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 require_once __DIR__ . '/Maintenance.php';
  25
  26 /**
  27  * Maintenance script to refresh link tables.
  28  *
  29  * @ingroup Maintenance
  30  */
  31 class RefreshLinks extends Maintenance {
  32         /** @var int|bool */
  33         protected $namespace = false;
  34
  35         public function __construct() {
  36                 parent::__construct();
  37                 $this->addDescription( 'Refresh link tables' );
  38                 $this->addOption( 'dfn-only', 'Delete links from nonexistent articles only' );
  39                 $this->addOption( 'new-only', 'Only affect articles with just a single edit' );
  40                 $this->addOption( 'redirects-only', 'Only fix redirects, not all links' );
  41                 $this->addOption( 'old-redirects-only', 'Only fix redirects with no redirect table entry' );
  42                 $this->addOption( 'e', 'Last page id to refresh', false, true );
  43                 $this->addOption( 'dfn-chunk-size', 'Maximum number of existent IDs to check per ' .
  44                         'query, default 100000', false, true );
  45                 $this->addOption( 'namespace', 'Only fix pages in this namespace', false, true );
  46                 $this->addArg( 'start', 'Page_id to start from, default 1', false );
  47                 $this->setBatchSize( 100 );
  48         }
  49
  50         public function execute() {
  51                 // Note that there is a difference between not specifying the start
  52                 // and end IDs and using the minimum and maximum values from the page
  53                 // table. In the latter case, deleteLinksFromNonexistent() will not
  54                 // delete entries for nonexistent IDs that fall outside the range.
  55                 $start = (int)$this->getArg( 0 ) ?: null;
  56                 $end = (int)$this->getOption( 'e' ) ?: null;
  57                 $dfnChunkSize = (int)$this->getOption( 'dfn-chunk-size', 100000 );
  58                 $ns = $this->getOption( 'namespace' );
  59                 if ( $ns === null ) {
  60                         $this->namespace = false;
  61                 } else {
  62                         $this->namespace = (int)$ns;
  63                 }
  64                 if ( !$this->hasOption( 'dfn-only' ) ) {
  65                         $new = $this->getOption( 'new-only', false );
  66                         $redir = $this->getOption( 'redirects-only', false );
  67                         $oldRedir = $this->getOption( 'old-redirects-only', false );
  68                         $this->doRefreshLinks( $start, $new, $end, $redir, $oldRedir );
  69                         $this->deleteLinksFromNonexistent( null, null, $this->mBatchSize, $dfnChunkSize );
  70                 } else {
  71                         $this->deleteLinksFromNonexistent( $start, $end, $this->mBatchSize, $dfnChunkSize );
  72                 }
  73         }
  74
  75         private function namespaceCond() {
  76                 return $this->namespace !== false
  77                         ? [ 'page_namespace' => $this->namespace ]
  78                         : [];
  79         }
  80
  81         /**
  82          * Do the actual link refreshing.
  83          * @param int|null $start Page_id to start from
  84          * @param bool $newOnly Only do pages with 1 edit
  85          * @param int|null $end Page_id to stop at
  86          * @param bool $redirectsOnly Only fix redirects
  87          * @param bool $oldRedirectsOnly Only fix redirects without redirect entries
  88          */
  89         private function doRefreshLinks( $start, $newOnly = false,
  90                 $end = null, $redirectsOnly = false, $oldRedirectsOnly = false
  91         ) {
  92                 $reportingInterval = 100;
  93                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
  94
  95                 if ( $start === null ) {
  96                         $start = 1;
  97                 }
  98
  99                 // Give extensions a chance to optimize settings
 100                 Hooks::run( 'MaintenanceRefreshLinksInit', [ $this ] );
 101
 102                 $what = $redirectsOnly ? "redirects" : "links";
 103
 104                 if ( $oldRedirectsOnly ) {
 105                         # This entire code path is cut-and-pasted from below.  Hurrah.
