maintenance/refreshLinks.php

   1 <?php
   2 /**
   3  * Refresh link tables.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Maintenance
  22  */
  23
  24 use MediaWiki\MediaWikiServices;
  25 use Wikimedia\Rdbms\IDatabase;
  26
  27 require_once __DIR__ . '/Maintenance.php';
  28
  29 /**
  30  * Maintenance script to refresh link tables.
  31  *
  32  * @ingroup Maintenance
  33  */
  34 class RefreshLinks extends Maintenance {
  35         const REPORTING_INTERVAL = 100;
  36
  37         /** @var int|bool */
  38         protected $namespace = false;
  39
  40         public function __construct() {
  41                 parent::__construct();
  42                 $this->addDescription( 'Refresh link tables' );
  43                 $this->addOption( 'dfn-only', 'Delete links from nonexistent articles only' );
  44                 $this->addOption( 'new-only', 'Only affect articles with just a single edit' );
  45                 $this->addOption( 'redirects-only', 'Only fix redirects, not all links' );
  46                 $this->addOption( 'old-redirects-only', 'Only fix redirects with no redirect table entry' );
  47                 $this->addOption( 'e', 'Last page id to refresh', false, true );
  48                 $this->addOption( 'dfn-chunk-size', 'Maximum number of existent IDs to check per ' .
  49                         'query, default 100000', false, true );
  50                 $this->addOption( 'namespace', 'Only fix pages in this namespace', false, true );
  51                 $this->addOption( 'category', 'Only fix pages in this category', false, true );
  52                 $this->addOption( 'tracking-category', 'Only fix pages in this tracking category', false, true );
  53                 $this->addArg( 'start', 'Page_id to start from, default 1', false );
  54                 $this->setBatchSize( 100 );
  55         }
  56
  57         public function execute() {
  58                 // Note that there is a difference between not specifying the start
  59                 // and end IDs and using the minimum and maximum values from the page
  60                 // table. In the latter case, deleteLinksFromNonexistent() will not
  61                 // delete entries for nonexistent IDs that fall outside the range.
  62                 $start = (int)$this->getArg( 0 ) ?: null;
  63                 $end = (int)$this->getOption( 'e' ) ?: null;
  64                 $dfnChunkSize = (int)$this->getOption( 'dfn-chunk-size', 100000 );
  65                 $ns = $this->getOption( 'namespace' );
  66                 if ( $ns === null ) {
  67                         $this->namespace = false;
  68                 } else {
  69                         $this->namespace = (int)$ns;
  70                 }
  71                 if ( ( $category = $this->getOption( 'category', false ) ) !== false ) {
  72                         $title = Title::makeTitleSafe( NS_CATEGORY, $category );
  73                         if ( !$title ) {
  74                                 $this->fatalError( "'$category' is an invalid category name!\n" );
  75                         }
  76                         $this->refreshCategory( $title );
  77                 } elseif ( ( $category = $this->getOption( 'tracking-category', false ) ) !== false ) {
  78                         $this->refreshTrackingCategory( $category );
  79                 } elseif ( !$this->hasOption( 'dfn-only' ) ) {
  80                         $new = $this->hasOption( 'new-only' );
  81                         $redir = $this->hasOption( 'redirects-only' );
  82                         $oldRedir = $this->hasOption( 'old-redirects-only' );
  83                         $this->doRefreshLinks( $start, $new, $end, $redir, $oldRedir );
  84                         $this->deleteLinksFromNonexistent( null, null, $this->getBatchSize(), $dfnChunkSize );
  85                 } else {
  86                         $this->deleteLinksFromNonexistent( $start, $end, $this->getBatchSize(), $dfnChunkSize );
  87                 }
  88         }
  89
  90         private function namespaceCond() {
  91                 return $this->namespace !== false
  92                         ? [ 'page_namespace' => $this->namespace ]
  93                         : [];
  94         }
  95
  96         /**
  97          * Do the actual link refreshing.
