← All changes | includes/seo/class-sitemap-generator.php +1781 -230 2.0.0 → 2.9.0 View file →

		@@ -21,8 +21,13 @@
21	21	if ( ! defined( 'ABSPATH' ) ) {
22	22	exit;
23	23	}
24	24
	25	+// Filename derivation and web-root removal are shared with the deactivator and
	26	+// with uninstall.php, which runs without an autoloader — so they live in a
	27	+// plain function file both can require. See includes/cleanup-webroot.php.
	28	+require_once __DIR__ . '/../cleanup-webroot.php';
	29	+
25	30	/**
26	31	* Sitemap Generator Class
27	32	*
28	33	* Generates and manages XML sitemaps for search engine optimization.
		@@ -33,13 +38,221 @@
33	38	*/
34	39	class Sitemap_Generator extends Abstract_SEO_Manager {
35	40
36	41	/**
	42	+ * Memoised WooCommerce page IDs kept out of the sitemap. Null until resolved.
	43	+ *
	44	+ * @since 2.0.1
	45	+ * @var int[]|null
	46	+ */
	47	+ private ?array $woocommerce_excluded_page_ids = null;
	48	+
	49	+ /**
	50	+ * Inclusion flag -> the child sitemap it controls.
	51	+ *
	52	+ * Shared by the child-list builder and by the "did the caller change what
	53	+ * the sitemap includes?" check in maybe_promote_to_index().
	54	+ *
	55	+ * @since 1.31.0
	56	+ * @var array<string, string>
	57	+ */
	58	+ private const INCLUSION_CHILD_TYPES = [
	59	+ 'include_posts' => 'posts',
	60	+ 'include_pages' => 'pages',
	61	+ 'include_categories' => 'categories',
	62	+ 'include_tags' => 'tags',
	63	+ ];
	64	+
	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	+ /**
37	83	* Supported sitemap types
38	84	*
39	85	* @since 1.0.0
40	86	* @var array
41	87	*/
	88	+ /**
	89	+ * How many post IDs to hydrate at a time while walking the sitemap set.
	90	+ *
	91	+ * @since 2.0.1
	92	+ * @var int
	93	+ */
	94	+ private const ID_WALK_CHUNK = 500;
	95	+
	96	+ /**
	97	+ * How long a content or settings change is debounced before the sitemap is
	98	+ * rebuilt, in seconds. Coalesces bulk edits into a single regeneration.
	99	+ *
	100	+ * @since 2.2.1
	101	+ * @var int
	102	+ */
	103	+ private const REGENERATION_DEBOUNCE = 30;
	104	+
	105	+ /**
	106	+ * How far past its due time the scheduled rebuild may sit before a request
	107	+ * takes it over, in seconds.
	108	+ *
	109	+ * WP-Cron is request-driven, so on a site running DISABLE_WP_CRON, blocking
	110	+ * loopback requests, or seeing very little traffic the event never fires and
	111	+ * the sitemap silently stops updating (#629). The grace keeps the fast path
	112	+ * (cron) in charge under normal conditions.
	113	+ *
	114	+ * @since 2.2.1
	115	+ * @var int
	116	+ */
	117	+ private const REGENERATION_TAKEOVER_GRACE = 120;
	118	+
	119	+ /**
	120	+ * Longest backoff between takeover attempts after a failed rebuild, so a
	121	+ * persistently failing generation cannot run on every admin request.
	122	+ *
	123	+ * @since 2.2.1
	124	+ * @var int
	125	+ */
	126	+ private const REGENERATION_MAX_BACKOFF = 3600;
	127	+
	128	+ /**
	129	+ * Option holding the rebuild that content/settings changes are still waiting
	130	+ * on: `since`, `source` ('content'|'settings'), `attempts`, `next_attempt`.
	131	+ * Absent means the served sitemap is up to date with what triggered it.
	132	+ *
	133	+ * @since 2.2.1
	134	+ * @var string
	135	+ */
	136	+ public const REGENERATION_PENDING_OPTION = 'thinkrank_sitemap_regeneration_pending';
	137	+
	138	+ /**
	139	+ * Option holding the last automatic-regeneration failure (`message`,
	140	+ * `source`, `time`), so the failure is visible instead of swallowed.
	141	+ *
	142	+ * @since 2.2.1
	143	+ * @var string
	144	+ */
	145	+ public const REGENERATION_ERROR_OPTION = 'thinkrank_sitemap_regeneration_error';
	146	+
	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	+ /**
	238	+ * Transient guarding against two generations running at once. Shared with
	239	+ * Sitemap_Endpoint's manual generate route so an automatic rebuild and a
	240	+ * manual one cannot write the same files concurrently.
	241	+ *
	242	+ * @since 2.2.1
	243	+ * @var string
	244	+ */
	245	+ public const GENERATION_LOCK_TRANSIENT = 'thinkrank_sitemap_generation_lock';
	246	+
	247	+ /**
	248	+ * How many term IDs to hydrate at a time while walking a taxonomy.
	249	+ *
	250	+ * @since 2.0.1
	251	+ * @var int
	252	+ */
	253	+ private const TERM_WALK_CHUNK = 1000;
	254	+
42	255	private array $sitemap_types = [
43	256	'posts' => [
44	257	'name' => 'Posts',
45	258	'post_types' => ['post'],
		@@ -164,15 +377,10 @@
164	377	*/
165	378	public function generate_sitemap(array $options = []): string {
166	379	$settings = $this->get_settings('site');
167	380
168		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	381	+ $xml = $this->xml_prolog($settings, 'sitemap');
169	382
170		- // Add XSL stylesheet only if styling is enabled
171		- if (!empty($settings['enable_styling'])) {
172		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
173		- }
174		-
175	383	// Add image namespace if images are enabled
176	384	if (!empty($settings['include_images'])) {
177	385	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
178	386	} else {
		@@ -284,8 +492,34 @@
284	492	];
285	493	}
286	494
287	495	/**
	496	+ * Sitemap keys outside the defaults.
	497	+ *
	498	+ * @since 2.0.1
	499	+ *
	500	+ * @return string[]
	501	+ */
	502	+ protected function additional_setting_keys(): array {
	503	+ return ['selected_preset'];
	504	+ }
	505	+
	506	+ /**
	507	+ * Inclusion flags are per post type and per taxonomy.
	508	+ *
	509	+ * A site registering a `product` post type stores `include_product`; an
	510	+ * enumerated list would go stale on the next registration, so the family
	511	+ * is matched instead.
	512	+ *
	513	+ * @since 2.0.1
	514	+ *
	515	+ * @return string[]
	516	+ */
	517	+ protected function dynamic_setting_key_patterns(): array {
	518	+ return ['/^include_[a-z0-9_]+$/', '/^exclude_[a-z0-9_]+$/'];
	519	+ }
	520	+
	521	+ /**
288	522	* Get default settings for a context type (implements interface)
289	523	*
290	524	* @since 1.0.0
291	525	*
		@@ -308,8 +542,15 @@
308	542	]
309	543	],
310	544	'use_sitemap_index' => false,
311	545
	546	+ // How the sitemap reaches crawlers. 'auto' keeps the historical
	547	+ // behaviour wherever the web root is writable, and only falls back
	548	+ // to serving the sitemap from PHP where writing a file is
	549	+ // impossible — previously a hard failure with nothing served
	550	+ // (#752).
	551	+ 'delivery_mode' => 'auto',
	552	+
312	553	// General Settings
313	554	'links_per_sitemap' => 1000,
314	555	'include_images' => true,
315	556	'include_featured_images' => false,
		@@ -331,8 +572,17 @@
331	572	// Advanced Options
332	573	'enable_styling' => true,
333	574	'custom_url_pattern' => 'sitemap-{type}.xml',
334	575
	576	+ // Stylesheet branding (#639). Both colours default to empty, not
	577	+ // to the stock hexes: empty means the stylesheet's own value
	578	+ // stands, so a site that never opens this screen renders exactly
	579	+ // as it did before the setting existed.
	580	+ 'styling_logo' => false,
	581	+ 'styling_logo_url' => '',
	582	+ 'styling_color_main' => '',
	583	+ 'styling_color_accent' => '',
	584	+
335	585	// Generation tracking
336	586	'last_generated' => ''
337	587	];
338	588	}
		@@ -337,8 +587,49 @@
337	587	];
338	588	}
339	589
340	590	/**
	591	+ * Normalize the stylesheet branding values on the way into the store.
	592	+ *
	593	+ * The generic sanitizer only runs `sanitize_text_field()` over a string,
	594	+ * which happily keeps "red" or "rebeccapurple" as a colour. Nothing
	595	+ * downstream can use those — {@see Sitemap_Stylesheet::render()} skips any
	596	+ * value it cannot read as a hex colour — so storing them would report a
	597	+ * successful save of a setting that changes nothing, and `get-sitemap-
	598	+ * settings` would hand an agent back a colour the sitemap does not use.
	599	+ * Reducing here instead keeps the store and the rendering in agreement.
	600	+ *
	601	+ * @since 2.7.0
	602	+ *
	603	+ * @param array $settings Settings to sanitize.
	604	+ * @param string $context_type Context the save is for.
	605	+ * @return array
	606	+ */
	607	+ protected function sanitize_settings(array $settings, string $context_type = 'site'): array {
	608	+ $sanitized = parent::sanitize_settings($settings, $context_type);
	609	+
	610	+ foreach (['styling_color_main', 'styling_color_accent'] as $key) {
	611	+ if (array_key_exists($key, $sanitized)) {
	612	+ $sanitized[$key] = Sitemap_Stylesheet::hex($sanitized[$key]);
	613	+ }
	614	+ }
	615	+
	616	+ if (array_key_exists('styling_logo_url', $sanitized)) {
	617	+ $sanitized['styling_logo_url'] = esc_url_raw((string) $sanitized['styling_logo_url']);
	618	+ }
	619	+
	620	+ // A mode this build cannot act on has to be stored as the fallback
	621	+ // rather than kept verbatim, or get-sitemap-settings reports a delivery
	622	+ // mode the site does not actually apply.
	623	+ if (array_key_exists('delivery_mode', $sanitized)) {
	624	+ $mode = sanitize_key((string) $sanitized['delivery_mode']);
	625	+ $sanitized['delivery_mode'] = in_array($mode, self::DELIVERY_MODES, true) ? $mode : 'auto';
	626	+ }
	627	+
	628	+ return $sanitized;
	629	+ }
	630	+
	631	+ /**
341	632	* Get settings schema definition (implements interface)
342	633	*
343	634	* @since 1.0.0
344	635	*
		@@ -352,8 +643,15 @@
352	643	'title' => 'Enable Sitemap',
353	644	'description' => 'Generate XML sitemap for search engines',
354	645	'default' => true
355	646	],
	647	+ 'delivery_mode' => [
	648	+ 'type' => 'string',
	649	+ 'title' => 'Sitemap Delivery',
	650	+ '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',
	651	+ 'enum' => self::DELIVERY_MODES,
	652	+ 'default' => 'auto'
	653	+ ],
356	654	'include_posts' => [
357	655	'type' => 'boolean',
358	656	'title' => 'Include Posts',
359	657	'description' => 'Include blog posts in sitemap',
		@@ -394,8 +692,32 @@
394	692	'type' => 'string',
395	693	'title' => 'Last Generated',
396	694	'description' => 'Timestamp of last sitemap generation',
397	695	'default' => ''
	696	+ ],
	697	+ 'styling_logo' => [
	698	+ 'type' => 'boolean',
	699	+ 'title' => 'Show Logo On Sitemap',
	700	+ 'description' => 'Show a logo above the sitemap heading',
	701	+ 'default' => false
	702	+ ],
	703	+ 'styling_logo_url' => [
	704	+ 'type' => 'string',
	705	+ 'title' => 'Sitemap Logo',
	706	+ 'description' => 'Logo image URL. Empty falls back to the site icon',
	707	+ 'default' => ''
	708	+ ],
	709	+ 'styling_color_main' => [
	710	+ 'type' => 'string',
	711	+ 'title' => 'Sitemap Main Color',
	712	+ 'description' => 'Hex color for the sitemap header, links and table head. Empty keeps the stock palette',
	713	+ 'default' => ''
	714	+ ],
	715	+ 'styling_color_accent' => [
	716	+ 'type' => 'string',
	717	+ 'title' => 'Sitemap Accent Color',
	718	+ 'description' => 'Hex color for the header gradient and link hovers. Empty keeps the stock palette',
	719	+ 'default' => ''
398	720	]
399	721	];
400	722	}
401	723
		@@ -412,9 +734,14 @@
412	734	* @return string XML URL entry
413	735	*/
414	736	private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string {
415	737	$xml = " <url>\n";
416		- $xml .= " <loc>" . esc_url($url) . "</loc>\n";
	738	+ // Every <loc> in every sitemap passes through here, which is why the
	739	+ // scheme preference is applied at this one point rather than at each
	740	+ // of the dozen collectors that build URLs (#638). An http sitemap on
	741	+ // an https site hands search engines the wrong address for the whole
	742	+ // site at once.
	743	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($url)) . "</loc>\n";
417	744	// Omit <lastmod> when unknown (empty) — a fabricated timestamp is worse
418	745	// than no timestamp, and an absent lastmod is valid per the spec.
419	746	if (!empty($lastmod)) {
420	747	$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n";
		@@ -424,9 +751,9 @@
424	751
425	752	// Add image entries if provided
426	753	foreach ($images as $image) {
427	754	$xml .= " <image:image>\n";
428		- $xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n";
	755	+ $xml .= " <image:loc>" . esc_url(Url_Scheme::apply((string) $image['url'])) . "</image:loc>\n";
429	756
430	757	if (!empty($image['title'])) {
431	758	$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n";
432	759	}
		@@ -539,9 +866,18 @@
539	866	if (empty($all_ids)) {
540	867	return;
541	868	}
542	869
543		- foreach (array_chunk($all_ids, 500) as $chunk) {
	870	+ // Walk the ID list with a moving window rather than array_chunk().
