| @@ -38,8 +38,9 @@ | ||
| 38 | 38 | 'activitypub_reprocess_outbox' => 'hourly', |
| 39 | 39 | 'activitypub_outbox_purge' => 'daily', |
| 40 | 40 | 'activitypub_inbox_purge' => 'daily', |
| 41 | 41 | 'activitypub_ap_post_purge' => 'daily', |
| 42 | + 'activitypub_tombstone_purge' => 'daily', | |
| 42 | 43 | 'activitypub_sync_blocklist_subscriptions' => 'weekly', |
| 43 | 44 | ); |
| 44 | 45 | |
| 45 | 46 | /** |
| @@ -51,17 +52,24 @@ | ||
| 51 | 52 | |
| 52 | 53 | /** |
| 53 | 54 | * Get the pause between async batches (in seconds). |
| 54 | 55 | * |
| 56 | + * @param string|false|null $hook Optional. The async batch hook being scheduled. Default current action. | |
| 57 | + * | |
| 55 | 58 | * @return int The pause in seconds. |
| 56 | 59 | */ |
| 57 | - public static function get_retry_delay() { | |
| 60 | + public static function get_retry_delay( $hook = null ) { | |
| 61 | + if ( null === $hook ) { | |
| 62 | + $hook = \current_action(); | |
| 63 | + } | |
| 64 | + | |
| 58 | 65 | /** |
| 59 | 66 | * Filters the pause between async batches (in seconds). |
| 60 | 67 | * |
| 61 | - * @param int $async_batch_pause The pause in seconds. Default 30. | |
| 68 | + * @param int $async_batch_pause The pause in seconds. Default 30. | |
| 69 | + * @param string|false|null $hook The async batch hook being scheduled. | |
| 62 | 70 | */ |
| 63 | - return apply_filters( 'activitypub_scheduler_async_batch_pause', 30 ); | |
| 71 | + return \apply_filters( 'activitypub_scheduler_async_batch_pause', 30, $hook ); | |
| 64 | 72 | } |
| 65 | 73 | |
| 66 | 74 | /** |
| 67 | 75 | * Initialize the class, registering WordPress hooks. |
| @@ -81,8 +89,9 @@ | ||
| 81 | 89 | \add_action( 'activitypub_reprocess_outbox', array( self::class, 'reprocess_outbox' ) ); |
| 82 | 90 | \add_action( 'activitypub_outbox_purge', array( self::class, 'purge_outbox' ) ); |
| 83 | 91 | \add_action( 'activitypub_inbox_purge', array( self::class, 'purge_inbox' ) ); |
| 84 | 92 | \add_action( 'activitypub_ap_post_purge', array( self::class, 'purge_ap_posts' ) ); |
| 93 | + \add_action( 'activitypub_tombstone_purge', array( self::class, 'purge_tombstones' ) ); | |
| 85 | 94 | \add_action( 'activitypub_inbox_create_item', array( self::class, 'process_inbox_activity' ) ); |
| 86 | 95 | \add_action( 'activitypub_sync_blocklist_subscriptions', array( Blocklist_Subscriptions::class, 'sync_all' ) ); |
| 87 | 96 | |
| 88 | 97 | \add_action( 'post_activitypub_add_to_outbox', array( self::class, 'schedule_outbox_activity_for_federation' ) ); |
| @@ -159,9 +168,9 @@ | ||
| 159 | 168 | */ |
| 160 | 169 | public static function register_schedules() { |
| 161 | 170 | foreach ( self::SCHEDULES as $hook => $recurrence ) { |
| 162 | 171 | if ( ! \wp_next_scheduled( $hook ) ) { |
| 163 | - \wp_schedule_event( time(), $recurrence, $hook ); | |
| 172 | + \wp_schedule_event( \time(), $recurrence, $hook ); | |
| 164 | 173 | } |
| 165 | 174 | } |
| 166 | 175 | |
| 167 | 176 | // Schedule monthly stats collection for the 1st of each month. |
| @@ -183,9 +192,9 @@ | ||
| 183 | 192 | * |
| 184 | 193 | * @return void |
| 185 | 194 | */ |
| 186 | 195 | public static function deregister_schedules() { |
| 187 | - foreach ( array_keys( self::SCHEDULES ) as $hook ) { | |
| 196 | + foreach ( \array_keys( self::SCHEDULES ) as $hook ) { | |
| 188 | 197 | \wp_unschedule_hook( $hook ); |
| 189 | 198 | } |
| 190 | 199 | |
| 191 | 200 | // Statistics schedules. |
| @@ -214,13 +223,13 @@ | ||
| 214 | 223 | $now = \current_time( 'timestamp' ); // phpcs:ignore WordPress.DateTime.CurrentTimeTimestamp.Requested |
