| @@ -251,8 +251,33 @@ | ||
| 251 | 251 | * @var int |
| 252 | 252 | */ |
| 253 | 253 | private const TERM_WALK_CHUNK = 1000; |
| 254 | 254 | |
| 255 | + /** | |
| 256 | + * Exception code for an automatic rebuild stopped short of the memory limit. | |
| 257 | + * | |
| 258 | + * @since 2.10.1 | |
| 259 | + * @var int | |
| 260 | + */ | |
| 261 | + private const MEMORY_ABORT_CODE = 4290; | |
| 262 | + | |
| 263 | + /** | |
| 264 | + * Whether the chunked walks should stop before the memory limit. Only on | |
| 265 | + * for automatic rebuilds, whose failure is recorded and retried. | |
| 266 | + * | |
| 267 | + * @since 2.10.1 | |
| 268 | + * @var bool | |
| 269 | + */ | |
| 270 | + private bool $memory_guard = false; | |
| 271 | + | |
| 272 | + /** | |
| 273 | + * Largest memory cost of one walked chunk in this rebuild, in bytes. | |
| 274 | + * | |
| 275 | + * @since 2.10.1 | |
| 276 | + * @var int | |
| 277 | + */ | |
| 278 | + private int $walk_chunk_cost = 0; | |
| 279 | + | |
| 255 | 280 | private array $sitemap_types = [ |
| 256 | 281 | 'posts' => [ |
| 257 | 282 | 'name' => 'Posts', |
| 258 | 283 | 'post_types' => ['post'], |
| @@ -279,25 +304,29 @@ | ||
| 279 | 304 | ] |
| 280 | 305 | ]; |
| 281 | 306 | |
| 282 | 307 | /** |
| 308 | + * The generator the content-change listeners share, built on the first | |
| 309 | + * change of a request. | |
| 310 | + * | |
| 311 | + * @since 2.10.1 | |
| 312 | + * @var self|null | |
| 313 | + */ | |
| 314 | + private static ?self $listener = null; | |
| 315 | + | |
| 316 | + /** | |
| 283 | 317 | * Constructor |
| 284 | 318 | * |
| 285 | 319 | * @since 1.0.0 |
| 320 | + * @since 2.10.1 Registers no hooks, whatever `$register_hooks` says. The | |
| 321 | + * content-change listeners are registered once, at bootstrap, | |
| 322 | + * by register_content_listeners(). | |
| 286 | 323 | * |
| 287 | - * @param bool $register_hooks Optional. Whether to register the auto-generation | |
| 288 | - * hooks. Pass false for a read-only instance built | |
| 289 | - * solely to query settings — the hooks are bound to | |
| 290 | - * `$this`, so a second hook-registering instance | |
| 291 | - * would run `handle_content_change()` twice per save. | |
| 324 | + * @param bool $register_hooks Unused since 2.10.1; kept so existing callers, | |
| 325 | + * Pro's included, keep working. | |
| 292 | 326 | */ |
| 293 | 327 | public function __construct(bool $register_hooks = true) { |
| 294 | 328 | parent::__construct('sitemap'); |
| 295 | - | |
| 296 | - // Initialize auto-generation hooks | |
| 297 | - if ($register_hooks) { | |
| 298 | - $this->init_auto_generation_hooks(); | |
| 299 | - } | |
| 300 | 329 | } |
| 301 | 330 | |
| 302 | 331 | /** |
| 303 | 332 | * Filter the args of a sitemap post query. |
| @@ -341,34 +370,154 @@ | ||
| 341 | 370 | return (array) apply_filters('thinkrank_sitemap_term_query_args', $args); |
| 342 | 371 | } |
| 343 | 372 | |
| 344 | 373 | /** |
| 345 | - * Initialize WordPress hooks for auto-generation | |
| 374 | + * Register the content-change listeners that queue an automatic rebuild. | |
| 346 | 375 | * |
| 347 | - * @since 1.0.0 | |
| 376 | + * Called once per request, at plugin bootstrap, next to the WP-Cron | |
| 377 | + * listeners that run the rebuild these queue (both in | |
| 378 | + * Plugin::register_sitemap_cron_listeners(), on plugins_loaded). The | |
| 379 | + * constructor used to register them, so they existed only in a request | |
| 380 | + * that happened to build a generator: the REST endpoint, the setup wizard, | |
| 381 | + * an MCP ability. Block-editor saves go through REST and were heard. A | |
| 382 | + * scheduled post published by WP-Cron, Quick Edit, the classic editor and | |
| 383 | + * WP-CLI were not, so the post stayed out of the sitemap with nothing | |
| 384 | + * pending to recover it (#824). Each instance also added its own set, so | |
| 385 | + * one REST save ran the handlers once per generator built. | |
| 386 | + * | |
| 387 | + * The generator is built on the first change and shared for the rest of | |
| 388 | + * the request, so a bulk edit does not construct one per post. | |
| 389 | + * | |
| 390 | + * @since 2.10.1 | |
| 348 | 391 | * @return void |
| 349 | 392 | */ |
| 350 | - private function init_auto_generation_hooks(): void { | |
| 351 | - // Content change hooks - use priority 20 to run after other plugins | |
| 352 | - add_action('save_post', [$this, 'handle_content_change'], 20, 2); | |
| 353 | - add_action('delete_post', [$this, 'handle_content_deletion'], 20); | |
| 354 | - add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20); | |
| 355 | - add_action('untrash_post', [$this, 'handle_content_change_by_id'], 20); | |
| 393 | + public static function register_content_listeners(): void { | |
| 394 | + // Priority 20, to run after other plugins. | |
| 395 | + add_action('save_post', static function (int $post_id, \WP_Post $post): void { | |
| 396 | + self::listener()->handle_content_change($post_id, $post); | |
| 397 | + }, 20, 2); | |
| 398 | + add_action('delete_post', static function (int $post_id): void { | |
| 399 | + self::listener()->handle_content_deletion($post_id); | |
| 400 | + }, 20); | |
| 401 | + add_action('wp_trash_post', static function (int $post_id): void { | |
| 402 | + self::listener()->handle_content_deletion($post_id); | |
| 403 | + }, 20); | |
| 404 | + add_action('untrash_post', static function (int $post_id): void { | |
| 405 | + self::listener()->handle_content_change_by_id($post_id); | |
| 406 | + }, 20); | |
| 356 | 407 | |
| 357 | - // Taxonomy change hooks | |
| 358 | - add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3); | |
| 359 | - add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3); | |
| 360 | - add_action('delete_term', [$this, 'handle_taxonomy_change'], 20, 3); | |
| 408 | + foreach (['created_term', 'edited_term', 'delete_term'] as $hook) { | |
