Add local caching for remote actor avatars (#2610)

Author: pfefferle

.github/changelog/2610-from-description

@@ -0,0 +1,4 @@
+Significance: minor
+Type: added
+
+Add local caching for remote actor avatars.

includes/class-attachments.php

@@ -8,6 +8,7 @@
 namespace Activitypub;
 
 use Activitypub\Collection\Posts;
+use Activitypub\Collection\Remote_Actors;
 
 /**
  * Attachments processor class.
@@ -27,18 +28,33 @@ class Attachments {
 	 */
 	public static $comments_dir = '/activitypub/comments/';
 
+	/**
+	 * Directory for storing actor avatar files.
+	 *
+	 * @var string
+	 */
+	public static $actors_dir = '/activitypub/actors/';
+
 	/**
 	 * Maximum width for imported images.
 	 *
 	 * @var int
 	 */
 	const MAX_IMAGE_DIMENSION = 1200;
 
+	/**
+	 * Maximum width for actor avatars.
+	 *
+	 * @var int
+	 */
+	const MAX_AVATAR_DIMENSION = 512;
+
 	/**
 	 * Initialize the class and set up filters.
 	 */
 	public static function init() {
 		\add_action( 'before_delete_post', array( self::class, 'delete_ap_posts_directory' ) );
+		\add_action( 'before_delete_post', array( self::class, 'delete_actors_directory' ) );
 	}
 
 	/**
@@ -207,7 +223,18 @@ private static function import_files_for_object( $attachments, $object_id, $obje
 	 */
 	private static function get_storage_paths( $object_id, $object_type ) {
 		$upload_dir = \wp_upload_dir();
-		$sub_dir    = 'comment' === $object_type ? self::$comments_dir : self::$ap_posts_dir;
+
+		switch ( $object_type ) {
+			case 'comment':
+				$sub_dir = self::$comments_dir;
+				break;
+			case 'actor':
+				$sub_dir = self::$actors_dir;
+				break;
+			default:
+				$sub_dir = self::$ap_posts_dir;
+				break;
+		}
 
 		return array(
 			'basedir' => $upload_dir['basedir'] . $sub_dir . $object_id,
@@ -958,4 +985,62 @@ private static function get_files_gallery_block( $files ) {
 
 		return $gallery;
 	}
+
+	/**
+	 * Save a remote actor's avatar locally.
+	 *
+	 * Downloads the avatar image, optimizes it, and stores it in the actors directory.
+	 * Returns the local URL for the saved avatar.
+	 *
+	 * @param int    $actor_id   The local actor post ID.
+	 * @param string $avatar_url The remote avatar URL.
+	 *
+	 * @return string|false The local avatar URL on success, false on failure.
+	 */
+	public static function save_actor_avatar( $actor_id, $avatar_url ) {
+		// Validate actor_id is a positive integer to prevent path traversal.
+		$actor_id = (int) $actor_id;
+		if ( $actor_id <= 0 ) {
+			return false;
+		}
+
+		if ( empty( $avatar_url ) || ! \filter_var( $avatar_url, FILTER_VALIDATE_URL ) ) {
+			return false;
+		}
+
+		// Delete existing avatar files before saving new one.
+		// This prevents accumulating old avatar files since save_file creates unique filenames.
+		self::delete_actors_directory( $actor_id );
+
+		$attachment_data = array( 'url' => $avatar_url );
+		$result          = self::save_file( $attachment_data, $actor_id, 'actor', self::MAX_AVATAR_DIMENSION );
+
+		if ( \is_wp_error( $result ) || ! isset( $result['url'] ) ) {
+			return false;
+		}
+
+		return $result['url'];
+	}
+
+	/**
+	 * Delete the activitypub files directory for an actor.
+	 *
+	 * @param int $actor_id The actor post ID.
+	 */
+	public static function delete_actors_directory( $actor_id ) {
+		if ( Remote_Actors::POST_TYPE !== \get_post_type( $actor_id ) ) {
+			return;
+		}
+
+		require_once ABSPATH . 'wp-admin/includes/file.php';
+
+		\WP_Filesystem();
+		global $wp_filesystem;
+
+		$activitypub_dir = self::get_storage_paths( $actor_id, 'actor' )['basedir'];
+
+		if ( $wp_filesystem->is_dir( $activitypub_dir ) ) {
+			$wp_filesystem->rmdir( $activitypub_dir, true );
+		}
+	}
 }

includes/collection/class-remote-actors.php

@@ -8,6 +8,7 @@
 namespace Activitypub\Collection;
 
 use Activitypub\Activity\Actor;
+use Activitypub\Attachments;
 use Activitypub\Http;
 use Activitypub\Sanitize;
 use Activitypub\Webfinger;
@@ -111,6 +112,11 @@ public static function create( $actor ) {
 			\kses_init_filters();
 		}
 
