| 1 |
<?php |
| 2 |
/** |
| 3 |
* Remote Posts collection file. |
| 4 |
* |
| 5 |
* @package Activitypub |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Activitypub\Collection; |
| 9 |
|
| 10 |
use Activitypub\Emoji; |
| 11 |
use Activitypub\Sanitize; |
| 12 |
|
| 13 |
use function Activitypub\generate_post_summary; |
| 14 |
use function Activitypub\object_to_uri; |
| 15 |
use function Activitypub\process_remote_media; |
| 16 |
|
| 17 |
/** |
| 18 |
* Remote Posts collection. |
| 19 |
* |
| 20 |
* Provides methods to retrieve, create, update, and manage remote |
| 21 |
* ActivityPub posts (articles, notes, media, etc.) received via |
| 22 |
* Server-to-Server (S2S) federation. |
| 23 |
* |
| 24 |
* @see Posts for local posts created via Client-to-Server (C2S) outbox. |
| 25 |
*/ |
| 26 |
class Remote_Posts { |
| 27 |
/** |
| 28 |
* The post type for the posts. |
| 29 |
* |
| 30 |
* @var string |
| 31 |
*/ |
| 32 |
const POST_TYPE = 'ap_post'; |
| 33 |
|
| 34 |
/** |
| 35 |
* Maximum number of remote post items to keep. |
| 36 |
* |
| 37 |
* @var int |
| 38 |
*/ |
| 39 |
const MAX_ITEMS = 5000; |
| 40 |
|
| 41 |
/** |
| 42 |
* Number of items to process per batch during purge. |
| 43 |
* |
| 44 |
* @var int |
| 45 |
*/ |
| 46 |
const PURGE_BATCH_SIZE = 100; |
| 47 |
|
| 48 |
/** |
| 49 |
* Maximum seconds a purge run may take before yielding. |
| 50 |
* |
| 51 |
* @var int |
| 52 |
*/ |
| 53 |
const PURGE_TIMEOUT = 30; |
| 54 |
|
| 55 |
/** |
| 56 |
* Add an object to the collection. |
| 57 |
* |
| 58 |
* @param array $activity The activity object data. |
| 59 |
* @param int|int[] $recipients The id(s) of the local blog-user(s). |
| 60 |
* |
| 61 |
* @return \WP_Post|\WP_Error The object post or WP_Error on failure. |
| 62 |
*/ |
| 63 |
public static function add( $activity, $recipients ) { |
| 64 |
$recipients = (array) $recipients; |
| 65 |
$activity_object = $activity['object']; |
| 66 |
|
| 67 |
$existing = self::get_by_guid( $activity_object['id'] ); |
| 68 |
// If post exists, call update instead. |
| 69 |
if ( ! \is_wp_error( $existing ) ) { |
| 70 |
return self::update( $activity, $recipients ); |
| 71 |
} |
| 72 |
|
| 73 |
// An actor may only create posts attributed to itself; only the actor is signature-bound, not attributedTo. |
| 74 |
if ( object_to_uri( $activity['actor'] ?? '' ) !== object_to_uri( $activity_object['attributedTo'] ?? '' ) ) { |
| 75 |
return new \WP_Error( |
| 76 |
'activitypub_create_unauthorized', |
| 77 |
\__( 'The Create actor does not match the object attributedTo.', 'activitypub' ), |
| 78 |
array( 'status' => 403 ) |
| 79 |
); |
| 80 |
} |
| 81 |
|
| 82 |
// Post doesn't exist, create new post. |
| 83 |
$actor = Remote_Actors::fetch_by_uri( object_to_uri( $activity_object['attributedTo'] ) ); |
| 84 |
|
| 85 |
if ( \is_wp_error( $actor ) ) { |
| 86 |
return $actor; |
| 87 |
} |
| 88 |
|
| 89 |
$post_array = self::activity_to_post( $activity_object ); |
| 90 |
$post_id = \wp_insert_post( $post_array, true ); |
| 91 |
|
| 92 |
if ( \is_wp_error( $post_id ) ) { |
| 93 |
return $post_id; |
| 94 |
} |
| 95 |
|
| 96 |
\add_post_meta( $post_id, '_activitypub_remote_actor_id', $actor->ID ); |
| 97 |
|
| 98 |
// Add recipients as separate meta entries after post is created. |
| 99 |
foreach ( $recipients as $user_id ) { |
| 100 |
