# the GNU General Public License and is only of importance to you if
36# you choose to contribute your changes and enhancements to the
37# community by submitting them to Best Practical Solutions, LLC.)
38#
39# By intentionally submitting any modifications, corrections or
40# derivatives to this work, or any other work intended for use with
41# Request Tracker, to Best Practical Solutions, LLC, you confirm that
42# you are the copyright holder for those contributions and you grant
43# Best Practical Solutions, LLC a nonexclusive, worldwide, irrevocable,
44# royalty-free, perpetual, license to use, copy, create derivative
45# works based on those contributions, and sublicense and distribute
46# those contributions and any derivatives thereof.
47#
48# END BPS TAGGED BLOCK }}}
49use strict;
50use warnings;
51no warnings 'once';
52
53# fix lib paths, some may be relative
af59614d 54BEGIN { # BEGIN RT CMD BOILERPLATE
84fb5b46 55 require File::Spec;
af59614d 56 require Cwd;
84fb5b46
MKG
57 my @libs = ("lib", "local/lib");
58 my $bin_path;
59
60 for my $lib (@libs) {
61 unless ( File::Spec->file_name_is_absolute($lib) ) {
af59614d 62 $bin_path ||= ( File::Spec->splitpath(Cwd::abs_path(__FILE__)) )[1];
84fb5b46
MKG
63 $lib = File::Spec->catfile( $bin_path, File::Spec->updir, $lib );
64 }
65 unshift @INC, $lib;
66 }
af59614d 67
84fb5b46
MKG
68}
69
70BEGIN {
71 use RT;
72 RT::LoadConfig();
73 RT::Init();
74};
75use RT::Interface::CLI ();
76
77my %OPT = (
78 help => 0,
79 debug => 0,
80);
81my @OPT_LIST = qw(help|h! debug!);
82
83my $db_type = RT->Config->Get('DatabaseType');
84if ( $db_type eq 'Pg' ) {
85 %OPT = (
86 %OPT,
87 limit => 0,
88 all => 0,
89 );
90 push @OPT_LIST, 'limit=i', 'all!';
91}
92elsif ( $db_type eq 'mysql' ) {
93 %OPT = (
94 %OPT,
95 limit => 0,
96 all => 0,
97 xmlpipe2 => 0,
98 );
99 push @OPT_LIST, 'limit=i', 'all!', 'xmlpipe2!';
100}
101elsif ( $db_type eq 'Oracle' ) {
102 %OPT = (
103 %OPT,
104 memory => '2M',
105 );
106 push @OPT_LIST, qw(memory=s);
107}
108
109use Getopt::Long qw(GetOptions);
110GetOptions( \%OPT, @OPT_LIST );
111
112if ( $OPT{'help'} ) {
113 RT::Interface::CLI->ShowHelp(
114 Sections => 'NAME|DESCRIPTION|'. uc($db_type),
115 );
116}
117
118my $fts_config = RT->Config->Get('FullTextSearch') || {};
119unless ( $fts_config->{'Enable'} ) {
120 print STDERR <<EOT;
121
122Full text search is disabled in your RT configuration. Run
b5747ff2 123/www/var/rt/sbin/rt-setup-fulltext-index to configure and enable it.
84fb5b46
MKG
124
125EOT
126 exit 1;
127}
128unless ( $fts_config->{'Indexed'} ) {
129 print STDERR <<EOT;
130
131Full text search is enabled in your RT configuration, but not with any
132full-text database indexing -- hence this tool is not required. Read
133the documentation for %FullTextSearch in your RT_Config for more details.
