← All changes | includes/seo/class-sitemap-generator.php +1449 -135 2.6.0 → 2.14.1 View file →

		@@ -62,8 +62,25 @@
62	62	'include_tags' => 'tags',
63	63	];
64	64
65	65	/**
	66	+ * Child sitemap type -> the object that actually supplies its entries.
	67	+ *
	68	+ * A child normally carries its object's own slug, but the sitemap presets UI
	69	+ * writes a display name for the WooCommerce taxonomy child
	70	+ * ('product_categories', not 'product_cat'), so a saved child list can name a
	71	+ * type no post type or taxonomy answers to. Resolving through here lets those
	72	+ * children take the same generic path as every other custom taxonomy instead
	73	+ * of needing a case of their own (#690).
	74	+ *
	75	+ * @since 2.7.0
	76	+ * @var array<string, string>
	77	+ */
	78	+ private const CHILD_TYPE_ALIASES = [
	79	+ 'product_categories' => 'product_cat',
	80	+ ];
	81	+
	82	+ /**
66	83	* Supported sitemap types
67	84	*
68	85	* @since 1.0.0
69	86	* @var array
		@@ -127,8 +144,98 @@
127	144	*/
128	145	public const REGENERATION_ERROR_OPTION = 'thinkrank_sitemap_regeneration_error';
129	146
130	147	/**
	148	+ * Delivery modes accepted by the `delivery_mode` setting.
	149	+ *
	150	+ * Mirrors LLMs_Txt_Manager::DELIVERY_MODES, which solved the same problem
	151	+ * for llms.txt. `auto` is the only value a site should normally need.
	152	+ *
	153	+ * @since 2.9.0
	154	+ * @var string[]
	155	+ */
	156	+ public const DELIVERY_MODES = ['auto', 'static', 'dynamic'];
	157	+
	158	+ /**
	159	+ * Cache group for documents rendered on the dynamic path.
	160	+ *
	161	+ * @since 2.9.0
	162	+ * @var string
	163	+ */
	164	+ private const DYNAMIC_CACHE_PREFIX = 'thinkrank_sitemap_doc_';
	165	+
	166	+ /**
	167	+ * How long a dynamically rendered document is cached.
	168	+ *
	169	+ * Invalidated by content and settings changes through
	170	+ * {@see self::flush_dynamic_cache()}, so this is only the backstop for a
	171	+ * change nothing hooked.
	172	+ *
	173	+ * @since 2.9.0
	174	+ * @var int
	175	+ */
	176	+ private const DYNAMIC_CACHE_TTL = 12 * HOUR_IN_SECONDS;
	177	+
	178	+ /**
	179	+ * How long the render lock is held before it is assumed abandoned.
	180	+ *
	181	+ * Long enough for a large site's full build, short enough that a request
	182	+ * killed mid-build does not lock the endpoint out for meaningfully long.
	183	+ *
	184	+ * @since 2.9.0
	185	+ * @var int
	186	+ */
	187	+ private const RENDER_LOCK_TTL = 60;
	188	+
	189	+ /**
	190	+ * Cached stand-in for "this site does not publish that name".
	191	+ *
	192	+ * published_document_names() lists what the configuration *could* produce,
	193	+ * but a child whose type is excluded produces nothing. Without a negative
	194	+ * entry those names miss the cache forever, so every request for one
	195	+ * rebuilt the entire sitemap — the same cost the positive cache exists to
	196	+ * avoid, on a public endpoint (#754 review).
	197	+ *
	198	+ * @since 2.9.0
	199	+ * @var string
	200	+ */
	201	+ private const ABSENT_MARKER = "\0thinkrank-absent";
	202	+
	203	+ /**
	204	+ * How many times, and how long, a losing request waits for the winner.
	205	+ *
	206	+ * Bounded at roughly a second in total: past that, building a second copy
	207	+ * costs less than making a crawler wait.
	208	+ *
	209	+ * @since 2.9.0
	210	+ * @var int
	211	+ */
	212	+ private const RENDER_LOCK_WAIT_ATTEMPTS = 4;
	213	+
	214	+ /**
	215	+ * @since 2.9.0
	216	+ * @var int
	217	+ */
	218	+ private const RENDER_LOCK_WAIT_MICROSECONDS = 250000;
	219	+
	220	+ /**
	221	+ * Where generated documents go instead of disk, when set.
	222	+ *
	223	+ * Every sitemap document this class produces — segments, the index, the
	224	+ * single flat file and local-sitemap.xml — is published through the one
	225	+ * writer, {@see self::save_sitemap_to_file()}. Swapping that writer for a
	226	+ * collector is therefore all it takes to render the same bytes without a
	227	+ * filesystem, which is what dynamic delivery needs (#752). Doing it here
	228	+ * rather than duplicating the build pipeline is deliberate: a second
	229	+ * pipeline would drift from this one, and the index in particular is
	230	+ * assembled from whatever the children actually produced.
	231	+ *
	232	+ * @since 2.9.0
	233	+ * @var callable|null
	234	+ */
	235	+ private $document_sink = null;
	236	+
	237	+ /**
131	238	* Transient guarding against two generations running at once. Shared with
132	239	* Sitemap_Endpoint's manual generate route so an automatic rebuild and a
133	240	* manual one cannot write the same files concurrently.
134	241	*
		@@ -144,8 +251,33 @@
144	251	* @var int
145	252	*/
146	253	private const TERM_WALK_CHUNK = 1000;
147	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	+
148	280	private array $sitemap_types = [
149	281	'posts' => [
150	282	'name' => 'Posts',
151	283	'post_types' => ['post'],
		@@ -172,25 +304,29 @@
172	304	]
173	305	];
174	306
175	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 $listener = null;
	315	+
	316	+ /**
176	317	* Constructor
177	318	*
178	319	* @since 1.0.0
	320	+ * @since 2.10.1 Registers no hooks, whatever `$register_hooks` says. The
	321	+ * content-change listeners are registered once, at bootstrap,
	322	+ * by register_content_listeners().
179	323	*
180		- * @param bool $register_hooks Optional. Whether to register the auto-generation
181		- * hooks. Pass false for a read-only instance built
182		- * solely to query settings — the hooks are bound to
183		- * `$this`, so a second hook-registering instance
184		- * 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.
185	326	*/
186	327	public function __construct(bool $register_hooks = true) {
187	328	parent::__construct('sitemap');
188		-
189		- // Initialize auto-generation hooks
190		- if ($register_hooks) {
191		- $this->init_auto_generation_hooks();
192		- }
193	329	}
194	330
195	331	/**
196	332	* Filter the args of a sitemap post query.
		@@ -234,34 +370,154 @@
234	370	return (array) apply_filters('thinkrank_sitemap_term_query_args', $args);
235	371	}
236	372
237	373	/**
238		- * Initialize WordPress hooks for auto-generation
	374	+ * Register the content-change listeners that queue an automatic rebuild.
239	375	*
240		- * @since 1.0.0
	376	+ * Called once per request, at plugin bootstrap, next to the WP-Cron
	377	+ * listeners that run the rebuild these queue (both in
	378	+ * Plugin::register_sitemap_cron_listeners(), on plugins_loaded). The
	379	+ * constructor used to register them, so they existed only in a request
	380	+ * that happened to build a generator: the REST endpoint, the setup wizard,
	381	+ * an MCP ability. Block-editor saves go through REST and were heard. A
	382	+ * scheduled post published by WP-Cron, Quick Edit, the classic editor and
	383	+ * WP-CLI were not, so the post stayed out of the sitemap with nothing
	384	+ * pending to recover it (#824). Each instance also added its own set, so
	385	+ * one REST save ran the handlers once per generator built.
	386	+ *
	387	+ * The generator is built on the first change and shared for the rest of
	388	+ * the request, so a bulk edit does not construct one per post.
	389	+ *
	390	+ * @since 2.10.1
241	391	* @return void
242	392	*/
243		- private function init_auto_generation_hooks(): void {
244		- // Content change hooks - use priority 20 to run after other plugins
245		- add_action('save_post', [$this, 'handle_content_change'], 20, 2);
246		- add_action('delete_post', [$this, 'handle_content_deletion'], 20);
247		- add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20);
248		- 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);
249	407
250		- // Taxonomy change hooks
251		- add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3);
252		- add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3);
253		- 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	+ }
254	413
255		- // NOTE: the WP-Cron regeneration listeners (thinkrank_regenerate_sitemap
256		- // and thinkrank_regenerate_sitemap_settings) are registered at plugin
257		- // bootstrap (Plugin::register_sitemap_cron_listeners(), on plugins_loaded)
258		- // rather than here. A cron run never builds this class via the REST
259		- // endpoint (no rest_api_init), so registering them in the constructor
260		- // 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);
261	431	}
262	432
263	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	+ /**
264	520	* Generate XML sitemap
265	521	*
266	522	* @since 1.0.0
267	523	*
		@@ -270,9 +526,9 @@
270	526	*/
271	527	public function generate_sitemap(array $options = []): string {
272	528	$settings = $this->get_settings('site');
273	529
274		- $xml = $this->xml_prolog($settings, 'sitemap.xsl');
	530	+ $xml = $this->xml_prolog($settings, 'sitemap');
275	531
276	532	// Add image namespace if images are enabled
277	533	if (!empty($settings['include_images'])) {
278	534	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
		@@ -435,8 +691,15 @@
435	691	]
436	692	],
437	693	'use_sitemap_index' => false,
438	694
	695	+ // How the sitemap reaches crawlers. 'auto' keeps the historical
	696	+ // behaviour wherever the web root is writable, and only falls back
	697	+ // to serving the sitemap from PHP where writing a file is
	698	+ // impossible — previously a hard failure with nothing served
	699	+ // (#752).
	700	+ 'delivery_mode' => 'auto',
	701	+
439	702	// General Settings
440	703	'links_per_sitemap' => 1000,
441	704	'include_images' => true,
442	705	'include_featured_images' => false,
		@@ -458,8 +721,17 @@
458	721	// Advanced Options
459	722	'enable_styling' => true,
460	723	'custom_url_pattern' => 'sitemap-{type}.xml',
461	724
	725	+ // Stylesheet branding (#639). Both colours default to empty, not
	726	+ // to the stock hexes: empty means the stylesheet's own value
	727	+ // stands, so a site that never opens this screen renders exactly
	728	+ // as it did before the setting existed.
	729	+ 'styling_logo' => false,
	730	+ 'styling_logo_url' => '',
	731	+ 'styling_color_main' => '',
	732	+ 'styling_color_accent' => '',
	733	+
462	734	// Generation tracking
463	735	'last_generated' => ''
464	736	];
465	737	}
		@@ -464,8 +736,49 @@
464	736	];
465	737	}
466	738
467	739	/**
	740	+ * Normalize the stylesheet branding values on the way into the store.
