1 files changed, 118 insertions, 25 deletions

diff --git a/includes/HTMLCacheUpdate.php b/includes/HTMLCacheUpdate.php
index bd63c072..6f90b2d9 100644
--- a/includes/HTMLCacheUpdate.php
+++ b/includes/HTMLCacheUpdate.php
@@ -25,38 +25,119 @@
  */
 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();
 	}
 
 	public function doUpdate() {
-		# Fetch the IDs
-		$numRows = $this->mCache->getNumLinks( $this->mTable );
+		if ( $this->mStart || $this->mEnd ) {
+			$this->doPartialUpdate();
+			return;
+		}
 
-		if ( $numRows != 0 ) {
-			if ( $numRows > $this->mRowsPerJob ) {
-				$this->insertJobs();
+		# 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->invalidate();
+				$this->invalidateTitles( $titleArray );
 			}
 		}
 		wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
 	}
 
+	/**
+	 * 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();
+		$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;
+			}
+			$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 );
+	}
+
 	protected function insertJobs() {
 		$batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
 		if ( !$batches ) {
 			return;
 		}
+		$jobs = array();
 		foreach ( $batches as $batch ) {
 			$params = array(
 				'table' => $this->mTable,
@@ -68,17 +149,20 @@ class HTMLCacheUpdate
 		Job::batchInsert( $jobs );
 	}
 
-
 	/**
-	 * Invalidate a set of pages, right now
+	 * Invalidate a range of pages, right now
+	 * @deprecated
 	 */
 	public function invalidate( $startId = false, $endId = false ) {
-		global $wgUseFileCache, $wgUseSquid;
-
 		$titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
-		if ( $titleArray->count() == 0 ) {
-			return;
-		}
+		$this->invalidateTitles( $titleArray );
+	}
+
+	/**
+	 * Invalidate an array (or iterator) of Title objects, right now
+	 */
+	protected function invalidateTitles( $titleArray ) {
+		global $wgUseFileCache, $wgUseSquid;
 
 		$dbw = wfGetDB( DB_MASTER );
 		$timestamp = $dbw->timestamp();
@@ -88,12 +172,20 @@ class HTMLCacheUpdate
 		foreach ( $titleArray as $title ) {
 			$ids[] = $title->getArticleID();
 		}
+
+		if ( !$ids ) {
+			return;
+		}
+
 		# Update page_touched
-		$dbw->update( 'page',
-			array( 'page_touched' => $timestamp ),
-			array( 'page_id IN (' . $dbw->makeList( $ids ) . ')' ),
-			__METHOD__
-		);
+		$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 ) {
@@ -108,6 +200,7 @@ class HTMLCacheUpdate
 			}
 		}
 	}
+
 }
 
 /**
@@ -121,9 +214,9 @@ class HTMLCacheUpdateJob extends Job {
 
 	/**
 	 * Construct a job
-	 * @param Title $title The title linked to
-	 * @param array $params Job parameters (table, start and end page_ids)
-	 * @param integer $id job_id
+	 * @param $title Title: the title linked to
+	 * @param $params Array: job parameters (table, start and end page_ids)
+	 * @param $id Integer: job id
 	 */
 	function __construct( $title, $params, $id = 0 ) {
 		parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
@@ -133,8 +226,8 @@ class HTMLCacheUpdateJob extends Job {
 	}
 
 	public function run() {
-		$update = new HTMLCacheUpdate( $this->title, $this->table );
-		$update->invalidate( $this->start, $this->end );
+		$update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
+		$update->doUpdate();
 		return true;
 	}
 }