Make the doc building for search aware of the revision

Added an optional RevisionRecord param to:
- ContentHandler::getParserOutputForIndexing
- ContentHandler::getDataForSearchIndex
- the SearchDataForIndex hook

So that they have a chance to build the content related to a specific
revision.

Ultimately we'd like to make this parameter mandatory.

Bug: T317309
Depends-On: I8b220cd6c4aeeca1d924bdd527409b8602318944
Depends-On: I8616b611caab3f5fa97ff0e655b19c3034304597
Change-Id: I3298ce7591069eb32f624b2c9fbb6de58ae04a29

RELEASE-NOTES-1.40

@@ -66,6 +66,10 @@ For notes on 1.39.x and older releases, see HISTORY.
   passed to var_dump(), to make its use for debugging more feasible.
 * Added 'GetBlockErrorMessageKey' hook, allow extensions'
   block error messages to be received and displayed by BlockErrorFormatter.
+* Added an optional RevisionRecord param to:
+  - ContentHandler::getParserOutputForIndexing
+  - ContentHandler::getDataForSearchIndex
+  - the SearchDataForIndex hook is deprecated in favor of SearchDataForIndex2
 * …
 
 === External library changes in 1.40 ===

includes/HookContainer/DeprecatedHooks.php

@@ -61,6 +61,7 @@ class DeprecatedHooks {
 		'SkinTemplateNavigation' => [ 'deprecatedVersion' => '1.39' ],
 		'SkinTemplateNavigation::SpecialPage' => [ 'deprecatedVersion' => '1.39' ],
 		'PersonalUrls' => [ 'deprecatedVersion' => '1.39' ],
+		'SearchDataForIndex' => [ 'deprecatedVersion' => '1.40', 'silent' => true ],
 	];
 
 	/**

includes/HookContainer/HookRunner.php

@@ -77,6 +77,7 @@ class HookRunner implements
 	\MediaWiki\Content\Hook\PageContentLanguageHook,
 	\MediaWiki\Content\Hook\PlaceNewSectionHook,
 	\MediaWiki\Content\Hook\SearchDataForIndexHook,
+	\MediaWiki\Content\Hook\SearchDataForIndex2Hook,
 	\MediaWiki\Specials\Contribute\Hook\ContributeCardsHook,
 	\MediaWiki\Diff\Hook\AbortDiffCacheHook,
 	\MediaWiki\Diff\Hook\ArticleContentOnDiffHook,
@@ -3294,15 +3295,22 @@ class HookRunner implements
 		);
 	}
 
-	public function onSearchDataForIndex( &$fields, $handler, $page, $output,
-		$engine
-	) {
+	public function onSearchDataForIndex( &$fields, $handler, $page, $output, $engine ) {
 		return $this->container->run(
 			'SearchDataForIndex',
 			[ &$fields, $handler, $page, $output, $engine ]
 		);
 	}
 
+	public function onSearchDataForIndex2( array &$fields, \ContentHandler $handler,
+		\WikiPage $page, \ParserOutput $output, \SearchEngine $engine, RevisionRecord $revision
+	) {
+		return $this->container->run(
+			'SearchDataForIndex2',
+			[ &$fields, $handler, $page, $output, $engine, $revision ]
+		);
+	}
+
 	public function onSearchGetNearMatch( $term, &$title ) {
 		return $this->container->run(
 			'SearchGetNearMatch',

includes/content/ContentHandler.php

@@ -39,6 +39,7 @@ use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\Revision\SlotRecord;
 use MediaWiki\Revision\SlotRenderingProvider;
 use MediaWiki\Search\ParserOutputSearchDataExtractor;
+use Wikimedia\Assert\Assert;
 use Wikimedia\ScopedCallback;
 
 /**
@@ -1380,16 +1381,32 @@ abstract class ContentHandler {
 	 * @param WikiPage $page Page to index
 	 * @param ParserOutput $output
 	 * @param SearchEngine $engine Search engine for which we are indexing
-	 * @return array Map of name=>value for fields
+	 * @param RevisionRecord|null $revision Revision content to fetch if provided or use the latest revision
+	 *                                      from WikiPage::getRevisionRecord() if not
+	 * @return array Map of name=>value for fields, an empty array is returned if the latest
+	 *               revision cannot be retrieved.
 	 * @since 1.28
 	 */
 	public function getDataForSearchIndex(
 		WikiPage $page,
 		ParserOutput $output,
-		SearchEngine $engine
+		SearchEngine $engine,
+		RevisionRecord $revision = null
 	) {
 		$fieldData = [];
-		$content = $page->getContent();
+		$revision = $revision ?? $page->getRevisionRecord();
+		if ( $revision === null ) {
+			LoggerFactory::getInstance( 'search' )->warning(
+				"Called getDataForSearchIndex on the page {page_id} for which the " .
+				"latest revision cannot be loaded.",
+				[ "page_id" => $page->getId() ]
+			);
+			return [];
+		}
+		Assert::invariant( $revision->getPageId() === $page->getId(),
+			'$revision and $page must target the same page_id' );
+
+		$content = $revision->getContent( SlotRecord::MAIN );
 
