Collect selective update statistics from LintUpdate job

This ensures that all parsoid parses are accounted for in our
statistics.  In the future we might want to query the cache for
an existing 'dirty' parse in this codepath to potentially allow
for selective update, but for now assume that selective updates
are not possible here.

Bug: T371713
Depends-On: I5b8c7ab48d5a1d6c1e311149fcac6abdc523aa13
Change-Id: I391e928175f60a1ff2e5c181e20ed72efe4dfd66

extension.json

@@ -31,6 +31,7 @@
 				"LinkRenderer",
 				"JobQueueGroup",
 				"WikiPageFactory",
+				"StatsFactory",
 				"Linter.CategoryManager",
 				"Linter.TotalsLookup",
 				"Linter.Database",

includes/Hooks.php

@@ -45,6 +45,7 @@ use MediaWiki\Storage\Hook\RevisionDataUpdatesHook;
 use MediaWiki\Title\Title;
 use MediaWiki\User\UserIdentity;
 use Skin;
+use Wikimedia\Stats\StatsFactory;
 use WikiPage;
 
 class Hooks implements
@@ -59,6 +60,7 @@ class Hooks implements
 	private LinkRenderer $linkRenderer;
 	private JobQueueGroup $jobQueueGroup;
 	private WikiPageFactory $wikiPageFactory;
+	private StatsFactory $statsFactory;
 	private CategoryManager $categoryManager;
 	private TotalsLookup $totalsLookup;
 	private Database $database;
@@ -68,6 +70,7 @@ class Hooks implements
 	 * @param LinkRenderer $linkRenderer
 	 * @param JobQueueGroup $jobQueueGroup
 	 * @param WikiPageFactory $wikiPageFactory
+	 * @param StatsFactory $statsFactory
 	 * @param CategoryManager $categoryManager
 	 * @param TotalsLookup $totalsLookup
 	 * @param Database $database
@@ -76,6 +79,7 @@ class Hooks implements
 		LinkRenderer $linkRenderer,
 		JobQueueGroup $jobQueueGroup,
 		WikiPageFactory $wikiPageFactory,
+		StatsFactory $statsFactory,
 		CategoryManager $categoryManager,
 		TotalsLookup $totalsLookup,
 		Database $database,
@@ -84,6 +88,7 @@ class Hooks implements
 		$this->linkRenderer = $linkRenderer;
 		$this->jobQueueGroup = $jobQueueGroup;
 		$this->wikiPageFactory = $wikiPageFactory;
+		$this->statsFactory = $statsFactory;
 		$this->categoryManager = $categoryManager;
 		$this->totalsLookup = $totalsLookup;
 		$this->database = $database;
@@ -359,7 +364,8 @@ class Hooks implements
 
 		$updates[] = new LintUpdate(
 			$this->wikiPageFactory,
-			$renderedRevision
+			$this->statsFactory,
+			$renderedRevision,
 		);
 	}
 }

includes/LintUpdate.php

@@ -24,22 +24,28 @@ use MediaWiki\Content\Renderer\ContentParseParams;
 use MediaWiki\Content\TextContent;
 use MediaWiki\Deferred\DataUpdate;
 use MediaWiki\Logger\LoggerFactory;
+use MediaWiki\MainConfigNames;
+use MediaWiki\MediaWikiServices;
 use MediaWiki\Page\WikiPageFactory;
 use MediaWiki\Revision\RenderedRevision;
 use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\Revision\SlotRecord;
+use Wikimedia\Stats\StatsFactory;
 
 class LintUpdate extends DataUpdate {
 
 	private WikiPageFactory $wikiPageFactory;
+	private StatsFactory $statsFactory;
 	private RenderedRevision $renderedRevision;
 
 	public function __construct(
 		WikiPageFactory $wikiPageFactory,
+		StatsFactory $statsFactory,
 		RenderedRevision $renderedRevision
 	) {
 		parent::__construct();
 		$this->wikiPageFactory = $wikiPageFactory;
+		$this->statsFactory = $statsFactory;
 		$this->renderedRevision = $renderedRevision;
 	}
 
@@ -62,6 +68,10 @@ class LintUpdate extends DataUpdate {
 
 		$pOptions = $page->makeParserOptions( 'canonical' );
 		$pOptions->setUseParsoid();
+		$pOptions->setRenderReason( 'LintUpdate' );
+
+		// XXX no previous output available on this code path
+		$previousOutput = null;
 
 		LoggerFactory::getInstance( 'Linter' )->debug(
 			'{method}: Parsing {page}',
@@ -83,9 +93,34 @@ class LintUpdate extends DataUpdate {
 			$pOptions,
 			// no need to generate HTML
 			false,
-			// XXX no previous output available
-			null
+			$previousOutput
 		);
-		$content->getContentHandler()->getParserOutput( $content, $cpoParams );
+		$output = $content->getContentHandler()->getParserOutput( $content, $cpoParams );
+
+		// T371713: Temporary statistics collection code to determine
+		// feasibility of Parsoid selective update
+		$sampleRate = MediaWikiServices::getInstance()->getMainConfig()->get(
+			MainConfigNames::ParsoidSelectiveUpdateSampleRate
+		);
+		$doSample = ( $sampleRate && mt_rand( 1, $sampleRate ) === 1 );
+		if ( $doSample ) {
+			$labels = [
+				'source' => 'LintUpdate',
+				'type' => 'full',
+				'reason' => $pOptions->getRenderReason(),
+				'parser' => 'parsoid',
+				'opportunistic' => 'false',
+			];
+			$totalStat = $this->statsFactory
+				->getCounter( 'parsercache_selective_total' );
+			$timeStat = $this->statsFactory
+				->getCounter( 'parsercache_selective_cpu_seconds' );
+			foreach ( $labels as $key => $value ) {
+				$totalStat->setLabel( $key, $value );
+				$timeStat->setLabel( $key, $value );
+			}
+			$totalStat->increment();
+			$timeStat->incrementBy( $output->getTimeProfile( 'cpu' ) );
+		}
 	}
 }

tests/phpunit/LintUpdateTest.php

@@ -32,6 +32,7 @@ use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\Revision\SlotRecord;
 use MediaWikiIntegrationTestCase;
 use RefreshLinksJob;
+use Wikimedia\Stats\StatsFactory;
 use WikiPage;
 use WikitextContent;
 
@@ -200,6 +201,7 @@ class LintUpdateTest extends MediaWikiIntegrationTestCase {
 
 		return new LintUpdate(
 			$wikiPageFactory,
+			StatsFactory::newNull(),
 			$renderedRevision
 		);
 	}