← All changes | includes/seo/class-sitemap-generator.php +2144 -151 2.1.0 → 2.14.1 View file →

		@@ -62,8 +62,25 @@
62	62	'include_tags' => 'tags',
63	63	];
64	64
65	65	/**
	66	+ * Child sitemap type -> the object that actually supplies its entries.
	67	+ *
	68	+ * A child normally carries its object's own slug, but the sitemap presets UI
	69	+ * writes a display name for the WooCommerce taxonomy child
	70	+ * ('product_categories', not 'product_cat'), so a saved child list can name a
	71	+ * type no post type or taxonomy answers to. Resolving through here lets those
	72	+ * children take the same generic path as every other custom taxonomy instead
	73	+ * of needing a case of their own (#690).
	74	+ *
	75	+ * @since 2.7.0
	76	+ * @var array<string, string>
	77	+ */
	78	+ private const CHILD_TYPE_ALIASES = [
	79	+ 'product_categories' => 'product_cat',
	80	+ ];
	81	+
	82	+ /**
66	83	* Supported sitemap types
67	84	*
68	85	* @since 1.0.0
69	86	* @var array
		@@ -76,8 +93,159 @@
76	93	*/
77	94	private const ID_WALK_CHUNK = 500;
78	95
79	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	+ /**
80	248	* How many term IDs to hydrate at a time while walking a taxonomy.
81	249	*
82	250	* @since 2.0.1
83	251	* @var int
		@@ -83,8 +251,33 @@
83	251	* @var int
84	252	*/
85	253	private const TERM_WALK_CHUNK = 1000;
86	254
	255	+ /**
	256	+ * Exception code for an automatic rebuild stopped short of the memory limit.
	257	+ *
	258	+ * @since 2.10.1
	259	+ * @var int
	260	+ */
	261	+ private const MEMORY_ABORT_CODE = 4290;
	262	+
	263	+ /**
	264	+ * Whether the chunked walks should stop before the memory limit. Only on
	265	+ * for automatic rebuilds, whose failure is recorded and retried.
	266	+ *
	267	+ * @since 2.10.1
	268	+ * @var bool
	269	+ */
	270	+ private bool $memory_guard = false;
	271	+
	272	+ /**
	273	+ * Largest memory cost of one walked chunk in this rebuild, in bytes.
	274	+ *
	275	+ * @since 2.10.1
	276	+ * @var int
	277	+ */
	278	+ private int $walk_chunk_cost = 0;
	279	+
87	280	private array $sitemap_types = [
88	281	'posts' => [
89	282	'name' => 'Posts',
90	283	'post_types' => ['post'],
		@@ -111,25 +304,29 @@
111	304	]
112	305	];
113	306
114	307	/**
	308	+ * The generator the content-change listeners share, built on the first
	309	+ * change of a request.
	310	+ *
	311	+ * @since 2.10.1
	312	+ * @var self|null
	313	+ */
	314	+ private static ?self $listener = null;
	315	+
	316	+ /**
115	317	* Constructor
116	318	*
117	319	* @since 1.0.0
	320	+ * @since 2.10.1 Registers no hooks, whatever `$register_hooks` says. The
	321	+ * content-change listeners are registered once, at bootstrap,
	322	+ * by register_content_listeners().
118	323	*
119		- * @param bool $register_hooks Optional. Whether to register the auto-generation
120		- * hooks. Pass false for a read-only instance built
121		- * solely to query settings — the hooks are bound to
122		- * `$this`, so a second hook-registering instance
123		- * would run `handle_content_change()` twice per save.
	324	+ * @param bool $register_hooks Unused since 2.10.1; kept so existing callers,
	325	+ * Pro's included, keep working.
124	326	*/
125	327	public function __construct(bool $register_hooks = true) {
126	328	parent::__construct('sitemap');
127		-
128		- // Initialize auto-generation hooks
129		- if ($register_hooks) {
130		- $this->init_auto_generation_hooks();
131		- }
132	329	}
133	330
134	331	/**
135	332	* Filter the args of a sitemap post query.
		@@ -173,34 +370,154 @@
173	370	return (array) apply_filters('thinkrank_sitemap_term_query_args', $args);
174	371	}
175	372
176	373	/**
177		- * Initialize WordPress hooks for auto-generation
	374	+ * Register the content-change listeners that queue an automatic rebuild.
178	375	*
179		- * @since 1.0.0
	376	+ * Called once per request, at plugin bootstrap, next to the WP-Cron
	377	+ * listeners that run the rebuild these queue (both in
	378	+ * Plugin::register_sitemap_cron_listeners(), on plugins_loaded). The
	379	+ * constructor used to register them, so they existed only in a request
	380	+ * that happened to build a generator: the REST endpoint, the setup wizard,
	381	+ * an MCP ability. Block-editor saves go through REST and were heard. A
	382	+ * scheduled post published by WP-Cron, Quick Edit, the classic editor and
	383	+ * WP-CLI were not, so the post stayed out of the sitemap with nothing
	384	+ * pending to recover it (#824). Each instance also added its own set, so
	385	+ * one REST save ran the handlers once per generator built.
	386	+ *
	387	+ * The generator is built on the first change and shared for the rest of
	388	+ * the request, so a bulk edit does not construct one per post.
	389	+ *
	390	+ * @since 2.10.1
180	391	* @return void
181	392	*/
182		- private function init_auto_generation_hooks(): void {
183		- // Content change hooks - use priority 20 to run after other plugins
184		- add_action('save_post', [$this, 'handle_content_change'], 20, 2);
185		- add_action('delete_post', [$this, 'handle_content_deletion'], 20);
186		- add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20);
187		- add_action('untrash_post', [$this, 'handle_content_change_by_id'], 20);
	393	+ public static function register_content_listeners(): void {
	394	+ // Priority 20, to run after other plugins.
	395	+ add_action('save_post', static function (int $post_id, \WP_Post $post): void {
	396	+ self::listener()->handle_content_change($post_id, $post);
	397	+ }, 20, 2);
	398	+ add_action('delete_post', static function (int $post_id): void {
	399	+ self::listener()->handle_content_deletion($post_id);
	400	+ }, 20);
	401	+ add_action('wp_trash_post', static function (int $post_id): void {
	402	+ self::listener()->handle_content_deletion($post_id);
	403	+ }, 20);
	404	+ add_action('untrash_post', static function (int $post_id): void {
	405	+ self::listener()->handle_content_change_by_id($post_id);
	406	+ }, 20);
188	407
189		- // Taxonomy change hooks
190		- add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3);
191		- add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3);
192		- add_action('delete_term', [$this, 'handle_taxonomy_change'], 20, 3);
	408	+ foreach (['created_term', 'edited_term', 'delete_term'] as $hook) {
	409	+ add_action($hook, static function (int $term_id, int $tt_id, string $taxonomy): void {
	410	+ self::listener()->handle_taxonomy_change($term_id, $tt_id, $taxonomy);
	411	+ }, 20, 3);
	412	+ }
193	413
194		- // NOTE: the WP-Cron regeneration listeners (thinkrank_regenerate_sitemap
195		- // and thinkrank_regenerate_sitemap_settings) are registered at plugin
196		- // bootstrap (Plugin::register_sitemap_cron_listeners(), on plugins_loaded)
197		- // rather than here. A cron run never builds this class via the REST
198		- // endpoint (no rest_api_init), so registering them in the constructor
199		- // would leave the scheduled events with no listener at cron time.
	414	+ // The sitemap leaves out what the robots tag noindexes and what
	415	+ // ThinkRank redirects (#911), so a change to either decides what the
	416	+ // published files should list. Neither is a post or term save, and
	417	+ // without these the files kept the old answer until unrelated content
	418	+ // changed.
	419	+ foreach (self::ROBOTS_SETTINGS_OPTIONS as $option) {
	420	+ add_action('update_option_' . $option, static function ($old_value, $value): void {
	421	+ self::handle_robots_settings_change($old_value, $value);
	422	+ }, 20, 2);
	423	+ add_action('add_option_' . $option, static function ($name, $value): void {
	424	+ self::handle_robots_settings_change([], $value);
	425	+ }, 20, 2);
	426	+ }
	427	+
	428	+ add_action('thinkrank_object_redirect_saved', static function (): void {
	429	+ self::listener()->schedule_regeneration();
	430	+ }, 20, 0);
200	431	}
201	432
202	433	/**
	434	+ * Options holding robots directives the sitemap's item filter reads.
	435	+ *
	436	+ * @since 2.15.0
	437	+ * @var string[]
	438	+ */
	439	+ private const ROBOTS_SETTINGS_OPTIONS = [
	440	+ 'thinkrank_global_seo_settings',
	441	+ 'thinkrank_global_robot_meta_settings',
	442	+ ];
	443	+
	444	+ /**
	445	+ * Queue a rebuild when a saved robots setting changes a noindex decision.
	446	+ *
	447	+ * Both options carry far more than robots directives (titles, schema,
	448	+ * feature switches), so only a change to a noindex outcome rebuilds.
	449	+ *
	450	+ * @since 2.15.0
	451	+ *
	452	+ * @param mixed $old_value Previous option value.
	453	+ * @param mixed $value New option value.
	454	+ * @return void
	455	+ */
	456	+ public static function handle_robots_settings_change($old_value, $value): void {
	457	+ if (self::noindex_fingerprint($old_value) === self::noindex_fingerprint($value)) {
	458	+ return;
	459	+ }
	460	+
	461	+ // The matrix memoises the option for the request, and a rebuild that
	462	+ // runs in this request must read the value just saved.
	463	+ Content_Type_Settings::flush_cache();
	464	+
	465	+ self::listener()->schedule_regeneration();
	466	+ }
	467	+
	468	+ /**
	469	+ * The noindex decisions a robots option makes, keyed by what they apply to.
	470	+ *
	471	+ * Reads both shapes: the flat site-wide directives, and the per-entity
	472	+ * rows, where a row's directives apply only while its robots switch is on.
	473	+ * A row that is on but stores no `noindex` key changes nothing, matching
	474	+ * the array_merge() the robots tag does.
	475	+ *
	476	+ * @since 2.15.0
	477	+ *
	478	+ * @param mixed $value Option value.
	479	+ * @return array<string, bool|null>
	480	+ */
	481	+ private static function noindex_fingerprint($value): array {
	482	+ if (!is_array($value)) {
	483	+ return [];
	484	+ }
	485	+
	486	+ $decisions = [];
	487	+
	488	+ if (array_key_exists('noindex', $value) && !is_array($value['noindex'])) {
	489	+ $decisions['*'] = !empty($value['noindex']);
	490	+ }
	491	+
	492	+ foreach ($value as $key => $row) {
	493	+ if (!is_array($row)) {
	494	+ continue;
	495	+ }
	496	+
	497	+ $robots = $row['robots_meta'] ?? null;
	498	+
	499	+ $decisions[(string) $key] = !empty($row['robots_meta_enabled']) && is_array($robots) && array_key_exists('noindex', $robots)
	500	+ ? !empty($robots['noindex'])
	501	+ : null;
	502	+ }
	503	+
	504	+ ksort($decisions);
	505	+
	506	+ return $decisions;
	507	+ }
	508	+
	509	+ /**
	510	+ * The generator the content-change listeners share.
	511	+ *
	512	+ * @since 2.10.1
	513	+ * @return self
	514	+ */
	515	+ private static function listener(): self {
	516	+ return self::$listener ??= new self(false);
	517	+ }
	518	+
	519	+ /**
203	520	* Generate XML sitemap
204	521	*
205	522	* @since 1.0.0
206	523	*
		@@ -209,15 +526,10 @@
209	526	*/
210	527	public function generate_sitemap(array $options = []): string {
211	528	$settings = $this->get_settings('site');
212	529
213		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	530	+ $xml = $this->xml_prolog($settings, 'sitemap');
214	531
215		- // Add XSL stylesheet only if styling is enabled
216		- if (!empty($settings['enable_styling'])) {
217		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
218		- }
219		-
220	532	// Add image namespace if images are enabled
221	533	if (!empty($settings['include_images'])) {
222	534	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
223	535	} else {
		@@ -379,8 +691,15 @@
379	691	]
380	692	],
381	693	'use_sitemap_index' => false,
382	694
	695	+ // How the sitemap reaches crawlers. 'auto' keeps the historical
	696	+ // behaviour wherever the web root is writable, and only falls back
	697	+ // to serving the sitemap from PHP where writing a file is
	698	+ // impossible — previously a hard failure with nothing served
	699	+ // (#752).
	700	+ 'delivery_mode' => 'auto',
	701	+
383	702	// General Settings
384	703	'links_per_sitemap' => 1000,
385	704	'include_images' => true,
386	705	'include_featured_images' => false,
		@@ -402,8 +721,17 @@
402	721	// Advanced Options
403	722	'enable_styling' => true,
404	723	'custom_url_pattern' => 'sitemap-{type}.xml',
405	724
	725	+ // Stylesheet branding (#639). Both colours default to empty, not
	726	+ // to the stock hexes: empty means the stylesheet's own value
	727	+ // stands, so a site that never opens this screen renders exactly
	728	+ // as it did before the setting existed.
	729	+ 'styling_logo' => false,
	730	+ 'styling_logo_url' => '',
	731	+ 'styling_color_main' => '',
	732	+ 'styling_color_accent' => '',
	733	+
406	734	// Generation tracking
407	735	'last_generated' => ''
408	736	];
409	737	}
		@@ -408,8 +736,49 @@
408	736	];
409	737	}
410	738
411	739	/**
	740	+ * Normalize the stylesheet branding values on the way into the store.
	741	+ *
	742	+ * The generic sanitizer only runs `sanitize_text_field()` over a string,
	743	+ * which happily keeps "red" or "rebeccapurple" as a colour. Nothing
	744	+ * downstream can use those — {@see Sitemap_Stylesheet::render()} skips any
	745	+ * value it cannot read as a hex colour — so storing them would report a
	746	+ * successful save of a setting that changes nothing, and `get-sitemap-
	747	+ * settings` would hand an agent back a colour the sitemap does not use.
	748	+ * Reducing here instead keeps the store and the rendering in agreement.
	749	+ *
	750	+ * @since 2.7.0
	751	+ *
	752	+ * @param array $settings Settings to sanitize.
	753	+ * @param string $context_type Context the save is for.
	754	+ * @return array
	755	+ */
	756	+ protected function sanitize_settings(array $settings, string $context_type = 'site'): array {
	757	+ $sanitized = parent::sanitize_settings($settings, $context_type);
	758	+
	759	+ foreach (['styling_color_main', 'styling_color_accent'] as $key) {
	760	+ if (array_key_exists($key, $sanitized)) {
	761	+ $sanitized[$key] = Sitemap_Stylesheet::hex($sanitized[$key]);
	762	+ }
	763	+ }
	764	+
	765	+ if (array_key_exists('styling_logo_url', $sanitized)) {
	766	+ $sanitized['styling_logo_url'] = esc_url_raw((string) $sanitized['styling_logo_url']);
	767	+ }
	768	+
	769	+ // A mode this build cannot act on has to be stored as the fallback
	770	+ // rather than kept verbatim, or get-sitemap-settings reports a delivery
	771	+ // mode the site does not actually apply.
