Refactor avatar handling to use remote actor meta

.github/changelog/2373-from-description

@@ -0,0 +1,4 @@
+Significance: minor
+Type: changed
+
+Refactored avatar handling into a new system that stores and manages avatars per remote actor, improving reliability and preparing for future caching support.

activitypub.php

@@ -70,6 +70,7 @@ function rest_init() {
 function plugin_init() {
 	\add_action( 'init', array( __NAMESPACE__ . '\Activitypub', 'init' ) );
 	\add_action( 'init', array( __NAMESPACE__ . '\Attachments', 'init' ) );
+	\add_action( 'init', array( __NAMESPACE__ . '\Avatars', 'init' ) );
 	\add_action( 'init', array( __NAMESPACE__ . '\Comment', 'init' ) );
 	\add_action( 'init', array( __NAMESPACE__ . '\Dispatcher', 'init' ) );
 	\add_action( 'init', array( __NAMESPACE__ . '\Embed', 'init' ) );

includes/class-activitypub.php

@@ -23,8 +23,6 @@ public static function init() {
 		\add_action( 'init', array( self::class, 'theme_compat' ), 11 );
 		\add_action( 'init', array( self::class, 'register_user_meta' ), 11 );
 
-		\add_filter( 'pre_get_avatar_data', array( self::class, 'pre_get_avatar_data' ), 11, 2 );
-
 		\add_action( 'wp_trash_post', array( self::class, 'trash_post' ), 1 );
 		\add_action( 'untrash_post', array( self::class, 'untrash_post' ), 1 );
 
@@ -90,56 +88,6 @@ public static function uninstall() {
 		Options::delete();
 	}
 
-	/**
-	 * Replaces the default avatar.
-	 *
-	 * @param array             $args        Arguments passed to get_avatar_data(), after processing.
-	 * @param int|string|object $id_or_email A user ID, email address, or comment object.
-	 *
-	 * @return array $args
-	 */
-	public static function pre_get_avatar_data( $args, $id_or_email ) {
-		if (
-			! $id_or_email instanceof \WP_Comment ||
-			! isset( $id_or_email->comment_type ) ||
-			$id_or_email->user_id
-		) {
-			return $args;
-		}
-
-		/**
-		 * Filter allowed comment types for avatars.
-		 *
-		 * @param array $allowed_comment_types Array of allowed comment types.
-		 */
-		$allowed_comment_types = \apply_filters( 'get_avatar_comment_types', array( 'comment' ) );
-		if ( ! \in_array( $id_or_email->comment_type ?: 'comment', $allowed_comment_types, true ) ) { // phpcs:ignore Universal.Operators.DisallowShortTernary
-			return $args;
-		}
-
-		// Check if comment has an avatar.
-		$avatar = \get_comment_meta( $id_or_email->comment_ID, 'avatar_url', true );
-
-		if ( $avatar ) {
-			if ( empty( $args['class'] ) ) {
-				$args['class'] = array();
-			} elseif ( \is_string( $args['class'] ) ) {
-				$args['class'] = \explode( ' ', $args['class'] );
-			}
-
-			/** This filter is documented in wp-includes/link-template.php */
-			$args['url']     = \apply_filters( 'get_avatar_url', $avatar, $id_or_email, $args );
-			$args['class'][] = 'avatar';
-			$args['class'][] = 'avatar-activitypub';
-			$args['class'][] = 'avatar-' . (int) $args['size'];
-			$args['class'][] = 'photo';
-			$args['class'][] = 'u-photo';
-			$args['class']   = \array_unique( $args['class'] );
-		}
-
-		return $args;
-	}
-
 	/**
 	 * Store permalink in meta, to send delete Activity.
 	 *

