includes/HTMLCacheUpdate.php

   1 <?php
   2
   3 /**
   4  * Class to invalidate the HTML cache of all the pages linking to a given title.
   5  * Small numbers of links will be done immediately, large numbers are pushed onto
   6  * the job queue.
   7  *
   8  * This class is designed to work efficiently with small numbers of links, and
   9  * to work reasonably well with up to ~10^5 links. Above ~10^6 links, the memory
  10  * and time requirements of loading all backlinked IDs in doUpdate() might become
  11  * prohibitive. The requirements measured at Wikimedia are approximately:
  12  *
  13  *   memory: 48 bytes per row
  14  *   time: 16us per row for the query plus processing
  15  *
  16  * The reason this query is done is to support partitioning of the job
  17  * by backlinked ID. The memory issue could be allieviated by doing this query in
  18  * batches, but of course LIMIT with an offset is inefficient on the DB side.
  19  *
  20  * The class is nevertheless a vast improvement on the previous method of using
  21  * Image::getLinksTo() and Title::touchArray(), which uses about 2KB of memory per
  22  * link.
  23  */
  24 class HTMLCacheUpdate
  25 {
  26         public $mTitle, $mTable, $mPrefix;
  27         public $mRowsPerJob, $mRowsPerQuery;
  28
  29         function __construct( $titleTo, $table ) {
  30                 global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  31
  32                 $this->mTitle = $titleTo;
  33                 $this->mTable = $table;
  34                 $this->mRowsPerJob = $wgUpdateRowsPerJob;
  35                 $this->mRowsPerQuery = $wgUpdateRowsPerQuery;
  36         }
  37
  38         function doUpdate() {
  39                 # Fetch the IDs
  40                 $cond = $this->getToCondition();
  41                 $dbr = wfGetDB( DB_SLAVE );
  42                 $res = $dbr->select( $this->mTable, $this->getFromField(), $cond, __METHOD__ );
  43                 $resWrap = new ResultWrapper( $dbr, $res );
  44                 if ( $dbr->numRows( $res ) != 0 ) {
  45                         if ( $dbr->numRows( $res ) > $this->mRowsPerJob ) {
  46                                 $this->insertJobs( $resWrap );
  47                         } else {
  48                                 $this->invalidateIDs( $resWrap );
  49                         }
  50                 }
  51                 $dbr->freeResult( $res );
  52         }
  53
  54         function insertJobs( ResultWrapper $res ) {
  55                 $numRows = $res->numRows();
  56                 $numBatches = ceil( $numRows / $this->mRowsPerJob );
  57                 $realBatchSize = $numRows / $numBatches;
  58                 $start = false;
  59                 $jobs = array();
  60                 do {
  61                         for ( $i = 0; $i < $realBatchSize - 1; $i++ ) {
  62                                 $row = $res->fetchRow();
  63                                 if ( $row ) {
  64                                         $id = $row[0];
  65                                 } else {
  66                                         $id = false;
  67                                         break;
  68                                 }
  69                         }
  70
  71                         $params = array(
  72                                 'table' => $this->mTable,
  73                                 'start' => $start,
  74                                 'end' => ( $id !== false ? $id - 1 : false ),
  75                         );
  76                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
  77
  78                         $start = $id;
  79                 } while ( $start );
  80
  81                 Job::batchInsert( $jobs );
  82         }
  83
  84         function getPrefix() {
  85                 static $prefixes = array(
  86                         'pagelinks' => 'pl',
  87                         'imagelinks' => 'il',
  88                         'categorylinks' => 'cl',
  89                         'templatelinks' => 'tl',
  90
  91                         # Not needed
  92                         # 'externallinks' => 'el',
  93                         # 'langlinks' => 'll'
  94                 );
  95
  96                 if ( is_null( $this->mPrefix ) ) {
  97                         $this->mPrefix = $prefixes[$this->mTable];
  98                         if ( is_null( $this->mPrefix ) ) {
  99                                 throw new MWException( "Invalid table type \"{$this->mTable}\" in " . __CLASS__ );
 100                         }
 101                 }
 102                 return $this->mPrefix;
 103         }
 104
 105         function getFromField() {
 106                 return $this->getPrefix() . '_from';
 107         }
 108
 109         function getToCondition() {
 110                 switch ( $this->mTable ) {
 111                         case 'pagelinks':
 112                                 return array(
 113                                         'pl_namespace' => $this->mTitle->getNamespace(),
