← All changes | includes/seo/class-sitemap-generator.php +2582 -196 1.29.0 → 2.14.2 View file →

		@@ -21,8 +21,13 @@
21	21	if ( ! defined( 'ABSPATH' ) ) {
22	22	exit;
23	23	}
24	24
	25	+// Filename derivation and web-root removal are shared with the deactivator and
	26	+// with uninstall.php, which runs without an autoloader — so they live in a
	27	+// plain function file both can require. See includes/cleanup-webroot.php.
	28	+require_once __DIR__ . '/../cleanup-webroot.php';
	29	+
25	30	/**
26	31	* Sitemap Generator Class
27	32	*
28	33	* Generates and manages XML sitemaps for search engine optimization.
		@@ -33,13 +38,246 @@
33	38	*/
34	39	class Sitemap_Generator extends Abstract_SEO_Manager {
35	40
36	41	/**
	42	+ * Memoised WooCommerce page IDs kept out of the sitemap. Null until resolved.
	43	+ *
	44	+ * @since 2.0.1
	45	+ * @var int[]|null
	46	+ */
	47	+ private ?array $woocommerce_excluded_page_ids = null;
	48	+
	49	+ /**
	50	+ * Inclusion flag -> the child sitemap it controls.
	51	+ *
	52	+ * Shared by the child-list builder and by the "did the caller change what
	53	+ * the sitemap includes?" check in maybe_promote_to_index().
	54	+ *
	55	+ * @since 1.31.0
	56	+ * @var array<string, string>
	57	+ */
	58	+ private const INCLUSION_CHILD_TYPES = [
	59	+ 'include_posts' => 'posts',
	60	+ 'include_pages' => 'pages',
	61	+ 'include_categories' => 'categories',
	62	+ 'include_tags' => 'tags',
	63	+ ];
	64	+
	65	+ /**
	66	+ * Child sitemap type -> the object that actually supplies its entries.
	67	+ *
	68	+ * A child normally carries its object's own slug, but the sitemap presets UI
	69	+ * writes a display name for the WooCommerce taxonomy child
	70	+ * ('product_categories', not 'product_cat'), so a saved child list can name a
	71	+ * type no post type or taxonomy answers to. Resolving through here lets those
	72	+ * children take the same generic path as every other custom taxonomy instead
	73	+ * of needing a case of their own (#690).
	74	+ *
	75	+ * @since 2.7.0
	76	+ * @var array<string, string>
	77	+ */
	78	+ private const CHILD_TYPE_ALIASES = [
	79	+ 'product_categories' => 'product_cat',
	80	+ ];
	81	+
	82	+ /**
37	83	* Supported sitemap types
38	84	*
39	85	* @since 1.0.0
40	86	* @var array
41	87	*/
	88	+ /**
	89	+ * How many post IDs to hydrate at a time while walking the sitemap set.
	90	+ *
	91	+ * @since 2.0.1
	92	+ * @var int
	93	+ */
	94	+ private const ID_WALK_CHUNK = 500;
	95	+
	96	+ /**
	97	+ * How long a content or settings change is debounced before the sitemap is
	98	+ * rebuilt, in seconds. Coalesces bulk edits into a single regeneration.
	99	+ *
	100	+ * @since 2.2.1
	101	+ * @var int
	102	+ */
	103	+ private const REGENERATION_DEBOUNCE = 30;
	104	+
	105	+ /**
	106	+ * How far past its due time the scheduled rebuild may sit before a request
	107	+ * takes it over, in seconds.
	108	+ *
	109	+ * WP-Cron is request-driven, so on a site running DISABLE_WP_CRON, blocking
	110	+ * loopback requests, or seeing very little traffic the event never fires and
	111	+ * the sitemap silently stops updating (#629). The grace keeps the fast path
	112	+ * (cron) in charge under normal conditions.
	113	+ *
	114	+ * @since 2.2.1
	115	+ * @var int
	116	+ */
	117	+ private const REGENERATION_TAKEOVER_GRACE = 120;
	118	+
	119	+ /**
	120	+ * Longest backoff between takeover attempts after a failed rebuild, so a
	121	+ * persistently failing generation cannot run on every admin request.
	122	+ *
	123	+ * @since 2.2.1
	124	+ * @var int
	125	+ */
	126	+ private const REGENERATION_MAX_BACKOFF = 3600;
	127	+
	128	+ /**
	129	+ * Option holding the rebuild that content/settings changes are still waiting
	130	+ * on: `since`, `source` ('content'|'settings'), `attempts`, `next_attempt`.
	131	+ * Absent means the served sitemap is up to date with what triggered it.
	132	+ *
	133	+ * @since 2.2.1
	134	+ * @var string
	135	+ */
	136	+ public const REGENERATION_PENDING_OPTION = 'thinkrank_sitemap_regeneration_pending';
	137	+
	138	+ /**
	139	+ * Option holding the last automatic-regeneration failure (`message`,
	140	+ * `source`, `time`), so the failure is visible instead of swallowed.
	141	+ *
	142	+ * @since 2.2.1
	143	+ * @var string
	144	+ */
	145	+ public const REGENERATION_ERROR_OPTION = 'thinkrank_sitemap_regeneration_error';
	146	+
	147	+ /**
	148	+ * Delivery modes accepted by the `delivery_mode` setting.
	149	+ *
	150	+ * Mirrors LLMs_Txt_Manager::DELIVERY_MODES, which solved the same problem
	151	+ * for llms.txt. `auto` is the only value a site should normally need.
	152	+ *
	153	+ * @since 2.9.0
	154	+ * @var string[]
	155	+ */
	156	+ public const DELIVERY_MODES = ['auto', 'static', 'dynamic'];
	157	+
	158	+ /**
	159	+ * Cache group for documents rendered on the dynamic path.
	160	+ *
	161	+ * @since 2.9.0
	162	+ * @var string
	163	+ */
	164	+ private const DYNAMIC_CACHE_PREFIX = 'thinkrank_sitemap_doc_';
	165	+
	166	+ /**
	167	+ * How long a dynamically rendered document is cached.
	168	+ *
	169	+ * Invalidated by content and settings changes through
	170	+ * {@see self::flush_dynamic_cache()}, so this is only the backstop for a
	171	+ * change nothing hooked.
	172	+ *
	173	+ * @since 2.9.0
	174	+ * @var int
	175	+ */
	176	+ private const DYNAMIC_CACHE_TTL = 12 * HOUR_IN_SECONDS;
	177	+
	178	+ /**
	179	+ * How long the render lock is held before it is assumed abandoned.
	180	+ *
	181	+ * Long enough for a large site's full build, short enough that a request
	182	+ * killed mid-build does not lock the endpoint out for meaningfully long.
	183	+ *
	184	+ * @since 2.9.0
	185	+ * @var int
	186	+ */
	187	+ private const RENDER_LOCK_TTL = 60;
	188	+
	189	+ /**
	190	+ * Cached stand-in for "this site does not publish that name".
	191	+ *
	192	+ * published_document_names() lists what the configuration *could* produce,
	193	+ * but a child whose type is excluded produces nothing. Without a negative
	194	+ * entry those names miss the cache forever, so every request for one
	195	+ * rebuilt the entire sitemap — the same cost the positive cache exists to
	196	+ * avoid, on a public endpoint (#754 review).
	197	+ *
	198	+ * @since 2.9.0
	199	+ * @var string
	200	+ */
	201	+ private const ABSENT_MARKER = "\0thinkrank-absent";
	202	+
	203	+ /**
	204	+ * How many times, and how long, a losing request waits for the winner.
	205	+ *
	206	+ * Bounded at roughly a second in total: past that, building a second copy
	207	+ * costs less than making a crawler wait.
	208	+ *
	209	+ * @since 2.9.0
	210	+ * @var int
	211	+ */
	212	+ private const RENDER_LOCK_WAIT_ATTEMPTS = 4;
	213	+
	214	+ /**
	215	+ * @since 2.9.0
	216	+ * @var int
	217	+ */
	218	+ private const RENDER_LOCK_WAIT_MICROSECONDS = 250000;
	219	+
	220	+ /**
	221	+ * Where generated documents go instead of disk, when set.
	222	+ *
	223	+ * Every sitemap document this class produces — segments, the index, the
	224	+ * single flat file and local-sitemap.xml — is published through the one
	225	+ * writer, {@see self::save_sitemap_to_file()}. Swapping that writer for a
	226	+ * collector is therefore all it takes to render the same bytes without a
	227	+ * filesystem, which is what dynamic delivery needs (#752). Doing it here
	228	+ * rather than duplicating the build pipeline is deliberate: a second
	229	+ * pipeline would drift from this one, and the index in particular is
	230	+ * assembled from whatever the children actually produced.
	231	+ *
	232	+ * @since 2.9.0
	233	+ * @var callable|null
	234	+ */
	235	+ private $document_sink = null;
	236	+
	237	+ /**
	238	+ * Transient guarding against two generations running at once. Shared with
	239	+ * Sitemap_Endpoint's manual generate route so an automatic rebuild and a
	240	+ * manual one cannot write the same files concurrently.