| 215 | 224 | $year = (int) \gmdate( 'Y', $now ); |
| 216 | 225 | |
| 217 | 226 | // Get December 1st 3:00 AM for this year. |
| 218 | - $this_year_dec_first = \strtotime( sprintf( '%d-12-01 03:00:00', $year ) ); | |
| 227 | + $this_year_dec_first = \strtotime( \sprintf( '%d-12-01 03:00:00', $year ) ); | |
| 219 | 228 | |
| 220 | 229 | // If we're already past this year's December 1st, schedule for next year. |
| 221 | 230 | if ( $now >= $this_year_dec_first ) { |
| 222 | - return \strtotime( sprintf( '%d-12-01 03:00:00', $year + 1 ) ); | |
| 231 | + return \strtotime( \sprintf( '%d-12-01 03:00:00', $year + 1 ) ); | |
| 223 | 232 | } |
| 224 | 233 | |
| 225 | 234 | return $this_year_dec_first; |
| 226 | 235 | } |
| @@ -230,24 +239,17 @@ | ||
| 230 | 239 | * |
| 231 | 240 | * @param int $outbox_item_id The outbox item ID. |
| 232 | 241 | */ |
| 233 | 242 | public static function unschedule_events_for_item( $outbox_item_id ) { |
| 234 | - $event_args = array( | |
| 235 | - $outbox_item_id, | |
| 236 | - Dispatcher::get_batch_size(), | |
| 237 | - \get_post_meta( $outbox_item_id, '_activitypub_outbox_offset', true ) ?: 0, // phpcs:ignore | |
| 238 | - ); | |
| 239 | - | |
| 240 | 243 | \delete_post_meta( $outbox_item_id, '_activitypub_outbox_offset' ); |
| 241 | 244 | |
| 242 | 245 | $timestamp = \wp_next_scheduled( 'activitypub_process_outbox', array( $outbox_item_id ) ); |
| 243 | 246 | \wp_unschedule_event( $timestamp, 'activitypub_process_outbox', array( $outbox_item_id ) ); |
| 244 | 247 | |
| 245 | - $timestamp = \wp_next_scheduled( 'activitypub_send_activity', $event_args ); | |
| 246 | - \wp_unschedule_event( $timestamp, 'activitypub_send_activity', $event_args ); | |
| 248 | + self::unschedule_outbox_delivery_batches( $outbox_item_id ); | |
| 247 | 249 | |
| 248 | 250 | // Invalidate any retries for this outbox item. |
| 249 | - foreach ( _get_cron_array() as $timestamp => $cron ) { | |
| 251 | + foreach ( \_get_cron_array() as $timestamp => $cron ) { | |
| 250 | 252 | if ( ! isset( $cron['activitypub_retry_activity'] ) ) { |
| 251 | 253 | continue; |
| 252 | 254 | } |
| 253 | 255 | |
| @@ -264,9 +266,9 @@ | ||
| 264 | 266 | */ |
| 265 | 267 | public static function update_remote_actors() { |
| 266 | 268 | $number = 5; |
| 267 | 269 | |
| 268 | - if ( defined( 'DISABLE_WP_CRON' ) && DISABLE_WP_CRON ) { | |
| 270 | + if ( \defined( 'DISABLE_WP_CRON' ) && DISABLE_WP_CRON ) { | |
| 269 | 271 | $number = 50; |
| 270 | 272 | } |
| 271 | 273 | |
| 272 | 274 | /** |
| @@ -273,18 +275,56 @@ | ||
| 273 | 275 | * Filter the number of remote Actors to update. |
| 274 | 276 | * |
| 275 | 277 | * @param int $number The number of remote Actors to update. |
| 276 | 278 | */ |
| 277 | - $number = apply_filters( 'activitypub_update_remote_actors_number', $number ); | |
| 279 | + $number = \apply_filters( 'activitypub_update_remote_actors_number', $number ); | |
| 278 | 280 | $actors = Remote_Actors::get_outdated( $number ); |
| 279 | 281 | |
| 280 | 282 | foreach ( $actors as $actor ) { |
| 281 | - $meta = get_remote_metadata_by_actor( $actor->guid, false ); | |
| 283 | + /* | |
| 284 | + * Use Http::get_remote_object() directly here. | |
| 285 | + * get_remote_metadata_by_actor() short-circuits to the locally | |
| 286 | + * cached ap_actor CPT via Remote_Actors::fetch_by_uri() and never | |
| 287 | + * makes an HTTP request when the actor is already cached, so the | |
| 288 | + * upsert would just rewrite the same stale data and this refresh | |
