* Small numbers of links will be done immediately, large numbers are pushed onto
* the job queue.
*
- * This class is designed to work efficiently with small numbers of links, and
+ * This class is designed to work efficiently with small numbers of links, and
* to work reasonably well with up to ~10^5 links. Above ~10^6 links, the memory
* and time requirements of loading all backlinked IDs in doUpdate() might become
* prohibitive. The requirements measured at Wikimedia are approximately:
- *
+ *
* memory: 48 bytes per row
* time: 16us per row for the query plus processing
*
* The reason this query is done is to support partitioning of the job
- * by backlinked ID. The memory issue could be allieviated by doing this query in
+ * by backlinked ID. The memory issue could be allieviated by doing this query in
* batches, but of course LIMIT with an offset is inefficient on the DB side.
*
- * The class is nevertheless a vast improvement on the previous method of using
+ * The class is nevertheless a vast improvement on the previous method of using
* Image::getLinksTo() and Title::touchArray(), which uses about 2KB of memory per
* link.
+ *
+ * @ingroup Cache
*/
class HTMLCacheUpdate
{
$this->mTable = $table;
$this->mRowsPerJob = $wgUpdateRowsPerJob;
$this->mRowsPerQuery = $wgUpdateRowsPerQuery;
+ $this->mCache = $this->mTitle->getBacklinkCache();
}
- function doUpdate() {
+ public function doUpdate() {
# Fetch the IDs
- $cond = $this->getToCondition();
- $dbr = wfGetDB( DB_SLAVE );
- $res = $dbr->select( $this->mTable, $this->getFromField(), $cond, __METHOD__ );
- $resWrap = new ResultWrapper( $dbr, $res );
- if ( $dbr->numRows( $res ) != 0 ) {
- if ( $dbr->numRows( $res ) > $this->mRowsPerJob ) {
- $this->insertJobs( $resWrap );
+ $numRows = $this->mCache->getNumLinks( $this->mTable );
+
+ if ( $numRows != 0 ) {
+ if ( $numRows > $this->mRowsPerJob ) {
+ $this->insertJobs();
} else {
- $this->invalidateIDs( $resWrap );
+ $this->invalidate();
}
}
- $dbr->freeResult( $res );
+ wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
}
- function insertJobs( ResultWrapper $res ) {
- $numRows = $res->numRows();
- $numBatches = ceil( $numRows / $this->mRowsPerJob );
- $realBatchSize = $numRows / $numBatches;
- $start = false;
- $jobs = array();
- do {
- for ( $i = 0; $i < $realBatchSize - 1; $i++ ) {
- $row = $res->fetchRow();
- if ( $row ) {
- $id = $row[0];
- } else {
- $id = false;
- break;
- }
- }
-
+ protected function insertJobs() {
+ $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
+ if ( !$batches ) {
+ return;
+ }
+ foreach ( $batches as $batch ) {
$params = array(
'table' => $this->mTable,
- 'start' => $start,
- 'end' => ( $id !== false ? $id - 1 : false ),
+ 'start' => $batch[0],
+ 'end' => $batch[1],
);
$jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
-
- $start = $id;
- } while ( $start );
-
+ }
Job::batchInsert( $jobs );
}
- function getPrefix() {
- static $prefixes = array(
- 'pagelinks' => 'pl',
- 'imagelinks' => 'il',
- 'categorylinks' => 'cl',
- 'templatelinks' => 'tl',
-
- # Not needed
- # 'externallinks' => 'el',
- # 'langlinks' => 'll'
- );
-
- if ( is_null( $this->mPrefix ) ) {
- $this->mPrefix = $prefixes[$this->mTable];
- if ( is_null( $this->mPrefix ) ) {
- throw new MWException( "Invalid table type \"{$this->mTable}\" in " . __CLASS__ );
- }
- }
- return $this->mPrefix;
- }
-
