MediaWiki master
rebuildtextindex.php
Go to the documentation of this file.
1<?php
21// @codeCoverageIgnoreStart
22require_once __DIR__ . '/Maintenance.php';
23// @codeCoverageIgnoreEnd
24
31
42 private const RTI_CHUNK_SIZE = 500;
43
44 public function __construct() {
45 parent::__construct();
46 $this->addDescription( 'Rebuild search index table from scratch' );
47 }
48
50 public function getDbType() {
51 return Maintenance::DB_ADMIN;
52 }
53
54 public function execute() {
55 // Shouldn't be needed for Postgres
56 $dbw = $this->getPrimaryDB();
57 if ( $dbw->getType() == 'postgres' ) {
58 $this->fatalError( "This script is not needed when using Postgres.\n" );
59 }
60
61 if ( $dbw->getType() == 'sqlite' ) {
62 if ( !DatabaseSqlite::getFulltextSearchModule() ) {
63 $this->fatalError( "Your version of SQLite module for PHP doesn't "
64 . "support full-text search (FTS3).\n" );
65 }
66 }
67
68 if ( $dbw->getType() == 'mysql' ) {
69 $this->dropMysqlTextIndex();
70 $this->clearSearchIndex();
71 $this->populateSearchIndex();
72 $this->createMysqlTextIndex();
73 } else {
74 $this->clearSearchIndex();
75 $this->populateSearchIndex();
76 }
77
78 $this->output( "Done.\n" );
79 }
80
84 protected function populateSearchIndex() {
85 $dbw = $this->getPrimaryDB();
86 $res = $dbw->newSelectQueryBuilder()
87 ->select( [ 'count' => 'MAX(page_id)' ] )
88 ->from( 'page' )
89 ->caller( __METHOD__ )->fetchResultSet();
90 $s = $res->fetchObject();
91 $count = $s->count;
92 $this->output( "Rebuilding index fields for {$count} pages...\n" );
93 $n = 0;
94
95 $revStore = $this->getServiceContainer()->getRevisionStore();
96 $queryBuilderTemplate = $revStore->newSelectQueryBuilder( $dbw )
97 ->joinPage()
98 ->joinComment();
99
100 while ( $n < $count ) {
101 if ( $n ) {
102 $this->output( $n . "\n" );
103 }
104 $end = $n + self::RTI_CHUNK_SIZE - 1;
105 $queryBuilder = clone $queryBuilderTemplate;
106 $res = $queryBuilder->where( [
107 $dbw->expr( 'page_id', '>=', $n )->and( 'page_id', '<=', $end ),
108 'page_latest = rev_id'
109 ] )->caller( __METHOD__ )->fetchResultSet();
110
111 foreach ( $res as $s ) {
112
113 // T268673 Prevent failure of WikiPage.php: Invalid or virtual namespace -1 given
114 if ( $s->page_namespace < 0 ) {
115 continue;
116 }
117
118 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
119 try {
120 $revRecord = $revStore->newRevisionFromRow( $s );
121 $content = $revRecord->getContent( SlotRecord::MAIN );
122
123 $u = new SearchUpdate( $s->page_id, $title, $content );
124 $u->doUpdate();
125 } catch ( MWContentSerializationException $ex ) {
126 $this->output( "Failed to deserialize content of revision {$s->rev_id} of page "
127 . "`" . $title->getPrefixedDBkey() . "`!\n" );
128 }
129 }
130 $n += self::RTI_CHUNK_SIZE;
131 }
132 }
133
137 private function dropMysqlTextIndex() {
138 $dbw = $this->getDB( DB_PRIMARY );
139 $searchindex = $dbw->tableName( 'searchindex' );
140 if ( $dbw->indexExists( 'searchindex', 'si_title', __METHOD__ ) ) {
141 $this->output( "Dropping index...\n" );
142 $sql = "ALTER TABLE $searchindex DROP INDEX si_title, DROP INDEX si_text";
143 $dbw->query( $sql, __METHOD__ );
144 }
145 }
146
150 private function createMysqlTextIndex() {
151 $dbw = $this->getPrimaryDB();
152 $searchindex = $dbw->tableName( 'searchindex' );
153 $this->output( "\nRebuild the index...\n" );
154 foreach ( [ 'si_title', 'si_text' ] as $field ) {
155 $sql = "ALTER TABLE $searchindex ADD FULLTEXT $field ($field)";
156 $dbw->query( $sql, __METHOD__ );
157 }
158 }
159
163 private function clearSearchIndex() {
164 $dbw = $this->getPrimaryDB();
165 $this->output( 'Clearing searchindex table...' );
166 $dbw->newDeleteQueryBuilder()
167 ->deleteFrom( 'searchindex' )
168 ->where( '*' )
169 ->caller( __METHOD__ )->execute();
170 $this->output( "Done\n" );
171 }
172}
173
174// @codeCoverageIgnoreStart
175$maintClass = RebuildTextIndex::class;
176require_once RUN_MAINTENANCE_IF_MAIN;
177// @codeCoverageIgnoreEnd
Exception representing a failure to serialize or unserialize a content object.
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
output( $out, $channel=null)
Throw some output to the user.
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
getServiceContainer()
Returns the main service container.
addDescription( $text)
Set the description text.
Value object representing a content slot associated with a page revision.
Database independent search index updater.
Represents a title within MediaWiki.
Definition Title.php:78
Rebuild search index table from scratch.
execute()
Do the actual work.
getDbType()
Does the script need different DB access? By default, we give Maintenance scripts normal rights to th...
__construct()
Default constructor.
populateSearchIndex()
Populates the search index with content from all pages.
This is the SQLite database abstraction layer.
const DB_PRIMARY
Definition defines.php:28