	741	+ *
	742	+ * The generic sanitizer only runs `sanitize_text_field()` over a string,
	743	+ * which happily keeps "red" or "rebeccapurple" as a colour. Nothing
	744	+ * downstream can use those — {@see Sitemap_Stylesheet::render()} skips any
	745	+ * value it cannot read as a hex colour — so storing them would report a
	746	+ * successful save of a setting that changes nothing, and `get-sitemap-
	747	+ * settings` would hand an agent back a colour the sitemap does not use.
	748	+ * Reducing here instead keeps the store and the rendering in agreement.
	749	+ *
	750	+ * @since 2.7.0
	751	+ *
	752	+ * @param array $settings Settings to sanitize.
	753	+ * @param string $context_type Context the save is for.
	754	+ * @return array
	755	+ */
	756	+ protected function sanitize_settings(array $settings, string $context_type = 'site'): array {
	757	+ $sanitized = parent::sanitize_settings($settings, $context_type);
	758	+
	759	+ foreach (['styling_color_main', 'styling_color_accent'] as $key) {
	760	+ if (array_key_exists($key, $sanitized)) {
	761	+ $sanitized[$key] = Sitemap_Stylesheet::hex($sanitized[$key]);
	762	+ }
	763	+ }
	764	+
	765	+ if (array_key_exists('styling_logo_url', $sanitized)) {
	766	+ $sanitized['styling_logo_url'] = esc_url_raw((string) $sanitized['styling_logo_url']);
	767	+ }
	768	+
	769	+ // A mode this build cannot act on has to be stored as the fallback
	770	+ // rather than kept verbatim, or get-sitemap-settings reports a delivery
	771	+ // mode the site does not actually apply.
	772	+ if (array_key_exists('delivery_mode', $sanitized)) {
	773	+ $mode = sanitize_key((string) $sanitized['delivery_mode']);
	774	+ $sanitized['delivery_mode'] = in_array($mode, self::DELIVERY_MODES, true) ? $mode : 'auto';
	775	+ }
	776	+
	777	+ return $sanitized;
	778	+ }
	779	+
	780	+ /**
468	781	* Get settings schema definition (implements interface)
469	782	*
470	783	* @since 1.0.0
471	784	*
		@@ -479,8 +792,15 @@
479	792	'title' => 'Enable Sitemap',
480	793	'description' => 'Generate XML sitemap for search engines',
481	794	'default' => true
482	795	],
	796	+ 'delivery_mode' => [
	797	+ 'type' => 'string',
	798	+ 'title' => 'Sitemap Delivery',
	799	+ 'description' => 'How the sitemap is served: auto picks static when the WordPress root is writable and dynamic when it is not, static writes files to the web root, dynamic serves the sitemap from WordPress with no files written',
	800	+ 'enum' => self::DELIVERY_MODES,
	801	+ 'default' => 'auto'
	802	+ ],
483	803	'include_posts' => [
484	804	'type' => 'boolean',
485	805	'title' => 'Include Posts',
486	806	'description' => 'Include blog posts in sitemap',
		@@ -521,8 +841,32 @@
521	841	'type' => 'string',
522	842	'title' => 'Last Generated',
523	843	'description' => 'Timestamp of last sitemap generation',
524	844	'default' => ''
	845	+ ],
	846	+ 'styling_logo' => [
	847	+ 'type' => 'boolean',
	848	+ 'title' => 'Show Logo On Sitemap',
	849	+ 'description' => 'Show a logo above the sitemap heading',
	850	+ 'default' => false
	851	+ ],
	852	+ 'styling_logo_url' => [
	853	+ 'type' => 'string',
	854	+ 'title' => 'Sitemap Logo',
	855	+ 'description' => 'Logo image URL. Empty falls back to the site icon',
	856	+ 'default' => ''
	857	+ ],
	858	+ 'styling_color_main' => [
	859	+ 'type' => 'string',
	860	+ 'title' => 'Sitemap Main Color',
	861	+ 'description' => 'Hex color for the sitemap header, links and table head. Empty keeps the stock palette',
	862	+ 'default' => ''
	863	+ ],
	864	+ 'styling_color_accent' => [
	865	+ 'type' => 'string',
	866	+ 'title' => 'Sitemap Accent Color',
	867	+ 'description' => 'Hex color for the header gradient and link hovers. Empty keeps the stock palette',
	868	+ 'default' => ''
525	869	]
526	870	];
527	871	}
528	872
		@@ -539,9 +883,14 @@
539	883	* @return string XML URL entry
540	884	*/
541	885	private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string {
542	886	$xml = " <url>\n";
543		- $xml .= " <loc>" . esc_url($url) . "</loc>\n";
	887	+ // Every <loc> in every sitemap passes through here, which is why the
	888	+ // scheme preference is applied at this one point rather than at each
	889	+ // of the dozen collectors that build URLs (#638). An http sitemap on
	890	+ // an https site hands search engines the wrong address for the whole
	891	+ // site at once.
	892	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($url)) . "</loc>\n";
544	893	// Omit <lastmod> when unknown (empty) — a fabricated timestamp is worse
545	894	// than no timestamp, and an absent lastmod is valid per the spec.
546	895	if (!empty($lastmod)) {
547	896	$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n";
		@@ -551,9 +900,9 @@
551	900
552	901	// Add image entries if provided
553	902	foreach ($images as $image) {
554	903	$xml .= " <image:image>\n";
555		- $xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n";
	904	+ $xml .= " <image:loc>" . esc_url(Url_Scheme::apply((string) $image['url'])) . "</image:loc>\n";
556	905
557	906	if (!empty($image['title'])) {
558	907	$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n";
559	908	}
		@@ -674,8 +1023,11 @@
674	1023	// well-bounded routine with no ceiling (#402).
675	1024	$total = count($all_ids);
676	1025
677	1026	for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) {
	1027	+ $this->assert_memory_headroom();
	1028	+ $chunk_start = memory_get_usage(true);
	1029	+
678	1030	$chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK);
679	1031
680	1032	$posts = get_posts($this->filter_query_args([
681	1033	'post_type' => $post_types,
		@@ -684,8 +1036,12 @@
684	1036	'post__in' => $chunk,
685	1037	'orderby' => 'post__in', // preserve the resolved order
686	1038	]));
687	1039
	1040	+ // get_featured_image() hydrates each featured image under the
	1041	+ // attachment's own ID, which the chunk's post IDs do not reach.
	1042	+ $attachment_ids = [];
	1043	+
688	1044	foreach ($posts as $post) {
689	1045	if ($this->should_include_in_sitemap($post, $settings)) {
690	1046	/**
691	1047	* Filter a sitemap entry's permalink.
		@@ -706,14 +1062,25 @@
706	1062	$priority = $this->calculate_intelligent_priority($post, $post->post_type);
707	1063	$changefreq = $this->calculate_change_frequency($post, $post->post_type);
708	1064	$images = $this->extract_post_images($post, $settings);
709	1065
	1066	+ $thumbnail_id = (int) get_post_thumbnail_id($post);
	1067	+ if ($thumbnail_id > 0) {
	1068	+ $attachment_ids[] = $thumbnail_id;
	1069	+ }
	1070	+
710	1071	yield $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images);
711	1072	}
712	1073	}
713	1074
714		- // Free the hydrated chunk before loading the next one.
	1075	+ // Free the hydrated chunk before loading the next one — including
	1076	+ // the copies get_posts() left in the runtime object cache.
715	1077	unset($posts);
	1078	+ $this->release_walk_memory(
	1079	+ $chunk_start,
	1080	+ $this->chunk_post_cache_groups($post_types),
	1081	+ array_merge($chunk, $attachment_ids)
	1082	+ );
716	1083	}
717	1084	}
718	1085
719	1086	/**
		@@ -773,8 +1140,11 @@
773	1140	// Same moving window as the post walk above, for the same reason.
774	1141	$total = count($all_ids);
775	1142
776	1143	for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) {
	1144	+ $this->assert_memory_headroom();
	1145	+ $chunk_start = memory_get_usage(true);
	1146	+
777	1147	$chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK);
778	1148
779	1149	$terms = get_terms($this->filter_term_query_args([
780	1150	'taxonomy' => $taxonomy,
		@@ -787,12 +1157,12 @@
787	1157	continue;
788	1158	}
789	1159
790	1160	foreach ($terms as $term) {
791		- // A term the user marked noindex must not be advertised in the
792		- // sitemap: the robots tag now honours term meta, so listing it
793		- // here would have the sitemap contradict the page's own tag.
794		- if ($this->term_is_noindexed((int) $term->term_id)) {
	1161	+ // A term whose archive says noindex (its own override or its
	1162	+ // taxonomy's), or that redirects, must not be advertised: the
	1163	+ // sitemap would contradict the page's own signal (#911).
	1164	+ if (!Indexability::is_indexable_term($term)) {
795	1165	continue;
796	1166	}
797	1167
798	1168	$url = get_term_link($term);
		@@ -803,8 +1173,9 @@
803	1173	}
804	1174	}
805	1175
806	1176	unset($terms);
	1177	+ $this->release_walk_memory($chunk_start, ['terms', 'term_meta'], $chunk);
807	1178	}
808	1179	}
809	1180
810	1181	/**
		@@ -819,19 +1190,25 @@
819	1190	* kept a shadowing file behind (#510) or had a competitor's deleted.
820	1191	*
821	1192	* @since 2.1.1
822	1193	*
823		- * @param array $settings Sitemap settings (read for `enable_styling`).
824		- * @param string $stylesheet Stylesheet basename in static/xsl/.
	1194	+ * The stylesheet URL is served by {@see Sitemap_Stylesheet}, not read off
	1195	+ * disk by the web server, because a static file cannot carry the site's own
	1196	+ * logo and colours (#639). It is a fixed URL: the palette is applied per
	1197	+ * request, so changing a brand colour needs no regeneration and shows up on
	1198	+ * sitemaps published long before.
	1199	+ *
	1200	+ * @param array $settings Sitemap settings (read for `enable_styling`).
	1201	+ * @param string $variant Stylesheet variant, `sitemap` or `index`.
825	1202	* @return string Prolog lines, newline-terminated.
826	1203	*/
827		- private function xml_prolog(array $settings, string $stylesheet): string {
	1204	+ private function xml_prolog(array $settings, string $variant): string {
828	1205	$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
829	1206	$xml .= THINKRANK_SITEMAP_MARKER . "\n";
830	1207
831	1208	// The stylesheet is presentation only, so it stays opt-in.
832	1209	if (!empty($settings['enable_styling'])) {
833		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/' . $stylesheet) . '"?>' . "\n";
	1210	+ $xml .= '<?xml-stylesheet type="text/xsl" href="' . esc_url(Sitemap_Stylesheet::url($variant)) . '"?>' . "\n";
834	1211	}
835	1212
836	1213	return $xml;
837	1214	}
		@@ -846,9 +1223,9 @@
846	1223	* @param bool $with_image_ns Include the image sitemap namespace
847	1224	* @return string Full sitemap XML
848	1225	*/
849	1226	private function wrap_urlset(array $entries, array $settings, bool $with_image_ns): string {
850		- $xml = $this->xml_prolog($settings, 'sitemap.xsl');
	1227	+ $xml = $this->xml_prolog($settings, 'sitemap');
851	1228
852	1229	if ($with_image_ns && !empty($settings['include_images'])) {
853	1230	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
854	1231	} else {
		@@ -1090,14 +1467,15 @@
1090	1467	if (is_wp_error($all_terms)) {
1091	1468	return [];
1092	1469	}
1093	1470
1094		- // Drop terms the user marked noindex. This path feeds the single general
1095		- // sitemap while collect_taxonomy_entries_iter() feeds the segmented ones,
1096		- // so both need the filter or the two disagree about the same term.