| 289 | + * would be a no-op. The Update handler documents the same trap. | |
| 290 | + */ | |
| 291 | + $meta = Http::get_remote_object( $actor->guid, false ); | |
| 282 | 292 | |
| 283 | - if ( empty( $meta ) || ! is_array( $meta ) || is_wp_error( $meta ) ) { | |
| 293 | + if ( empty( $meta ) || ! \is_array( $meta ) || \is_wp_error( $meta ) ) { | |
| 284 | 294 | Remote_Actors::add_error( $actor->ID, 'Failed to fetch or parse metadata' ); |
| 285 | 295 | } else { |
| 286 | - $id = Remote_Actors::upsert( $meta ); | |
| 296 | + /* | |
| 297 | + * Only refresh when the remote still reports the same identity. A | |
| 298 | + * different (or missing) id means a Move or a malformed response; | |
| 299 | + * applying it would rewrite the cached guid in place and could | |
| 300 | + * collide with another cached actor, so leave the record alone. | |
| 301 | + * Updating by the known post ID otherwise refreshes it without the | |
| 302 | + * redundant get_by_uri() lookup upsert() would do. | |
| 303 | + */ | |
| 304 | + $fetched_id = isset( $meta['id'] ) && \is_string( $meta['id'] ) ? \esc_url_raw( $meta['id'] ) : ''; | |
| 305 | + if ( $fetched_id !== $actor->guid ) { | |
| 306 | + /* | |
| 307 | + * Bump only the modified date, directly, so the skipped actor | |
| 308 | + * drops out of the outdated queue and is not re-fetched every | |
| 309 | + * run. A direct write avoids the save_post hooks wp_update_post() | |
| 310 | + * fires (which would needlessly clear the cached avatar); the | |
| 311 | + * record is intentionally left unchanged otherwise. | |
| 312 | + */ | |
| 313 | + global $wpdb; | |
| 314 | + $wpdb->update( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching | |
| 315 | + $wpdb->posts, | |
| 316 | + array( | |
| 317 | + 'post_modified' => \current_time( 'mysql' ), | |
| 318 | + 'post_modified_gmt' => \current_time( 'mysql', true ), | |
| 319 | + ), | |
| 320 | + array( 'ID' => $actor->ID ) | |
| 321 | + ); | |
| 322 | + \clean_post_cache( $actor->ID ); | |
| 323 | + continue; | |
| 324 | + } | |
| 325 | + | |
| 326 | + $id = Remote_Actors::update( $actor->ID, $meta ); | |
| 287 | 327 | if ( \is_wp_error( $id ) ) { |
| 288 | 328 | continue; |
| 289 | 329 | } |
| 290 | 330 | Remote_Actors::clear_errors( $id ); |
| @@ -297,9 +337,9 @@ | ||
| 297 | 337 | */ |
| 298 | 338 | public static function cleanup_remote_actors() { |
| 299 | 339 | $number = 5; |
| 300 | 340 | |
| 301 | - if ( defined( 'DISABLE_WP_CRON' ) && DISABLE_WP_CRON ) { | |
| 341 | + if ( \defined( 'DISABLE_WP_CRON' ) && DISABLE_WP_CRON ) { | |
| 302 | 342 | $number = 50; |
| 303 | 343 | } |
| 304 | 344 | |
| 305 | 345 | /** |
| @@ -306,9 +346,9 @@ | ||
| 306 | 346 | * Filter the number of remote Actors to clean up. |
| 307 | 347 | * |
| 308 | 348 | * @param int $number The number of remote Actors to clean up. |
| 309 | 349 | */ |
| 310 | - $number = apply_filters( 'activitypub_cleanup_remote_actors_number', $number ); | |
| 350 | + $number = \apply_filters( 'activitypub_cleanup_remote_actors_number', $number ); | |
| 311 | 351 | $actors = Remote_Actors::get_faulty( $number ); |
| 312 | 352 | |
| 313 | 353 | foreach ( $actors as $actor ) { |
| 314 | 354 | $meta = get_remote_metadata_by_actor( $actor->guid, false ); |
| @@ -314,9 +354,9 @@ | ||
| 314 | 354 | $meta = get_remote_metadata_by_actor( $actor->guid, false ); |
| 315 | 355 | |
| 316 | 356 | if ( Tombstone::exists( $meta ) ) { |
| 317 | 357 | \wp_delete_post( $actor->ID ); |
| 318 | - } elseif ( empty( $meta ) || ! is_array( $meta ) || \is_wp_error( $meta ) ) { | |