self::add_recipient( $post_id, $user_id ); |
| 101 |
} |
| 102 |
|
| 103 |
self::add_taxonomies( $post_id, $activity_object ); |
| 104 |
|
| 105 |
return \get_post( $post_id ); |
| 106 |
} |
| 107 |
|
| 108 |
/** |
| 109 |
* Get an object from the collection. |
| 110 |
* |
| 111 |
* @param int $id The object ID. |
| 112 |
* |
| 113 |
* @return \WP_Post|null The post object or null on failure. |
| 114 |
*/ |
| 115 |
public static function get( $id ) { |
| 116 |
return \get_post( $id ); |
| 117 |
} |
| 118 |
|
| 119 |
/** |
| 120 |
* Get an object by its GUID. |
| 121 |
* |
| 122 |
* @param string $guid The object GUID. |
| 123 |
* |
| 124 |
* @return \WP_Post|\WP_Error The object post or WP_Error on failure. |
| 125 |
*/ |
| 126 |
public static function get_by_guid( $guid ) { |
| 127 |
global $wpdb; |
| 128 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching |
| 129 |
$post_id = $wpdb->get_var( |
| 130 |
$wpdb->prepare( |
| 131 |
"SELECT ID FROM $wpdb->posts WHERE guid=%s AND post_type=%s", |
| 132 |
\esc_url( $guid ), |
| 133 |
self::POST_TYPE |
| 134 |
) |
| 135 |
); |
| 136 |
|
| 137 |
if ( ! $post_id ) { |
| 138 |
return new \WP_Error( |
| 139 |
'activitypub_post_not_found', |
| 140 |
\__( 'Post not found', 'activitypub' ), |
| 141 |
array( 'status' => 404 ) |
| 142 |
); |
| 143 |
} |
| 144 |
|
| 145 |
return \get_post( $post_id ); |
| 146 |
} |
| 147 |
|
| 148 |
/** |
| 149 |
* Update an object in the collection. |
| 150 |
* |
| 151 |
* @param array $activity The activity object data. |
| 152 |
* @param int|int[] $recipients The id(s) of the local blog-user(s). |
| 153 |
* |
| 154 |
* @return \WP_Post|\WP_Error The updated object post or WP_Error on failure. |
| 155 |
*/ |
| 156 |
public static function update( $activity, $recipients ) { |
| 157 |
$recipients = (array) $recipients; |
| 158 |
|
| 159 |
$post = self::get_by_guid( $activity['object']['id'] ); |
| 160 |
if ( \is_wp_error( $post ) ) { |
| 161 |
return $post; |
| 162 |
} |
| 163 |
|
| 164 |
/* |
| 165 |
* Only the post's author may update it. When the activity carries an actor (every |
| 166 |
* signature-verified inbound activity does), compare it against the remote actor |
| 167 |
* stored when the post was first cached (its guid is the actor URI), so a remote |
| 168 |
* server cannot overwrite another host's cached post by sending an Update whose |
| 169 |
* object.id points at a post it does not own. The actor must be used here, not the |
| 170 |
* payload's attributedTo, because only the actor is bound to the HTTP signature. |
| 171 |
*/ |
| 172 |
if ( isset( $activity['actor'] ) ) { |
| 173 |
$owner = \get_post( (int) \get_post_meta( $post->ID, '_activitypub_remote_actor_id', true ) ); |
| 174 |
if ( ! $owner instanceof \WP_Post || object_to_uri( $activity['actor'] ) !== $owner->guid ) { |
| 175 |
return new \WP_Error( |
| 176 |
'activitypub_update_forbidden', |
| 177 |
\__( 'Update failed: the actor does not own this post.', 'activitypub' ), |
| 178 |
array( 'status' => 403 ) |
| 179 |
); |
| 180 |
} |
| 181 |
} |
| 182 |
|
| 183 |
$post_array = self::activity_to_post( $activity['object'] ); |
| 184 |
$post_array['ID'] = $post->ID; |
| 185 |
$post_id = \wp_update_post( $post_array, true ); |
| 186 |
|
| 187 |
if ( \is_wp_error( $post_id ) ) { |
| 188 |
return $post_id; |
| 189 |
} |
| 190 |
|
| 191 |