	241	+ *
	242	+ * @since 2.2.1
	243	+ * @var string
	244	+ */
	245	+ public const GENERATION_LOCK_TRANSIENT = 'thinkrank_sitemap_generation_lock';
	246	+
	247	+ /**
	248	+ * How many term IDs to hydrate at a time while walking a taxonomy.
	249	+ *
	250	+ * @since 2.0.1
	251	+ * @var int
	252	+ */
	253	+ private const TERM_WALK_CHUNK = 1000;
	254	+
	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	+
42	280	private array $sitemap_types = [
43	281	'posts' => [
44	282	'name' => 'Posts',
45	283	'post_types' => ['post'],
		@@ -66,17 +304,29 @@
66	304	]
67	305	];
68	306
69	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 $listener = null;
	315	+
	316	+ /**
70	317	* Constructor
71	318	*
72	319	* @since 1.0.0
	320	+ * @since 2.10.1 Registers no hooks, whatever `$register_hooks` says. The
	321	+ * content-change listeners are registered once, at bootstrap,
	322	+ * by register_content_listeners().
	323	+ *
	324	+ * @param bool $register_hooks Unused since 2.10.1; kept so existing callers,
	325	+ * Pro's included, keep working.
73	326	*/
74		- public function __construct() {
	327	+ public function __construct(bool $register_hooks = true) {
75	328	parent::__construct('sitemap');
76		-
77		- // Initialize auto-generation hooks
78		- $this->init_auto_generation_hooks();
79	329	}
80	330
81	331	/**
82	332	* Filter the args of a sitemap post query.
		@@ -120,34 +370,154 @@
120	370	return (array) apply_filters('thinkrank_sitemap_term_query_args', $args);
121	371	}
122	372
123	373	/**
124		- * Initialize WordPress hooks for auto-generation
	374	+ * Register the content-change listeners that queue an automatic rebuild.
125	375	*
126		- * @since 1.0.0
	376	+ * Called once per request, at plugin bootstrap, next to the WP-Cron
	377	+ * listeners that run the rebuild these queue (both in
	378	+ * Plugin::register_sitemap_cron_listeners(), on plugins_loaded). The
	379	+ * constructor used to register them, so they existed only in a request
	380	+ * that happened to build a generator: the REST endpoint, the setup wizard,
	381	+ * an MCP ability. Block-editor saves go through REST and were heard. A
	382	+ * scheduled post published by WP-Cron, Quick Edit, the classic editor and
	383	+ * WP-CLI were not, so the post stayed out of the sitemap with nothing
	384	+ * pending to recover it (#824). Each instance also added its own set, so
	385	+ * one REST save ran the handlers once per generator built.
	386	+ *
	387	+ * The generator is built on the first change and shared for the rest of
	388	+ * the request, so a bulk edit does not construct one per post.
	389	+ *
	390	+ * @since 2.10.1
127	391	* @return void
128	392	*/
129		- private function init_auto_generation_hooks(): void {
130		- // Content change hooks - use priority 20 to run after other plugins
131		- add_action('save_post', [$this, 'handle_content_change'], 20, 2);
132		- add_action('delete_post', [$this, 'handle_content_deletion'], 20);
133		- add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20);
134		- 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);
135	407
136		- // Taxonomy change hooks
137		- add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3);
138		- add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3);
139		- 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	+ }
140	413
141		- // NOTE: the WP-Cron regeneration listeners (thinkrank_regenerate_sitemap
142		- // and thinkrank_regenerate_sitemap_settings) are registered at plugin
143		- // bootstrap (Plugin::register_sitemap_cron_listeners(), on plugins_loaded)
144		- // rather than here. A cron run never builds this class via the REST
145		- // endpoint (no rest_api_init), so registering them in the constructor
146		- // 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);
147	431	}
148	432
149	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	+ /**
150	520	* Generate XML sitemap
151	521	*
152	522	* @since 1.0.0
153	523	*
		@@ -156,15 +526,10 @@
156	526	*/
157	527	public function generate_sitemap(array $options = []): string {
158	528	$settings = $this->get_settings('site');
159	529
160		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	530	+ $xml = $this->xml_prolog($settings, 'sitemap');
161	531
162		- // Add XSL stylesheet only if styling is enabled
163		- if (!empty($settings['enable_styling'])) {
164		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
165		- }
166		-
167	532	// Add image namespace if images are enabled
168	533	if (!empty($settings['include_images'])) {
169	534	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
170	535	} else {
		@@ -276,8 +641,34 @@
276	641	];
277	642	}
278	643
279	644	/**
	645	+ * Sitemap keys outside the defaults.
	646	+ *
	647	+ * @since 2.0.1
	648	+ *
	649	+ * @return string[]
	650	+ */
	651	+ protected function additional_setting_keys(): array {
	652	+ return ['selected_preset'];
	653	+ }
	654	+
	655	+ /**
	656	+ * Inclusion flags are per post type and per taxonomy.
	657	+ *
	658	+ * A site registering a `product` post type stores `include_product`; an
	659	+ * enumerated list would go stale on the next registration, so the family
	660	+ * is matched instead.
	661	+ *
	662	+ * @since 2.0.1
	663	+ *
	664	+ * @return string[]
	665	+ */
	666	+ protected function dynamic_setting_key_patterns(): array {
	667	+ return ['/^include_[a-z0-9_]+$/', '/^exclude_[a-z0-9_]+$/'];
	668	+ }
	669	+
	670	+ /**
280	671	* Get default settings for a context type (implements interface)
281	672	*
282	673	* @since 1.0.0
283	674	*
		@@ -300,8 +691,15 @@
300	691	]
301	692	],
302	693	'use_sitemap_index' => false,
303	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	+
304	702	// General Settings
305	703	'links_per_sitemap' => 1000,
306	704	'include_images' => true,
307	705	'include_featured_images' => false,
		@@ -323,8 +721,17 @@
323	721	// Advanced Options
324	722	'enable_styling' => true,
325	723	'custom_url_pattern' => 'sitemap-{type}.xml',
326	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	+
327	734	// Generation tracking
328	735	'last_generated' => ''
329	736	];
330	737	}
		@@ -329,8 +736,49 @@
329	736	];
330	737	}
331	738
332	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	+ /**
333	781	* Get settings schema definition (implements interface)
334	782	*
335	783	* @since 1.0.0
336	784	*
		@@ -344,8 +792,15 @@
344	792	'title' => 'Enable Sitemap',
345	793	'description' => 'Generate XML sitemap for search engines',
346	794	'default' => true
347	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	+ ],
348	803	'include_posts' => [
349	804	'type' => 'boolean',
350	805	'title' => 'Include Posts',
351	806	'description' => 'Include blog posts in sitemap',
		@@ -386,8 +841,32 @@
386	841	'type' => 'string',
387	842	'title' => 'Last Generated',
388	843	'description' => 'Timestamp of last sitemap generation',
389	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' => ''
390	869	]
391	870	];
392	871	}
393	872
		@@ -404,9 +883,14 @@
404	883	* @return string XML URL entry
405	884	*/
406	885	private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string {
407	886	$xml = " <url>\n";
408		- $xml .= " <loc>" . esc_url($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";
409	893	// Omit <lastmod> when unknown (empty) — a fabricated timestamp is worse
410	894	// than no timestamp, and an absent lastmod is valid per the spec.
411	895	if (!empty($lastmod)) {
412	896	$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n";
		@@ -416,9 +900,12 @@
416	900
417	901	// Add image entries if provided
418	902	foreach ($images as $image) {
419	903	$xml .= " <image:image>\n";
420		- $xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n";
	904	+ // The sitemap protocol wants an escaped URL, and WordPress hands back
	905	+ // attachment URLs with non-ASCII filenames unencoded (esc_url() does
	906	+ // not encode them either), so write the percent-encoded form (#924).
	907	+ $xml .= " <image:loc>" . esc_url(\ThinkRank\Core\Url_Validator::to_ascii(Url_Scheme::apply((string) $image['url']))) . "</image:loc>\n";
421	908
422	909	if (!empty($image['title'])) {
423	910	$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n";
424	911	}
		@@ -531,9 +1018,21 @@
531	1018	if (empty($all_ids)) {
532	1019	return;
533	1020	}
534	1021
535		- foreach (array_chunk($all_ids, 500) as $chunk) {
	1022	+ // Walk the ID list with a moving window rather than array_chunk().
	1023	+ // array_chunk() builds a second array holding every element again, so
	1024	+ // peak memory was twice the ID list — on a 100k-post site that is ~16MB
	1025	+ // where ~8MB is needed, and this walk is the one part of an otherwise
	1026	+ // well-bounded routine with no ceiling (#402).
