| 1 |
<?php |
| 2 |
/** |
| 3 |
* Tombstone class file. |
| 4 |
* |
| 5 |
* @package Activitypub |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Activitypub; |
| 9 |
|
| 10 |
use Activitypub\Activity\Base_Object; |
| 11 |
|
| 12 |
/** |
| 13 |
* ActivityPub Tombstone Class. |
| 14 |
* |
| 15 |
* Handles detection and management of tombstoned (deleted) ActivityPub resources. |
| 16 |
* A tombstone in ActivityPub represents a deleted object that was previously available. |
| 17 |
* This class provides methods to detect tombstones across various data formats including |
| 18 |
* URLs, ActivityPub objects, arrays, and WordPress error responses. |
| 19 |
* |
| 20 |
* @see https://www.w3.org/TR/activitystreams-vocabulary/#dfn-tombstone |
| 21 |
*/ |
| 22 |
class Tombstone { |
| 23 |
/** |
| 24 |
* HTTP status codes that indicate a tombstoned resource. |
| 25 |
* |
| 26 |
* - 404: Not Found - Resource no longer exists |
| 27 |
* - 410: Gone - Resource was intentionally removed |
| 28 |
* |
| 29 |
* @var int[] Array of HTTP status codes indicating tombstones. |
| 30 |
*/ |
| 31 |
private static $codes = array( 404, 410 ); |
| 32 |
|
| 33 |
/** |
| 34 |
* The custom post type used to store local tombstones. |
| 35 |
* |
| 36 |
* @var string |
| 37 |
*/ |
| 38 |
const POST_TYPE = 'ap_tombstone'; |
| 39 |
|
| 40 |
/** |
| 41 |
* Check if a tombstone exists for the given resource. |
| 42 |
* |
| 43 |
* This is the main entry point for tombstone detection. It accepts various |
| 44 |
* data types and routes them to the appropriate checking method: |
| 45 |
* - URLs (string): Checks remote or local tombstone status |
| 46 |
* - WP_Error objects: Checks for tombstone-indicating HTTP status codes |
| 47 |
* - Arrays: Checks for ActivityPub Tombstone type |
| 48 |
* - Objects: Checks for ActivityPub Tombstone type or Base_Object instances |
| 49 |
* |
| 50 |
* @param string|\WP_Error|array|object $various The resource data to check for tombstone status. |
| 51 |
* Can be a URL, error object, ActivityPub array, or object. |
| 52 |
* |
| 53 |
* @return bool True if the resource is tombstoned, false otherwise. |
| 54 |
*/ |
| 55 |
public static function exists( $various ) { |
| 56 |
if ( \is_wp_error( $various ) ) { |
| 57 |
return self::exists_in_error( $various ); |
| 58 |
} |
| 59 |
|
| 60 |
if ( \is_string( $various ) ) { |
| 61 |
if ( is_same_domain( $various ) ) { |
| 62 |
return self::exists_local( $various ); |
| 63 |
} |
| 64 |
return self::exists_remote( $various ); |
| 65 |
} |
| 66 |
|
| 67 |
if ( \is_array( $various ) ) { |
| 68 |
return self::check_array( $various ); |
| 69 |
} |
| 70 |
|
| 71 |
if ( \is_object( $various ) ) { |
| 72 |
return self::check_object( $various ); |
| 73 |
} |
| 74 |
|
| 75 |
return false; |
| 76 |
} |
| 77 |
|
| 78 |
/** |
| 79 |
* Check if a remote URL is tombstoned. |
| 80 |
* |
| 81 |
* Makes an HTTP request to the remote URL with ActivityPub headers |
| 82 |
* and checks for tombstone indicators: |
| 83 |
* - HTTP 404/410 status codes |
| 84 |
* - ActivityPub Tombstone object type in response body |
| 85 |
* |
| 86 |