	1471	+ // Drop terms that are not indexable destinations. This path feeds the
	1472	+ // single general sitemap while collect_taxonomy_entries_iter() feeds
	1473	+ // the segmented ones, so both need the filter or the two disagree about
	1474	+ // the same term.
1097	1475	$all_terms = array_values(array_filter(
1098	1476	$all_terms,
1099		- fn($term) => !$this->term_is_noindexed((int) $term->term_id)
	1477	+ static fn($term) => $term instanceof \WP_Term && Indexability::is_indexable_term($term)
1100	1478	));
1101	1479
1102	1480	// Group terms by taxonomy
1103	1481	return $this->group_terms_by_taxonomy($all_terms);
		@@ -1279,17 +1657,16 @@
1279	1657	if (!in_array($post->post_status, ['publish', 'private'], true)) {
1280	1658	return false;
1281	1659	}
1282	1660
1283		- // Check if post overrides robots and sets noindex.
1284		- if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) {
1285		- $raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true);
1286		- if (is_string($raw) && $raw !== '') {
1287		- $robots = json_decode($raw, true);
1288		- if (is_array($robots) && !empty($robots['noindex'])) {
1289		- return false;
1290		- }
1291		- }
	1661	+ // A URL whose page says noindex, or that ThinkRank redirects, is not a
	1662	+ // destination. Only the per-post noindex used to be read here, so a
	1663	+ // post type set to No-index still had every item listed, and redirected
	1664	+ // posts were submitted as "Page with redirect" (#911). The password
	1665	+ // check stays with the setting above: listing protected posts is a
	1666	+ // choice this sitemap has always offered.
	1667	+ if (Indexability::is_post_noindexed($post) || Indexability::is_post_redirected($post)) {
	1668	+ return false;
1292	1669	}
1293	1670
1294	1671	return true;
1295	1672	}
		@@ -1328,35 +1705,8 @@
1328	1705	return $ids;
1329	1706	}
1330	1707
1331	1708	/**
1332		- * Whether a term carries an explicit noindex override.
1333		- *
1334		- * Mirrors the post-side check in should_include_post(); terms store the same
1335		- * `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written
1336		- * by the update-term-seo ability and by the SEO importer.
1337		- *
1338		- * @since 1.31.0
1339		- *
1340		- * @param int $term_id Term to test.
1341		- * @return bool True when the term is marked noindex.
1342		- */
1343		- private function term_is_noindexed(int $term_id): bool {
1344		- if (!(bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true)) {
1345		- return false;
1346		- }
1347		-
1348		- $raw = get_term_meta($term_id, '_thinkrank_robots_meta', true);
1349		- if (!is_string($raw) || $raw === '') {
1350		- return false;
1351		- }
1352		-
1353		- $robots = json_decode($raw, true);
1354		-
1355		- return is_array($robots) && !empty($robots['noindex']);
1356		- }
1357		-
1358		- /**
1359	1709	* Count total URLs in sitemap
1360	1710	*
1361	1711	* @since 1.0.0
1362	1712	*
		@@ -1635,8 +1985,14 @@
1635	1985	* @param string $source Either 'content' or 'settings'.
1636	1986	* @return void
1637	1987	*/
1638	1988	private function mark_regeneration_pending(string $source): void {
	1989	+ // Whatever made the static files stale made the rendered ones stale
	1990	+ // too. Invalidating here rather than only on the rebuild keeps the two
	1991	+ // delivery modes reacting to exactly the same triggers, which is the
	1992	+ // only way a dynamic site stays as fresh as a static one (#752).
	1993	+ $this->flush_dynamic_cache();
	1994	+
1639	1995	$pending = get_option(self::REGENERATION_PENDING_OPTION, []);
1640	1996	$pending = is_array($pending) ? $pending : [];
1641	1997
1642	1998	$since = !empty($pending['since']) ? (int) $pending['since'] : time();
		@@ -1642,11 +1998,13 @@
1642	1998	$since = !empty($pending['since']) ? (int) $pending['since'] : time();
1643	1999	$current = isset($pending['source']) ? (string) $pending['source'] : '';
1644	2000	$source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
1645	2001
	2002	+ // Merged so the bookkeeping a rebuild keeps on the marker (`started`,
	2003	+ // `memory_limit`) survives an edit made while it is outstanding.
1646	2004	update_option(
1647	2005	self::REGENERATION_PENDING_OPTION,
1648		- [
	2006	+ array_merge($pending, [
1649	2007	'since' => $since,
1650	2008	'source' => $source,
1651	2009	'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 0,
1652	2010	'next_attempt' => !empty($pending['next_attempt'])
		@@ -1654,9 +2012,9 @@
1654	2012	: time() + self::REGENERATION_TAKEOVER_GRACE,
1655	2013	// Bumped on every change so a rebuild can tell whether the edit
1656	2014	// it started for is still the newest one outstanding.
1657	2015	'revision' => (!empty($pending['revision']) ? (int) $pending['revision'] : 0) + 1,
1658		- ],
	2016	+ ]),
1659	2017	true
1660	2018	);
1661	2019	}
1662	2020
		@@ -1702,8 +2060,18 @@
1702	2060	$revision !== null
1703	2061	&& is_array($pending)
1704	2062	&& (int) ($pending['revision'] ?? 0) !== $revision
1705	2063	) {
	2064	+ // This attempt succeeded, so drop what it claimed: the newer change
	2065	+ // waits the grace a fresh edit gets, not a failure backoff it never
	2066	+ // earned. Not zero: that edit queued its own debounced event, and
	2067	+ // a marker due at once had the next admin request rebuild in its
	2068	+ // shutdown and the event rebuild again seconds later.
	2069	+ unset($pending['started'], $pending['memory_limit']);
	2070	+ $pending['attempts'] = 0;
	2071	+ $pending['next_attempt'] = time() + self::REGENERATION_TAKEOVER_GRACE;
	2072	+
	2073	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
1706	2074	return;
1707	2075	}
1708	2076
1709	2077	delete_option(self::REGENERATION_PENDING_OPTION);
		@@ -1709,8 +2077,69 @@
1709	2077	delete_option(self::REGENERATION_PENDING_OPTION);
1710	2078	}
1711	2079
1712	2080	/**
	2081	+ * Record the attempt that is about to run before it runs.
	2082	+ *
	2083	+ * A PHP fatal — the memory limit or max_execution_time — is not a
	2084	+ * Throwable, so no catch or finally around the generation runs when the
	2085	+ * process dies, and a failure recorded afterwards was never recorded at
	2086	+ * all: `attempts` stayed 0, the backoff never applied, and the next request
	2087	+ * started the same doomed rebuild again (a fatal every few minutes for as
	2088	+ * long as an admin was logged in). Claiming the attempt up front makes the
	2089	+ * backoff hold even when nothing after this line gets to run, and leaves a
	2090	+ * `started` stamp the next attempt can recognise as an interrupted one.
	2091	+ *
	2092	+ * @since 2.10.1
	2093	+ * @return void
	2094	+ */
	2095	+ private function claim_regeneration_attempt(): void {
	2096	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	2097	+
	2098	+ // Nothing outstanding (e.g. a manual generation already satisfied it):
	2099	+ // there is no marker to retry from, so nothing to claim.
	2100	+ if (!is_array($pending) || empty($pending['since'])) {
	2101	+ return;
	2102	+ }
	2103	+
	2104	+ if (!empty($pending['started'])) {
	2105	+ // The previous attempt claimed itself and never reported back.
	2106	+ update_option(
	2107	+ self::REGENERATION_ERROR_OPTION,
	2108	+ [
	2109	+ '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'),
	2110	+ 'source' => isset($pending['source']) ? (string) $pending['source'] : 'content',
	2111	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 1,
	2112	+ 'time' => (int) $pending['started'],
	2113	+ ],
	2114	+ false
	2115	+ );
	2116	+ }
	2117	+
	2118	+ $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1;
	2119	+
	2120	+ $pending['attempts'] = $attempts;
	2121	+ $pending['next_attempt'] = time() + $this->regeneration_backoff($attempts);
	2122	+ $pending['started'] = time();
	2123	+
	2124	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
	2125	+ }
	2126	+
	2127	+ /**
	2128	+ * Delay before the next takeover after the given number of attempts.
	2129	+ *
	2130	+ * @since 2.10.1
	2131	+ * @param int $attempts Attempts made so far (1 or more).
	2132	+ * @return int Seconds.
	2133	+ */
	2134	+ private function regeneration_backoff(int $attempts): int {
	2135	+ return (int) min(
	2136	+ self::REGENERATION_TAKEOVER_GRACE * (2 ** min(max($attempts, 1), 10)),
	2137	+ self::REGENERATION_MAX_BACKOFF
	2138	+ );
	2139	+ }
	2140	+
	2141	+ /**
1713	2142	* Record a failed regeneration instead of discarding it.
1714	2143	*
1715	2144	* Keeps the pending marker in place so the rebuild is retried, but backs the
1716	2145	* next attempt off exponentially (capped) so a persistently failing
		@@ -1716,22 +2145,30 @@
1716	2145	* next attempt off exponentially (capped) so a persistently failing
1717	2146	* generation cannot run on every admin request.
1718	2147	*
1719	2148	* @since 2.2.1
1720		- * @param string $message Failure detail.
1721		- * @param string $source Either 'content' or 'settings'.
	2149	+ * @since 2.10.1 Accepts the memory limit a rebuild had to stop short of, and
	2150	+ * does not count an attempt claim_regeneration_attempt()
	2151	+ * already counted.
	2152	+ * @param string $message Failure detail.
	2153	+ * @param string $source Either 'content' or 'settings'.
	2154	+ * @param int|null $memory_limit Memory limit (bytes) the rebuild stopped
	2155	+ * short of, when that was the failure.
1722	2156	* @return void
1723	2157	*/
1724		- private function record_regeneration_failure(string $message, string $source): void {
	2158	+ private function record_regeneration_failure(string $message, string $source, ?int $memory_limit = null): void {
1725	2159	$pending = get_option(self::REGENERATION_PENDING_OPTION, []);
1726	2160	$pending = is_array($pending) ? $pending : [];
1727		- $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1;
	2161	+ $attempts = !empty($pending['attempts']) ? (int) $pending['attempts'] : 0;
1728	2162
1729		- $backoff = min(
1730		- self::REGENERATION_TAKEOVER_GRACE * (2 ** min($attempts, 10)),
1731		- self::REGENERATION_MAX_BACKOFF
1732		- );
	2163	+ // An attempt that claimed itself up front has already been counted.