 		if ( $content ) {
 			$searchDataExtractor = new ParserOutputSearchDataExtractor();
@@ -1408,6 +1425,8 @@ abstract class ContentHandler {
 		}
 
 		$this->getHookRunner()->onSearchDataForIndex( $fieldData, $this, $page, $output, $engine );
+		$this->getHookRunner()->onSearchDataForIndex2( $fieldData, $this, $page, $output, $engine, $revision );
+
 		return $fieldData;
 	}
 
@@ -1426,10 +1445,15 @@ abstract class ContentHandler {
 	 *
 	 * @param WikiPage $page
 	 * @param ParserCache|null $cache deprecated since 1.38 and won't have any effect
+	 * @param RevisionRecord|null $revision
 	 * @return ParserOutput|null null when the ParserOutput cannot be obtained
 	 * @see ParserOutputAccess::getParserOutput() for failure modes
 	 */
-	public function getParserOutputForIndexing( WikiPage $page, ParserCache $cache = null ) {
+	public function getParserOutputForIndexing(
+		WikiPage $page,
+		ParserCache $cache = null,
+		RevisionRecord $revision = null
+	) {
 		// TODO: MCR: ContentHandler should be called per slot, not for the whole page.
 		// See T190066.
 		$parserOptions = $page->makeParserOptions( 'canonical' );
@@ -1437,7 +1461,7 @@ abstract class ContentHandler {
 		return $parserOutputAccess->getParserOutput(
 			$page,
 			$parserOptions,
-			null,
+			$revision,
 			ParserOutputAccess::OPT_NO_UPDATE_CACHE
 		)->getValue();
 	}

includes/content/FileContentHandler.php

@@ -1,6 +1,7 @@
 <?php
 
 use MediaWiki\MediaWikiServices;
+use MediaWiki\Revision\RevisionRecord;
 
 /**
  * Content handler for File: files
@@ -36,7 +37,8 @@ class FileContentHandler extends WikitextContentHandler {
 	public function getDataForSearchIndex(
 		WikiPage $page,
 		ParserOutput $parserOutput,
-		SearchEngine $engine
+		SearchEngine $engine,
+		?RevisionRecord $revision = null
 	) {
 		$fields = [];
 

includes/content/Hook/SearchDataForIndex2Hook.php

@@ -0,0 +1,42 @@
+<?php
+
+namespace MediaWiki\Content\Hook;
+
+use ContentHandler;
+use MediaWiki\Revision\RevisionRecord;
+use ParserOutput;
+use SearchEngine;
+use WikiPage;
+
+/**
+ * This is a hook handler interface, see docs/Hooks.md.
+ * Use the hook name "SearchDataForIndex" to register handlers implementing this interface.
+ *
+ * @stable to implement
+ * @ingroup Hooks
+ */
+interface SearchDataForIndex2Hook {
+
+	/**
+	 * Use this hook to add data to search document. Allows you to add any data to
+	 * the field map used to index the document.
+	 *
+	 * @since 1.40
+	 *
+	 * @param array &$fields Array of name => value pairs for fields
+	 * @param ContentHandler $handler ContentHandler for the content being indexed
+	 * @param WikiPage $page WikiPage that is being indexed
+	 * @param ParserOutput $output ParserOutput that is produced from the page
+	 * @param SearchEngine $engine SearchEngine for which the indexing is intended
+	 * @param RevisionRecord $revision RevisionRecord being indexed
+	 * @return bool|void True or no return value to continue or false to abort
+	 */
+	public function onSearchDataForIndex2(
+		array &$fields,
+		ContentHandler $handler,
+		WikiPage $page,
+		ParserOutput $output,
+		SearchEngine $engine,
+		RevisionRecord $revision
+	);
+}

includes/content/Hook/SearchDataForIndexHook.php

@@ -12,6 +12,7 @@ use WikiPage;
  * Use the hook name "SearchDataForIndex" to register handlers implementing this interface.
  *
  * @stable to implement
+ * @deprecated since 1.40, use SearchDataForIndexHook2 instead.
  * @ingroup Hooks
  */
 interface SearchDataForIndexHook {

includes/content/TextContentHandler.php

@@ -27,6 +27,7 @@ use MediaWiki\Content\Renderer\ContentParseParams;
 use MediaWiki\Content\Transform\PreSaveTransformParams;
 use MediaWiki\MainConfigNames;
 use MediaWiki\MediaWikiServices;
+use MediaWiki\Revision\RevisionRecord;
 