* @param string $url The remote URL to check for tombstone status. |
| 87 |
* |
| 88 |
* @return bool True if the remote URL is tombstoned, false otherwise. |
| 89 |
*/ |
| 90 |
public static function exists_remote( $url ) { |
| 91 |
/** |
| 92 |
* Fires before checking if the URL is a tombstone. |
| 93 |
* |
| 94 |
* @param string $url The URL to check. |
| 95 |
*/ |
| 96 |
\do_action( 'activitypub_pre_http_is_tombstone', $url ); |
| 97 |
|
| 98 |
$response = Http::get( $url ); |
| 99 |
|
| 100 |
if ( ! \is_wp_error( $response ) ) { |
| 101 |
$data = \wp_remote_retrieve_body( $response ); |
| 102 |
$data = \json_decode( $data, true ); |
| 103 |
|
| 104 |
return self::check_array( $data ); |
| 105 |
} |
| 106 |
|
| 107 |
if ( \in_array( (int) $response->get_error_code(), self::$codes, true ) ) { |
| 108 |
return true; |
| 109 |
} |
| 110 |
|
| 111 |
return false; |
| 112 |
} |
| 113 |
|
| 114 |
/** |
| 115 |
* Check if a local URL is tombstoned. |
| 116 |
* |
| 117 |
* Matches by the MD5 hash of the normalized URL stored in `post_name`. |
| 118 |
* Falls back to the legacy `activitypub_tombstone_urls` option for |
| 119 |
* tombstones that have not yet been migrated. |
| 120 |
* |
| 121 |
* @param string $url The local URL to check for tombstone status. |
| 122 |
* |
| 123 |
* @return bool True if the local URL is tombstoned, false otherwise. |
| 124 |
*/ |
| 125 |
public static function exists_local( $url ) { |
| 126 |
if ( ! \is_string( $url ) || '' === $url ) { |
| 127 |
return false; |
| 128 |
} |
| 129 |
|
| 130 |
$normalized = normalize_url( $url ); |
| 131 |
|
| 132 |
if ( ! empty( self::find_post_ids_by_url( $normalized ) ) ) { |
| 133 |
return true; |
| 134 |
} |
| 135 |
|
| 136 |
/* |
| 137 |
* Fallback to the legacy option during migration. Once the option is |
| 138 |
* deleted (migration complete), get_option returns false and the |
| 139 |
* is_array() guard short-circuits immediately. |
| 140 |
*/ |
| 141 |
$legacy = \get_option( 'activitypub_tombstone_urls', false ); |
| 142 |
if ( \is_array( $legacy ) && \in_array( $normalized, $legacy, true ) ) { |
| 143 |
return true; |
| 144 |
} |
| 145 |
|
| 146 |
return false; |
| 147 |
} |
| 148 |
|
| 149 |
/** |
| 150 |
* Check if a WP_Error object indicates a tombstoned resource. |
| 151 |
* |
| 152 |
* Examines the error data for HTTP status codes that indicate tombstones. |
| 153 |
* This is typically used when HTTP requests return error responses. |
| 154 |
* |
| 155 |
* @param \WP_Error $wp_error The WordPress error object to examine. |
| 156 |
* |
| 157 |
* @return bool True if the error indicates a tombstoned resource, false otherwise. |
| 158 |
*/ |
| 159 |
public static function exists_in_error( $wp_error ) { |
| 160 |
if ( ! \is_wp_error( $wp_error ) ) { |
| 161 |
return false; |
| 162 |
} |
| 163 |
|
| 164 |
$data = $wp_error->get_error_data(); |
| 165 |
if ( isset( $data['status'] ) && \in_array( (int) $data['status'], self::$codes, true ) ) { |
| 166 |
return true; |
| 167 |
} |
| 168 |
|
| 169 |
return false; |
| 170 |
} |
| 171 |
|
| 172 |
/** |
| 173 |
* Check if an array represents an ActivityPub Tombstone object. |