	772	+ if (array_key_exists('delivery_mode', $sanitized)) {
	773	+ $mode = sanitize_key((string) $sanitized['delivery_mode']);
	774	+ $sanitized['delivery_mode'] = in_array($mode, self::DELIVERY_MODES, true) ? $mode : 'auto';
	775	+ }
	776	+
	777	+ return $sanitized;
	778	+ }
	779	+
	780	+ /**
412	781	* Get settings schema definition (implements interface)
413	782	*
414	783	* @since 1.0.0
415	784	*
		@@ -423,8 +792,15 @@
423	792	'title' => 'Enable Sitemap',
424	793	'description' => 'Generate XML sitemap for search engines',
425	794	'default' => true
426	795	],
	796	+ 'delivery_mode' => [
	797	+ 'type' => 'string',
	798	+ 'title' => 'Sitemap Delivery',
	799	+ 'description' => 'How the sitemap is served: auto picks static when the WordPress root is writable and dynamic when it is not, static writes files to the web root, dynamic serves the sitemap from WordPress with no files written',
	800	+ 'enum' => self::DELIVERY_MODES,
	801	+ 'default' => 'auto'
	802	+ ],
427	803	'include_posts' => [
428	804	'type' => 'boolean',
429	805	'title' => 'Include Posts',
430	806	'description' => 'Include blog posts in sitemap',
		@@ -465,8 +841,32 @@
465	841	'type' => 'string',
466	842	'title' => 'Last Generated',
467	843	'description' => 'Timestamp of last sitemap generation',
468	844	'default' => ''
	845	+ ],
	846	+ 'styling_logo' => [
	847	+ 'type' => 'boolean',
	848	+ 'title' => 'Show Logo On Sitemap',
	849	+ 'description' => 'Show a logo above the sitemap heading',
	850	+ 'default' => false
	851	+ ],
	852	+ 'styling_logo_url' => [
	853	+ 'type' => 'string',
	854	+ 'title' => 'Sitemap Logo',
	855	+ 'description' => 'Logo image URL. Empty falls back to the site icon',
	856	+ 'default' => ''
	857	+ ],
	858	+ 'styling_color_main' => [
	859	+ 'type' => 'string',
	860	+ 'title' => 'Sitemap Main Color',
	861	+ 'description' => 'Hex color for the sitemap header, links and table head. Empty keeps the stock palette',
	862	+ 'default' => ''
	863	+ ],
	864	+ 'styling_color_accent' => [
	865	+ 'type' => 'string',
	866	+ 'title' => 'Sitemap Accent Color',
	867	+ 'description' => 'Hex color for the header gradient and link hovers. Empty keeps the stock palette',
	868	+ 'default' => ''
469	869	]
470	870	];
471	871	}
472	872
		@@ -483,9 +883,14 @@
483	883	* @return string XML URL entry
484	884	*/
485	885	private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string {
486	886	$xml = " <url>\n";
487		- $xml .= " <loc>" . esc_url($url) . "</loc>\n";
	887	+ // Every <loc> in every sitemap passes through here, which is why the
	888	+ // scheme preference is applied at this one point rather than at each
	889	+ // of the dozen collectors that build URLs (#638). An http sitemap on
	890	+ // an https site hands search engines the wrong address for the whole
	891	+ // site at once.
	892	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($url)) . "</loc>\n";
488	893	// Omit <lastmod> when unknown (empty) — a fabricated timestamp is worse
489	894	// than no timestamp, and an absent lastmod is valid per the spec.
490	895	if (!empty($lastmod)) {
491	896	$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n";
		@@ -495,9 +900,9 @@
495	900
496	901	// Add image entries if provided
497	902	foreach ($images as $image) {
498	903	$xml .= " <image:image>\n";
499		- $xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n";
	904	+ $xml .= " <image:loc>" . esc_url(Url_Scheme::apply((string) $image['url'])) . "</image:loc>\n";
500	905
501	906	if (!empty($image['title'])) {
502	907	$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n";
503	908	}
		@@ -618,8 +1023,11 @@
618	1023	// well-bounded routine with no ceiling (#402).
619	1024	$total = count($all_ids);
620	1025
621	1026	for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) {
	1027	+ $this->assert_memory_headroom();
	1028	+ $chunk_start = memory_get_usage(true);
	1029	+
622	1030	$chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK);
623	1031
624	1032	$posts = get_posts($this->filter_query_args([
625	1033	'post_type' => $post_types,
		@@ -628,8 +1036,12 @@
628	1036	'post__in' => $chunk,
629	1037	'orderby' => 'post__in', // preserve the resolved order
630	1038	]));
631	1039
	1040	+ // get_featured_image() hydrates each featured image under the
	1041	+ // attachment's own ID, which the chunk's post IDs do not reach.
	1042	+ $attachment_ids = [];
	1043	+
632	1044	foreach ($posts as $post) {
633	1045	if ($this->should_include_in_sitemap($post, $settings)) {
634	1046	/**
635	1047	* Filter a sitemap entry's permalink.
		@@ -650,14 +1062,25 @@
650	1062	$priority = $this->calculate_intelligent_priority($post, $post->post_type);
651	1063	$changefreq = $this->calculate_change_frequency($post, $post->post_type);
652	1064	$images = $this->extract_post_images($post, $settings);
653	1065
	1066	+ $thumbnail_id = (int) get_post_thumbnail_id($post);
	1067	+ if ($thumbnail_id > 0) {
	1068	+ $attachment_ids[] = $thumbnail_id;
	1069	+ }
	1070	+
654	1071	yield $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images);
655	1072	}
656	1073	}
657	1074
658		- // Free the hydrated chunk before loading the next one.
	1075	+ // Free the hydrated chunk before loading the next one — including
	1076	+ // the copies get_posts() left in the runtime object cache.
659	1077	unset($posts);
	1078	+ $this->release_walk_memory(
	1079	+ $chunk_start,
	1080	+ $this->chunk_post_cache_groups($post_types),
	1081	+ array_merge($chunk, $attachment_ids)
	1082	+ );
660	1083	}
661	1084	}
662	1085
663	1086	/**
		@@ -717,8 +1140,11 @@
717	1140	// Same moving window as the post walk above, for the same reason.
718	1141	$total = count($all_ids);
719	1142
720	1143	for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) {
	1144	+ $this->assert_memory_headroom();
	1145	+ $chunk_start = memory_get_usage(true);
	1146	+
721	1147	$chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK);
722	1148
723	1149	$terms = get_terms($this->filter_term_query_args([
724	1150	'taxonomy' => $taxonomy,
		@@ -731,12 +1157,12 @@
731	1157	continue;
732	1158	}
733	1159
734	1160	foreach ($terms as $term) {
735		- // A term the user marked noindex must not be advertised in the
736		- // sitemap: the robots tag now honours term meta, so listing it
737		- // here would have the sitemap contradict the page's own tag.
738		- if ($this->term_is_noindexed((int) $term->term_id)) {
	1161	+ // A term whose archive says noindex (its own override or its
	1162	+ // taxonomy's), or that redirects, must not be advertised: the
	1163	+ // sitemap would contradict the page's own signal (#911).
	1164	+ if (!Indexability::is_indexable_term($term)) {
739	1165	continue;
740	1166	}
741	1167
742	1168	$url = get_term_link($term);
		@@ -747,12 +1173,48 @@
747	1173	}
748	1174	}
749	1175
750	1176	unset($terms);
	1177	+ $this->release_walk_memory($chunk_start, ['terms', 'term_meta'], $chunk);
751	1178	}
752	1179	}
753	1180
754	1181	/**
	1182	+ * The XML declaration, ownership marker and optional stylesheet every
	1183	+ * sitemap document opens with.
	1184	+ *
	1185	+ * The marker is written unconditionally, and that is the point: removal on
	1186	+ * deactivate and uninstall deletes a web-root sitemap only when the file
	1187	+ * says it is ours, and our filenames are the canonical ones another SEO
	1188	+ * plugin writes too (#515). Tying the proof to `enable_styling` — the one
	1189	+ * marker older versions left — would mean a site with styling off either
	1190	+ * kept a shadowing file behind (#510) or had a competitor's deleted.
	1191	+ *
	1192	+ * @since 2.1.1
	1193	+ *
	1194	+ * The stylesheet URL is served by {@see Sitemap_Stylesheet}, not read off
	1195	+ * disk by the web server, because a static file cannot carry the site's own
	1196	+ * logo and colours (#639). It is a fixed URL: the palette is applied per
	1197	+ * request, so changing a brand colour needs no regeneration and shows up on
	1198	+ * sitemaps published long before.
	1199	+ *
	1200	+ * @param array $settings Sitemap settings (read for `enable_styling`).
	1201	+ * @param string $variant Stylesheet variant, `sitemap` or `index`.
	1202	+ * @return string Prolog lines, newline-terminated.
	1203	+ */
	1204	+ private function xml_prolog(array $settings, string $variant): string {
	1205	+ $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1206	+ $xml .= THINKRANK_SITEMAP_MARKER . "\n";
	1207	+
	1208	+ // The stylesheet is presentation only, so it stays opt-in.
	1209	+ if (!empty($settings['enable_styling'])) {
	1210	+ $xml .= '<?xml-stylesheet type="text/xsl" href="' . esc_url(Sitemap_Stylesheet::url($variant)) . '"?>' . "\n";
	1211	+ }
	1212	+
	1213	+ return $xml;
	1214	+ }
	1215	+
	1216	+ /**
755	1217	* Wrap a set of <url> entry strings in a complete <urlset> document.
756	1218	*
757	1219	* @since 1.14.0
758	1220	*
		@@ -761,14 +1223,10 @@
761	1223	* @param bool $with_image_ns Include the image sitemap namespace
762	1224	* @return string Full sitemap XML
763	1225	*/
764	1226	private function wrap_urlset(array $entries, array $settings, bool $with_image_ns): string {
765		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1227	+ $xml = $this->xml_prolog($settings, 'sitemap');
766	1228
767		- if (!empty($settings['enable_styling'])) {
768		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
769		- }
770		-
771	1229	if ($with_image_ns && !empty($settings['include_images'])) {
772	1230	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
773	1231	} else {
774	1232	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
		@@ -945,9 +1403,16 @@
945	1403	'_builtin' => false
946	1404	], 'names');
947	1405
948	1406	foreach ($custom_taxonomies as $taxonomy) {
949		- if ($this->should_include_taxonomy($taxonomy)) {
	1407	+ if (!$this->should_include_taxonomy($taxonomy)) {
	1408	+ continue;
	1409	+ }
	1410	+
	1411	+ // Same as the post-type walk above: an explicit per-taxonomy flag
	1412	+ // now decides, and an unset flag keeps the previous "included"
	1413	+ // behaviour (#660).
	1414	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $settings)) {
950	1415	$taxonomies[] = $taxonomy;
951	1416	}
952	1417	}
953	1418
		@@ -1002,14 +1467,15 @@
1002	1467	if (is_wp_error($all_terms)) {
1003	1468	return [];
1004	1469	}
1005	1470
1006		- // Drop terms the user marked noindex. This path feeds the single general
1007		- // sitemap while collect_taxonomy_entries_iter() feeds the segmented ones,
1008		- // so both need the filter or the two disagree about the same term.
	1471	+ // Drop terms that are not indexable destinations. This path feeds the
	1472	+ // single general sitemap while collect_taxonomy_entries_iter() feeds
	1473	+ // the segmented ones, so both need the filter or the two disagree about
	1474	+ // the same term.
1009	1475	$all_terms = array_values(array_filter(
1010	1476	$all_terms,
1011		- fn($term) => !$this->term_is_noindexed((int) $term->term_id)
	1477	+ static fn($term) => $term instanceof \WP_Term && Indexability::is_indexable_term($term)
1012	1478	));
1013	1479
1014	1480	// Group terms by taxonomy
1015	1481	return $this->group_terms_by_taxonomy($all_terms);
		@@ -1191,17 +1657,16 @@
1191	1657	if (!in_array($post->post_status, ['publish', 'private'], true)) {
1192	1658	return false;
1193	1659	}
1194	1660
1195		- // Check if post overrides robots and sets noindex.
1196		- if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) {
1197		- $raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true);
1198		- if (is_string($raw) && $raw !== '') {
1199		- $robots = json_decode($raw, true);
1200		- if (is_array($robots) && !empty($robots['noindex'])) {
1201		- return false;
1202		- }
1203		- }
	1661	+ // A URL whose page says noindex, or that ThinkRank redirects, is not a
	1662	+ // destination. Only the per-post noindex used to be read here, so a
	1663	+ // post type set to No-index still had every item listed, and redirected
	1664	+ // posts were submitted as "Page with redirect" (#911). The password
	1665	+ // check stays with the setting above: listing protected posts is a
	1666	+ // choice this sitemap has always offered.
	1667	+ if (Indexability::is_post_noindexed($post) || Indexability::is_post_redirected($post)) {
	1668	+ return false;
1204	1669	}
1205	1670
1206	1671	return true;
1207	1672	}
		@@ -1240,35 +1705,8 @@
1240	1705	return $ids;
1241	1706	}
1242	1707
1243	1708	/**
1244		- * Whether a term carries an explicit noindex override.
1245		- *
1246		- * Mirrors the post-side check in should_include_post(); terms store the same
1247		- * `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written
1248		- * by the update-term-seo ability and by the SEO importer.
1249		- *
1250		- * @since 1.31.0
1251		- *
1252		- * @param int $term_id Term to test.
1253		- * @return bool True when the term is marked noindex.
1254		- */
1255		- private function term_is_noindexed(int $term_id): bool {
1256		- if (!(bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true)) {
1257		- return false;
1258		- }
1259		-
1260		- $raw = get_term_meta($term_id, '_thinkrank_robots_meta', true);
1261		- if (!is_string($raw) || $raw === '') {
1262		- return false;
1263		- }
1264		-
1265		- $robots = json_decode($raw, true);
1266		-
1267		- return is_array($robots) && !empty($robots['noindex']);
1268		- }
1269		-
1270		- /**
1271	1709	* Count total URLs in sitemap
1272	1710	*
1273	1711	* @since 1.0.0
1274	1712	*
		@@ -1431,9 +1869,15 @@
1431	1869	if (!empty($settings['include_pages'])) {
1432	1870	$post_types[] = 'page';
1433	1871	}
1434	1872
1435		- // Auto-detect public custom post types that should be included
	1873	+ // Auto-detect public custom post types that should be included.
	1874	+ //
	1875	+ // A custom type's `include_<slug>` / `exclude_<slug>` flag is honoured
	1876	+ // here (#660). It was previously stored — additional_setting_keys()
	1877	+ // has always let those keys through — but never read, so a CPT was in
	1878	+ // the sitemap whatever the setting said. Unset still means included, so
	1879	+ // a site that never touched the flag is unaffected.