| 409 | + add_action($hook, static function (int $term_id, int $tt_id, string $taxonomy): void { | |
| 410 | + self::listener()->handle_taxonomy_change($term_id, $tt_id, $taxonomy); | |
| 411 | + }, 20, 3); | |
| 412 | + } | |
| 361 | 413 | |
| 362 | - // NOTE: the WP-Cron regeneration listeners (thinkrank_regenerate_sitemap | |
| 363 | - // and thinkrank_regenerate_sitemap_settings) are registered at plugin | |
| 364 | - // bootstrap (Plugin::register_sitemap_cron_listeners(), on plugins_loaded) | |
| 365 | - // rather than here. A cron run never builds this class via the REST | |
| 366 | - // endpoint (no rest_api_init), so registering them in the constructor | |
| 367 | - // would leave the scheduled events with no listener at cron time. | |
| 414 | + // The sitemap leaves out what the robots tag noindexes and what | |
| 415 | + // ThinkRank redirects (#911), so a change to either decides what the | |
| 416 | + // published files should list. Neither is a post or term save, and | |
| 417 | + // without these the files kept the old answer until unrelated content | |
| 418 | + // changed. | |
| 419 | + foreach (self::ROBOTS_SETTINGS_OPTIONS as $option) { | |
| 420 | + add_action('update_option_' . $option, static function ($old_value, $value): void { | |
| 421 | + self::handle_robots_settings_change($old_value, $value); | |
| 422 | + }, 20, 2); | |
| 423 | + add_action('add_option_' . $option, static function ($name, $value): void { | |
| 424 | + self::handle_robots_settings_change([], $value); | |
| 425 | + }, 20, 2); | |
| 426 | + } | |
| 427 | + | |
| 428 | + add_action('thinkrank_object_redirect_saved', static function (): void { | |
| 429 | + self::listener()->schedule_regeneration(); | |
| 430 | + }, 20, 0); | |
| 368 | 431 | } |
| 369 | 432 | |
| 370 | 433 | /** |
| 434 | + * Options holding robots directives the sitemap's item filter reads. | |
| 435 | + * | |
| 436 | + * @since 2.15.0 | |
| 437 | + * @var string[] | |
| 438 | + */ | |
| 439 | + private const ROBOTS_SETTINGS_OPTIONS = [ | |
| 440 | + 'thinkrank_global_seo_settings', | |
| 441 | + 'thinkrank_global_robot_meta_settings', | |
| 442 | + ]; | |
| 443 | + | |
| 444 | + /** | |
| 445 | + * Queue a rebuild when a saved robots setting changes a noindex decision. | |
| 446 | + * | |
| 447 | + * Both options carry far more than robots directives (titles, schema, | |
| 448 | + * feature switches), so only a change to a noindex outcome rebuilds. | |
| 449 | + * | |
| 450 | + * @since 2.15.0 | |
| 451 | + * | |
| 452 | + * @param mixed $old_value Previous option value. | |
| 453 | + * @param mixed $value New option value. | |
| 454 | + * @return void | |
| 455 | + */ | |
| 456 | + public static function handle_robots_settings_change($old_value, $value): void { | |
| 457 | + if (self::noindex_fingerprint($old_value) === self::noindex_fingerprint($value)) { | |
| 458 | + return; | |
| 459 | + } | |
| 460 | + | |
| 461 | + // The matrix memoises the option for the request, and a rebuild that | |
| 462 | + // runs in this request must read the value just saved. | |
| 463 | + Content_Type_Settings::flush_cache(); | |
| 464 | + | |
| 465 | + self::listener()->schedule_regeneration(); | |
| 466 | + } | |
| 467 | + | |
| 468 | + /** | |
| 469 | + * The noindex decisions a robots option makes, keyed by what they apply to. | |
| 470 | + * | |
| 471 | + * Reads both shapes: the flat site-wide directives, and the per-entity | |
| 472 | + * rows, where a row's directives apply only while its robots switch is on. | |
| 473 | + * A row that is on but stores no `noindex` key changes nothing, matching | |
| 474 | + * the array_merge() the robots tag does. | |
| 475 | + * | |
| 476 | + * @since 2.15.0 | |
| 477 | + * | |
| 478 | + * @param mixed $value Option value. | |
| 479 | + * @return array<string, bool|null> | |
| 480 | + */ | |
| 481 | + private static function noindex_fingerprint($value): array { | |
| 482 | + if (!is_array($value)) { | |
| 483 | + return []; | |
| 484 | + } | |
| 485 | + | |
| 486 | + $decisions = []; | |
| 487 | + | |
| 488 | + if (array_key_exists('noindex', $value) && !is_array($value['noindex'])) { | |
| 489 | + $decisions['*'] = !empty($value['noindex']); | |
| 490 | + } | |
| 491 | + | |
| 492 | + foreach ($value as $key => $row) { | |
| 493 | + if (!is_array($row)) { | |
| 494 | + continue; | |
| 495 | + } | |
| 496 | + | |
| 497 | + $robots = $row['robots_meta'] ?? null; | |
| 498 | + | |
| 499 | + $decisions[(string) $key] = !empty($row['robots_meta_enabled']) && is_array($robots) && array_key_exists('noindex', $robots) | |
| 500 | + ? !empty($robots['noindex']) | |
| 501 | + : null; | |
| 502 | + } | |
| 503 | + | |
| 504 | + ksort($decisions); | |
| 505 | + | |
| 506 | + return $decisions; | |
| 507 | + } | |
| 508 | + | |
| 509 | + /** | |
| 510 | + * The generator the content-change listeners share. | |
| 511 | + * | |
| 512 | + * @since 2.10.1 | |
| 513 | + * @return self | |
| 514 | + */ | |
| 515 | + private static function listener(): self { | |
| 516 | + return self::$listener ??= new self(false); | |
| 517 | + } | |
| 518 | + | |
| 519 | + /** | |
| 371 | 520 | * Generate XML sitemap |
| 372 | 521 | * |
| 373 | 522 | * @since 1.0.0 |
| 374 | 523 | * |
| @@ -751,9 +900,12 @@ | ||
| 751 | 900 | |
| 752 | 901 | // Add image entries if provided |
| 753 | 902 | foreach ($images as $image) { |
| 754 | 903 | $xml .= " <image:image>\n"; |
| 755 | - $xml .= " <image:loc>" . esc_url(Url_Scheme::apply((string) $image['url'])) . "</image:loc>\n"; | |
| 904 | + // The sitemap protocol wants an escaped URL, and WordPress hands back | |
| 905 | + // attachment URLs with non-ASCII filenames unencoded (esc_url() does | |
| 906 | + // not encode them either), so write the percent-encoded form (#924). | |
| 907 | + $xml .= " <image:loc>" . esc_url(\ThinkRank\Core\Url_Validator::to_ascii(Url_Scheme::apply((string) $image['url']))) . "</image:loc>\n"; | |