	1027	+ $total = count($all_ids);
	1028	+
	1029	+ for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) {
	1030	+ $this->assert_memory_headroom();
	1031	+ $chunk_start = memory_get_usage(true);
	1032	+
	1033	+ $chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK);
	1034	+
536	1035	$posts = get_posts($this->filter_query_args([
537	1036	'post_type' => $post_types,
538	1037	'post_status' => 'publish',
539	1038	'numberposts' => count($chunk),
		@@ -540,22 +1039,51 @@
540	1039	'post__in' => $chunk,
541	1040	'orderby' => 'post__in', // preserve the resolved order
542	1041	]));
543	1042
	1043	+ // get_featured_image() hydrates each featured image under the
	1044	+ // attachment's own ID, which the chunk's post IDs do not reach.
	1045	+ $attachment_ids = [];
	1046	+
544	1047	foreach ($posts as $post) {
545	1048	if ($this->should_include_in_sitemap($post, $settings)) {
546		- $url = get_permalink($post);
	1049	+ /**
	1050	+ * Filter a sitemap entry's permalink.
	1051	+ *
	1052	+ * The multilingual manager uses this to generate each
	1053	+ * translation's URL in its OWN language: the sitemap query
	1054	+ * deliberately runs with suppress_filters, and the cron
	1055	+ * rebuild runs with no language context at all, so a bare
	1056	+ * get_permalink() resolved every translation to the
	1057	+ * default-language URL — N entries sharing one <loc> (#409).
	1058	+ *
	1059	+ * @since 2.0.1
	1060	+ * @param string $url Permalink as WordPress resolved it.
	1061	+ * @param \WP_Post $post Post the entry describes.
	1062	+ */
	1063	+ $url = apply_filters('thinkrank_sitemap_post_permalink', get_permalink($post), $post);
547	1064	$lastmod = gmdate('c', strtotime($post->post_modified_gmt));
548	1065	$priority = $this->calculate_intelligent_priority($post, $post->post_type);
549	1066	$changefreq = $this->calculate_change_frequency($post, $post->post_type);
550	1067	$images = $this->extract_post_images($post, $settings);
551	1068
	1069	+ $thumbnail_id = (int) get_post_thumbnail_id($post);
	1070	+ if ($thumbnail_id > 0) {
	1071	+ $attachment_ids[] = $thumbnail_id;
	1072	+ }
	1073	+
552	1074	yield $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images);
553	1075	}
554	1076	}
555	1077
556		- // Free the hydrated chunk before loading the next one.
	1078	+ // Free the hydrated chunk before loading the next one — including
	1079	+ // the copies get_posts() left in the runtime object cache.
557	1080	unset($posts);
	1081	+ $this->release_walk_memory(
	1082	+ $chunk_start,
	1083	+ $this->chunk_post_cache_groups($post_types),
	1084	+ array_merge($chunk, $attachment_ids)
	1085	+ );
558	1086	}
559	1087	}
560	1088
561	1089	/**
		@@ -611,9 +1139,17 @@
611	1139	if (is_wp_error($all_ids) || empty($all_ids)) {
612	1140	return;
613	1141	}
614	1142
615		- foreach (array_chunk($all_ids, 1000) as $chunk) {
	1143	+ // Same moving window as the post walk above, for the same reason.
	1144	+ $total = count($all_ids);
	1145	+
	1146	+ for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) {
	1147	+ $this->assert_memory_headroom();
	1148	+ $chunk_start = memory_get_usage(true);
	1149	+
	1150	+ $chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK);
	1151	+
616	1152	$terms = get_terms($this->filter_term_query_args([
617	1153	'taxonomy' => $taxonomy,
618	1154	'include' => $chunk,
619	1155	'orderby' => 'include', // preserve the resolved order
		@@ -624,8 +1160,15 @@
624	1160	continue;
625	1161	}
626	1162
627	1163	foreach ($terms as $term) {
	1164	+ // A term whose archive says noindex (its own override or its
	1165	+ // taxonomy's), or that redirects, must not be advertised: the
	1166	+ // sitemap would contradict the page's own signal (#911).
	1167	+ if (!Indexability::is_indexable_term($term)) {
	1168	+ continue;
	1169	+ }
	1170	+
628	1171	$url = get_term_link($term);
629	1172	if (!is_wp_error($url)) {
630	1173	// Omit lastmod for terms — the generation time is not a real
631	1174	// modification time and would mislabel every term as just-changed.
		@@ -633,12 +1176,48 @@
633	1176	}
634	1177	}
635	1178
636	1179	unset($terms);
	1180	+ $this->release_walk_memory($chunk_start, ['terms', 'term_meta'], $chunk);
637	1181	}
638	1182	}
639	1183
640	1184	/**
	1185	+ * The XML declaration, ownership marker and optional stylesheet every
	1186	+ * sitemap document opens with.
	1187	+ *
	1188	+ * The marker is written unconditionally, and that is the point: removal on
	1189	+ * deactivate and uninstall deletes a web-root sitemap only when the file
	1190	+ * says it is ours, and our filenames are the canonical ones another SEO
	1191	+ * plugin writes too (#515). Tying the proof to `enable_styling` — the one
	1192	+ * marker older versions left — would mean a site with styling off either
	1193	+ * kept a shadowing file behind (#510) or had a competitor's deleted.
	1194	+ *
	1195	+ * @since 2.1.1
	1196	+ *
	1197	+ * The stylesheet URL is served by {@see Sitemap_Stylesheet}, not read off
	1198	+ * disk by the web server, because a static file cannot carry the site's own
	1199	+ * logo and colours (#639). It is a fixed URL: the palette is applied per
	1200	+ * request, so changing a brand colour needs no regeneration and shows up on
	1201	+ * sitemaps published long before.
	1202	+ *
	1203	+ * @param array $settings Sitemap settings (read for `enable_styling`).
	1204	+ * @param string $variant Stylesheet variant, `sitemap` or `index`.
	1205	+ * @return string Prolog lines, newline-terminated.
	1206	+ */
	1207	+ private function xml_prolog(array $settings, string $variant): string {
	1208	+ $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1209	+ $xml .= THINKRANK_SITEMAP_MARKER . "\n";
	1210	+
	1211	+ // The stylesheet is presentation only, so it stays opt-in.
	1212	+ if (!empty($settings['enable_styling'])) {
	1213	+ $xml .= '<?xml-stylesheet type="text/xsl" href="' . esc_url(Sitemap_Stylesheet::url($variant)) . '"?>' . "\n";
	1214	+ }
	1215	+
	1216	+ return $xml;
	1217	+ }
	1218	+
	1219	+ /**
641	1220	* Wrap a set of <url> entry strings in a complete <urlset> document.
642	1221	*
643	1222	* @since 1.14.0
644	1223	*
		@@ -647,14 +1226,10 @@
647	1226	* @param bool $with_image_ns Include the image sitemap namespace
648	1227	* @return string Full sitemap XML
649	1228	*/
650	1229	private function wrap_urlset(array $entries, array $settings, bool $with_image_ns): string {
651		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
	1230	+ $xml = $this->xml_prolog($settings, 'sitemap');
652	1231
653		- if (!empty($settings['enable_styling'])) {
654		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap.xsl') . '"?>' . "\n";
655		- }
656		-
657	1232	if ($with_image_ns && !empty($settings['include_images'])) {
658	1233	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
659	1234	} else {
660	1235	$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
		@@ -778,9 +1353,9 @@
778	1353	if (preg_match('/src=["\']([^"\']+)["\']/', $img_tag, $src_match)) {
779	1354	$image_url = $src_match[1];
780	1355
781	1356	// Skip if not a valid URL or external image
782		- if (!filter_var($image_url, FILTER_VALIDATE_URL)) {
	1357	+ if (!\ThinkRank\Core\Url_Validator::is_valid($image_url)) {
783	1358	continue;
784	1359	}
785	1360
786	1361	// Extract title and alt attributes
		@@ -831,9 +1406,16 @@
831	1406	'_builtin' => false
832	1407	], 'names');
833	1408
834	1409	foreach ($custom_taxonomies as $taxonomy) {
835		- if ($this->should_include_taxonomy($taxonomy)) {
	1410	+ if (!$this->should_include_taxonomy($taxonomy)) {
	1411	+ continue;
	1412	+ }
	1413	+
	1414	+ // Same as the post-type walk above: an explicit per-taxonomy flag
	1415	+ // now decides, and an unset flag keeps the previous "included"
	1416	+ // behaviour (#660).
	1417	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $settings)) {
836	1418	$taxonomies[] = $taxonomy;
837	1419	}
838	1420	}
839	1421
		@@ -888,8 +1470,17 @@
888	1470	if (is_wp_error($all_terms)) {
889	1471	return [];
890	1472	}
891	1473
	1474	+ // Drop terms that are not indexable destinations. This path feeds the
	1475	+ // single general sitemap while collect_taxonomy_entries_iter() feeds
	1476	+ // the segmented ones, so both need the filter or the two disagree about
	1477	+ // the same term.