// Add new recipients using add_recipient (handles deduplication). |
| 192 |
foreach ( $recipients as $user_id ) { |
| 193 |
self::add_recipient( $post_id, $user_id ); |
| 194 |
} |
| 195 |
|
| 196 |
self::add_taxonomies( $post_id, $activity['object'] ); |
| 197 |
|
| 198 |
return \get_post( $post_id ); |
| 199 |
} |
| 200 |
|
| 201 |
/** |
| 202 |
* Delete an object from the collection. |
| 203 |
* |
| 204 |
* @param int $id The object ID. |
| 205 |
* |
| 206 |
* @return \WP_Post|false|null Post data on success, false or null on failure. |
| 207 |
*/ |
| 208 |
public static function delete( $id ) { |
| 209 |
return \wp_delete_post( $id, true ); |
| 210 |
} |
| 211 |
|
| 212 |
/** |
| 213 |
* Delete an object from the collection by its GUID. |
| 214 |
* |
| 215 |
* @param string $guid The object GUID. |
| 216 |
* |
| 217 |
* @return \WP_Post|\WP_Error|false|null Post data on success, false or null on failure, or WP_Error if no post to delete. |
| 218 |
*/ |
| 219 |
public static function delete_by_guid( $guid ) { |
| 220 |
$post = self::get_by_guid( $guid ); |
| 221 |
if ( \is_wp_error( $post ) ) { |
| 222 |
return $post; |
| 223 |
} |
| 224 |
|
| 225 |
return self::delete( $post->ID ); |
| 226 |
} |
| 227 |
|
| 228 |
/** |
| 229 |
* Extract hashtag names from ActivityPub tag array. |
| 230 |
* |
| 231 |
* @param array $tags Array of ActivityPub tags. |
| 232 |
* |
| 233 |
* @return array Array of normalized hashtag names (without # prefix, trimmed, sanitized). |
| 234 |
*/ |
| 235 |
public static function extract_hashtags( $tags ) { |
| 236 |
$hashtags = array(); |
| 237 |
|
| 238 |
if ( empty( $tags ) || ! \is_array( $tags ) ) { |
| 239 |
return $hashtags; |
| 240 |
} |
| 241 |
|
| 242 |
foreach ( $tags as $tag ) { |
| 243 |
if ( isset( $tag['type'] ) && 'Hashtag' === $tag['type'] && isset( $tag['name'] ) ) { |
| 244 |
// Strip # prefix, trim whitespace, and sanitize. |
| 245 |
$normalized = \trim( \ltrim( $tag['name'], '#' ) ); |
| 246 |
$normalized = \wp_strip_all_tags( $normalized ); |
| 247 |
|
| 248 |
if ( ! empty( $normalized ) ) { |
| 249 |
$hashtags[] = $normalized; |
| 250 |
} |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
return $hashtags; |
| 255 |
} |
| 256 |
|
| 257 |
/** |
| 258 |
* Remove hashtags from content. |
| 259 |
* |
| 260 |
* Removes hashtags that appear at the end of the content. |
| 261 |
* Handles both plain text and HTML content, including hashtags within anchor tags. |
| 262 |
* |
| 263 |
* @param string $content The content to process. |
| 264 |
* @param array $tags Array of tag objects from activity (with 'type' and 'name' keys). |
| 265 |
* |
| 266 |
* @return string The content with trailing hashtags removed. |
| 267 |
*/ |
| 268 |
public static function remove_hashtags( $content, $tags ) { |
| 269 |
if ( empty( $content ) || empty( $tags ) || ! \is_array( $tags ) ) { |
| 270 |
return $content; |
| 271 |
} |
| 272 |
|
| 273 |
// Extract and normalize hashtags from tag objects. |
| 274 |
$normalized_tags = self::extract_hashtags( $tags ); |
| 275 |
|
| 276 |
if ( empty( $normalized_tags ) ) { |
| 277 |
return $content; |
| 278 |
} |
| 279 |
|
| 280 |
// Build pattern to match trailing hashtags (at end of content or before closing tags). |
| 281 |
$tag_patterns = array(); |
| 282 |
foreach ( $normalized_tags as $tag ) { |
| 283 |
$escaped_tag = \preg_quote( $tag, '/' ); |
