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