MediaWiki  master
findMissingActors.php
Go to the documentation of this file.
1 <?php
29 
30 require_once __DIR__ . '/Maintenance.php';
31 
38 
42  private $userFactory;
43 
47  private $userNameUtils;
48 
52  private $loadBalancer;
53 
57  private $lbFactory;
58 
63 
65  private $tables;
66 
67  public function __construct() {
68  parent::__construct();
69 
70  $this->addDescription( 'Find and fix invalid actor IDs.' );
71  $this->addOption( 'field', 'The name of a database field to process',
72  true, true );
73  $this->addOption( 'type', 'Which type of invalid actors to find or fix, '
74  . 'missing or broken (with empty actor_name which can\'t be associated '
75  . 'with an existing user).',
76  false, true );
77  $this->addOption( 'skip', 'A comma-separated list of actor IDs to skip.',
78  false, true );
79  $this->addOption( 'overwrite-with', 'Replace invalid actors with this user. '
80  . 'Typically, this would be "Unknown user", but it could be any reserved '
81  . 'system user (per $wgReservedUsernames) or locally registered user. '
82  . 'If not given, invalid actors will only be listed, not fixed. '
83  . 'You will be prompted for confirmation before data is written. ',
84  false, true );
85 
86  $this->setBatchSize( 1000 );
87  }
88 
89  public function initializeServices(
90  ?UserFactory $userFactory = null,
93  ?LBFactory $lbFactory = null,
95  ) {
96  $services = MediaWikiServices::getInstance();
97 
98  $this->userFactory = $userFactory ?? $this->userFactory ?? $services->getUserFactory();
99  $this->userNameUtils = $userNameUtils ?? $this->userNameUtils ?? $services->getUserNameUtils();
100  $this->loadBalancer = $loadBalancer ?? $this->loadBalancer ?? $services->getDBLoadBalancer();
101  $this->lbFactory = $lbFactory ?? $this->lbFactory ?? $services->getDBLoadBalancerFactory();
102  $this->actorNormalization = $actorNormalization ?? $this->actorNormalization ??
103  $services->getActorNormalization();
104  }
105 
109  private function getTables() {
110  if ( !$this->tables ) {
111  $tables = [
112  'ar_actor' => [ 'archive', 'ar_actor', 'ar_id' ],
113  'ipb_by_actor' => [ 'ipblocks', 'ipb_by_actor', 'ipb_id' ], // no index on ipb_by_actor!
114  'img_actor' => [ 'image', 'img_actor', 'img_name' ],
115  'oi_actor' => [ 'oldimage', 'oi_actor', 'oi_archive_name' ], // no index on oi_archive_name!
116  'fa_actor' => [ 'filearchive', 'fa_actor', 'fa_id' ],
117  'rc_actor' => [ 'recentchanges', 'rc_actor', 'rc_id' ],
118  'log_actor' => [ 'logging', 'log_actor', 'log_id' ],
119  'rev_actor' => [ 'revision', 'rev_actor', 'rev_id' ],
120  ];
121  $this->tables = $tables;
122  }
123  return $this->tables;
124  }
125 
130  private function getTableInfo( $field ) {
131  $tables = $this->getTables();
132  return $tables[$field] ?? null;
133  }
134 
144  private function getNewActorId() {
145  $name = $this->getOption( 'overwrite-with' );
146 
147  if ( $name === null ) {
148  return null;
149  }
150 
151  $user = $this->userFactory->newFromName( $name );
152 
153  if ( !$user ) {
154  $this->fatalError( "Not a valid user name: '$name'" );
155  }
156 
157  $name = $this->userNameUtils->getCanonical( $name, UserRigorOptions::RIGOR_NONE );
158 
159  if ( $user->isRegistered() ) {
160  $this->output( "Using existing user: '$user'\n" );
161  } elseif ( !$this->userNameUtils->isValid( $name ) ) {
162  $this->fatalError( "Not a valid user name: '$name'" );
163  } elseif ( !$this->userNameUtils->isUsable( $name ) ) {
164  $this->output( "Using system user: '$name'\n" );
165  } else {
166  $this->fatalError( "Unknown user: '$name'" );
167  }
168 
169  $dbw = $this->loadBalancer->getConnectionRef( DB_PRIMARY );
170  $actorId = $this->actorNormalization->acquireActorId( $user, $dbw );
171 
172  if ( !$actorId ) {
173  $this->fatalError( "Failed to acquire an actor ID for user '$user'" );
174  }
175 
176  $this->output( "Replacement actor ID is $actorId.\n" );
177  return $actorId;
178  }
179 
180  public function execute() {
181  $this->initializeServices();
182 
183  $field = $this->getOption( 'field' );
184  if ( !$this->getTableInfo( $field ) ) {
185  $this->fatalError( "Unknown field: $field.\n" );
186  }
187 
188  $type = $this->getOption( 'type', 'missing' );
189  if ( $type !== 'missing' && $type !== 'broken' ) {
190  $this->fatalError( "Unknown type: $type.\n" );
191  }
192 
193  $skip = $this->parseIntList( $this->getOption( 'skip', '' ) );
194  $overwrite = $this->getNewActorId();
195 
196  $bad = $this->findBadActors( $field, $type, $skip );
197 
198  if ( $bad && $overwrite ) {
199  $this->output( "\n" );
200  $this->output( "Do you want to OVERWRITE the listed actor IDs?\n" );
201  $this->output( "Information about the invalid IDs will be lost!\n" );
202  $this->output( "\n" );
203  $confirm = self::readconsole( 'Type "yes" to continue: ' );
204 
205  if ( $confirm === 'yes' ) {
206  $this->overwriteActorIDs( $field, array_keys( $bad ), $overwrite );
207  } else {
208  $this->fatalError( 'Aborted.' );
209  }
210  }
211 
212  $this->output( "Done.\n" );
213  }
214 
224  private function findBadActors( $field, $type, $skip ) {
225  [ $table, $actorField, $idField ] = $this->getTableInfo( $field );
226  $this->output( "Finding invalid actor IDs in $table.$actorField...\n" );
227 
228  $dbr = $this->loadBalancer->getConnectionRef(
229  DB_REPLICA,
230  [ 'maintenance', 'vslow', 'slow' ]
231  );
232 
233  /*
234  We are building an SQL query like this one here, performing a left join
235  to detect rows in $table that lack a matching row in the actor table.