| 284 |
$tag_patterns[] = '(?:<a[^>]*>\s*)?#' . $escaped_tag . '(?=\s|<|$)(?:\s*<\/a>)?'; |
| 285 |
} |
| 286 |
|
| 287 |
/* |
| 288 |
* Pattern explanation: |
| 289 |
* Match one or more hashtags (plain or in anchor tags) at the end of content. |
| 290 |
* The pattern matches trailing hashtags before closing HTML tags or at end of string. |
| 291 |
*/ |
| 292 |
$pattern = '/(?:\s+(?:' . \implode( '|', $tag_patterns ) . '))+(?=\s*(?:<\/[^>]+>)*\s*$)/i'; |
| 293 |
$content = \preg_replace( $pattern, '', $content ); |
| 294 |
|
| 295 |
// Clean up any extra whitespace at end of paragraphs. |
| 296 |
$content = \preg_replace( '/<p>\s*<\/p>/', '', $content ); |
| 297 |
$content = \preg_replace( '/\s+<\/p>/', '</p>', $content ); |
| 298 |
$content = \preg_replace( '/\s+<\/strong>/', '</strong>', $content ); |
| 299 |
|
| 300 |
return \trim( $content ); |
| 301 |
} |
| 302 |
|
| 303 |
/** |
| 304 |
* Convert an activity to a post array. |
| 305 |
* |
| 306 |
* @param array $activity The activity array. |
| 307 |
* |
| 308 |
* @return array|\WP_Error The post array or WP_Error on failure. |
| 309 |
*/ |
| 310 |
private static function activity_to_post( $activity ) { |
| 311 |
if ( ! \is_array( $activity ) ) { |
| 312 |
return new \WP_Error( 'invalid_activity', \__( 'Invalid activity format', 'activitypub' ) ); |
| 313 |
} |
| 314 |
|
| 315 |
$gm_date = \gmdate( 'Y-m-d H:i:s', \strtotime( $activity['published'] ?? 'now' ) ); |
| 316 |
|
| 317 |
// Sanitize content and remove hashtags. |
| 318 |
$content = isset( $activity['content'] ) ? Sanitize::content( $activity['content'] ) : ''; |
| 319 |
$content = self::remove_hashtags( $content, $activity['tag'] ?? array() ); |
| 320 |
$content = Emoji::wrap_in_content( $content, $activity ); |
| 321 |
|
| 322 |
// Process remote media: wrap inline images and append attachments. |
| 323 |
$attachments = self::extract_attachments( $activity ); |
| 324 |
$content = process_remote_media( $content, $attachments ); |
| 325 |
|
| 326 |
return array( |
| 327 |
'post_title' => isset( $activity['name'] ) ? \wp_strip_all_tags( $activity['name'] ) : '', |
| 328 |
'post_content' => $content, |
| 329 |
'post_excerpt' => isset( $activity['summary'] ) ? \wp_strip_all_tags( $activity['summary'] ) : generate_post_summary( $activity['content'] ?? '' ), |
| 330 |
'post_status' => 'publish', |
| 331 |
'post_type' => self::POST_TYPE, |
| 332 |
'post_date_gmt' => $gm_date, |
| 333 |
'post_date' => \get_date_from_gmt( $gm_date ), |
| 334 |
'guid' => isset( $activity['id'] ) ? \esc_url_raw( $activity['id'] ) : '', |
| 335 |
); |
| 336 |
} |
| 337 |
|
| 338 |
/** |
| 339 |
* Add taxonomies to the object post. |
| 340 |
* |
| 341 |
* @param int $post_id The post ID. |
| 342 |
* @param array $activity_object The activity object data. |
| 343 |
*/ |
| 344 |
private static function add_taxonomies( $post_id, $activity_object ) { |
| 345 |
// Save Object Type as Taxonomy item. |
| 346 |
\wp_set_post_terms( $post_id, array( $activity_object['type'] ), 'ap_object_type' ); |
| 347 |
|
| 348 |
// Save the Hashtags as Taxonomy items. |
| 349 |
$tags = self::extract_hashtags( $activity_object['tag'] ?? array() ); |
| 350 |
|
| 351 |
\wp_set_post_terms( $post_id, $tags, 'ap_tag' ); |
| 352 |
} |
| 353 |
|
| 354 |
/** |
| 355 |
* Extract media attachments from an activity object. |