	1478	+ $all_terms = array_values(array_filter(
	1479	+ $all_terms,
	1480	+ static fn($term) => $term instanceof \WP_Term && Indexability::is_indexable_term($term)
	1481	+ ));
	1482	+
892	1483	// Group terms by taxonomy
893	1484	return $this->group_terms_by_taxonomy($all_terms);
894	1485	}
895	1486
		@@ -1038,8 +1629,17 @@
1038	1629	* @param array $settings Sitemap settings
1039	1630	* @return bool Whether to include in sitemap
1040	1631	*/
1041	1632	private function should_include_in_sitemap(\WP_Post $post, array $settings): bool {
	1633	+ // The WooCommerce cart, checkout and account pages are transactional,
	1634	+ // never indexable, and generate_default_robots_rules() already emits a
	1635	+ // Disallow for each of them. Listing them here submitted URLs our own
	1636	+ // robots.txt blocks, which Search Console reports as "Submitted URL
	1637	+ // blocked by robots.txt". Yoast and Rank Math exclude the same three.
	1638	+ if (in_array($post->ID, $this->woocommerce_excluded_page_ids(), true)) {
	1639	+ return false;
	1640	+ }
	1641	+
1042	1642	// Respect user setting for password protected content
1043	1643	if (!empty($post->post_password) && !empty($settings['exclude_password_protected'])) {
1044	1644	return false;
1045	1645	}
		@@ -1060,20 +1660,53 @@
1060	1660	if (!in_array($post->post_status, ['publish', 'private'], true)) {
1061	1661	return false;
1062	1662	}
1063	1663
1064		- // Check if post overrides robots and sets noindex.
1065		- if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) {
1066		- $raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true);
1067		- if (is_string($raw) && $raw !== '') {
1068		- $robots = json_decode($raw, true);
1069		- if (is_array($robots) && !empty($robots['noindex'])) {
1070		- return false;
	1664	+ // A URL whose page says noindex, or that ThinkRank redirects, is not a
	1665	+ // destination. Only the per-post noindex used to be read here, so a
	1666	+ // post type set to No-index still had every item listed, and redirected
	1667	+ // posts were submitted as "Page with redirect" (#911). The password
	1668	+ // check stays with the setting above: listing protected posts is a
	1669	+ // choice this sitemap has always offered.
	1670	+ if (Indexability::is_post_noindexed($post) || Indexability::is_post_redirected($post)) {
	1671	+ return false;
	1672	+ }
	1673	+
	1674	+ return true;
	1675	+ }
	1676	+
	1677	+ /**
	1678	+ * WooCommerce pages that must never reach the sitemap.
	1679	+ *
	1680	+ * Resolved through wc_get_page_id() so a store that moved or renamed its
	1681	+ * cart/checkout/account pages is still matched. Returns an empty list when
	1682	+ * WooCommerce is not active. Memoised — should_include_in_sitemap() runs
	1683	+ * once per post.
	1684	+ *
	1685	+ * @since 2.0.1
	1686	+ *
	1687	+ * @return int[] Page IDs to exclude.
	1688	+ */
	1689	+ private function woocommerce_excluded_page_ids(): array {
	1690	+ if ($this->woocommerce_excluded_page_ids !== null) {
	1691	+ return $this->woocommerce_excluded_page_ids;
	1692	+ }
	1693	+
	1694	+ $ids = [];
	1695	+
	1696	+ if (function_exists('wc_get_page_id')) {
	1697	+ foreach (['cart', 'checkout', 'myaccount'] as $page) {
	1698	+ $id = (int) wc_get_page_id($page);
	1699	+ // wc_get_page_id() returns -1 when the page is not configured.
	1700	+ if ($id > 0) {
	1701	+ $ids[] = $id;
1071	1702	}
1072	1703	}
1073	1704	}
1074	1705
1075		- return true;
	1706	+ $this->woocommerce_excluded_page_ids = $ids;
	1707	+
	1708	+ return $ids;
1076	1709	}
1077	1710
1078	1711	/**
1079	1712	* Count total URLs in sitemap
		@@ -1239,9 +1872,15 @@
1239	1872	if (!empty($settings['include_pages'])) {
1240	1873	$post_types[] = 'page';
1241	1874	}
1242	1875
1243		- // Auto-detect public custom post types that should be included
	1876	+ // Auto-detect public custom post types that should be included.
	1877	+ //
	1878	+ // A custom type's `include_<slug>` / `exclude_<slug>` flag is honoured
	1879	+ // here (#660). It was previously stored — additional_setting_keys()
	1880	+ // has always let those keys through — but never read, so a CPT was in
	1881	+ // the sitemap whatever the setting said. Unset still means included, so
	1882	+ // a site that never touched the flag is unaffected.
1244	1883	$custom_post_types = get_post_types([
1245	1884	'public' => true,
1246	1885	'_builtin' => false
1247	1886	], 'names');
		@@ -1246,9 +1885,13 @@
1246	1885	'_builtin' => false
1247	1886	], 'names');
1248	1887
1249	1888	foreach ($custom_post_types as $post_type) {
1250		- if ($this->should_include_post_type($post_type)) {
	1889	+ if (!$this->should_include_post_type($post_type)) {
	1890	+ continue;
	1891	+ }
	1892	+
	1893	+ if (\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $post_type, $settings)) {
1251	1894	$post_types[] = $post_type;
1252	1895	}
1253	1896	}
1254	1897
		@@ -1269,18 +1912,11 @@
1269	1912	// Templately's internal `templately_library` store — which are template
1270	1913	// records, not standalone indexable URLs. BetterDocs `docs` and
1271	1914	// WooCommerce `product` register exclude_from_search => false, so they
1272	1915	// remain included.
1273		- if (!is_post_type_viewable($post_type)) {
1274		- return false;
1275		- }
1276		-
1277		- $post_type_obj = get_post_type_object($post_type);
1278		- if (!$post_type_obj || !empty($post_type_obj->exclude_from_search)) {
1279		- return false;
1280		- }
1281		-
1282		- return true;
	1916	+ // The predicate lives in Content_Type_Settings so the matrix can ask
	1917	+ // the same question before offering a switch for this post type.
	1918	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_post_type($post_type);
1283	1919	}
1284	1920
1285	1921	/**
1286	1922	* Check if taxonomy should trigger regeneration
		@@ -1289,11 +1925,13 @@
1289	1925	* @param string $taxonomy Taxonomy slug
1290	1926	* @return bool True if taxonomy should trigger regeneration
1291	1927	*/
1292	1928	private function should_include_taxonomy(string $taxonomy): bool {
1293		- // Include all public taxonomies (presets control which sitemaps are created)
1294		- $taxonomy_obj = get_taxonomy($taxonomy);
1295		- return $taxonomy_obj && $taxonomy_obj->public;
	1929	+ // Public taxonomies only, and only the ones this generator can actually
	1930	+ // emit — the same predicate the content-type matrix asks before it
	1931	+ // offers a sitemap switch for one (presets still control which
	1932	+ // sitemaps are created).
	1933	+ return \ThinkRank\SEO\Content_Type_Settings::sitemap_accepts_taxonomy($taxonomy);
1296	1934	}
1297	1935
1298	1936	/**
1299	1937	* Schedule debounced sitemap regeneration
		@@ -1301,16 +1939,679 @@
1301	1939	* @since 1.0.0
1302	1940	* @return void
1303	1941	*/
1304	1942	private function schedule_debounced_regeneration(): void {
1305		- // Clear any existing scheduled regeneration
1306		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap');
	1943	+ $this->mark_regeneration_pending('content');
	1944	+ $this->debounce_event('thinkrank_regenerate_sitemap');
	1945	+ }
1307	1946
1308		- // Schedule regeneration in 30 seconds to debounce rapid changes
1309		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap');
	1947	+ /**
	1948	+ * Schedule (or keep) the debounced single event behind a regeneration hook.
	1949	+ *
	1950	+ * An event that is already due is left alone. WP-Cron only runs when a
	1951	+ * request arrives, so on a site with DISABLE_WP_CRON, a blocked loopback or
	1952	+ * little traffic an overdue event can sit in the queue for a long time —
	1953	+ * clearing and re-scheduling it on every save pushed the rebuild
	1954	+ * permanently 30 seconds into the future and the sitemap never updated
	1955	+ * (#629). Debouncing only against an event that has not come due yet keeps
	1956	+ * the bulk-edit coalescing without starving the rebuild.
	1957	+ *
	1958	+ * @since 2.2.1
	1959	+ * @param string $hook Regeneration hook to debounce.
	1960	+ * @return void
	1961	+ */
	1962	+ private function debounce_event(string $hook): void {
	1963	+ $next = wp_next_scheduled($hook);
	1964	+
	1965	+ if ($next !== false) {
	1966	+ if ($next <= time()) {
	1967	+ return;
	1968	+ }
	1969	+
	1970	+ wp_clear_scheduled_hook($hook);
	1971	+ }
	1972	+
	1973	+ wp_schedule_single_event(time() + self::REGENERATION_DEBOUNCE, $hook);
1310	1974	}
1311	1975
1312	1976	/**
	1977	+ * Record that a rebuild is outstanding, so an overdue one can be taken over
	1978	+ * by a later request and its staleness surfaced in the UI.