	871	+ // array_chunk() builds a second array holding every element again, so
	872	+ // peak memory was twice the ID list — on a 100k-post site that is ~16MB
	873	+ // where ~8MB is needed, and this walk is the one part of an otherwise
	874	+ // well-bounded routine with no ceiling (#402).
	875	+ $total = count($all_ids);
	876	+
	877	+ for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) {
	878	+ $chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK);
	879	+
544	880	$posts = get_posts($this->filter_query_args([
545	881	'post_type' => $post_types,
546	882	'post_status' => 'publish',
547	883	'numberposts' => count($chunk),
		@@ -550,9 +886,23 @@
550	886	]));
551	887
552	888	foreach ($posts as $post) {
553	889	if ($this->should_include_in_sitemap($post, $settings)) {
554		- $url = get_permalink($post);
	890	+ /**
	891	+ * Filter a sitemap entry's permalink.
	892	+ *
	893	+ * The multilingual manager uses this to generate each
	894	+ * translation's URL in its OWN language: the sitemap query
	895	+ * deliberately runs with suppress_filters, and the cron
	896	+ * rebuild runs with no language context at all, so a bare
	897	+ * get_permalink() resolved every translation to the
	898	+ * default-language URL — N entries sharing one <loc> (#409).
	899	+ *
	900	+ * @since 2.0.1
	901	+ * @param string $url Permalink as WordPress resolved it.
	902	+ * @param \WP_Post $post Post the entry describes.
	903	+ */
	904	+ $url = apply_filters('thinkrank_sitemap_post_permalink', get_permalink($post), $post);
555	905	$lastmod = gmdate('c', strtotime($post->post_modified_gmt));
556	906	$priority = $this->calculate_intelligent_priority($post, $post->post_type);
557	907	$changefreq = $this->calculate_change_frequency($post, $post->post_type);
558	908	$images = $this->extract_post_images($post, $settings);
		@@ -619,9 +969,14 @@
619	969	if (is_wp_error($all_ids) || empty($all_ids)) {
620	970	return;
621	971	}
622	972
623		- foreach (array_chunk($all_ids, 1000) as $chunk) {
	973	+ // Same moving window as the post walk above, for the same reason.
	974	+ $total = count($all_ids);
	975	+
	976	+ for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) {
	977	+ $chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK);
	978	+
624	979	$terms = get_terms($this->filter_term_query_args([
625	980	'taxonomy' => $taxonomy,
626	981	'include' => $chunk,
627	982	'orderby' => 'include', // preserve the resolved order
		@@ -652,8 +1007,43 @@
652	1007	}
653	1008	}
654	1009
655	1010	/**
	1011	+ * The XML declaration, ownership marker and optional stylesheet every
	1012	+ * sitemap document opens with.
	1013	+ *
	1014	+ * The marker is written unconditionally, and that is the point: removal on
	1015	+ * deactivate and uninstall deletes a web-root sitemap only when the file
	1016	+ * says it is ours, and our filenames are the canonical ones another SEO
	1017	+ * plugin writes too (#515). Tying the proof to `enable_styling` — the one
	1018	+ * marker older versions left — would mean a site with styling off either
	1019	+ * kept a shadowing file behind (#510) or had a competitor's deleted.
	1020	+ *
	1021	+ * @since 2.1.1
	1022	+ *
	1023	+ * The stylesheet URL is served by {@see Sitemap_Stylesheet}, not read off
	1024	+ * disk by the web server, because a static file cannot carry the site's own
	1025	+ * logo and colours (#639). It is a fixed URL: the palette is applied per
	1026	+ * request, so changing a brand colour needs no regeneration and shows up on
	1027	+ * sitemaps published long before.
	1028	+ *
	1029	+ * @param array $settings Sitemap settings (read for `enable_styling`).
	1030	+ * @param string $variant Stylesheet variant, `sitemap` or `index`.
	1031	+ * @return string Prolog lines, newline-terminated.
	1032	+ */
	1033	+ private function xml_prolog(array $settings, string $variant): string {
	1034	+ $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1035	+ $xml .= THINKRANK_SITEMAP_MARKER . "\n";
	1036	+
	1037	+ // The stylesheet is presentation only, so it stays opt-in.
	1038	+ if (!empty($settings['enable_styling'])) {
	1039	+ $xml .= '<?xml-stylesheet type="text/xsl" href="' . esc_url(Sitemap_Stylesheet::url($variant)) . '"?>' . "\n";
	1040	+ }
	1041	+
	1042	+ return $xml;
	1043	+ }
	1044	+
	1045	+ /**
656	1046	* Wrap a set of <url> entry strings in a complete <urlset> document.
657	1047	*
658	1048	* @since 1.14.0
659	1049	*
		@@ -662,14 +1052,10 @@
662	1052	* @param bool $with_image_ns Include the image sitemap namespace
663	1053	* @return string Full sitemap XML
664	1054	*/
665	1055	private function wrap_urlset(array $entries, array $settings, bool $with_image_ns): string {
666		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1056	+ $xml = $this->xml_prolog($settings, 'sitemap');
667	1057
668		- if (!empty($settings['enable_styling'])) {
669		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
670		- }
671		-
672	1058	if ($with_image_ns && !empty($settings['include_images'])) {
673	1059	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
674	1060	} else {
675	1061	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
		@@ -846,9 +1232,16 @@
846	1232	'_builtin' => false
847	1233	], 'names');
848	1234
849	1235	foreach ($custom_taxonomies as $taxonomy) {
850		- if ($this->should_include_taxonomy($taxonomy)) {
	1236	+ if (!$this->should_include_taxonomy($taxonomy)) {
	1237	+ continue;
	1238	+ }
	1239	+
	1240	+ // Same as the post-type walk above: an explicit per-taxonomy flag
	1241	+ // now decides, and an unset flag keeps the previous "included"
	1242	+ // behaviour (#660).
	1243	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $settings)) {
851	1244	$taxonomies[] = $taxonomy;
852	1245	}
853	1246	}
854	1247
		@@ -1061,8 +1454,17 @@
1061	1454	* @param array $settings Sitemap settings
1062	1455	* @return bool Whether to include in sitemap
1063	1456	*/
1064	1457	private function should_include_in_sitemap(\WP_Post $post, array $settings): bool {
	1458	+ // The WooCommerce cart, checkout and account pages are transactional,
	1459	+ // never indexable, and generate_default_robots_rules() already emits a
	1460	+ // Disallow for each of them. Listing them here submitted URLs our own
	1461	+ // robots.txt blocks, which Search Console reports as "Submitted URL
	1462	+ // blocked by robots.txt". Yoast and Rank Math exclude the same three.
	1463	+ if (in_array($post->ID, $this->woocommerce_excluded_page_ids(), true)) {
	1464	+ return false;
	1465	+ }
	1466	+
1065	1467	// Respect user setting for password protected content
1066	1468	if (!empty($post->post_password) && !empty($settings['exclude_password_protected'])) {
1067	1469	return false;
1068	1470	}
		@@ -1098,8 +1500,42 @@
1098	1500	return true;
1099	1501	}
1100	1502
1101	1503	/**
	1504	+ * WooCommerce pages that must never reach the sitemap.
	1505	+ *
	1506	+ * Resolved through wc_get_page_id() so a store that moved or renamed its
	1507	+ * cart/checkout/account pages is still matched. Returns an empty list when
	1508	+ * WooCommerce is not active. Memoised — should_include_in_sitemap() runs
	1509	+ * once per post.
	1510	+ *
	1511	+ * @since 2.0.1
	1512	+ *
	1513	+ * @return int[] Page IDs to exclude.
	1514	+ */
	1515	+ private function woocommerce_excluded_page_ids(): array {
	1516	+ if ($this->woocommerce_excluded_page_ids !== null) {
	1517	+ return $this->woocommerce_excluded_page_ids;
	1518	+ }
	1519	+
	1520	+ $ids = [];
	1521	+
	1522	+ if (function_exists('wc_get_page_id')) {
	1523	+ foreach (['cart', 'checkout', 'myaccount'] as $page) {
	1524	+ $id = (int) wc_get_page_id($page);
	1525	+ // wc_get_page_id() returns -1 when the page is not configured.
	1526	+ if ($id > 0) {
	1527	+ $ids[] = $id;
	1528	+ }
	1529	+ }
	1530	+ }
	1531	+
	1532	+ $this->woocommerce_excluded_page_ids = $ids;
	1533	+
	1534	+ return $ids;
	1535	+ }
	1536	+
	1537	+ /**
1102	1538	* Whether a term carries an explicit noindex override.
1103	1539	*
1104	1540	* Mirrors the post-side check in should_include_post(); terms store the same
1105	1541	* `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written
		@@ -1289,9 +1725,15 @@
1289	1725	if (!empty($settings['include_pages'])) {
1290	1726	$post_types[] = 'page';
1291	1727	}
1292	1728
1293		- // Auto-detect public custom post types that should be included
	1729	+ // Auto-detect public custom post types that should be included.
	1730	+ //
	1731	+ // A custom type's `include_<slug>` / `exclude_<slug>` flag is honoured
	1732	+ // here (#660). It was previously stored — additional_setting_keys()
	1733	+ // has always let those keys through — but never read, so a CPT was in
	1734	+ // the sitemap whatever the setting said. Unset still means included, so
	1735	+ // a site that never touched the flag is unaffected.
1294	1736	$custom_post_types = get_post_types([
1295	1737	'public' => true,
1296	1738	'_builtin' => false
1297	1739	], 'names');
		@@ -1296,9 +1738,13 @@
1296	1738	'_builtin' => false
1297	1739	], 'names');
1298	1740
1299	1741	foreach ($custom_post_types as $post_type) {
1300		- if ($this->should_include_post_type($post_type)) {
	1742	+ if (!$this->should_include_post_type($post_type)) {
	1743	+ continue;
	1744	+ }
	1745	+
	1746	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $post_type, $settings)) {
1301	1747	$post_types[] = $post_type;
1302	1748	}
1303	1749	}
1304	1750
		@@ -1319,18 +1765,11 @@
1319	1765	// Templately's internal `templately_library` store — which are template
1320	1766	// records, not standalone indexable URLs. BetterDocs `docs` and
1321	1767	// WooCommerce `product` register exclude_from_search => false, so they
1322	1768	// remain included.
1323		- if (!is_post_type_viewable($post_type)) {
1324		- return false;
1325		- }
1326		-
1327		- $post_type_obj = get_post_type_object($post_type);
1328		- if (!$post_type_obj || !empty($post_type_obj->exclude_from_search)) {
1329		- return false;
1330		- }
1331		-
1332		- return true;
	1769	+ // The predicate lives in Content_Type_Settings so the matrix can ask
	1770	+ // the same question before offering a switch for this post type.
	1771	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_post_type($post_type);
1333	1772	}
1334	1773
1335	1774	/**
1336	1775	* Check if taxonomy should trigger regeneration
		@@ -1339,11 +1778,13 @@
1339	1778	* @param string $taxonomy Taxonomy slug
1340	1779	* @return bool True if taxonomy should trigger regeneration
1341	1780	*/
1342	1781	private function should_include_taxonomy(string $taxonomy): bool {
1343		- // Include all public taxonomies (presets control which sitemaps are created)
1344		- $taxonomy_obj = get_taxonomy($taxonomy);
1345		- return $taxonomy_obj && $taxonomy_obj->public;
	1782	+ // Public taxonomies only, and only the ones this generator can actually
	1783	+ // emit — the same predicate the content-type matrix asks before it
	1784	+ // offers a sitemap switch for one (presets still control which
	1785	+ // sitemaps are created).
	1786	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_taxonomy($taxonomy);
1346	1787	}
1347	1788
1348	1789	/**
1349	1790	* Schedule debounced sitemap regeneration
		@@ -1351,16 +1792,356 @@
1351	1792	* @since 1.0.0
1352	1793	* @return void
1353	1794	*/
1354	1795	private function schedule_debounced_regeneration(): void {
1355		- // Clear any existing scheduled regeneration
1356		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap');
	1796	+ $this->mark_regeneration_pending('content');
	1797	+ $this->debounce_event('thinkrank_regenerate_sitemap');
	1798	+ }
1357	1799
1358		- // Schedule regeneration in 30 seconds to debounce rapid changes
1359		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap');
	1800	+ /**
	1801	+ * Schedule (or keep) the debounced single event behind a regeneration hook.
	1802	+ *
	1803	+ * An event that is already due is left alone. WP-Cron only runs when a
	1804	+ * request arrives, so on a site with DISABLE_WP_CRON, a blocked loopback or
	1805	+ * little traffic an overdue event can sit in the queue for a long time —
	1806	+ * clearing and re-scheduling it on every save pushed the rebuild
	1807	+ * permanently 30 seconds into the future and the sitemap never updated
	1808	+ * (#629). Debouncing only against an event that has not come due yet keeps
	1809	+ * the bulk-edit coalescing without starving the rebuild.
	1810	+ *
	1811	+ * @since 2.2.1
	1812	+ * @param string $hook Regeneration hook to debounce.
