X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FHTMLCacheUpdate.php;h=7c4731b0ad53805ea678042545bd8d4365b63e63;hb=42fe290d16f66ba3d719bc6149cc79570e55a6b3;hp=9a0b6a0846916beec51a7233be95a70186813e73;hpb=5fef2333d43faed0571878e848d6c40ade2c9810;p=lhc%2Fweb%2Fwiklou.git

diff --git a/includes/HTMLCacheUpdate.php b/includes/HTMLCacheUpdate.php
index 9a0b6a0846..7c4731b0ad 100644
--- a/includes/HTMLCacheUpdate.php
+++ b/includes/HTMLCacheUpdate.php
@@ -5,187 +5,209 @@
  * 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
 {
-	public $mTitle, $mTable, $mPrefix;
+	public $mTitle, $mTable, $mPrefix, $mStart, $mEnd;
 	public $mRowsPerJob, $mRowsPerQuery;
 
-	function __construct( $titleTo, $table ) {
+	function __construct( $titleTo, $table, $start = false, $end = false ) {
 		global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
 
 		$this->mTitle = $titleTo;
 		$this->mTable = $table;
+		$this->mStart = $start;
+		$this->mEnd = $end;
 		$this->mRowsPerJob = $wgUpdateRowsPerJob;
 		$this->mRowsPerQuery = $wgUpdateRowsPerQuery;
+		$this->mCache = $this->mTitle->getBacklinkCache();
 	}
 
-	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 );
+	public function doUpdate() {
+		if ( $this->mStart || $this->mEnd ) {
+			$this->doPartialUpdate();
+			return;
+		}
+
+		# Get an estimate of the number of rows from the BacklinkCache
+		$numRows = $this->mCache->getNumLinks( $this->mTable );
+		if ( $numRows > $this->mRowsPerJob * 2 ) {
+			# Do fast cached partition
+			$this->insertJobs();
+		} else {
+			# Get the links from the DB
+			$titleArray = $this->mCache->getLinks( $this->mTable );
+			# Check if the row count estimate was correct
+			if ( $titleArray->count() > $this->mRowsPerJob * 2 ) {
+				# Not correct, do accurate partition
+				wfDebug( __METHOD__.": row count estimate was incorrect, repartitioning\n" );
+				$this->insertJobsFromTitles( $titleArray );
 			} else {
-				$this->invalidateIDs( $resWrap );
+				$this->invalidateTitles( $titleArray );
 			}
 		}
-		$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;
+	/**
+	 * Update some of the backlinks, defined by a page ID range
+	 */
+	protected function doPartialUpdate() {
+		$titleArray = $this->mCache->getLinks( $this->mTable, $this->mStart, $this->mEnd );
+		if ( $titleArray->count() <= $this->mRowsPerJob * 2 ) {
+			# This partition is small enough, do the update
+			$this->invalidateTitles( $titleArray );
+		} else {
+			# Partitioning was excessively inaccurate. Divide the job further.
+			# This can occur when a large number of links are added in a short 
+			# period of time, say by updating a heavily-used template.
+			$this->insertJobsFromTitles( $titleArray );
+		}
+	}
+
+	/**
+	 * Partition the current range given by $this->mStart and $this->mEnd,
+	 * using a pre-calculated title array which gives the links in that range.
+	 * Queue the resulting jobs.
+	 */
+	protected function insertJobsFromTitles( $titleArray ) {
+		# We make subpartitions in the sense that the start of the first job
+		# will be the start of the parent partition, and the end of the last
+		# job will be the end of the parent partition.
 		$jobs = array();
-		do {
-			for ( $i = 0; $i < $realBatchSize - 1; $i++ ) {
-				$row = $res->fetchRow();
-				if ( $row ) {
-					$id = $row[0];
-				} else {
-					$id = false;
-					break;
-				}
-			}
-			if ( $id !== false ) {
-				// One less on the end to avoid duplicating the boundary
-				$job = new HTMLCacheUpdateJob( $this->mTitle, $this->mTable, $start, $id - 1 );
-			} else {
-				$job = new HTMLCacheUpdateJob( $this->mTitle, $this->mTable, $start, false );
+		$start = $this->mStart; # start of the current job
+		$numTitles = 0;
+		foreach ( $titleArray as $title ) {
+			$id = $title->getArticleID();
+			# $numTitles is now the number of titles in the current job not 
+			# including the current ID
+			if ( $numTitles >= $this->mRowsPerJob ) {
+				# Add a job up to but not including the current ID
+				$params = array(
+					'table' => $this->mTable,
+					'start' => $start,
+					'end' => $id - 1
+				);
+				$jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
+				$start = $id;
+				$numTitles = 0;
 			}
-			$jobs[] = $job;
-
-			$start = $id;
-		} while ( $start );
+			$numTitles++;
+		}
+		# Last job
+		$params = array(
+			'table' => $this->mTable,
+			'start' => $start,
+			'end' => $this->mEnd
+		);
+		$jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
+		wfDebug( __METHOD__.": repartitioning into " . count( $jobs ) . " jobs\n" );
+
+		if ( count( $jobs ) < 2 ) {
+			# I don't think this is possible at present, but handling this case
+			# makes the code a bit more robust against future code updates and 
+			# avoids a potential infinite loop of repartitioning
+			wfDebug( __METHOD__.": repartitioning failed!\n" );
+			$this->invalidateTitles( $titleArray );
+			return;
+		}
 