134
135EOT
136 exit 1;
137}
138
139if ( $db_type eq 'Oracle' ) {
140 my $index = $fts_config->{'IndexName'} || 'rt_fts_index';
141 $RT::Handle->dbh->do(
142 "begin ctx_ddl.sync_index(?, ?); end;", undef,
143 $index, $OPT{'memory'}
144 );
145 exit;
146} elsif ( $db_type eq 'mysql' ) {
147 unless ($OPT{'xmlpipe2'}) {
148 print STDERR <<EOT;
149
150Updates to the external Sphinx index are done via running the sphinx
151`indexer` tool:
152
153 indexer rt
154
155EOT
156 exit 1;
157 }
158}
159
160my @types = qw(text html);
161foreach my $type ( @types ) {
162 REDO:
163 my $attachments = attachments($type);
164 $attachments->Limit(
165 FIELD => 'id',
166 OPERATOR => '>',
167 VALUE => last_indexed($type)
168 );
169 $attachments->OrderBy( FIELD => 'id', ORDER => 'asc' );
170 $attachments->RowsPerPage( $OPT{'limit'} || 100 );
171
172 my $found = 0;
173 while ( my $a = $attachments->Next ) {
174 next if filter( $type, $a );
175 debug("Found attachment #". $a->id );
176 my $txt = extract($type, $a) or next;
177 $found++;
178 process( $type, $a, $txt );
179 debug("Processed attachment #". $a->id );
180 }
181 finalize( $type, $attachments ) if $found;
182 clean( $type );
183 goto REDO if $OPT{'all'} and $attachments->Count == ($OPT{'limit'} || 100)
184}
185
186sub attachments {
187 my $type = shift;
188 my $res = RT::Attachments->new( RT->SystemUser );
189 my $txn_alias = $res->Join(
190 ALIAS1 => 'main',
191 FIELD1 => 'TransactionId',
192 TABLE2 => 'Transactions',
193 FIELD2 => 'id',
194 );
195 $res->Limit(
196 ALIAS => $txn_alias,
197 FIELD => 'ObjectType',
198 VALUE => 'RT::Ticket',
199 );
200 my $ticket_alias = $res->Join(
201 ALIAS1 => $txn_alias,
202 FIELD1 => 'ObjectId',
203 TABLE2 => 'Tickets',
204 FIELD2 => 'id',
205 );
206 $res->Limit(
207 ALIAS => $ticket_alias,
208 FIELD => 'Status',
209 OPERATOR => '!=',
210 VALUE => 'deleted'
211 );
212
213 return goto_specific(
214 suffix => $type,
215 error => "Don't know how to find $type attachments",
216 arguments => [$res],
217 );
218}
219
220sub last_indexed {
221 my ($type) = (@_);
222 return goto_specific(
223 suffix => $db_type,
224 error => "Don't know how to find last indexed $type attachment for $db_type DB",
225 arguments => \@_,
226 );
227}
228
229sub filter {
230 my $type = shift;
231 return goto_specific(
232 suffix => $type,
233 arguments => \@_,
234 );
235}
236
237sub extract {
238 my $type = shift;
239 return goto_specific(
240 suffix => $type,
241 error => "No way to convert $type attachment into text",
242 arguments => \@_,
243 );
244}
245
246sub process {
247 return goto_specific(
248 suffix => $db_type,
249 error => "No processer for $db_type DB",
250 arguments => \@_,
251 );
252}
253
254sub finalize {
255 return goto_specific(
256 suffix => $db_type,
257 arguments => \@_,
258 );
259}
260
261sub clean {
262 return goto_specific(
263 suffix => $db_type,
264 arguments => \@_,
265 );
266}
267
268{
269sub last_indexed_mysql {
270 my $type = shift;
271 my $attr = $RT::System->FirstAttribute('LastIndexedAttachments');
272 return 0 unless $attr;
273 return 0 unless exists $attr->{ $type };
274 return $attr->{ $type } || 0;
275}
276
277sub process_mysql {
278 my ($type, $attachment, $text) = (@_);
279
280 my $doc = sphinx_template();
281
282 my $element = $doc->createElement('sphinx:document');
283 $element->setAttribute( id => $attachment->id );
284 $element->appendTextChild( content => $$text );
285
286 $doc->documentElement->appendChild( $element );
287}
288
289my $doc = undef;
290sub sphinx_template {
291 return $doc if $doc;
292
293 require XML::LibXML;
294 $doc = XML::LibXML::Document->new('1.0', 'UTF-8');
295 my $root = $doc->createElement('sphinx:docset');
296 $doc->setDocumentElement( $root );
297
298 my $schema = $doc->createElement('sphinx:schema');
299 $root->appendChild( $schema );
300 foreach ( qw(content) ) {
301 my $field = $doc->createElement('sphinx:field');
302 $field->setAttribute( name => $_ );
303 $schema->appendChild( $field );
304 }
305
306 return $doc;
307}
308
309sub finalize_mysql {
310 my ($type, $attachments) = @_;