1436	1880	$custom_post_types = get_post_types([
1437	1881	'public' => true,
1438	1882	'_builtin' => false
1439	1883	], 'names');
		@@ -1438,9 +1882,13 @@
1438	1882	'_builtin' => false
1439	1883	], 'names');
1440	1884
1441	1885	foreach ($custom_post_types as $post_type) {
1442		- if ($this->should_include_post_type($post_type)) {
	1886	+ if (!$this->should_include_post_type($post_type)) {
	1887	+ continue;
	1888	+ }
	1889	+
	1890	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $post_type, $settings)) {
1443	1891	$post_types[] = $post_type;
1444	1892	}
1445	1893	}
1446	1894
		@@ -1461,18 +1909,11 @@
1461	1909	// Templately's internal `templately_library` store — which are template
1462	1910	// records, not standalone indexable URLs. BetterDocs `docs` and
1463	1911	// WooCommerce `product` register exclude_from_search => false, so they
1464	1912	// remain included.
1465		- if (!is_post_type_viewable($post_type)) {
1466		- return false;
1467		- }
1468		-
1469		- $post_type_obj = get_post_type_object($post_type);
1470		- if (!$post_type_obj || !empty($post_type_obj->exclude_from_search)) {
1471		- return false;
1472		- }
1473		-
1474		- return true;
	1913	+ // The predicate lives in Content_Type_Settings so the matrix can ask
	1914	+ // the same question before offering a switch for this post type.
	1915	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_post_type($post_type);
1475	1916	}
1476	1917
1477	1918	/**
1478	1919	* Check if taxonomy should trigger regeneration
		@@ -1481,11 +1922,13 @@
1481	1922	* @param string $taxonomy Taxonomy slug
1482	1923	* @return bool True if taxonomy should trigger regeneration
1483	1924	*/
1484	1925	private function should_include_taxonomy(string $taxonomy): bool {
1485		- // Include all public taxonomies (presets control which sitemaps are created)
1486		- $taxonomy_obj = get_taxonomy($taxonomy);
1487		- return $taxonomy_obj && $taxonomy_obj->public;
	1926	+ // Public taxonomies only, and only the ones this generator can actually
	1927	+ // emit — the same predicate the content-type matrix asks before it
	1928	+ // offers a sitemap switch for one (presets still control which
	1929	+ // sitemaps are created).
	1930	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_taxonomy($taxonomy);
1488	1931	}
1489	1932
1490	1933	/**
1491	1934	* Schedule debounced sitemap regeneration
		@@ -1493,16 +1936,679 @@
1493	1936	* @since 1.0.0
1494	1937	* @return void
1495	1938	*/
1496	1939	private function schedule_debounced_regeneration(): void {
1497		- // Clear any existing scheduled regeneration
1498		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap');
	1940	+ $this->mark_regeneration_pending('content');
	1941	+ $this->debounce_event('thinkrank_regenerate_sitemap');
	1942	+ }
1499	1943
1500		- // Schedule regeneration in 30 seconds to debounce rapid changes
1501		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap');
	1944	+ /**
	1945	+ * Schedule (or keep) the debounced single event behind a regeneration hook.
	1946	+ *
	1947	+ * An event that is already due is left alone. WP-Cron only runs when a
	1948	+ * request arrives, so on a site with DISABLE_WP_CRON, a blocked loopback or
	1949	+ * little traffic an overdue event can sit in the queue for a long time —
	1950	+ * clearing and re-scheduling it on every save pushed the rebuild
	1951	+ * permanently 30 seconds into the future and the sitemap never updated
	1952	+ * (#629). Debouncing only against an event that has not come due yet keeps
	1953	+ * the bulk-edit coalescing without starving the rebuild.
	1954	+ *
	1955	+ * @since 2.2.1
	1956	+ * @param string $hook Regeneration hook to debounce.
	1957	+ * @return void
	1958	+ */
	1959	+ private function debounce_event(string $hook): void {
	1960	+ $next = wp_next_scheduled($hook);
	1961	+
	1962	+ if ($next !== false) {
	1963	+ if ($next <= time()) {
	1964	+ return;
	1965	+ }
	1966	+
	1967	+ wp_clear_scheduled_hook($hook);
	1968	+ }
	1969	+
	1970	+ wp_schedule_single_event(time() + self::REGENERATION_DEBOUNCE, $hook);
1502	1971	}
1503	1972
1504	1973	/**
	1974	+ * Record that a rebuild is outstanding, so an overdue one can be taken over
	1975	+ * by a later request and its staleness surfaced in the UI.
	1976	+ *
	1977	+ * `since` is the *oldest* outstanding change: it is what the takeover grace
	1978	+ * and the admin staleness warning are measured from, so successive edits
	1979	+ * must not push it forward. A settings change outranks a content change —
	1980	+ * it rebuilds regardless of the auto_generate toggle and handles a sitemap
	1981	+ * that has just been disabled — so once one is outstanding it stays the
	1982	+ * recorded source until the rebuild lands.
	1983	+ *
	1984	+ * @since 2.2.1
	1985	+ * @param string $source Either 'content' or 'settings'.
	1986	+ * @return void
	1987	+ */
	1988	+ private function mark_regeneration_pending(string $source): void {
	1989	+ // Whatever made the static files stale made the rendered ones stale
	1990	+ // too. Invalidating here rather than only on the rebuild keeps the two
	1991	+ // delivery modes reacting to exactly the same triggers, which is the
	1992	+ // only way a dynamic site stays as fresh as a static one (#752).
	1993	+ $this->flush_dynamic_cache();
	1994	+
	1995	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1996	+ $pending = is_array($pending) ? $pending : [];
	1997	+
	1998	+ $since = !empty($pending['since']) ? (int) $pending['since'] : time();
	1999	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	2000	+ $source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	2001	+
	2002	+ // Merged so the bookkeeping a rebuild keeps on the marker (`started`,
	2003	+ // `memory_limit`) survives an edit made while it is outstanding.
	2004	+ update_option(
	2005	+ self::REGENERATION_PENDING_OPTION,
	2006	+ array_merge($pending, [
	2007	+ 'since' => $since,
	2008	+ 'source' => $source,
	2009	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 0,
	2010	+ 'next_attempt' => !empty($pending['next_attempt'])
	2011	+ ? (int) $pending['next_attempt']
	2012	+ : time() + self::REGENERATION_TAKEOVER_GRACE,
	2013	+ // Bumped on every change so a rebuild can tell whether the edit
	2014	+ // it started for is still the newest one outstanding.
	2015	+ 'revision' => (!empty($pending['revision']) ? (int) $pending['revision'] : 0) + 1,
	2016	+ ]),
	2017	+ true
	2018	+ );
	2019	+ }
	2020	+
	2021	+ /**
	2022	+ * The revision of the outstanding rebuild, for
	2023	+ * {@see mark_regeneration_complete()} to compare against once it is done.
	2024	+ *
	2025	+ * @since 2.2.1
	2026	+ * @return int Current revision, 0 when nothing is outstanding.
	2027	+ */
	2028	+ private function current_regeneration_revision(): int {
	2029	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2030	+
	2031	+ return (is_array($pending) && !empty($pending['revision'])) ? (int) $pending['revision'] : 0;
	2032	+ }
	2033	+
	2034	+ /**
	2035	+ * Clear the outstanding-rebuild marker and any recorded failure.
	2036	+ *
	2037	+ * Public because a manual generation satisfies whatever the automatic path
	2038	+ * was still waiting to write.
	2039	+ *
	2040	+ * @since 2.2.1
	2041	+ * @return void
	2042	+ */
	2043	+ public function mark_regeneration_complete(?int $revision = null): void {
	2044	+ // The write succeeded, so whatever failure was on record is history.
	2045	+ if (get_option(self::REGENERATION_ERROR_OPTION, null) !== null) {
	2046	+ delete_option(self::REGENERATION_ERROR_OPTION);
	2047	+ }
	2048	+
	2049	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	2050	+
	2051	+ if ($pending === null) {
	2052	+ return;
	2053	+ }
	2054	+
	2055	+ // A change that landed while this rebuild was running is not covered by
	2056	+ // the files it just wrote, so it has to stay outstanding — otherwise, on
	2057	+ // a site where WP-Cron never fires, clearing the marker would strand it
	2058	+ // exactly the way #629 stranded everything.
	2059	+ if (
	2060	+ $revision !== null
	2061	+ && is_array($pending)
	2062	+ && (int) ($pending['revision'] ?? 0) !== $revision
	2063	+ ) {
	2064	+ // This attempt succeeded, so drop what it claimed: the newer change
	2065	+ // waits the grace a fresh edit gets, not a failure backoff it never
	2066	+ // earned. Not zero: that edit queued its own debounced event, and
	2067	+ // a marker due at once had the next admin request rebuild in its
	2068	+ // shutdown and the event rebuild again seconds later.
	2069	+ unset($pending['started'], $pending['memory_limit']);
	2070	+ $pending['attempts'] = 0;
	2071	+ $pending['next_attempt'] = time() + self::REGENERATION_TAKEOVER_GRACE;
	2072	+
	2073	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
	2074	+ return;
	2075	+ }
	2076	+
	2077	+ delete_option(self::REGENERATION_PENDING_OPTION);
	2078	+ }
	2079	+
	2080	+ /**
	2081	+ * Record the attempt that is about to run before it runs.
	2082	+ *
	2083	+ * A PHP fatal — the memory limit or max_execution_time — is not a
	2084	+ * Throwable, so no catch or finally around the generation runs when the
	2085	+ * process dies, and a failure recorded afterwards was never recorded at
	2086	+ * all: `attempts` stayed 0, the backoff never applied, and the next request
	2087	+ * started the same doomed rebuild again (a fatal every few minutes for as
	2088	+ * long as an admin was logged in). Claiming the attempt up front makes the
	2089	+ * backoff hold even when nothing after this line gets to run, and leaves a
	2090	+ * `started` stamp the next attempt can recognise as an interrupted one.
	2091	+ *
	2092	+ * @since 2.10.1
	2093	+ * @return void
	2094	+ */
	2095	+ private function claim_regeneration_attempt(): void {
	2096	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	2097	+
	2098	+ // Nothing outstanding (e.g. a manual generation already satisfied it):
	2099	+ // there is no marker to retry from, so nothing to claim.
	2100	+ if (!is_array($pending) || empty($pending['since'])) {
	2101	+ return;
	2102	+ }
	2103	+
	2104	+ if (!empty($pending['started'])) {
	2105	+ // The previous attempt claimed itself and never reported back.
	2106	+ update_option(
	2107	+ self::REGENERATION_ERROR_OPTION,
	2108	+ [
	2109	+ 'message' => __('The previous automatic sitemap rebuild stopped before it finished, most likely because PHP ran out of memory or time. It is retried with a growing delay. If this keeps happening, raise the PHP memory_limit or max_execution_time, or run WP-Cron from a system cron.', 'thinkrank'),
	2110	+ 'source' => isset($pending['source']) ? (string) $pending['source'] : 'content',
	2111	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 1,
	2112	+ 'time' => (int) $pending['started'],
	2113	+ ],
	2114	+ false
	2115	+ );
	2116	+ }
	2117	+
	2118	+ $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1;
	2119	+
	2120	+ $pending['attempts'] = $attempts;
	2121	+ $pending['next_attempt'] = time() + $this->regeneration_backoff($attempts);
	2122	+ $pending['started'] = time();
	2123	+
	2124	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
	2125	+ }
	2126	+
	2127	+ /**
	2128	+ * Delay before the next takeover after the given number of attempts.
	2129	+ *
	2130	+ * @since 2.10.1
	2131	+ * @param int $attempts Attempts made so far (1 or more).
	2132	+ * @return int Seconds.
	2133	+ */
	2134	+ private function regeneration_backoff(int $attempts): int {
	2135	+ return (int) min(
	2136	+ self::REGENERATION_TAKEOVER_GRACE * (2 ** min(max($attempts, 1), 10)),
	2137	+ self::REGENERATION_MAX_BACKOFF
	2138	+ );
	2139	+ }
	2140	+
	2141	+ /**
	2142	+ * Record a failed regeneration instead of discarding it.
	2143	+ *
	2144	+ * Keeps the pending marker in place so the rebuild is retried, but backs the
	2145	+ * next attempt off exponentially (capped) so a persistently failing
	2146	+ * generation cannot run on every admin request.
	2147	+ *
	2148	+ * @since 2.2.1
	2149	+ * @since 2.10.1 Accepts the memory limit a rebuild had to stop short of, and
	2150	+ * does not count an attempt claim_regeneration_attempt()
	2151	+ * already counted.
	2152	+ * @param string $message Failure detail.
	2153	+ * @param string $source Either 'content' or 'settings'.
	2154	+ * @param int|null $memory_limit Memory limit (bytes) the rebuild stopped
	2155	+ * short of, when that was the failure.
	2156	+ * @return void
	2157	+ */
	2158	+ private function record_regeneration_failure(string $message, string $source, ?int $memory_limit = null): void {
	2159	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2160	+ $pending = is_array($pending) ? $pending : [];
	2161	+ $attempts = !empty($pending['attempts']) ? (int) $pending['attempts'] : 0;
	2162	+
	2163	+ // An attempt that claimed itself up front has already been counted.
	2164	+ if (empty($pending['started'])) {
	2165	+ $attempts++;
	2166	+ }
	2167	+ $attempts = max($attempts, 1);
	2168	+
	2169	+ $backoff = $this->regeneration_backoff($attempts);
	2170	+
	2171	+ // Same precedence mark_regeneration_pending() enforces: a settings
	2172	+ // rebuild outranks a content one and must not be downgraded by a failed
	2173	+ // attempt. Overwriting it routed the retry back through the content
	2174	+ // path, where should_auto_generate() can be false and the completion
	2175	+ // marker then discards the settings rebuild entirely. Only the
	2176	+ // outstanding rebuild is upgraded — the recorded error keeps reporting
	2177	+ // whichever attempt actually failed.
	2178	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	2179	+ $pending_source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	2180	+
	2181	+ $marker = [
	2182	+ 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(),
	2183	+ 'source' => $pending_source,
	2184	+ 'attempts' => $attempts,
	2185	+ 'next_attempt' => time() + $backoff,
	2186	+ 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0,
	2187	+ ];
	2188	+
	2189	+ // Remembered so has_memory_for_retry() can keep requests with no more
	2190	+ // memory than this from repeating the same attempt.
	2191	+ if ($memory_limit !== null && $memory_limit > 0) {
	2192	+ $marker['memory_limit'] = $memory_limit;
	2193	+ }
	2194	+
	2195	+ update_option(self::REGENERATION_PENDING_OPTION, $marker, true);
	2196	+
	2197	+ update_option(
	2198	+ self::REGENERATION_ERROR_OPTION,
	2199	+ [
	2200	+ 'message' => $message,
	2201	+ 'source' => $source,
	2202	+ 'attempts' => $attempts,
	2203	+ 'time' => time(),
	2204	+ ],
	2205	+ false
	2206	+ );
	2207	+
	2208	+ if (defined('WP_DEBUG') && WP_DEBUG) {
	2209	+ // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
	2210	+ error_log(sprintf('ThinkRank: sitemap %s regeneration failed — %s', $source, $message));
	2211	+ }
	2212	+ }
	2213	+
	2214	+ /**
	2215	+ * Is a rebuild outstanding and past the point where WP-Cron should have run
	2216	+ * it?
	2217	+ *
	2218	+ * Deliberately cheap — one autoloaded option read — because it is consulted
	2219	+ * on every admin request to decide whether the takeover is needed.