| 756 | 908 | |
| 757 | 909 | if (!empty($image['title'])) { |
| 758 | 910 | $xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n"; |
| 759 | 911 | } |
| @@ -874,8 +1026,11 @@ | ||
| 874 | 1026 | // well-bounded routine with no ceiling (#402). |
| 875 | 1027 | $total = count($all_ids); |
| 876 | 1028 | |
| 877 | 1029 | for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) { |
| 1030 | + $this->assert_memory_headroom(); | |
| 1031 | + $chunk_start = memory_get_usage(true); | |
| 1032 | + | |
| 878 | 1033 | $chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK); |
| 879 | 1034 | |
| 880 | 1035 | $posts = get_posts($this->filter_query_args([ |
| 881 | 1036 | 'post_type' => $post_types, |
| @@ -884,8 +1039,12 @@ | ||
| 884 | 1039 | 'post__in' => $chunk, |
| 885 | 1040 | 'orderby' => 'post__in', // preserve the resolved order |
| 886 | 1041 | ])); |
| 887 | 1042 | |
| 1043 | + // get_featured_image() hydrates each featured image under the | |
| 1044 | + // attachment's own ID, which the chunk's post IDs do not reach. | |
| 1045 | + $attachment_ids = []; | |
| 1046 | + | |
| 888 | 1047 | foreach ($posts as $post) { |
| 889 | 1048 | if ($this->should_include_in_sitemap($post, $settings)) { |
| 890 | 1049 | /** |
| 891 | 1050 | * Filter a sitemap entry's permalink. |
| @@ -906,14 +1065,25 @@ | ||
| 906 | 1065 | $priority = $this->calculate_intelligent_priority($post, $post->post_type); |
| 907 | 1066 | $changefreq = $this->calculate_change_frequency($post, $post->post_type); |
| 908 | 1067 | $images = $this->extract_post_images($post, $settings); |
| 909 | 1068 | |
| 1069 | + $thumbnail_id = (int) get_post_thumbnail_id($post); | |
| 1070 | + if ($thumbnail_id > 0) { | |
| 1071 | + $attachment_ids[] = $thumbnail_id; | |
| 1072 | + } | |
| 1073 | + | |
| 910 | 1074 | yield $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images); |
| 911 | 1075 | } |
| 912 | 1076 | } |
| 913 | 1077 | |
| 914 | - // Free the hydrated chunk before loading the next one. | |
| 1078 | + // Free the hydrated chunk before loading the next one — including | |
| 1079 | + // the copies get_posts() left in the runtime object cache. | |
| 915 | 1080 | unset($posts); |
| 1081 | + $this->release_walk_memory( | |
| 1082 | + $chunk_start, | |
| 1083 | + $this->chunk_post_cache_groups($post_types), | |
| 1084 | + array_merge($chunk, $attachment_ids) | |
| 1085 | + ); | |
| 916 | 1086 | } |
| 917 | 1087 | } |
| 918 | 1088 | |
| 919 | 1089 | /** |
| @@ -973,8 +1143,11 @@ | ||
| 973 | 1143 | // Same moving window as the post walk above, for the same reason. |
| 974 | 1144 | $total = count($all_ids); |
| 975 | 1145 | |
| 976 | 1146 | for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) { |
| 1147 | + $this->assert_memory_headroom(); | |
| 1148 | + $chunk_start = memory_get_usage(true); | |
| 1149 | + | |
| 977 | 1150 | $chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK); |
| 978 | 1151 | |
| 979 | 1152 | $terms = get_terms($this->filter_term_query_args([ |
| 980 | 1153 | 'taxonomy' => $taxonomy, |
| @@ -987,12 +1160,12 @@ | ||
| 987 | 1160 | continue; |
| 988 | 1161 | } |
| 989 | 1162 | |
| 990 | 1163 | foreach ($terms as $term) { |
| 991 | - // A term the user marked noindex must not be advertised in the | |
| 992 | - // sitemap: the robots tag now honours term meta, so listing it | |
| 993 | - // here would have the sitemap contradict the page's own tag. | |
| 994 | - if ($this->term_is_noindexed((int) $term->term_id)) { | |
| 1164 | + // A term whose archive says noindex (its own override or its | |
| 1165 | + // taxonomy's), or that redirects, must not be advertised: the | |
| 1166 | + // sitemap would contradict the page's own signal (#911). | |
| 1167 | + if (!Indexability::is_indexable_term($term)) { | |
| 995 | 1168 | continue; |
| 996 | 1169 | } |
| 997 | 1170 | |
| 998 | 1171 | $url = get_term_link($term); |
| @@ -1003,8 +1176,9 @@ | ||
| 1003 | 1176 | } |
| 1004 | 1177 | } |
| 1005 | 1178 | |
| 1006 | 1179 | unset($terms); |
| 1180 | + $this->release_walk_memory($chunk_start, ['terms', 'term_meta'], $chunk); | |
| 1007 | 1181 | } |
| 1008 | 1182 | } |
| 1009 | 1183 | |
| 1010 | 1184 | /** |
| @@ -1179,9 +1353,9 @@ | ||
| 1179 | 1353 | if (preg_match('/src=["\']([^"\']+)["\']/', $img_tag, $src_match)) { |
| 1180 | 1354 | $image_url = $src_match[1]; |
| 1181 | 1355 | |
| 1182 | 1356 | // Skip if not a valid URL or external image |
| 1183 | - if (!filter_var($image_url, FILTER_VALIDATE_URL)) { | |
| 1357 | + if (!\ThinkRank\Core\Url_Validator::is_valid($image_url)) { | |
| 1184 | 1358 | continue; |
| 1185 | 1359 | } |
| 1186 | 1360 | |
| 1187 | 1361 | // Extract title and alt attributes |
| @@ -1296,14 +1470,15 @@ | ||
| 1296 | 1470 | if (is_wp_error($all_terms)) { |
| 1297 | 1471 | return []; |
| 1298 | 1472 | } |
| 1299 | 1473 | |
| 1300 | - // Drop terms the user marked noindex. This path feeds the single general | |
| 1301 | - // sitemap while collect_taxonomy_entries_iter() feeds the segmented ones, | |
| 1302 | - // so both need the filter or the two disagree about the same term. | |
| 1474 | + // Drop terms that are not indexable destinations. This path feeds the | |
| 1475 | + // single general sitemap while collect_taxonomy_entries_iter() feeds | |
| 1476 | + // the segmented ones, so both need the filter or the two disagree about | |
| 1477 | + // the same term. | |
| 1303 | 1478 | $all_terms = array_values(array_filter( |
| 1304 | 1479 | $all_terms, |
| 1305 | - fn($term) => !$this->term_is_noindexed((int) $term->term_id) | |
| 1480 | + static fn($term) => $term instanceof \WP_Term && Indexability::is_indexable_term($term) | |
| 1306 | 1481 | )); |
| 1307 | 1482 | |
| 1308 | 1483 | // Group terms by taxonomy |
| 1309 | 1484 | return $this->group_terms_by_taxonomy($all_terms); |
| @@ -1485,17 +1660,16 @@ | ||
| 1485 | 1660 | if (!in_array($post->post_status, ['publish', 'private'], true)) { |
