[Summary view] [Print] [Text view]

   1  <?php
   2  /**
   3   * HTML cache invalidation of all pages linking to a given title.
   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 Cache
  22   */
  23  
  24  /**
  25   * Job to purge the cache for all pages that link to or use another page or file
  26   *
  27   * This job comes in a few variants:
  28   *   - a) Recursive jobs to purge caches for backlink pages for a given title.
  29   *        These jobs have have (recursive:true,table:<table>) set.
  30   *   - b) Jobs to purge caches for a set of titles (the job title is ignored).
  31   *          These jobs have have (pages:(<page ID>:(<namespace>,<title>),...) set.
  32   *
  33   * @ingroup JobQueue
  34   */
  35  class HTMLCacheUpdateJob extends Job {
  36  	function __construct( $title, $params = '' ) {
  37          parent::__construct( 'htmlCacheUpdate', $title, $params );
  38          // Base backlink purge jobs can be de-duplicated
  39          $this->removeDuplicates = ( !isset( $params['range'] ) && !isset( $params['pages'] ) );
  40      }
  41  
  42  	function run() {
  43          global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  44  
  45          static $expected = array( 'recursive', 'pages' ); // new jobs have one of these
  46  
  47          $oldRangeJob = false;
  48          if ( !array_intersect( array_keys( $this->params ), $expected ) ) {
  49              // B/C for older job params formats that lack these fields:
  50              // a) base jobs with just ("table") and b) range jobs with ("table","start","end")
  51              if ( isset( $this->params['start'] ) && isset( $this->params['end'] ) ) {
  52                  $oldRangeJob = true;
  53              } else {
  54                  $this->params['recursive'] = true; // base job
  55              }
  56          }
  57  
  58          // Job to purge all (or a range of) backlink pages for a page
  59          if ( !empty( $this->params['recursive'] ) ) {
  60              // Convert this into no more than $wgUpdateRowsPerJob HTMLCacheUpdateJob per-title
  61              // jobs and possibly a recursive HTMLCacheUpdateJob job for the rest of the backlinks
  62              $jobs = BacklinkJobUtils::partitionBacklinkJob(
  63                  $this,
  64                  $wgUpdateRowsPerJob,
  65                  $wgUpdateRowsPerQuery, // jobs-per-title
  66                  // Carry over information for de-duplication
  67                  array( 'params' => $this->getRootJobParams() )
  68              );
  69              JobQueueGroup::singleton()->push( $jobs );
  70          // Job to purge pages for for a set of titles
  71          } elseif ( isset( $this->params['pages'] ) ) {
  72              $this->invalidateTitles( $this->params['pages'] );
  73          // B/C for job to purge a range of backlink pages for a given page
  74          } elseif ( $oldRangeJob ) {
  75              $titleArray = $this->title->getBacklinkCache()->getLinks(
  76                  $this->params['table'], $this->params['start'], $this->params['end'] );
  77  
  78              $pages = array(); // same format BacklinkJobUtils uses
  79              foreach ( $titleArray as $tl ) {
  80                  $pages[$tl->getArticleId()] = array( $tl->getNamespace(), $tl->getDbKey() );
  81              }
  82  
  83              $jobs = array();
  84              foreach ( array_chunk( $pages, $wgUpdateRowsPerJob ) as $pageChunk ) {
  85                  $jobs[] = new HTMLCacheUpdateJob( $this->title,
  86                      array(
  87                          'table' => $this->params['table'],
  88                          'pages' => $pageChunk
  89                      ) + $this->getRootJobParams() // carry over information for de-duplication
  90                  );
  91              }
  92              JobQueueGroup::singleton()->push( $jobs );
  93          }
  94  
  95          return true;
  96      }
  97  
  98      /**
  99       * @param array $pages Map of (page ID => (namespace, DB key)) entries
 100       */
 101  	protected function invalidateTitles( array $pages ) {
 102          global $wgUpdateRowsPerQuery, $wgUseFileCache, $wgUseSquid;
 103  
 104          // Get all page IDs in this query into an array
 105          $pageIds = array_keys( $pages );
 106          if ( !$pageIds ) {
 107              return;
 108          }
 109  
 110          $dbw = wfGetDB( DB_MASTER );
 111  
 112          // The page_touched field will need to be bumped for these pages.
 113          // Only bump it to the present time if no "rootJobTimestamp" was known.
 114          // If it is known, it can be used instead, which avoids invalidating output
 115          // that was in fact generated *after* the relevant dependency change time
 116          // (e.g. template edit). This is particularily useful since refreshLinks jobs
 117          // save back parser output and usually run along side htmlCacheUpdate jobs;
 118          // their saved output would be invalidated by using the current timestamp.
 119          if ( isset( $this->params['rootJobTimestamp'] ) ) {
 120              $touchTimestamp = $this->params['rootJobTimestamp'];
 121          } else {
 122              $touchTimestamp = wfTimestampNow();
 123          }
 124  
 125          // Update page_touched (skipping pages already touched since the root job).
 126          // Check $wgUpdateRowsPerQuery for sanity; batch jobs are sized by that already.
 127          foreach ( array_chunk( $pageIds, $wgUpdateRowsPerQuery ) as $batch ) {
 128              $dbw->update( 'page',
 129                  array( 'page_touched' => $dbw->timestamp( $touchTimestamp ) ),
 130                  array( 'page_id' => $batch,
 131                      // don't invalidated pages that were already invalidated
 132                      "page_touched < " . $dbw->addQuotes( $dbw->timestamp( $touchTimestamp ) )
 133                  ),
 134                  __METHOD__
 135              );
 136          }
 137          // Get the list of affected pages (races only mean something else did the purge)
 138          $titleArray = TitleArray::newFromResult( $dbw->select(
 139              'page',
 140              array( 'page_namespace', 'page_title' ),
 141              array( 'page_id' => $pageIds, 'page_touched' => $dbw->timestamp( $touchTimestamp ) ),
 142              __METHOD__
 143          ) );
 144  
 145          // Update squid
 146          if ( $wgUseSquid ) {
 147              $u = SquidUpdate::newFromTitles( $titleArray );
 148              $u->doUpdate();
 149          }
 150  
 151          // Update file cache
 152          if ( $wgUseFileCache ) {
 153              foreach ( $titleArray as $title ) {
 154                  HTMLFileCache::clearFileCache( $title );
 155              }
 156          }
 157      }
 158  
 159  	public function workItemCount() {
 160          return isset( $this->params['pages'] ) ? count( $this->params['pages'] ) : 1;
 161      }
 162  }