 114                                         'pl_title' => $this->mTitle->getDBkey()
 115                                 );
 116                         case 'templatelinks':
 117                                 return array(
 118                                         'tl_namespace' => $this->mTitle->getNamespace(),
 119                                         'tl_title' => $this->mTitle->getDBkey()
 120                                 );
 121                         case 'imagelinks':
 122                                 return array( 'il_to' => $this->mTitle->getDBkey() );
 123                         case 'categorylinks':
 124                                 return array( 'cl_to' => $this->mTitle->getDBkey() );
 125                 }
 126                 throw new MWException( 'Invalid table type in ' . __CLASS__ );
 127         }
 128
 129         /**
 130          * Invalidate a set of IDs, right now
 131          */
 132         function invalidateIDs( ResultWrapper $res ) {
 133                 global $wgUseFileCache, $wgUseSquid;
 134
 135                 if ( $res->numRows() == 0 ) {
 136                         return;
 137                 }
 138
 139                 $dbw = wfGetDB( DB_MASTER );
 140                 $timestamp = $dbw->timestamp();
 141                 $done = false;
 142
 143                 while ( !$done ) {
 144                         # Get all IDs in this query into an array
 145                         $ids = array();
 146                         for ( $i = 0; $i < $this->mRowsPerQuery; $i++ ) {
 147                                 $row = $res->fetchRow();
 148                                 if ( $row ) {
 149                                         $ids[] = $row[0];
 150                                 } else {
 151                                         $done = true;
 152                                         break;
 153                                 }
 154                         }
 155
 156                         if ( !count( $ids ) ) {
 157                                 break;
 158                         }
 159
 160                         # Update page_touched
 161                         $dbw->update( 'page',
 162                                 array( 'page_touched' => $timestamp ),
 163                                 array( 'page_id IN (' . $dbw->makeList( $ids ) . ')' ),
 164                                 __METHOD__
 165                         );
 166
 167                         # Update squid
 168                         if ( $wgUseSquid || $wgUseFileCache ) {
 169                                 $titles = Title::newFromIDs( $ids );
 170                                 if ( $wgUseSquid ) {
 171                                         $u = SquidUpdate::newFromTitles( $titles );
 172                                         $u->doUpdate();
 173                                 }
 174
 175                                 # Update file cache
 176                                 if  ( $wgUseFileCache ) {
 177                                         foreach ( $titles as $title ) {
 178                                                 $cm = new HTMLFileCache($title);
 179                                                 @unlink($cm->fileCacheName());
 180                                         }
 181                                 }
 182                         }
 183                 }
 184         }
 185 }
 186
 187 /**
 188  * @todo document (e.g. one-sentence top-level class description).
 189  */
 190 class HTMLCacheUpdateJob extends Job {
 191         var $table, $start, $end;
 192
 193         /**
 194          * Construct a job
 195          * @param Title $title The title linked to
 196          * @param array $params Job parameters (table, start and end page_ids)
 197          * @param integer $id job_id
 198          */
 199         function __construct( $title, $params, $id = 0 ) {
 200                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 201                 $this->table = $params['table'];
 202                 $this->start = $params['start'];
 203                 $this->end = $params['end'];
 204         }
 205
 206         function run() {
 207                 $update = new HTMLCacheUpdate( $this->title, $this->table );
 208
 209                 $fromField = $update->getFromField();
 210                 $conds = $update->getToCondition();
 211                 if ( $this->start ) {
 212                         $conds[] = "$fromField >= {$this->start}";
 213                 }
 214                 if ( $this->end ) {
 215                         $conds[] = "$fromField <= {$this->end}";
 216                 }
 217
 218                 $dbr = wfGetDB( DB_SLAVE );
 219                 $res = $dbr->select( $this->table, $fromField, $conds, __METHOD__ );
 220                 $update->invalidateIDs( new ResultWrapper( $dbr, $res ) );
 221                 $dbr->freeResult( $res );
 222
 223                 return true;
 224         }
 225 }
 226