| 1486 | 1661 | return false; |
| 1487 | 1662 | } |
| 1488 | 1663 | |
| 1489 | - // Check if post overrides robots and sets noindex. | |
| 1490 | - if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) { | |
| 1491 | - $raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true); | |
| 1492 | - if (is_string($raw) && $raw !== '') { | |
| 1493 | - $robots = json_decode($raw, true); | |
| 1494 | - if (is_array($robots) && !empty($robots['noindex'])) { | |
| 1495 | - return false; | |
| 1496 | - } | |
| 1497 | - } | |
| 1664 | + // A URL whose page says noindex, or that ThinkRank redirects, is not a | |
| 1665 | + // destination. Only the per-post noindex used to be read here, so a | |
| 1666 | + // post type set to No-index still had every item listed, and redirected | |
| 1667 | + // posts were submitted as "Page with redirect" (#911). The password | |
| 1668 | + // check stays with the setting above: listing protected posts is a | |
| 1669 | + // choice this sitemap has always offered. | |
| 1670 | + if (Indexability::is_post_noindexed($post) || Indexability::is_post_redirected($post)) { | |
| 1671 | + return false; | |
| 1498 | 1672 | } |
| 1499 | 1673 | |
| 1500 | 1674 | return true; |
| 1501 | 1675 | } |
| @@ -1534,35 +1708,8 @@ | ||
| 1534 | 1708 | return $ids; |
| 1535 | 1709 | } |
| 1536 | 1710 | |
| 1537 | 1711 | /** |
| 1538 | - * Whether a term carries an explicit noindex override. | |
| 1539 | - * | |
| 1540 | - * Mirrors the post-side check in should_include_post(); terms store the same | |
| 1541 | - * `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written | |
| 1542 | - * by the update-term-seo ability and by the SEO importer. | |
| 1543 | - * | |
| 1544 | - * @since 1.31.0 | |
| 1545 | - * | |
| 1546 | - * @param int $term_id Term to test. | |
| 1547 | - * @return bool True when the term is marked noindex. | |
| 1548 | - */ | |
| 1549 | - private function term_is_noindexed(int $term_id): bool { | |
| 1550 | - if (!(bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true)) { | |
| 1551 | - return false; | |
| 1552 | - } | |
| 1553 | - | |
| 1554 | - $raw = get_term_meta($term_id, '_thinkrank_robots_meta', true); | |
| 1555 | - if (!is_string($raw) || $raw === '') { | |
| 1556 | - return false; | |
| 1557 | - } | |
| 1558 | - | |
| 1559 | - $robots = json_decode($raw, true); | |
| 1560 | - | |
| 1561 | - return is_array($robots) && !empty($robots['noindex']); | |
| 1562 | - } | |
| 1563 | - | |
| 1564 | - /** | |
| 1565 | 1712 | * Count total URLs in sitemap |
| 1566 | 1713 | * |
| 1567 | 1714 | * @since 1.0.0 |
| 1568 | 1715 | * |
| @@ -1854,11 +2001,13 @@ | ||
| 1854 | 2001 | $since = !empty($pending['since']) ? (int) $pending['since'] : time(); |
| 1855 | 2002 | $current = isset($pending['source']) ? (string) $pending['source'] : ''; |
| 1856 | 2003 | $source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content'; |
| 1857 | 2004 | |
| 2005 | + // Merged so the bookkeeping a rebuild keeps on the marker (`started`, | |
| 2006 | + // `memory_limit`) survives an edit made while it is outstanding. | |
| 1858 | 2007 | update_option( |
| 1859 | 2008 | self::REGENERATION_PENDING_OPTION, |
| 1860 | - [ | |
| 2009 | + array_merge($pending, [ | |
| 1861 | 2010 | 'since' => $since, |
| 1862 | 2011 | 'source' => $source, |
| 1863 | 2012 | 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 0, |
| 1864 | 2013 | 'next_attempt' => !empty($pending['next_attempt']) |
| @@ -1866,9 +2015,9 @@ | ||
| 1866 | 2015 | : time() + self::REGENERATION_TAKEOVER_GRACE, |
| 1867 | 2016 | // Bumped on every change so a rebuild can tell whether the edit |
| 1868 | 2017 | // it started for is still the newest one outstanding. |
| 1869 | 2018 | 'revision' => (!empty($pending['revision']) ? (int) $pending['revision'] : 0) + 1, |
| 1870 | - ], | |
| 2019 | + ]), | |
| 1871 | 2020 | true |
| 1872 | 2021 | ); |
| 1873 | 2022 | } |
| 1874 | 2023 | |
| @@ -1914,8 +2063,18 @@ | ||
| 1914 | 2063 | $revision !== null |
| 1915 | 2064 | && is_array($pending) |
| 1916 | 2065 | && (int) ($pending['revision'] ?? 0) !== $revision |
| 1917 | 2066 | ) { |
| 2067 | + // This attempt succeeded, so drop what it claimed: the newer change | |
| 2068 | + // waits the grace a fresh edit gets, not a failure backoff it never | |
| 2069 | + // earned. Not zero: that edit queued its own debounced event, and | |
| 2070 | + // a marker due at once had the next admin request rebuild in its | |
| 2071 | + // shutdown and the event rebuild again seconds later. | |
| 2072 | + unset($pending['started'], $pending['memory_limit']); | |
| 2073 | + $pending['attempts'] = 0; | |
| 2074 | + $pending['next_attempt'] = time() + self::REGENERATION_TAKEOVER_GRACE; | |
| 2075 | + | |
| 2076 | + update_option(self::REGENERATION_PENDING_OPTION, $pending, true); | |
| 1918 | 2077 | return; |
| 1919 | 2078 | } |
| 1920 | 2079 | |
| 1921 | 2080 | delete_option(self::REGENERATION_PENDING_OPTION); |
| @@ -1921,8 +2080,69 @@ | ||
| 1921 | 2080 | delete_option(self::REGENERATION_PENDING_OPTION); |
| 1922 | 2081 | } |
| 1923 | 2082 | |
| 1924 | 2083 | /** |
| 2084 | + * Record the attempt that is about to run before it runs. | |
| 2085 | + * | |
| 2086 | + * A PHP fatal — the memory limit or max_execution_time — is not a | |
| 2087 | + * Throwable, so no catch or finally around the generation runs when the | |
| 2088 | + * process dies, and a failure recorded afterwards was never recorded at | |
| 2089 | + * all: `attempts` stayed 0, the backoff never applied, and the next request | |
| 2090 | + * started the same doomed rebuild again (a fatal every few minutes for as | |
| 2091 | + * long as an admin was logged in). Claiming the attempt up front makes the | |
| 2092 | + * backoff hold even when nothing after this line gets to run, and leaves a | |
| 2093 | + * `started` stamp the next attempt can recognise as an interrupted one. | |
| 2094 | + * | |