- function getFromField() {
- return $this->getPrefix() . '_from';
- }
-
- function getToCondition() {
- switch ( $this->mTable ) {
- case 'pagelinks':
- return array(
- 'pl_namespace' => $this->mTitle->getNamespace(),
- 'pl_title' => $this->mTitle->getDBkey()
- );
- case 'templatelinks':
- return array(
- 'tl_namespace' => $this->mTitle->getNamespace(),
- 'tl_title' => $this->mTitle->getDBkey()
- );
- case 'imagelinks':
- return array( 'il_to' => $this->mTitle->getDBkey() );
- case 'categorylinks':
- return array( 'cl_to' => $this->mTitle->getDBkey() );
- }
- throw new MWException( 'Invalid table type in ' . __CLASS__ );
- }
/**
- * Invalidate a set of IDs, right now
+ * Invalidate a set of pages, right now
*/
- function invalidateIDs( ResultWrapper $res ) {
+ public function invalidate( $startId = false, $endId = false ) {
global $wgUseFileCache, $wgUseSquid;
- if ( $res->numRows() == 0 ) {
+ $titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
+ if ( $titleArray->count() == 0 ) {
return;
}
$dbw = wfGetDB( DB_MASTER );
$timestamp = $dbw->timestamp();
- $done = false;
-
- while ( !$done ) {
- # Get all IDs in this query into an array
- $ids = array();
- for ( $i = 0; $i < $this->mRowsPerQuery; $i++ ) {
- $row = $res->fetchRow();
- if ( $row ) {
- $ids[] = $row[0];
- } else {
- $done = true;
- break;
- }
- }
- if ( !count( $ids ) ) {
- break;
- }
-
- # Update page_touched
- $dbw->update( 'page',
- array( 'page_touched' => $timestamp ),
- array( 'page_id IN (' . $dbw->makeList( $ids ) . ')' ),
- __METHOD__
- );
+ # Get all IDs in this query into an array
+ $ids = array();
+ foreach ( $titleArray as $title ) {
+ $ids[] = $title->getArticleID();
+ }
+ # Update page_touched
+ $dbw->update( 'page',
+ array( 'page_touched' => $timestamp ),
+ array( 'page_id IN (' . $dbw->makeList( $ids ) . ')' ),
+ __METHOD__
+ );
+
+ # Update squid
+ if ( $wgUseSquid ) {
+ $u = SquidUpdate::newFromTitles( $titleArray );
+ $u->doUpdate();
+ }
- # Update squid
- if ( $wgUseSquid || $wgUseFileCache ) {
- $titles = Title::newFromIDs( $ids );
- if ( $wgUseSquid ) {
- $u = SquidUpdate::newFromTitles( $titles );
- $u->doUpdate();
- }
-
- # Update file cache
- if ( $wgUseFileCache ) {
- foreach ( $titles as $title ) {
- $cm = new HTMLFileCache($title);
- @unlink($cm->fileCacheName());
- }
- }
+ # Update file cache
+ if ( $wgUseFileCache ) {
+ foreach ( $titleArray as $title ) {
+ HTMLFileCache::clearFileCache( $title );
}
}
}
}
/**
- * @todo document (e.g. one-sentence top-level class description).
+ * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
+ * job gets called from the queue.
+ *
+ * @ingroup JobQueue
*/
class HTMLCacheUpdateJob extends Job {
var $table, $start, $end;
$this->end = $params['end'];
}
- function run() {
+ public function run() {
$update = new HTMLCacheUpdate( $this->title, $this->table );
-
- $fromField = $update->getFromField();
- $conds = $update->getToCondition();
- if ( $this->start ) {
- $conds[] = "$fromField >= {$this->start}";
- }
- if ( $this->end ) {
- $conds[] = "$fromField <= {$this->end}";
- }
-
- $dbr = wfGetDB( DB_SLAVE );
- $res = $dbr->select( $this->table, $fromField, $conds, __METHOD__ );
- $update->invalidateIDs( new ResultWrapper( $dbr, $res ) );
- $dbr->freeResult( $res );
-
+ $update->invalidate( $this->start, $this->end );
return true;
}
}
-
scripts.mit.edu / autoinstalls / mediawiki.git / blobdiff