  98          * @param int|null $start Page_id to start from
  99          * @param bool $newOnly Only do pages with 1 edit
 100          * @param int|null $end Page_id to stop at
 101          * @param bool $redirectsOnly Only fix redirects
 102          * @param bool $oldRedirectsOnly Only fix redirects without redirect entries
 103          */
 104         private function doRefreshLinks( $start, $newOnly = false,
 105                 $end = null, $redirectsOnly = false, $oldRedirectsOnly = false
 106         ) {
 107                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 108
 109                 if ( $start === null ) {
 110                         $start = 1;
 111                 }
 112
 113                 // Give extensions a chance to optimize settings
 114                 Hooks::run( 'MaintenanceRefreshLinksInit', [ $this ] );
 115
 116                 $what = $redirectsOnly ? "redirects" : "links";
 117
 118                 if ( $oldRedirectsOnly ) {
 119                         # This entire code path is cut-and-pasted from below.  Hurrah.
 120
 121                         $conds = [
 122                                 "page_is_redirect=1",
 123                                 "rd_from IS NULL",
 124                                 self::intervalCond( $dbr, 'page_id', $start, $end ),
 125                         ] + $this->namespaceCond();
 126
 127                         $res = $dbr->select(
 128                                 [ 'page', 'redirect' ],
 129                                 'page_id',
 130                                 $conds,
 131                                 __METHOD__,
 132                                 [],
 133                                 [ 'redirect' => [ "LEFT JOIN", "page_id=rd_from" ] ]
 134                         );
 135                         $num = $res->numRows();
 136                         $this->output( "Refreshing $num old redirects from $start...\n" );
 137
 138                         $i = 0;
 139
 140                         foreach ( $res as $row ) {
 141                                 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
 142                                         $this->output( "$i\n" );
 143                                         wfWaitForSlaves();
 144                                 }
 145                                 $this->fixRedirect( $row->page_id );
 146                         }
 147                 } elseif ( $newOnly ) {
 148                         $this->output( "Refreshing $what from " );
 149                         $res = $dbr->select( 'page',
 150                                 [ 'page_id' ],
 151                                 [
 152                                         'page_is_new' => 1,
 153                                         self::intervalCond( $dbr, 'page_id', $start, $end ),
 154                                 ] + $this->namespaceCond(),
 155                                 __METHOD__
 156                         );
 157                         $num = $res->numRows();
 158                         $this->output( "$num new articles...\n" );
 159
 160                         $i = 0;
 161                         foreach ( $res as $row ) {
 162                                 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
 163                                         $this->output( "$i\n" );
 164                                         wfWaitForSlaves();
 165                                 }
 166                                 if ( $redirectsOnly ) {
 167                                         $this->fixRedirect( $row->page_id );
 168                                 } else {
 169                                         self::fixLinksFromArticle( $row->page_id, $this->namespace );
 170                                 }
 171                         }
 172                 } else {
 173                         if ( !$end ) {
 174                                 $maxPage = $dbr->selectField( 'page', 'max(page_id)', '', __METHOD__ );
 175                                 $maxRD = $dbr->selectField( 'redirect', 'max(rd_from)', '', __METHOD__ );
 176                                 $end = max( $maxPage, $maxRD );
 177                         }
 178                         $this->output( "Refreshing redirects table.\n" );
 179                         $this->output( "Starting from page_id $start of $end.\n" );
 180
 181                         for ( $id = $start; $id <= $end; $id++ ) {
 182                                 if ( !( $id % self::REPORTING_INTERVAL ) ) {
 183                                         $this->output( "$id\n" );
 184                                         wfWaitForSlaves();
 185                                 }
 186                                 $this->fixRedirect( $id );
 187                         }
 188
 189                         if ( !$redirectsOnly ) {
 190                                 $this->output( "Refreshing links tables.\n" );
 191                                 $this->output( "Starting from page_id $start of $end.\n" );
 192
 193                                 for ( $id = $start; $id <= $end; $id++ ) {
 194                                         if ( !( $id % self::REPORTING_INTERVAL ) ) {
 195                                                 $this->output( "$id\n" );
 196                                                 wfWaitForSlaves();
 197                                         }
 198                                         self::fixLinksFromArticle( $id, $this->namespace );
 199                                 }
 200                         }
 201                 }
 202         }
 203
 204         /**
 205          * Update the redirect entry for a given page.