| 2095 | + * @since 2.10.1 | |
| 2096 | + * @return void | |
| 2097 | + */ | |
| 2098 | + private function claim_regeneration_attempt(): void { | |
| 2099 | + $pending = get_option(self::REGENERATION_PENDING_OPTION, null); | |
| 2100 | + | |
| 2101 | + // Nothing outstanding (e.g. a manual generation already satisfied it): | |
| 2102 | + // there is no marker to retry from, so nothing to claim. | |
| 2103 | + if (!is_array($pending) || empty($pending['since'])) { | |
| 2104 | + return; | |
| 2105 | + } | |
| 2106 | + | |
| 2107 | + if (!empty($pending['started'])) { | |
| 2108 | + // The previous attempt claimed itself and never reported back. | |
| 2109 | + update_option( | |
| 2110 | + self::REGENERATION_ERROR_OPTION, | |
| 2111 | + [ | |
| 2112 | + 'message' => __('The previous automatic sitemap rebuild stopped before it finished, most likely because PHP ran out of memory or time. It is retried with a growing delay. If this keeps happening, raise the PHP memory_limit or max_execution_time, or run WP-Cron from a system cron.', 'thinkrank'), | |
| 2113 | + 'source' => isset($pending['source']) ? (string) $pending['source'] : 'content', | |
| 2114 | + 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 1, | |
| 2115 | + 'time' => (int) $pending['started'], | |
| 2116 | + ], | |
| 2117 | + false | |
| 2118 | + ); | |
| 2119 | + } | |
| 2120 | + | |
| 2121 | + $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1; | |
| 2122 | + | |
| 2123 | + $pending['attempts'] = $attempts; | |
| 2124 | + $pending['next_attempt'] = time() + $this->regeneration_backoff($attempts); | |
| 2125 | + $pending['started'] = time(); | |
| 2126 | + | |
| 2127 | + update_option(self::REGENERATION_PENDING_OPTION, $pending, true); | |
| 2128 | + } | |
| 2129 | + | |
| 2130 | + /** | |
| 2131 | + * Delay before the next takeover after the given number of attempts. | |
| 2132 | + * | |
| 2133 | + * @since 2.10.1 | |
| 2134 | + * @param int $attempts Attempts made so far (1 or more). | |
| 2135 | + * @return int Seconds. | |
| 2136 | + */ | |
| 2137 | + private function regeneration_backoff(int $attempts): int { | |
| 2138 | + return (int) min( | |
| 2139 | + self::REGENERATION_TAKEOVER_GRACE * (2 ** min(max($attempts, 1), 10)), | |
| 2140 | + self::REGENERATION_MAX_BACKOFF | |
| 2141 | + ); | |
| 2142 | + } | |
| 2143 | + | |
| 2144 | + /** | |
| 1925 | 2145 | * Record a failed regeneration instead of discarding it. |
| 1926 | 2146 | * |
| 1927 | 2147 | * Keeps the pending marker in place so the rebuild is retried, but backs the |
| 1928 | 2148 | * next attempt off exponentially (capped) so a persistently failing |
| @@ -1928,22 +2148,30 @@ | ||
| 1928 | 2148 | * next attempt off exponentially (capped) so a persistently failing |
| 1929 | 2149 | * generation cannot run on every admin request. |
| 1930 | 2150 | * |
| 1931 | 2151 | * @since 2.2.1 |
| 1932 | - * @param string $message Failure detail. | |
| 1933 | - * @param string $source Either 'content' or 'settings'. | |
| 2152 | + * @since 2.10.1 Accepts the memory limit a rebuild had to stop short of, and | |
| 2153 | + * does not count an attempt claim_regeneration_attempt() | |
| 2154 | + * already counted. | |
| 2155 | + * @param string $message Failure detail. | |
| 2156 | + * @param string $source Either 'content' or 'settings'. | |
| 2157 | + * @param int|null $memory_limit Memory limit (bytes) the rebuild stopped | |
| 2158 | + * short of, when that was the failure. | |
| 1934 | 2159 | * @return void |
| 1935 | 2160 | */ |
| 1936 | - private function record_regeneration_failure(string $message, string $source): void { | |
| 2161 | + private function record_regeneration_failure(string $message, string $source, ?int $memory_limit = null): void { | |
| 1937 | 2162 | $pending = get_option(self::REGENERATION_PENDING_OPTION, []); |
| 1938 | 2163 | $pending = is_array($pending) ? $pending : []; |
| 1939 | - $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1; | |
| 2164 | + $attempts = !empty($pending['attempts']) ? (int) $pending['attempts'] : 0; | |
| 1940 | 2165 | |
| 1941 | - $backoff = min( | |
| 1942 | - self::REGENERATION_TAKEOVER_GRACE * (2 ** min($attempts, 10)), | |
| 1943 | - self::REGENERATION_MAX_BACKOFF | |
| 1944 | - ); | |
| 2166 | + // An attempt that claimed itself up front has already been counted. | |
| 2167 | + if (empty($pending['started'])) { | |
| 2168 | + $attempts++; | |
| 2169 | + } | |
| 2170 | + $attempts = max($attempts, 1); | |
| 1945 | 2171 | |
| 2172 | + $backoff = $this->regeneration_backoff($attempts); | |
| 2173 | + | |
| 1946 | 2174 | // Same precedence mark_regeneration_pending() enforces: a settings |
| 1947 | 2175 | // rebuild outranks a content one and must not be downgraded by a failed |
| 1948 | 2176 | // attempt. Overwriting it routed the retry back through the content |
| 1949 | 2177 | // path, where should_auto_generate() can be false and the completion |
| @@ -1952,20 +2180,24 @@ | ||
| 1952 | 2180 | // whichever attempt actually failed. |
| 1953 | 2181 | $current = isset($pending['source']) ? (string) $pending['source'] : ''; |
| 1954 | 2182 | $pending_source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content'; |
| 1955 | 2183 | |
| 1956 | - update_option( | |
| 1957 | - self::REGENERATION_PENDING_OPTION, | |
| 1958 | - [ | |
| 1959 | - 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(), | |
| 1960 | - 'source' => $pending_source, | |
| 1961 | - 'attempts' => $attempts, | |
| 1962 | - 'next_attempt' => time() + $backoff, | |
| 1963 | - 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0, | |
| 1964 | - ], | |
| 1965 | - true | |
| 1966 | - ); | |
| 2184 | + $marker = [ | |
| 2185 | + 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(), | |
| 2186 | + 'source' => $pending_source, | |
| 2187 | + 'attempts' => $attempts, | |
| 2188 | + 'next_attempt' => time() + $backoff, | |
| 2189 | + 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0, | |