	1813	+ * @return void
	1814	+ */
	1815	+ private function debounce_event(string $hook): void {
	1816	+ $next = wp_next_scheduled($hook);
	1817	+
	1818	+ if ($next !== false) {
	1819	+ if ($next <= time()) {
	1820	+ return;
	1821	+ }
	1822	+
	1823	+ wp_clear_scheduled_hook($hook);
	1824	+ }
	1825	+
	1826	+ wp_schedule_single_event(time() + self::REGENERATION_DEBOUNCE, $hook);
1360	1827	}
1361	1828
1362	1829	/**
	1830	+ * Record that a rebuild is outstanding, so an overdue one can be taken over
	1831	+ * by a later request and its staleness surfaced in the UI.
	1832	+ *
	1833	+ * `since` is the *oldest* outstanding change: it is what the takeover grace
	1834	+ * and the admin staleness warning are measured from, so successive edits
	1835	+ * must not push it forward. A settings change outranks a content change —
	1836	+ * it rebuilds regardless of the auto_generate toggle and handles a sitemap
	1837	+ * that has just been disabled — so once one is outstanding it stays the
	1838	+ * recorded source until the rebuild lands.
	1839	+ *
	1840	+ * @since 2.2.1
	1841	+ * @param string $source Either 'content' or 'settings'.
	1842	+ * @return void
	1843	+ */
	1844	+ private function mark_regeneration_pending(string $source): void {
	1845	+ // Whatever made the static files stale made the rendered ones stale
	1846	+ // too. Invalidating here rather than only on the rebuild keeps the two
	1847	+ // delivery modes reacting to exactly the same triggers, which is the
	1848	+ // only way a dynamic site stays as fresh as a static one (#752).
	1849	+ $this->flush_dynamic_cache();
	1850	+
	1851	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1852	+ $pending = is_array($pending) ? $pending : [];
	1853	+
	1854	+ $since = !empty($pending['since']) ? (int) $pending['since'] : time();
	1855	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	1856	+ $source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	1857	+
	1858	+ update_option(
	1859	+ self::REGENERATION_PENDING_OPTION,
	1860	+ [
	1861	+ 'since' => $since,
	1862	+ 'source' => $source,
	1863	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 0,
	1864	+ 'next_attempt' => !empty($pending['next_attempt'])
	1865	+ ? (int) $pending['next_attempt']
	1866	+ : time() + self::REGENERATION_TAKEOVER_GRACE,
	1867	+ // Bumped on every change so a rebuild can tell whether the edit
	1868	+ // it started for is still the newest one outstanding.
	1869	+ 'revision' => (!empty($pending['revision']) ? (int) $pending['revision'] : 0) + 1,
	1870	+ ],
	1871	+ true
	1872	+ );
	1873	+ }
	1874	+
	1875	+ /**
	1876	+ * The revision of the outstanding rebuild, for
	1877	+ * {@see mark_regeneration_complete()} to compare against once it is done.
	1878	+ *
	1879	+ * @since 2.2.1
	1880	+ * @return int Current revision, 0 when nothing is outstanding.
	1881	+ */
	1882	+ private function current_regeneration_revision(): int {
	1883	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1884	+
	1885	+ return (is_array($pending) && !empty($pending['revision'])) ? (int) $pending['revision'] : 0;
	1886	+ }
	1887	+
	1888	+ /**
	1889	+ * Clear the outstanding-rebuild marker and any recorded failure.
	1890	+ *
	1891	+ * Public because a manual generation satisfies whatever the automatic path
	1892	+ * was still waiting to write.
	1893	+ *
	1894	+ * @since 2.2.1
	1895	+ * @return void
	1896	+ */
	1897	+ public function mark_regeneration_complete(?int $revision = null): void {
	1898	+ // The write succeeded, so whatever failure was on record is history.
	1899	+ if (get_option(self::REGENERATION_ERROR_OPTION, null) !== null) {
	1900	+ delete_option(self::REGENERATION_ERROR_OPTION);
	1901	+ }
	1902	+
	1903	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	1904	+
	1905	+ if ($pending === null) {
	1906	+ return;
	1907	+ }
	1908	+
	1909	+ // A change that landed while this rebuild was running is not covered by
	1910	+ // the files it just wrote, so it has to stay outstanding — otherwise, on
	1911	+ // a site where WP-Cron never fires, clearing the marker would strand it
	1912	+ // exactly the way #629 stranded everything.
	1913	+ if (
	1914	+ $revision !== null
	1915	+ && is_array($pending)
	1916	+ && (int) ($pending['revision'] ?? 0) !== $revision
	1917	+ ) {
	1918	+ return;
	1919	+ }
	1920	+
	1921	+ delete_option(self::REGENERATION_PENDING_OPTION);
	1922	+ }
	1923	+
	1924	+ /**
	1925	+ * Record a failed regeneration instead of discarding it.
	1926	+ *
	1927	+ * Keeps the pending marker in place so the rebuild is retried, but backs the
	1928	+ * next attempt off exponentially (capped) so a persistently failing
	1929	+ * generation cannot run on every admin request.
	1930	+ *
	1931	+ * @since 2.2.1
	1932	+ * @param string $message Failure detail.
	1933	+ * @param string $source Either 'content' or 'settings'.
	1934	+ * @return void
	1935	+ */
	1936	+ private function record_regeneration_failure(string $message, string $source): void {
	1937	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1938	+ $pending = is_array($pending) ? $pending : [];
	1939	+ $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1;
	1940	+
	1941	+ $backoff = min(
	1942	+ self::REGENERATION_TAKEOVER_GRACE * (2 ** min($attempts, 10)),
	1943	+ self::REGENERATION_MAX_BACKOFF
	1944	+ );
	1945	+
	1946	+ // Same precedence mark_regeneration_pending() enforces: a settings
	1947	+ // rebuild outranks a content one and must not be downgraded by a failed
	1948	+ // attempt. Overwriting it routed the retry back through the content
	1949	+ // path, where should_auto_generate() can be false and the completion
	1950	+ // marker then discards the settings rebuild entirely. Only the
	1951	+ // outstanding rebuild is upgraded — the recorded error keeps reporting
	1952	+ // whichever attempt actually failed.
	1953	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	1954	+ $pending_source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	1955	+
	1956	+ update_option(
	1957	+ self::REGENERATION_PENDING_OPTION,
	1958	+ [
	1959	+ 'since' => !empty($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	+ );
	1967	+
	1968	+ update_option(
	1969	+ self::REGENERATION_ERROR_OPTION,
	1970	+ [
	1971	+ 'message' => $message,
	1972	+ 'source' => $source,
	1973	+ 'attempts' => $attempts,
	1974	+ 'time' => time(),
	1975	+ ],
	1976	+ false
	1977	+ );
	1978	+
	1979	+ if (defined('WP_DEBUG') && WP_DEBUG) {
	1980	+ // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
	1981	+ error_log(sprintf('ThinkRank: sitemap %s regeneration failed — %s', $source, $message));
	1982	+ }
	1983	+ }
	1984	+
	1985	+ /**
	1986	+ * Is a rebuild outstanding and past the point where WP-Cron should have run
	1987	+ * it?
	1988	+ *
	1989	+ * Deliberately cheap — one autoloaded option read — because it is consulted
	1990	+ * on every admin request to decide whether the takeover is needed.
	1991	+ *
	1992	+ * @since 2.2.1
	1993	+ * @return bool True when a request should rebuild the sitemap itself.
	1994	+ */
	1995	+ public static function has_overdue_regeneration(): bool {
	1996	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1997	+
	1998	+ if (!is_array($pending) || empty($pending['since'])) {
	1999	+ return false;
	2000	+ }
	2001	+
	2002	+ $due = !empty($pending['next_attempt'])
	2003	+ ? (int) $pending['next_attempt']
	2004	+ : (int) $pending['since'] + self::REGENERATION_TAKEOVER_GRACE;
	2005	+
	2006	+ return time() >= $due;
	2007	+ }
	2008	+
	2009	+ /**
	2010	+ * Rebuild the sitemap in-request when WP-Cron has not delivered.
	2011	+ *
	2012	+ * Hooked on `shutdown` for admin, REST and CLI requests only (see
	2013	+ * Plugin::register_sitemap_cron_listeners()), so the work happens after the
	2014	+ * response has been sent and never adds latency to a visitor page view.
	2015	+ *
	2016	+ * @since 2.2.1
	2017	+ * @return void
	2018	+ */
	2019	+ public function run_overdue_regeneration(): void {
	2020	+ if (!self::has_overdue_regeneration()) {
	2021	+ return;
	2022	+ }
	2023	+
	2024	+ // Cron is running: it is about to do exactly this work.
	2025	+ if (wp_doing_cron()) {
	2026	+ return;
	2027	+ }
	2028	+
	2029	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2030	+ $source = (is_array($pending) && isset($pending['source'])) ? (string) $pending['source'] : 'content';
	2031	+
	2032	+ if ($source === 'settings') {
	2033	+ $this->regenerate_sitemap_from_settings();
	2034	+ return;
	2035	+ }
	2036	+
	2037	+ $this->auto_regenerate_sitemap();
	2038	+ }
	2039	+
	2040	+ /**
	2041	+ * Acquire the shared generation lock.
	2042	+ *
	2043	+ * @since 2.2.1
	2044	+ * @return bool True when this process may generate.
	2045	+ */
	2046	+ private function acquire_generation_lock(): bool {
	2047	+ if (get_transient(self::GENERATION_LOCK_TRANSIENT)) {
	2048	+ return false;
	2049	+ }
	2050	+
	2051	+ set_transient(self::GENERATION_LOCK_TRANSIENT, time(), 5 * MINUTE_IN_SECONDS);
	2052	+
	2053	+ return true;
	2054	+ }
	2055	+
	2056	+ /**
	2057	+ * Release the shared generation lock.
	2058	+ *
	2059	+ * @since 2.2.1
	2060	+ * @return void
	2061	+ */
	2062	+ private function release_generation_lock(): void {
	2063	+ delete_transient(self::GENERATION_LOCK_TRANSIENT);
	2064	+ }
	2065	+
	2066	+ /**
	2067	+ * Report how automatic regeneration is faring, for the admin UI.
	2068	+ *
	2069	+ * The feature used to fail invisibly: `last_generated` simply stopped
	2070	+ * advancing and nothing drew attention to it (#629).
	2071	+ *
	2072	+ * @since 2.2.1
	2073	+ * @return array Health payload.
	2074	+ */
	2075	+ public function get_regeneration_health(): array {
	2076	+ $settings = $this->get_settings('site');
	2077	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2078	+ $pending = is_array($pending) ? $pending : [];
	2079	+ $error = get_option(self::REGENERATION_ERROR_OPTION, []);
	2080	+ $error = is_array($error) ? $error : [];
	2081	+
	2082	+ $since = !empty($pending['since']) ? (int) $pending['since'] : 0;
	2083	+
	2084	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap');
	2085	+ if ($next_scheduled === false) {
	2086	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap_settings');
	2087	+ }
	2088	+
	2089	+ return [
	2090	+ 'auto_generate' => !empty($settings['auto_generate']),
	2091	+ 'last_generated' => $settings['last_generated'] ?? '',
	2092	+ 'pending_since' => $since ? gmdate('c', $since) : null,
	2093	+ 'pending_seconds' => $since ? max(0, time() - $since) : 0,
	2094	+ 'next_scheduled' => $next_scheduled ? gmdate('c', (int) $next_scheduled) : null,
	2095	+ 'cron_disabled' => defined('DISABLE_WP_CRON') && DISABLE_WP_CRON,
	2096	+ 'stale' => $this->is_sitemap_stale($settings),
	2097	+ 'last_error' => !empty($error['message'])
	2098	+ ? [
	2099	+ 'message' => (string) $error['message'],
	2100	+ 'source' => isset($error['source']) ? (string) $error['source'] : 'content',
	2101	+ 'time' => !empty($error['time']) ? gmdate('c', (int) $error['time']) : null,
	2102	+ ]
	2103	+ : null,
	2104	+ ];
	2105	+ }
	2106	+
	2107	+ /**
	2108	+ * Has published content changed since the served sitemap was last written?
	2109	+ *
	2110	+ * Uses core's cached last-modified lookup, which only considers published
	2111	+ * posts — the same content the sitemap covers.
	2112	+ *
	2113	+ * @since 2.2.1
	2114	+ * @param array $settings Sitemap settings.
	2115	+ * @return bool True when the sitemap is behind the content.
	2116	+ */
	2117	+ private function is_sitemap_stale(array $settings): bool {
	2118	+ if (empty($settings['enabled']) || empty($settings['last_generated'])) {
	2119	+ // Never generated is already reported separately by the UI.
	2120	+ return false;
	2121	+ }
	2122	+
	2123	+ $generated = strtotime((string) $settings['last_generated']);
	2124	+ if (!$generated) {
	2125	+ return false;
	2126	+ }
	2127	+
	2128	+ $modified = get_lastpostmodified('gmt');
	2129	+ if (!$modified) {
	2130	+ return false;
	2131	+ }
	2132	+
	2133	+ $modified = strtotime($modified . ' UTC');
	2134	+ if (!$modified) {
	2135	+ return false;
	2136	+ }
	2137	+
	2138	+ // A minute of slack keeps a rebuild that ran alongside the edit from
	2139	+ // reporting itself as stale.
	2140	+ return $modified > ($generated + MINUTE_IN_SECONDS);
	2141	+ }
	2142	+
	2143	+ /**
1363	2144	* Public entry point to debounce-rebuild the sitemap after a settings change
1364	2145	* (e.g. toggling inclusion rules via REST or the MCP ability), so the served
1365	2146	* file reflects the new settings instead of going stale until a content edit.
1366	2147	*
		@@ -1369,10 +2150,10 @@
1369	2150	public function schedule_regeneration(): void {
1370	2151	// Debounce against rapid successive saves, but use the settings-specific
1371	2152	// hook so the rebuild runs regardless of the auto_generate toggle (which
1372	2153	// only governs content-change-triggered regeneration).
