MediaWiki  master
orphans.php
Go to the documentation of this file.
1 <?php
31 require_once __DIR__ . '/Maintenance.php';
32 
35 
42 class Orphans extends Maintenance {
43  public function __construct() {
44  parent::__construct();
45  $this->addDescription( "Look for 'orphan' revisions hooked to pages which don't exist\n" .
46  "and 'childless' pages with no revisions\n" .
47  "Then, kill the poor widows and orphans\n" .
48  "Man this is depressing"
49  );
50  $this->addOption( 'fix', 'Actually fix broken entries' );
51  }
52 
53  public function execute() {
54  $this->checkOrphans( $this->hasOption( 'fix' ) );
55  $this->checkSeparation( $this->hasOption( 'fix' ) );
56  }
57 
63  private function lockTables( $db, $extraTable = [] ) {
64  $tbls = [ 'page', 'revision', 'redirect' ];
65  if ( $extraTable ) {
66  $tbls = array_merge( $tbls, $extraTable );
67  }
68  $db->lockTables( [], $tbls, __METHOD__ );
69  }
70 
75  private function checkOrphans( $fix ) {
76  $dbw = $this->getDB( DB_MASTER );
77  $commentStore = CommentStore::getStore();
78 
79  if ( $fix ) {
80  $this->lockTables( $dbw );
81  }
82 
83  $commentQuery = $commentStore->getJoin( 'rev_comment' );
84  $actorQuery = ActorMigration::newMigration()->getJoin( 'rev_user' );
85 
86  $this->output( "Checking for orphan revision table entries... "
87  . "(this may take a while on a large wiki)\n" );
88  $result = $dbw->select(
89  [ 'revision', 'page' ] + $commentQuery['tables'] + $actorQuery['tables'],
90  [ 'rev_id', 'rev_page', 'rev_timestamp' ] + $commentQuery['fields'] + $actorQuery['fields'],
91  [ 'page_id' => null ],
92  __METHOD__,
93  [],
94  [ 'page' => [ 'LEFT JOIN', [ 'rev_page=page_id' ] ] ] + $commentQuery['joins']
95  + $actorQuery['joins']
96  );
97  $orphans = $result->numRows();
98  if ( $orphans > 0 ) {
99  $this->output( "$orphans orphan revisions...\n" );
100  $this->output( sprintf(
101  "%10s %10s %14s %20s %s\n",
102  'rev_id', 'rev_page', 'rev_timestamp', 'rev_user_text', 'rev_comment'
103  ) );
104 
105  $contLang = MediaWikiServices::getInstance()->getContentLanguage();
106  foreach ( $result as $row ) {
107  $comment = $commentStore->getComment( 'rev_comment', $row )->text;
108  if ( $comment !== '' ) {
109  $comment = '(' . $contLang->truncateForVisual( $comment, 40 ) . ')';
110  }
111  $rev_user_text = $contLang->truncateForVisual( $row->rev_user_text, 20 );
112  # pad $rev_user_text to 20 characters. Note that this may
113  # yield poor results if $rev_user_text contains combining
114  # or half-width characters. Alas.
115  if ( mb_strlen( $rev_user_text ) < 20 ) {
116  $rev_user_text = str_repeat( ' ', 20 - mb_strlen( $rev_user_text ) );
117  }
118  $this->output( sprintf( "%10d %10d %14s %s %s\n",
119  $row->rev_id,
120  $row->rev_page,
121  $row->rev_timestamp,
122  $rev_user_text,
123  $comment ) );
124  if ( $fix ) {
125  $dbw->delete( 'revision', [ 'rev_id' => $row->rev_id ], __METHOD__ );
126  }
127  }
128  if ( !$fix ) {
129  $this->output( "Run again with --fix to remove these entries automatically.\n" );
130  }
131  } else {
132  $this->output( "No orphans! Yay!\n" );
133  }
134 
135  if ( $fix ) {
136  $dbw->unlockTables( __METHOD__ );
137  }
138  }
139 
144  private function checkSeparation( $fix ) {
145  $dbw = $this->getDB( DB_MASTER );
146  $page = $dbw->tableName( 'page' );
147  $revision = $dbw->tableName( 'revision' );
148 
149  if ( $fix ) {
150  $this->lockTables( $dbw, [ 'user', 'text' ] );
151  }
152 
153  $this->output( "\nChecking for pages whose page_latest links are incorrect... "
154  . "(this may take a while on a large wiki)\n" );
155  $result = $dbw->query( "
156  SELECT *
157  FROM $page LEFT JOIN $revision ON page_latest=rev_id
158  ", __METHOD__ );
159  $found = 0;
160  $revLookup = MediaWikiServices::getInstance()->getRevisionLookup();
161  foreach ( $result as $row ) {
162  $result2 = $dbw->query( "
163  SELECT MAX(rev_timestamp) as max_timestamp
164  FROM $revision
165  WHERE rev_page=" . (int)( $row->page_id ),
166  __METHOD__
167  );
168  $row2 = $dbw->fetchObject( $result2 );
169  if ( $row2 ) {
170  if ( $row->rev_timestamp != $row2->max_timestamp ) {
171  if ( $found == 0 ) {
172  $this->output( sprintf( "%10s %10s %14s %14s\n",
173  'page_id', 'rev_id', 'timestamp', 'max timestamp' ) );
174  }
175  ++$found;
176  $this->output( sprintf( "%10d %10d %14s %14s\n",
177  $row->page_id,
178  $row->page_latest,
179  $row->rev_timestamp,
180  $row2->max_timestamp ) );
181  if ( $fix ) {
182  # ...