| 2190 | + ]; | |
| 1967 | 2191 | |
| 2192 | + // Remembered so has_memory_for_retry() can keep requests with no more | |
| 2193 | + // memory than this from repeating the same attempt. | |
| 2194 | + if ($memory_limit !== null && $memory_limit > 0) { | |
| 2195 | + $marker['memory_limit'] = $memory_limit; | |
| 2196 | + } | |
| 2197 | + | |
| 2198 | + update_option(self::REGENERATION_PENDING_OPTION, $marker, true); | |
| 2199 | + | |
| 1968 | 2200 | update_option( |
| 1969 | 2201 | self::REGENERATION_ERROR_OPTION, |
| 1970 | 2202 | [ |
| 1971 | 2203 | 'message' => $message, |
| @@ -2020,16 +2252,34 @@ | ||
| 2020 | 2252 | if (!self::has_overdue_regeneration()) { |
| 2021 | 2253 | return; |
| 2022 | 2254 | } |
| 2023 | 2255 | |
| 2024 | - // Cron is running: it is about to do exactly this work. | |
| 2256 | + $pending = get_option(self::REGENERATION_PENDING_OPTION, []); | |
| 2257 | + $pending = is_array($pending) ? $pending : []; | |
| 2258 | + $source = isset($pending['source']) ? (string) $pending['source'] : 'content'; | |
| 2259 | + | |
| 2260 | + // Cron is running and its own event for this rebuild is still queued | |
| 2261 | + // and due: it is about to do exactly this work. Only then — an event | |
| 2262 | + // that has already been consumed (e.g. it fired while another process | |
| 2263 | + // held the generation lock) is never re-queued, and returning here | |
| 2264 | + // unconditionally left the rebuild to requests that could not finish it. | |
| 2025 | 2265 | if (wp_doing_cron()) { |
| 2266 | + $hook = $source === 'settings' ? 'thinkrank_regenerate_sitemap_settings' : 'thinkrank_regenerate_sitemap'; | |
| 2267 | + $next = wp_next_scheduled($hook); | |
| 2268 | + | |
| 2269 | + if ($next !== false && $next <= time()) { | |
| 2270 | + return; | |
| 2271 | + } | |
| 2272 | + } | |
| 2273 | + | |
| 2274 | + // The last attempt had to stop short of this process's memory limit. | |
| 2275 | + // Retrying at the same (or a lower) limit only repeats that, so leave | |
| 2276 | + // the rebuild to a process with more room — WP-CLI, a system cron, or a | |
| 2277 | + // host with a higher limit — instead of burning it on every request. | |
| 2278 | + if (!$this->has_memory_for_retry($pending)) { | |
| 2026 | 2279 | return; |
| 2027 | 2280 | } |
| 2028 | 2281 | |
| 2029 | - $pending = get_option(self::REGENERATION_PENDING_OPTION, []); | |
| 2030 | - $source = (is_array($pending) && isset($pending['source'])) ? (string) $pending['source'] : 'content'; | |
| 2031 | - | |
| 2032 | 2282 | if ($source === 'settings') { |
| 2033 | 2283 | $this->regenerate_sitemap_from_settings(); |
| 2034 | 2284 | return; |
| 2035 | 2285 | } |
| @@ -2063,8 +2313,228 @@ | ||
| 2063 | 2313 | delete_transient(self::GENERATION_LOCK_TRANSIENT); |
| 2064 | 2314 | } |
| 2065 | 2315 | |
| 2066 | 2316 | /** |
| 2317 | + * This process's PHP memory limit in bytes. | |
| 2318 | + * | |
| 2319 | + * @since 2.10.1 | |
| 2320 | + * @return int Bytes, or -1 when unlimited (or unreadable). | |
| 2321 | + */ | |
| 2322 | + private function current_memory_limit(): int { | |
| 2323 | + $limit = (string) ini_get('memory_limit'); | |
| 2324 | + | |
| 2325 | + if ($limit === '' || $limit === '-1') { | |
| 2326 | + return -1; | |
| 2327 | + } | |
| 2328 | + | |
| 2329 | + $bytes = (int) wp_convert_hr_to_bytes($limit); | |
| 2330 | + | |
| 2331 | + return $bytes > 0 ? $bytes : -1; | |
| 2332 | + } | |
| 2333 | + | |
| 2334 | + /** | |
| 2335 | + * May this process retry a rebuild that last stopped at the memory limit? | |
| 2336 | + * | |
| 2337 | + * Raises the limit the way wp-admin does first, so a request that can get | |
| 2338 | + * more room than the failed attempt had is still allowed to try. | |
| 2339 | + * | |
| 2340 | + * @since 2.10.1 | |
| 2341 | + * @param array $pending The pending marker. | |
| 2342 | + * @return bool True when there is no recorded memory failure, or this | |
| 2343 | + * process has more memory than the attempt that failed. | |
| 2344 | + */ | |
| 2345 | + private function has_memory_for_retry(array $pending): bool { | |
| 2346 | + if (empty($pending['memory_limit'])) { | |
| 2347 | + return true; | |
| 2348 | + } | |
| 2349 | + | |
| 2350 | + wp_raise_memory_limit('admin'); | |
| 2351 | + | |
| 2352 | + $limit = $this->current_memory_limit(); | |
| 2353 | + | |
| 2354 | + return $limit === -1 || $limit > (int) $pending['memory_limit']; | |
| 2355 | + } | |
| 2356 | + | |
| 2357 | + /** | |
| 2358 | + * Release what one walked chunk left behind. | |
| 2359 | + * | |
| 2360 | + * Hydrating a chunk through get_posts()/get_terms() also stores every | |
| 2361 | + * object and its meta in the in-process object cache, which nothing | |
| 2362 | + * empties until the request ends. Unsetting the chunk therefore freed | |
| 2363 | + * nothing, and the walk grew with the size of the site instead of the size | |
| 2364 | + * of a chunk — about 1.3 GB on a 45k-post site. | |
| 2365 | + * | |
| 2366 | + * A persistent object cache that supports it drops only its in-process | |
| 2367 | + * copy (`flush_runtime`); the shared store keeps its data. WordPress's | |
| 2368 | + * default cache has no shared store, and flushing it would empty every | |
| 2369 | + * group for the rest of the request (options, the queried object, other | |
| 2370 | + * plugins' data), so there only the chunk's own entries are deleted. A | |
| 2371 | + * persistent cache without `flush_runtime` is left alone: deleting from it | |
| 2372 | + * would evict the objects for every other request too. | |
| 2373 | + * | |
| 2374 | + * @since 2.10.1 | |
| 2375 | + * @param int $chunk_start memory_get_usage(true) before the chunk was hydrated. | |
| 2376 | + * @param array $groups Cache groups keyed by the chunk's object IDs. | |
| 2377 | + * @param int[] $ids The chunk's object IDs, plus any objects it | |