1373		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap_settings');
1374		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap_settings');
	2154	+ $this->mark_regeneration_pending('settings');
	2155	+ $this->debounce_event('thinkrank_regenerate_sitemap_settings');
1375	2156	}
1376	2157
1377	2158	/**
1378	2159	* Rebuild the served sitemap after an explicit settings change.
		@@ -1384,8 +2165,15 @@
1384	2165	*
1385	2166	* @return void
1386	2167	*/
1387	2168	public function regenerate_sitemap_from_settings(): void {
	2169	+ if (!$this->acquire_generation_lock()) {
	2170	+ // A manual generation (or another request's takeover) is already
	2171	+ // writing the files; the pending marker survives so this rebuild is
	2172	+ // retried rather than lost.
	2173	+ return;
	2174	+ }
	2175	+
1388	2176	try {
1389	2177	$settings = $this->get_settings('site');
1390	2178	if (empty($settings['enabled'])) {
1391	2179	// The sitemap was disabled: remove the previously generated static
		@@ -1391,13 +2179,31 @@
1391	2179	// The sitemap was disabled: remove the previously generated static
1392	2180	// files so the web server stops serving a stale sitemap that
1393	2181	// crawlers would otherwise keep fetching.
1394	2182	$this->delete_published_sitemaps();
	2183	+ $this->mark_regeneration_complete();
1395	2184	return;
1396	2185	}
1397		- $this->generate_and_save($settings);
	2186	+
	2187	+ $revision = $this->current_regeneration_revision();
	2188	+
	2189	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2190	+ $this->switch_to_dynamic_delivery($settings, $revision, 'settings');
	2191	+ return;
	2192	+ }
	2193	+
	2194	+ if ($this->generate_and_save($settings)) {
	2195	+ $this->mark_regeneration_complete($revision);
	2196	+ } else {
	2197	+ $this->record_regeneration_failure(
	2198	+ $this->write_failure_message(),
	2199	+ 'settings'
	2200	+ );
	2201	+ }
1398	2202	} catch (\Throwable $e) {
1399		- // Settings-triggered regeneration failed - details in exception.
	2203	+ $this->record_regeneration_failure($e->getMessage(), 'settings');
	2204	+ } finally {
	2205	+ $this->release_generation_lock();
1400	2206	}
1401	2207	}
1402	2208
1403	2209	/**
		@@ -1402,18 +2208,23 @@
1402	2208
1403	2209	/**
1404	2210	* Remove every static sitemap file ThinkRank publishes to the web root.
1405	2211	*
1406		- * Called when the sitemap feature is disabled, and by the cleanup route, so
1407		- * /sitemap.xml, /sitemap_index.xml, the segmented children (incl. paginated
1408		- * -N pages), and /local-sitemap.xml stop being served. Only ThinkRank's own
1409		- * filenames are targeted; WordPress core's wp-sitemap.xml and any other
1410		- * plugin's sitemap in the web root are left untouched.
	2212	+ * Called when the sitemap feature is disabled, by the cleanup route, and by
	2213	+ * both removal paths, so /sitemap.xml, /sitemap_index.xml, the segmented
	2214	+ * children (incl. paginated -N pages), and /local-sitemap.xml stop being
	2215	+ * served. Only ThinkRank's own filenames are targeted; WordPress core's
	2216	+ * wp-sitemap.xml and any other plugin's sitemap in the web root are left
	2217	+ * untouched.
1411	2218	*
1412	2219	* @since 1.31.0 Returns the filenames removed, and accepts the settings to
1413	2220	* derive them from, so a caller that already read them (and
1414	2221	* needs to report what went) does not have to re-read or
1415	2222	* re-derive the name list.
	2223	+ * @since 2.1.0 Delegates to thinkrank_webroot_delete_sitemaps(). Uninstall
	2224	+ * needs the same removal but has no autoloader to reach this
	2225	+ * class, so the logic moved to includes/cleanup-webroot.php
	2226	+ * and this stays as the in-plugin entry point.
1416	2227	*
1417	2228	* @param array|null $settings Optional. Sitemap settings; defaults to the
1418	2229	* saved site settings.
1419	2230	* @return array{deleted: string[], failed: string[]} Basenames removed, and
		@@ -1419,123 +2230,58 @@
1419	2230	* @return array{deleted: string[], failed: string[]} Basenames removed, and
1420	2231	* those that existed but could not be removed.
1421	2232	*/
1422	2233	public function delete_published_sitemaps(?array $settings = null): array {
1423		- $settings = $settings ?? $this->get_settings('site');
1424		- $deleted = [];
1425		- $failed = [];
1426		-
1427		- // Base filenames to remove. Derive the child/index names from the stored
1428		- // sitemap_urls so a custom custom_url_pattern (e.g. seo-{type}.xml) is
1429		- // honored, not just the default sitemap-*.xml naming. Include the current
1430		- // primary + the default names as a safety net.
1431		- $names = [
1432		- 'sitemap.xml',
1433		- 'sitemap_index.xml',
1434		- 'local-sitemap.xml',
1435		- $this->get_primary_sitemap_filename($settings),
1436		- ];
1437		-
1438		- foreach ((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : []) as $config) {
1439		- if (empty($config['url'])) {
1440		- continue;
1441		- }
1442		- $name = basename((string) wp_parse_url($config['url'], PHP_URL_PATH));
1443		- if ($name !== '') {
1444		- $names[] = $name;
1445		- }
1446		- }
1447		-
1448		- // Segment names for every type we could have published under the current
1449		- // url pattern, not just the ones stored in sitemap_urls. Two states leave
1450		- // a published child unlisted there: settings that drifted from what is on
1451		- // disk (a single-file entry while an index and its children are live), and
1452		- // a content type that has since been switched off. Deriving the names
1453		- // from the pattern reaches those without falling back to a 'sitemap-*.xml'
1454		- // glob, which would also match another plugin's segments.
1455		- $names = array_merge($names, $this->publishable_segment_filenames($settings));
1456		-
1457		- foreach (array_unique(array_filter($names)) as $name) {
1458		- // Remove the file itself and any paginated -N variants of its stem
1459		- // (e.g. seo-posts.xml plus seo-posts-2.xml, seo-posts-3.xml…).
1460		- $path = ABSPATH . $name;
1461		- if (file_exists($path)) {
1462		- wp_delete_file($path);
1463		- // wp_delete_file() returns nothing, so confirm by re-checking.
1464		- if (file_exists($path)) {
1465		- $failed[] = $name;
1466		- } else {
1467		- $deleted[] = $name;
1468		- }
1469		- }
1470		- if (preg_match('/^(.*)\.xml$/i', $name, $m)) {
1471		- // Pagination pages only — a numeric suffix on this exact stem.
1472		- // Globbing '<stem>-*.xml' matched any name that merely started
1473		- // with the stem, so the default 'sitemap.xml' entry pulled in
1474		- // every sitemap-*.xml in the root, including another plugin's.
1475		- $paged_pattern = '/^' . preg_quote($m[1], '/') . '-\d+\.xml$/i';
1476		-
1477		- foreach (glob(ABSPATH . $m[1] . '-*.xml') ?: [] as $paged) {
1478		- $paged_name = basename($paged);
1479		- if (!preg_match($paged_pattern, $paged_name)) {
1480		- continue;
1481		- }
1482		-
1483		- wp_delete_file($paged);
1484		- if (file_exists($paged)) {
1485		- $failed[] = $paged_name;
1486		- } else {
1487		- $deleted[] = $paged_name;
1488		- }
1489		- }
1490		- }
1491		- }
1492		-
1493		- return [
1494		- 'deleted' => array_values(array_unique($deleted)),
1495		- 'failed' => array_values(array_unique($failed)),
1496		- ];
	2234	+ return thinkrank_webroot_delete_sitemaps($settings ?? $this->get_settings('site'));
1497	2235	}
1498	2236
1499	2237	/**
1500	2238	* Every child-sitemap filename this site could have published.
1501	2239	*
1502		- * Formats the configured url pattern against each type ThinkRank segments by
1503		- * — the four built-ins plus every public custom post type and public custom
1504		- * taxonomy — ignoring whether that type is currently included. The point is
1505		- * to recognise our own filenames, and a type that was published and later
1506		- * disabled still left a file behind.
1507		- *
1508	2240	* @since 1.31.0
	2241	+ * @since 2.1.0 Delegates to thinkrank_webroot_segment_filenames().
1509	2242	*
1510	2243	* @param array $settings Sitemap settings (read for `custom_url_pattern`).
1511	2244	* @return string[] Basenames, e.g. ['sitemap-posts.xml', 'sitemap-pages.xml'].
1512	2245	*/
1513	2246	private function publishable_segment_filenames(array $settings): array {
1514		- $pattern = (string) ($settings['custom_url_pattern'] ?? 'sitemap-{type}.xml');
1515		- if (strpos($pattern, '{type}') === false) {
1516		- return [];
1517		- }
	2247	+ return thinkrank_webroot_segment_filenames($settings);
	2248	+ }
1518	2249
1519		- $types = ['posts', 'pages', 'categories', 'tags'];
1520		-
1521		- foreach (get_post_types(['public' => true, '_builtin' => false], 'names') as $cpt) {
1522		- $types[] = (string) $cpt;
1523		- }
1524		-
1525		- foreach (get_taxonomies(['public' => true, '_builtin' => false], 'names') as $taxonomy) {
1526		- $types[] = (string) $taxonomy;
1527		- }
1528		-
1529		- $names = [];
1530		- foreach (array_unique($types) as $type) {
1531		- $name = basename(str_replace('{type}', $type, $pattern));
1532		- if ($name !== '') {
1533		- $names[] = $name;
1534		- }
1535		- }
1536		-
1537		- return $names;
	2250	+ /**
	2251	+ * Can this web-root file be shown to be a sitemap ThinkRank wrote?
	2252	+ *
	2253	+ * The generator deletes as often as cleanup does — a segment that dropped
	2254	+ * out of the set, a pagination page beyond the new count, the local sitemap
	2255	+ * after the business identity was cleared — and until 2.1.1 it did all
	2256	+ * three by filename alone. That is the #515 bug on a far more frequent
	2257	+ * trigger: our names are the canonical ones, so an ordinary regeneration
	2258	+ * (post save, term change, settings save) destroyed RankMath's
	2259	+ * `sitemap-tags.xml` and `local-sitemap.xml` with no deactivation involved.
	2260	+ *
	2261	+ * Every name the generator derives comes from the current settings — the
	2262	+ * url pattern, the configured `sitemap_urls`, `local-sitemap.xml` — but the
	2263	+ * ownership test is still asked with `$name_derived = false`, which switches
	2264	+ * off the legacy fallback for the whole generator side.
	2265	+ *
	2266	+ * The fallback exists to recover a pre-2.1.1 file written with
	2267	+ * `enable_styling` off, which carries neither marker. That recovery belongs
	2268	+ * to the once-off cleanup paths. Here it can only do harm: this method runs
	2269	+ * on every post save, and everything this version writes carries
	2270	+ * THINKRANK_SITEMAP_MARKER, so after the site's first regeneration an
	2271	+ * unmarked file at one of our names is by definition somebody else's — and
	2272	+ * deleting it on an ordinary regeneration is #515 through the more common
	2273	+ * door. The cost is a stale unmarked segment left on disk until deactivation
	2274	+ * picks it up, which is the safe direction to fail in.
	2275	+ *
	2276	+ * @since 2.1.1
	2277	+ *
	2278	+ * @param string $path Absolute path to a file in the web root.
	2279	+ * @param array $settings Sitemap settings.
	2280	+ * @return bool True when the file may be deleted.
	2281	+ */
	2282	+ private function webroot_sitemap_is_ours(string $path, array $settings): bool {
	2283	+ return thinkrank_webroot_sitemap_is_ours($path, $settings, false);
1538	2284	}
1539	2285
1540	2286	/**
1541	2287	* Auto-regenerate sitemap (called by scheduled action)
		@@ -1543,20 +2289,48 @@
1543	2289	* @since 1.0.0
1544	2290	* @return void
1545	2291	*/
1546	2292	public function auto_regenerate_sitemap(): void {
	2293	+ if (!$this->acquire_generation_lock()) {
	2294	+ // A manual generation (or another request's takeover) is already
	2295	+ // writing the files; the pending marker survives so this rebuild is
	2296	+ // retried rather than lost.
	2297	+ return;
	2298	+ }
	2299	+
1547	2300	try {
1548	2301	// Double-check that auto-generation is still enabled
1549	2302	if (!$this->should_auto_generate()) {
	2303	+ // Nothing outstanding can be delivered while the feature is off,
	2304	+ // so drop the marker rather than let the takeover retry forever.
	2305	+ $this->mark_regeneration_complete();
1550	2306	return;
1551	2307	}
1552	2308
1553		- $this->generate_and_save($this->get_settings('site'));
	2309	+ $revision = $this->current_regeneration_revision();
	2310	+ $settings = $this->get_settings('site');
1554	2311
1555		- // Sitemap auto-regenerated successfully
	2312	+ // See regenerate_sitemap_from_settings(): nothing to write.
	2313	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2314	+ $this->switch_to_dynamic_delivery($settings, $revision, 'content');
	2315	+ return;
	2316	+ }
1556	2317
	2318	+ if ($this->generate_and_save($settings)) {
	2319	+ $this->mark_regeneration_complete($revision);
	2320	+ } else {
	2321	+ // Previously this returned quietly and last_generated simply
	2322	+ // stopped advancing, leaving the site owner with no way to learn
	2323	+ // the sitemap had stopped updating (#629).
