OutputTransform: Fix double IDs on headings

Based on Ifeaaba1d0215e6f67f889a09c02879cc9079aa19

Bug: T366083
Co-Authored-by: Bartosz Dziewoński <dziewonski@fastmail.fm>
Change-Id: I2712e0fa9272106e8cd686980f847ee7f6385b6f
(cherry picked from commit 5757066096a0eac7f722e63aa3722e068915d33a)

includes/OutputTransform/Stages/HandleSectionLinks.php

@@ -155,6 +155,10 @@ class HandleSectionLinks extends ContentTextTransformStage {
 		$link, $fallbackAnchor, string $wrapperType
 	) {
 		$anchorEscaped = htmlspecialchars( $anchor, ENT_COMPAT );
+		$idAttr = " id=\"$anchorEscaped\"";
+		if ( isset( $attrs['id'] ) ) {
+			$idAttr = '';
+		}
 		$fallback = '';
 		if ( $fallbackAnchor !== false && $fallbackAnchor !== $anchor ) {
 			$fallbackAnchor = htmlspecialchars( $fallbackAnchor, ENT_COMPAT );
@@ -164,16 +168,16 @@ class HandleSectionLinks extends ContentTextTransformStage {
 		switch ( $wrapperType ) {
 			case 'legacy':
 				return "<h$level" . Html::expandAttributes( $attrs ) . ">"
-					. "$fallback<span class=\"mw-headline\" id=\"$anchorEscaped\">$html</span>"
+					. "$fallback<span class=\"mw-headline\"$idAttr>$html</span>"
 					. $link
 					. "</h$level>";
 			case 'mwheading':
 				return "<div class=\"mw-heading mw-heading$level\">"
-					. "<h$level id=\"$anchorEscaped\"" . Html::expandAttributes( $attrs ) . ">$fallback$html</h$level>"
+					. "<h$level$idAttr" . Html::expandAttributes( $attrs ) . ">$fallback$html</h$level>"
 					. $link
 					. "</div>";
 			case 'none':
-				return "<h$level id=\"$anchorEscaped\"" . Html::expandAttributes( $attrs ) . ">$fallback$html</h$level>"
+				return "<h$level$idAttr" . Html::expandAttributes( $attrs ) . ">$fallback$html</h$level>"
 					. $link;
 			default:
 				throw new LogicException( "Bad wrapper type: $wrapperType" );

includes/parser/Parser.php

@@ -4361,9 +4361,20 @@ class Parser {
 			// conveniently also giving us a way to handle French spaces (T324763)
 			$safeHeadline = $this->tidy->tidy( $safeHeadline, [ Sanitizer::class, 'armorFrenchSpaces' ] );
 
+			// Wrap the safe headline to parse the heading attributes
+			// Literal HTML tags should be sanitized at this point
+			// cleanUpTocLine will strip the headline tag
+			$wrappedHeadline = "<h$level" . $matches['attrib'][$headlineCount] . $safeHeadline . "</h$level>";
+
 			// Parse the heading contents as HTML. This makes it easier to strip out some HTML tags,
 			// and ensures that we generate balanced HTML at the end (T218330).
-			$headlineDom = DOMUtils::parseHTMLToFragment( $domDocument, $safeHeadline );
+			$headlineDom = DOMUtils::parseHTMLToFragment( $domDocument, $wrappedHeadline );
+
+			// Extract a user defined id on the heading
+			// A heading is expected as the first child and could be asserted
+			$h = $headlineDom->firstChild;
+			$headingId = ( $h instanceof Element && DOMUtils::isHeading( $h ) ) ?
+				DOMCompat::getAttribute( $h, 'id' ) : null;
 
 			$this->cleanUpTocLine( $headlineDom );
 
@@ -4372,12 +4383,17 @@ class Parser {
 
 			# For the anchor, strip out HTML-y stuff period
 			$safeHeadline = trim( $headlineDom->textContent );
+
 			# Save headline for section edit hint before it's normalized for the link
 			$headlineHint = htmlspecialchars( $safeHeadline );
 
 			$safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
 			$safeHeadline = self::normalizeSectionName( $safeHeadline );
 
+			if ( $headingId !== null && $headingId !== '' ) {
+				$safeHeadline = $headingId;
+			}
+
 			$fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
 			$linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
 			$safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );

tests/parser/headings.txt

@@ -2353,10 +2353,10 @@ wgParserEnableLegacyHeadingDOM=false
 !! wikitext
 <h3 id="asdf">odd</h3>
 !! html/php
-<h3 id="odd" id="asdf">odd</h3>
+<h3 id="asdf">odd</h3>
 !! metadata/php
 Sections:
- h3 index: toclevel:1 number:1 title:NULL off:NULL anchor/linkAnchor:odd line:odd
+ h3 index: toclevel:1 number:1 title:NULL off:NULL anchor/linkAnchor:asdf line:odd
 !! html/parsoid
 <section data-mw-section-id="0"></section><section data-mw-section-id="-1"><h3 id="asdf" data-parsoid='{"stx":"html","reusedId":true}'>odd</h3></section>
 !! metadata/parsoid

tests/parser/legacyHeadings.txt

@@ -2350,10 +2350,10 @@ wgParserEnableLegacyHeadingDOM=true
 !! wikitext
 <h3 id="asdf">odd</h3>
 !! html/php
-<h3 id="asdf"><span class="mw-headline" id="odd">odd</span></h3>
+<h3 id="asdf"><span class="mw-headline">odd</span></h3>
 !! metadata/php
 Sections:
- h3 index: toclevel:1 number:1 title:NULL off:NULL anchor/linkAnchor:odd line:odd
+ h3 index: toclevel:1 number:1 title:NULL off:NULL anchor/linkAnchor:asdf line:odd
 !! html/parsoid
 <section data-mw-section-id="0"></section><section data-mw-section-id="-1"><h3 id="asdf" data-parsoid='{"stx":"html","reusedId":true}'>odd</h3></section>
 !! metadata/parsoid