	2164	+ if (empty($pending['started'])) {
	2165	+ $attempts++;
	2166	+ }
	2167	+ $attempts = max($attempts, 1);
1733	2168
	2169	+ $backoff = $this->regeneration_backoff($attempts);
	2170	+
1734	2171	// Same precedence mark_regeneration_pending() enforces: a settings
1735	2172	// rebuild outranks a content one and must not be downgraded by a failed
1736	2173	// attempt. Overwriting it routed the retry back through the content
1737	2174	// path, where should_auto_generate() can be false and the completion
		@@ -1740,20 +2177,24 @@
1740	2177	// whichever attempt actually failed.
1741	2178	$current = isset($pending['source']) ? (string) $pending['source'] : '';
1742	2179	$pending_source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
1743	2180
1744		- update_option(
1745		- self::REGENERATION_PENDING_OPTION,
1746		- [
1747		- 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(),
1748		- 'source' => $pending_source,
1749		- 'attempts' => $attempts,
1750		- 'next_attempt' => time() + $backoff,
1751		- 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0,
1752		- ],
1753		- true
1754		- );
	2181	+ $marker = [
	2182	+ 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(),
	2183	+ 'source' => $pending_source,
	2184	+ 'attempts' => $attempts,
	2185	+ 'next_attempt' => time() + $backoff,
	2186	+ 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0,
	2187	+ ];
1755	2188
	2189	+ // Remembered so has_memory_for_retry() can keep requests with no more
	2190	+ // memory than this from repeating the same attempt.
	2191	+ if ($memory_limit !== null && $memory_limit > 0) {
	2192	+ $marker['memory_limit'] = $memory_limit;
	2193	+ }
	2194	+
	2195	+ update_option(self::REGENERATION_PENDING_OPTION, $marker, true);
	2196	+
1756	2197	update_option(
1757	2198	self::REGENERATION_ERROR_OPTION,
1758	2199	[
1759	2200	'message' => $message,
		@@ -1808,16 +2249,34 @@
1808	2249	if (!self::has_overdue_regeneration()) {
1809	2250	return;
1810	2251	}
1811	2252
1812		- // Cron is running: it is about to do exactly this work.
	2253	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2254	+ $pending = is_array($pending) ? $pending : [];
	2255	+ $source = isset($pending['source']) ? (string) $pending['source'] : 'content';
	2256	+
	2257	+ // Cron is running and its own event for this rebuild is still queued
	2258	+ // and due: it is about to do exactly this work. Only then — an event
	2259	+ // that has already been consumed (e.g. it fired while another process
	2260	+ // held the generation lock) is never re-queued, and returning here
	2261	+ // unconditionally left the rebuild to requests that could not finish it.
1813	2262	if (wp_doing_cron()) {
	2263	+ $hook = $source === 'settings' ? 'thinkrank_regenerate_sitemap_settings' : 'thinkrank_regenerate_sitemap';
	2264	+ $next = wp_next_scheduled($hook);
	2265	+
	2266	+ if ($next !== false && $next <= time()) {
	2267	+ return;
	2268	+ }
	2269	+ }
	2270	+
	2271	+ // The last attempt had to stop short of this process's memory limit.
	2272	+ // Retrying at the same (or a lower) limit only repeats that, so leave
	2273	+ // the rebuild to a process with more room — WP-CLI, a system cron, or a
	2274	+ // host with a higher limit — instead of burning it on every request.
	2275	+ if (!$this->has_memory_for_retry($pending)) {
1814	2276	return;
1815	2277	}
1816	2278
1817		- $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
1818		- $source = (is_array($pending) && isset($pending['source'])) ? (string) $pending['source'] : 'content';
1819		-
1820	2279	if ($source === 'settings') {
1821	2280	$this->regenerate_sitemap_from_settings();
1822	2281	return;
1823	2282	}
		@@ -1851,8 +2310,228 @@
1851	2310	delete_transient(self::GENERATION_LOCK_TRANSIENT);
1852	2311	}
1853	2312
1854	2313	/**
	2314	+ * This process's PHP memory limit in bytes.
	2315	+ *
	2316	+ * @since 2.10.1
	2317	+ * @return int Bytes, or -1 when unlimited (or unreadable).
	2318	+ */
	2319	+ private function current_memory_limit(): int {
	2320	+ $limit = (string) ini_get('memory_limit');
	2321	+
	2322	+ if ($limit === '' || $limit === '-1') {
	2323	+ return -1;
	2324	+ }
	2325	+
	2326	+ $bytes = (int) wp_convert_hr_to_bytes($limit);
	2327	+
	2328	+ return $bytes > 0 ? $bytes : -1;
	2329	+ }
	2330	+
	2331	+ /**
	2332	+ * May this process retry a rebuild that last stopped at the memory limit?
	2333	+ *
	2334	+ * Raises the limit the way wp-admin does first, so a request that can get
	2335	+ * more room than the failed attempt had is still allowed to try.
	2336	+ *
	2337	+ * @since 2.10.1
	2338	+ * @param array $pending The pending marker.
	2339	+ * @return bool True when there is no recorded memory failure, or this
	2340	+ * process has more memory than the attempt that failed.
	2341	+ */
	2342	+ private function has_memory_for_retry(array $pending): bool {
	2343	+ if (empty($pending['memory_limit'])) {
	2344	+ return true;
	2345	+ }
	2346	+
	2347	+ wp_raise_memory_limit('admin');
	2348	+
	2349	+ $limit = $this->current_memory_limit();
	2350	+
	2351	+ return $limit === -1 || $limit > (int) $pending['memory_limit'];
	2352	+ }
	2353	+
	2354	+ /**
	2355	+ * Release what one walked chunk left behind.
	2356	+ *
	2357	+ * Hydrating a chunk through get_posts()/get_terms() also stores every
	2358	+ * object and its meta in the in-process object cache, which nothing
	2359	+ * empties until the request ends. Unsetting the chunk therefore freed
	2360	+ * nothing, and the walk grew with the size of the site instead of the size
	2361	+ * of a chunk — about 1.3 GB on a 45k-post site.
	2362	+ *
	2363	+ * A persistent object cache that supports it drops only its in-process
	2364	+ * copy (`flush_runtime`); the shared store keeps its data. WordPress's
	2365	+ * default cache has no shared store, and flushing it would empty every
	2366	+ * group for the rest of the request (options, the queried object, other
	2367	+ * plugins' data), so there only the chunk's own entries are deleted. A
	2368	+ * persistent cache without `flush_runtime` is left alone: deleting from it
	2369	+ * would evict the objects for every other request too.
	2370	+ *
	2371	+ * @since 2.10.1
	2372	+ * @param int $chunk_start memory_get_usage(true) before the chunk was hydrated.
	2373	+ * @param array $groups Cache groups keyed by the chunk's object IDs.
	2374	+ * @param int[] $ids The chunk's object IDs, plus any objects it
	2375	+ * hydrated under their own (featured images).
	2376	+ * @return void
	2377	+ * @throws \Error See assert_memory_headroom().
	2378	+ */
	2379	+ private function release_walk_memory(int $chunk_start, array $groups, array $ids): void {
	2380	+ // What one chunk costs before it is released: the margin the next one
	2381	+ // needs. Measured in the same real allocated size assert_memory_headroom()
	2382	+ // compares against the limit, so the two are the same unit.
	2383	+ $this->walk_chunk_cost = max($this->walk_chunk_cost, memory_get_usage(true) - $chunk_start);
	2384	+
	2385	+ if (wp_using_ext_object_cache()) {
	2386	+ if (
	2387	+ function_exists('wp_cache_supports')
	2388	+ && wp_cache_supports('flush_runtime')
	2389	+ && function_exists('wp_cache_flush_runtime')
	2390	+ ) {
	2391	+ wp_cache_flush_runtime();
	2392	+ }
	2393	+ } elseif (!empty($ids)) {
	2394	+ foreach ($groups as $group) {
	2395	+ wp_cache_delete_multiple($ids, $group);
	2396	+ }
	2397	+ }
	2398	+
	2399	+ $this->assert_memory_headroom();
	2400	+ }
	2401	+
	2402	+ /**
	2403	+ * Cache groups get_posts() fills per post for the given post types.
	2404	+ *
	2405	+ * The post, its meta, and one relationships group per taxonomy the post
	2406	+ * type uses (update_object_term_cache()). Term objects themselves are
	2407	+ * bounded by the number of terms, not posts, so they are left cached.
	2408	+ *
	2409	+ * @since 2.10.1
	2410	+ * @param string[] $post_types Post types being walked.
	2411	+ * @return string[] Cache groups keyed by post ID.
	2412	+ */
	2413	+ private function chunk_post_cache_groups(array $post_types): array {
	2414	+ $groups = ['posts', 'post_meta'];
	2415	+
	2416	+ foreach (get_object_taxonomies($post_types) as $taxonomy) {
	2417	+ $groups[] = $taxonomy . '_relationships';
	2418	+ }
	2419	+
	2420	+ return array_values(array_unique($groups));
	2421	+ }
	2422	+
	2423	+ /**
	2424	+ * Stop an automatic rebuild before the memory limit rather than at it.
	2425	+ *
	2426	+ * A PHP memory fatal skips every catch and finally, so the lock, the
	2427	+ * failure record and the backoff are all lost with it, while stopping here
	2428	+ * is an ordinary, fully recorded failure. Checked before each chunk is
	2429	+ * hydrated, against a margin of at least the largest chunk seen so far.
	2430	+ *
	2431	+ * It throws an \Error, not an \Exception, on purpose: the per-segment
	2432	+ * catch (\Exception) blocks in generate_multiple_sitemaps() would otherwise
	2433	+ * swallow it and carry on — writing an index without the aborted segments
	2434	+ * and then pruning their files as orphans. Only the automatic rebuild's
	2435	+ * catch (\Throwable) is meant to see it.
	2436	+ *
	2437	+ * @since 2.10.1
	2438	+ * @return void
	2439	+ * @throws \Error When the automatic rebuild is close to the memory limit.
	2440	+ */
	2441	+ private function assert_memory_headroom(): void {
	2442	+ if (!$this->memory_guard) {
	2443	+ return;
	2444	+ }
	2445	+
	2446	+ $limit = $this->current_memory_limit();
	2447	+ if ($limit === -1) {
	2448	+ return;
	2449	+ }
	2450	+
	2451	+ // A fifth of the limit (at least 32 MB) for writing the files, or one
	2452	+ // and a half of the costliest chunk if that is more — and never more
	2453	+ // than half the limit either way. Without that outer cap a single
	2454	+ // anomalously expensive chunk (500 posts of serialised page-builder or
	2455	+ // ACF meta reaches hundreds of megabytes) puts the margin above the
	2456	+ // limit itself, so every later check aborts at any usage at all, the
	2457	+ // failure records this process's limit, and has_memory_for_retry()
	2458	+ // then refuses every process that has the same limit. A site that
	2459	+ // never actually ran out of memory would stop rebuilding until WP-CLI
	2460	+ // or a system cron happened to run.
