MediaWiki  master
cleanupTitles.php
Go to the documentation of this file.
1 <?php
29 
30 require_once __DIR__ . '/cleanupTable.inc';
31 
37 class TitleCleanup extends TableCleanup {
38  public function __construct() {
39  parent::__construct();
40  $this->addDescription( 'Script to clean up broken, unparseable titles' );
41  $this->batchSize = 1000;
42  }
43 
47  protected function processRow( $row ) {
48  $display = Title::makeName( $row->page_namespace, $row->page_title );
49  $verified = MediaWikiServices::getInstance()->getContentLanguage()->normalize( $display );
50  $title = Title::newFromText( $verified );
51 
52  if ( !is_null( $title )
53  && $title->canExist()
54  && $title->getNamespace() == $row->page_namespace
55  && $title->getDBkey() === $row->page_title
56  ) {
57  $this->progress( 0 ); // all is fine
58 
59  return;
60  }
61 
62  if ( $row->page_namespace == NS_FILE && $this->fileExists( $row->page_title ) ) {
63  $this->output( "file $row->page_title needs cleanup, please run cleanupImages.php.\n" );
64  $this->progress( 0 );
65  } elseif ( is_null( $title ) ) {
66  $this->output( "page $row->page_id ($display) is illegal.\n" );
67  $this->moveIllegalPage( $row );
68  $this->progress( 1 );
69  } else {
70  $this->output( "page $row->page_id ($display) doesn't match self.\n" );
71  $this->moveInconsistentPage( $row, $title );
72  $this->progress( 1 );
73  }
74  }
75 
80  protected function fileExists( $name ) {
81  // XXX: Doesn't actually check for file existence, just presence of image record.
82  // This is reasonable, since cleanupImages.php only iterates over the image table.
83  $dbr = $this->getDB( DB_REPLICA );
84  $row = $dbr->selectRow( 'image', [ 'img_name' ], [ 'img_name' => $name ], __METHOD__ );
85 
86  return $row !== false;
87  }
88 
92  protected function moveIllegalPage( $row ) {
93  $legal = 'A-Za-z0-9_/\\\\-';
94  $legalized = preg_replace_callback( "!([^$legal])!",
95  [ $this, 'hexChar' ],
96  $row->page_title );
97  if ( $legalized == '.' ) {
98  $legalized = '(dot)';
99  }
100  if ( $legalized == '_' ) {
101  $legalized = '(space)';
102  }
103  $legalized = 'Broken/' . $legalized;
104 
105  $title = Title::newFromText( $legalized );
106  if ( is_null( $title ) ) {
107  $clean = 'Broken/id:' . $row->page_id;
108  $this->output( "Couldn't legalize; form '$legalized' still invalid; using '$clean'\n" );
109  $title = Title::newFromText( $clean );
110  } elseif ( $title->exists() ) {
111  $clean = 'Broken/id:' . $row->page_id;
112  $this->output( "Legalized for '$legalized' exists; using '$clean'\n" );
113  $title = Title::newFromText( $clean );
114  }
115 
116  $dest = $title->getDBkey();
117  if ( $this->dryrun ) {
118  $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
119  "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
120  } else {
121  $this->output( "renaming $row->page_id ($row->page_namespace," .
122  "'$row->page_title') to ($row->page_namespace,'$dest')\n" );
123  $dbw = $this->getDB( DB_MASTER );
124  $dbw->update( 'page',
125  [ 'page_title' => $dest ],
126  [ 'page_id' => $row->page_id ],
127  __METHOD__ );
128  }
129  }
130 
135  protected function moveInconsistentPage( $row, Title $title ) {
136  if ( $title->exists( Title::READ_LATEST )
137  || $title->getInterwiki()
138  || !$title->canExist()
139  ) {
140  $titleImpossible = $title->getInterwiki() || !$title->canExist();
141  if ( $titleImpossible ) {
142  $prior = $title->getPrefixedDBkey();
143  } else {
144  $prior = $title->getDBkey();
145  }
146 
147  # Old cleanupTitles could move articles there. See T25147.
148  $ns = $row->page_namespace;
149  if ( $ns < 0 ) {
150  $ns = 0;
151  }
152 
153  # Namespace which no longer exists. Put the page in the main namespace
154  # since we don't have any idea of the old namespace name. See T70501.
155  if ( !MediaWikiServices::getInstance()->getNamespaceInfo()->exists( $ns ) ) {
156  $ns = 0;
157  }
158 
159  if ( !$titleImpossible && !$title->exists() ) {
160  // Looks like the current title, after cleaning it up, is valid and available
161  $clean = $prior;
162  } else {
163  $clean = 'Broken/' . $prior;
164  }
165  $verified = Title::makeTitleSafe( $ns, $clean );
166  if ( !$verified || $verified->exists() ) {
167  $blah = "Broken/id:" . $row->page_id;
168  $this->output( "Couldn't legalize; form '$clean' exists; using '$blah'\n" );
169  $verified = Title::makeTitleSafe( $ns, $blah );
170  }
171  $title = $verified;
172  }
173  if ( is_null( $title ) ) {
174  $this->fatalError( "Something awry; empty title." );
175  }
176  $ns = $title->getNamespace();
177  $dest = $title->getDBkey();
178 
179  if ( $this->dryrun ) {
180  $this->output( "DRY RUN: would rename $row->page_id ($row->page_namespace," .
181  "'$row->page_title') to ($ns,'$dest')\n" );
182  } else {
183  $this->output( "renaming $row->page_id ($row->page_namespace," .
184  "'$row->page_title') to ($ns,'$dest')\n" );
185  $dbw = $this->getDB( DB_MASTER );
186  $dbw->update( 'page',
187  [
188  'page_namespace' => $ns,
189  'page_title' => $dest
190  ],
191  [ 'page_id' => $row->page_id ],
192  __METHOD__ );
193  MediaWikiServices::getInstance()->getLinkCache()->clear();
194  }
195  }
196 }
197 
198 $maintClass = TitleCleanup::class;
199 require_once RUN_MAINTENANCE_IF_MAIN;
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
const DB_MASTER
Definition: defines.php:26
fileExists( $name)
getDBkey()
Get the main part with underscores.
Definition: Title.php:1014
Generic class to cleanup a database table.
canExist()
Can this title represent a page in the wiki&#39;s database?
Definition: Title.php:1189
addDescription( $text)
Set the description text.
getNamespace()
Get the namespace index, i.e.
Definition: Title.php:1035
const NS_FILE
Definition: Defines.php:66
output( $out, $channel=null)
Throw some output to the user.
getInterwiki()
Get the interwiki prefix.
Definition: Title.php:924
static makeTitleSafe( $ns, $title, $fragment='', $interwiki='')
Create a new Title from a namespace index and a DB key.
Definition: Title.php:610
progress( $updated)
moveInconsistentPage( $row, Title $title)
$maintClass
moveIllegalPage( $row)
exists( $flags=0)
Check if page exists.
Definition: Title.php:4132
const DB_REPLICA
Definition: defines.php:25
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Maintenance script to clean up broken, unparseable titles.
static makeName( $ns, $title, $fragment='', $interwiki='', $canonicalNamespace=false)
Make a prefixed DB key from a DB key and a namespace index.
Definition: Title.php:811
processRow( $row)
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
getPrefixedDBkey()
Get the prefixed database key form.
Definition: Title.php:1841
static newFromText( $text, $defaultNamespace=NS_MAIN)
Create a new Title from text, such as what one would find in a link.
Definition: Title.php:317