311 sphinx_template()->toFH(*STDOUT, 1);
312}
313
314sub clean_mysql {
315 $doc = undef;
316}
317
318}
319
320sub last_indexed_pg {
321 my $type = shift;
322 my $attachments = attachments( $type );
323 my $alias = 'main';
324 if ( $fts_config->{'Table'} && $fts_config->{'Table'} ne 'Attachments' ) {
325 $alias = $attachments->Join(
326 TYPE => 'left',
327 FIELD1 => 'id',
328 TABLE2 => $fts_config->{'Table'},
329 FIELD2 => 'id',
330 );
331 }
332 $attachments->Limit(
333 ALIAS => $alias,
334 FIELD => $fts_config->{'Column'},
335 OPERATOR => 'IS NOT',
336 VALUE => 'NULL',
337 );
338 $attachments->OrderBy( FIELD => 'id', ORDER => 'desc' );
339 $attachments->RowsPerPage( 1 );
340 my $res = $attachments->First;
341 return 0 unless $res;
342 return $res->id;
343}
344
345sub process_pg {
346 my ($type, $attachment, $text) = (@_);
347
348 my $dbh = $RT::Handle->dbh;
349 my $table = $fts_config->{'Table'};
350 my $column = $fts_config->{'Column'};
351
352 my $query;
353 if ( $table ) {
354 if ( my ($id) = $dbh->selectrow_array("SELECT id FROM $table WHERE id = ?", undef, $attachment->id) ) {
355 $query = "UPDATE $table SET $column = to_tsvector(?) WHERE id = ?";
356 } else {
357 $query = "INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)";
358 }
359 } else {
360 $query = "UPDATE Attachments SET $column = to_tsvector(?) WHERE id = ?";
361 }
362
363 my $status = eval { $dbh->do( $query, undef, $$text, $attachment->id ) };
364 unless ( $status ) {
5b0d0914 365 if ( $dbh->err == 7 && $dbh->state eq '54000' ) {
01e3b242 366 warn "Attachment @{[$attachment->id]} cannot be indexed. Most probably it contains too many unique words. Error: ". $dbh->errstr;
5b0d0914 367 } elsif ( $dbh->err == 7 && $dbh->state eq '22021' ) {
01e3b242 368 warn "Attachment @{[$attachment->id]} cannot be indexed. Most probably it contains invalid UTF8 bytes. Error: ". $dbh->errstr;
84fb5b46
MKG
369 } else {
370 die "error: ". $dbh->errstr;
371 }
403d7b0b
MKG
372
373 # Insert an empty tsvector, so we count this row as "indexed"
374 # for purposes of knowing where to pick up
375 eval { $dbh->do( $query, undef, "", $attachment->id ) }
376 or die "Failed to insert empty tsvector: " . $dbh->errstr;
84fb5b46
MKG
377 }
378}
379
380sub attachments_text {
381 my $res = shift;
382 $res->Limit( FIELD => 'ContentType', VALUE => 'text/plain' );
383 return $res;
384}
385
386sub extract_text {
387 my $attachment = shift;
388 my $text = $attachment->Content;
389 return undef unless defined $text && length($text);
390 return \$text;
391}
392
393sub attachments_html {
394 my $res = shift;
395 $res->Limit( FIELD => 'ContentType', VALUE => 'text/html' );
396 return $res;
397}
398
399sub filter_html {
400 my $attachment = shift;
401 if ( my $parent = $attachment->ParentObj ) {
402# skip html parts that are alternatives
403 return 1 if $parent->id
404 && $parent->ContentType eq 'mulitpart/alternative';
405 }
406 return 0;
407}
408
409sub extract_html {
410 my $attachment = shift;
411 my $text = $attachment->Content;
412 return undef unless defined $text && length($text);
413# TODO: html -> text
414 return \$text;
415}
416
417sub goto_specific {
418 my %args = (@_);
419
420 my $func = (caller(1))[3];
421 $func =~ s/.*:://;
422 my $call = $func ."_". lc $args{'suffix'};
423 unless ( defined &$call ) {
424 return undef unless $args{'error'};
425 require Carp; Carp::croak( $args{'error'} );
426 }
427 @_ = @{ $args{'arguments'} };
428 goto &$call;
429}
430
431
432# helper functions
433sub debug { print @_, "\n" if $OPT{debug}; 1 }
434sub error { $RT::Logger->error(_(@_)); 1 }
435sub warning { $RT::Logger->warn(_(@_)); 1 }
436
437=head1 NAME
438
439rt-fulltext-indexer - Indexer for full text search
440
441=head1 DESCRIPTION
442
443This is a helper script to keep full text indexes in sync with data.
444Read F<docs/full_text_indexing.pod> for complete details on how and when
445to run it.
446
447=head1 AUTHOR
448
449Ruslan Zakirov E<lt>ruz@bestpractical.comE<gt>,
450Alex Vandiver E<lt>alexmv@bestpractical.comE<gt>
451
452=cut
453