	2461	+ $headroom = (int) min(
	2462	+ max(
	2463	+ min(max($limit * 0.2, 32 * MB_IN_BYTES), $limit * 0.5),
	2464	+ $this->walk_chunk_cost * 1.5
	2465	+ ),
	2466	+ $limit * 0.5
	2467	+ );
	2468	+
	2469	+ // The real allocated size, which is what PHP enforces memory_limit
	2470	+ // against; memory_get_usage(false) reports only what is handed out of
	2471	+ // those allocations and so understates the margin by the allocator's
	2472	+ // slack.
	2473	+ $usage = memory_get_usage(true);
	2474	+
	2475	+ if ($usage > $limit - $headroom) {
	2476	+ throw new \Error(
	2477	+ sprintf(
	2478	+ /* translators: 1: memory in use, 2: PHP memory limit. */
	2479	+ esc_html__('The sitemap rebuild was stopped at %1$s of the %2$s PHP memory limit, before PHP would have run out of memory. It will be retried by a process with more memory (WP-CLI or a system cron). To let it finish in the admin, raise the PHP memory_limit.', 'thinkrank'),
	2480	+ esc_html((string) size_format($usage)),
	2481	+ esc_html((string) size_format($limit))
	2482	+ ),
	2483	+ // phpcs:ignore WordPress.Security.EscapeOutput.ExceptionNotEscaped -- an integer class constant, not output.
	2484	+ self::MEMORY_ABORT_CODE
	2485	+ );
	2486	+ }
	2487	+ }
	2488	+
	2489	+ /**
	2490	+ * Run an automatic rebuild's generation with the fatal-safe bookkeeping.
	2491	+ *
	2492	+ * @since 2.10.1
	2493	+ * @param array $settings Sitemap settings.
	2494	+ * @return bool Whatever generate_and_save() returned.
	2495	+ * @throws \Throwable Whatever generation throws, after the memory guard is
	2496	+ * switched back off.
	2497	+ */
	2498	+ private function generate_for_regeneration(array $settings): bool {
	2499	+ // Same headroom wp-admin gives itself; a no-op when the limit is
	2500	+ // already higher or unlimited.
	2501	+ wp_raise_memory_limit('admin');
	2502	+
	2503	+ $this->claim_regeneration_attempt();
	2504	+ $this->memory_guard = true;
	2505	+ $this->walk_chunk_cost = 0;
	2506	+
	2507	+ try {
	2508	+ return $this->generate_and_save($settings);
	2509	+ } finally {
	2510	+ $this->memory_guard = false;
	2511	+ }
	2512	+ }
	2513	+
	2514	+ /**
	2515	+ * Record a failure thrown by an automatic rebuild.
	2516	+ *
	2517	+ * @since 2.10.1
	2518	+ * @param \Throwable $e What was thrown.
	2519	+ * @param string $source Either 'content' or 'settings'.
	2520	+ * @return void
	2521	+ */
	2522	+ private function record_thrown_regeneration_failure(\Throwable $e, string $source): void {
	2523	+ $memory_limit = null;
	2524	+
	2525	+ if ($e instanceof \Error && $e->getCode() === self::MEMORY_ABORT_CODE) {
	2526	+ $memory_limit = $this->current_memory_limit();
	2527	+ $memory_limit = $memory_limit > 0 ? $memory_limit : null;
	2528	+ }
	2529	+
	2530	+ $this->record_regeneration_failure($e->getMessage(), $source, $memory_limit);
	2531	+ }
	2532	+
	2533	+ /**
1855	2534	* Report how automatic regeneration is faring, for the admin UI.
1856	2535	*
1857	2536	* The feature used to fail invisibly: `last_generated` simply stopped
1858	2537	* advancing and nothing drew attention to it (#629).
		@@ -1950,16 +2629,20 @@
1950	2629	* auto_generate setting: the user deliberately changed inclusion rules and
1951	2630	* expects the served file to reflect them even if content-triggered
1952	2631	* auto-generation is turned off. Still respects the master `enabled` flag.
1953	2632	*
1954		- * @return void
	2633	+ * @since 2.10.0 Reports whether the served sitemap was actually rebuilt, so
	2634	+ * a caller can say so rather than assume it (#764). Existing
	2635	+ * callers that ignore the return are unaffected.
	2636	+ *
	2637	+ * @return bool True when the served sitemap now reflects the settings.
1955	2638	*/
1956		- public function regenerate_sitemap_from_settings(): void {
	2639	+ public function regenerate_sitemap_from_settings(): bool {
1957	2640	if (!$this->acquire_generation_lock()) {
1958	2641	// A manual generation (or another request's takeover) is already
1959	2642	// writing the files; the pending marker survives so this rebuild is
1960	2643	// retried rather than lost.
1961		- return;
	2644	+ return false;
1962	2645	}
1963	2646
1964	2647	try {
1965	2648	$settings = $this->get_settings('site');
		@@ -1966,25 +2649,56 @@
1966	2649	if (empty($settings['enabled'])) {
1967	2650	// The sitemap was disabled: remove the previously generated static
1968	2651	// files so the web server stops serving a stale sitemap that
1969	2652	// crawlers would otherwise keep fetching.
1970		- $this->delete_published_sitemaps();
	2653	+ //
	2654	+ // A file that could not be removed is still being served, so
	2655	+ // this is not a success. Reporting one here would tell a caller
	2656	+ // the sitemap was gone while the web server kept answering with
	2657	+ // it, which is the failure this return value exists to prevent
	2658	+ // (#764).
	2659	+ $removal = $this->delete_published_sitemaps($settings);
	2660	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2661	+
	2662	+ if (!empty($stuck)) {
	2663	+ $this->record_regeneration_failure(
	2664	+ $this->stuck_files_message($stuck, true),
	2665	+ 'settings'
	2666	+ );
	2667	+
	2668	+ return false;
	2669	+ }
	2670	+
1971	2671	$this->mark_regeneration_complete();
1972		- return;
	2672	+
	2673	+ return true;
1973	2674	}
1974	2675
1975	2676	$revision = $this->current_regeneration_revision();
1976	2677
1977		- if ($this->generate_and_save($settings)) {
	2678	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2679	+ // Returns false when a static file is stuck in the web root:
	2680	+ // the server keeps serving that file in preference to WordPress,
	2681	+ // so the switch has not taken effect (#764).
	2682	+ return $this->switch_to_dynamic_delivery($settings, $revision, 'settings');
	2683	+ }
	2684	+
	2685	+ if ($this->generate_for_regeneration($settings)) {
1978	2686	$this->mark_regeneration_complete($revision);
1979		- } else {
1980		- $this->record_regeneration_failure(
1981		- __('The sitemap files could not be written to the site root.', 'thinkrank'),
1982		- 'settings'
1983		- );
	2687	+
	2688	+ return true;
1984	2689	}
	2690	+
	2691	+ $this->record_regeneration_failure(
	2692	+ $this->write_failure_message(),
	2693	+ 'settings'
	2694	+ );
	2695	+
	2696	+ return false;
1985	2697	} catch (\Throwable $e) {
1986		- $this->record_regeneration_failure($e->getMessage(), 'settings');
	2698	+ $this->record_thrown_regeneration_failure($e, 'settings');
	2699	+
	2700	+ return false;
1987	2701	} finally {
1988	2702	$this->release_generation_lock();
1989	2703	}
1990	2704	}
		@@ -1989,8 +2703,27 @@
1989	2703	}
1990	2704	}
1991	2705
1992	2706	/**
	2707	+ * When a rebuild has been outstanding since, or 0 when none is.
	2708	+ *
	2709	+ * Lets a caller report an honest "saved, but the served file has not caught
	2710	+ * up yet" instead of a bare success (#764).
	2711	+ *
	2712	+ * @since 2.10.0
	2713	+ * @return int Unix timestamp, or 0 when nothing is pending.
	2714	+ */
	2715	+ public static function regeneration_pending_since(): int {
	2716	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2717	+
	2718	+ if (!is_array($pending) || empty($pending['since'])) {
	2719	+ return 0;
	2720	+ }
	2721	+
	2722	+ return (int) $pending['since'];
	2723	+ }
	2724	+
	2725	+ /**
1993	2726	* Remove every static sitemap file ThinkRank publishes to the web root.
1994	2727	*
1995	2728	* Called when the sitemap feature is disabled, by the cleanup route, and by
1996	2729	* both removal paths, so /sitemap.xml, /sitemap_index.xml, the segmented
		@@ -2089,10 +2822,17 @@
2089	2822	return;
2090	2823	}
2091	2824
2092	2825	$revision = $this->current_regeneration_revision();
	2826	+ $settings = $this->get_settings('site');
2093	2827
2094		- if ($this->generate_and_save($this->get_settings('site'))) {
	2828	+ // See regenerate_sitemap_from_settings(): nothing to write.
	2829	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2830	+ $this->switch_to_dynamic_delivery($settings, $revision, 'content');
	2831	+ return;
	2832	+ }
	2833	+
	2834	+ if ($this->generate_for_regeneration($settings)) {
2095	2835	$this->mark_regeneration_complete($revision);
2096	2836	} else {
2097	2837	// Previously this returned quietly and last_generated simply
2098	2838	// stopped advancing, leaving the site owner with no way to learn
		@@ -2097,14 +2837,14 @@
2097	2837	// Previously this returned quietly and last_generated simply
2098	2838	// stopped advancing, leaving the site owner with no way to learn
2099	2839	// the sitemap had stopped updating (#629).
2100	2840	$this->record_regeneration_failure(
2101		- __('The sitemap files could not be written to the site root.', 'thinkrank'),
	2841	+ $this->write_failure_message(),
2102	2842	'content'
2103	2843	);
2104	2844	}
2105	2845	} catch (\Throwable $e) {
2106		- $this->record_regeneration_failure($e->getMessage(), 'content');
	2846	+ $this->record_thrown_regeneration_failure($e, 'content');
2107	2847	} finally {
2108	2848	$this->release_generation_lock();
2109	2849	}
2110	2850	}
		@@ -2121,8 +2861,26 @@
2121	2861	* @param array $settings Sitemap settings.
2122	2862	* @return bool True when the sitemap files were written.
2123	2863	*/
2124	2864	public function generate_and_save(array $settings): bool {
	2865	+ // Dynamic delivery publishes no files, so writing them here would put a
	2866	+ // static copy back in the web root for the server to serve in place of
	2867	+ // the dynamic route. Guarding at each call site left gaps — the
	2868	+ // snapshot migrator's post-import regeneration had none — so the rule
	2869	+ // lives with the writing instead.
	2870	+ //
	2871	+ // `is_collecting()` is the exception that makes dynamic delivery work
	2872	+ // at all: render_document() and collect_documents() reach this same
	2873	+ // method with the writer swapped for a collector, and that is precisely
	2874	+ // the dynamic build. Only a real write is skipped.