| 356 |
* |
| 357 |
* Extracts attachments with URL, alt text, and media type for appending to content. |
| 358 |
* |
| 359 |
* @param array $activity_object The activity object data. |
| 360 |
* |
| 361 |
* @return array Array of attachments with 'url', 'alt', and 'type' keys. |
| 362 |
*/ |
| 363 |
private static function extract_attachments( $activity_object ) { |
| 364 |
if ( empty( $activity_object['attachment'] ) || ! \is_array( $activity_object['attachment'] ) ) { |
| 365 |
return array(); |
| 366 |
} |
| 367 |
|
| 368 |
$attachments = array(); |
| 369 |
foreach ( $activity_object['attachment'] as $attachment ) { |
| 370 |
if ( \is_object( $attachment ) ) { |
| 371 |
$attachment = \get_object_vars( $attachment ); |
| 372 |
} |
| 373 |
|
| 374 |
if ( empty( $attachment['url'] ) ) { |
| 375 |
continue; |
| 376 |
} |
| 377 |
|
| 378 |
$mime_type = $attachment['mediaType'] ?? ''; |
| 379 |
|
| 380 |
if ( \str_starts_with( $mime_type, 'video/' ) ) { |
| 381 |
$type = 'video'; |
| 382 |
} elseif ( \str_starts_with( $mime_type, 'audio/' ) ) { |
| 383 |
$type = 'audio'; |
| 384 |
} else { |
| 385 |
$type = 'image'; |
| 386 |
} |
| 387 |
|
| 388 |
$attachments[] = array( |
| 389 |
'url' => $attachment['url'], |
| 390 |
'alt' => $attachment['name'] ?? '', |
| 391 |
'type' => $type, |
| 392 |
); |
| 393 |
} |
| 394 |
|
| 395 |
return $attachments; |
| 396 |
} |
| 397 |
|
| 398 |
/** |
| 399 |
* Get posts by remote actor. |
| 400 |
* |
| 401 |
* @param string $actor The remote actor URI. |
| 402 |
* |
| 403 |
* @return array Array of WP_Post objects. |
| 404 |
*/ |
| 405 |
public static function get_by_remote_actor( $actor ) { |
| 406 |
$remote_actor = Remote_Actors::fetch_by_uri( $actor ); |
| 407 |
|
| 408 |
if ( \is_wp_error( $remote_actor ) ) { |
| 409 |
return array(); |
| 410 |
} |
| 411 |
|
| 412 |
return self::get_by_remote_actor_id( $remote_actor->ID ); |
| 413 |
} |
| 414 |
|
| 415 |
/** |
| 416 |
* Get posts by remote actor ID. |
| 417 |
* |
| 418 |
* @param int $actor_id The remote actor post ID. |
| 419 |
* |
| 420 |
* @return array Array of WP_Post objects. |
| 421 |
*/ |
| 422 |
public static function get_by_remote_actor_id( $actor_id ) { |
| 423 |
$query = new \WP_Query( |
| 424 |
array( |
| 425 |
'post_type' => self::POST_TYPE, |
| 426 |
'posts_per_page' => -1, |
| 427 |
// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key |
| 428 |
'meta_key' => '_activitypub_remote_actor_id', |
| 429 |
// phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_value |
| 430 |
'meta_value' => $actor_id, |
| 431 |
) |
| 432 |
); |
| 433 |
|
| 434 |
return $query->posts; |
| 435 |
} |
| 436 |
|
| 437 |
/** |
| 438 |
* Get all recipients for a post. |
| 439 |
* |
| 440 |
* @param int $post_id The post ID. |
| 441 |
* |
| 442 |
* @return int[] Array of user IDs who are recipients. |
| 443 |
*/ |
| 444 |
public static function get_recipients( $post_id ) { |
| 445 |
// Get all meta values with key '_activitypub_user_id' (single => false). |
| 446 |
$recipients = \get_post_meta( $post_id, '_activitypub_user_id', false ); |
| 447 |
$recipients = \array_map( 'intval', $recipients ); |
| 448 |
|
| 449 |
return $recipients; |
| 450 |
} |
| 451 |
|
| 452 |
/** |
| 453 |
* Check if a user is a recipient of a post. |
| 454 |
* |
| 455 |
* @param int $post_id The post ID. |
| 456 |