 206          *
 207          * This methods bypasses the "redirect" table to get the redirect target,
 208          * and parses the page's content to fetch it. This allows to be sure that
 209          * the redirect target is up to date and valid.
 210          * This is particularly useful when modifying namespaces to be sure the
 211          * entry in the "redirect" table points to the correct page and not to an
 212          * invalid one.
 213          *
 214          * @param int $id The page ID to check
 215          */
 216         private function fixRedirect( $id ) {
 217                 $page = WikiPage::newFromID( $id );
 218                 $dbw = $this->getDB( DB_MASTER );
 219
 220                 if ( $page === null ) {
 221                         // This page doesn't exist (any more)
 222                         // Delete any redirect table entry for it
 223                         $dbw->delete( 'redirect', [ 'rd_from' => $id ],
 224                                 __METHOD__ );
 225
 226                         return;
 227                 } elseif ( $this->namespace !== false
 228                         && !$page->getTitle()->inNamespace( $this->namespace )
 229                 ) {
 230                         return;
 231                 }
 232
 233                 $rt = null;
 234                 $content = $page->getContent( Revision::RAW );
 235                 if ( $content !== null ) {
 236                         $rt = $content->getUltimateRedirectTarget();
 237                 }
 238
 239                 if ( $rt === null ) {
 240                         // The page is not a redirect
 241                         // Delete any redirect table entry for it
 242                         $dbw->delete( 'redirect', [ 'rd_from' => $id ], __METHOD__ );
 243                         $fieldValue = 0;
 244                 } else {
 245                         $page->insertRedirectEntry( $rt );
 246                         $fieldValue = 1;
 247                 }
 248
 249                 // Update the page table to be sure it is an a consistent state
 250                 $dbw->update( 'page', [ 'page_is_redirect' => $fieldValue ],
 251                         [ 'page_id' => $id ], __METHOD__ );
 252         }
 253
 254         /**
 255          * Run LinksUpdate for all links on a given page_id
 256          * @param int $id The page_id
 257          * @param int|bool $ns Only fix links if it is in this namespace
 258          */
 259         public static function fixLinksFromArticle( $id, $ns = false ) {
 260                 $page = WikiPage::newFromID( $id );
 261
 262                 MediaWikiServices::getInstance()->getLinkCache()->clear();
 263
 264                 if ( $page === null ) {
 265                         return;
 266                 } elseif ( $ns !== false
 267                         && !$page->getTitle()->inNamespace( $ns ) ) {
 268                         return;
 269                 }
 270
 271                 $content = $page->getContent( Revision::RAW );
 272                 if ( $content === null ) {
 273                         return;
 274                 }
 275
 276                 $updates = $content->getSecondaryDataUpdates(
 277                         $page->getTitle(), /* $old = */ null, /* $recursive = */ false );
 278                 foreach ( $updates as $update ) {
 279                         DeferredUpdates::addUpdate( $update );
 280                         DeferredUpdates::doUpdates();
 281                 }
 282         }
 283
 284         /**
 285          * Removes non-existing links from pages from pagelinks, imagelinks,
 286          * categorylinks, templatelinks, externallinks, interwikilinks, langlinks and redirect tables.
 287          *
 288          * @param int|null $start Page_id to start from
 289          * @param int|null $end Page_id to stop at
 290          * @param int $batchSize The size of deletion batches
 291          * @param int $chunkSize Maximum number of existent IDs to check per query
 292          *
 293          * @author Merlijn van Deen <valhallasw@arctus.nl>
 294          */
 295         private function deleteLinksFromNonexistent( $start = null, $end = null, $batchSize = 100,
 296                 $chunkSize = 100000
 297         ) {
 298                 wfWaitForSlaves();
 299                 $this->output( "Deleting illegal entries from the links tables...\n" );
 300                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 301                 do {
 302                         // Find the start of the next chunk. This is based only
 303                         // on existent page_ids.