 106
 107                         $conds = [
 108                                 "page_is_redirect=1",
 109                                 "rd_from IS NULL",
 110                                 self::intervalCond( $dbr, 'page_id', $start, $end ),
 111                         ] + $this->namespaceCond();
 112
 113                         $res = $dbr->select(
 114                                 [ 'page', 'redirect' ],
 115                                 'page_id',
 116                                 $conds,
 117                                 __METHOD__,
 118                                 [],
 119                                 [ 'redirect' => [ "LEFT JOIN", "page_id=rd_from" ] ]
 120                         );
 121                         $num = $res->numRows();
 122                         $this->output( "Refreshing $num old redirects from $start...\n" );
 123
 124                         $i = 0;
 125
 126                         foreach ( $res as $row ) {
 127                                 if ( !( ++$i % $reportingInterval ) ) {
 128                                         $this->output( "$i\n" );
 129                                         wfWaitForSlaves();
 130                                 }
 131                                 $this->fixRedirect( $row->page_id );
 132                         }
 133                 } elseif ( $newOnly ) {
 134                         $this->output( "Refreshing $what from " );
 135                         $res = $dbr->select( 'page',
 136                                 [ 'page_id' ],
 137                                 [
 138                                         'page_is_new' => 1,
 139                                         self::intervalCond( $dbr, 'page_id', $start, $end ),
 140                                 ] + $this->namespaceCond(),
 141                                 __METHOD__
 142                         );
 143                         $num = $res->numRows();
 144                         $this->output( "$num new articles...\n" );
 145
 146                         $i = 0;
 147                         foreach ( $res as $row ) {
 148                                 if ( !( ++$i % $reportingInterval ) ) {
 149                                         $this->output( "$i\n" );
 150                                         wfWaitForSlaves();
 151                                 }
 152                                 if ( $redirectsOnly ) {
 153                                         $this->fixRedirect( $row->page_id );
 154                                 } else {
 155                                         self::fixLinksFromArticle( $row->page_id, $this->namespace );
 156                                 }
 157                         }
 158                 } else {
 159                         if ( !$end ) {
 160                                 $maxPage = $dbr->selectField( 'page', 'max(page_id)', false );
 161                                 $maxRD = $dbr->selectField( 'redirect', 'max(rd_from)', false );
 162                                 $end = max( $maxPage, $maxRD );
 163                         }
 164                         $this->output( "Refreshing redirects table.\n" );
 165                         $this->output( "Starting from page_id $start of $end.\n" );
 166
 167                         for ( $id = $start; $id <= $end; $id++ ) {
 168
 169                                 if ( !( $id % $reportingInterval ) ) {
 170                                         $this->output( "$id\n" );
 171                                         wfWaitForSlaves();
 172                                 }
 173                                 $this->fixRedirect( $id );
 174                         }
 175
 176                         if ( !$redirectsOnly ) {
 177                                 $this->output( "Refreshing links tables.\n" );
 178                                 $this->output( "Starting from page_id $start of $end.\n" );
 179
 180                                 for ( $id = $start; $id <= $end; $id++ ) {
 181
 182                                         if ( !( $id % $reportingInterval ) ) {
 183                                                 $this->output( "$id\n" );
 184                                                 wfWaitForSlaves();
 185                                         }
 186                                         self::fixLinksFromArticle( $id, $this->namespace );
 187                                 }
 188                         }
 189                 }
 190         }
 191
 192         /**
 193          * Update the redirect entry for a given page.
 194          *
 195          * This methods bypasses the "redirect" table to get the redirect target,
 196          * and parses the page's content to fetch it. This allows to be sure that
 197          * the redirect target is up to date and valid.
 198          * This is particularly useful when modifying namespaces to be sure the
 199          * entry in the "redirect" table points to the correct page and not to an
 200          * invalid one.
