PageTriage should not compile article not in queue

1. Fix it so it only compiles data in the queue
2. a maintenance script to clean the junk data

Change-Id: I0bf48b1cf75c96720e24e27b07751282dec58dd2

PageTriage.hooks.php

@@ -131,7 +131,11 @@ public static function onArticleSaveComplete( $article, $user, $text, $summary,
 			return true;
 		}
 
-		$acp = ArticleCompileProcessor::newFromPageId( array( $article->getId() ) );
+		// if there is a previous revision, it's safe to check against slave database
+		$validateDb = $revision->getPrevious() ? DB_SLAVE : DB_MASTER;
+
+		// false will enforce a validation against pagetriage_page table
+		$acp = ArticleCompileProcessor::newFromPageId( array( $article->getId() ), false, $validateDb );
 		if ( $acp ) {
 			// Register the article object so we can get the content and other useful information
 			// this is primarily for replication delay from slave

includes/ArticleMetadata.php

@@ -9,7 +9,7 @@ class ArticleMetadata {
 
 	/**
 	 * @param $pageId array - list of page id
-	 * @param $validated bool - whether the page ids are validated
+	 * @param $validated bool - whether the page ids have been validated
 	 * @param $validateDb const - DB_MASTER/DB_SLAVE
 	 */
 	public function __construct( array $pageId, $validated = true, $validateDb = DB_MASTER ) {
@@ -186,7 +186,7 @@ public function getMetadata() {
 			// Compile the data if it is not available, this is a very rare case unless
 			// the metadata gets deleted manually
 			if ( $articles ) {
-				$acp = ArticleCompileProcessor::newFromPageId( $articles );
+				$acp = ArticleCompileProcessor::newFromPageId( $articles, false, DB_SLAVE );
 				if ( $acp ) {
 					$pageData += $acp->compileMetadata();
 				}

tools/cleanupPageTriagePageTags.php

@@ -0,0 +1,62 @@
+<?php
+/**
+ * Remove page with namespace other than NS_MAIN/NS_USER from pagetriage queue
+ *
+ * @ingroup Maintenance
+ */
+
+require_once( dirname( __FILE__ ) . '/../../../maintenance/Maintenance.php' );
+
+/**
+ * Maintenance script that removes data from pagetriage_page_tags with page_id not in pagetriage_page
+ *
+ * @ingroup Maintenance
+ */
+class CleanupPageTriagePageTags extends Maintenance {
+
+	protected $batchSize = 100;
+
+	public function execute() {
+		$dbw = wfGetDB( DB_MASTER );
+		$dbr = wfGetDB( DB_SLAVE );
+
+		$count = $this->batchSize;
+		$start = 0;
+
+		while ( $count == $this->batchSize ) {
+			$res = $dbr->select(
+				array( 'pagetriage_page_tags', 'pagetriage_page' ),
+				array( 'DISTINCT ptrpt_page_id AS page_id' ),
+				array(
+					'ptrpt_page_id > ' . $start,
+					'ptrp_page_id IS NULL'
+				),
+				__METHOD__,
+				array( 'LIMIT' => $this->batchSize, 'ORDER BY' => 'ptrpt_page_id' ),
+				array( 'pagetriage_page' => array( 'LEFT JOIN', 'ptrp_page_id = ptrpt_page_id' ) )
+			);
+
+			$page = array();
+			foreach( $res as $row ) {
+				$page[] = $row->page_id;
+				$start  = $row->page_id;
+			};
+			$count = count( $page );
+
+			if ( $count > 0 ) {
+				$dbw->delete(
+					'pagetriage_page_tags',
+					array( 'ptrpt_page_id' => $page ),
+					__METHOD__
+				);
+
+				$this->output( "processing " . $count . "\n" );
+				wfWaitForSlaves();
+			}
+
+		}
+	}
+}
+
+$maintClass = 'CleanupPageTriagePageTags'; // Tells it to run the class
+require_once( RUN_MAINTENANCE_IF_MAIN );