* @param int $user_id The user ID to check. |
| 457 |
* |
| 458 |
* @return bool True if user is a recipient, false otherwise. |
| 459 |
*/ |
| 460 |
public static function has_recipient( $post_id, $user_id ) { |
| 461 |
$recipients = self::get_recipients( $post_id ); |
| 462 |
|
| 463 |
return \in_array( (int) $user_id, $recipients, true ); |
| 464 |
} |
| 465 |
|
| 466 |
/** |
| 467 |
* Add a recipient to an existing post. |
| 468 |
* |
| 469 |
* @param int $post_id The post ID. |
| 470 |
* @param int $user_id The user ID to add. |
| 471 |
* |
| 472 |
* @return bool True on success, false on failure. |
| 473 |
*/ |
| 474 |
public static function add_recipient( $post_id, $user_id ) { |
| 475 |
$user_id = (int) $user_id; |
| 476 |
// Allow 0 for blog user, but reject negative values. |
| 477 |
if ( $user_id < 0 ) { |
| 478 |
return false; |
| 479 |
} |
| 480 |
|
| 481 |
// Check if already a recipient. |
| 482 |
if ( self::has_recipient( $post_id, $user_id ) ) { |
| 483 |
return true; |
| 484 |
} |
| 485 |
|
| 486 |
// Add new recipient as separate meta entry. |
| 487 |
return (bool) \add_post_meta( $post_id, '_activitypub_user_id', $user_id, false ); |
| 488 |
} |
| 489 |
|
| 490 |
/** |
| 491 |
* Add multiple recipients to an existing post. |
| 492 |
* |
| 493 |
* @param int $post_id The post ID. |
| 494 |
* @param int[] $user_ids The user ID or array of user IDs to add. |
| 495 |
*/ |
| 496 |
public static function add_recipients( $post_id, $user_ids ) { |
| 497 |
foreach ( $user_ids as $user_id ) { |
| 498 |
self::add_recipient( $post_id, $user_id ); |
| 499 |
} |
| 500 |
} |
| 501 |
|
| 502 |
/** |
| 503 |
* Remove a recipient from a post. |
| 504 |
* |
| 505 |
* @param int $post_id The post ID. |
| 506 |
* @param int $user_id The user ID to remove. |
| 507 |
* |
| 508 |
* @return bool True on success, false on failure. |
| 509 |
*/ |
| 510 |
public static function remove_recipient( $post_id, $user_id ) { |
| 511 |
$user_id = (int) $user_id; |
| 512 |
|
| 513 |
// Allow 0 for blog user, but reject negative values. |
| 514 |
if ( $user_id < 0 ) { |
| 515 |
return false; |
| 516 |
} |
| 517 |
|
| 518 |
// Delete the specific meta entry with this value. |
| 519 |
return \delete_post_meta( $post_id, '_activitypub_user_id', $user_id ); |
| 520 |
} |
| 521 |
|
| 522 |
/** |
| 523 |
* Delete all posts. |
| 524 |
* |
| 525 |
* Used during plugin uninstall to clean up all remote posts. |
| 526 |
* |
| 527 |
* @return int The number of posts deleted. |
| 528 |
*/ |
| 529 |
public static function delete_all() { |
| 530 |
$post_ids = \get_posts( |
| 531 |
array( |
| 532 |
'post_type' => self::POST_TYPE, |
| 533 |
'post_status' => array( 'any', 'trash', 'auto-draft' ), |
| 534 |
'fields' => 'ids', |
| 535 |
'numberposts' => -1, |
| 536 |
) |
| 537 |
); |
| 538 |
|
| 539 |
foreach ( $post_ids as $post_id ) { |
| 540 |
\wp_delete_post( $post_id, true ); |
| 541 |
} |
| 542 |
|
| 543 |
return count( $post_ids ); |
| 544 |
} |
| 545 |
|
| 546 |
/** |
| 547 |
* Purge old remote posts. |
| 548 |
* |
| 549 |
* Deletes remote posts older than the specified number of days, |
| 550 |
* but preserves posts that have comments from local users |
| 551 |
* as these indicate meaningful local interactions. |
| 552 |
* |
| 553 |
* @param int $days Number of days to keep items. Items older than this will be deleted. |
