Merge "REST compare: bundle section offset information"

includes/Rest/Handler/CompareHandler.php

@@ -11,6 +11,7 @@ use MediaWiki\Revision\RevisionLookup;
 use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\Revision\SlotRecord;
 use MediaWiki\Revision\SuppressedDataException;
+use Parser;
 use RequestContext;
 use TextContent;
 use User;
@@ -24,18 +25,26 @@ class CompareHandler extends Handler {
 	/** @var PermissionManager */
 	private $permissionManager;
 
+	/** @var Parser */
+	private $parser;
+
 	/** @var User */
 	private $user;
 
 	/** @var RevisionRecord[] */
 	private $revisions = [];
 
+	/** @var string[] */
+	private $textCache = [];
+
 	public function __construct(
 		RevisionLookup $revisionLookup,
-		PermissionManager $permissionManager
+		PermissionManager $permissionManager,
+		Parser $parser
 	) {
 		$this->revisionLookup = $revisionLookup;
 		$this->permissionManager = $permissionManager;
+		$this->parser = $parser;
 
 		// @todo Inject this, when there is a good way to do that
 		$this->user = RequestContext::getMain()->getUser();
@@ -61,10 +70,12 @@ class CompareHandler extends Handler {
 			'from' => [
 				'id' => $fromRev->getId(),
 				'slot_role' => $this->getRole(),
+				'sections' => $this->getSectionInfo( 'from' )
 			],
 			'to' => [
 				'id' => $toRev->getId(),
-				'slot_role' => $this->getRole()
+				'slot_role' => $this->getRole(),
+				'sections' => $this->getSectionInfo( 'to' )
 			],
 			'diff' => [ 'PLACEHOLDER' => null ]
 		];
@@ -129,29 +140,32 @@ class CompareHandler extends Handler {
 	}
 
 	private function getRevisionText( $paramName ) {
-		$revision = $this->getRevision( $paramName );
-		try {
-			$content = $revision
-				->getSlot( $this->getRole(), RevisionRecord::FOR_THIS_USER, $this->user )
-				->getContent()
-				->convert( CONTENT_MODEL_TEXT );
-			if ( $content instanceof TextContent ) {
-				return $content->getText();
-			} else {
+		if ( !isset( $this->textCache[$paramName] ) ) {
+			$revision = $this->getRevision( $paramName );
+			try {
+				$content = $revision
+					->getSlot( $this->getRole(), RevisionRecord::FOR_THIS_USER, $this->user )
+					->getContent()
+					->convert( CONTENT_MODEL_TEXT );
+				if ( $content instanceof TextContent ) {
+					$this->textCache[$paramName] = $content->getText();
+				} else {
+					throw new LocalizedHttpException(
+						new MessageValue(
+							'rest-compare-wrong-content',
+							[ $this->getRole(), $paramName ]
+						),
+						400 );
+				}
+			} catch ( SuppressedDataException $e ) {
 				throw new LocalizedHttpException(
-					new MessageValue(
-						'rest-compare-wrong-content',
-						[ $this->getRole(), $paramName ]
-					),
-					400 );
+					new MessageValue( 'rest-compare-inaccessible', [ $paramName ] ), 403 );
+			} catch ( RevisionAccessException $e ) {
+				throw new LocalizedHttpException(
+					new MessageValue( 'rest-compare-nonexistent', [ $paramName ] ), 404 );
 			}
-		} catch ( SuppressedDataException $e ) {
-			throw new LocalizedHttpException(
-				new MessageValue( 'rest-compare-inaccessible', [ $paramName ] ), 403 );
-		} catch ( RevisionAccessException $e ) {
-			throw new LocalizedHttpException(
-				new MessageValue( 'rest-compare-nonexistent', [ $paramName ] ), 404 );
 		}
+		return $this->textCache[$paramName];
 	}
 
 	/**
@@ -169,6 +183,27 @@ class CompareHandler extends Handler {
 		return wikidiff2_inline_json_diff( $fromText, $toText, 2 );
 	}
 
+	/**
+	 * @return array
+	 */
+	private function getSectionInfo( $paramName ) {
+		$text = $this->getRevisionText( $paramName );
+		$parserSections = $this->parser->getFlatSectionInfo( $text );
+		$sections = [];
+		foreach ( $parserSections as $i => $parserSection ) {
+			// Skip section zero, which comes before the first heading, since
+			// its offset is always zero, so the client can assume its location.
+			if ( $i !== 0 ) {
+				$sections[] = [
+					'level' => $parserSection['level'],
+					'heading' => $parserSection['heading'],
+					'offset' => $parserSection['offset'],
+				];
+			}
+		}
+		return $sections;
+	}
+
 	public function getParamSettings() {
 		return [
 			'from' => [

includes/Rest/coreRoutes.json

@@ -28,7 +28,8 @@
 		"class": "MediaWiki\\Rest\\Handler\\CompareHandler",
 		"services": [
 			"RevisionLookup",
-			"PermissionManager"
+			"PermissionManager",
+			"Parser"
 		]
 	},
 	{

includes/parser/Parser.php

@@ -5880,6 +5880,72 @@ class Parser {
 		return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
 	}
 
+	/**
+	 * Get an array of preprocessor section information.
+	 *
+	 * Preprocessor sections are those identified by wikitext-style syntax, not
+	 * HTML-style syntax. Templates are not expanded, so these sections do not
+	 * include sections created by templates or parser functions. This is the
+	 * same definition of a section as used by section editing, but not the
+	 * same as TOC generation.
+	 *
+	 * These sections are typically smaller than those acted on by getSection() and
+	 * replaceSection() since they are not nested. Section nesting could be
+	 * reconstructed from the heading levels.
+	 *
+	 * The return value is an array of associative array info structures. Each
+	 * associative array contains the following keys, describing a section:
+	 *
+	 *  - index: An integer identifying the section.
+	 *  - level: The heading level, e.g. 1 for <h1>. For the section before the
+	 *    the first heading, this will be 0.
+	 *  - offset: The byte offset within the wikitext at which the section starts
+	 *  - heading: The wikitext for the header which introduces the section,
+	 *    including equals signs. For the section before the first heading, this
+	 *    will be an empty string.
+	 *  - text: The complete text of the section.
+	 *
+	 * @param string $text
+	 * @return array[]
+	 */
+	public function getFlatSectionInfo( $text ) {
+		$magicScopeVariable = $this->lock();
+		$this->startParse( null, new ParserOptions, self::OT_PLAIN, true );
+		$frame = $this->getPreprocessor()->newFrame();
+		$root = $this->preprocessToDom( $text, 0 );
+		$node = $root->getFirstChild();
+		$offset = 0;
+		$currentSection = [
+			'index' => 0,
+			'level' => 0,
+			'offset' => 0,
+			'heading' => '',
+			'text' => ''
+		];
+		$sections = [];
+
+		while ( $node ) {
+			$nodeText = $frame->expand( $node, PPFrame::RECOVER_ORIG );
+			if ( $node->getName() === 'h' ) {
+				$bits = $node->splitHeading();
+				$sections[] = $currentSection;
+				$currentSection = [
+					'index' => $bits['i'],
+					'level' => $bits['level'],
+					'offset' => $offset,
+					'heading' => $nodeText,
+					'text' => $nodeText
+				];
+			} else {
+				$currentSection['text'] .= $nodeText;
+			}
+			$offset += strlen( $nodeText );
+			$node = $node->getNextSibling();
+		}
+		$sections[] = $currentSection;
+		return $sections;
+	}
+
 	/**
 	 * Get the ID of the revision we are parsing
 	 *