From 91316d49c248a32de9cb5dad67d8f52cf40094e3 Mon Sep 17 00:00:00 2001
From: Isabelle Hurbain-Palatin <ihurbainpalatin@wikimedia.org>
Date: Mon, 18 Nov 2024 16:56:03 +0100
Subject: [PATCH] Normalize ref html before comparison

If the ref generated HTML contains attributes such as data-parsoid or
about, the string comparison fails, and two references that have the
same content end up being seen as different.
This patch filters out data-parsoid and about attributes from the
generated HTML.

Bug: T380152
Change-Id: I4c201c469d21ff9cb54f746386ce53b8d636361d
---
 src/Parsoid/References.php | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/src/Parsoid/References.php b/src/Parsoid/References.php
index 046316cbc..ec1bfaaf6 100644
--- a/src/Parsoid/References.php
+++ b/src/Parsoid/References.php
@@ -10,6 +10,7 @@ use Cite\Cite;
 use Cite\MarkSymbolRenderer;
 use Closure;
 use MediaWiki\Config\Config;
+use MediaWiki\HtmlHelper;
 use MediaWiki\MediaWikiServices;
 use stdClass;
 use Wikimedia\Message\MessageValue;
@@ -27,6 +28,7 @@ use Wikimedia\Parsoid\NodeData\DataMw;
 use Wikimedia\Parsoid\NodeData\DataMwError;
 use Wikimedia\Parsoid\NodeData\DataParsoid;
 use Wikimedia\Parsoid\Utils\DOMCompat;
+use Wikimedia\RemexHtml\Serializer\SerializerNode;
 
 /**
  * @license GPL-2.0-or-later
@@ -226,7 +228,7 @@ class References extends ExtensionTagHandler {
 					// Ideally, we should strip the mw:Cite/Follow wrappers before comparing
 					// But, we are going to ignore this edge case as not worth the complexity.
 					$html = $extApi->domToHtml( $c, true, false );
-					$contentDiffers = ( $html !== $ref->cachedHtml );
+					$contentDiffers = ( $this->normalizeRef( $html ) !== $this->normalizeRef( $ref->cachedHtml ) );
 				}
 			} else {
 				if ( $refsData->inReferencesContent() ) {
@@ -878,4 +880,17 @@ class References extends ExtensionTagHandler {
 
 		return false;
 	}
+
+	private function normalizeRef( string $s ): string {
+		return HtmlHelper::modifyElements( $s,
+			static function ( SerializerNode $node ): bool {
+				return isset( $node->attrs['data-parsoid'] ) || isset( $node->attrs['about'] );
+			},
+			static function ( SerializerNode $node ): SerializerNode {
+				unset( $node->attrs['data-parsoid'] );
+				unset( $node->attrs['about'] );
+				return $node;
+			}
+		);
+	}
 }