	2220	+ *
	2221	+ * @since 2.2.1
	2222	+ * @return bool True when a request should rebuild the sitemap itself.
	2223	+ */
	2224	+ public static function has_overdue_regeneration(): bool {
	2225	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2226	+
	2227	+ if (!is_array($pending) || empty($pending['since'])) {
	2228	+ return false;
	2229	+ }
	2230	+
	2231	+ $due = !empty($pending['next_attempt'])
	2232	+ ? (int) $pending['next_attempt']
	2233	+ : (int) $pending['since'] + self::REGENERATION_TAKEOVER_GRACE;
	2234	+
	2235	+ return time() >= $due;
	2236	+ }
	2237	+
	2238	+ /**
	2239	+ * Rebuild the sitemap in-request when WP-Cron has not delivered.
	2240	+ *
	2241	+ * Hooked on `shutdown` for admin, REST and CLI requests only (see
	2242	+ * Plugin::register_sitemap_cron_listeners()), so the work happens after the
	2243	+ * response has been sent and never adds latency to a visitor page view.
	2244	+ *
	2245	+ * @since 2.2.1
	2246	+ * @return void
	2247	+ */
	2248	+ public function run_overdue_regeneration(): void {
	2249	+ if (!self::has_overdue_regeneration()) {
	2250	+ return;
	2251	+ }
	2252	+
	2253	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2254	+ $pending = is_array($pending) ? $pending : [];
	2255	+ $source = isset($pending['source']) ? (string) $pending['source'] : 'content';
	2256	+
	2257	+ // Cron is running and its own event for this rebuild is still queued
	2258	+ // and due: it is about to do exactly this work. Only then — an event
	2259	+ // that has already been consumed (e.g. it fired while another process
	2260	+ // held the generation lock) is never re-queued, and returning here
	2261	+ // unconditionally left the rebuild to requests that could not finish it.
	2262	+ if (wp_doing_cron()) {
	2263	+ $hook = $source === 'settings' ? 'thinkrank_regenerate_sitemap_settings' : 'thinkrank_regenerate_sitemap';
	2264	+ $next = wp_next_scheduled($hook);
	2265	+
	2266	+ if ($next !== false && $next <= time()) {
	2267	+ return;
	2268	+ }
	2269	+ }
	2270	+
	2271	+ // The last attempt had to stop short of this process's memory limit.
	2272	+ // Retrying at the same (or a lower) limit only repeats that, so leave
	2273	+ // the rebuild to a process with more room — WP-CLI, a system cron, or a
	2274	+ // host with a higher limit — instead of burning it on every request.
	2275	+ if (!$this->has_memory_for_retry($pending)) {
	2276	+ return;
	2277	+ }
	2278	+
	2279	+ if ($source === 'settings') {
	2280	+ $this->regenerate_sitemap_from_settings();
	2281	+ return;
	2282	+ }
	2283	+
	2284	+ $this->auto_regenerate_sitemap();
	2285	+ }
	2286	+
	2287	+ /**
	2288	+ * Acquire the shared generation lock.
	2289	+ *
	2290	+ * @since 2.2.1
	2291	+ * @return bool True when this process may generate.
	2292	+ */
	2293	+ private function acquire_generation_lock(): bool {
	2294	+ if (get_transient(self::GENERATION_LOCK_TRANSIENT)) {
	2295	+ return false;
	2296	+ }
	2297	+
	2298	+ set_transient(self::GENERATION_LOCK_TRANSIENT, time(), 5 * MINUTE_IN_SECONDS);
	2299	+
	2300	+ return true;
	2301	+ }
	2302	+
	2303	+ /**
	2304	+ * Release the shared generation lock.
	2305	+ *
	2306	+ * @since 2.2.1
	2307	+ * @return void
	2308	+ */
	2309	+ private function release_generation_lock(): void {
	2310	+ delete_transient(self::GENERATION_LOCK_TRANSIENT);
	2311	+ }
	2312	+
	2313	+ /**
	2314	+ * This process's PHP memory limit in bytes.
	2315	+ *
	2316	+ * @since 2.10.1
	2317	+ * @return int Bytes, or -1 when unlimited (or unreadable).
	2318	+ */
	2319	+ private function current_memory_limit(): int {
	2320	+ $limit = (string) ini_get('memory_limit');
	2321	+
	2322	+ if ($limit === '' || $limit === '-1') {
	2323	+ return -1;
	2324	+ }
	2325	+
	2326	+ $bytes = (int) wp_convert_hr_to_bytes($limit);
	2327	+
	2328	+ return $bytes > 0 ? $bytes : -1;
	2329	+ }
	2330	+
	2331	+ /**
	2332	+ * May this process retry a rebuild that last stopped at the memory limit?
	2333	+ *
	2334	+ * Raises the limit the way wp-admin does first, so a request that can get
	2335	+ * more room than the failed attempt had is still allowed to try.
	2336	+ *
	2337	+ * @since 2.10.1
	2338	+ * @param array $pending The pending marker.
	2339	+ * @return bool True when there is no recorded memory failure, or this
	2340	+ * process has more memory than the attempt that failed.
	2341	+ */
	2342	+ private function has_memory_for_retry(array $pending): bool {
	2343	+ if (empty($pending['memory_limit'])) {
	2344	+ return true;
	2345	+ }
	2346	+
	2347	+ wp_raise_memory_limit('admin');
	2348	+
	2349	+ $limit = $this->current_memory_limit();
	2350	+
	2351	+ return $limit === -1 || $limit > (int) $pending['memory_limit'];
	2352	+ }
	2353	+
	2354	+ /**
	2355	+ * Release what one walked chunk left behind.
	2356	+ *
	2357	+ * Hydrating a chunk through get_posts()/get_terms() also stores every
	2358	+ * object and its meta in the in-process object cache, which nothing
	2359	+ * empties until the request ends. Unsetting the chunk therefore freed
	2360	+ * nothing, and the walk grew with the size of the site instead of the size
	2361	+ * of a chunk — about 1.3 GB on a 45k-post site.
	2362	+ *
	2363	+ * A persistent object cache that supports it drops only its in-process
	2364	+ * copy (`flush_runtime`); the shared store keeps its data. WordPress's
	2365	+ * default cache has no shared store, and flushing it would empty every
	2366	+ * group for the rest of the request (options, the queried object, other
	2367	+ * plugins' data), so there only the chunk's own entries are deleted. A
	2368	+ * persistent cache without `flush_runtime` is left alone: deleting from it
	2369	+ * would evict the objects for every other request too.
	2370	+ *
	2371	+ * @since 2.10.1
	2372	+ * @param int $chunk_start memory_get_usage(true) before the chunk was hydrated.
	2373	+ * @param array $groups Cache groups keyed by the chunk's object IDs.
	2374	+ * @param int[] $ids The chunk's object IDs, plus any objects it
	2375	+ * hydrated under their own (featured images).
	2376	+ * @return void
	2377	+ * @throws \Error See assert_memory_headroom().
	2378	+ */
	2379	+ private function release_walk_memory(int $chunk_start, array $groups, array $ids): void {
	2380	+ // What one chunk costs before it is released: the margin the next one
	2381	+ // needs. Measured in the same real allocated size assert_memory_headroom()
	2382	+ // compares against the limit, so the two are the same unit.
	2383	+ $this->walk_chunk_cost = max($this->walk_chunk_cost, memory_get_usage(true) - $chunk_start);
	2384	+
	2385	+ if (wp_using_ext_object_cache()) {
	2386	+ if (
	2387	+ function_exists('wp_cache_supports')
	2388	+ && wp_cache_supports('flush_runtime')
	2389	+ && function_exists('wp_cache_flush_runtime')
	2390	+ ) {
	2391	+ wp_cache_flush_runtime();
	2392	+ }
	2393	+ } elseif (!empty($ids)) {
	2394	+ foreach ($groups as $group) {
	2395	+ wp_cache_delete_multiple($ids, $group);
	2396	+ }
	2397	+ }
	2398	+
	2399	+ $this->assert_memory_headroom();
	2400	+ }
	2401	+
	2402	+ /**
	2403	+ * Cache groups get_posts() fills per post for the given post types.
	2404	+ *
	2405	+ * The post, its meta, and one relationships group per taxonomy the post
	2406	+ * type uses (update_object_term_cache()). Term objects themselves are
	2407	+ * bounded by the number of terms, not posts, so they are left cached.
	2408	+ *
	2409	+ * @since 2.10.1
	2410	+ * @param string[] $post_types Post types being walked.
	2411	+ * @return string[] Cache groups keyed by post ID.
	2412	+ */
	2413	+ private function chunk_post_cache_groups(array $post_types): array {
	2414	+ $groups = ['posts', 'post_meta'];
	2415	+
	2416	+ foreach (get_object_taxonomies($post_types) as $taxonomy) {
	2417	+ $groups[] = $taxonomy . '_relationships';
	2418	+ }
	2419	+
	2420	+ return array_values(array_unique($groups));
	2421	+ }
	2422	+
	2423	+ /**
	2424	+ * Stop an automatic rebuild before the memory limit rather than at it.
	2425	+ *
	2426	+ * A PHP memory fatal skips every catch and finally, so the lock, the
	2427	+ * failure record and the backoff are all lost with it, while stopping here
	2428	+ * is an ordinary, fully recorded failure. Checked before each chunk is
	2429	+ * hydrated, against a margin of at least the largest chunk seen so far.
	2430	+ *
	2431	+ * It throws an \Error, not an \Exception, on purpose: the per-segment
	2432	+ * catch (\Exception) blocks in generate_multiple_sitemaps() would otherwise
	2433	+ * swallow it and carry on — writing an index without the aborted segments
	2434	+ * and then pruning their files as orphans. Only the automatic rebuild's
	2435	+ * catch (\Throwable) is meant to see it.
	2436	+ *
	2437	+ * @since 2.10.1
	2438	+ * @return void
	2439	+ * @throws \Error When the automatic rebuild is close to the memory limit.
	2440	+ */
	2441	+ private function assert_memory_headroom(): void {
	2442	+ if (!$this->memory_guard) {
	2443	+ return;
	2444	+ }
	2445	+
	2446	+ $limit = $this->current_memory_limit();
	2447	+ if ($limit === -1) {
	2448	+ return;
	2449	+ }
	2450	+
	2451	+ // A fifth of the limit (at least 32 MB) for writing the files, or one
	2452	+ // and a half of the costliest chunk if that is more — and never more
	2453	+ // than half the limit either way. Without that outer cap a single
	2454	+ // anomalously expensive chunk (500 posts of serialised page-builder or
	2455	+ // ACF meta reaches hundreds of megabytes) puts the margin above the
	2456	+ // limit itself, so every later check aborts at any usage at all, the
	2457	+ // failure records this process's limit, and has_memory_for_retry()
	2458	+ // then refuses every process that has the same limit. A site that
	2459	+ // never actually ran out of memory would stop rebuilding until WP-CLI
	2460	+ // or a system cron happened to run.
	2461	+ $headroom = (int) min(
	2462	+ max(
	2463	+ min(max($limit * 0.2, 32 * MB_IN_BYTES), $limit * 0.5),
	2464	+ $this->walk_chunk_cost * 1.5
	2465	+ ),
	2466	+ $limit * 0.5
	2467	+ );
	2468	+
	2469	+ // The real allocated size, which is what PHP enforces memory_limit
	2470	+ // against; memory_get_usage(false) reports only what is handed out of
	2471	+ // those allocations and so understates the margin by the allocator's
	2472	+ // slack.
	2473	+ $usage = memory_get_usage(true);
	2474	+
	2475	+ if ($usage > $limit - $headroom) {
	2476	+ throw new \Error(
	2477	+ sprintf(
	2478	+ /* translators: 1: memory in use, 2: PHP memory limit. */
	2479	+ esc_html__('The sitemap rebuild was stopped at %1$s of the %2$s PHP memory limit, before PHP would have run out of memory. It will be retried by a process with more memory (WP-CLI or a system cron). To let it finish in the admin, raise the PHP memory_limit.', 'thinkrank'),
	2480	+ esc_html((string) size_format($usage)),
	2481	+ esc_html((string) size_format($limit))
	2482	+ ),
	2483	+ // phpcs:ignore WordPress.Security.EscapeOutput.ExceptionNotEscaped -- an integer class constant, not output.
	2484	+ self::MEMORY_ABORT_CODE
	2485	+ );
	2486	+ }
	2487	+ }
	2488	+
	2489	+ /**
	2490	+ * Run an automatic rebuild's generation with the fatal-safe bookkeeping.
	2491	+ *
	2492	+ * @since 2.10.1
	2493	+ * @param array $settings Sitemap settings.
	2494	+ * @return bool Whatever generate_and_save() returned.
	2495	+ * @throws \Throwable Whatever generation throws, after the memory guard is
	2496	+ * switched back off.
	2497	+ */
	2498	+ private function generate_for_regeneration(array $settings): bool {
	2499	+ // Same headroom wp-admin gives itself; a no-op when the limit is
	2500	+ // already higher or unlimited.
	2501	+ wp_raise_memory_limit('admin');
	2502	+
	2503	+ $this->claim_regeneration_attempt();
	2504	+ $this->memory_guard = true;
	2505	+ $this->walk_chunk_cost = 0;
	2506	+
	2507	+ try {
	2508	+ return $this->generate_and_save($settings);
	2509	+ } finally {
	2510	+ $this->memory_guard = false;
	2511	+ }
	2512	+ }
	2513	+
	2514	+ /**
	2515	+ * Record a failure thrown by an automatic rebuild.
	2516	+ *
	2517	+ * @since 2.10.1
	2518	+ * @param \Throwable $e What was thrown.
	2519	+ * @param string $source Either 'content' or 'settings'.
	2520	+ * @return void
	2521	+ */
	2522	+ private function record_thrown_regeneration_failure(\Throwable $e, string $source): void {
	2523	+ $memory_limit = null;
	2524	+
	2525	+ if ($e instanceof \Error && $e->getCode() === self::MEMORY_ABORT_CODE) {
	2526	+ $memory_limit = $this->current_memory_limit();
	2527	+ $memory_limit = $memory_limit > 0 ? $memory_limit : null;
	2528	+ }
	2529	+
	2530	+ $this->record_regeneration_failure($e->getMessage(), $source, $memory_limit);
	2531	+ }
	2532	+
	2533	+ /**
	2534	+ * Report how automatic regeneration is faring, for the admin UI.
	2535	+ *
	2536	+ * The feature used to fail invisibly: `last_generated` simply stopped
	2537	+ * advancing and nothing drew attention to it (#629).
	2538	+ *
	2539	+ * @since 2.2.1
	2540	+ * @return array Health payload.