183  $maxId = $dbw->selectField(
184  'revision',
185  'rev_id',
186  [
187  'rev_page' => $row->page_id,
188  'rev_timestamp' => $row2->max_timestamp
189  ],
190  __METHOD__
191  );
192  $this->output( "... updating to revision $maxId\n" );
193  $maxRev = $revLookup->getRevisionById( $maxId );
194  $title = Title::makeTitle( $row->page_namespace, $row->page_title );
195  $article = WikiPage::factory( $title );
196  $article->updateRevisionOn( $dbw, $maxRev );
197  }
198  }
199  } else {
200  $this->output( "wtf\n" );
201  }
202  }
203 
204  if ( $found ) {
205  $this->output( "Found $found pages with incorrect latest revision.\n" );
206  } else {
207  $this->output( "No pages with incorrect latest revision. Yay!\n" );
208  }
209  if ( !$fix && $found > 0 ) {
210  $this->output( "Run again with --fix to remove these entries automatically.\n" );
211  }
212 
213  if ( $fix ) {
214  $dbw->unlockTables( __METHOD__ );
215  }
216  }
217 }
218 
219 $maintClass = Orphans::class;
220 require_once RUN_MAINTENANCE_IF_MAIN;
RUN_MAINTENANCE_IF_MAIN
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:38
MediaWiki\MediaWikiServices
MediaWikiServices is the service locator for the application scope of MediaWiki.
Definition: MediaWikiServices.php:160
Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition: Maintenance.php:327
Maintenance
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:55
ActorMigration\newMigration
static newMigration()
Static constructor.
Definition: ActorMigration.php:140
WikiPage\factory
static factory(Title $title)
Create a WikiPage object of the appropriate class for the given title.
Definition: WikiPage.php:159
Orphans\__construct
__construct()
Default constructor.
Definition: orphans.php:43
Orphans\checkSeparation
checkSeparation( $fix)
Check for pages where page_latest is wrong.
Definition: orphans.php:144
Maintenance\addOption
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
Definition: Maintenance.php:245
$title
$title
Definition: testCompression.php:38
Title\makeTitle
static makeTitle( $ns, $title, $fragment='', $interwiki='')
Create a new Title from a namespace index and a DB key.
Definition: Title.php:591
DB_MASTER
const DB_MASTER
Definition: defines.php:26
Orphans\checkOrphans
checkOrphans( $fix)
Check for orphan revisions.
Definition: orphans.php:75
Orphans\lockTables
lockTables( $db, $extraTable=[])
Lock the appropriate tables for the script.
Definition: orphans.php:63
Maintenance\getDB
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
Definition: Maintenance.php:1366
Orphans
Maintenance script that looks for 'orphan' revisions hooked to pages which don't exist and 'childless...
Definition: orphans.php:42
Maintenance\output
output( $out, $channel=null)
Throw some output to the user.
Definition: Maintenance.php:434
Orphans\execute
execute()
Do the actual work.
Definition: orphans.php:53
Maintenance\hasOption
hasOption( $name)
Checks to see if a particular option exists.
Definition: Maintenance.php:266
CommentStore\getStore
static getStore()
Definition: CommentStore.php:120
Wikimedia\Rdbms\IMaintainableDatabase
Advanced database interface for IDatabase handles that include maintenance methods.
Definition: IMaintainableDatabase.php:38
$maintClass
$maintClass
Definition: orphans.php:219