| 2378 | + * hydrated under their own (featured images). | |
| 2379 | + * @return void | |
| 2380 | + * @throws \Error See assert_memory_headroom(). | |
| 2381 | + */ | |
| 2382 | + private function release_walk_memory(int $chunk_start, array $groups, array $ids): void { | |
| 2383 | + // What one chunk costs before it is released: the margin the next one | |
| 2384 | + // needs. Measured in the same real allocated size assert_memory_headroom() | |
| 2385 | + // compares against the limit, so the two are the same unit. | |
| 2386 | + $this->walk_chunk_cost = max($this->walk_chunk_cost, memory_get_usage(true) - $chunk_start); | |
| 2387 | + | |
| 2388 | + if (wp_using_ext_object_cache()) { | |
| 2389 | + if ( | |
| 2390 | + function_exists('wp_cache_supports') | |
| 2391 | + && wp_cache_supports('flush_runtime') | |
| 2392 | + && function_exists('wp_cache_flush_runtime') | |
| 2393 | + ) { | |
| 2394 | + wp_cache_flush_runtime(); | |
| 2395 | + } | |
| 2396 | + } elseif (!empty($ids)) { | |
| 2397 | + foreach ($groups as $group) { | |
| 2398 | + wp_cache_delete_multiple($ids, $group); | |
| 2399 | + } | |
| 2400 | + } | |
| 2401 | + | |
| 2402 | + $this->assert_memory_headroom(); | |
| 2403 | + } | |
| 2404 | + | |
| 2405 | + /** | |
| 2406 | + * Cache groups get_posts() fills per post for the given post types. | |
| 2407 | + * | |
| 2408 | + * The post, its meta, and one relationships group per taxonomy the post | |
| 2409 | + * type uses (update_object_term_cache()). Term objects themselves are | |
| 2410 | + * bounded by the number of terms, not posts, so they are left cached. | |
| 2411 | + * | |
| 2412 | + * @since 2.10.1 | |
| 2413 | + * @param string[] $post_types Post types being walked. | |
| 2414 | + * @return string[] Cache groups keyed by post ID. | |
| 2415 | + */ | |
| 2416 | + private function chunk_post_cache_groups(array $post_types): array { | |
| 2417 | + $groups = ['posts', 'post_meta']; | |
| 2418 | + | |
| 2419 | + foreach (get_object_taxonomies($post_types) as $taxonomy) { | |
| 2420 | + $groups[] = $taxonomy . '_relationships'; | |
| 2421 | + } | |
| 2422 | + | |
| 2423 | + return array_values(array_unique($groups)); | |
| 2424 | + } | |
| 2425 | + | |
| 2426 | + /** | |
| 2427 | + * Stop an automatic rebuild before the memory limit rather than at it. | |
| 2428 | + * | |
| 2429 | + * A PHP memory fatal skips every catch and finally, so the lock, the | |
| 2430 | + * failure record and the backoff are all lost with it, while stopping here | |
| 2431 | + * is an ordinary, fully recorded failure. Checked before each chunk is | |
| 2432 | + * hydrated, against a margin of at least the largest chunk seen so far. | |
| 2433 | + * | |
| 2434 | + * It throws an \Error, not an \Exception, on purpose: the per-segment | |
| 2435 | + * catch (\Exception) blocks in generate_multiple_sitemaps() would otherwise | |
| 2436 | + * swallow it and carry on — writing an index without the aborted segments | |
| 2437 | + * and then pruning their files as orphans. Only the automatic rebuild's | |
| 2438 | + * catch (\Throwable) is meant to see it. | |
| 2439 | + * | |
| 2440 | + * @since 2.10.1 | |
| 2441 | + * @return void | |
| 2442 | + * @throws \Error When the automatic rebuild is close to the memory limit. | |
| 2443 | + */ | |
| 2444 | + private function assert_memory_headroom(): void { | |
| 2445 | + if (!$this->memory_guard) { | |
| 2446 | + return; | |
| 2447 | + } | |
| 2448 | + | |
| 2449 | + $limit = $this->current_memory_limit(); | |
| 2450 | + if ($limit === -1) { | |
| 2451 | + return; | |
| 2452 | + } | |
| 2453 | + | |
| 2454 | + // A fifth of the limit (at least 32 MB) for writing the files, or one | |
| 2455 | + // and a half of the costliest chunk if that is more — and never more | |
| 2456 | + // than half the limit either way. Without that outer cap a single | |
| 2457 | + // anomalously expensive chunk (500 posts of serialised page-builder or | |
| 2458 | + // ACF meta reaches hundreds of megabytes) puts the margin above the | |
| 2459 | + // limit itself, so every later check aborts at any usage at all, the | |
| 2460 | + // failure records this process's limit, and has_memory_for_retry() | |
| 2461 | + // then refuses every process that has the same limit. A site that | |
| 2462 | + // never actually ran out of memory would stop rebuilding until WP-CLI | |
| 2463 | + // or a system cron happened to run. | |
| 2464 | + $headroom = (int) min( | |
| 2465 | + max( | |
| 2466 | + min(max($limit * 0.2, 32 * MB_IN_BYTES), $limit * 0.5), | |
| 2467 | + $this->walk_chunk_cost * 1.5 | |
| 2468 | + ), | |
| 2469 | + $limit * 0.5 | |
| 2470 | + ); | |
| 2471 | + | |
| 2472 | + // The real allocated size, which is what PHP enforces memory_limit | |
| 2473 | + // against; memory_get_usage(false) reports only what is handed out of | |
| 2474 | + // those allocations and so understates the margin by the allocator's | |
| 2475 | + // slack. | |
| 2476 | + $usage = memory_get_usage(true); | |
| 2477 | + | |
| 2478 | + if ($usage > $limit - $headroom) { | |
| 2479 | + throw new \Error( | |
| 2480 | + sprintf( | |
| 2481 | + /* translators: 1: memory in use, 2: PHP memory limit. */ | |
| 2482 | + esc_html__('The sitemap rebuild was stopped at %1$s of the %2$s PHP memory limit, before PHP would have run out of memory. It will be retried by a process with more memory (WP-CLI or a system cron). To let it finish in the admin, raise the PHP memory_limit.', 'thinkrank'), | |
| 2483 | + esc_html((string) size_format($usage)), | |
| 2484 | + esc_html((string) size_format($limit)) | |
| 2485 | + ), | |
| 2486 | + // phpcs:ignore WordPress.Security.EscapeOutput.ExceptionNotEscaped -- an integer class constant, not output. | |
| 2487 | + self::MEMORY_ABORT_CODE | |
| 2488 | + ); | |
| 2489 | + } | |
| 2490 | + } | |
| 2491 | + | |
| 2492 | + /** | |
| 2493 | + * Run an automatic rebuild's generation with the fatal-safe bookkeeping. | |