 /**
  * Base content handler implementation for flat text contents.
@@ -157,9 +158,10 @@ class TextContentHandler extends ContentHandler {
 	public function getDataForSearchIndex(
 		WikiPage $page,
 		ParserOutput $output,
-		SearchEngine $engine
+		SearchEngine $engine,
+		?RevisionRecord $revision = null
 	) {
-		$fields = parent::getDataForSearchIndex( $page, $output, $engine );
+		$fields = parent::getDataForSearchIndex( $page, $output, $engine, $revision );
 		$fields['language'] =
 			$this->getPageLanguage( $page->getTitle(), $page->getContent() )->getCode();
 		return $fields;

includes/content/WikitextContentHandler.php

@@ -29,6 +29,7 @@ use MediaWiki\Content\Transform\PreSaveTransformParams;
 use MediaWiki\Languages\LanguageNameUtils;
 use MediaWiki\MediaWikiServices;
 use MediaWiki\Parser\ParserOutputFlags;
+use MediaWiki\Revision\RevisionRecord;
 
 /**
  * Content handler for wiki text pages.
@@ -157,9 +158,10 @@ class WikitextContentHandler extends TextContentHandler {
 	public function getDataForSearchIndex(
 		WikiPage $page,
 		ParserOutput $parserOutput,
-		SearchEngine $engine
+		SearchEngine $engine,
+		?RevisionRecord $revision = null
 	) {
-		$fields = parent::getDataForSearchIndex( $page, $parserOutput, $engine );
+		$fields = parent::getDataForSearchIndex( $page, $parserOutput, $engine, $revision );
 
 		$structure = new WikiTextStructure( $parserOutput );
 		$fields['heading'] = $structure->headings();
@@ -172,7 +174,7 @@ class WikitextContentHandler extends TextContentHandler {
 		// Until we have full first-class content handler for files, we invoke it explicitly here
 		if ( $page->getTitle()->getNamespace() === NS_FILE ) {
 			$fields = array_merge( $fields,
-					$this->getFileHandler()->getDataForSearchIndex( $page, $parserOutput, $engine ) );
+					$this->getFileHandler()->getDataForSearchIndex( $page, $parserOutput, $engine, $revision ) );
 		}
 		return $fields;
 	}

tests/phpunit/includes/content/ContentHandlerTest.php

@@ -444,9 +444,9 @@ class ContentHandlerTest extends MediaWikiIntegrationTestCase {
 				$fields['testDataField'] = 'test content';
 			} );
 
-		$contentRenderer = $this->getServiceContainer()->getContentRenderer();
-		$output = $contentRenderer->getParserOutput( $page->getContent(), $title );
-		$data = $page->getContentHandler()->getDataForSearchIndex( $page, $output, $mockEngine );
+		$revision = $page->getRevisionRecord();
+		$output = $page->getContentHandler()->getParserOutputForIndexing( $page, null,  $revision );
+		$data = $page->getContentHandler()->getDataForSearchIndex( $page, $output, $mockEngine, $revision );
 		$this->assertArrayHasKey( 'text', $data );
 		$this->assertArrayHasKey( 'text_bytes', $data );
 		$this->assertArrayHasKey( 'language', $data );
@@ -461,8 +461,9 @@ class ContentHandlerTest extends MediaWikiIntegrationTestCase {
 	public function testParserOutputForIndexing() {
 		$title = Title::newFromText( 'Smithee', NS_MAIN );
 		$page = $this->getServiceContainer()->getWikiPageFactory()->newFromTitle( $title );
+		$revision = $page->getRevisionRecord();
 
-		$out = $page->getContentHandler()->getParserOutputForIndexing( $page );
+		$out = $page->getContentHandler()->getParserOutputForIndexing( $page, null, $revision );
 		$this->assertInstanceOf( ParserOutput::class, $out );
 		$this->assertStringContainsString( 'one who smiths', $out->getRawText() );
 	}