| 174 |
* |
| 175 |
* Examines the array for the ActivityPub 'type' property set to 'Tombstone'. |
| 176 |
* This follows the ActivityStreams specification for tombstone objects. |
| 177 |
* |
| 178 |
* @param array|mixed $data The array data to check. Non-arrays return false. |
| 179 |
* |
| 180 |
* @return bool True if the array represents a Tombstone object, false otherwise. |
| 181 |
*/ |
| 182 |
private static function check_array( $data ) { |
| 183 |
if ( ! \is_array( $data ) ) { |
| 184 |
return false; |
| 185 |
} |
| 186 |
|
| 187 |
if ( isset( $data['type'] ) && 'Tombstone' === $data['type'] ) { |
| 188 |
return true; |
| 189 |
} |
| 190 |
|
| 191 |
return false; |
| 192 |
} |
| 193 |
|
| 194 |
/** |
| 195 |
* Check if an object represents an ActivityPub Tombstone. |
| 196 |
* |
| 197 |
* Checks for tombstone indicators in objects: |
| 198 |
* - Standard objects: 'type' property set to 'Tombstone' |
| 199 |
* - Base_Object instances: Uses get_type() method to check for 'Tombstone' |
| 200 |
* |
| 201 |
* @param object|mixed $data The object data to check. Non-objects return false. |
| 202 |
* |
| 203 |
* @return bool True if the object represents a Tombstone, false otherwise. |
| 204 |
*/ |
| 205 |
private static function check_object( $data ) { |
| 206 |
if ( ! \is_object( $data ) ) { |
| 207 |
return false; |
| 208 |
} |
| 209 |
|
| 210 |
if ( isset( $data->type ) && 'Tombstone' === $data->type ) { |
| 211 |
return true; |
| 212 |
} |
| 213 |
|
| 214 |
if ( $data instanceof Base_Object && 'Tombstone' === $data->get_type() ) { |
| 215 |
return true; |
| 216 |
} |
| 217 |
|
| 218 |
return false; |
| 219 |
} |
| 220 |
|
| 221 |
/** |
| 222 |
* Look up tombstone post IDs by canonical URL. |
| 223 |
* |
| 224 |
* The MD5 of the normalized URL is unique per URL, so a successful |
| 225 |
* `bury()` produces exactly one row and the canonical lookup is enough. |
| 226 |
* |
| 227 |
* @since 8.3.0 |
| 228 |
* |
| 229 |
* @param string $normalized The normalized URL (scheme stripped). |
| 230 |
* @return int[] Post IDs (zero or one entry under normal operation). |
| 231 |
*/ |
| 232 |
private static function find_post_ids_by_url( $normalized ) { |
| 233 |
global $wpdb; |
| 234 |
|
| 235 |
/* |
| 236 |
* `bury()` is idempotent on the MD5 slug, so a successful insert |
| 237 |
* produces exactly one row per URL. `LIMIT 1` matches that invariant |
| 238 |
* and keeps the query cheap on the hot `exists_local()` path. |
| 239 |
*/ |
| 240 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery |
| 241 |
$ids = $wpdb->get_col( |
| 242 |
$wpdb->prepare( |
| 243 |
"SELECT ID FROM {$wpdb->posts} WHERE post_type = %s AND post_name = %s LIMIT 1", |
| 244 |
self::POST_TYPE, |
| 245 |
\md5( $normalized ) |
| 246 |
) |
| 247 |
); |
| 248 |
|
| 249 |
return \array_map( 'intval', $ids ); |
| 250 |
} |
| 251 |
|
| 252 |
/** |
| 253 |
* Add one or more URLs to the local tombstone registry. |
| 254 |
* |
| 255 |
* "Buries" URLs by adding them to the local tombstone URL registry. |
| 256 |
* URLs are normalized before storage; duplicate calls for the same URL |
| 257 |