includes/class-avatars.php

@@ -0,0 +1,82 @@
+<?php
+/**
+ * Avatars class file.
+ *
+ * @package Activitypub
+ */
+
+namespace Activitypub;
+
+use Activitypub\Collection\Remote_Actors;
+
+/**
+ * ActivityPub Avatars class.
+ */
+class Avatars {
+	/**
+	 * Initialize the class, registering WordPress hooks.
+	 */
+	public static function init() {
+		\add_filter( 'pre_get_avatar_data', array( self::class, 'pre_get_avatar_data' ), 11, 2 );
+	}
+
+	/**
+	 * Replaces the default avatar.
+	 *
+	 * @param array             $args        Arguments passed to get_avatar_data(), after processing.
+	 * @param int|string|object $id_or_email A user ID, email address, or comment object.
+	 *
+	 * @return array $args
+	 */
+	public static function pre_get_avatar_data( $args, $id_or_email ) {
+		if (
+			! $id_or_email instanceof \WP_Comment ||
+			! isset( $id_or_email->comment_type ) ||
+			$id_or_email->user_id
+		) {
+			return $args;
+		}
+
+		/**
+		 * Filter allowed comment types for avatars.
+		 *
+		 * @param array $allowed_comment_types Array of allowed comment types.
+		 */
+		$allowed_comment_types = \apply_filters( 'get_avatar_comment_types', array( 'comment' ) );
+		if ( ! \in_array( $id_or_email->comment_type ?: 'comment', $allowed_comment_types, true ) ) { // phpcs:ignore Universal.Operators.DisallowShortTernary
+			return $args;
+		}
+
+		$avatar = null;
+
+		// First, try to get avatar from remote actor.
+		$remote_actor_id = \get_comment_meta( $id_or_email->comment_ID, '_activitypub_remote_actor_id', true );
+		if ( $remote_actor_id ) {
+			$avatar = Remote_Actors::get_avatar_url( $remote_actor_id );
+		}
+
+		// Fall back to avatar_url comment meta for backward compatibility.
+		if ( ! $avatar ) {
+			$avatar = \get_comment_meta( $id_or_email->comment_ID, 'avatar_url', true );
+		}
+
+		if ( $avatar ) {
+			if ( empty( $args['class'] ) ) {
+				$args['class'] = array();
+			} elseif ( \is_string( $args['class'] ) ) {
+				$args['class'] = \explode( ' ', $args['class'] );
+			}
+
+			/** This filter is documented in wp-includes/link-template.php */
+			$args['url']     = \apply_filters( 'get_avatar_url', $avatar, $id_or_email, $args );
+			$args['class'][] = 'avatar';
+			$args['class'][] = 'avatar-activitypub';
+			$args['class'][] = 'avatar-' . (int) $args['size'];
+			$args['class'][] = 'photo';
+			$args['class'][] = 'u-photo';
+			$args['class']   = \array_unique( $args['class'] );
+		}
+
+		return $args;
+	}
+}

includes/class-migration.php

@@ -31,6 +31,7 @@ public static function init() {
 		Scheduler::register_async_batch_callback( 'activitypub_update_comment_counts', array( self::class, 'update_comment_counts' ) );
 		Scheduler::register_async_batch_callback( 'activitypub_create_post_outbox_items', array( self::class, 'create_post_outbox_items' ) );
 		Scheduler::register_async_batch_callback( 'activitypub_create_comment_outbox_items', array( self::class, 'create_comment_outbox_items' ) );
+		Scheduler::register_async_batch_callback( 'activitypub_migrate_avatar_to_remote_actors', array( self::class, 'migrate_avatar_to_remote_actors' ) );
 	}
 
 	/**
@@ -211,6 +212,7 @@ public static function maybe_migrate() {
 
 		if ( \version_compare( $version_from_db, 'unreleased', '<' ) ) {
 			self::clean_up_inbox();
+			\wp_schedule_single_event( \time(), 'activitypub_migrate_avatar_to_remote_actors' );
 		}
 
 		// Ensure all required cron schedules are registered.
@@ -1079,4 +1081,78 @@ private static function clean_up_inbox() {
 			\wp_delete_post( $post_id, true );
 		}
 	}
+
+	/**
+	 * Migrate avatar URLs from comment meta to remote actors in batches.
+	 *
+	 * This migration:
+	 * 1. Finds all comments with ActivityPub protocol and avatar_url meta
+	 * 2. Looks up the remote actor by comment_author_url
+	 * 3. Adds _activitypub_remote_actor_id to comment meta
+	 * 4. Stores avatar_url in remote actor post meta
+	 *
+	 * Note: We don't use offset because as we add _activitypub_remote_actor_id,
+	 * comments are filtered out of the query. We just keep fetching the next
+	 * batch until no more comments match the criteria.
+	 *
+	 * @param int $batch_size Optional. Number of comments to process per batch. Default 50.
+	 * @return array|null Array with batch size if there are more comments to process, null otherwise.
+	 */
+	public static function migrate_avatar_to_remote_actors( $batch_size = 50 ) {
+		global $wpdb;
+
+		/*
+		 * Get comments with avatar_url meta that don't have _activitypub_remote_actor_id yet.
+		 * Uses conditional aggregation to reduce JOINs from 3 to 1, improving query performance.
+		 * Filters meta_key before GROUP BY to reduce rows processed during aggregation.
+		 * No offset needed - as we process comments, they're filtered out by the HAVING clause.
+		 */
+		$comments = $wpdb->get_results( // phpcs:ignore WordPress.DB.DirectDatabaseQuery
+			$wpdb->prepare(
+				"SELECT c.comment_ID, c.comment_author_url,
+					MAX(CASE WHEN cm.meta_key = 'avatar_url' THEN cm.meta_value END) AS avatar_url,
+					MAX(CASE WHEN cm.meta_key = 'protocol' THEN cm.meta_value END) AS protocol,
+					MAX(CASE WHEN cm.meta_key = '_activitypub_remote_actor_id' THEN cm.meta_value END) AS remote_actor_id
+				FROM {$wpdb->comments} c
+				INNER JOIN {$wpdb->commentmeta} cm ON c.comment_ID = cm.comment_id
+				WHERE cm.meta_key IN ('avatar_url', 'protocol', '_activitypub_remote_actor_id')
+				GROUP BY c.comment_ID, c.comment_author_url
+				HAVING protocol = 'activitypub'
+					AND avatar_url IS NOT NULL
+					AND (remote_actor_id IS NULL OR remote_actor_id = '')
+				LIMIT %d",
+				$batch_size
+			)
+		);
+
+		foreach ( $comments as $comment ) {
+			if ( empty( $comment->comment_author_url ) ) {
+				continue;
+			}
+
+			// Try to get the remote actor by URI.
+			$remote_actor = Remote_Actors::fetch_by_uri( $comment->comment_author_url );
+
+			// If we have a valid remote actor, store the reference.
+			if ( ! \is_wp_error( $remote_actor ) ) {
+				// Add _activitypub_remote_actor_id to comment meta.
+				\add_comment_meta( $comment->comment_ID, '_activitypub_remote_actor_id', $remote_actor->ID, true );
+
+				// Ensure avatar is stored on remote actor if not already present.
+				$existing_avatar = \get_post_meta( $remote_actor->ID, '_activitypub_avatar_url', true );
+				if ( empty( $existing_avatar ) && ! empty( $comment->avatar_url ) ) {
+					\update_post_meta( $remote_actor->ID, '_activitypub_avatar_url', \esc_url_raw( $comment->avatar_url ) );
+				}
+			}
+		}
+
+		// Return batch info if there are more comments to process.
+		if ( count( $comments ) === $batch_size ) {
+			return array(
+				'batch_size' => $batch_size,
+			);
+		}
+
+		return null;
+	}
 }