	2541	+ */
	2542	+ public function get_regeneration_health(): array {
	2543	+ $settings = $this->get_settings('site');
	2544	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2545	+ $pending = is_array($pending) ? $pending : [];
	2546	+ $error = get_option(self::REGENERATION_ERROR_OPTION, []);
	2547	+ $error = is_array($error) ? $error : [];
	2548	+
	2549	+ $since = !empty($pending['since']) ? (int) $pending['since'] : 0;
	2550	+
	2551	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap');
	2552	+ if ($next_scheduled === false) {
	2553	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap_settings');
	2554	+ }
	2555	+
	2556	+ return [
	2557	+ 'auto_generate' => !empty($settings['auto_generate']),
	2558	+ 'last_generated' => $settings['last_generated'] ?? '',
	2559	+ 'pending_since' => $since ? gmdate('c', $since) : null,
	2560	+ 'pending_seconds' => $since ? max(0, time() - $since) : 0,
	2561	+ 'next_scheduled' => $next_scheduled ? gmdate('c', (int) $next_scheduled) : null,
	2562	+ 'cron_disabled' => defined('DISABLE_WP_CRON') && DISABLE_WP_CRON,
	2563	+ 'stale' => $this->is_sitemap_stale($settings),
	2564	+ 'last_error' => !empty($error['message'])
	2565	+ ? [
	2566	+ 'message' => (string) $error['message'],
	2567	+ 'source' => isset($error['source']) ? (string) $error['source'] : 'content',
	2568	+ 'time' => !empty($error['time']) ? gmdate('c', (int) $error['time']) : null,
	2569	+ ]
	2570	+ : null,
	2571	+ ];
	2572	+ }
	2573	+
	2574	+ /**
	2575	+ * Has published content changed since the served sitemap was last written?
	2576	+ *
	2577	+ * Uses core's cached last-modified lookup, which only considers published
	2578	+ * posts — the same content the sitemap covers.
	2579	+ *
	2580	+ * @since 2.2.1
	2581	+ * @param array $settings Sitemap settings.
	2582	+ * @return bool True when the sitemap is behind the content.
	2583	+ */
	2584	+ private function is_sitemap_stale(array $settings): bool {
	2585	+ if (empty($settings['enabled']) || empty($settings['last_generated'])) {
	2586	+ // Never generated is already reported separately by the UI.
	2587	+ return false;
	2588	+ }
	2589	+
	2590	+ $generated = strtotime((string) $settings['last_generated']);
	2591	+ if (!$generated) {
	2592	+ return false;
	2593	+ }
	2594	+
	2595	+ $modified = get_lastpostmodified('gmt');
	2596	+ if (!$modified) {
	2597	+ return false;
	2598	+ }
	2599	+
	2600	+ $modified = strtotime($modified . ' UTC');
	2601	+ if (!$modified) {
	2602	+ return false;
	2603	+ }
	2604	+
	2605	+ // A minute of slack keeps a rebuild that ran alongside the edit from
	2606	+ // reporting itself as stale.
	2607	+ return $modified > ($generated + MINUTE_IN_SECONDS);
	2608	+ }
	2609	+
	2610	+ /**
1505	2611	* Public entry point to debounce-rebuild the sitemap after a settings change
1506	2612	* (e.g. toggling inclusion rules via REST or the MCP ability), so the served
1507	2613	* file reflects the new settings instead of going stale until a content edit.
1508	2614	*
		@@ -1511,10 +2617,10 @@
1511	2617	public function schedule_regeneration(): void {
1512	2618	// Debounce against rapid successive saves, but use the settings-specific
1513	2619	// hook so the rebuild runs regardless of the auto_generate toggle (which
1514	2620	// only governs content-change-triggered regeneration).
1515		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap_settings');
1516		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap_settings');
	2621	+ $this->mark_regeneration_pending('settings');
	2622	+ $this->debounce_event('thinkrank_regenerate_sitemap_settings');
1517	2623	}
1518	2624
1519	2625	/**
1520	2626	* Rebuild the served sitemap after an explicit settings change.
		@@ -1523,11 +2629,22 @@
1523	2629	* auto_generate setting: the user deliberately changed inclusion rules and
1524	2630	* expects the served file to reflect them even if content-triggered
1525	2631	* auto-generation is turned off. Still respects the master `enabled` flag.
1526	2632	*
1527		- * @return void
	2633	+ * @since 2.10.0 Reports whether the served sitemap was actually rebuilt, so
	2634	+ * a caller can say so rather than assume it (#764). Existing
	2635	+ * callers that ignore the return are unaffected.
	2636	+ *
	2637	+ * @return bool True when the served sitemap now reflects the settings.
1528	2638	*/
1529		- public function regenerate_sitemap_from_settings(): void {
	2639	+ public function regenerate_sitemap_from_settings(): bool {
	2640	+ if (!$this->acquire_generation_lock()) {
	2641	+ // A manual generation (or another request's takeover) is already
	2642	+ // writing the files; the pending marker survives so this rebuild is
	2643	+ // retried rather than lost.
	2644	+ return false;
	2645	+ }
	2646	+
1530	2647	try {
1531	2648	$settings = $this->get_settings('site');
1532	2649	if (empty($settings['enabled'])) {
1533	2650	// The sitemap was disabled: remove the previously generated static
		@@ -1532,18 +2649,81 @@
1532	2649	if (empty($settings['enabled'])) {
1533	2650	// The sitemap was disabled: remove the previously generated static
1534	2651	// files so the web server stops serving a stale sitemap that
1535	2652	// crawlers would otherwise keep fetching.
1536		- $this->delete_published_sitemaps();
1537		- return;
	2653	+ //
	2654	+ // A file that could not be removed is still being served, so
	2655	+ // this is not a success. Reporting one here would tell a caller
	2656	+ // the sitemap was gone while the web server kept answering with
	2657	+ // it, which is the failure this return value exists to prevent
	2658	+ // (#764).
	2659	+ $removal = $this->delete_published_sitemaps($settings);
	2660	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2661	+
	2662	+ if (!empty($stuck)) {
	2663	+ $this->record_regeneration_failure(
	2664	+ $this->stuck_files_message($stuck, true),
	2665	+ 'settings'
	2666	+ );
	2667	+
	2668	+ return false;
	2669	+ }
	2670	+
	2671	+ $this->mark_regeneration_complete();
	2672	+
	2673	+ return true;
1538	2674	}
1539		- $this->generate_and_save($settings);
	2675	+
	2676	+ $revision = $this->current_regeneration_revision();
	2677	+
	2678	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2679	+ // Returns false when a static file is stuck in the web root:
	2680	+ // the server keeps serving that file in preference to WordPress,
	2681	+ // so the switch has not taken effect (#764).
	2682	+ return $this->switch_to_dynamic_delivery($settings, $revision, 'settings');
	2683	+ }
	2684	+
	2685	+ if ($this->generate_for_regeneration($settings)) {
	2686	+ $this->mark_regeneration_complete($revision);
	2687	+
	2688	+ return true;
	2689	+ }
	2690	+
	2691	+ $this->record_regeneration_failure(
	2692	+ $this->write_failure_message(),
	2693	+ 'settings'
	2694	+ );
	2695	+
	2696	+ return false;
1540	2697	} catch (\Throwable $e) {
1541		- // Settings-triggered regeneration failed - details in exception.
	2698	+ $this->record_thrown_regeneration_failure($e, 'settings');
	2699	+
	2700	+ return false;
	2701	+ } finally {
	2702	+ $this->release_generation_lock();
1542	2703	}
1543	2704	}
1544	2705
1545	2706	/**
	2707	+ * When a rebuild has been outstanding since, or 0 when none is.
	2708	+ *
	2709	+ * Lets a caller report an honest "saved, but the served file has not caught
	2710	+ * up yet" instead of a bare success (#764).
	2711	+ *
	2712	+ * @since 2.10.0
	2713	+ * @return int Unix timestamp, or 0 when nothing is pending.
	2714	+ */
	2715	+ public static function regeneration_pending_since(): int {
	2716	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2717	+
	2718	+ if (!is_array($pending) || empty($pending['since'])) {
	2719	+ return 0;
	2720	+ }
	2721	+
	2722	+ return (int) $pending['since'];
	2723	+ }
	2724	+
	2725	+ /**
1546	2726	* Remove every static sitemap file ThinkRank publishes to the web root.
1547	2727	*
1548	2728	* Called when the sitemap feature is disabled, by the cleanup route, and by
1549	2729	* both removal paths, so /sitemap.xml, /sitemap_index.xml, the segmented
		@@ -1583,8 +2763,44 @@
1583	2763	return thinkrank_webroot_segment_filenames($settings);
1584	2764	}
1585	2765
1586	2766	/**
	2767	+ * Can this web-root file be shown to be a sitemap ThinkRank wrote?
	2768	+ *
	2769	+ * The generator deletes as often as cleanup does — a segment that dropped
	2770	+ * out of the set, a pagination page beyond the new count, the local sitemap
	2771	+ * after the business identity was cleared — and until 2.1.1 it did all
	2772	+ * three by filename alone. That is the #515 bug on a far more frequent
	2773	+ * trigger: our names are the canonical ones, so an ordinary regeneration
	2774	+ * (post save, term change, settings save) destroyed RankMath's
	2775	+ * `sitemap-tags.xml` and `local-sitemap.xml` with no deactivation involved.
	2776	+ *
	2777	+ * Every name the generator derives comes from the current settings — the
	2778	+ * url pattern, the configured `sitemap_urls`, `local-sitemap.xml` — but the
	2779	+ * ownership test is still asked with `$name_derived = false`, which switches
	2780	+ * off the legacy fallback for the whole generator side.
	2781	+ *
	2782	+ * The fallback exists to recover a pre-2.1.1 file written with
	2783	+ * `enable_styling` off, which carries neither marker. That recovery belongs
	2784	+ * to the once-off cleanup paths. Here it can only do harm: this method runs
	2785	+ * on every post save, and everything this version writes carries
	2786	+ * THINKRANK_SITEMAP_MARKER, so after the site's first regeneration an
	2787	+ * unmarked file at one of our names is by definition somebody else's — and
	2788	+ * deleting it on an ordinary regeneration is #515 through the more common
	2789	+ * door. The cost is a stale unmarked segment left on disk until deactivation
	2790	+ * picks it up, which is the safe direction to fail in.
	2791	+ *
	2792	+ * @since 2.1.1
	2793	+ *
	2794	+ * @param string $path Absolute path to a file in the web root.
	2795	+ * @param array $settings Sitemap settings.
	2796	+ * @return bool True when the file may be deleted.
	2797	+ */
	2798	+ private function webroot_sitemap_is_ours(string $path, array $settings): bool {
	2799	+ return thinkrank_webroot_sitemap_is_ours($path, $settings, false);
	2800	+ }
	2801	+
	2802	+ /**
1587	2803	* Auto-regenerate sitemap (called by scheduled action)
1588	2804	*
1589	2805	* @since 1.0.0
1590	2806	* @return void
		@@ -1589,20 +2805,48 @@
1589	2805	* @since 1.0.0
1590	2806	* @return void
1591	2807	*/
1592	2808	public function auto_regenerate_sitemap(): void {
	2809	+ if (!$this->acquire_generation_lock()) {
	2810	+ // A manual generation (or another request's takeover) is already
	2811	+ // writing the files; the pending marker survives so this rebuild is
	2812	+ // retried rather than lost.
	2813	+ return;
	2814	+ }
	2815	+
1593	2816	try {
1594	2817	// Double-check that auto-generation is still enabled
1595	2818	if (!$this->should_auto_generate()) {
	2819	+ // Nothing outstanding can be delivered while the feature is off,
	2820	+ // so drop the marker rather than let the takeover retry forever.
	2821	+ $this->mark_regeneration_complete();
1596	2822	return;
1597	2823	}
1598	2824
1599		- $this->generate_and_save($this->get_settings('site'));
	2825	+ $revision = $this->current_regeneration_revision();
	2826	+ $settings = $this->get_settings('site');
1600	2827
1601		- // Sitemap auto-regenerated successfully
	2828	+ // See regenerate_sitemap_from_settings(): nothing to write.
	2829	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2830	+ $this->switch_to_dynamic_delivery($settings, $revision, 'content');
	2831	+ return;
	2832	+ }
1602	2833
	2834	+ if ($this->generate_for_regeneration($settings)) {
	2835	+ $this->mark_regeneration_complete($revision);
	2836	+ } else {
	2837	+ // Previously this returned quietly and last_generated simply
	2838	+ // stopped advancing, leaving the site owner with no way to learn
	2839	+ // the sitemap had stopped updating (#629).
	2840	+ $this->record_regeneration_failure(
	2841	+ $this->write_failure_message(),
	2842	+ 'content'
	2843	+ );
	2844	+ }
1603	2845	} catch (\Throwable $e) {
1604		- // Sitemap auto-regeneration failed - error details available in exception
	2846	+ $this->record_thrown_regeneration_failure($e, 'content');
	2847	+ } finally {
	2848	+ $this->release_generation_lock();
1605	2849	}
1606	2850	}
1607	2851
1608	2852	/**
		@@ -1617,8 +2861,26 @@
1617	2861	* @param array $settings Sitemap settings.
1618	2862	* @return bool True when the sitemap files were written.
1619	2863	*/
1620	2864	public function generate_and_save(array $settings): bool {
	2865	+ // Dynamic delivery publishes no files, so writing them here would put a
	2866	+ // static copy back in the web root for the server to serve in place of
	2867	+ // the dynamic route. Guarding at each call site left gaps — the
	2868	+ // snapshot migrator's post-import regeneration had none — so the rule
	2869	+ // lives with the writing instead.
	2870	+ //
	2871	+ // `is_collecting()` is the exception that makes dynamic delivery work
	2872	+ // at all: render_document() and collect_documents() reach this same
	2873	+ // method with the writer swapped for a collector, and that is precisely
	2874	+ // the dynamic build. Only a real write is skipped.
	2875	+ if (!$this->is_collecting() && 'dynamic' === $this->resolve_delivery_mode($settings)) {
	2876	+ // Whatever prompted this call changed the sitemap's content, so the
	2877	+ // rendered copies must not outlive it.
	2878	+ $this->flush_dynamic_cache();
	2879	+
	2880	+ return true;
	2881	+ }
	2882	+
1621	2883	// Index mode is driven by the use_sitemap_index toggle (not merely by how
1622	2884	// many sitemap_urls happen to be configured). When the toggle is on but
1623	2885	// no child sitemaps are set up yet, synthesize the per-type segmented set
1624	2886	// so we emit a real <sitemapindex> with paginated children instead of a
		@@ -1629,16 +2891,29 @@
1629	2891	$results = $this->generate_multiple_sitemaps($settings);
1630	2892	$written = !empty($results['success']);
1631	2893	} else {
1632	2894	$xml = $this->generate_sitemap($settings);
1633		- $written = $this->save_sitemap_to_file($xml, $this->get_primary_sitemap_filename($settings));
	2895	+ $primary = $this->get_primary_sitemap_filename($settings);
	2896	+ $written = $this->save_sitemap_to_file($xml, $primary);
1634	2897
1635	2898	// Local business sitemap is a standalone file, regenerated on the
1636	2899	// single-sitemap path too (this is the default mode).
1637	2900	$this->regenerate_local_sitemap($settings);
	2901	+
	2902	+ // Switching out of index mode leaves sitemap_index.xml and every
	2903	+ // child on disk, still served and never refreshed again. The index
	2904	+ // path already prunes what it no longer owns; this path never did,
	2905	+ // so the site kept serving two sitemap trees (#563). Ownership is
	2906	+ // still tested per file, so another plugin's sitemap at one of our
	2907	+ // names is never touched (#515).