| 358 | + } elseif ( empty( $meta ) || ! \is_array( $meta ) || \is_wp_error( $meta ) ) { | |
| 319 | 359 | if ( Remote_Actors::count_errors( $actor->ID ) >= 5 ) { |
| 320 | 360 | \wp_schedule_single_event( \time(), 'activitypub_delete_remote_actor_interactions', array( $actor->guid ) ); |
| 321 | 361 | \wp_schedule_single_event( \time(), 'activitypub_delete_remote_actor_posts', array( $actor->guid ) ); |
| 322 | 362 | \wp_delete_post( $actor->ID ); |
| @@ -343,9 +383,9 @@ | ||
| 343 | 383 | public static function schedule_outbox_activity_for_federation( $id, $offset = 3 ) { |
| 344 | 384 | $hook = 'activitypub_process_outbox'; |
| 345 | 385 | $args = array( $id ); |
| 346 | 386 | |
| 347 | - if ( false === wp_next_scheduled( $hook, $args ) ) { | |
| 387 | + if ( false === \wp_next_scheduled( $hook, $args ) ) { | |
| 348 | 388 | \wp_schedule_single_event( |
| 349 | 389 | \time() + $offset, |
| 350 | 390 | $hook, |
| 351 | 391 | $args |
| @@ -368,9 +408,9 @@ | ||
| 368 | 408 | |
| 369 | 409 | foreach ( $ids as $id ) { |
| 370 | 410 | // Bail if there is a pending batch. |
| 371 | 411 | $offset = \get_post_meta( $id, '_activitypub_outbox_offset', true ) ?: 0; // phpcs:ignore |
| 372 | - if ( \wp_next_scheduled( 'activitypub_send_activity', array( $id, Dispatcher::get_batch_size(), $offset ) ) ) { | |
| 412 | + if ( self::has_scheduled_outbox_delivery_batch( $id, $offset ) ) { | |
| 373 | 413 | return; |
| 374 | 414 | } |
| 375 | 415 | |
| 376 | 416 | // Bail if there is a batch in progress. |
| @@ -404,8 +444,20 @@ | ||
| 404 | 444 | Remote_Posts::purge( \get_option( 'activitypub_ap_post_purge_days', ACTIVITYPUB_AP_POST_PURGE_DAYS ) ); |
| 405 | 445 | } |
| 406 | 446 | |
| 407 | 447 | /** |
| 448 | + * Daily cron handler that purges expired tombstones. | |
| 449 | + * | |
| 450 | + * Retention is non-urgent: large backlogs (e.g. after retention is first enforced) | |
| 451 | + * drain across multiple daily runs. | |
| 452 | + * | |
| 453 | + * @since 8.3.0 | |
| 454 | + */ | |
| 455 | + public static function purge_tombstones() { | |
| 456 | + Tombstone::purge(); | |
| 457 | + } | |
| 458 | + | |
| 459 | + /** | |
| 408 | 460 | * Process cached inbox activity. |
| 409 | 461 | * |
| 410 | 462 | * Retrieves all collected user IDs for an activity and processes them together. |
| 411 | 463 | * |
| @@ -420,9 +472,9 @@ | ||
| 420 | 472 | |
| 421 | 473 | $data = \json_decode( $inbox_item->post_content, true ); |
| 422 | 474 | // Reconstruct activity from inbox post. |
| 423 | 475 | $activity = Activity::init_from_array( $data ); |
| 424 | - $type = \Activitypub\camel_to_snake_case( $activity->get_type() ); | |
| 476 | + $type = camel_to_snake_case( $activity->get_type() ); | |
| 425 | 477 | $context = Inbox::CONTEXT_INBOX; |
| 426 | 478 | $user_ids = Inbox::get_recipients( $inbox_item->ID ); |
| 427 | 479 | |
| 428 | 480 | /** |
| @@ -527,13 +579,76 @@ | ||
| 527 | 579 | self::unlock( $key ); |
| 528 | 580 | |
| 529 | 581 | if ( ! empty( $next ) ) { |
| 530 | 582 | // Schedule the next run, adding the result to the arguments. |
| 531 | - \wp_schedule_single_event( \time() + self::get_retry_delay(), \current_action(), \array_values( $next ) ); | |
| 583 | + \wp_schedule_single_event( \time() + self::get_retry_delay( \current_action() ), \current_action(), \array_values( $next ) ); | |
| 532 | 584 | } |
| 533 | 585 | } |
| 534 | 586 | |
| 535 | 587 | /** |
| 588 | + * Whether an outbox item already has a scheduled delivery batch at an offset. | |
| 589 | + * | |
| 590 | + * @param int $outbox_item_id The outbox item ID. | |