	2875	+ if (!$this->is_collecting() && 'dynamic' === $this->resolve_delivery_mode($settings)) {
	2876	+ // Whatever prompted this call changed the sitemap's content, so the
	2877	+ // rendered copies must not outlive it.
	2878	+ $this->flush_dynamic_cache();
	2879	+
	2880	+ return true;
	2881	+ }
	2882	+
2125	2883	// Index mode is driven by the use_sitemap_index toggle (not merely by how
2126	2884	// many sitemap_urls happen to be configured). When the toggle is on but
2127	2885	// no child sitemaps are set up yet, synthesize the per-type segmented set
2128	2886	// so we emit a real <sitemapindex> with paginated children instead of a
		@@ -2149,9 +2907,13 @@
2149	2907	// names is never touched (#515).
2150	2908	$this->prune_orphaned_segments($settings, [['filename' => basename($primary)]]);
2151	2909	}
2152	2910
2153		- if ($written) {
	2911	+ // last_generated describes what is on disk. A dynamic render publishes
	2912	+ // nothing, so advancing it would report a static publication that never
	2913	+ // happened and would let primary_sitemap_file_exists() callers believe
	2914	+ // there is a file to serve.
	2915	+ if ($written && !$this->is_collecting()) {
2154	2916	$settings['last_generated'] = gmdate('c');
2155	2917	$this->save_settings('site', null, $settings);
2156	2918	}
2157	2919
		@@ -2158,8 +2920,433 @@
2158	2920	return $written;
2159	2921	}
2160	2922
2161	2923	/**
	2924	+ * Whether this instance is rendering documents rather than publishing them.
	2925	+ *
	2926	+ * @since 2.9.0
	2927	+ *
	2928	+ * @return bool
	2929	+ */
	2930	+ private function is_collecting(): bool {
	2931	+ return $this->document_sink !== null;
	2932	+ }
	2933	+
	2934	+ /**
	2935	+ * Complete a regeneration that delivers dynamically, retiring stale files.
	2936	+ *
	2937	+ * Dynamic delivery renders nothing to disk, but that is only half the job.
	2938	+ * A web server hands back an existing `/sitemap.xml` without ever loading
	2939	+ * WordPress, so any file left over from a previous static generation goes on
	2940	+ * being served forever and {@see \ThinkRank\Frontend\SEO_Manager
	2941	+ * ::maybe_serve_sitemap()} is never reached. Switching to dynamic while
	2942	+ * leaving those files in place would therefore appear to do nothing at all.
	2943	+ *
	2944	+ * Both transitions matter and they differ:
	2945	+ *
	2946	+ * - An explicit switch to `dynamic` happens on a site whose root is usually
	2947	+ * still writable, so the files can simply be removed.
	2948	+ * - An `auto` site that becomes read-only cannot remove them, because
	2949	+ * deleting an entry needs write permission on the directory that holds
	2950	+ * it. There the stale sitemap really is stuck in front of us, and the
	2951	+ * honest outcome is a recorded failure naming it rather than a rebuild
	2952	+ * reported as complete (#754 review).
	2953	+ *
	2954	+ * Ownership is tested per file by the shared helper, so another plugin's
	2955	+ * sitemap at one of our names is never deleted (#515).
	2956	+ *
	2957	+ * @since 2.9.0
	2958	+ *
	2959	+ * @param array $settings Sitemap settings.
	2960	+ * @param int $revision Revision this rebuild is completing.
	2961	+ * @param string $source 'settings' or 'content', for the failure record.
	2962	+ * @return void
	2963	+ */
	2964	+ private function switch_to_dynamic_delivery(array $settings, int $revision, string $source): bool {
	2965	+ $this->flush_dynamic_cache();
	2966	+
	2967	+ $removal = $this->delete_published_sitemaps($settings);
	2968	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2969	+
	2970	+ if (!empty($stuck)) {
	2971	+ $this->record_regeneration_failure($this->stuck_files_message($stuck), $source);
	2972	+
	2973	+ return false;
	2974	+ }
	2975	+
	2976	+ $this->mark_regeneration_complete($revision);
	2977	+
	2978	+ return true;
	2979	+ }
	2980	+
	2981	+ /**
	2982	+ * Why a stale file left in the web root means the change has not landed.
	2983	+ *
	2984	+ * Shared by every path that removes published files, so they cannot
	2985	+ * describe the same situation differently (#764).
	2986	+ *
	2987	+ * The two situations that reach it differ in what WordPress is doing, and
	2988	+ * the message has to say which. After a switch to dynamic delivery
	2989	+ * WordPress IS serving the sitemap and the files shadow it. After the
	2990	+ * sitemap is switched off WordPress serves nothing, so the one message
	2991	+ * used to tell a site owner who had just disabled the sitemap that it was
	2992	+ * "being served from WordPress", which is the opposite of what they did.
	2993	+ *
	2994	+ * @since 2.10.0
	2995	+ * @since 2.10.0 Public, so the REST endpoint uses it rather than a copy;
	2996	+ * takes $sitemap_disabled for the disabled path.
	2997	+ *
	2998	+ * @param string[] $stuck Basenames that could not be removed.
	2999	+ * @param bool $sitemap_disabled True when the files outlived disabling
	3000	+ * the sitemap rather than a switch to
	3001	+ * dynamic delivery.
	3002	+ * @return string
	3003	+ */
	3004	+ public function stuck_files_message(array $stuck, bool $sitemap_disabled = false): string {
	3005	+ if ($sitemap_disabled) {
	3006	+ return sprintf(
	3007	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3008	+ __('The sitemap is disabled, but these files are still in the site root and your web server is still serving them: %1$s. They could not be removed because %2$s is not writable. Delete them, or ask your host to make the WordPress root writable.', 'thinkrank'),
	3009	+ implode(', ', $stuck),
	3010	+ untrailingslashit(ABSPATH)
	3011	+ );
	3012	+ }
	3013	+
	3014	+ return sprintf(
	3015	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3016	+ __('The sitemap is being served from WordPress, but these files are still in the site root and your web server will keep serving them instead: %1$s. They could not be removed because %2$s is not writable. Delete them, or ask your host to make the WordPress root writable.', 'thinkrank'),
	3017	+ implode(', ', $stuck),
	3018	+ untrailingslashit(ABSPATH)
	3019	+ );
	3020	+ }
	3021	+
	3022	+ /**
	3023	+ * What to tell the site owner when publishing the files failed.
	3024	+ *
	3025	+ * The old wording stated the symptom and stopped there, so the reported
	3026	+ * cause was a guess and this reached support as a plugin fault rather than
	3027	+ * a folder permission (#752, #753). When the root is demonstrably
	3028	+ * unwritable, say that, and say what to do about it.
	3029	+ *
	3030	+ * @since 2.9.0
	3031	+ *
	3032	+ * @return string
	3033	+ */
	3034	+ private function write_failure_message(): string {
	3035	+ if (!wp_is_writable(ABSPATH)) {
	3036	+ return sprintf(
	3037	+ /* translators: %s: absolute path to the WordPress root. */
	3038	+ __('The sitemap could not be written because the folder %s is not writable by PHP. Ask your host to make the WordPress root writable, or set Sitemap Delivery to Dynamic to serve the sitemap without writing files.', 'thinkrank'),
	3039	+ untrailingslashit(ABSPATH)
	3040	+ );
	3041	+ }
	3042	+
	3043	+ return __('The sitemap files could not be written to the site root.', 'thinkrank');
	3044	+ }
	3045	+
	3046	+ /**
	3047	+ * How this site delivers its sitemap.
	3048	+ *
	3049	+ * `auto` is resolved on whether the web root can be written. That is the
	3050	+ * right signal here (unlike llms.txt, where the question is whether the
	3051	+ * server applies the .htaccess charset block): a site whose root is
	3052	+ * read-only cannot publish a sitemap file at all, and before this existed
	3053	+ * the feature simply failed with "The sitemap files could not be written to
	3054	+ * the site root." and served nothing (#752).
	3055	+ *
	3056	+ * @since 2.9.0
	3057	+ *
	3058	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3059	+ * @return string One of 'static' or 'dynamic'. Never 'auto'.
	3060	+ */
	3061	+ public function resolve_delivery_mode(?array $settings = null): string {
	3062	+ $settings = $settings ?? $this->get_settings('site');
	3063	+ $mode = (string) ($settings['delivery_mode'] ?? 'auto');
	3064	+
	3065	+ if ('static' === $mode || 'dynamic' === $mode) {
	3066	+ return $mode;
	3067	+ }
	3068	+
	3069	+ return wp_is_writable(ABSPATH) ? 'static' : 'dynamic';
	3070	+ }
	3071	+
	3072	+ /**
	3073	+ * Render one published sitemap document without touching the filesystem.
	3074	+ *
	3075	+ * Runs the ordinary build pipeline with the writer swapped for a collector,
	3076	+ * so the bytes returned here are the bytes the static path would have
	3077	+ * written. `SitemapDeliveryParityTest` asserts that equivalence rather than
	3078	+ * trusting it.
	3079	+ *
	3080	+ * The whole set is built to answer for one file, because the index can only
	3081	+ * be assembled from the children that were actually produced. The result is
	3082	+ * cached per document, so that cost is paid once per change and not once
	3083	+ * per crawler request.
	3084	+ *
	3085	+ * @since 2.9.0
	3086	+ *
	3087	+ * @param string $filename Published file name, e.g. 'sitemap.xml'.
	3088	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3089	+ * @return string|null XML, or null when this site does not publish that name.
	3090	+ */
	3091	+ public function render_document(string $filename, ?array $settings = null): ?string {
	3092	+ $filename = basename($filename);
	3093	+ $settings = $settings ?? $this->get_settings('site');
	3094	+
	3095	+ if (empty($settings['enabled'])) {
	3096	+ return null;
	3097	+ }
	3098	+
	3099	+ $cached = get_transient($this->dynamic_cache_key($filename));
	3100	+ if (self::ABSENT_MARKER === $cached) {
	3101	+ return null;
	3102	+ }
	3103	+ if (is_string($cached) && '' !== $cached) {
	3104	+ return $cached;
	3105	+ }
	3106	+
	3107	+ // A miss builds the whole set, because the index can only be assembled
	3108	+ // from the children that were actually produced. Caching only the
	3109	+ // requested document therefore made a crawler walking the index and its
	3110	+ // children rebuild the entire site's sitemap once per file — every post
	3111	+ // and taxonomy query repeated N times on a public endpoint (#754
	3112	+ // review). The set is built once and stored in full.
	3113	+ return $this->stream_documents($settings, $filename);
	3114	+ }
	3115	+
	3116	+ /**
	3117	+ * Build every document, caching each as it is produced, keeping one.