	1979	+ *
	1980	+ * `since` is the *oldest* outstanding change: it is what the takeover grace
	1981	+ * and the admin staleness warning are measured from, so successive edits
	1982	+ * must not push it forward. A settings change outranks a content change —
	1983	+ * it rebuilds regardless of the auto_generate toggle and handles a sitemap
	1984	+ * that has just been disabled — so once one is outstanding it stays the
	1985	+ * recorded source until the rebuild lands.
	1986	+ *
	1987	+ * @since 2.2.1
	1988	+ * @param string $source Either 'content' or 'settings'.
	1989	+ * @return void
	1990	+ */
	1991	+ private function mark_regeneration_pending(string $source): void {
	1992	+ // Whatever made the static files stale made the rendered ones stale
	1993	+ // too. Invalidating here rather than only on the rebuild keeps the two
	1994	+ // delivery modes reacting to exactly the same triggers, which is the
	1995	+ // only way a dynamic site stays as fresh as a static one (#752).
	1996	+ $this->flush_dynamic_cache();
	1997	+
	1998	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	1999	+ $pending = is_array($pending) ? $pending : [];
	2000	+
	2001	+ $since = !empty($pending['since']) ? (int) $pending['since'] : time();
	2002	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	2003	+ $source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	2004	+
	2005	+ // Merged so the bookkeeping a rebuild keeps on the marker (`started`,
	2006	+ // `memory_limit`) survives an edit made while it is outstanding.
	2007	+ update_option(
	2008	+ self::REGENERATION_PENDING_OPTION,
	2009	+ array_merge($pending, [
	2010	+ 'since' => $since,
	2011	+ 'source' => $source,
	2012	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 0,
	2013	+ 'next_attempt' => !empty($pending['next_attempt'])
	2014	+ ? (int) $pending['next_attempt']
	2015	+ : time() + self::REGENERATION_TAKEOVER_GRACE,
	2016	+ // Bumped on every change so a rebuild can tell whether the edit
	2017	+ // it started for is still the newest one outstanding.
	2018	+ 'revision' => (!empty($pending['revision']) ? (int) $pending['revision'] : 0) + 1,
	2019	+ ]),
	2020	+ true
	2021	+ );
	2022	+ }
	2023	+
	2024	+ /**
	2025	+ * The revision of the outstanding rebuild, for
	2026	+ * {@see mark_regeneration_complete()} to compare against once it is done.
	2027	+ *
	2028	+ * @since 2.2.1
	2029	+ * @return int Current revision, 0 when nothing is outstanding.
	2030	+ */
	2031	+ private function current_regeneration_revision(): int {
	2032	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2033	+
	2034	+ return (is_array($pending) && !empty($pending['revision'])) ? (int) $pending['revision'] : 0;
	2035	+ }
	2036	+
	2037	+ /**
	2038	+ * Clear the outstanding-rebuild marker and any recorded failure.
	2039	+ *
	2040	+ * Public because a manual generation satisfies whatever the automatic path
	2041	+ * was still waiting to write.
	2042	+ *
	2043	+ * @since 2.2.1
	2044	+ * @return void
	2045	+ */
	2046	+ public function mark_regeneration_complete(?int $revision = null): void {
	2047	+ // The write succeeded, so whatever failure was on record is history.
	2048	+ if (get_option(self::REGENERATION_ERROR_OPTION, null) !== null) {
	2049	+ delete_option(self::REGENERATION_ERROR_OPTION);
	2050	+ }
	2051	+
	2052	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	2053	+
	2054	+ if ($pending === null) {
	2055	+ return;
	2056	+ }
	2057	+
	2058	+ // A change that landed while this rebuild was running is not covered by
	2059	+ // the files it just wrote, so it has to stay outstanding — otherwise, on
	2060	+ // a site where WP-Cron never fires, clearing the marker would strand it
	2061	+ // exactly the way #629 stranded everything.
	2062	+ if (
	2063	+ $revision !== null
	2064	+ && is_array($pending)
	2065	+ && (int) ($pending['revision'] ?? 0) !== $revision
	2066	+ ) {
	2067	+ // This attempt succeeded, so drop what it claimed: the newer change
	2068	+ // waits the grace a fresh edit gets, not a failure backoff it never
	2069	+ // earned. Not zero: that edit queued its own debounced event, and
	2070	+ // a marker due at once had the next admin request rebuild in its
	2071	+ // shutdown and the event rebuild again seconds later.
	2072	+ unset($pending['started'], $pending['memory_limit']);
	2073	+ $pending['attempts'] = 0;
	2074	+ $pending['next_attempt'] = time() + self::REGENERATION_TAKEOVER_GRACE;
	2075	+
	2076	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
	2077	+ return;
	2078	+ }
	2079	+
	2080	+ delete_option(self::REGENERATION_PENDING_OPTION);
	2081	+ }
	2082	+
	2083	+ /**
	2084	+ * Record the attempt that is about to run before it runs.
	2085	+ *
	2086	+ * A PHP fatal — the memory limit or max_execution_time — is not a
	2087	+ * Throwable, so no catch or finally around the generation runs when the
	2088	+ * process dies, and a failure recorded afterwards was never recorded at
	2089	+ * all: `attempts` stayed 0, the backoff never applied, and the next request
	2090	+ * started the same doomed rebuild again (a fatal every few minutes for as
	2091	+ * long as an admin was logged in). Claiming the attempt up front makes the
	2092	+ * backoff hold even when nothing after this line gets to run, and leaves a
	2093	+ * `started` stamp the next attempt can recognise as an interrupted one.
	2094	+ *
	2095	+ * @since 2.10.1
	2096	+ * @return void
	2097	+ */
	2098	+ private function claim_regeneration_attempt(): void {
	2099	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, null);
	2100	+
	2101	+ // Nothing outstanding (e.g. a manual generation already satisfied it):
	2102	+ // there is no marker to retry from, so nothing to claim.
	2103	+ if (!is_array($pending) || empty($pending['since'])) {
	2104	+ return;
	2105	+ }
	2106	+
	2107	+ if (!empty($pending['started'])) {
	2108	+ // The previous attempt claimed itself and never reported back.
	2109	+ update_option(
	2110	+ self::REGENERATION_ERROR_OPTION,
	2111	+ [
	2112	+ 'message' => __('The previous automatic sitemap rebuild stopped before it finished, most likely because PHP ran out of memory or time. It is retried with a growing delay. If this keeps happening, raise the PHP memory_limit or max_execution_time, or run WP-Cron from a system cron.', 'thinkrank'),
	2113	+ 'source' => isset($pending['source']) ? (string) $pending['source'] : 'content',
	2114	+ 'attempts' => !empty($pending['attempts']) ? (int) $pending['attempts'] : 1,
	2115	+ 'time' => (int) $pending['started'],
	2116	+ ],
	2117	+ false
	2118	+ );
	2119	+ }
	2120	+
	2121	+ $attempts = (!empty($pending['attempts']) ? (int) $pending['attempts'] : 0) + 1;
	2122	+
	2123	+ $pending['attempts'] = $attempts;
	2124	+ $pending['next_attempt'] = time() + $this->regeneration_backoff($attempts);
	2125	+ $pending['started'] = time();
	2126	+
	2127	+ update_option(self::REGENERATION_PENDING_OPTION, $pending, true);
	2128	+ }
	2129	+
	2130	+ /**
	2131	+ * Delay before the next takeover after the given number of attempts.
	2132	+ *
	2133	+ * @since 2.10.1
	2134	+ * @param int $attempts Attempts made so far (1 or more).
	2135	+ * @return int Seconds.
	2136	+ */
	2137	+ private function regeneration_backoff(int $attempts): int {
	2138	+ return (int) min(
	2139	+ self::REGENERATION_TAKEOVER_GRACE * (2 ** min(max($attempts, 1), 10)),
	2140	+ self::REGENERATION_MAX_BACKOFF
	2141	+ );
	2142	+ }
	2143	+
	2144	+ /**
	2145	+ * Record a failed regeneration instead of discarding it.
	2146	+ *
	2147	+ * Keeps the pending marker in place so the rebuild is retried, but backs the
	2148	+ * next attempt off exponentially (capped) so a persistently failing
	2149	+ * generation cannot run on every admin request.
	2150	+ *
	2151	+ * @since 2.2.1
	2152	+ * @since 2.10.1 Accepts the memory limit a rebuild had to stop short of, and
	2153	+ * does not count an attempt claim_regeneration_attempt()
	2154	+ * already counted.
	2155	+ * @param string $message Failure detail.
	2156	+ * @param string $source Either 'content' or 'settings'.
	2157	+ * @param int|null $memory_limit Memory limit (bytes) the rebuild stopped
	2158	+ * short of, when that was the failure.
	2159	+ * @return void
	2160	+ */
	2161	+ private function record_regeneration_failure(string $message, string $source, ?int $memory_limit = null): void {
	2162	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2163	+ $pending = is_array($pending) ? $pending : [];
	2164	+ $attempts = !empty($pending['attempts']) ? (int) $pending['attempts'] : 0;
	2165	+
	2166	+ // An attempt that claimed itself up front has already been counted.