236 
237  In this example, $field is 'log_actor', so $table is 'logging',
238  $actorField is 'log_actor', and $idField is 'log_id'.
239  Further, $skip is [ 1, 2, 3, 4 ] and the batch size is 1000.
240 
241  SELECT log_id
242  FROM logging
243  LEFT JOIN actor ON log_actor = actor_id
244  WHERE actor_id IS NULL
245  AND log_actor NOT IN (1, 2, 3, 4)
246  LIMIT 1000;
247  */
248 
249  $conds = $type == 'missing'
250  ? [ 'actor_id' => null ]
251  : [ 'actor_name' => '' ];
252 
253  if ( $skip ) {
254  $conds[] = $actorField . ' NOT IN ( ' . $dbr->makeList( $skip ) . ' ) ';
255  }
256 
257  $queryBuilder = $dbr->newSelectQueryBuilder();
258  $queryBuilder->table( $table )
259  ->fields( [ $actorField, $idField ] )
260  ->conds( $conds )
261  ->leftJoin( 'actor', null, [ "$actorField = actor_id" ] )
262  ->limit( $this->getBatchSize() )
263  ->caller( __METHOD__ );
264 
265  $res = $queryBuilder->fetchResultSet();
266  $count = $res->numRows();
267 
268  $bad = [];
269 
270  if ( $count ) {
271  $this->output( "\t\tID\tACTOR\n" );
272  }
273 
274  foreach ( $res as $row ) {
275  $id = $row->$idField;
276  $actor = (int)( $row->$actorField );
277 
278  $bad[$id] = $actor;
279  $this->output( "\t\t$id\t$actor\n" );
280  }
281 
282  $this->output( "\tFound $count invalid actor IDs.\n" );
283 
284  if ( $count >= $this->getBatchSize() ) {
285  $this->output( "\tBatch size reached, run again after fixing the current batch.\n" );
286  }
287 
288  return $bad;
289  }
290 
300  private function overwriteActorIDs( $field, array $ids, int $overwrite ) {
301  [ $table, $actorField, $idField ] = $this->getTableInfo( $field );
302 
303  $count = count( $ids );
304  $this->output( "OVERWRITING $count actor IDs in $table.$actorField with $overwrite...\n" );
305 
306  $dbw = $this->loadBalancer->getConnectionRef( DB_PRIMARY );
307 
308  $dbw->update( $table, [ $actorField => $overwrite ], [ $idField => $ids ], __METHOD__ );
309 
310  $count = $dbw->affectedRows();
311 
312  $this->lbFactory->waitForReplication();
313  $this->output( "\tUpdated $count rows.\n" );
314 
315  return $count;
316  }
317 
318 }
319 
320 $maintClass = FindMissingActors::class;
321 require_once RUN_MAINTENANCE_IF_MAIN;
Maintenance script for finding and replacing invalid actor IDs, see T261325 and T307738.
findBadActors( $field, $type, $skip)
Find rows that have bad actor IDs.
execute()
Do the actual work.
UserNameUtils null $userNameUtils
ActorNormalization $actorNormalization
getNewActorId()
Returns the actor ID of the user specified with the –overwrite-with option, or null if –overwrite-wit...
initializeServices(?UserFactory $userFactory=null, ?UserNameUtils $userNameUtils=null, ?LoadBalancer $loadBalancer=null, ?LBFactory $lbFactory=null, ?ActorNormalization $actorNormalization=null)
UserFactory null $userFactory
overwriteActorIDs( $field, array $ids, int $overwrite)
Overwrite the actor ID in a given set of rows.
__construct()
Default constructor.
LoadBalancer null $loadBalancer
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:66
output( $out, $channel=null)
Throw some output to the user.
static readconsole( $prompt='> ')
Prompt the console for input.
getBatchSize()
Returns batch size.
parseIntList( $text)
Utility function to parse a string (perhaps from a command line option) into a list of integers (perh...
addDescription( $text)
Set the description text.
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
getOption( $name, $default=null)
Get an option, or return the default.
setBatchSize( $s=0)
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Service locator for MediaWiki core services.
Creates User objects.
Definition: UserFactory.php:38
UserNameUtils service.
An interface for generating database load balancers.
Definition: LBFactory.php:43
Database connection, tracking, load balancing, and transaction manager for a cluster.
Service for dealing with the actor table.
Shared interface for rigor levels when dealing with User methods.
const DB_REPLICA
Definition: defines.php:26
const DB_PRIMARY
Definition: defines.php:28