	3118	+ *
	3119	+ * A miss has to build the whole set, because the index can only be
	3120	+ * assembled from the children that were actually produced. It does not have
	3121	+ * to *hold* the whole set: the static path never keeps more than one page
	3122	+ * in memory, writing each to disk as it goes, and buffering every
	3123	+ * document's XML to return one of them undid that on the request path,
	3124	+ * where a large site's entire sitemap corpus would sit in a single PHP
	3125	+ * process (#754 review).
	3126	+ *
	3127	+ * So the sink writes each document straight to its cache entry and lets it
	3128	+ * go, retaining only the one this request is answering. Peak retention is
	3129	+ * one document, whatever the site's size.
	3130	+ *
	3131	+ * Concurrency: the first request through takes a short lock and does the
	3132	+ * work. One that finds the lock held waits a bounded moment for the winner
	3133	+ * to publish, then builds anyway, because serving a correct sitemap late
	3134	+ * beats serving none.
	3135	+ *
	3136	+ * @since 2.9.0
	3137	+ *
	3138	+ * @param array $settings Sitemap settings.
	3139	+ * @param string $wanted Document this request is answering.
	3140	+ * @return string|null XML for $wanted, or null when the site does not publish it.
	3141	+ */
	3142	+ private function stream_documents(array $settings, string $wanted): ?string {
	3143	+ $lock = self::DYNAMIC_CACHE_PREFIX . 'lock';
	3144	+
	3145	+ if (!$this->acquire_render_lock($lock)) {
	3146	+ for ($attempt = 0; $attempt < self::RENDER_LOCK_WAIT_ATTEMPTS; $attempt++) {
	3147	+ usleep(self::RENDER_LOCK_WAIT_MICROSECONDS);
	3148	+
	3149	+ $cached = get_transient($this->dynamic_cache_key($wanted));
	3150	+ if (self::ABSENT_MARKER === $cached) {
	3151	+ return null;
	3152	+ }
	3153	+ if (is_string($cached) && '' !== $cached) {
	3154	+ return $cached;
	3155	+ }
	3156	+ }
	3157	+ }
	3158	+
	3159	+ $kept = null;
	3160	+ // Names only. Keeping the bodies here would be the very retention this
	3161	+ // method exists to avoid.
	3162	+ $produced = [];
	3163	+
	3164	+ $previous = $this->document_sink;
	3165	+ $this->document_sink = function (string $name, string $xml) use (&$kept, &$produced, $wanted): void {
	3166	+ $produced[$name] = true;
	3167	+ set_transient($this->dynamic_cache_key($name), $xml, self::DYNAMIC_CACHE_TTL);
	3168	+
	3169	+ if ($name === $wanted) {
	3170	+ $kept = $xml;
	3171	+ }
	3172	+ };
	3173	+
	3174	+ try {
	3175	+ $this->generate_and_save($settings);
	3176	+
	3177	+ // Names the configuration lists but this build did not produce get
	3178	+ // a negative entry, so asking for one again is a cache hit rather
	3179	+ // than another full rebuild.
	3180	+ $absent = $this->published_document_names($settings);
	3181	+
	3182	+ // Also the exact name this request asked for: a paginated page past
	3183	+ // the end of a stem is a legitimate request shape that the base
	3184	+ // list cannot enumerate, and without an entry it would rebuild on
	3185	+ // every hit.
	3186	+ $absent[] = $wanted;
	3187	+
	3188	+ foreach (array_unique($absent) as $name) {
	3189	+ if (!isset($produced[$name])) {
	3190	+ set_transient($this->dynamic_cache_key($name), self::ABSENT_MARKER, self::DYNAMIC_CACHE_TTL);
	3191	+ }
	3192	+ }
	3193	+ } finally {
	3194	+ $this->document_sink = $previous;
	3195	+ delete_transient($lock);
	3196	+ }
	3197	+
	3198	+ return $kept;
	3199	+ }
	3200	+
	3201	+ /**
	3202	+ * Take the render lock, if it is free.
	3203	+ *
	3204	+ * Not atomic across processes, and deliberately so: the fallback for losing
	3205	+ * a race is duplicated work, never a wrong or missing sitemap, so a
	3206	+ * heavier primitive would buy nothing here.
	3207	+ *
	3208	+ * @since 2.9.0
	3209	+ *
	3210	+ * @param string $lock Lock transient name.
	3211	+ * @return bool True when this request holds the lock.
	3212	+ */
	3213	+ private function acquire_render_lock(string $lock): bool {
	3214	+ if (false !== get_transient($lock)) {
	3215	+ return false;
	3216	+ }
	3217	+
	3218	+ set_transient($lock, time(), self::RENDER_LOCK_TTL);
	3219	+
	3220	+ return true;
	3221	+ }
	3222	+
	3223	+ /**
	3224	+ * Build every document this site publishes and return them all.
	3225	+ *
	3226	+ * Verification and tooling only. This retains the whole set in memory, so
	3227	+ * it must never be used to answer a request: {@see self::stream_documents()}
	3228	+ * is the serving path and keeps one document at a time regardless of site
	3229	+ * size (#754 review). `SitemapDeliveryParityTest` enforces that separation
	3230	+ * by failing if the request path routes back through here.
	3231	+ *
	3232	+ * @since 2.9.0
	3233	+ *
	3234	+ * @param array $settings Sitemap settings.
	3235	+ * @return array<string,string> Filename => XML.
	3236	+ */
	3237	+ public function collect_documents(array $settings): array {
	3238	+ $documents = [];
	3239	+
	3240	+ $previous = $this->document_sink;
	3241	+ $this->document_sink = static function (string $name, string $xml) use (&$documents): void {
	3242	+ $documents[$name] = $xml;
	3243	+ };
	3244	+
	3245	+ try {
	3246	+ $this->generate_and_save($settings);
	3247	+ } finally {
	3248	+ $this->document_sink = $previous;
	3249	+ }
	3250	+
	3251	+ return $documents;
	3252	+ }
	3253	+
	3254	+ /**
	3255	+ * The file names this site publishes, without building their contents.
	3256	+ *
	3257	+ * Used by the request router to decide whether a URL is ours before doing
	3258	+ * any work. Cheap: it reads the configured child list rather than querying
	3259	+ * for entries.
	3260	+ *
	3261	+ * @since 2.9.0
	3262	+ *
	3263	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3264	+ * @return string[] File names, including paginated pages that may exist.
	3265	+ */
	3266	+ public function published_document_names(?array $settings = null): array {
	3267	+ $settings = $settings ?? $this->get_settings('site');
	3268	+ $resolved = $this->maybe_promote_to_index($settings);
	3269	+
	3270	+ $names = [$this->get_primary_sitemap_filename($settings), 'local-sitemap.xml'];
	3271	+
	3272	+ foreach ((array) ($resolved['sitemap_urls'] ?? []) as $child) {
	3273	+ if (!is_array($child) || empty($child['enabled'])) {
	3274	+ continue;
	3275	+ }
	3276	+
	3277	+ $path = (string) wp_parse_url((string) ($child['url'] ?? ''), PHP_URL_PATH);
	3278	+ if ('' !== $path) {
	3279	+ $names[] = basename($path);
	3280	+ }
	3281	+ }
	3282	+
	3283	+ return array_values(array_unique(array_filter($names)));
	3284	+ }
	3285	+
	3286	+ /**
	3287	+ * Does this site publish a document under that name?
	3288	+ *
	3289	+ * Not a plain membership test against {@see self::published_document_names()}:
	3290	+ * that lists the configured children, and a child over the per-file URL cap
	3291	+ * is split into `<stem>-2.xml`, `<stem>-3.xml` and so on, with every page
	3292	+ * listed in the index. Gating the request router on the base list alone
	3293	+ * therefore 404'd exactly the pages the index points at, which is worse than
	3294	+ * not serving them at all.
	3295	+ *
	3296	+ * Page counts are not knowable without building, so the stem is what is
	3297	+ * matched; a page that does not exist is answered by the build finding
	3298	+ * nothing for it, and is then cached as absent.
	3299	+ *
	3300	+ * @since 2.9.0
	3301	+ *
	3302	+ * @param string $name Requested file name.
	3303	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3304	+ * @return bool
	3305	+ */
	3306	+ public function publishes_document_name(string $name, ?array $settings = null): bool {
	3307	+ $names = $this->published_document_names($settings);
	3308	+
	3309	+ if (in_array($name, $names, true)) {
	3310	+ return true;
	3311	+ }
	3312	+
	3313	+ if (!preg_match('/^(.*)-\d+\.xml$/i', $name, $m)) {
	3314	+ return false;
	3315	+ }
	3316	+
	3317	+ return in_array($m[1] . '.xml', $names, true);
	3318	+ }
	3319	+
	3320	+ /**
	3321	+ * Transient key for a rendered document.
	3322	+ *
	3323	+ * @since 2.9.0
	3324	+ *
	3325	+ * @param string $filename Published file name.
	3326	+ * @return string
	3327	+ */
	3328	+ private function dynamic_cache_key(string $filename): string {
	3329	+ return self::DYNAMIC_CACHE_PREFIX . md5($filename);
	3330	+ }
	3331	+
	3332	+ /**
	3333	+ * Drop every cached dynamic document.
	3334	+ *
	3335	+ * Called from the same places that mark the static files stale, so the two
	3336	+ * delivery modes invalidate on identical triggers.
	3337	+ *
	3338	+ * @since 2.9.0
	3339	+ *
	3340	+ * @return void
	3341	+ */
	3342	+ public function flush_dynamic_cache(): void {
	3343	+ foreach ($this->published_document_names() as $name) {
	3344	+ delete_transient($this->dynamic_cache_key($name));
	3345	+ }
	3346	+ }
	3347	+
	3348	+ /**
2162	3349	* Resolve index-vs-single mode, synthesizing child sitemaps when needed.
2163	3350	*
2164	3351	* - When use_sitemap_index is on but no child sitemaps are configured, build
2165	3352	* the per-type segmented set so a real <sitemapindex> is produced (#127).
		@@ -2354,8 +3541,18 @@
2354	3541	// File validation failed - error details available in exception
2355	3542	return false;
2356	3543	}
2357	3544
	3545	+ // Dynamic delivery: hand the document to the collector instead of the
	3546	+ // filesystem. Reported as published, because for this run it is — the
	3547	+ // caller's success/failure bookkeeping and the index assembly both key
	3548	+ // off this return value.
	3549	+ if ($this->document_sink !== null) {
	3550	+ ($this->document_sink)($filename, $sitemap_xml);
	3551	+
	3552	+ return true;
	3553	+ }
	3554	+
2358	3555	$sitemap_path = ABSPATH . $filename;
2359	3556
2360	3557	// Use WordPress filesystem API for better security
2361	3558	global $wp_filesystem;
		@@ -2430,8 +3627,38 @@
2430	3627
2431	3628	$urls[] = $this->build_child_sitemap_entry($cpt, $pattern);
2432	3629	}
2433	3630
	3631	+ // Public custom taxonomies get the same treatment (#690). Flat mode has
	3632	+ // always walked them through get_enabled_taxonomies(); index mode built
	3633	+ // its children from the list above and never consulted a taxonomy at
	3634	+ // all, so every custom-taxonomy archive silently vanished from the
	3635	+ // sitemap the moment a site switched modes — and the per-taxonomy switch
	3636	+ // the matrix writes had nothing to act on. Same two tests the post-type
	3637	+ // walk applies, in the same order.
