MediaWiki master
cleanupTitles.php
Go to the documentation of this file.
1<?php
29
30require_once __DIR__ . '/TableCleanup.php';
31
38 public function __construct() {
39 parent::__construct();
40 $this->addDescription( 'Script to clean up broken, unparseable titles' );
41 $this->setBatchSize( 1000 );
42 }
43
47 protected function processRow( $row ) {
48 $display = Title::makeName( $row->page_namespace, $row->page_title );
49 $verified = $this->getServiceContainer()->getContentLanguage()->normalize( $display );
50 $title = Title::newFromText( $verified );
51
52 if ( $title !== null
53 && $title->canExist()
54 && $title->getNamespace() == $row->page_namespace
55 && $title->getDBkey() === $row->page_title
56 ) {
57 // all is fine
58 $this->progress( 0 );
59
60 return;
61 }
62
63 if ( $row->page_namespace == NS_FILE && $this->fileExists( $row->page_title ) ) {
64 $this->output( "file $row->page_title needs cleanup, please run cleanupImages.php.\n" );
65 $this->progress( 0 );
66 } elseif ( $title === null ) {
67 $this->output( "page $row->page_id ($display) is illegal.\n" );
68 $this->moveIllegalPage( $row );
69 $this->progress( 1 );
70 } else {
71 $this->output( "page $row->page_id ($display) doesn't match self.\n" );
72 $this->moveInconsistentPage( $row, $title );
73 $this->progress( 1 );
74 }
75 }
76
81 protected function fileExists( $name ) {
82 // XXX: Doesn't actually check for file existence, just presence of image record.
83 // This is reasonable, since cleanupImages.php only iterates over the image table.
84 $dbr = $this->getReplicaDB();
85 $row = $dbr->newSelectQueryBuilder()
86 ->select( '*' )
87 ->from( 'image' )
88 ->where( [ 'img_name' => $name ] )
89 ->caller( __METHOD__ )
90 ->fetchRow();
91
92 return $row !== false;
93 }
94
98 protected function moveIllegalPage( $row ) {
99 $legal = 'A-Za-z0-9_/\\\\-';
100 $legalized = preg_replace_callback( "!([^$legal])!",
101 [ $this, 'hexChar' ],
102 $row->page_title );
103 if ( $legalized == '.' ) {
104 $legalized = '(dot)';
105 }
106 if ( $legalized == '_' ) {
107 $legalized = '(space)';
108 }
109 $legalized = 'Broken/' . $legalized;
110
111 $title = Title::newFromText( $legalized );
112 if ( $title === null ) {
113 $clean = 'Broken/id:' . $row->page_id;
114 $this->output( "Couldn't legalize; form '$legalized' still invalid; using '$clean'\n" );
115 $title = Title::newFromText( $clean );
116 } elseif ( $title->exists() ) {
117 $clean = 'Broken/id:' . $row->page_id;
118 $this->output( "Legalized for '$legalized' exists; using '$clean'\n" );
119 $title = Title::newFromText( $clean );
120 }
121
122 $dest = $title->getDBkey();
123 if ( $this->dryrun ) {
124 $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
125 "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
126 } else {
127 $this->output( "renaming $row->page_id ($row->page_namespace," .
128 "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
129 $this->getPrimaryDB()
130 ->newUpdateQueryBuilder()
131 ->update( 'page' )
132 ->set( [ 'page_title' => $dest ] )
133 ->where( [ 'page_id' => $row->page_id ] )
134 ->caller( __METHOD__ )->execute();
135 }
136 }
137
142 protected function moveInconsistentPage( $row, Title $title ) {
143 if ( $title->exists( IDBAccessObject::READ_LATEST )
144 || $title->getInterwiki()
145 || !$title->canExist()
146 ) {
147 $titleImpossible = $title->getInterwiki() || !$title->canExist();
148 if ( $titleImpossible ) {
149 $prior = $title->getPrefixedDBkey();
150 } else {
151 $prior = $title->getDBkey();
152 }
153
154 # Old cleanupTitles could move articles there. See T25147.
155 $ns = $row->page_namespace;
156 if ( $ns < 0 ) {
157 $ns = 0;
158 }
159
160 # Namespace which no longer exists. Put the page in the main namespace
161 # since we don't have any idea of the old namespace name. See T70501.
162 if ( !$this->getServiceContainer()->getNamespaceInfo()->exists( $ns ) ) {
163 $ns = 0;
164 }
165
166 if ( !$titleImpossible && !$title->exists() ) {
167 // Looks like the current title, after cleaning it up, is valid and available
168 $clean = $prior;
169 } else {
170 $clean = 'Broken/' . $prior;
171 }
172 $verified = Title::makeTitleSafe( $ns, $clean );
173 if ( !$verified || $verified->exists() ) {
174 $blah = "Broken/id:" . $row->page_id;
175 $this->output( "Couldn't legalize; form '$clean' exists; using '$blah'\n" );
176 $verified = Title::makeTitleSafe( $ns, $blah );
177 }
178 $title = $verified;
179 }
180 if ( $title === null ) {
181 $this->fatalError( "Something awry; empty title." );
182 }
183 $ns = $title->getNamespace();
184 $dest = $title->getDBkey();
185
186 if ( $this->dryrun ) {
187 $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
188 "'$row->page_title') to ($ns,'$dest')\n" );
189 } else {
190 $this->output( "renaming $row->page_id ($row->page_namespace," .
191 "'$row->page_title') to ($ns,'$dest')\n" );
192 $this->getPrimaryDB()
193 ->newUpdateQueryBuilder()
194 ->update( 'page' )
195 ->set( [
196 'page_namespace' => $ns,
197 'page_title' => $dest
198 ] )
199 ->where( [ 'page_id' => $row->page_id ] )
200 ->caller( __METHOD__ )->execute();
201 $this->getServiceContainer()->getLinkCache()->clear();
202 }
203 }
204}
205
206$maintClass = TitleCleanup::class;
207require_once RUN_MAINTENANCE_IF_MAIN;
const NS_FILE
Definition Defines.php:71
output( $out, $channel=null)
Throw some output to the user.
getServiceContainer()
Returns the main service container.
addDescription( $text)
Set the description text.
setBatchSize( $s=0)
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Represents a title within MediaWiki.
Definition Title.php:79
canExist()
Can this title represent a page in the wiki's database?
Definition Title.php:1213
exists( $flags=0)
Check if page exists.
Definition Title.php:3146
getNamespace()
Get the namespace index, i.e.
Definition Title.php:1045
getInterwiki()
Get the interwiki prefix.
Definition Title.php:946
getDBkey()
Get the main part with underscores.
Definition Title.php:1036
getPrefixedDBkey()
Get the prefixed database key form.
Definition Title.php:1850
Generic class to cleanup a database table.
progress( $updated)
Maintenance script to clean up broken, unparseable titles.
moveIllegalPage( $row)
moveInconsistentPage( $row, Title $title)
__construct()
Default constructor.
fileExists( $name)
$maintClass