 		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__ );
-			}
+	protected function insertJobs() {
+		$batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
+		if ( !$batches ) {
+			return;
 		}
-		return $this->mPrefix;
-	}
-	
-	function getFromField() {
-		return $this->getPrefix() . '_from';
+		$jobs = array();
+		foreach ( $batches as $batch ) {
+			$params = array(
+				'table' => $this->mTable,
+				'start' => $batch[0],
+				'end' => $batch[1],
+			);
+			$jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
+		}
+		Job::batchInsert( $jobs );
 	}
 
-	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 range of pages, right now
+	 * @deprecated
+	 */
+	public function invalidate( $startId = false, $endId = false ) {
+		$titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
+		$this->invalidateTitles( $titleArray );
 	}
 
 	/**
-	 * Invalidate a set of IDs, right now
+	 * Invalidate an array (or iterator) of Title objects, right now
 	 */
-	function invalidateIDs( ResultWrapper $res ) {
+	protected function invalidateTitles( $titleArray ) {
 		global $wgUseFileCache, $wgUseSquid;
 
-		if ( $res->numRows() == 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 ) . ')' ),
+		# Get all IDs in this query into an array
+		$ids = array();
+		foreach ( $titleArray as $title ) {
+			$ids[] = $title->getArticleID();
+		}
+
+		if ( !$ids ) {
+			return;
+		}
+
+		# Update page_touched
+		$batches = array_chunk( $ids, $this->mRowsPerQuery );
+		foreach ( $batches as $batch ) {
+			$dbw->update( 'page',
+				array( 'page_touched' => $timestamp ),
+				array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 				__METHOD__
 			);
+		}
 
-			# 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 squid
+		if ( $wgUseSquid ) {
+			$u = SquidUpdate::newFromTitles( $titleArray );
+			$u->doUpdate();
+		}
+
+		# 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;
@@ -193,40 +215,19 @@ class HTMLCacheUpdateJob extends Job {
 	/**
 	 * Construct a job
 	 * @param Title $title The title linked to
-	 * @param string $table The name of the link table.
-	 * @param integer $start Beginning page_id or false for open interval
-	 * @param integer $end End page_id or false for open interval
+	 * @param array $params Job parameters (table, start and end page_ids)
 	 * @param integer $id job_id
 	 */
-	function __construct( $title, $table, $start, $end, $id = 0 ) {
-		$params = array(
-			'table' => $table, 
-			'start' => $start, 
-			'end' => $end );
+	function __construct( $title, $params, $id = 0 ) {
 		parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
-		$this->table = $table;
-		$this->start = intval( $start );
-		$this->end = intval( $end );
+		$this->table = $params['table'];
+		$this->start = $params['start'];
+		$this->end = $params['end'];
 	}
 
-	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 );
-
+	public function run() {
+		$update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
+		$update->doUpdate();
 		return true;
 	}
 }
-?>