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