Incoming Feed Items: Fix in-article hash links (#426)

akirk · Jan 2, 2025 · 06fdf72 · 06fdf72
1 parent 2336ac2
commit 06fdf72
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 6 deletions.
diff --git a/feed-parsers/class-feed-parser-simplepie.php b/feed-parsers/class-feed-parser-simplepie.php
@@ -293,7 +293,7 @@ public function process_items( $items, $url ) {
 				array(
 					'permalink' => $item->get_permalink(),
 					'title'     => $title,
-					'content'   => $this->convert_relative_urls_to_absolute_urls( $item->get_content(), $url ),
+					'content'   => $this->convert_relative_urls_to_absolute_urls( $item->get_content(), $item->get_permalink() ),
 				)
 			);
 
@@ -347,7 +347,10 @@ public function process_items( $items, $url ) {
 			}
 
 			if ( is_object( $item->get_author() ) ) {
-				$feed_item->author = \wp_strip_all_tags( $item->get_author()->name );
+				$feed_item->author = $item->get_author()->name;
+				// Strip HTML tags, even if they are escaped.
+				$feed_item->author = htmlspecialchars_decode( $feed_item->author, ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML401 );
+				$feed_item->author = \wp_strip_all_tags( $feed_item->author );
 			}
 
 			$feed_item->date         = $item->get_gmdate( 'U' );

diff --git a/feed-parsers/class-feed-parser.php b/feed-parsers/class-feed-parser.php
@@ -78,19 +78,34 @@ public function discover_available_feeds( $content, $url ) { // phpcs:ignore Gen
 	 * Convert relative URLs to absolute ones in incoming content.
 	 *
 	 * @param      string $html   The html.
-	 * @param      string $url    The url of the feed.
+	 * @param      string $permalink  The permalink of the feed.
 	 *
 	 * @return     string  The HTML with URLs replaced to their absolute represenation.
 	 */
-	public function convert_relative_urls_to_absolute_urls( $html, $url ) {
+	public function convert_relative_urls_to_absolute_urls( $html, $permalink ) {
 		if ( ! $html ) {
 			$html = '';
 		}
+
+		// Strip off the hash.
+		$permalink = strtok( $permalink, '#' );
+
 		// For now this only converts links and image srcs.
 		return preg_replace_callback(
 			'~(src|href)=(?:"([^"]+)|\'([^\']+))~i',
-			function ( $m ) use ( $url ) {
-				return str_replace( $m[2], Mf2\resolveUrl( $url, $m[2] ), $m[0] );
+			function ( $m ) use ( $permalink ) {
+				if ( str_starts_with( $m[2], '#' ) ) {
+					// Don't update hash-only links.
+					return $m[0];
+				}
+
+				if ( str_starts_with( $m[2], $permalink . '#' ) ) {
+					// Remove absolute URL from hashes.
+					return str_replace( $permalink, '', $m[0] );
+				}
+
+				// Convert relative URLs to absolute ones.
+				return str_replace( $m[2], Mf2\resolveUrl( $permalink, $m[2] ), $m[0] );
 			},
 			$html
 		);