* are a no-op because the `post_name` slug is the MD5 of the |
| 258 |
* normalized URL. |
| 259 |
* |
| 260 |
* @param string ...$urls The URLs to add to the tombstone registry. |
| 261 |
*/ |
| 262 |
public static function bury( ...$urls ) { |
| 263 |
foreach ( $urls as $url ) { |
| 264 |
if ( ! \filter_var( $url, \FILTER_VALIDATE_URL ) ) { |
| 265 |
continue; |
| 266 |
} |
| 267 |
|
| 268 |
$normalized = normalize_url( $url ); |
| 269 |
|
| 270 |
if ( ! empty( self::find_post_ids_by_url( $normalized ) ) ) { |
| 271 |
continue; |
| 272 |
} |
| 273 |
|
| 274 |
/* |
| 275 |
* Store the original URL in `guid` so it is human-readable and |
| 276 |
* survives `esc_url()` without scheme mangling. The hash slug |
| 277 |
* in `post_name` is what we actually key lookups on. |
| 278 |
*/ |
| 279 |
$post_id = \wp_insert_post( |
| 280 |
array( |
| 281 |
'post_type' => self::POST_TYPE, |
| 282 |
'post_status' => 'publish', |
| 283 |
'post_name' => \md5( $normalized ), |
| 284 |
'guid' => $url, |
| 285 |
'post_author' => 0, |
| 286 |
), |
| 287 |
true |
| 288 |
); |
| 289 |
|
| 290 |
if ( \is_wp_error( $post_id ) || ! $post_id ) { |
| 291 |
/** |
| 292 |
* Fires when `bury()` fails to write a tombstone row. |
| 293 |
* |
| 294 |
* The URL is silently not tombstoned in this case — the |
| 295 |
* request path will respond as it would for any other |
| 296 |
* non-existent post. Useful as a monitoring hook. |
| 297 |
* |
| 298 |
* @since 8.3.0 |
| 299 |
* |
| 300 |
* @param string $normalized The normalized URL that failed to bury. |
| 301 |
* @param \WP_Error|int|null $post_id The `wp_insert_post()` return value. |
| 302 |
*/ |
| 303 |
\do_action( 'activitypub_tombstone_bury_failed', $normalized, $post_id ); |
| 304 |
} |
| 305 |
} |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* Remove one or more URLs from the local tombstone registry. |
| 310 |
* |
| 311 |
* Removes URLs from the local tombstone URL registry. |
| 312 |
* URLs are normalized before comparison to ensure consistent matching. |
| 313 |
* This marks the URLs as no longer tombstoned for future local checks. |
| 314 |
* |
| 315 |
* @param string ...$urls The URLs to remove from the tombstone registry. |
| 316 |
*/ |
| 317 |
public static function remove( ...$urls ) { |
| 318 |
$normalized_urls = array(); |
| 319 |
foreach ( $urls as $url ) { |
| 320 |
if ( \filter_var( $url, \FILTER_VALIDATE_URL ) ) { |
| 321 |
$normalized_urls[] = normalize_url( $url ); |
| 322 |
} |
| 323 |
} |
| 324 |
|
| 325 |
if ( empty( $normalized_urls ) ) { |
| 326 |
return; |
| 327 |
} |
| 328 |
|
| 329 |
$normalized_urls = \array_values( \array_unique( $normalized_urls ) ); |
| 330 |
|
| 331 |
foreach ( $normalized_urls as $normalized ) { |
| 332 |
foreach ( self::find_post_ids_by_url( $normalized ) as $post_id ) { |
| 333 |
\wp_delete_post( $post_id, true ); |
| 334 |
} |
| 335 |
} |
| 336 |
|
| 337 |
$legacy = \get_option( 'activitypub_tombstone_urls', false ); |
| 338 |
if ( ! \is_array( $legacy ) ) { |
| 339 |
return; |
| 340 |
} |
| 341 |
|
| 342 |
$filtered = \array_values( \array_diff( $legacy, $normalized_urls ) ); |
