Use RemexHtml to properly remove thumbnails

edwardspec · Dec 30, 2024 · 1828250 · 1828250
1 parent 8247ebe
commit 1828250
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 11 deletions.
diff --git a/includes/HtmlSanitizer.php b/includes/HtmlSanitizer.php
@@ -46,6 +46,9 @@ public function startDocument( $fragmentNamespace, $fragmentName ) {
 
 			/** @inheritDoc */
 			public function element( SerializerNode $parent, SerializerNode $node, $contents ) {
+				$typeof = $node->attrs['typeof'] ?? null;
+				$classes = explode( ' ', $node->attrs['class'] ?? '' );
+
 				switch ( $node->name ) {
 					// Remove everything outside the <body> tag.
 					case 'head':
@@ -55,20 +58,33 @@ public function element( SerializerNode $parent, SerializerNode $node, $contents
 					case 'body':
 						return $contents;
 
-					case 'img':
-						// Remove the image tags: in 99,9% of cases they are too wide
-						// to be included into the calendar.
-						// Not needed in MediaWiki 1.40+ (already removed with the <span> below).
-						return '';
+					case 'a':
+						// MediaWiki 1.39 only: remove the links around non-thumbnail images.
+						if ( in_array( 'image', $classes ) ) {
+							return '';
+						}
+						break;
+
+					case 'div':
+						// MediaWiki 1.39 only: remove wrapper around the thumbnail.
+						if ( in_array( 'thumb', $classes ) ) {
+							return '';
+						}
+						break;
 
 					case 'span':
-						if ( ( $node->attrs['typeof'] ?? '' ) === 'mw:File' ) {
-							// Wrapper around the image.
+						// MediaWiki 1.40+: remove the wrappers around non-thumbnail images.
+						if ( $typeof === 'mw:File' ) {
 							return '';
 						}
 						break;
 
-					// TODO: properly remove <div class="thumb"> with all contents (currently hidden by CSS).
+					case 'figure':
+						// MediaWiki 1.40+: remove wrapper around the thumbnail.
+						if ( $typeof === 'mw:File/Thumb' ) {
+							return '';
+						}
+						break;
 
 					case 'p':
 						// Remove trailing newline inside <p> tags.

diff --git a/tests/phpunit/EventCalendarTest.php b/tests/phpunit/EventCalendarTest.php
@@ -774,10 +774,12 @@ public function testSnippetForCompressedRevision() {
 
 	/**
 	 * Verify that unwanted parts of HTML (such as images) are removed from the snippet.
+	 * @dataProvider dataProviderSnippetSanitizer
+	 * @param bool $isThumb False to add an image to the page, true to add a thumbnail.
 	 */
-	public function testSnippetSanitizer() {
-		$filename = 'Testimage.png';
-		$pageText = "Expected snippet [[File:$filename]]";
+	public function testSnippetSanitizer( $isThumb ) {
+		$filename = 'Testimage' . ( $isThumb ? '1' : '2' ) . '.png';
+		$pageText = 'Expected snippet [[File:' . $filename . ( $isThumb ? '|thumb' : '' ) . ']]';
 		$expectedSnippet = '<p>Expected snippet</p>';
 
 		// Upload a test file, so that [[File:]] syntax would create an actual thumbnail, not a redlink.
@@ -800,6 +802,17 @@ public function testSnippetSanitizer() {
 		$this->assertSame( $expectedSnippet, $actualData[0]['title'] );
 	}
 
+	/**
+	 * Provides datasets for testSnippetSanitizer().
+	 * @return array
+	 */
+	public function dataProviderSnippetSanitizer() {
+		return [
+			'image (not a thumbnail)' => [ false ],
+			'thumbnail' => [ true ]
+		];
+	}
+
 	/**
 	 * Verify that parameters like height=300 and aspectratio=1.5 are provided to JavaScript library.
 	 * @dataProvider dataProviderOptionalAttributes