| 2494 | + * | |
| 2495 | + * @since 2.10.1 | |
| 2496 | + * @param array $settings Sitemap settings. | |
| 2497 | + * @return bool Whatever generate_and_save() returned. | |
| 2498 | + * @throws \Throwable Whatever generation throws, after the memory guard is | |
| 2499 | + * switched back off. | |
| 2500 | + */ | |
| 2501 | + private function generate_for_regeneration(array $settings): bool { | |
| 2502 | + // Same headroom wp-admin gives itself; a no-op when the limit is | |
| 2503 | + // already higher or unlimited. | |
| 2504 | + wp_raise_memory_limit('admin'); | |
| 2505 | + | |
| 2506 | + $this->claim_regeneration_attempt(); | |
| 2507 | + $this->memory_guard = true; | |
| 2508 | + $this->walk_chunk_cost = 0; | |
| 2509 | + | |
| 2510 | + try { | |
| 2511 | + return $this->generate_and_save($settings); | |
| 2512 | + } finally { | |
| 2513 | + $this->memory_guard = false; | |
| 2514 | + } | |
| 2515 | + } | |
| 2516 | + | |
| 2517 | + /** | |
| 2518 | + * Record a failure thrown by an automatic rebuild. | |
| 2519 | + * | |
| 2520 | + * @since 2.10.1 | |
| 2521 | + * @param \Throwable $e What was thrown. | |
| 2522 | + * @param string $source Either 'content' or 'settings'. | |
| 2523 | + * @return void | |
| 2524 | + */ | |
| 2525 | + private function record_thrown_regeneration_failure(\Throwable $e, string $source): void { | |
| 2526 | + $memory_limit = null; | |
| 2527 | + | |
| 2528 | + if ($e instanceof \Error && $e->getCode() === self::MEMORY_ABORT_CODE) { | |
| 2529 | + $memory_limit = $this->current_memory_limit(); | |
| 2530 | + $memory_limit = $memory_limit > 0 ? $memory_limit : null; | |
| 2531 | + } | |
| 2532 | + | |
| 2533 | + $this->record_regeneration_failure($e->getMessage(), $source, $memory_limit); | |
| 2534 | + } | |
| 2535 | + | |
| 2536 | + /** | |
| 2067 | 2537 | * Report how automatic regeneration is faring, for the admin UI. |
| 2068 | 2538 | * |
| 2069 | 2539 | * The feature used to fail invisibly: `last_generated` simply stopped |
| 2070 | 2540 | * advancing and nothing drew attention to it (#629). |
| @@ -2214,9 +2684,9 @@ | ||
| 2214 | 2684 | // so the switch has not taken effect (#764). |
| 2215 | 2685 | return $this->switch_to_dynamic_delivery($settings, $revision, 'settings'); |
| 2216 | 2686 | } |
| 2217 | 2687 | |
| 2218 | - if ($this->generate_and_save($settings)) { | |
| 2688 | + if ($this->generate_for_regeneration($settings)) { | |
| 2219 | 2689 | $this->mark_regeneration_complete($revision); |
| 2220 | 2690 | |
| 2221 | 2691 | return true; |
| 2222 | 2692 | } |
| @@ -2227,9 +2697,9 @@ | ||
| 2227 | 2697 | ); |
| 2228 | 2698 | |
| 2229 | 2699 | return false; |
| 2230 | 2700 | } catch (\Throwable $e) { |
| 2231 | - $this->record_regeneration_failure($e->getMessage(), 'settings'); | |
| 2701 | + $this->record_thrown_regeneration_failure($e, 'settings'); | |
| 2232 | 2702 | |
| 2233 | 2703 | return false; |
| 2234 | 2704 | } finally { |
| 2235 | 2705 | $this->release_generation_lock(); |
| @@ -2363,9 +2833,9 @@ | ||
| 2363 | 2833 | $this->switch_to_dynamic_delivery($settings, $revision, 'content'); |
| 2364 | 2834 | return; |
| 2365 | 2835 | } |
| 2366 | 2836 | |
| 2367 | - if ($this->generate_and_save($settings)) { | |
| 2837 | + if ($this->generate_for_regeneration($settings)) { | |
| 2368 | 2838 | $this->mark_regeneration_complete($revision); |
| 2369 | 2839 | } else { |
| 2370 | 2840 | // Previously this returned quietly and last_generated simply |
| 2371 | 2841 | // stopped advancing, leaving the site owner with no way to learn |
| @@ -2375,9 +2845,9 @@ | ||
| 2375 | 2845 | 'content' |
| 2376 | 2846 | ); |
| 2377 | 2847 | } |
| 2378 | 2848 | } catch (\Throwable $e) { |
| 2379 | - $this->record_regeneration_failure($e->getMessage(), 'content'); | |
| 2849 | + $this->record_thrown_regeneration_failure($e, 'content'); | |
| 2380 | 2850 | } finally { |
| 2381 | 2851 | $this->release_generation_lock(); |
| 2382 | 2852 | } |
| 2383 | 2853 | } |
| @@ -3415,12 +3885,30 @@ | ||
| 3415 | 3885 | $buffer = []; |
| 3416 | 3886 | } |
| 3417 | 3887 | } |
| 3418 | 3888 | |
| 3419 | - // Flush the trailing partial page, or a single empty page when the | |
| 3420 | - // type had no entries at all (parity with the previous behavior of | |
| 3421 | - // always writing at least one page per configured child). | |
| 3422 | - if (!empty($buffer) || $page === 0) { | |
| 3889 | + // Flush the trailing partial page. | |
| 3890 | + // | |
| 3891 | + // A type that produced nothing writes no page at all in index | |
| 3892 | + // mode (#836). It used to write one empty urlset and list it in | |
| 3893 | + // the index, so a crawler was asked to fetch a file that | |
| 3894 | + // answers with no URLs — Search Console reports an empty | |
| 3895 | + // sitemap referenced from an index as a warning, and the fetch | |
| 3896 | + // is wasted on every pass. On this site three of eleven | |
| 3897 | + // children were empty: a post type with nothing published and | |
| 3898 | + // two taxonomies with no terms. | |
| 3899 | + // | |
| 3900 | + // Single-file mode still writes its one page even when empty, | |
| 3901 | + // because that file IS the site's /sitemap.xml and a 404 there | |
| 3902 | + // is worse than an empty urlset. Nothing lists it, so it costs | |
| 3903 | + // no crawl budget. | |
| 3904 | + // | |
| 3905 | + // Skipping the write also keeps the filename out of | |
| 3906 | + // $results['sitemaps_generated'], which is what | |
| 3907 | + // prune_orphaned_segments() treats as "written this run" — so | |
| 3908 | + // a type that empties after previously publishing has its | |
| 3909 | + // stale file deleted rather than left serving. | |
| 3910 | + if (!empty($buffer) || ($page === 0 && $index_config === null)) { | |
| 3423 | 3911 | $page++; |
| 3424 | 3912 | $this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children); |
| 3425 | 3913 | } |
| 3426 | 3914 | |