 201          *
 202          * @param int $id The page ID to check
 203          */
 204         private function fixRedirect( $id ) {
 205                 $page = WikiPage::newFromID( $id );
 206                 $dbw = $this->getDB( DB_MASTER );
 207
 208                 if ( $page === null ) {
 209                         // This page doesn't exist (any more)
 210                         // Delete any redirect table entry for it
 211                         $dbw->delete( 'redirect', [ 'rd_from' => $id ],
 212                                 __METHOD__ );
 213
 214                         return;
 215                 } elseif ( $this->namespace !== false
 216                         && !$page->getTitle()->inNamespace( $this->namespace )
 217                 ) {
 218                         return;
 219                 }
 220
 221                 $rt = null;
 222                 $content = $page->getContent( Revision::RAW );
 223                 if ( $content !== null ) {
 224                         $rt = $content->getUltimateRedirectTarget();
 225                 }
 226
 227                 if ( $rt === null ) {
 228                         // The page is not a redirect
 229                         // Delete any redirect table entry for it
 230                         $dbw->delete( 'redirect', [ 'rd_from' => $id ], __METHOD__ );
 231                         $fieldValue = 0;
 232                 } else {
 233                         $page->insertRedirectEntry( $rt );
 234                         $fieldValue = 1;
 235                 }
 236
 237                 // Update the page table to be sure it is an a consistent state
 238                 $dbw->update( 'page', [ 'page_is_redirect' => $fieldValue ],
 239                         [ 'page_id' => $id ], __METHOD__ );
 240         }
 241
 242         /**
 243          * Run LinksUpdate for all links on a given page_id
 244          * @param int $id The page_id
 245          * @param int|bool $ns Only fix links if it is in this namespace
 246          */
 247         public static function fixLinksFromArticle( $id, $ns = false ) {
 248                 $page = WikiPage::newFromID( $id );
 249
 250                 LinkCache::singleton()->clear();
 251
 252                 if ( $page === null ) {
 253                         return;
 254                 } elseif ( $ns !== false
 255                         && !$page->getTitle()->inNamespace( $ns ) ) {
 256                         return;
 257                 }
 258
 259                 $content = $page->getContent( Revision::RAW );
 260                 if ( $content === null ) {
 261                         return;
 262                 }
 263
 264                 foreach ( $content->getSecondaryDataUpdates( $page->getTitle() ) as $update ) {
 265                         DeferredUpdates::addUpdate( $update );
 266                 }
 267         }
 268
 269         /**
 270          * Removes non-existing links from pages from pagelinks, imagelinks,
 271          * categorylinks, templatelinks, externallinks, interwikilinks, langlinks and redirect tables.
 272          *
 273          * @param int|null $start Page_id to start from
 274          * @param int|null $end Page_id to stop at
 275          * @param int $batchSize The size of deletion batches
 276          * @param int $chunkSize Maximum number of existent IDs to check per query
 277          *
 278          * @author Merlijn van Deen <valhallasw@arctus.nl>
 279          */
 280         private function deleteLinksFromNonexistent( $start = null, $end = null, $batchSize = 100,
 281                 $chunkSize = 100000
 282         ) {
 283                 wfWaitForSlaves();
 284                 $this->output( "Deleting illegal entries from the links tables...\n" );
 285                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 286                 do {
 287                         // Find the start of the next chunk. This is based only
 288                         // on existent page_ids.
 289                         $nextStart = $dbr->selectField(
 290                                 'page',
 291                                 'page_id',
 292                                 [ self::intervalCond( $dbr, 'page_id', $start, $end ) ]
 293                                 + $this->namespaceCond(),
 294                                 __METHOD__,
 295                                 [ 'ORDER BY' => 'page_id', 'OFFSET' => $chunkSize ]
 296                         );
 297
 298                         if ( $nextStart !== false ) {
 299                                 // To find the end of the current chunk, subtract one.
 300                                 // This will serve to limit the number of rows scanned in
 301                                 // dfnCheckInterval(), per query, to at most the sum of
 302                                 // the chunk size and deletion batch size.
 303                                 $chunkEnd = $nextStart - 1;
 304                         } else {
 305                                 // This is the last chunk. Check all page_ids up to $end.