 304                         $nextStart = $dbr->selectField(
 305                                 'page',
 306                                 'page_id',
 307                                 [ self::intervalCond( $dbr, 'page_id', $start, $end ) ]
 308                                 + $this->namespaceCond(),
 309                                 __METHOD__,
 310                                 [ 'ORDER BY' => 'page_id', 'OFFSET' => $chunkSize ]
 311                         );
 312
 313                         if ( $nextStart !== false ) {
 314                                 // To find the end of the current chunk, subtract one.
 315                                 // This will serve to limit the number of rows scanned in
 316                                 // dfnCheckInterval(), per query, to at most the sum of
 317                                 // the chunk size and deletion batch size.
 318                                 $chunkEnd = $nextStart - 1;
 319                         } else {
 320                                 // This is the last chunk. Check all page_ids up to $end.
 321                                 $chunkEnd = $end;
 322                         }
 323
 324                         $fmtStart = $start !== null ? "[$start" : '(-INF';
 325                         $fmtChunkEnd = $chunkEnd !== null ? "$chunkEnd]" : 'INF)';
 326                         $this->output( "  Checking interval $fmtStart, $fmtChunkEnd\n" );
 327                         $this->dfnCheckInterval( $start, $chunkEnd, $batchSize );
 328
 329                         $start = $nextStart;
 330
 331                 } while ( $nextStart !== false );
 332         }
 333
 334         /**
 335          * @see RefreshLinks::deleteLinksFromNonexistent()
 336          * @param int|null $start Page_id to start from
 337          * @param int|null $end Page_id to stop at
 338          * @param int $batchSize The size of deletion batches
 339          */
 340         private function dfnCheckInterval( $start = null, $end = null, $batchSize = 100 ) {
 341                 $dbw = $this->getDB( DB_MASTER );
 342                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 343
 344                 $linksTables = [ // table name => page_id field
 345                         'pagelinks' => 'pl_from',
 346                         'imagelinks' => 'il_from',
 347                         'categorylinks' => 'cl_from',
 348                         'templatelinks' => 'tl_from',
 349                         'externallinks' => 'el_from',
 350                         'iwlinks' => 'iwl_from',
 351                         'langlinks' => 'll_from',
 352                         'redirect' => 'rd_from',
 353                         'page_props' => 'pp_page',
 354                 ];
 355
 356                 foreach ( $linksTables as $table => $field ) {
 357                         $this->output( "    $table: 0" );
 358                         $tableStart = $start;
 359                         $counter = 0;
 360                         do {
 361                                 $ids = $dbr->selectFieldValues(
 362                                         $table,
 363                                         $field,
 364                                         [
 365                                                 self::intervalCond( $dbr, $field, $tableStart, $end ),
 366                                                 "$field NOT IN ({$dbr->selectSQLText( 'page', 'page_id' )})",
 367                                         ],
 368                                         __METHOD__,
 369                                         [ 'DISTINCT', 'ORDER BY' => $field, 'LIMIT' => $batchSize ]
 370                                 );
 371
 372                                 $numIds = count( $ids );
 373                                 if ( $numIds ) {
 374                                         $counter += $numIds;
 375                                         $dbw->delete( $table, [ $field => $ids ], __METHOD__ );
 376                                         $this->output( ", $counter" );
 377                                         $tableStart = $ids[$numIds - 1] + 1;
 378                                         wfWaitForSlaves();
 379                                 }
 380
 381                         } while ( $numIds >= $batchSize && ( $end === null || $tableStart <= $end ) );
 382
 383                         $this->output( " deleted.\n" );
 384                 }
 385         }
 386
 387         /**
 388          * Build a SQL expression for a closed interval (i.e. BETWEEN).