	3638	+ $taken = array_column($urls, 'type');
	3639	+
	3640	+ foreach (get_taxonomies(['public' => true, '_builtin' => false], 'names') as $taxonomy) {
	3641	+ if (!$this->should_include_taxonomy($taxonomy)) {
	3642	+ continue;
	3643	+ }
	3644	+
	3645	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $inclusions)) {
	3646	+ continue;
	3647	+ }
	3648	+
	3649	+ // Post types and taxonomies are separate registries, so a site can
	3650	+ // hold both a `foo` post type and a `foo` taxonomy. They would
	3651	+ // resolve to one filename, and stream_type_entries() answers post
	3652	+ // types first, so the second child would list the first one's file
	3653	+ // twice in the index rather than adding anything.
	3654	+ if (in_array($taxonomy, $taken, true)) {
	3655	+ continue;
	3656	+ }
	3657	+
	3658	+ $urls[] = $this->build_child_sitemap_entry($taxonomy, $pattern);
	3659	+ }
	3660	+
2434	3661	return $urls;
2435	3662	}
2436	3663
2437	3664	/**
		@@ -2588,8 +3815,22 @@
2588	3815	&& !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $type, $settings)) {
2589	3816	continue;
2590	3817	}
2591	3818
	3819	+ // The taxonomy counterpart of the two guards above (#690). A child
	3820	+ // list saved while a taxonomy was still included keeps naming it, so
	3821	+ // without this, excluding one in the matrix would still rewrite and
	3822	+ // re-list the file the user asked not to have. The built-in
	3823	+ // aggregates are named 'categories'/'tags' rather than
	3824	+ // 'category'/'post_tag', so taxonomy_exists() leaves them to the
	3825	+ // inclusion-flag check below.
	3826	+ $child_taxonomy = self::CHILD_TYPE_ALIASES[$type] ?? $type;
	3827	+ if (taxonomy_exists($child_taxonomy)
	3828	+ && (!$this->should_include_taxonomy($child_taxonomy)
	3829	+ || !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $child_taxonomy, $settings))) {
	3830	+ continue;
	3831	+ }
	3832	+
2592	3833	// The built-in aggregates carry their switch in the inclusion flag
2593	3834	// rather than under a post type name, and they are the four most
2594	3835	// people actually use. build_segmented_sitemap_urls() drops a child
2595	3836	// whose flag is empty; regenerating from a list saved while it was
		@@ -2641,12 +3882,30 @@
2641	3882	$buffer = [];
2642	3883	}
2643	3884	}
2644	3885
2645		- // Flush the trailing partial page, or a single empty page when the
2646		- // type had no entries at all (parity with the previous behavior of
2647		- // always writing at least one page per configured child).
2648		- if (!empty($buffer) || $page === 0) {
	3886	+ // Flush the trailing partial page.
	3887	+ //
	3888	+ // A type that produced nothing writes no page at all in index
	3889	+ // mode (#836). It used to write one empty urlset and list it in
	3890	+ // the index, so a crawler was asked to fetch a file that
	3891	+ // answers with no URLs — Search Console reports an empty
	3892	+ // sitemap referenced from an index as a warning, and the fetch
	3893	+ // is wasted on every pass. On this site three of eleven
	3894	+ // children were empty: a post type with nothing published and
	3895	+ // two taxonomies with no terms.
	3896	+ //
	3897	+ // Single-file mode still writes its one page even when empty,
	3898	+ // because that file IS the site's /sitemap.xml and a 404 there
	3899	+ // is worse than an empty urlset. Nothing lists it, so it costs
	3900	+ // no crawl budget.
	3901	+ //
	3902	+ // Skipping the write also keeps the filename out of
	3903	+ // $results['sitemaps_generated'], which is what
	3904	+ // prune_orphaned_segments() treats as "written this run" — so
	3905	+ // a type that empties after previously publishing has its
	3906	+ // stale file deleted rather than left serving.
	3907	+ if (!empty($buffer) || ($page === 0 && $index_config === null)) {
2649	3908	$page++;
2650	3909	$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children);
2651	3910	}
2652	3911
		@@ -2725,8 +3984,15 @@
2725	3984	* @param array $generated Entries from $results['sitemaps_generated'].
2726	3985	* @return string[] Basenames removed.
2727	3986	*/
2728	3987	private function prune_orphaned_segments(array $settings, array $generated): array {
	3988	+ // Rendering for a request, not publishing: there is nothing on disk
	3989	+ // this run owns, and a dynamic render must never delete the files a
	3990	+ // site's previous static mode left behind.
	3991	+ if ($this->is_collecting()) {
	3992	+ return [];
	3993	+ }
	3994	+
2729	3995	$kept = [];
2730	3996	foreach ($generated as $entry) {
2731	3997	if (!empty($entry['filename'])) {
2732	3998	$kept[strtolower((string) $entry['filename'])] = true;
		@@ -2828,9 +4094,9 @@
2828	4094	// publishes under too (this method mirrors it deliberately), so on
2829	4095	// a migrated site the file at that path may never have been ours
2830	4096	// to delete (#515).
2831	4097	$path = ABSPATH . 'local-sitemap.xml';
2832		- if (file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) {
	4098	+ if (!$this->is_collecting() && file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) {
2833	4099	wp_delete_file($path);
2834	4100	}
2835	4101	return false;
2836	4102	}
		@@ -2852,8 +4118,25 @@
2852	4118	*
2853	4119	* @since 1.15.x
2854	4120	* @return array Zero or one URL entry
2855	4121	*/
	4122	+ /**
	4123	+ * Does this site publish a local business sitemap right now?
	4124	+ *
	4125	+ * The same gate {@see self::regenerate_local_sitemap()} applies, asked
	4126	+ * without writing anything. Callers that need to know whether the document
	4127	+ * exists must not test the filesystem: under dynamic delivery it is served
	4128	+ * from PHP and there is no file, which is how `local-sitemap.xml` came to be
	4129	+ * dropped from robots.txt on exactly those sites (#752).
	4130	+ *
	4131	+ * @since 2.9.0
	4132	+ *
	4133	+ * @return bool True when the local sitemap has content to publish.
	4134	+ */
	4135	+ public function publishes_local_sitemap(): bool {
	4136	+ return !empty($this->collect_local_entries());
	4137	+ }
	4138	+
2856	4139	private function collect_local_entries(): array {
2857	4140	if (!class_exists('ThinkRank\\SEO\\Site_Identity_Manager')) {
2858	4141	require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-site-identity-manager.php';
2859	4142	}
		@@ -2925,17 +4208,43 @@
2925	4208	case 'products':
2926	4209	$entries = post_type_exists('product') ? $this->collect_post_entries_iter(['product'], $settings) : [];
2927	4210	return ['entries' => $entries, 'image_ns' => true];
2928	4211
2929		- case 'product_categories':
2930		- $entries = taxonomy_exists('product_cat') ? $this->collect_taxonomy_entries_iter('product_cat', $settings) : [];
2931		- return ['entries' => $entries, 'image_ns' => false];
	4212	+ default:
	4213	+ // Resolve a preset's display name to the object it streams, so
	4214	+ // 'product_categories' is an ordinary taxonomy child rather than
	4215	+ // a case of its own (#690).
	4216	+ $alias = self::CHILD_TYPE_ALIASES[$type] ?? null;
	4217	+ $object = $alias ?? $type;
2932	4218
2933		- default:
2934	4219	$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names');
2935		- if (in_array($type, $custom_post_types, true)) {
2936		- return ['entries' => $this->collect_post_entries_iter([$type], $settings), 'image_ns' => true];
	4220	+ if (in_array($object, $custom_post_types, true)) {
	4221	+ return ['entries' => $this->collect_post_entries_iter([$object], $settings), 'image_ns' => true];
2937	4222	}
	4223	+
	4224	+ // Custom taxonomies reach index mode here, streamed through the
	4225	+ // same iterator flat mode uses so the two modes emit identical
	4226	+ // URLs for the same settings.
	4227	+ if (taxonomy_exists($object) && $this->should_include_taxonomy($object)) {
	4228	+ return ['entries' => $this->collect_taxonomy_entries_iter($object, $settings), 'image_ns' => false];
	4229	+ }
	4230	+
	4231	+ // An aliased child whose object is gone (WooCommerce deactivated)
	4232	+ // keeps writing the empty file it always wrote. Returning null
	4233	+ // here would hand it the whole-site fallback instead, dumping
	4234	+ // every URL on the site into a file named for products.
	4235	+ //
	4236	+ // A registered taxonomy this generator will not emit
	4237	+ // (`post_format`, `nav_menu`, a non-public one) needs the same
	4238	+ // answer for the same reason. generate_multiple_sitemaps()
	4239	+ // skips those before they reach here, so nothing takes this
	4240	+ // path today — but it is the one branch where falling through
	4241	+ // to null is silently catastrophic rather than merely wrong,
	4242	+ // and the guard keeping it unreachable lives in another method.
	4243	+ if ($alias !== null || taxonomy_exists($object)) {
	4244	+ return ['entries' => [], 'image_ns' => false];
	4245	+ }
	4246	+
2938	4247	return null;
2939	4248	}
2940	4249	}
2941	4250
		@@ -2988,8 +4297,13 @@
2988	4297	* @param array $settings Sitemap settings, for the ownership test.
2989	4298	* @return void
2990	4299	*/
2991	4300	private function cleanup_stale_pages(string $base_url, int $current_pages, array $settings): void {
	4301	+ // See prune_orphaned_segments(): a dynamic render deletes nothing.
	4302	+ if ($this->is_collecting()) {
	4303	+ return;
	4304	+ }
	4305	+
2992	4306	$filename = basename(wp_parse_url($base_url, PHP_URL_PATH));
2993	4307	if (!preg_match('/^(.*)\.xml$/i', $filename, $m)) {
2994	4308	return;
2995	4309	}
		@@ -3114,9 +4428,9 @@
3114	4428	* @param array $settings Sitemap settings
3115	4429	* @return string Sitemap index XML
3116	4430	*/
3117	4431	private function generate_sitemap_index(array $children, array $settings): string {
3118		- $xml = $this->xml_prolog($settings, 'sitemap-index.xsl');
	4432	+ $xml = $this->xml_prolog($settings, 'index');
3119	4433	$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
3120	4434
3121	4435	$site_url = home_url();
3122	4436
		@@ -3129,9 +4443,9 @@
3129	4443	$sitemap_url = $site_url . $sitemap_url;
3130	4444	}
3131	4445
3132	4446	$xml .= " <sitemap>\n";
3133		- $xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n";
	4447	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($sitemap_url)) . "</loc>\n";
3134	4448	$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n";
3135	4449	$xml .= " </sitemap>\n";
3136	4450	}
3137	4451