	2908	+ $this->prune_orphaned_segments($settings, [['filename' => basename($primary)]]);
1638	2909	}
1639	2910
1640		- if ($written) {
	2911	+ // last_generated describes what is on disk. A dynamic render publishes
	2912	+ // nothing, so advancing it would report a static publication that never
	2913	+ // happened and would let primary_sitemap_file_exists() callers believe
	2914	+ // there is a file to serve.
	2915	+ if ($written && !$this->is_collecting()) {
1641	2916	$settings['last_generated'] = gmdate('c');
1642	2917	$this->save_settings('site', null, $settings);
1643	2918	}
1644	2919
		@@ -1645,8 +2920,433 @@
1645	2920	return $written;
1646	2921	}
1647	2922
1648	2923	/**
	2924	+ * Whether this instance is rendering documents rather than publishing them.
	2925	+ *
	2926	+ * @since 2.9.0
	2927	+ *
	2928	+ * @return bool
	2929	+ */
	2930	+ private function is_collecting(): bool {
	2931	+ return $this->document_sink !== null;
	2932	+ }
	2933	+
	2934	+ /**
	2935	+ * Complete a regeneration that delivers dynamically, retiring stale files.
	2936	+ *
	2937	+ * Dynamic delivery renders nothing to disk, but that is only half the job.
	2938	+ * A web server hands back an existing `/sitemap.xml` without ever loading
	2939	+ * WordPress, so any file left over from a previous static generation goes on
	2940	+ * being served forever and {@see \ThinkRank\Frontend\SEO_Manager
	2941	+ * ::maybe_serve_sitemap()} is never reached. Switching to dynamic while
	2942	+ * leaving those files in place would therefore appear to do nothing at all.
	2943	+ *
	2944	+ * Both transitions matter and they differ:
	2945	+ *
	2946	+ * - An explicit switch to `dynamic` happens on a site whose root is usually
	2947	+ * still writable, so the files can simply be removed.
	2948	+ * - An `auto` site that becomes read-only cannot remove them, because
	2949	+ * deleting an entry needs write permission on the directory that holds
	2950	+ * it. There the stale sitemap really is stuck in front of us, and the
	2951	+ * honest outcome is a recorded failure naming it rather than a rebuild
	2952	+ * reported as complete (#754 review).
	2953	+ *
	2954	+ * Ownership is tested per file by the shared helper, so another plugin's
	2955	+ * sitemap at one of our names is never deleted (#515).
	2956	+ *
	2957	+ * @since 2.9.0
	2958	+ *
	2959	+ * @param array $settings Sitemap settings.
	2960	+ * @param int $revision Revision this rebuild is completing.
	2961	+ * @param string $source 'settings' or 'content', for the failure record.
	2962	+ * @return void
	2963	+ */
	2964	+ private function switch_to_dynamic_delivery(array $settings, int $revision, string $source): bool {
	2965	+ $this->flush_dynamic_cache();
	2966	+
	2967	+ $removal = $this->delete_published_sitemaps($settings);
	2968	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2969	+
	2970	+ if (!empty($stuck)) {
	2971	+ $this->record_regeneration_failure($this->stuck_files_message($stuck), $source);
	2972	+
	2973	+ return false;
	2974	+ }
	2975	+
	2976	+ $this->mark_regeneration_complete($revision);
	2977	+
	2978	+ return true;
	2979	+ }
	2980	+
	2981	+ /**
	2982	+ * Why a stale file left in the web root means the change has not landed.
	2983	+ *
	2984	+ * Shared by every path that removes published files, so they cannot
	2985	+ * describe the same situation differently (#764).
	2986	+ *
	2987	+ * The two situations that reach it differ in what WordPress is doing, and
	2988	+ * the message has to say which. After a switch to dynamic delivery
	2989	+ * WordPress IS serving the sitemap and the files shadow it. After the
	2990	+ * sitemap is switched off WordPress serves nothing, so the one message
	2991	+ * used to tell a site owner who had just disabled the sitemap that it was
	2992	+ * "being served from WordPress", which is the opposite of what they did.
	2993	+ *
	2994	+ * @since 2.10.0
	2995	+ * @since 2.10.0 Public, so the REST endpoint uses it rather than a copy;
	2996	+ * takes $sitemap_disabled for the disabled path.
	2997	+ *
	2998	+ * @param string[] $stuck Basenames that could not be removed.
	2999	+ * @param bool $sitemap_disabled True when the files outlived disabling
	3000	+ * the sitemap rather than a switch to
	3001	+ * dynamic delivery.
	3002	+ * @return string
	3003	+ */
	3004	+ public function stuck_files_message(array $stuck, bool $sitemap_disabled = false): string {
	3005	+ if ($sitemap_disabled) {
	3006	+ return sprintf(
	3007	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3008	+ __('The sitemap is disabled, but these files are still in the site root and your web server is still serving them: %1$s. They could not be removed because %2$s is not writable. Delete them, or ask your host to make the WordPress root writable.', 'thinkrank'),
	3009	+ implode(', ', $stuck),
	3010	+ untrailingslashit(ABSPATH)
	3011	+ );
	3012	+ }
	3013	+
	3014	+ return sprintf(
	3015	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3016	+ __('The sitemap is being served from WordPress, but these files are still in the site root and your web server will keep serving them instead: %1$s. They could not be removed because %2$s is not writable. Delete them, or ask your host to make the WordPress root writable.', 'thinkrank'),
	3017	+ implode(', ', $stuck),
	3018	+ untrailingslashit(ABSPATH)
	3019	+ );
	3020	+ }
	3021	+
	3022	+ /**
	3023	+ * What to tell the site owner when publishing the files failed.
	3024	+ *
	3025	+ * The old wording stated the symptom and stopped there, so the reported
	3026	+ * cause was a guess and this reached support as a plugin fault rather than
	3027	+ * a folder permission (#752, #753). When the root is demonstrably
	3028	+ * unwritable, say that, and say what to do about it.
	3029	+ *
	3030	+ * @since 2.9.0
	3031	+ *
	3032	+ * @return string
	3033	+ */
	3034	+ private function write_failure_message(): string {
	3035	+ if (!wp_is_writable(ABSPATH)) {
	3036	+ return sprintf(
	3037	+ /* translators: %s: absolute path to the WordPress root. */
	3038	+ __('The sitemap could not be written because the folder %s is not writable by PHP. Ask your host to make the WordPress root writable, or set Sitemap Delivery to Dynamic to serve the sitemap without writing files.', 'thinkrank'),
	3039	+ untrailingslashit(ABSPATH)
	3040	+ );
	3041	+ }
	3042	+
	3043	+ return __('The sitemap files could not be written to the site root.', 'thinkrank');
	3044	+ }
	3045	+
	3046	+ /**
	3047	+ * How this site delivers its sitemap.
	3048	+ *
	3049	+ * `auto` is resolved on whether the web root can be written. That is the
	3050	+ * right signal here (unlike llms.txt, where the question is whether the
	3051	+ * server applies the .htaccess charset block): a site whose root is
	3052	+ * read-only cannot publish a sitemap file at all, and before this existed
	3053	+ * the feature simply failed with "The sitemap files could not be written to
	3054	+ * the site root." and served nothing (#752).
	3055	+ *
	3056	+ * @since 2.9.0
	3057	+ *
	3058	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3059	+ * @return string One of 'static' or 'dynamic'. Never 'auto'.
	3060	+ */
	3061	+ public function resolve_delivery_mode(?array $settings = null): string {
	3062	+ $settings = $settings ?? $this->get_settings('site');
	3063	+ $mode = (string) ($settings['delivery_mode'] ?? 'auto');
	3064	+
	3065	+ if ('static' === $mode || 'dynamic' === $mode) {
	3066	+ return $mode;
	3067	+ }
	3068	+
	3069	+ return wp_is_writable(ABSPATH) ? 'static' : 'dynamic';
	3070	+ }
	3071	+
	3072	+ /**
	3073	+ * Render one published sitemap document without touching the filesystem.
	3074	+ *
	3075	+ * Runs the ordinary build pipeline with the writer swapped for a collector,
	3076	+ * so the bytes returned here are the bytes the static path would have
	3077	+ * written. `SitemapDeliveryParityTest` asserts that equivalence rather than
	3078	+ * trusting it.
	3079	+ *
	3080	+ * The whole set is built to answer for one file, because the index can only
	3081	+ * be assembled from the children that were actually produced. The result is
	3082	+ * cached per document, so that cost is paid once per change and not once
	3083	+ * per crawler request.
	3084	+ *
	3085	+ * @since 2.9.0
	3086	+ *
	3087	+ * @param string $filename Published file name, e.g. 'sitemap.xml'.
	3088	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3089	+ * @return string|null XML, or null when this site does not publish that name.
	3090	+ */
	3091	+ public function render_document(string $filename, ?array $settings = null): ?string {
	3092	+ $filename = basename($filename);
	3093	+ $settings = $settings ?? $this->get_settings('site');
	3094	+
	3095	+ if (empty($settings['enabled'])) {
	3096	+ return null;
	3097	+ }
	3098	+
	3099	+ $cached = get_transient($this->dynamic_cache_key($filename));
	3100	+ if (self::ABSENT_MARKER === $cached) {
	3101	+ return null;
	3102	+ }
	3103	+ if (is_string($cached) && '' !== $cached) {
	3104	+ return $cached;
	3105	+ }
	3106	+
	3107	+ // A miss builds the whole set, because the index can only be assembled
	3108	+ // from the children that were actually produced. Caching only the
	3109	+ // requested document therefore made a crawler walking the index and its
	3110	+ // children rebuild the entire site's sitemap once per file — every post
	3111	+ // and taxonomy query repeated N times on a public endpoint (#754
	3112	+ // review). The set is built once and stored in full.
	3113	+ return $this->stream_documents($settings, $filename);
	3114	+ }
	3115	+
	3116	+ /**
	3117	+ * Build every document, caching each as it is produced, keeping one.
	3118	+ *
	3119	+ * A miss has to build the whole set, because the index can only be
	3120	+ * assembled from the children that were actually produced. It does not have
	3121	+ * to *hold* the whole set: the static path never keeps more than one page
	3122	+ * in memory, writing each to disk as it goes, and buffering every
	3123	+ * document's XML to return one of them undid that on the request path,
	3124	+ * where a large site's entire sitemap corpus would sit in a single PHP
	3125	+ * process (#754 review).
	3126	+ *
	3127	+ * So the sink writes each document straight to its cache entry and lets it
	3128	+ * go, retaining only the one this request is answering. Peak retention is
	3129	+ * one document, whatever the site's size.
	3130	+ *
	3131	+ * Concurrency: the first request through takes a short lock and does the
	3132	+ * work. One that finds the lock held waits a bounded moment for the winner
	3133	+ * to publish, then builds anyway, because serving a correct sitemap late
	3134	+ * beats serving none.
	3135	+ *
	3136	+ * @since 2.9.0
	3137	+ *
	3138	+ * @param array $settings Sitemap settings.
	3139	+ * @param string $wanted Document this request is answering.
	3140	+ * @return string|null XML for $wanted, or null when the site does not publish it.
	3141	+ */
	3142	+ private function stream_documents(array $settings, string $wanted): ?string {
	3143	+ $lock = self::DYNAMIC_CACHE_PREFIX . 'lock';
	3144	+
	3145	+ if (!$this->acquire_render_lock($lock)) {
	3146	+ for ($attempt = 0; $attempt < self::RENDER_LOCK_WAIT_ATTEMPTS; $attempt++) {
	3147	+ usleep(self::RENDER_LOCK_WAIT_MICROSECONDS);
	3148	+
	3149	+ $cached = get_transient($this->dynamic_cache_key($wanted));
	3150	+ if (self::ABSENT_MARKER === $cached) {
	3151	+ return null;
	3152	+ }
	3153	+ if (is_string($cached) && '' !== $cached) {
	3154	+ return $cached;
	3155	+ }
	3156	+ }
	3157	+ }
	3158	+
	3159	+ $kept = null;
	3160	+ // Names only. Keeping the bodies here would be the very retention this
	3161	+ // method exists to avoid.
	3162	+ $produced = [];
	3163	+
	3164	+ $previous = $this->document_sink;
	3165	+ $this->document_sink = function (string $name, string $xml) use (&$kept, &$produced, $wanted): void {
	3166	+ $produced[$name] = true;
	3167	+ set_transient($this->dynamic_cache_key($name), $xml, self::DYNAMIC_CACHE_TTL);
	3168	+
	3169	+ if ($name === $wanted) {
	3170	+ $kept = $xml;
	3171	+ }
	3172	+ };
	3173	+
	3174	+ try {
	3175	+ $this->generate_and_save($settings);
	3176	+
	3177	+ // Names the configuration lists but this build did not produce get
	3178	+ // a negative entry, so asking for one again is a cache hit rather
	3179	+ // than another full rebuild.
	3180	+ $absent = $this->published_document_names($settings);
	3181	+
	3182	+ // Also the exact name this request asked for: a paginated page past
	3183	+ // the end of a stem is a legitimate request shape that the base
	3184	+ // list cannot enumerate, and without an entry it would rebuild on
	3185	+ // every hit.
	3186	+ $absent[] = $wanted;
	3187	+
	3188	+ foreach (array_unique($absent) as $name) {
	3189	+ if (!isset($produced[$name])) {
	3190	+ set_transient($this->dynamic_cache_key($name), self::ABSENT_MARKER, self::DYNAMIC_CACHE_TTL);
	3191	+ }
	3192	+ }
	3193	+ } finally {
	3194	+ $this->document_sink = $previous;
	3195	+ delete_transient($lock);
	3196	+ }
	3197	+
	3198	+ return $kept;
	3199	+ }
	3200	+
	3201	+ /**
	3202	+ * Take the render lock, if it is free.
	3203	+ *
	3204	+ * Not atomic across processes, and deliberately so: the fallback for losing
	3205	+ * a race is duplicated work, never a wrong or missing sitemap, so a
	3206	+ * heavier primitive would buy nothing here.
	3207	+ *
	3208	+ * @since 2.9.0
	3209	+ *
	3210	+ * @param string $lock Lock transient name.
	3211	+ * @return bool True when this request holds the lock.
	3212	+ */
	3213	+ private function acquire_render_lock(string $lock): bool {
	3214	+ if (false !== get_transient($lock)) {
	3215	+ return false;
	3216	+ }
	3217	+
	3218	+ set_transient($lock, time(), self::RENDER_LOCK_TTL);
	3219	+
	3220	+ return true;
	3221	+ }
	3222	+
	3223	+ /**
	3224	+ * Build every document this site publishes and return them all.
	3225	+ *
	3226	+ * Verification and tooling only. This retains the whole set in memory, so
	3227	+ * it must never be used to answer a request: {@see self::stream_documents()}
	3228	+ * is the serving path and keeps one document at a time regardless of site
	3229	+ * size (#754 review). `SitemapDeliveryParityTest` enforces that separation
	3230	+ * by failing if the request path routes back through here.