	2167	+ if (empty($pending['started'])) {
	2168	+ $attempts++;
	2169	+ }
	2170	+ $attempts = max($attempts, 1);
	2171	+
	2172	+ $backoff = $this->regeneration_backoff($attempts);
	2173	+
	2174	+ // Same precedence mark_regeneration_pending() enforces: a settings
	2175	+ // rebuild outranks a content one and must not be downgraded by a failed
	2176	+ // attempt. Overwriting it routed the retry back through the content
	2177	+ // path, where should_auto_generate() can be false and the completion
	2178	+ // marker then discards the settings rebuild entirely. Only the
	2179	+ // outstanding rebuild is upgraded — the recorded error keeps reporting
	2180	+ // whichever attempt actually failed.
	2181	+ $current = isset($pending['source']) ? (string) $pending['source'] : '';
	2182	+ $pending_source = ($current === 'settings' || $source === 'settings') ? 'settings' : 'content';
	2183	+
	2184	+ $marker = [
	2185	+ 'since' => !empty($pending['since']) ? (int) $pending['since'] : time(),
	2186	+ 'source' => $pending_source,
	2187	+ 'attempts' => $attempts,
	2188	+ 'next_attempt' => time() + $backoff,
	2189	+ 'revision' => !empty($pending['revision']) ? (int) $pending['revision'] : 0,
	2190	+ ];
	2191	+
	2192	+ // Remembered so has_memory_for_retry() can keep requests with no more
	2193	+ // memory than this from repeating the same attempt.
	2194	+ if ($memory_limit !== null && $memory_limit > 0) {
	2195	+ $marker['memory_limit'] = $memory_limit;
	2196	+ }
	2197	+
	2198	+ update_option(self::REGENERATION_PENDING_OPTION, $marker, true);
	2199	+
	2200	+ update_option(
	2201	+ self::REGENERATION_ERROR_OPTION,
	2202	+ [
	2203	+ 'message' => $message,
	2204	+ 'source' => $source,
	2205	+ 'attempts' => $attempts,
	2206	+ 'time' => time(),
	2207	+ ],
	2208	+ false
	2209	+ );
	2210	+
	2211	+ if (defined('WP_DEBUG') && WP_DEBUG) {
	2212	+ // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
	2213	+ error_log(sprintf('ThinkRank: sitemap %s regeneration failed — %s', $source, $message));
	2214	+ }
	2215	+ }
	2216	+
	2217	+ /**
	2218	+ * Is a rebuild outstanding and past the point where WP-Cron should have run
	2219	+ * it?
	2220	+ *
	2221	+ * Deliberately cheap — one autoloaded option read — because it is consulted
	2222	+ * on every admin request to decide whether the takeover is needed.
	2223	+ *
	2224	+ * @since 2.2.1
	2225	+ * @return bool True when a request should rebuild the sitemap itself.
	2226	+ */
	2227	+ public static function has_overdue_regeneration(): bool {
	2228	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2229	+
	2230	+ if (!is_array($pending) || empty($pending['since'])) {
	2231	+ return false;
	2232	+ }
	2233	+
	2234	+ $due = !empty($pending['next_attempt'])
	2235	+ ? (int) $pending['next_attempt']
	2236	+ : (int) $pending['since'] + self::REGENERATION_TAKEOVER_GRACE;
	2237	+
	2238	+ return time() >= $due;
	2239	+ }
	2240	+
	2241	+ /**
	2242	+ * Rebuild the sitemap in-request when WP-Cron has not delivered.
	2243	+ *
	2244	+ * Hooked on `shutdown` for admin, REST and CLI requests only (see
	2245	+ * Plugin::register_sitemap_cron_listeners()), so the work happens after the
	2246	+ * response has been sent and never adds latency to a visitor page view.
	2247	+ *
	2248	+ * @since 2.2.1
	2249	+ * @return void
	2250	+ */
	2251	+ public function run_overdue_regeneration(): void {
	2252	+ if (!self::has_overdue_regeneration()) {
	2253	+ return;
	2254	+ }
	2255	+
	2256	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2257	+ $pending = is_array($pending) ? $pending : [];
	2258	+ $source = isset($pending['source']) ? (string) $pending['source'] : 'content';
	2259	+
	2260	+ // Cron is running and its own event for this rebuild is still queued
	2261	+ // and due: it is about to do exactly this work. Only then — an event
	2262	+ // that has already been consumed (e.g. it fired while another process
	2263	+ // held the generation lock) is never re-queued, and returning here
	2264	+ // unconditionally left the rebuild to requests that could not finish it.
	2265	+ if (wp_doing_cron()) {
	2266	+ $hook = $source === 'settings' ? 'thinkrank_regenerate_sitemap_settings' : 'thinkrank_regenerate_sitemap';
	2267	+ $next = wp_next_scheduled($hook);
	2268	+
	2269	+ if ($next !== false && $next <= time()) {
	2270	+ return;
	2271	+ }
	2272	+ }
	2273	+
	2274	+ // The last attempt had to stop short of this process's memory limit.
	2275	+ // Retrying at the same (or a lower) limit only repeats that, so leave
	2276	+ // the rebuild to a process with more room — WP-CLI, a system cron, or a
	2277	+ // host with a higher limit — instead of burning it on every request.
	2278	+ if (!$this->has_memory_for_retry($pending)) {
	2279	+ return;
	2280	+ }
	2281	+
	2282	+ if ($source === 'settings') {
	2283	+ $this->regenerate_sitemap_from_settings();
	2284	+ return;
	2285	+ }
	2286	+
	2287	+ $this->auto_regenerate_sitemap();
	2288	+ }
	2289	+
	2290	+ /**
	2291	+ * Acquire the shared generation lock.
	2292	+ *
	2293	+ * @since 2.2.1
	2294	+ * @return bool True when this process may generate.
	2295	+ */
	2296	+ private function acquire_generation_lock(): bool {
	2297	+ if (get_transient(self::GENERATION_LOCK_TRANSIENT)) {
	2298	+ return false;
	2299	+ }
	2300	+
	2301	+ set_transient(self::GENERATION_LOCK_TRANSIENT, time(), 5 * MINUTE_IN_SECONDS);
	2302	+
	2303	+ return true;
	2304	+ }
	2305	+
	2306	+ /**
	2307	+ * Release the shared generation lock.
	2308	+ *
	2309	+ * @since 2.2.1
	2310	+ * @return void
	2311	+ */
	2312	+ private function release_generation_lock(): void {
	2313	+ delete_transient(self::GENERATION_LOCK_TRANSIENT);
	2314	+ }
	2315	+
	2316	+ /**
	2317	+ * This process's PHP memory limit in bytes.
	2318	+ *
	2319	+ * @since 2.10.1
	2320	+ * @return int Bytes, or -1 when unlimited (or unreadable).
	2321	+ */
	2322	+ private function current_memory_limit(): int {
	2323	+ $limit = (string) ini_get('memory_limit');
	2324	+
	2325	+ if ($limit === '' || $limit === '-1') {
	2326	+ return -1;
	2327	+ }
	2328	+
	2329	+ $bytes = (int) wp_convert_hr_to_bytes($limit);
	2330	+
	2331	+ return $bytes > 0 ? $bytes : -1;
	2332	+ }
	2333	+
	2334	+ /**
	2335	+ * May this process retry a rebuild that last stopped at the memory limit?
	2336	+ *
	2337	+ * Raises the limit the way wp-admin does first, so a request that can get
	2338	+ * more room than the failed attempt had is still allowed to try.
	2339	+ *
	2340	+ * @since 2.10.1
	2341	+ * @param array $pending The pending marker.
	2342	+ * @return bool True when there is no recorded memory failure, or this
	2343	+ * process has more memory than the attempt that failed.
	2344	+ */
	2345	+ private function has_memory_for_retry(array $pending): bool {
	2346	+ if (empty($pending['memory_limit'])) {
	2347	+ return true;
	2348	+ }
	2349	+
	2350	+ wp_raise_memory_limit('admin');
	2351	+
	2352	+ $limit = $this->current_memory_limit();
	2353	+
	2354	+ return $limit === -1 || $limit > (int) $pending['memory_limit'];
	2355	+ }
	2356	+
	2357	+ /**
	2358	+ * Release what one walked chunk left behind.
	2359	+ *
	2360	+ * Hydrating a chunk through get_posts()/get_terms() also stores every
	2361	+ * object and its meta in the in-process object cache, which nothing
	2362	+ * empties until the request ends. Unsetting the chunk therefore freed
	2363	+ * nothing, and the walk grew with the size of the site instead of the size
	2364	+ * of a chunk — about 1.3 GB on a 45k-post site.
	2365	+ *
	2366	+ * A persistent object cache that supports it drops only its in-process
	2367	+ * copy (`flush_runtime`); the shared store keeps its data. WordPress's
	2368	+ * default cache has no shared store, and flushing it would empty every
	2369	+ * group for the rest of the request (options, the queried object, other
	2370	+ * plugins' data), so there only the chunk's own entries are deleted. A
	2371	+ * persistent cache without `flush_runtime` is left alone: deleting from it
	2372	+ * would evict the objects for every other request too.