	2324	+ $this->record_regeneration_failure(
	2325	+ $this->write_failure_message(),
	2326	+ 'content'
	2327	+ );
	2328	+ }
1557	2329	} catch (\Throwable $e) {
1558		- // Sitemap auto-regeneration failed - error details available in exception
	2330	+ $this->record_regeneration_failure($e->getMessage(), 'content');
	2331	+ } finally {
	2332	+ $this->release_generation_lock();
1559	2333	}
1560	2334	}
1561	2335
1562	2336	/**
		@@ -1571,8 +2345,26 @@
1571	2345	* @param array $settings Sitemap settings.
1572	2346	* @return bool True when the sitemap files were written.
1573	2347	*/
1574	2348	public function generate_and_save(array $settings): bool {
	2349	+ // Dynamic delivery publishes no files, so writing them here would put a
	2350	+ // static copy back in the web root for the server to serve in place of
	2351	+ // the dynamic route. Guarding at each call site left gaps — the
	2352	+ // snapshot migrator's post-import regeneration had none — so the rule
	2353	+ // lives with the writing instead.
	2354	+ //
	2355	+ // `is_collecting()` is the exception that makes dynamic delivery work
	2356	+ // at all: render_document() and collect_documents() reach this same
	2357	+ // method with the writer swapped for a collector, and that is precisely
	2358	+ // the dynamic build. Only a real write is skipped.
	2359	+ if (!$this->is_collecting() && 'dynamic' === $this->resolve_delivery_mode($settings)) {
	2360	+ // Whatever prompted this call changed the sitemap's content, so the
	2361	+ // rendered copies must not outlive it.
	2362	+ $this->flush_dynamic_cache();
	2363	+
	2364	+ return true;
	2365	+ }
	2366	+
1575	2367	// Index mode is driven by the use_sitemap_index toggle (not merely by how
1576	2368	// many sitemap_urls happen to be configured). When the toggle is on but
1577	2369	// no child sitemaps are set up yet, synthesize the per-type segmented set
1578	2370	// so we emit a real <sitemapindex> with paginated children instead of a
		@@ -1583,16 +2375,29 @@
1583	2375	$results = $this->generate_multiple_sitemaps($settings);
1584	2376	$written = !empty($results['success']);
1585	2377	} else {
1586	2378	$xml = $this->generate_sitemap($settings);
1587		- $written = $this->save_sitemap_to_file($xml, $this->get_primary_sitemap_filename($settings));
	2379	+ $primary = $this->get_primary_sitemap_filename($settings);
	2380	+ $written = $this->save_sitemap_to_file($xml, $primary);
1588	2381
1589	2382	// Local business sitemap is a standalone file, regenerated on the
1590	2383	// single-sitemap path too (this is the default mode).
1591	2384	$this->regenerate_local_sitemap($settings);
	2385	+
	2386	+ // Switching out of index mode leaves sitemap_index.xml and every
	2387	+ // child on disk, still served and never refreshed again. The index
	2388	+ // path already prunes what it no longer owns; this path never did,
	2389	+ // so the site kept serving two sitemap trees (#563). Ownership is
	2390	+ // still tested per file, so another plugin's sitemap at one of our
	2391	+ // names is never touched (#515).
	2392	+ $this->prune_orphaned_segments($settings, [['filename' => basename($primary)]]);
1592	2393	}
1593	2394
1594		- if ($written) {
	2395	+ // last_generated describes what is on disk. A dynamic render publishes
	2396	+ // nothing, so advancing it would report a static publication that never
	2397	+ // happened and would let primary_sitemap_file_exists() callers believe
	2398	+ // there is a file to serve.
	2399	+ if ($written && !$this->is_collecting()) {
1595	2400	$settings['last_generated'] = gmdate('c');
1596	2401	$this->save_settings('site', null, $settings);
1597	2402	}
1598	2403
		@@ -1599,8 +2404,398 @@
1599	2404	return $written;
1600	2405	}
1601	2406
1602	2407	/**
	2408	+ * Whether this instance is rendering documents rather than publishing them.
	2409	+ *
	2410	+ * @since 2.9.0
	2411	+ *
	2412	+ * @return bool
	2413	+ */
	2414	+ private function is_collecting(): bool {
	2415	+ return $this->document_sink !== null;
	2416	+ }
	2417	+
	2418	+ /**
	2419	+ * Complete a regeneration that delivers dynamically, retiring stale files.
	2420	+ *
	2421	+ * Dynamic delivery renders nothing to disk, but that is only half the job.
	2422	+ * A web server hands back an existing `/sitemap.xml` without ever loading
	2423	+ * WordPress, so any file left over from a previous static generation goes on
	2424	+ * being served forever and {@see \ThinkRank\Frontend\SEO_Manager
	2425	+ * ::maybe_serve_sitemap()} is never reached. Switching to dynamic while
	2426	+ * leaving those files in place would therefore appear to do nothing at all.
	2427	+ *
	2428	+ * Both transitions matter and they differ:
	2429	+ *
	2430	+ * - An explicit switch to `dynamic` happens on a site whose root is usually
	2431	+ * still writable, so the files can simply be removed.
	2432	+ * - An `auto` site that becomes read-only cannot remove them, because
	2433	+ * deleting an entry needs write permission on the directory that holds
	2434	+ * it. There the stale sitemap really is stuck in front of us, and the
	2435	+ * honest outcome is a recorded failure naming it rather than a rebuild
	2436	+ * reported as complete (#754 review).
	2437	+ *
	2438	+ * Ownership is tested per file by the shared helper, so another plugin's
	2439	+ * sitemap at one of our names is never deleted (#515).
	2440	+ *
	2441	+ * @since 2.9.0
	2442	+ *
	2443	+ * @param array $settings Sitemap settings.
	2444	+ * @param int $revision Revision this rebuild is completing.
	2445	+ * @param string $source 'settings' or 'content', for the failure record.
	2446	+ * @return void
	2447	+ */
	2448	+ private function switch_to_dynamic_delivery(array $settings, int $revision, string $source): void {
	2449	+ $this->flush_dynamic_cache();
	2450	+
	2451	+ $removal = $this->delete_published_sitemaps($settings);
	2452	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2453	+
	2454	+ if (!empty($stuck)) {
	2455	+ $this->record_regeneration_failure(
	2456	+ sprintf(
	2457	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	2458	+ __('The sitemap is being served from WordPress, but these files are still in the site root and your web server will keep serving them instead: %1$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'),
	2459	+ implode(', ', $stuck),
	2460	+ untrailingslashit(ABSPATH)
	2461	+ ),
	2462	+ $source
	2463	+ );
	2464	+
	2465	+ return;
	2466	+ }
	2467	+
	2468	+ $this->mark_regeneration_complete($revision);
	2469	+ }
	2470	+
	2471	+ /**
	2472	+ * What to tell the site owner when publishing the files failed.
	2473	+ *
	2474	+ * The old wording stated the symptom and stopped there, so the reported
	2475	+ * cause was a guess and this reached support as a plugin fault rather than
	2476	+ * a folder permission (#752, #753). When the root is demonstrably
	2477	+ * unwritable, say that, and say what to do about it.
	2478	+ *
	2479	+ * @since 2.9.0
	2480	+ *
	2481	+ * @return string
	2482	+ */
	2483	+ private function write_failure_message(): string {
	2484	+ if (!wp_is_writable(ABSPATH)) {
	2485	+ return sprintf(
	2486	+ /* translators: %s: absolute path to the WordPress root. */
	2487	+ __('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'),
	2488	+ untrailingslashit(ABSPATH)
	2489	+ );
	2490	+ }
	2491	+
	2492	+ return __('The sitemap files could not be written to the site root.', 'thinkrank');
	2493	+ }
	2494	+
	2495	+ /**
	2496	+ * How this site delivers its sitemap.
	2497	+ *
	2498	+ * `auto` is resolved on whether the web root can be written. That is the
	2499	+ * right signal here (unlike llms.txt, where the question is whether the
	2500	+ * server applies the .htaccess charset block): a site whose root is
	2501	+ * read-only cannot publish a sitemap file at all, and before this existed
	2502	+ * the feature simply failed with "The sitemap files could not be written to
	2503	+ * the site root." and served nothing (#752).
	2504	+ *
	2505	+ * @since 2.9.0
	2506	+ *
	2507	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	2508	+ * @return string One of 'static' or 'dynamic'. Never 'auto'.
	2509	+ */
	2510	+ public function resolve_delivery_mode(?array $settings = null): string {
	2511	+ $settings = $settings ?? $this->get_settings('site');
	2512	+ $mode = (string) ($settings['delivery_mode'] ?? 'auto');
	2513	+
	2514	+ if ('static' === $mode || 'dynamic' === $mode) {
	2515	+ return $mode;
	2516	+ }
	2517	+
	2518	+ return wp_is_writable(ABSPATH) ? 'static' : 'dynamic';
	2519	+ }
	2520	+
	2521	+ /**
	2522	+ * Render one published sitemap document without touching the filesystem.
	2523	+ *
	2524	+ * Runs the ordinary build pipeline with the writer swapped for a collector,
	2525	+ * so the bytes returned here are the bytes the static path would have
	2526	+ * written. `SitemapDeliveryParityTest` asserts that equivalence rather than
	2527	+ * trusting it.
	2528	+ *
	2529	+ * The whole set is built to answer for one file, because the index can only
	2530	+ * be assembled from the children that were actually produced. The result is
	2531	+ * cached per document, so that cost is paid once per change and not once
	2532	+ * per crawler request.
	2533	+ *
	2534	+ * @since 2.9.0
	2535	+ *
	2536	+ * @param string $filename Published file name, e.g. 'sitemap.xml'.
	2537	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	2538	+ * @return string|null XML, or null when this site does not publish that name.
	2539	+ */
	2540	+ public function render_document(string $filename, ?array $settings = null): ?string {
	2541	+ $filename = basename($filename);
	2542	+ $settings = $settings ?? $this->get_settings('site');
	2543	+
	2544	+ if (empty($settings['enabled'])) {
	2545	+ return null;
	2546	+ }
	2547	+
	2548	+ $cached = get_transient($this->dynamic_cache_key($filename));
	2549	+ if (self::ABSENT_MARKER === $cached) {
	2550	+ return null;
	2551	+ }
	2552	+ if (is_string($cached) && '' !== $cached) {
	2553	+ return $cached;
	2554	+ }
	2555	+
	2556	+ // A miss builds the whole set, because the index can only be assembled
	2557	+ // from the children that were actually produced. Caching only the
	2558	+ // requested document therefore made a crawler walking the index and its
	2559	+ // children rebuild the entire site's sitemap once per file — every post
	2560	+ // and taxonomy query repeated N times on a public endpoint (#754
	2561	+ // review). The set is built once and stored in full.
	2562	+ return $this->stream_documents($settings, $filename);
	2563	+ }
	2564	+
	2565	+ /**
	2566	+ * Build every document, caching each as it is produced, keeping one.
	2567	+ *
	2568	+ * A miss has to build the whole set, because the index can only be
	2569	+ * assembled from the children that were actually produced. It does not have
	2570	+ * to *hold* the whole set: the static path never keeps more than one page
	2571	+ * in memory, writing each to disk as it goes, and buffering every
	2572	+ * document's XML to return one of them undid that on the request path,
	2573	+ * where a large site's entire sitemap corpus would sit in a single PHP
	2574	+ * process (#754 review).
	2575	+ *
	2576	+ * So the sink writes each document straight to its cache entry and lets it
	2577	+ * go, retaining only the one this request is answering. Peak retention is
	2578	+ * one document, whatever the site's size.
	2579	+ *
	2580	+ * Concurrency: the first request through takes a short lock and does the
	2581	+ * work. One that finds the lock held waits a bounded moment for the winner
	2582	+ * to publish, then builds anyway, because serving a correct sitemap late
	2583	+ * beats serving none.
	2584	+ *
	2585	+ * @since 2.9.0
	2586	+ *
	2587	+ * @param array $settings Sitemap settings.
	2588	+ * @param string $wanted Document this request is answering.
	2589	+ * @return string|null XML for $wanted, or null when the site does not publish it.
	2590	+ */
	2591	+ private function stream_documents(array $settings, string $wanted): ?string {
	2592	+ $lock = self::DYNAMIC_CACHE_PREFIX . 'lock';
	2593	+
	2594	+ if (!$this->acquire_render_lock($lock)) {
	2595	+ for ($attempt = 0; $attempt < self::RENDER_LOCK_WAIT_ATTEMPTS; $attempt++) {
	2596	+ usleep(self::RENDER_LOCK_WAIT_MICROSECONDS);
	2597	+
	2598	+ $cached = get_transient($this->dynamic_cache_key($wanted));
	2599	+ if (self::ABSENT_MARKER === $cached) {
	2600	+ return null;
	2601	+ }
	2602	+ if (is_string($cached) && '' !== $cached) {
	2603	+ return $cached;
	2604	+ }
	2605	+ }
	2606	+ }
	2607	+
	2608	+ $kept = null;
	2609	+ // Names only. Keeping the bodies here would be the very retention this
	2610	+ // method exists to avoid.
	2611	+ $produced = [];
	2612	+
	2613	+ $previous = $this->document_sink;
	2614	+ $this->document_sink = function (string $name, string $xml) use (&$kept, &$produced, $wanted): void {
	2615	+ $produced[$name] = true;
	2616	+ set_transient($this->dynamic_cache_key($name), $xml, self::DYNAMIC_CACHE_TTL);
	2617	+
	2618	+ if ($name === $wanted) {
	2619	+ $kept = $xml;
	2620	+ }
	2621	+ };
	2622	+
	2623	+ try {
	2624	+ $this->generate_and_save($settings);
	2625	+
	2626	+ // Names the configuration lists but this build did not produce get
	2627	+ // a negative entry, so asking for one again is a cache hit rather
	2628	+ // than another full rebuild.