| 343 |
if ( \count( $filtered ) === \count( $legacy ) ) { |
| 344 |
return; |
| 345 |
} |
| 346 |
|
| 347 |
if ( empty( $filtered ) ) { |
| 348 |
\delete_option( 'activitypub_tombstone_urls' ); |
| 349 |
} else { |
| 350 |
\update_option( 'activitypub_tombstone_urls', $filtered ); |
| 351 |
} |
| 352 |
} |
| 353 |
|
| 354 |
/** |
| 355 |
* Delete every tombstone post and the legacy option. |
| 356 |
* |
| 357 |
* Used during plugin uninstall to clean up all local tombstones. |
| 358 |
* |
| 359 |
* @since 8.3.0 |
| 360 |
* |
| 361 |
* @return int The number of tombstone posts deleted. |
| 362 |
*/ |
| 363 |
public static function delete_all() { |
| 364 |
global $wpdb; |
| 365 |
|
| 366 |
$post_ids = \array_map( |
| 367 |
'intval', |
| 368 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery |
| 369 |
$wpdb->get_col( |
| 370 |
$wpdb->prepare( |
| 371 |
"SELECT ID FROM {$wpdb->posts} WHERE post_type = %s", |
| 372 |
self::POST_TYPE |
| 373 |
) |
| 374 |
) |
| 375 |
); |
| 376 |
|
| 377 |
$deleted = 0; |
| 378 |
foreach ( $post_ids as $post_id ) { |
| 379 |
if ( \wp_delete_post( $post_id, true ) ) { |
| 380 |
++$deleted; |
| 381 |
} |
| 382 |
} |
| 383 |
|
| 384 |
\delete_option( 'activitypub_tombstone_urls' ); |
| 385 |
|
| 386 |
return $deleted; |
| 387 |
} |
| 388 |
|
| 389 |
/** |
| 390 |
* Delete tombstones older than the retention window. |
| 391 |
* |
| 392 |
* Processes up to `$batch_size` tombstones per call. Retention is |
| 393 |
* non-urgent: large backlogs drain across multiple daily runs of the |
| 394 |
* `activitypub_tombstone_purge` cron event. |
| 395 |
* |
| 396 |
* @since 8.3.0 |
| 397 |
* |
| 398 |
* @param int $batch_size Max number of tombstones to delete per call. |
| 399 |
* @return int The number of tombstones deleted. |
| 400 |
*/ |
| 401 |
public static function purge( $batch_size = 200 ) { |
| 402 |
/** |
| 403 |
* Filters the retention window for local tombstones, in days. |
| 404 |
* |
| 405 |
* Set to 0 or a negative value to disable automatic purge. |
| 406 |
* |
| 407 |
* @since 8.3.0 |
| 408 |
* |
| 409 |
* @param int $days Retention window in days. Default 90. |
| 410 |
*/ |
| 411 |
$days = (int) \apply_filters( 'activitypub_tombstone_retention_days', 90 ); |
| 412 |
|
| 413 |
if ( $days <= 0 ) { |
| 414 |
return 0; |
| 415 |
} |
| 416 |
|
| 417 |
$cutoff = \gmdate( 'Y-m-d H:i:s', \time() - $days * DAY_IN_SECONDS ); |
| 418 |
|
| 419 |
$ids = \get_posts( |
| 420 |
array( |
| 421 |
'post_type' => self::POST_TYPE, |
| 422 |
'post_status' => 'publish', |
| 423 |
'posts_per_page' => (int) $batch_size, |
| 424 |
'fields' => 'ids', |
| 425 |
'orderby' => 'date', |
| 426 |
'order' => 'ASC', |
| 427 |
'no_found_rows' => true, |
| 428 |
'date_query' => array( |
| 429 |
array( |
| 430 |
'column' => 'post_date_gmt', |
| 431 |
'before' => $cutoff, |
| 432 |
), |
| 433 |
), |
| 434 |
) |
| 435 |
); |
| 436 |
|
| 437 |
$deleted = 0; |
| 438 |
foreach ( $ids as $id ) { |
| 439 |
if ( \wp_delete_post( (int) $id, true ) ) { |
| 440 |
++$deleted; |
| 441 |
} |
| 442 |
} |
| 443 |
|
| 444 |
return $deleted; |
| 445 |
} |
| 446 |
} |
| 447 |
|