	3231	+ *
	3232	+ * @since 2.9.0
	3233	+ *
	3234	+ * @param array $settings Sitemap settings.
	3235	+ * @return array<string,string> Filename => XML.
	3236	+ */
	3237	+ public function collect_documents(array $settings): array {
	3238	+ $documents = [];
	3239	+
	3240	+ $previous = $this->document_sink;
	3241	+ $this->document_sink = static function (string $name, string $xml) use (&$documents): void {
	3242	+ $documents[$name] = $xml;
	3243	+ };
	3244	+
	3245	+ try {
	3246	+ $this->generate_and_save($settings);
	3247	+ } finally {
	3248	+ $this->document_sink = $previous;
	3249	+ }
	3250	+
	3251	+ return $documents;
	3252	+ }
	3253	+
	3254	+ /**
	3255	+ * The file names this site publishes, without building their contents.
	3256	+ *
	3257	+ * Used by the request router to decide whether a URL is ours before doing
	3258	+ * any work. Cheap: it reads the configured child list rather than querying
	3259	+ * for entries.
	3260	+ *
	3261	+ * @since 2.9.0
	3262	+ *
	3263	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3264	+ * @return string[] File names, including paginated pages that may exist.
	3265	+ */
	3266	+ public function published_document_names(?array $settings = null): array {
	3267	+ $settings = $settings ?? $this->get_settings('site');
	3268	+ $resolved = $this->maybe_promote_to_index($settings);
	3269	+
	3270	+ $names = [$this->get_primary_sitemap_filename($settings), 'local-sitemap.xml'];
	3271	+
	3272	+ foreach ((array) ($resolved['sitemap_urls'] ?? []) as $child) {
	3273	+ if (!is_array($child) || empty($child['enabled'])) {
	3274	+ continue;
	3275	+ }
	3276	+
	3277	+ $path = (string) wp_parse_url((string) ($child['url'] ?? ''), PHP_URL_PATH);
	3278	+ if ('' !== $path) {
	3279	+ $names[] = basename($path);
	3280	+ }
	3281	+ }
	3282	+
	3283	+ return array_values(array_unique(array_filter($names)));
	3284	+ }
	3285	+
	3286	+ /**
	3287	+ * Does this site publish a document under that name?
	3288	+ *
	3289	+ * Not a plain membership test against {@see self::published_document_names()}:
	3290	+ * that lists the configured children, and a child over the per-file URL cap
	3291	+ * is split into `<stem>-2.xml`, `<stem>-3.xml` and so on, with every page
	3292	+ * listed in the index. Gating the request router on the base list alone
	3293	+ * therefore 404'd exactly the pages the index points at, which is worse than
	3294	+ * not serving them at all.
	3295	+ *
	3296	+ * Page counts are not knowable without building, so the stem is what is
	3297	+ * matched; a page that does not exist is answered by the build finding
	3298	+ * nothing for it, and is then cached as absent.
	3299	+ *
	3300	+ * @since 2.9.0
	3301	+ *
	3302	+ * @param string $name Requested file name.
	3303	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3304	+ * @return bool
	3305	+ */
	3306	+ public function publishes_document_name(string $name, ?array $settings = null): bool {
	3307	+ $names = $this->published_document_names($settings);
	3308	+
	3309	+ if (in_array($name, $names, true)) {
	3310	+ return true;
	3311	+ }
	3312	+
	3313	+ if (!preg_match('/^(.*)-\d+\.xml$/i', $name, $m)) {
	3314	+ return false;
	3315	+ }
	3316	+
	3317	+ return in_array($m[1] . '.xml', $names, true);
	3318	+ }
	3319	+
	3320	+ /**
	3321	+ * Transient key for a rendered document.
	3322	+ *
	3323	+ * @since 2.9.0
	3324	+ *
	3325	+ * @param string $filename Published file name.
	3326	+ * @return string
	3327	+ */
	3328	+ private function dynamic_cache_key(string $filename): string {
	3329	+ return self::DYNAMIC_CACHE_PREFIX . md5($filename);
	3330	+ }
	3331	+
	3332	+ /**
	3333	+ * Drop every cached dynamic document.
	3334	+ *
	3335	+ * Called from the same places that mark the static files stale, so the two
	3336	+ * delivery modes invalidate on identical triggers.
	3337	+ *
	3338	+ * @since 2.9.0
	3339	+ *
	3340	+ * @return void
	3341	+ */
	3342	+ public function flush_dynamic_cache(): void {
	3343	+ foreach ($this->published_document_names() as $name) {
	3344	+ delete_transient($this->dynamic_cache_key($name));
	3345	+ }
	3346	+ }
	3347	+
	3348	+ /**
1649	3349	* Resolve index-vs-single mode, synthesizing child sitemaps when needed.
1650	3350	*
1651	3351	* - When use_sitemap_index is on but no child sitemaps are configured, build
1652	3352	* the per-type segmented set so a real <sitemapindex> is produced (#127).
		@@ -1841,8 +3541,18 @@
1841	3541	// File validation failed - error details available in exception
1842	3542	return false;
1843	3543	}
1844	3544
	3545	+ // Dynamic delivery: hand the document to the collector instead of the
	3546	+ // filesystem. Reported as published, because for this run it is — the
	3547	+ // caller's success/failure bookkeeping and the index assembly both key
	3548	+ // off this return value.
	3549	+ if ($this->document_sink !== null) {
	3550	+ ($this->document_sink)($filename, $sitemap_xml);
	3551	+
	3552	+ return true;
	3553	+ }
	3554	+
1845	3555	$sitemap_path = ABSPATH . $filename;
1846	3556
1847	3557	// Use WordPress filesystem API for better security
1848	3558	global $wp_filesystem;
		@@ -1851,9 +3561,22 @@
1851	3561	WP_Filesystem();
1852	3562	}
1853	3563
1854	3564	if ($wp_filesystem) {
1855		- return $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3565	+ $written = $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3566	+
	3567	+ if ($written) {
	3568	+ // Every sitemap this version writes carries the ownership
	3569	+ // marker, so once one has been written an unmarked file at one
	3570	+ // of our names cannot be ours. Recording that retires the
	3571	+ // legacy fallback for this install — see
	3572	+ // thinkrank_webroot_sitemap_is_ours().
	3573	+ if (get_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION) !== '1') {
	3574	+ update_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION, '1', false);
	3575	+ }
	3576	+ }
	3577	+
	3578	+ return $written;
1856	3579	}
1857	3580
1858	3581	// WP_Filesystem initialization failed
1859	3582	return false;
		@@ -1887,13 +3610,55 @@
1887	3610
1888	3611	// Public custom post types each get a child sitemap (parity with the
1889	3612	// "complete" preset and with Rank Math, which lists every public CPT).
1890	3613	foreach (get_post_types(['public' => true, '_builtin' => false], 'names') as $cpt) {
1891		- if ($this->should_include_post_type($cpt)) {
1892		- $urls[] = $this->build_child_sitemap_entry($cpt, $pattern);
	3614	+ if (!$this->should_include_post_type($cpt)) {
	3615	+ continue;
1893	3616	}
	3617	+
	3618	+ // ...and the per-content-type sitemap switch (#660).
	3619	+ // get_enabled_post_types() already honours it, but this list is what
	3620	+ // index mode builds its children from — so without the same test a
	3621	+ // CPT the user had switched off still got its own child sitemap,
	3622	+ // created and streamed in full. The flags live in $inclusions, which
	3623	+ // is the settings array these children are derived from.
	3624	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $cpt, $inclusions)) {
	3625	+ continue;
	3626	+ }
	3627	+
	3628	+ $urls[] = $this->build_child_sitemap_entry($cpt, $pattern);
1894	3629	}
1895	3630
	3631	+ // Public custom taxonomies get the same treatment (#690). Flat mode has
	3632	+ // always walked them through get_enabled_taxonomies(); index mode built
	3633	+ // its children from the list above and never consulted a taxonomy at
	3634	+ // all, so every custom-taxonomy archive silently vanished from the
	3635	+ // sitemap the moment a site switched modes — and the per-taxonomy switch
	3636	+ // the matrix writes had nothing to act on. Same two tests the post-type
	3637	+ // walk applies, in the same order.
	3638	+ $taken = array_column($urls, 'type');
	3639	+
	3640	+ foreach (get_taxonomies(['public' => true, '_builtin' => false], 'names') as $taxonomy) {
	3641	+ if (!$this->should_include_taxonomy($taxonomy)) {
	3642	+ continue;
	3643	+ }
	3644	+
	3645	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $inclusions)) {
	3646	+ continue;
	3647	+ }
	3648	+
	3649	+ // Post types and taxonomies are separate registries, so a site can
	3650	+ // hold both a `foo` post type and a `foo` taxonomy. They would
	3651	+ // resolve to one filename, and stream_type_entries() answers post
	3652	+ // types first, so the second child would list the first one's file
	3653	+ // twice in the index rather than adding anything.
	3654	+ if (in_array($taxonomy, $taken, true)) {
	3655	+ continue;
	3656	+ }
	3657	+
	3658	+ $urls[] = $this->build_child_sitemap_entry($taxonomy, $pattern);
	3659	+ }
	3660	+
1896	3661	return $urls;
1897	3662	}
1898	3663
1899	3664	/**
		@@ -2040,8 +3805,48 @@
2040	3805	if (post_type_exists($type) && !$this->should_include_post_type($type)) {
2041	3806	continue;
2042	3807	}
2043	3808
	3809	+ // Same for the matrix switch: a child list saved before the user
	3810	+ // excluded this content type still names it, and regenerating from
	3811	+ // that list would rewrite the file they asked not to have. Built-in
	3812	+ // aggregates ('posts', 'pages', ...) are not post type names, so
	3813	+ // post_type_exists() keeps this to real custom post types.
	3814	+ if (post_type_exists($type)
	3815	+ && !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $type, $settings)) {
	3816	+ continue;
	3817	+ }
	3818	+
	3819	+ // The taxonomy counterpart of the two guards above (#690). A child
	3820	+ // list saved while a taxonomy was still included keeps naming it, so
	3821	+ // without this, excluding one in the matrix would still rewrite and
	3822	+ // re-list the file the user asked not to have. The built-in
	3823	+ // aggregates are named 'categories'/'tags' rather than
	3824	+ // 'category'/'post_tag', so taxonomy_exists() leaves them to the
	3825	+ // inclusion-flag check below.
	3826	+ $child_taxonomy = self::CHILD_TYPE_ALIASES[$type] ?? $type;
	3827	+ if (taxonomy_exists($child_taxonomy)
	3828	+ && (!$this->should_include_taxonomy($child_taxonomy)
	3829	+ || !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $child_taxonomy, $settings))) {
	3830	+ continue;
	3831	+ }
	3832	+
	3833	+ // The built-in aggregates carry their switch in the inclusion flag
	3834	+ // rather than under a post type name, and they are the four most
	3835	+ // people actually use. build_segmented_sitemap_urls() drops a child
	3836	+ // whose flag is empty; regenerating from a list saved while it was
	3837	+ // still on has to make the same decision, or turning Posts, Pages,
	3838	+ // Categories or Tags off in the matrix rewrites and re-lists the
	3839	+ // very file it was asked to remove.
	3840	+ // An absent flag means "not configured", which every other reader
	3841	+ // treats as included; only a flag that is present and off excludes.
	3842	+ $aggregate_flag = array_search($type, self::INCLUSION_CHILD_TYPES, true);
	3843	+ if ($aggregate_flag !== false
	3844	+ && array_key_exists($aggregate_flag, $settings)
	3845	+ && empty($settings[$aggregate_flag])) {
	3846	+ continue;
	3847	+ }
	3848	+
2044	3849	try {
2045	3850	// The single "general"/"WordPress" sitemap is one un-paginated file
2046	3851	// (there is no index to reference extra pages); it no longer drops
2047	3852	// overflow URLs.
		@@ -2077,18 +3882,36 @@
2077	3882	$buffer = [];
2078	3883	}
2079	3884	}
2080	3885
2081		- // Flush the trailing partial page, or a single empty page when the
2082		- // type had no entries at all (parity with the previous behavior of
2083		- // always writing at least one page per configured child).
2084		- if (!empty($buffer) || $page === 0) {
	3886	+ // Flush the trailing partial page.
	3887	+ //
	3888	+ // A type that produced nothing writes no page at all in index
	3889	+ // mode (#836). It used to write one empty urlset and list it in
	3890	+ // the index, so a crawler was asked to fetch a file that
	3891	+ // answers with no URLs — Search Console reports an empty
	3892	+ // sitemap referenced from an index as a warning, and the fetch
	3893	+ // is wasted on every pass. On this site three of eleven
	3894	+ // children were empty: a post type with nothing published and
	3895	+ // two taxonomies with no terms.
	3896	+ //
	3897	+ // Single-file mode still writes its one page even when empty,
	3898	+ // because that file IS the site's /sitemap.xml and a 404 there
	3899	+ // is worse than an empty urlset. Nothing lists it, so it costs
	3900	+ // no crawl budget.
	3901	+ //
	3902	+ // Skipping the write also keeps the filename out of
	3903	+ // $results['sitemaps_generated'], which is what
	3904	+ // prune_orphaned_segments() treats as "written this run" — so
	3905	+ // a type that empties after previously publishing has its
	3906	+ // stale file deleted rather than left serving.
	3907	+ if (!empty($buffer) || ($page === 0 && $index_config === null)) {
2085	3908	$page++;
2086	3909	$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children);
2087	3910	}
2088	3911
2089	3912	// Remove pages left over from a previous, larger generation.
2090		- $this->cleanup_stale_pages($sitemap_config['url'], $page);
	3913	+ $this->cleanup_stale_pages($sitemap_config['url'], $page, $settings);
2091	3914
2092	3915	} catch (\Exception $e) {
2093	3916	$results['errors'][] = "Error generating {$type} sitemap: " . $e->getMessage();
2094	3917	$results['success'] = false;
		@@ -2107,10 +3930,16 @@
2107	3930	$results['errors'][] = 'Error generating local sitemap: ' . $e->getMessage();
2108	3931	$results['success'] = false;
2109	3932	}
2110	3933
2111		- // Build the index last, from the child files actually generated.
	3934	+ // Build the index last, from the child files actually generated, plus the
	3935	+ // sitemaps other plugins own: those serve their own URLs and write no
	3936	+ // file here, so they are appended to the index only (#104).
2112	3937	if ($index_config !== null) {
	3938	+ foreach (self::additional_sitemaps() as $extra) {
	3939	+ $index_children[] = ['url' => $extra];
	3940	+ }
	3941	+
2113	3942	try {
2114	3943	$index_xml = $this->generate_sitemap_index($index_children, $settings);
2115	3944	$filename = basename(wp_parse_url($index_config['url'], PHP_URL_PATH));
2116	3945
		@@ -2155,8 +3984,15 @@
2155	3984	* @param array $generated Entries from $results['sitemaps_generated'].
2156	3985	* @return string[] Basenames removed.
2157	3986	*/
2158	3987	private function prune_orphaned_segments(array $settings, array $generated): array {
	3988	+ // Rendering for a request, not publishing: there is nothing on disk
	3989	+ // this run owns, and a dynamic render must never delete the files a
	3990	+ // site's previous static mode left behind.