 306                                 $chunkEnd = $end;
 307                         }
 308
 309                         $fmtStart = $start !== null ? "[$start" : '(-INF';
 310                         $fmtChunkEnd = $chunkEnd !== null ? "$chunkEnd]" : 'INF)';
 311                         $this->output( "  Checking interval $fmtStart, $fmtChunkEnd\n" );
 312                         $this->dfnCheckInterval( $start, $chunkEnd, $batchSize );
 313
 314                         $start = $nextStart;
 315
 316                 } while ( $nextStart !== false );
 317         }
 318
 319         /**
 320          * @see RefreshLinks::deleteLinksFromNonexistent()
 321          * @param int|null $start Page_id to start from
 322          * @param int|null $end Page_id to stop at
 323          * @param int $batchSize The size of deletion batches
 324          */
 325         private function dfnCheckInterval( $start = null, $end = null, $batchSize = 100 ) {
 326                 $dbw = $this->getDB( DB_MASTER );
 327                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 328
 329                 $linksTables = [ // table name => page_id field
 330                         'pagelinks' => 'pl_from',
 331                         'imagelinks' => 'il_from',
 332                         'categorylinks' => 'cl_from',
 333                         'templatelinks' => 'tl_from',
 334                         'externallinks' => 'el_from',
 335                         'iwlinks' => 'iwl_from',
 336                         'langlinks' => 'll_from',
 337                         'redirect' => 'rd_from',
 338                         'page_props' => 'pp_page',
 339                 ];
 340
 341                 foreach ( $linksTables as $table => $field ) {
 342                         $this->output( "    $table: 0" );
 343                         $tableStart = $start;
 344                         $counter = 0;
 345                         do {
 346                                 $ids = $dbr->selectFieldValues(
 347                                         $table,
 348                                         $field,
 349                                         [
 350                                                 self::intervalCond( $dbr, $field, $tableStart, $end ),
 351                                                 "$field NOT IN ({$dbr->selectSQLText( 'page', 'page_id' )})",
 352                                         ],
 353                                         __METHOD__,
 354                                         [ 'DISTINCT', 'ORDER BY' => $field, 'LIMIT' => $batchSize ]
 355                                 );
 356
 357                                 $numIds = count( $ids );
 358                                 if ( $numIds ) {
 359                                         $counter += $numIds;
 360                                         $dbw->delete( $table, [ $field => $ids ], __METHOD__ );
 361                                         $this->output( ", $counter" );
 362                                         $tableStart = $ids[$numIds - 1] + 1;
 363                                         wfWaitForSlaves();
 364                                 }
 365
 366                         } while ( $numIds >= $batchSize && ( $end === null || $tableStart <= $end ) );
 367
 368                         $this->output( " deleted.\n" );
 369                 }
 370         }
 371
 372         /**
 373          * Build a SQL expression for a closed interval (i.e. BETWEEN).
 374          *
 375          * By specifying a null $start or $end, it is also possible to create
 376          * half-bounded or unbounded intervals using this function.
 377          *
 378          * @param IDatabase $db Database connection
 379          * @param string $var Field name
 380          * @param mixed $start First value to include or null
 381          * @param mixed $end Last value to include or null
 382          */
 383         private static function intervalCond( IDatabase $db, $var, $start, $end ) {
 384                 if ( $start === null && $end === null ) {
 385                         return "$var IS NOT NULL";
 386                 } elseif ( $end === null ) {
 387                         return "$var >= {$db->addQuotes( $start )}";
 388                 } elseif ( $start === null ) {
 389                         return "$var <= {$db->addQuotes( $end )}";
 390                 } else {
 391                         return "$var BETWEEN {$db->addQuotes( $start )} AND {$db->addQuotes( $end )}";
 392                 }
 393         }
 394 }
 395
 396 $maintClass = 'RefreshLinks';
 397 require_once RUN_MAINTENANCE_IF_MAIN;