	2629	+ $absent = $this->published_document_names($settings);
	2630	+
	2631	+ // Also the exact name this request asked for: a paginated page past
	2632	+ // the end of a stem is a legitimate request shape that the base
	2633	+ // list cannot enumerate, and without an entry it would rebuild on
	2634	+ // every hit.
	2635	+ $absent[] = $wanted;
	2636	+
	2637	+ foreach (array_unique($absent) as $name) {
	2638	+ if (!isset($produced[$name])) {
	2639	+ set_transient($this->dynamic_cache_key($name), self::ABSENT_MARKER, self::DYNAMIC_CACHE_TTL);
	2640	+ }
	2641	+ }
	2642	+ } finally {
	2643	+ $this->document_sink = $previous;
	2644	+ delete_transient($lock);
	2645	+ }
	2646	+
	2647	+ return $kept;
	2648	+ }
	2649	+
	2650	+ /**
	2651	+ * Take the render lock, if it is free.
	2652	+ *
	2653	+ * Not atomic across processes, and deliberately so: the fallback for losing
	2654	+ * a race is duplicated work, never a wrong or missing sitemap, so a
	2655	+ * heavier primitive would buy nothing here.
	2656	+ *
	2657	+ * @since 2.9.0
	2658	+ *
	2659	+ * @param string $lock Lock transient name.
	2660	+ * @return bool True when this request holds the lock.
	2661	+ */
	2662	+ private function acquire_render_lock(string $lock): bool {
	2663	+ if (false !== get_transient($lock)) {
	2664	+ return false;
	2665	+ }
	2666	+
	2667	+ set_transient($lock, time(), self::RENDER_LOCK_TTL);
	2668	+
	2669	+ return true;
	2670	+ }
	2671	+
	2672	+ /**
	2673	+ * Build every document this site publishes and return them all.
	2674	+ *
	2675	+ * Verification and tooling only. This retains the whole set in memory, so
	2676	+ * it must never be used to answer a request: {@see self::stream_documents()}
	2677	+ * is the serving path and keeps one document at a time regardless of site
	2678	+ * size (#754 review). `SitemapDeliveryParityTest` enforces that separation
	2679	+ * by failing if the request path routes back through here.
	2680	+ *
	2681	+ * @since 2.9.0
	2682	+ *
	2683	+ * @param array $settings Sitemap settings.
	2684	+ * @return array<string,string> Filename => XML.
	2685	+ */
	2686	+ public function collect_documents(array $settings): array {
	2687	+ $documents = [];
	2688	+
	2689	+ $previous = $this->document_sink;
	2690	+ $this->document_sink = static function (string $name, string $xml) use (&$documents): void {
	2691	+ $documents[$name] = $xml;
	2692	+ };
	2693	+
	2694	+ try {
	2695	+ $this->generate_and_save($settings);
	2696	+ } finally {
	2697	+ $this->document_sink = $previous;
	2698	+ }
	2699	+
	2700	+ return $documents;
	2701	+ }
	2702	+
	2703	+ /**
	2704	+ * The file names this site publishes, without building their contents.
	2705	+ *
	2706	+ * Used by the request router to decide whether a URL is ours before doing
	2707	+ * any work. Cheap: it reads the configured child list rather than querying
	2708	+ * for entries.
	2709	+ *
	2710	+ * @since 2.9.0
	2711	+ *
	2712	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	2713	+ * @return string[] File names, including paginated pages that may exist.
	2714	+ */
	2715	+ public function published_document_names(?array $settings = null): array {
	2716	+ $settings = $settings ?? $this->get_settings('site');
	2717	+ $resolved = $this->maybe_promote_to_index($settings);
	2718	+
	2719	+ $names = [$this->get_primary_sitemap_filename($settings), 'local-sitemap.xml'];
	2720	+
	2721	+ foreach ((array) ($resolved['sitemap_urls'] ?? []) as $child) {
	2722	+ if (!is_array($child) || empty($child['enabled'])) {
	2723	+ continue;
	2724	+ }
	2725	+
	2726	+ $path = (string) wp_parse_url((string) ($child['url'] ?? ''), PHP_URL_PATH);
	2727	+ if ('' !== $path) {
	2728	+ $names[] = basename($path);
	2729	+ }
	2730	+ }
	2731	+
	2732	+ return array_values(array_unique(array_filter($names)));
	2733	+ }
	2734	+
	2735	+ /**
	2736	+ * Does this site publish a document under that name?
	2737	+ *
	2738	+ * Not a plain membership test against {@see self::published_document_names()}:
	2739	+ * that lists the configured children, and a child over the per-file URL cap
	2740	+ * is split into `<stem>-2.xml`, `<stem>-3.xml` and so on, with every page
	2741	+ * listed in the index. Gating the request router on the base list alone
	2742	+ * therefore 404'd exactly the pages the index points at, which is worse than
	2743	+ * not serving them at all.
	2744	+ *
	2745	+ * Page counts are not knowable without building, so the stem is what is
	2746	+ * matched; a page that does not exist is answered by the build finding
	2747	+ * nothing for it, and is then cached as absent.
	2748	+ *
	2749	+ * @since 2.9.0
	2750	+ *
	2751	+ * @param string $name Requested file name.
	2752	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	2753	+ * @return bool
	2754	+ */
	2755	+ public function publishes_document_name(string $name, ?array $settings = null): bool {
	2756	+ $names = $this->published_document_names($settings);
	2757	+
	2758	+ if (in_array($name, $names, true)) {
	2759	+ return true;
	2760	+ }
	2761	+
	2762	+ if (!preg_match('/^(.*)-\d+\.xml$/i', $name, $m)) {
	2763	+ return false;
	2764	+ }
	2765	+
	2766	+ return in_array($m[1] . '.xml', $names, true);
	2767	+ }
	2768	+
	2769	+ /**
	2770	+ * Transient key for a rendered document.
	2771	+ *
	2772	+ * @since 2.9.0
	2773	+ *
	2774	+ * @param string $filename Published file name.
	2775	+ * @return string
	2776	+ */
	2777	+ private function dynamic_cache_key(string $filename): string {
	2778	+ return self::DYNAMIC_CACHE_PREFIX . md5($filename);
	2779	+ }
	2780	+
	2781	+ /**
	2782	+ * Drop every cached dynamic document.
	2783	+ *
	2784	+ * Called from the same places that mark the static files stale, so the two
	2785	+ * delivery modes invalidate on identical triggers.
	2786	+ *
	2787	+ * @since 2.9.0
	2788	+ *
	2789	+ * @return void
	2790	+ */
	2791	+ public function flush_dynamic_cache(): void {
	2792	+ foreach ($this->published_document_names() as $name) {
	2793	+ delete_transient($this->dynamic_cache_key($name));
	2794	+ }
	2795	+ }
	2796	+
	2797	+ /**
1603	2798	* Resolve index-vs-single mode, synthesizing child sitemaps when needed.
1604	2799	*
1605	2800	* - When use_sitemap_index is on but no child sitemaps are configured, build
1606	2801	* the per-type segmented set so a real <sitemapindex> is produced (#127).
		@@ -1620,8 +2815,22 @@
1620	2815	$mode_supplied = array_key_exists('use_sitemap_index', $settings);
1621	2816	$urls_supplied = is_array($settings['sitemap_urls'] ?? null);
1622	2817	$has_children = count($urls_supplied ? $settings['sitemap_urls'] : []) > 1;
1623	2818
	2819	+ // Which inclusion flags did the caller actually name? The child list is
	2820	+ // the only thing that reads them, and inheriting a saved one skipped
	2821	+ // that — so on an index-mode site the include_* flags were enforced
	2822	+ // nowhere but in the browser, where SitemapGeneration.js recomputes
	2823	+ // sitemap_urls itself. Every non-UI client, the shipped
	2824	+ // `update-sitemap-settings` ability included, saved the flag and changed
	2825	+ // nothing (#398). Read from the payload for the same reason as above:
	2826	+ // after the merge every saved flag would look like one the caller named.
	2827	+ $named_inclusions = array_intersect(
	2828	+ array_keys(self::INCLUSION_CHILD_TYPES),
	2829	+ array_keys($settings)
	2830	+ );
	2831	+ $inclusions_supplied = (bool) $named_inclusions;
	2832	+
1624	2833	// Inclusion flags may be absent from a partial payload (e.g. the manual
1625	2834	// generate endpoint) — fall back to saved settings so synthesized child
1626	2835	// sitemaps reflect the real include_posts/pages/categories choices.
1627	2836	$inclusions = array_merge($saved, $settings);
		@@ -1657,12 +2866,26 @@
1657	2866	$settings['use_sitemap_index'] = $saved['use_sitemap_index'] ?? '';
1658	2867
1659	2868	// Inheriting the mode means inheriting its children too, unless the
1660	2869	// caller named its own set.
	2870	+ $saved_children = (is_array($saved['sitemap_urls'] ?? null) ? $saved['sitemap_urls'] : []);
1661	2871	if (!empty($settings['use_sitemap_index'])
1662	2872	&& !$urls_supplied
1663		- && count((is_array($saved['sitemap_urls'] ?? null) ? $saved['sitemap_urls'] : [])) > 1) {
1664		- $settings['sitemap_urls'] = $saved['sitemap_urls'];
	2873	+ && count($saved_children) > 1) {
	2874	+ // A named inclusion flag is applied *to* the inherited list, not
	2875	+ // used to regenerate it. build_segmented_sitemap_urls() also adds
	2876	+ // a child for every public custom post type, so rebuilding here
	2877	+ // would make `{include_pages: false}` — one thing off — silently
	2878	+ // switch on children the saved list never had (an Elementor
	2879	+ // internal CPT, a WooCommerce product feed). Only the flags the
	2880	+ // caller actually named change anything.
	2881	+ $settings['sitemap_urls'] = $inclusions_supplied
	2882	+ ? $this->apply_inclusion_flags_to_children($saved_children, $named_inclusions, $inclusions)
	2883	+ : $saved_children;
	2884	+
	2885	+ // The children are resolved either way — including when the
	2886	+ // caller switched the last one off, which leaves a bare index and
	2887	+ // is what they asked for.
1665	2888	$has_children = true;
1666	2889	}
1667	2890	}
1668	2891
		@@ -1723,26 +2946,9 @@
1723	2946	* @param array $settings Sitemap settings.
1724	2947	* @return string Sitemap filename.
1725	2948	*/
1726	2949	public function get_primary_sitemap_filename(array $settings): string {
1727		- $use_index = !empty($settings['use_sitemap_index']);
1728		-
1729		- foreach ((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : []) as $config) {
1730		- if (empty($config['enabled']) || empty($config['url'])) {
1731		- continue;
1732		- }
1733		-
1734		- if ($use_index !== (($config['type'] ?? '') === 'index')) {
1735		- continue;
1736		- }
1737		-
1738		- $path = wp_parse_url($config['url'], PHP_URL_PATH);
1739		- if (!empty($path)) {
1740		- return basename($path);
1741		- }
1742		- }
1743		-
1744		- return $use_index ? 'sitemap_index.xml' : 'sitemap.xml';
	2950	+ return thinkrank_webroot_primary_sitemap_filename($settings);
1745	2951	}
1746	2952
1747	2953	/**
1748	2954	* Public URL of the sitemap the site serves.
		@@ -1784,8 +2990,18 @@
1784	2990	// File validation failed - error details available in exception
1785	2991	return false;
1786	2992	}
1787	2993
	2994	+ // Dynamic delivery: hand the document to the collector instead of the
	2995	+ // filesystem. Reported as published, because for this run it is — the
	2996	+ // caller's success/failure bookkeeping and the index assembly both key
	2997	+ // off this return value.
	2998	+ if ($this->document_sink !== null) {
	2999	+ ($this->document_sink)($filename, $sitemap_xml);
	3000	+
	3001	+ return true;
	3002	+ }
	3003	+
1788	3004	$sitemap_path = ABSPATH . $filename;
1789	3005
1790	3006	// Use WordPress filesystem API for better security
1791	3007	global $wp_filesystem;
		@@ -1794,9 +3010,22 @@
1794	3010	WP_Filesystem();
1795	3011	}
1796	3012
1797	3013	if ($wp_filesystem) {
1798		- return $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3014	+ $written = $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3015	+
	3016	+ if ($written) {
	3017	+ // Every sitemap this version writes carries the ownership
	3018	+ // marker, so once one has been written an unmarked file at one
	3019	+ // of our names cannot be ours. Recording that retires the
	3020	+ // legacy fallback for this install — see
	3021	+ // thinkrank_webroot_sitemap_is_ours().