| 591 | + * @param int $offset The delivery offset. | |
| 592 | + * | |
| 593 | + * @return bool True when a matching delivery batch is scheduled. | |
| 594 | + */ | |
| 595 | + private static function has_scheduled_outbox_delivery_batch( $outbox_item_id, $offset ) { | |
| 596 | + return ! empty( self::get_scheduled_outbox_delivery_batches( $outbox_item_id, $offset ) ); | |
| 597 | + } | |
| 598 | + | |
| 599 | + /** | |
| 600 | + * Unschedule all pending delivery batches for an outbox item. | |
| 601 | + * | |
| 602 | + * @param int $outbox_item_id The outbox item ID. | |
| 603 | + */ | |
| 604 | + private static function unschedule_outbox_delivery_batches( $outbox_item_id ) { | |
| 605 | + foreach ( self::get_scheduled_outbox_delivery_batches( $outbox_item_id ) as $event ) { | |
| 606 | + \wp_unschedule_event( $event['timestamp'], 'activitypub_send_activity', $event['args'] ); | |
| 607 | + } | |
| 608 | + } | |
| 609 | + | |
| 610 | + /** | |
| 611 | + * Get scheduled delivery batches for an outbox item. | |
| 612 | + * | |
| 613 | + * The batch size is deliberately ignored because scheduled events may | |
| 614 | + * retain an older value after the admin changes the distribution mode. | |
| 615 | + * | |
| 616 | + * @param int $outbox_item_id The outbox item ID. | |
| 617 | + * @param int|null $offset Optional. Restrict results to this delivery offset. | |
| 618 | + * | |
| 619 | + * @return array Scheduled events with timestamp and args. | |
| 620 | + */ | |
| 621 | + private static function get_scheduled_outbox_delivery_batches( $outbox_item_id, $offset = null ) { | |
| 622 | + $events = array(); | |
| 623 | + | |
| 624 | + foreach ( \_get_cron_array() as $timestamp => $cron ) { | |
| 625 | + if ( empty( $cron['activitypub_send_activity'] ) ) { | |
| 626 | + continue; | |
| 627 | + } | |
| 628 | + | |
| 629 | + foreach ( $cron['activitypub_send_activity'] as $event ) { | |
| 630 | + $args = $event['args'] ?? array(); | |
| 631 | + | |
| 632 | + if ( ! isset( $args[0] ) || (int) $outbox_item_id !== (int) $args[0] ) { | |
| 633 | + continue; | |
| 634 | + } | |
| 635 | + | |
| 636 | + if ( null !== $offset && (int) ( $args[2] ?? 0 ) !== (int) $offset ) { | |
| 637 | + continue; | |
| 638 | + } | |
| 639 | + | |
| 640 | + $events[] = array( | |
| 641 | + 'timestamp' => $timestamp, | |
| 642 | + 'args' => $args, | |
| 643 | + ); | |
| 644 | + } | |
| 645 | + } | |
| 646 | + | |
| 647 | + return $events; | |
| 648 | + } | |
| 649 | + | |
| 650 | + /** | |
| 536 | 651 | * Locks the async batch process for individual callbacks to prevent simultaneous processing. |
| 537 | 652 | * |
| 538 | 653 | * @param string $key Serialized callback name. |
| 539 | 654 | * @return bool|int True if the lock was successful, timestamp of existing lock otherwise. |
| @@ -611,14 +726,14 @@ | ||
| 611 | 726 | if ( 'Create' !== $activity->get_type() ) { |
| 612 | 727 | return; |
| 613 | 728 | } |
| 614 | 729 | |
| 615 | - if ( ! is_object( $activity->get_object() ) ) { | |
| 730 | + if ( ! \is_object( $activity->get_object() ) ) { | |
| 616 | 731 | return; |
| 617 | 732 | } |
| 618 | 733 | |
| 619 | 734 | // Check if the object is an article, image, audio, video, event, or document and ignore profile updates and other activities. |
| 620 | - if ( ! in_array( $activity->get_object()->get_type(), Base_Object::TYPES, true ) ) { | |
| 735 | + if ( ! \in_array( $activity->get_object()->get_type(), Base_Object::TYPES, true ) ) { | |
| 621 | 736 | return; |
| 622 | 737 | } |
| 623 | 738 | |
| 624 | 739 | $announce = new Activity(); |