| 554 |
* |
| 555 |
* @return int The number of items deleted. |
| 556 |
*/ |
| 557 |
public static function purge( $days ) { |
| 558 |
if ( $days <= 0 ) { |
| 559 |
return 0; |
| 560 |
} |
| 561 |
|
| 562 |
$counts = \wp_count_posts( self::POST_TYPE ); |
| 563 |
$total = 0; |
| 564 |
foreach ( $counts as $count ) { |
| 565 |
$total += (int) $count; |
| 566 |
} |
| 567 |
|
| 568 |
if ( $total <= 200 ) { |
| 569 |
return 0; |
| 570 |
} |
| 571 |
|
| 572 |
global $wpdb; |
| 573 |
|
| 574 |
$deleted = 0; |
| 575 |
$cutoff = \gmdate( 'Y-m-d', \time() - ( $days * DAY_IN_SECONDS ) ); |
| 576 |
$start_time = \time(); |
| 577 |
$exclude = array(); |
| 578 |
|
| 579 |
// If total exceeds the hard cap, drop the date filter to purge oldest items first. |
| 580 |
$overflow = $total > self::MAX_ITEMS; |
| 581 |
$date_query = array( |
| 582 |
array( |
| 583 |
'before' => $cutoff, |
| 584 |
), |
| 585 |
); |
| 586 |
|
| 587 |
$query_args = array( |
| 588 |
'post_type' => self::POST_TYPE, |
| 589 |
'post_status' => 'any', |
| 590 |
'fields' => 'ids', |
| 591 |
'numberposts' => self::PURGE_BATCH_SIZE, |
| 592 |
'orderby' => 'date', |
| 593 |
'order' => 'ASC', |
| 594 |
); |
| 595 |
|
| 596 |
if ( ! $overflow ) { |
| 597 |
$query_args['date_query'] = $date_query; |
| 598 |
} |
| 599 |
|
| 600 |
do { |
| 601 |
$query_args['exclude'] = $exclude; |
| 602 |
$post_ids = \get_posts( $query_args ); |
| 603 |
|
| 604 |
if ( empty( $post_ids ) ) { |
| 605 |
break; |
| 606 |
} |
| 607 |
|
| 608 |
// Batch-fetch post IDs that have local user comments (single query per batch). |
| 609 |
$placeholders = \implode( ',', \array_fill( 0, \count( $post_ids ), '%d' ) ); |
| 610 |
|
| 611 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery |
| 612 |
$commented_post_ids = $wpdb->get_col( |
| 613 |
// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders |
| 614 |
$wpdb->prepare( "SELECT DISTINCT comment_post_ID FROM $wpdb->comments WHERE comment_post_ID IN ($placeholders) AND user_id > 0", $post_ids ) |
| 615 |
); |
| 616 |
$commented_post_ids = \array_flip( $commented_post_ids ); |
| 617 |
|
| 618 |
foreach ( $post_ids as $post_id ) { |
| 619 |
/** |
| 620 |
* Filter whether to preserve a specific ap_post from being purged. |
| 621 |
* |
| 622 |
* @param bool $preserve Whether to preserve this post. Default false. |
| 623 |
* @param int $post_id The ap_post ID being considered for deletion. |
| 624 |
* |
| 625 |
* @return bool Whether to preserve this post from deletion. |
| 626 |
*/ |
| 627 |
if ( \apply_filters( 'activitypub_preserve_ap_post', false, $post_id ) ) { |
| 628 |
$exclude[] = $post_id; |
| 629 |
continue; |
| 630 |
} |
| 631 |
|
| 632 |
// Preserve posts with comments from local users. |
| 633 |
if ( isset( $commented_post_ids[ $post_id ] ) ) { |
| 634 |
$exclude[] = $post_id; |
| 635 |
continue; |
| 636 |
} |
| 637 |
|
| 638 |
\wp_delete_post( $post_id, true ); |
| 639 |
++$deleted; |
| 640 |
} |
| 641 |
|
| 642 |
// Once we're back under the cap, re-apply the date filter. |
| 643 |
if ( $overflow && ( $total - $deleted ) <= self::MAX_ITEMS ) { |
| 644 |
$overflow = false; |
| 645 |
$query_args['date_query'] = $date_query; |
| 646 |
} |
| 647 |
} while ( ! empty( $post_ids ) && ( \time() - $start_time ) < self::PURGE_TIMEOUT ); |
| 648 |
|
| 649 |
return $deleted; |
| 650 |
} |
| 651 |
} |
| 652 |
|
