Improve tags regex

akirk · Jun 1, 2023 · cc8b2e6 · cc8b2e6
1 parent 7135cb6
commit cc8b2e6
Showing 1 changed file with 11 additions and 5 deletions.
diff --git a/includes/class-feed.php b/includes/class-feed.php
@@ -61,7 +61,7 @@ private function register_hooks() {
 		add_filter( 'pre_option_rss_use_excerpt', array( $this, 'feed_use_excerpt' ), 90 );
 		add_filter( 'friends_early_modify_feed_item', array( $this, 'apply_early_feed_rules' ), 10, 3 );
 		add_filter( 'friends_modify_feed_item', array( $this, 'apply_feed_rules' ), 10, 3 );
-		add_filter( 'friends_modify_feed_item', array( $this, 'extract_tags' ), 10, 3 );
+		add_filter( 'friends_modify_feed_item', array( $this, 'extract_tags' ) );
 
 		add_action( 'rss_item', array( $this, 'feed_additional_fields' ) );
 		add_action( 'rss2_item', array( $this, 'feed_additional_fields' ) );
@@ -442,11 +442,16 @@ public function validate_feed_catch_all( $catch_all ) {
 		return $catch_all;
 	}
 
-	public function extract_tags( $item, User_Feed $feed = null, User $friend_user = null ) {
-		// This is a variation of the ACTIVITYPUB_HASHTAGS_REGEXP.
-		if ( \preg_match_all( '/(?:(?<=\s)|(?<=<p>)|(?<=<br>)|^)#(\w+)(?:(?=\s|[[:punct:]]|$))/ui', \wp_strip_all_tags( $item->post_content ), $match ) ) {
+	public function extract_tags( $item ) {
+		// (?<!&) avoid to match HTML entities like &#039;.
+		// [\w-] to allow tags with dashes in them.
+		if ( \preg_match_all( '/(?<!&|\w)#([\w-]+)/ui', \wp_strip_all_tags( $item->post_content ), $match ) ) {
 			$tags = \implode( ', ', $match[1] );
-			$item->tags = $tags;
+			if ( is_array( $item->tags ) ) {
+				$item->tags = array_merge( $item->tags, $tags );
+			} else {
+				$item->tags = $tags;
+			}
 		}
 
 		return $item;
@@ -633,6 +638,7 @@ public function process_incoming_feed_items( array $items, User_Feed $user_feed
 						$modified_posts[] = $post_id;
 					}
 				}
+				wp_set_post_tags( $post_id, $item->tags );
 			} else {
 				$post_data['post_type']     = Friends::CPT;
 				$post_data['post_date_gmt'] = $item->date;