includes/collection/class-interactions.php

@@ -203,7 +203,37 @@ public static function get_by_actor( $actor ) {
 			),
 		);
 
-		return get_comments( $args );
+		return \get_comments( $args );
+	}
+
+	/**
+	 * Get interaction(s) by remote actor ID.
+	 *
+	 * This is an optimized query that uses the remote actor post ID directly
+	 * instead of querying by author_url.
+	 *
+	 * @param int $remote_actor_id The remote actor post ID.
+	 *
+	 * @return array The interactions as WP_Comment objects.
+	 */
+	public static function get_by_remote_actor_id( $remote_actor_id ) {
+		$args = array(
+			'nopaging'   => true,
+			// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query
+			'meta_query' => array(
+				'relation' => 'AND',
+				array(
+					'key'   => 'protocol',
+					'value' => 'activitypub',
+				),
+				array(
+					'key'   => '_activitypub_remote_actor_id',
+					'value' => $remote_actor_id,
+				),
+			),
+		);
+
+		return \get_comments( $args );
 	}
 
 	/**
@@ -305,8 +335,13 @@ public static function activity_to_comment( $activity ) {
 			),
 		);
 
-		if ( isset( $actor['icon']['url'] ) ) {
-			$comment_data['comment_meta']['avatar_url'] = \esc_url_raw( $actor['icon']['url'] );
+		// Store reference to remote actor post.
+		$actor_uri = object_to_uri( $activity['actor'] ?? null );
+		if ( $actor_uri ) {
+			$remote_actor = Remote_Actors::get_by_uri( $actor_uri );
+			if ( ! \is_wp_error( $remote_actor ) ) {
+				$comment_data['comment_meta']['_activitypub_remote_actor_id'] = $remote_actor->ID;
+			}
 		}
 
 		if ( isset( $activity['object']['url'] ) ) {

includes/collection/class-posts.php

@@ -221,18 +221,30 @@ private static function add_taxonomies( $post_id, $activity_object ) {
 	 */
 	public static function get_by_remote_actor( $actor ) {
 		$remote_actor = Remote_Actors::fetch_by_uri( $actor );
+
 		if ( \is_wp_error( $remote_actor ) ) {
 			return array();
 		}
 
+		return self::get_by_remote_actor_id( $remote_actor->ID );
+	}
+
+	/**
+	 * Get posts by remote actor ID.
+	 *
+	 * @param int $actor_id The remote actor post ID.
+	 *
+	 * @return array Array of WP_Post objects.
+	 */
+	public static function get_by_remote_actor_id( $actor_id ) {
 		$query = new \WP_Query(
 			array(
 				'post_type'      => self::POST_TYPE,
 				'posts_per_page' => -1,
 				// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key
 				'meta_key'       => '_activitypub_remote_actor_id',
 				// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_value
-				'meta_value'     => $remote_actor->ID,
+				'meta_value'     => $actor_id,
 			)
 		);