+		// Cache the actor's avatar locally.
+		if ( ! \is_wp_error( $post_id ) ) {
+			self::cache_avatar( $post_id, $actor );
+		}
+
 		return $post_id;
 	}
 
@@ -158,6 +164,11 @@ public static function update( $post, $actor ) {
 			\kses_init_filters();
 		}
 
+		// Re-cache the actor's avatar if it has changed.
+		if ( ! \is_wp_error( $post_id ) ) {
+			self::cache_avatar( $post_id, $actor );
+		}
+
 		return $post_id;
 	}
 
@@ -531,12 +542,6 @@ private static function prepare_custom_post_type( $actor ) {
 			'_activitypub_acct'  => $webfinger,
 		);
 
-		// Store avatar URL if available.
-		$icon = object_to_uri( $actor->get_icon() );
-		if ( $icon ) {
-			$meta_input['_activitypub_avatar_url'] = $icon;
-		}
-
 		return array(
 			'guid'         => \esc_url_raw( $actor->get_id() ),
 			'post_title'   => \wp_strip_all_tags( \wp_slash( $actor->get_name() ?: $actor->get_preferred_username() ) ),
@@ -653,6 +658,9 @@ public static function get_acct( $id ) {
 	/**
 	 * Get the avatar URL for a remote actor.
 	 *
+	 * Returns the locally cached avatar URL if available, otherwise falls back
+	 * to the default avatar.
+	 *
 	 * @param int $id The ID of the remote actor post.
 	 *
 	 * @return string The avatar URL or empty string if not found.
@@ -663,7 +671,7 @@ public static function get_avatar_url( $id ) {
 			return $avatar_url;
 		}
 
-		// If not found in meta, try to extract from post_content JSON.
+		// If not found in meta, try to extract from post_content JSON and cache it.
 		$post = \get_post( $id );
 		if ( ! $post || empty( $post->post_content ) ) {
 			return '';
@@ -677,9 +685,36 @@ public static function get_avatar_url( $id ) {
 			return $default_avatar_url;
 		}
 
-		$avatar_url = object_to_uri( $actor_data['icon'] );
-		// Cache it in meta for next time.
-		\update_post_meta( $id, '_activitypub_avatar_url', \esc_url_raw( $avatar_url ) );
+		return self::cache_avatar( $id, $actor_data );
+	}
+
+	/**
+	 * Cache a remote actor's avatar locally.
+	 *
+	 * Downloads the avatar image, optimizes it (resize/WebP), and stores it locally.
+	 *
+	 * @param int                $post_id The actor post ID.
+	 * @param Actor|array|object $actor   The actor object or data array.
+	 *
+	 * @return string|null The cached avatar URL, or null if no avatar.
+	 */
+	private static function cache_avatar( $post_id, $actor ) {
+		$data              = $actor instanceof Actor ? $actor->to_array() : (array) $actor;
+		$remote_avatar_url = object_to_uri( $data['icon'] ?? null );
+
+		if ( empty( $remote_avatar_url ) ) {
+			// No avatar to save, clean up any existing avatar.
+			Attachments::delete_actors_directory( $post_id );
+			\delete_post_meta( $post_id, '_activitypub_avatar_url' );
+			return null;
+		}
+
+		// Download and save the avatar locally.
+		$local_url = Attachments::save_actor_avatar( $post_id, $remote_avatar_url );
+
+		// Store the local URL if caching succeeded, otherwise store the remote URL.
+		$avatar_url = $local_url ?: $remote_avatar_url;
+		\update_post_meta( $post_id, '_activitypub_avatar_url', \esc_url_raw( $avatar_url ) );
 
 		return $avatar_url;
 	}