 389          *
 390          * By specifying a null $start or $end, it is also possible to create
 391          * half-bounded or unbounded intervals using this function.
 392          *
 393          * @param IDatabase $db
 394          * @param string $var Field name
 395          * @param mixed $start First value to include or null
 396          * @param mixed $end Last value to include or null
 397          * @return string
 398          */
 399         private static function intervalCond( IDatabase $db, $var, $start, $end ) {
 400                 if ( $start === null && $end === null ) {
 401                         return "$var IS NOT NULL";
 402                 } elseif ( $end === null ) {
 403                         return "$var >= {$db->addQuotes( $start )}";
 404                 } elseif ( $start === null ) {
 405                         return "$var <= {$db->addQuotes( $end )}";
 406                 } else {
 407                         return "$var BETWEEN {$db->addQuotes( $start )} AND {$db->addQuotes( $end )}";
 408                 }
 409         }
 410
 411         /**
 412          * Refershes links for pages in a tracking category
 413          *
 414          * @param string $category Category key
 415          */
 416         private function refreshTrackingCategory( $category ) {
 417                 $cats = $this->getPossibleCategories( $category );
 418
 419                 if ( !$cats ) {
 420                         $this->error( "Tracking category '$category' is disabled\n" );
 421                         // Output to stderr but don't bail out,
 422                 }
 423
 424                 foreach ( $cats as $cat ) {
 425                         $this->refreshCategory( $cat );
 426                 }
 427         }
 428
 429         /**
 430          * Refreshes links to a category
 431          *
 432          * @param Title $category
 433          */
 434         private function refreshCategory( Title $category ) {
 435                 $this->output( "Refreshing pages in category '{$category->getText()}'...\n" );
 436
 437                 $dbr = $this->getDB( DB_REPLICA );
 438                 $conds = [
 439                         'page_id=cl_from',
 440                         'cl_to' => $category->getDBkey(),
 441                 ];
 442                 if ( $this->namespace !== false ) {
 443                         $conds['page_namespace'] = $this->namespace;
 444                 }
 445
 446                 $i = 0;
 447                 $timestamp = '';
 448                 $lastId = 0;
 449                 do {
 450                         $finalConds = $conds;
 451                         $timestamp = $dbr->addQuotes( $timestamp );
 452                         $finalConds [] =
 453                                 "(cl_timestamp > $timestamp OR (cl_timestamp = $timestamp AND cl_from > $lastId))";
 454                         $res = $dbr->select( [ 'page', 'categorylinks' ],
 455                                 [ 'page_id', 'cl_timestamp' ],
 456                                 $finalConds,
 457                                 __METHOD__,
 458                                 [
 459                                         'ORDER BY' => [ 'cl_timestamp', 'cl_from' ],
 460                                         'LIMIT' => $this->getBatchSize(),
 461                                 ]
 462                         );
 463
 464                         foreach ( $res as $row ) {
 465                                 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
 466                                         $this->output( "$i\n" );
 467                                         wfWaitForSlaves();
 468                                 }
 469                                 $lastId = $row->page_id;
 470                                 $timestamp = $row->cl_timestamp;
 471                                 self::fixLinksFromArticle( $row->page_id );
 472                         }
 473
 474                 } while ( $res->numRows() == $this->getBatchSize() );
 475         }
 476
 477         /**
 478          * Returns a list of possible categories for a given tracking category key
 479          *
 480          * @param string $categoryKey
 481          * @return Title[]
 482          */
 483         private function getPossibleCategories( $categoryKey ) {
 484                 $trackingCategories = new TrackingCategories( $this->getConfig() );
 485                 $cats = $trackingCategories->getTrackingCategories();
 486                 if ( isset( $cats[$categoryKey] ) ) {
 487                         return $cats[$categoryKey]['cats'];
 488                 }
 489                 $this->fatalError( "Unknown tracking category {$categoryKey}\n" );
 490         }
 491 }
 492
 493 $maintClass = RefreshLinks::class;
 494 require_once RUN_MAINTENANCE_IF_MAIN;