	3022	+ if (get_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION) !== '1') {
	3023	+ update_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION, '1', false);
	3024	+ }
	3025	+ }
	3026	+
	3027	+ return $written;
1799	3028	}
1800	3029
1801	3030	// WP_Filesystem initialization failed
1802	3031	return false;
		@@ -1819,52 +3048,164 @@
1819	3048	*/
1820	3049	public function build_segmented_sitemap_urls(array $inclusions): array {
1821	3050	$pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml';
1822	3051
1823		- $entry = static function (string $url, string $type): array {
1824		- return [
1825		- 'url' => $url,
1826		- 'type' => $type,
1827		- 'enabled' => true,
1828		- 'last_checked' => null,
1829		- 'status' => 'unknown',
1830		- ];
1831		- };
1832		- $child = static function (string $type) use ($pattern, $entry): array {
1833		- $file = str_replace('{type}', $type, $pattern);
1834		- if (strpos($file, '/') !== 0) {
1835		- $file = '/' . $file;
	3052	+ $urls = [$this->sitemap_child_entry('/sitemap_index.xml', 'index')];
	3053	+
	3054	+ foreach (self::INCLUSION_CHILD_TYPES as $flag => $type) {
	3055	+ if (!empty($inclusions[$flag])) {
	3056	+ $urls[] = $this->build_child_sitemap_entry($type, $pattern);
1836	3057	}
1837		- return $entry($file, $type);
1838		- };
1839		-
1840		- $urls = [$entry('/sitemap_index.xml', 'index')];
1841		-
1842		- if (!empty($inclusions['include_posts'])) {
1843		- $urls[] = $child('posts');
1844	3058	}
1845		- if (!empty($inclusions['include_pages'])) {
1846		- $urls[] = $child('pages');
1847		- }
1848		- if (!empty($inclusions['include_categories'])) {
1849		- $urls[] = $child('categories');
1850		- }
1851		- if (!empty($inclusions['include_tags'])) {
1852		- $urls[] = $child('tags');
1853		- }
1854	3059
1855	3060	// Public custom post types each get a child sitemap (parity with the
1856	3061	// "complete" preset and with Rank Math, which lists every public CPT).
1857	3062	foreach (get_post_types(['public' => true, '_builtin' => false], 'names') as $cpt) {
1858		- if ($this->should_include_post_type($cpt)) {
1859		- $urls[] = $child($cpt);
	3063	+ if (!$this->should_include_post_type($cpt)) {
	3064	+ continue;
1860	3065	}
	3066	+
	3067	+ // ...and the per-content-type sitemap switch (#660).
	3068	+ // get_enabled_post_types() already honours it, but this list is what
	3069	+ // index mode builds its children from — so without the same test a
	3070	+ // CPT the user had switched off still got its own child sitemap,
	3071	+ // created and streamed in full. The flags live in $inclusions, which
	3072	+ // is the settings array these children are derived from.
	3073	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $cpt, $inclusions)) {
	3074	+ continue;
	3075	+ }
	3076	+
	3077	+ $urls[] = $this->build_child_sitemap_entry($cpt, $pattern);
1861	3078	}
1862	3079
	3080	+ // Public custom taxonomies get the same treatment (#690). Flat mode has
	3081	+ // always walked them through get_enabled_taxonomies(); index mode built
	3082	+ // its children from the list above and never consulted a taxonomy at
	3083	+ // all, so every custom-taxonomy archive silently vanished from the
	3084	+ // sitemap the moment a site switched modes — and the per-taxonomy switch
	3085	+ // the matrix writes had nothing to act on. Same two tests the post-type
	3086	+ // walk applies, in the same order.
	3087	+ $taken = array_column($urls, 'type');
	3088	+
	3089	+ foreach (get_taxonomies(['public' => true, '_builtin' => false], 'names') as $taxonomy) {
	3090	+ if (!$this->should_include_taxonomy($taxonomy)) {
	3091	+ continue;
	3092	+ }
	3093	+
	3094	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $inclusions)) {
	3095	+ continue;
	3096	+ }
	3097	+
	3098	+ // Post types and taxonomies are separate registries, so a site can
	3099	+ // hold both a `foo` post type and a `foo` taxonomy. They would
	3100	+ // resolve to one filename, and stream_type_entries() answers post
	3101	+ // types first, so the second child would list the first one's file
	3102	+ // twice in the index rather than adding anything.
	3103	+ if (in_array($taxonomy, $taken, true)) {
	3104	+ continue;
	3105	+ }
	3106	+
	3107	+ $urls[] = $this->build_child_sitemap_entry($taxonomy, $pattern);
	3108	+ }
	3109	+
1863	3110	return $urls;
1864	3111	}
1865	3112
1866	3113	/**
	3114	+ * Apply only the inclusion flags the caller named to an existing child list.
	3115	+ *
	3116	+ * The narrow counterpart to build_segmented_sitemap_urls(): that one
	3117	+ * regenerates the whole set from scratch, which is right when there is no set
	3118	+ * yet and wrong when there is. Rebuilding an existing list would add a child
	3119	+ * for every public custom post type it had never contained, so a payload that
	3120	+ * switches one thing off would switch others on. Here a flag adds or removes
	3121	+ * exactly its own child and leaves every other entry — custom post types,
	3122	+ * hand-added URLs, per-child enabled/status state — untouched (#398).
	3123	+ *
	3124	+ * @since 1.31.0
	3125	+ *
	3126	+ * @param array $children Existing child sitemap entries.
	3127	+ * @param string[] $named_inclusions Inclusion flag keys present in the payload.
	3128	+ * @param array $inclusions Merged settings, for the resolved flag
	3129	+ * values and custom_url_pattern.
	3130	+ * @return array Updated child sitemap entries.
	3131	+ */
	3132	+ private function apply_inclusion_flags_to_children(
	3133	+ array $children,
	3134	+ array $named_inclusions,
	3135	+ array $inclusions
	3136	+ ): array {
	3137	+ $pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml';
	3138	+
	3139	+ foreach ($named_inclusions as $flag) {
	3140	+ $type = self::INCLUSION_CHILD_TYPES[$flag];
	3141	+
	3142	+ $present = false;
	3143	+ foreach ($children as $entry) {
	3144	+ if (($entry['type'] ?? '') === $type) {
	3145	+ $present = true;
	3146	+ break;
	3147	+ }
	3148	+ }
	3149	+
	3150	+ if (empty($inclusions[$flag])) {
	3151	+ if ($present) {
	3152	+ $children = array_values(array_filter(
	3153	+ $children,
	3154	+ static function ($entry) use ($type): bool {
	3155	+ return (is_array($entry) ? ($entry['type'] ?? '') : '') !== $type;
	3156	+ }
	3157	+ ));
	3158	+ }
	3159	+ continue;
	3160	+ }
	3161	+
	3162	+ if (!$present) {
	3163	+ $children[] = $this->build_child_sitemap_entry($type, $pattern);
	3164	+ }
	3165	+ }
	3166	+
	3167	+ return $children;
	3168	+ }
	3169	+
	3170	+ /**
	3171	+ * Build one child sitemap entry, resolving its filename from the url pattern.
	3172	+ *
	3173	+ * @since 1.31.0
	3174	+ *
	3175	+ * @param string $type Child sitemap type (posts, pages, a post type name).
	3176	+ * @param string $pattern Filename pattern containing {type}.
	3177	+ * @return array Sitemap URL config.
	3178	+ */
	3179	+ private function build_child_sitemap_entry(string $type, string $pattern): array {
	3180	+ $file = str_replace('{type}', $type, $pattern);
	3181	+ if (strpos($file, '/') !== 0) {
	3182	+ $file = '/' . $file;
	3183	+ }
	3184	+
	3185	+ return $this->sitemap_child_entry($file, $type);
	3186	+ }
	3187	+
	3188	+ /**
	3189	+ * The shape generate_multiple_sitemaps() expects of a sitemap_urls entry.
	3190	+ *
	3191	+ * @since 1.31.0
	3192	+ *
	3193	+ * @param string $url Sitemap path.
	3194	+ * @param string $type Entry type.
	3195	+ * @return array Sitemap URL config.
	3196	+ */
	3197	+ private function sitemap_child_entry(string $url, string $type): array {
	3198	+ return [
	3199	+ 'url' => $url,
	3200	+ 'type' => $type,
	3201	+ 'enabled' => true,
	3202	+ 'last_checked' => null,
	3203	+ 'status' => 'unknown',
	3204	+ ];
	3205	+ }
	3206	+
	3207	+ /**
1867	3208	* Generate multiple sitemaps based on settings
1868	3209	*
1869	3210	* @since 1.0.0
1870	3211	* @param array $settings Sitemap settings
		@@ -1913,8 +3254,48 @@
1913	3254	if (post_type_exists($type) && !$this->should_include_post_type($type)) {
1914	3255	continue;
1915	3256	}
1916	3257
	3258	+ // Same for the matrix switch: a child list saved before the user
	3259	+ // excluded this content type still names it, and regenerating from
	3260	+ // that list would rewrite the file they asked not to have. Built-in
	3261	+ // aggregates ('posts', 'pages', ...) are not post type names, so
	3262	+ // post_type_exists() keeps this to real custom post types.
	3263	+ if (post_type_exists($type)
	3264	+ && !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $type, $settings)) {
	3265	+ continue;
	3266	+ }
	3267	+
	3268	+ // The taxonomy counterpart of the two guards above (#690). A child
	3269	+ // list saved while a taxonomy was still included keeps naming it, so
	3270	+ // without this, excluding one in the matrix would still rewrite and
	3271	+ // re-list the file the user asked not to have. The built-in
	3272	+ // aggregates are named 'categories'/'tags' rather than
	3273	+ // 'category'/'post_tag', so taxonomy_exists() leaves them to the
	3274	+ // inclusion-flag check below.
	3275	+ $child_taxonomy = self::CHILD_TYPE_ALIASES[$type] ?? $type;
	3276	+ if (taxonomy_exists($child_taxonomy)
	3277	+ && (!$this->should_include_taxonomy($child_taxonomy)
	3278	+ || !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $child_taxonomy, $settings))) {
	3279	+ continue;
	3280	+ }
	3281	+
	3282	+ // The built-in aggregates carry their switch in the inclusion flag
	3283	+ // rather than under a post type name, and they are the four most
	3284	+ // people actually use. build_segmented_sitemap_urls() drops a child
	3285	+ // whose flag is empty; regenerating from a list saved while it was
	3286	+ // still on has to make the same decision, or turning Posts, Pages,
	3287	+ // Categories or Tags off in the matrix rewrites and re-lists the
	3288	+ // very file it was asked to remove.
	3289	+ // An absent flag means "not configured", which every other reader
	3290	+ // treats as included; only a flag that is present and off excludes.
	3291	+ $aggregate_flag = array_search($type, self::INCLUSION_CHILD_TYPES, true);
	3292	+ if ($aggregate_flag !== false
	3293	+ && array_key_exists($aggregate_flag, $settings)
	3294	+ && empty($settings[$aggregate_flag])) {
	3295	+ continue;
	3296	+ }
	3297	+
1917	3298	try {
1918	3299	// The single "general"/"WordPress" sitemap is one un-paginated file
1919	3300	// (there is no index to reference extra pages); it no longer drops
1920	3301	// overflow URLs.
		@@ -1959,9 +3340,9 @@
1959	3340	$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children);
1960	3341	}
1961	3342
1962	3343	// Remove pages left over from a previous, larger generation.
1963		- $this->cleanup_stale_pages($sitemap_config['url'], $page);
	3344	+ $this->cleanup_stale_pages($sitemap_config['url'], $page, $settings);
1964	3345
1965	3346	} catch (\Exception $e) {
1966	3347	$results['errors'][] = "Error generating {$type} sitemap: " . $e->getMessage();
1967	3348	$results['success'] = false;
		@@ -1980,10 +3361,16 @@
1980	3361	$results['errors'][] = 'Error generating local sitemap: ' . $e->getMessage();
1981	3362	$results['success'] = false;
1982	3363	}
1983	3364
1984		- // Build the index last, from the child files actually generated.
	3365	+ // Build the index last, from the child files actually generated, plus the
	3366	+ // sitemaps other plugins own: those serve their own URLs and write no
	3367	+ // file here, so they are appended to the index only (#104).
1985	3368	if ($index_config !== null) {
	3369	+ foreach (self::additional_sitemaps() as $extra) {
	3370	+ $index_children[] = ['url' => $extra];
	3371	+ }
	3372	+
1986	3373	try {
1987	3374	$index_xml = $this->generate_sitemap_index($index_children, $settings);
1988	3375	$filename = basename(wp_parse_url($index_config['url'], PHP_URL_PATH));
1989	3376
		@@ -2028,8 +3415,15 @@
2028	3415	* @param array $generated Entries from $results['sitemaps_generated'].
2029	3416	* @return string[] Basenames removed.
2030	3417	*/
2031	3418	private function prune_orphaned_segments(array $settings, array $generated): array {
	3419	+ // Rendering for a request, not publishing: there is nothing on disk
	3420	+ // this run owns, and a dynamic render must never delete the files a
	3421	+ // site's previous static mode left behind.
	3422	+ if ($this->is_collecting()) {
	3423	+ return [];
	3424	+ }
	3425	+
2032	3426	$kept = [];
2033	3427	foreach ($generated as $entry) {
2034	3428	if (!empty($entry['filename'])) {
2035	3429	$kept[strtolower((string) $entry['filename'])] = true;
		@@ -2035,15 +3429,22 @@
2035	3429	$kept[strtolower((string) $entry['filename'])] = true;
2036	3430	}
2037	3431	}
2038	3432
2039		- // The index and the local business sitemap are written by their own
2040		- // paths and are not segments, so they are never orphans here.
2041		- $kept[strtolower(basename($this->get_primary_sitemap_filename($settings)))] = true;
	3433	+ // The current mode's primary and the local business sitemap are written
	3434	+ // by their own paths and are never orphans here.
	3435	+ $primary = strtolower(basename($this->get_primary_sitemap_filename($settings)));
	3436	+ $kept[$primary] = true;
2042	3437	$kept['local-sitemap.xml'] = true;
2043		- $kept['sitemap.xml'] = true;
2044		- $kept['sitemap_index.xml'] = true;
2045	3438
	3439	+ // The OTHER mode's primary is an orphan the moment the mode changes:
	3440	+ // index mode leaves sitemap.xml behind, flat mode leaves
	3441	+ // sitemap_index.xml and its children. Both used to be kept
	3442	+ // unconditionally, so the site served two sitemap trees and only ever
	3443	+ // refreshed one (#563). The children are already covered by the segment
	3444	+ // sweep below, which now sees them because the index is no longer kept.