	3991	+ if ($this->is_collecting()) {
	3992	+ return [];
	3993	+ }
	3994	+
2159	3995	$kept = [];
2160	3996	foreach ($generated as $entry) {
2161	3997	if (!empty($entry['filename'])) {
2162	3998	$kept[strtolower((string) $entry['filename'])] = true;
		@@ -2162,15 +3998,22 @@
2162	3998	$kept[strtolower((string) $entry['filename'])] = true;
2163	3999	}
2164	4000	}
2165	4001
2166		- // The index and the local business sitemap are written by their own
2167		- // paths and are not segments, so they are never orphans here.
2168		- $kept[strtolower(basename($this->get_primary_sitemap_filename($settings)))] = true;
	4002	+ // The current mode's primary and the local business sitemap are written
	4003	+ // by their own paths and are never orphans here.
	4004	+ $primary = strtolower(basename($this->get_primary_sitemap_filename($settings)));
	4005	+ $kept[$primary] = true;
2169	4006	$kept['local-sitemap.xml'] = true;
2170		- $kept['sitemap.xml'] = true;
2171		- $kept['sitemap_index.xml'] = true;
2172	4007
	4008	+ // The OTHER mode's primary is an orphan the moment the mode changes:
	4009	+ // index mode leaves sitemap.xml behind, flat mode leaves
	4010	+ // sitemap_index.xml and its children. Both used to be kept
	4011	+ // unconditionally, so the site served two sitemap trees and only ever
	4012	+ // refreshed one (#563). The children are already covered by the segment
	4013	+ // sweep below, which now sees them because the index is no longer kept.
	4014	+ $stale_primaries = array_diff(['sitemap.xml', 'sitemap_index.xml'], [$primary]);
	4015	+
2173	4016	$removed = [];
2174	4017
2175	4018	global $wp_filesystem;
2176	4019	if (!$wp_filesystem) {
		@@ -2180,9 +4023,11 @@
2180	4023	if (!$wp_filesystem) {
2181	4024	return $removed;
2182	4025	}
2183	4026
2184		- foreach ($this->publishable_segment_filenames($settings) as $candidate) {
	4027	+ $candidates = array_merge($this->publishable_segment_filenames($settings), $stale_primaries);
	4028	+
	4029	+ foreach ($candidates as $candidate) {
2185	4030	if (isset($kept[strtolower($candidate)])) {
2186	4031	continue;
2187	4032	}
2188	4033
		@@ -2201,8 +4046,13 @@
2201	4046	foreach ($paths as $path) {
2202	4047	if (!file_exists($path)) {
2203	4048	continue;
2204	4049	}
	4050	+ // A name we could have published is not proof we published
	4051	+ // this file: RankMath and core write at the same paths (#515).
	4052	+ if (!$this->webroot_sitemap_is_ours($path, $settings)) {
	4053	+ continue;
	4054	+ }
2205	4055	if ($wp_filesystem->delete($path)) {
2206	4056	$removed[] = basename($path);
2207	4057	}
2208	4058	}
		@@ -2238,11 +4088,15 @@
2238	4088	$settings = $settings ?? $this->get_settings('site');
2239	4089	$entries = $this->collect_local_entries();
2240	4090
2241	4091	if (empty($entries)) {
2242		- // Business identity was cleared — drop any file left from before.
	4092	+ // Business identity was cleared — drop the file we left from
	4093	+ // before, but only ours. `local-sitemap.xml` is the name Rank Math
	4094	+ // publishes under too (this method mirrors it deliberately), so on
	4095	+ // a migrated site the file at that path may never have been ours
	4096	+ // to delete (#515).
2243	4097	$path = ABSPATH . 'local-sitemap.xml';
2244		- if (file_exists($path)) {
	4098	+ if (!$this->is_collecting() && file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) {
2245	4099	wp_delete_file($path);
2246	4100	}
2247	4101	return false;
2248	4102	}
		@@ -2264,8 +4118,25 @@
2264	4118	*
2265	4119	* @since 1.15.x
2266	4120	* @return array Zero or one URL entry
2267	4121	*/
	4122	+ /**
	4123	+ * Does this site publish a local business sitemap right now?
	4124	+ *
	4125	+ * The same gate {@see self::regenerate_local_sitemap()} applies, asked
	4126	+ * without writing anything. Callers that need to know whether the document
	4127	+ * exists must not test the filesystem: under dynamic delivery it is served
	4128	+ * from PHP and there is no file, which is how `local-sitemap.xml` came to be
	4129	+ * dropped from robots.txt on exactly those sites (#752).
	4130	+ *
	4131	+ * @since 2.9.0
	4132	+ *
	4133	+ * @return bool True when the local sitemap has content to publish.
	4134	+ */
	4135	+ public function publishes_local_sitemap(): bool {
	4136	+ return !empty($this->collect_local_entries());
	4137	+ }
	4138	+
2268	4139	private function collect_local_entries(): array {
2269	4140	if (!class_exists('ThinkRank\\SEO\\Site_Identity_Manager')) {
2270	4141	require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-site-identity-manager.php';
2271	4142	}
		@@ -2337,17 +4208,43 @@
2337	4208	case 'products':
2338	4209	$entries = post_type_exists('product') ? $this->collect_post_entries_iter(['product'], $settings) : [];
2339	4210	return ['entries' => $entries, 'image_ns' => true];
2340	4211
2341		- case 'product_categories':
2342		- $entries = taxonomy_exists('product_cat') ? $this->collect_taxonomy_entries_iter('product_cat', $settings) : [];
2343		- return ['entries' => $entries, 'image_ns' => false];
	4212	+ default:
	4213	+ // Resolve a preset's display name to the object it streams, so
	4214	+ // 'product_categories' is an ordinary taxonomy child rather than
	4215	+ // a case of its own (#690).
	4216	+ $alias = self::CHILD_TYPE_ALIASES[$type] ?? null;
	4217	+ $object = $alias ?? $type;
2344	4218
2345		- default:
2346	4219	$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names');
2347		- if (in_array($type, $custom_post_types, true)) {
2348		- return ['entries' => $this->collect_post_entries_iter([$type], $settings), 'image_ns' => true];
	4220	+ if (in_array($object, $custom_post_types, true)) {
	4221	+ return ['entries' => $this->collect_post_entries_iter([$object], $settings), 'image_ns' => true];
2349	4222	}
	4223	+
	4224	+ // Custom taxonomies reach index mode here, streamed through the
	4225	+ // same iterator flat mode uses so the two modes emit identical
	4226	+ // URLs for the same settings.
	4227	+ if (taxonomy_exists($object) && $this->should_include_taxonomy($object)) {
	4228	+ return ['entries' => $this->collect_taxonomy_entries_iter($object, $settings), 'image_ns' => false];
	4229	+ }
	4230	+
	4231	+ // An aliased child whose object is gone (WooCommerce deactivated)
	4232	+ // keeps writing the empty file it always wrote. Returning null
	4233	+ // here would hand it the whole-site fallback instead, dumping
	4234	+ // every URL on the site into a file named for products.
	4235	+ //
	4236	+ // A registered taxonomy this generator will not emit
	4237	+ // (`post_format`, `nav_menu`, a non-public one) needs the same
	4238	+ // answer for the same reason. generate_multiple_sitemaps()
	4239	+ // skips those before they reach here, so nothing takes this
	4240	+ // path today — but it is the one branch where falling through
	4241	+ // to null is silently catastrophic rather than merely wrong,
	4242	+ // and the guard keeping it unreachable lives in another method.
	4243	+ if ($alias !== null || taxonomy_exists($object)) {
	4244	+ return ['entries' => [], 'image_ns' => false];
	4245	+ }
	4246	+
2350	4247	return null;
2351	4248	}
2352	4249	}
2353	4250
		@@ -2390,13 +4287,23 @@
2390	4287	* which has no -N suffix) is never touched.
2391	4288	*
2392	4289	* @since 1.14.0
2393	4290	*
	4291	+ * @since 2.1.1 Each candidate must pass the content ownership test — a
	4292	+ * `-N.xml` page of another plugin's sitemap paginates our
	4293	+ * stem exactly as ours does (#515).
	4294	+ *
2394	4295	* @param string $base_url Base (page 1) sitemap URL
2395	4296	* @param int $current_pages Number of pages generated this run
	4297	+ * @param array $settings Sitemap settings, for the ownership test.
2396	4298	* @return void
2397	4299	*/
2398		- private function cleanup_stale_pages(string $base_url, int $current_pages): void {
	4300	+ private function cleanup_stale_pages(string $base_url, int $current_pages, array $settings): void {
	4301	+ // See prune_orphaned_segments(): a dynamic render deletes nothing.
	4302	+ if ($this->is_collecting()) {
	4303	+ return;
	4304	+ }
	4305	+
2399	4306	$filename = basename(wp_parse_url($base_url, PHP_URL_PATH));
2400	4307	if (!preg_match('/^(.*)\.xml$/i', $filename, $m)) {
2401	4308	return;
2402	4309	}
		@@ -2413,9 +4320,11 @@
2413	4320
2414	4321	$candidates = glob(ABSPATH . $stem . '-*.xml') ?: [];
2415	4322	foreach ($candidates as $path) {
2416	4323	// Only delete numeric-suffixed pages beyond the current count.
2417		- if (preg_match('/-(\d+)\.xml$/', basename($path), $mm) && (int) $mm[1] > $current_pages) {
	4324	+ if (preg_match('/-(\d+)\.xml$/', basename($path), $mm)
	4325	+ && (int) $mm[1] > $current_pages
	4326	+ && $this->webroot_sitemap_is_ours($path, $settings)) {
2418	4327	$wp_filesystem->delete($path);
2419	4328	}
2420	4329	}
2421	4330	}
		@@ -2420,8 +4329,97 @@
2420	4329	}
2421	4330	}
2422	4331
2423	4332	/**
	4333	+ * Sitemap URLs contributed by other plugins.
	4334	+ *
	4335	+ * ThinkRank owns the sitemap index and the robots.txt `Sitemap:` lines, so a
	4336	+ * companion plugin that serves its own sitemap — Pro's news and video
	4337	+ * sitemaps, for instance — had no way to be discovered: it appeared in
	4338	+ * neither, leaving manual Search Console submission as the only route in
	4339	+ * (#104). Registering here puts a sitemap in the index when one exists, and
	4340	+ * in robots.txt when it does not.
	4341	+ *
	4342	+ * Callers get root-relative paths. Entries are normalised to a leading
	4343	+ * slash, de-duplicated, and anything that is not a non-empty string is
	4344	+ * dropped, so one badly-behaved callback cannot produce a malformed index.
	4345	+ *
	4346	+ * @since 2.3.1
	4347	+ *
	4348	+ * @return string[] Root-relative sitemap paths, e.g. ['/news-sitemap.xml'].
	4349	+ */
	4350	+ public static function additional_sitemaps(): array {
	4351	+ /**
	4352	+ * Filters the sitemaps contributed by other plugins.
	4353	+ *
	4354	+ * @since 2.3.1
	4355	+ *
	4356	+ * @param string[] $sitemaps Root-relative sitemap paths.
	4357	+ */
	4358	+ $sitemaps = apply_filters('thinkrank_additional_sitemaps', []);
	4359	+
	4360	+ if (!is_array($sitemaps)) {
	4361	+ return [];
	4362	+ }
	4363	+
	4364	+ // Both consumers resolve an entry with home_url(), which prefixes the
	4365	+ // install's own directory. Everything below is measured against that so
	4366	+ // an absolute URL is reduced to what home_url() will put back.
	4367	+ $home = wp_parse_url(home_url('/'));
	4368	+ $home_host = strtolower((string) ($home['host'] ?? ''));
	4369	+ $home_path = '/' . trim((string) ($home['path'] ?? ''), '/');
	4370	+
	4371	+ $clean = [];
	4372	+ foreach ($sitemaps as $sitemap) {
	4373	+ if (!is_string($sitemap)) {
	4374	+ continue;
	4375	+ }
	4376	+
	4377	+ $sitemap = trim($sitemap);
	4378	+ if ('' === $sitemap) {
	4379	+ continue;
	4380	+ }
	4381	+
	4382	+ // A full URL on this site is accepted and reduced to the part
	4383	+ // home_url() does not already supply, so a caller that reached for
	4384	+ // home_url() still lands in the right place — including on a
	4385	+ // subdirectory install, where keeping the whole path would repeat
	4386	+ // the directory. A URL on another host is dropped rather than
	4387	+ // rewritten: the sitemaps protocol will not accept a cross-host
	4388	+ // child anyway, and reusing its path would advertise a URL on this
	4389	+ // site that does not exist.
	4390	+ if (preg_match('#^(https?:)?//#i', $sitemap)) {
	4391	+ $parts = wp_parse_url('//' === substr($sitemap, 0, 2) ? 'https:' . $sitemap : $sitemap);
	4392	+ if (!is_array($parts)) {
	4393	+ continue;
	4394	+ }
	4395	+
	4396	+ if (strtolower((string) ($parts['host'] ?? '')) !== $home_host) {
	4397	+ continue;
	4398	+ }
	4399	+
	4400	+ $path = (string) ($parts['path'] ?? '');
	4401	+ if ('' === $path) {
	4402	+ continue;
	4403	+ }
	4404	+
	4405	+ if ('/' !== $home_path && ($path === $home_path || 0 === strpos($path, $home_path . '/'))) {
	4406	+ $path = substr($path, strlen($home_path));
	4407	+ }
	4408	+
	4409	+ // A sitemap served from a query string keeps it; dropping the
	4410	+ // query would point at a different document.
	4411	+ $query = (string) ($parts['query'] ?? '');
	4412	+ $sitemap = $path . ('' !== $query ? '?' . $query : '');
	4413	+ }
	4414	+
	4415	+ $clean[] = '/' . ltrim($sitemap, '/');
	4416	+ }
	4417	+
	4418	+ return array_values(array_unique($clean));
	4419	+ }
	4420	+
	4421	+ /**
2424	4422	* Generate sitemap index XML from the list of child sitemap files produced
2425	4423	* during generation (each already resolved to its final, possibly paginated,
2426	4424	* URL).
2427	4425	*
		@@ -2430,14 +4428,9 @@
2430	4428	* @param array $settings Sitemap settings
2431	4429	* @return string Sitemap index XML
2432	4430	*/
2433	4431	private function generate_sitemap_index(array $children, array $settings): string {
2434		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
2435		-
2436		- // Add XSL stylesheet only if styling is enabled
2437		- if (!empty($settings['enable_styling'])) {
2438		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap-index.xsl') . '"?>' . "\n";
2439		- }
	4432	+ $xml = $this->xml_prolog($settings, 'index');
2440	4433	$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
2441	4434
2442	4435	$site_url = home_url();
2443	4436
		@@ -2450,9 +4443,9 @@
2450	4443	$sitemap_url = $site_url . $sitemap_url;
2451	4444	}
2452	4445
2453	4446	$xml .= " <sitemap>\n";
2454		- $xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n";
	4447	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($sitemap_url)) . "</loc>\n";
2455	4448	$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n";
2456	4449	$xml .= " </sitemap>\n";
2457	4450	}
2458	4451