	2373	+ *
	2374	+ * @since 2.10.1
	2375	+ * @param int $chunk_start memory_get_usage(true) before the chunk was hydrated.
	2376	+ * @param array $groups Cache groups keyed by the chunk's object IDs.
	2377	+ * @param int[] $ids The chunk's object IDs, plus any objects it
	2378	+ * hydrated under their own (featured images).
	2379	+ * @return void
	2380	+ * @throws \Error See assert_memory_headroom().
	2381	+ */
	2382	+ private function release_walk_memory(int $chunk_start, array $groups, array $ids): void {
	2383	+ // What one chunk costs before it is released: the margin the next one
	2384	+ // needs. Measured in the same real allocated size assert_memory_headroom()
	2385	+ // compares against the limit, so the two are the same unit.
	2386	+ $this->walk_chunk_cost = max($this->walk_chunk_cost, memory_get_usage(true) - $chunk_start);
	2387	+
	2388	+ if (wp_using_ext_object_cache()) {
	2389	+ if (
	2390	+ function_exists('wp_cache_supports')
	2391	+ && wp_cache_supports('flush_runtime')
	2392	+ && function_exists('wp_cache_flush_runtime')
	2393	+ ) {
	2394	+ wp_cache_flush_runtime();
	2395	+ }
	2396	+ } elseif (!empty($ids)) {
	2397	+ foreach ($groups as $group) {
	2398	+ wp_cache_delete_multiple($ids, $group);
	2399	+ }
	2400	+ }
	2401	+
	2402	+ $this->assert_memory_headroom();
	2403	+ }
	2404	+
	2405	+ /**
	2406	+ * Cache groups get_posts() fills per post for the given post types.
	2407	+ *
	2408	+ * The post, its meta, and one relationships group per taxonomy the post
	2409	+ * type uses (update_object_term_cache()). Term objects themselves are
	2410	+ * bounded by the number of terms, not posts, so they are left cached.
	2411	+ *
	2412	+ * @since 2.10.1
	2413	+ * @param string[] $post_types Post types being walked.
	2414	+ * @return string[] Cache groups keyed by post ID.
	2415	+ */
	2416	+ private function chunk_post_cache_groups(array $post_types): array {
	2417	+ $groups = ['posts', 'post_meta'];
	2418	+
	2419	+ foreach (get_object_taxonomies($post_types) as $taxonomy) {
	2420	+ $groups[] = $taxonomy . '_relationships';
	2421	+ }
	2422	+
	2423	+ return array_values(array_unique($groups));
	2424	+ }
	2425	+
	2426	+ /**
	2427	+ * Stop an automatic rebuild before the memory limit rather than at it.
	2428	+ *
	2429	+ * A PHP memory fatal skips every catch and finally, so the lock, the
	2430	+ * failure record and the backoff are all lost with it, while stopping here
	2431	+ * is an ordinary, fully recorded failure. Checked before each chunk is
	2432	+ * hydrated, against a margin of at least the largest chunk seen so far.
	2433	+ *
	2434	+ * It throws an \Error, not an \Exception, on purpose: the per-segment
	2435	+ * catch (\Exception) blocks in generate_multiple_sitemaps() would otherwise
	2436	+ * swallow it and carry on — writing an index without the aborted segments
	2437	+ * and then pruning their files as orphans. Only the automatic rebuild's
	2438	+ * catch (\Throwable) is meant to see it.
	2439	+ *
	2440	+ * @since 2.10.1
	2441	+ * @return void
	2442	+ * @throws \Error When the automatic rebuild is close to the memory limit.
	2443	+ */
	2444	+ private function assert_memory_headroom(): void {
	2445	+ if (!$this->memory_guard) {
	2446	+ return;
	2447	+ }
	2448	+
	2449	+ $limit = $this->current_memory_limit();
	2450	+ if ($limit === -1) {
	2451	+ return;
	2452	+ }
	2453	+
	2454	+ // A fifth of the limit (at least 32 MB) for writing the files, or one
	2455	+ // and a half of the costliest chunk if that is more — and never more
	2456	+ // than half the limit either way. Without that outer cap a single
	2457	+ // anomalously expensive chunk (500 posts of serialised page-builder or
	2458	+ // ACF meta reaches hundreds of megabytes) puts the margin above the
	2459	+ // limit itself, so every later check aborts at any usage at all, the
	2460	+ // failure records this process's limit, and has_memory_for_retry()
	2461	+ // then refuses every process that has the same limit. A site that
	2462	+ // never actually ran out of memory would stop rebuilding until WP-CLI
	2463	+ // or a system cron happened to run.
	2464	+ $headroom = (int) min(
	2465	+ max(
	2466	+ min(max($limit * 0.2, 32 * MB_IN_BYTES), $limit * 0.5),
	2467	+ $this->walk_chunk_cost * 1.5
	2468	+ ),
	2469	+ $limit * 0.5
	2470	+ );
	2471	+
	2472	+ // The real allocated size, which is what PHP enforces memory_limit
	2473	+ // against; memory_get_usage(false) reports only what is handed out of
	2474	+ // those allocations and so understates the margin by the allocator's
	2475	+ // slack.
	2476	+ $usage = memory_get_usage(true);
	2477	+
	2478	+ if ($usage > $limit - $headroom) {
	2479	+ throw new \Error(
	2480	+ sprintf(
	2481	+ /* translators: 1: memory in use, 2: PHP memory limit. */
	2482	+ esc_html__('The sitemap rebuild was stopped at %1$s of the %2$s PHP memory limit, before PHP would have run out of memory. It will be retried by a process with more memory (WP-CLI or a system cron). To let it finish in the admin, raise the PHP memory_limit.', 'thinkrank'),
	2483	+ esc_html((string) size_format($usage)),
	2484	+ esc_html((string) size_format($limit))
	2485	+ ),
	2486	+ // phpcs:ignore WordPress.Security.EscapeOutput.ExceptionNotEscaped -- an integer class constant, not output.
	2487	+ self::MEMORY_ABORT_CODE
	2488	+ );
	2489	+ }
	2490	+ }
	2491	+
	2492	+ /**
	2493	+ * Run an automatic rebuild's generation with the fatal-safe bookkeeping.
	2494	+ *
	2495	+ * @since 2.10.1
	2496	+ * @param array $settings Sitemap settings.
	2497	+ * @return bool Whatever generate_and_save() returned.
	2498	+ * @throws \Throwable Whatever generation throws, after the memory guard is
	2499	+ * switched back off.
	2500	+ */
	2501	+ private function generate_for_regeneration(array $settings): bool {
	2502	+ // Same headroom wp-admin gives itself; a no-op when the limit is
	2503	+ // already higher or unlimited.
	2504	+ wp_raise_memory_limit('admin');
	2505	+
	2506	+ $this->claim_regeneration_attempt();
	2507	+ $this->memory_guard = true;
	2508	+ $this->walk_chunk_cost = 0;
	2509	+
	2510	+ try {
	2511	+ return $this->generate_and_save($settings);
	2512	+ } finally {
	2513	+ $this->memory_guard = false;
	2514	+ }
	2515	+ }
	2516	+
	2517	+ /**
	2518	+ * Record a failure thrown by an automatic rebuild.
	2519	+ *
	2520	+ * @since 2.10.1
	2521	+ * @param \Throwable $e What was thrown.
	2522	+ * @param string $source Either 'content' or 'settings'.
	2523	+ * @return void
	2524	+ */
	2525	+ private function record_thrown_regeneration_failure(\Throwable $e, string $source): void {
	2526	+ $memory_limit = null;
	2527	+
	2528	+ if ($e instanceof \Error && $e->getCode() === self::MEMORY_ABORT_CODE) {
	2529	+ $memory_limit = $this->current_memory_limit();
	2530	+ $memory_limit = $memory_limit > 0 ? $memory_limit : null;
	2531	+ }
	2532	+
	2533	+ $this->record_regeneration_failure($e->getMessage(), $source, $memory_limit);
	2534	+ }
	2535	+
	2536	+ /**
	2537	+ * Report how automatic regeneration is faring, for the admin UI.
	2538	+ *
	2539	+ * The feature used to fail invisibly: `last_generated` simply stopped
	2540	+ * advancing and nothing drew attention to it (#629).
	2541	+ *
	2542	+ * @since 2.2.1
	2543	+ * @return array Health payload.
