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