	3445	+ $stale_primaries = array_diff(['sitemap.xml', 'sitemap_index.xml'], [$primary]);
	3446	+
2046	3447	$removed = [];
2047	3448
2048	3449	global $wp_filesystem;
2049	3450	if (!$wp_filesystem) {
		@@ -2053,9 +3454,11 @@
2053	3454	if (!$wp_filesystem) {
2054	3455	return $removed;
2055	3456	}
2056	3457
2057		- foreach ($this->publishable_segment_filenames($settings) as $candidate) {
	3458	+ $candidates = array_merge($this->publishable_segment_filenames($settings), $stale_primaries);
	3459	+
	3460	+ foreach ($candidates as $candidate) {
2058	3461	if (isset($kept[strtolower($candidate)])) {
2059	3462	continue;
2060	3463	}
2061	3464
		@@ -2074,8 +3477,13 @@
2074	3477	foreach ($paths as $path) {
2075	3478	if (!file_exists($path)) {
2076	3479	continue;
2077	3480	}
	3481	+ // A name we could have published is not proof we published
	3482	+ // this file: RankMath and core write at the same paths (#515).
	3483	+ if (!$this->webroot_sitemap_is_ours($path, $settings)) {
	3484	+ continue;
	3485	+ }
2078	3486	if ($wp_filesystem->delete($path)) {
2079	3487	$removed[] = basename($path);
2080	3488	}
2081	3489	}
		@@ -2111,11 +3519,15 @@
2111	3519	$settings = $settings ?? $this->get_settings('site');
2112	3520	$entries = $this->collect_local_entries();
2113	3521
2114	3522	if (empty($entries)) {
2115		- // Business identity was cleared — drop any file left from before.
	3523	+ // Business identity was cleared — drop the file we left from
	3524	+ // before, but only ours. `local-sitemap.xml` is the name Rank Math
	3525	+ // publishes under too (this method mirrors it deliberately), so on
	3526	+ // a migrated site the file at that path may never have been ours
	3527	+ // to delete (#515).
2116	3528	$path = ABSPATH . 'local-sitemap.xml';
2117		- if (file_exists($path)) {
	3529	+ if (!$this->is_collecting() && file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) {
2118	3530	wp_delete_file($path);
2119	3531	}
2120	3532	return false;
2121	3533	}
		@@ -2137,8 +3549,25 @@
2137	3549	*
2138	3550	* @since 1.15.x
2139	3551	* @return array Zero or one URL entry
2140	3552	*/
	3553	+ /**
	3554	+ * Does this site publish a local business sitemap right now?
	3555	+ *
	3556	+ * The same gate {@see self::regenerate_local_sitemap()} applies, asked
	3557	+ * without writing anything. Callers that need to know whether the document
	3558	+ * exists must not test the filesystem: under dynamic delivery it is served
	3559	+ * from PHP and there is no file, which is how `local-sitemap.xml` came to be
	3560	+ * dropped from robots.txt on exactly those sites (#752).
	3561	+ *
	3562	+ * @since 2.9.0
	3563	+ *
	3564	+ * @return bool True when the local sitemap has content to publish.
	3565	+ */
	3566	+ public function publishes_local_sitemap(): bool {
	3567	+ return !empty($this->collect_local_entries());
	3568	+ }
	3569	+
2141	3570	private function collect_local_entries(): array {
2142	3571	if (!class_exists('ThinkRank\\SEO\\Site_Identity_Manager')) {
2143	3572	require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-site-identity-manager.php';
2144	3573	}
		@@ -2210,17 +3639,43 @@
2210	3639	case 'products':
2211	3640	$entries = post_type_exists('product') ? $this->collect_post_entries_iter(['product'], $settings) : [];
2212	3641	return ['entries' => $entries, 'image_ns' => true];
2213	3642
2214		- case 'product_categories':
2215		- $entries = taxonomy_exists('product_cat') ? $this->collect_taxonomy_entries_iter('product_cat', $settings) : [];
2216		- return ['entries' => $entries, 'image_ns' => false];
	3643	+ default:
	3644	+ // Resolve a preset's display name to the object it streams, so
	3645	+ // 'product_categories' is an ordinary taxonomy child rather than
	3646	+ // a case of its own (#690).
	3647	+ $alias = self::CHILD_TYPE_ALIASES[$type] ?? null;
	3648	+ $object = $alias ?? $type;
2217	3649
2218		- default:
2219	3650	$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names');
2220		- if (in_array($type, $custom_post_types, true)) {
2221		- return ['entries' => $this->collect_post_entries_iter([$type], $settings), 'image_ns' => true];
	3651	+ if (in_array($object, $custom_post_types, true)) {
	3652	+ return ['entries' => $this->collect_post_entries_iter([$object], $settings), 'image_ns' => true];
2222	3653	}
	3654	+
	3655	+ // Custom taxonomies reach index mode here, streamed through the
	3656	+ // same iterator flat mode uses so the two modes emit identical
	3657	+ // URLs for the same settings.
	3658	+ if (taxonomy_exists($object) && $this->should_include_taxonomy($object)) {
	3659	+ return ['entries' => $this->collect_taxonomy_entries_iter($object, $settings), 'image_ns' => false];
	3660	+ }
	3661	+
	3662	+ // An aliased child whose object is gone (WooCommerce deactivated)
	3663	+ // keeps writing the empty file it always wrote. Returning null
	3664	+ // here would hand it the whole-site fallback instead, dumping
	3665	+ // every URL on the site into a file named for products.
	3666	+ //
	3667	+ // A registered taxonomy this generator will not emit
	3668	+ // (`post_format`, `nav_menu`, a non-public one) needs the same
	3669	+ // answer for the same reason. generate_multiple_sitemaps()
	3670	+ // skips those before they reach here, so nothing takes this
	3671	+ // path today — but it is the one branch where falling through
	3672	+ // to null is silently catastrophic rather than merely wrong,
	3673	+ // and the guard keeping it unreachable lives in another method.
	3674	+ if ($alias !== null || taxonomy_exists($object)) {
	3675	+ return ['entries' => [], 'image_ns' => false];
	3676	+ }
	3677	+
2223	3678	return null;
2224	3679	}
2225	3680	}
2226	3681
		@@ -2263,13 +3718,23 @@
2263	3718	* which has no -N suffix) is never touched.
2264	3719	*
2265	3720	* @since 1.14.0
2266	3721	*
	3722	+ * @since 2.1.1 Each candidate must pass the content ownership test — a
	3723	+ * `-N.xml` page of another plugin's sitemap paginates our
	3724	+ * stem exactly as ours does (#515).
	3725	+ *
2267	3726	* @param string $base_url Base (page 1) sitemap URL
2268	3727	* @param int $current_pages Number of pages generated this run
	3728	+ * @param array $settings Sitemap settings, for the ownership test.
2269	3729	* @return void
2270	3730	*/
2271		- private function cleanup_stale_pages(string $base_url, int $current_pages): void {
	3731	+ private function cleanup_stale_pages(string $base_url, int $current_pages, array $settings): void {
	3732	+ // See prune_orphaned_segments(): a dynamic render deletes nothing.
	3733	+ if ($this->is_collecting()) {
	3734	+ return;
	3735	+ }
	3736	+
2272	3737	$filename = basename(wp_parse_url($base_url, PHP_URL_PATH));
2273	3738	if (!preg_match('/^(.*)\.xml$/i', $filename, $m)) {
2274	3739	return;
2275	3740	}
		@@ -2286,9 +3751,11 @@
2286	3751
2287	3752	$candidates = glob(ABSPATH . $stem . '-*.xml') ?: [];
2288	3753	foreach ($candidates as $path) {
2289	3754	// Only delete numeric-suffixed pages beyond the current count.
2290		- if (preg_match('/-(\d+)\.xml$/', basename($path), $mm) && (int) $mm[1] > $current_pages) {
	3755	+ if (preg_match('/-(\d+)\.xml$/', basename($path), $mm)
	3756	+ && (int) $mm[1] > $current_pages
	3757	+ && $this->webroot_sitemap_is_ours($path, $settings)) {
2291	3758	$wp_filesystem->delete($path);
2292	3759	}
2293	3760	}
2294	3761	}
		@@ -2293,8 +3760,97 @@
2293	3760	}
2294	3761	}
2295	3762
2296	3763	/**
	3764	+ * Sitemap URLs contributed by other plugins.
	3765	+ *
	3766	+ * ThinkRank owns the sitemap index and the robots.txt `Sitemap:` lines, so a
	3767	+ * companion plugin that serves its own sitemap — Pro's news and video
	3768	+ * sitemaps, for instance — had no way to be discovered: it appeared in
	3769	+ * neither, leaving manual Search Console submission as the only route in
	3770	+ * (#104). Registering here puts a sitemap in the index when one exists, and
	3771	+ * in robots.txt when it does not.
	3772	+ *
	3773	+ * Callers get root-relative paths. Entries are normalised to a leading
	3774	+ * slash, de-duplicated, and anything that is not a non-empty string is
	3775	+ * dropped, so one badly-behaved callback cannot produce a malformed index.
	3776	+ *
	3777	+ * @since 2.3.1
	3778	+ *
	3779	+ * @return string[] Root-relative sitemap paths, e.g. ['/news-sitemap.xml'].
	3780	+ */
	3781	+ public static function additional_sitemaps(): array {
	3782	+ /**
	3783	+ * Filters the sitemaps contributed by other plugins.
	3784	+ *
	3785	+ * @since 2.3.1
	3786	+ *
	3787	+ * @param string[] $sitemaps Root-relative sitemap paths.
	3788	+ */
	3789	+ $sitemaps = apply_filters('thinkrank_additional_sitemaps', []);
	3790	+
	3791	+ if (!is_array($sitemaps)) {
	3792	+ return [];
	3793	+ }
	3794	+
	3795	+ // Both consumers resolve an entry with home_url(), which prefixes the
	3796	+ // install's own directory. Everything below is measured against that so
	3797	+ // an absolute URL is reduced to what home_url() will put back.
	3798	+ $home = wp_parse_url(home_url('/'));
	3799	+ $home_host = strtolower((string) ($home['host'] ?? ''));
	3800	+ $home_path = '/' . trim((string) ($home['path'] ?? ''), '/');
	3801	+
	3802	+ $clean = [];
	3803	+ foreach ($sitemaps as $sitemap) {
	3804	+ if (!is_string($sitemap)) {
	3805	+ continue;
	3806	+ }
	3807	+
	3808	+ $sitemap = trim($sitemap);
	3809	+ if ('' === $sitemap) {
	3810	+ continue;
	3811	+ }
	3812	+
	3813	+ // A full URL on this site is accepted and reduced to the part
	3814	+ // home_url() does not already supply, so a caller that reached for
	3815	+ // home_url() still lands in the right place — including on a
	3816	+ // subdirectory install, where keeping the whole path would repeat
	3817	+ // the directory. A URL on another host is dropped rather than
	3818	+ // rewritten: the sitemaps protocol will not accept a cross-host
	3819	+ // child anyway, and reusing its path would advertise a URL on this
	3820	+ // site that does not exist.
	3821	+ if (preg_match('#^(https?:)?//#i', $sitemap)) {
	3822	+ $parts = wp_parse_url('//' === substr($sitemap, 0, 2) ? 'https:' . $sitemap : $sitemap);
	3823	+ if (!is_array($parts)) {
	3824	+ continue;
	3825	+ }
	3826	+
	3827	+ if (strtolower((string) ($parts['host'] ?? '')) !== $home_host) {
	3828	+ continue;
	3829	+ }
	3830	+
	3831	+ $path = (string) ($parts['path'] ?? '');
	3832	+ if ('' === $path) {
	3833	+ continue;
	3834	+ }
	3835	+
	3836	+ if ('/' !== $home_path && ($path === $home_path || 0 === strpos($path, $home_path . '/'))) {
	3837	+ $path = substr($path, strlen($home_path));
	3838	+ }
	3839	+
	3840	+ // A sitemap served from a query string keeps it; dropping the
	3841	+ // query would point at a different document.
	3842	+ $query = (string) ($parts['query'] ?? '');
	3843	+ $sitemap = $path . ('' !== $query ? '?' . $query : '');
	3844	+ }
	3845	+
	3846	+ $clean[] = '/' . ltrim($sitemap, '/');
	3847	+ }
	3848	+
	3849	+ return array_values(array_unique($clean));
	3850	+ }
	3851	+
	3852	+ /**
2297	3853	* Generate sitemap index XML from the list of child sitemap files produced
2298	3854	* during generation (each already resolved to its final, possibly paginated,
2299	3855	* URL).
2300	3856	*
		@@ -2303,14 +3859,9 @@
2303	3859	* @param array $settings Sitemap settings
2304	3860	* @return string Sitemap index XML
2305	3861	*/
2306	3862	private function generate_sitemap_index(array $children, array $settings): string {
2307		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
2308		-
2309		- // Add XSL stylesheet only if styling is enabled
2310		- if (!empty($settings['enable_styling'])) {
2311		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap-index.xsl') . '"?>' . "\n";
2312		- }
	3863	+ $xml = $this->xml_prolog($settings, 'index');
2313	3864	$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
2314	3865
2315	3866	$site_url = home_url();
2316	3867
		@@ -2323,9 +3874,9 @@
2323	3874	$sitemap_url = $site_url . $sitemap_url;
2324	3875	}
2325	3876
2326	3877	$xml .= " <sitemap>\n";
2327		- $xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n";
	3878	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($sitemap_url)) . "</loc>\n";
2328	3879	$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n";
2329	3880	$xml .= " </sitemap>\n";
2330	3881	}
2331	3882