	2544	+ */
	2545	+ public function get_regeneration_health(): array {
	2546	+ $settings = $this->get_settings('site');
	2547	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2548	+ $pending = is_array($pending) ? $pending : [];
	2549	+ $error = get_option(self::REGENERATION_ERROR_OPTION, []);
	2550	+ $error = is_array($error) ? $error : [];
	2551	+
	2552	+ $since = !empty($pending['since']) ? (int) $pending['since'] : 0;
	2553	+
	2554	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap');
	2555	+ if ($next_scheduled === false) {
	2556	+ $next_scheduled = wp_next_scheduled('thinkrank_regenerate_sitemap_settings');
	2557	+ }
	2558	+
	2559	+ return [
	2560	+ 'auto_generate' => !empty($settings['auto_generate']),
	2561	+ 'last_generated' => $settings['last_generated'] ?? '',
	2562	+ 'pending_since' => $since ? gmdate('c', $since) : null,
	2563	+ 'pending_seconds' => $since ? max(0, time() - $since) : 0,
	2564	+ 'next_scheduled' => $next_scheduled ? gmdate('c', (int) $next_scheduled) : null,
	2565	+ 'cron_disabled' => defined('DISABLE_WP_CRON') && DISABLE_WP_CRON,
	2566	+ 'stale' => $this->is_sitemap_stale($settings),
	2567	+ 'last_error' => !empty($error['message'])
	2568	+ ? [
	2569	+ 'message' => (string) $error['message'],
	2570	+ 'source' => isset($error['source']) ? (string) $error['source'] : 'content',
	2571	+ 'time' => !empty($error['time']) ? gmdate('c', (int) $error['time']) : null,
	2572	+ ]
	2573	+ : null,
	2574	+ ];
	2575	+ }
	2576	+
	2577	+ /**
	2578	+ * Has published content changed since the served sitemap was last written?
	2579	+ *
	2580	+ * Uses core's cached last-modified lookup, which only considers published
	2581	+ * posts — the same content the sitemap covers.
	2582	+ *
	2583	+ * @since 2.2.1
	2584	+ * @param array $settings Sitemap settings.
	2585	+ * @return bool True when the sitemap is behind the content.
	2586	+ */
	2587	+ private function is_sitemap_stale(array $settings): bool {
	2588	+ if (empty($settings['enabled']) || empty($settings['last_generated'])) {
	2589	+ // Never generated is already reported separately by the UI.
	2590	+ return false;
	2591	+ }
	2592	+
	2593	+ $generated = strtotime((string) $settings['last_generated']);
	2594	+ if (!$generated) {
	2595	+ return false;
	2596	+ }
	2597	+
	2598	+ $modified = get_lastpostmodified('gmt');
	2599	+ if (!$modified) {
	2600	+ return false;
	2601	+ }
	2602	+
	2603	+ $modified = strtotime($modified . ' UTC');
	2604	+ if (!$modified) {
	2605	+ return false;
	2606	+ }
	2607	+
	2608	+ // A minute of slack keeps a rebuild that ran alongside the edit from
	2609	+ // reporting itself as stale.
	2610	+ return $modified > ($generated + MINUTE_IN_SECONDS);
	2611	+ }
	2612	+
	2613	+ /**
1313	2614	* Public entry point to debounce-rebuild the sitemap after a settings change
1314	2615	* (e.g. toggling inclusion rules via REST or the MCP ability), so the served
1315	2616	* file reflects the new settings instead of going stale until a content edit.
1316	2617	*
		@@ -1319,10 +2620,10 @@
1319	2620	public function schedule_regeneration(): void {
1320	2621	// Debounce against rapid successive saves, but use the settings-specific
1321	2622	// hook so the rebuild runs regardless of the auto_generate toggle (which
1322	2623	// only governs content-change-triggered regeneration).
1323		- wp_clear_scheduled_hook('thinkrank_regenerate_sitemap_settings');
1324		- wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap_settings');
	2624	+ $this->mark_regeneration_pending('settings');
	2625	+ $this->debounce_event('thinkrank_regenerate_sitemap_settings');
1325	2626	}
1326	2627
1327	2628	/**
1328	2629	* Rebuild the served sitemap after an explicit settings change.
		@@ -1331,11 +2632,22 @@
1331	2632	* auto_generate setting: the user deliberately changed inclusion rules and
1332	2633	* expects the served file to reflect them even if content-triggered
1333	2634	* auto-generation is turned off. Still respects the master `enabled` flag.
1334	2635	*
1335		- * @return void
	2636	+ * @since 2.10.0 Reports whether the served sitemap was actually rebuilt, so
	2637	+ * a caller can say so rather than assume it (#764). Existing
	2638	+ * callers that ignore the return are unaffected.
	2639	+ *
	2640	+ * @return bool True when the served sitemap now reflects the settings.
1336	2641	*/
1337		- public function regenerate_sitemap_from_settings(): void {
	2642	+ public function regenerate_sitemap_from_settings(): bool {
	2643	+ if (!$this->acquire_generation_lock()) {
	2644	+ // A manual generation (or another request's takeover) is already
	2645	+ // writing the files; the pending marker survives so this rebuild is
	2646	+ // retried rather than lost.
	2647	+ return false;
	2648	+ }
	2649	+
1338	2650	try {
1339	2651	$settings = $this->get_settings('site');
1340	2652	if (empty($settings['enabled'])) {
1341	2653	// The sitemap was disabled: remove the previously generated static
		@@ -1340,64 +2652,155 @@
1340	2652	if (empty($settings['enabled'])) {
1341	2653	// The sitemap was disabled: remove the previously generated static
1342	2654	// files so the web server stops serving a stale sitemap that
1343	2655	// crawlers would otherwise keep fetching.
1344		- $this->delete_published_sitemaps();
1345		- return;
	2656	+ //
	2657	+ // A file that could not be removed is still being served, so
	2658	+ // this is not a success. Reporting one here would tell a caller
	2659	+ // the sitemap was gone while the web server kept answering with
	2660	+ // it, which is the failure this return value exists to prevent
	2661	+ // (#764).
	2662	+ $removal = $this->delete_published_sitemaps($settings);
	2663	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2664	+
	2665	+ if (!empty($stuck)) {
	2666	+ $this->record_regeneration_failure(
	2667	+ $this->stuck_files_message($stuck, true),
	2668	+ 'settings'
	2669	+ );
	2670	+
	2671	+ return false;
	2672	+ }
	2673	+
	2674	+ $this->mark_regeneration_complete();
	2675	+
	2676	+ return true;
1346	2677	}
1347		- $this->generate_and_save($settings);
	2678	+
	2679	+ $revision = $this->current_regeneration_revision();
	2680	+
	2681	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2682	+ // Returns false when a static file is stuck in the web root:
	2683	+ // the server keeps serving that file in preference to WordPress,
	2684	+ // so the switch has not taken effect (#764).
	2685	+ return $this->switch_to_dynamic_delivery($settings, $revision, 'settings');
	2686	+ }
	2687	+
	2688	+ if ($this->generate_for_regeneration($settings)) {
	2689	+ $this->mark_regeneration_complete($revision);
	2690	+
	2691	+ return true;
	2692	+ }
	2693	+
	2694	+ $this->record_regeneration_failure(
	2695	+ $this->write_failure_message(),
	2696	+ 'settings'
	2697	+ );
	2698	+
	2699	+ return false;
1348	2700	} catch (\Throwable $e) {
1349		- // Settings-triggered regeneration failed - details in exception.
	2701	+ $this->record_thrown_regeneration_failure($e, 'settings');
	2702	+
	2703	+ return false;
	2704	+ } finally {
	2705	+ $this->release_generation_lock();
1350	2706	}
1351	2707	}
1352	2708
1353	2709	/**
	2710	+ * When a rebuild has been outstanding since, or 0 when none is.
	2711	+ *
	2712	+ * Lets a caller report an honest "saved, but the served file has not caught
	2713	+ * up yet" instead of a bare success (#764).
	2714	+ *
	2715	+ * @since 2.10.0
	2716	+ * @return int Unix timestamp, or 0 when nothing is pending.
	2717	+ */
	2718	+ public static function regeneration_pending_since(): int {
	2719	+ $pending = get_option(self::REGENERATION_PENDING_OPTION, []);
	2720	+
	2721	+ if (!is_array($pending) || empty($pending['since'])) {
	2722	+ return 0;
	2723	+ }
	2724	+
	2725	+ return (int) $pending['since'];
	2726	+ }
	2727	+
	2728	+ /**
1354	2729	* Remove every static sitemap file ThinkRank publishes to the web root.
1355	2730	*
1356		- * Called when the sitemap feature is disabled so /sitemap.xml,
1357		- * /sitemap_index.xml, the segmented children (incl. paginated -N pages), and
1358		- * /local-sitemap.xml stop being served. Only ThinkRank's own filenames are
1359		- * targeted; WordPress core's wp-sitemap.xml is left untouched.
	2731	+ * Called when the sitemap feature is disabled, by the cleanup route, and by
	2732	+ * both removal paths, so /sitemap.xml, /sitemap_index.xml, the segmented
	2733	+ * children (incl. paginated -N pages), and /local-sitemap.xml stop being
	2734	+ * served. Only ThinkRank's own filenames are targeted; WordPress core's
	2735	+ * wp-sitemap.xml and any other plugin's sitemap in the web root are left
	2736	+ * untouched.
1360	2737	*
1361		- * @return void
	2738	+ * @since 1.31.0 Returns the filenames removed, and accepts the settings to
	2739	+ * derive them from, so a caller that already read them (and
	2740	+ * needs to report what went) does not have to re-read or
	2741	+ * re-derive the name list.
	2742	+ * @since 2.1.0 Delegates to thinkrank_webroot_delete_sitemaps(). Uninstall
	2743	+ * needs the same removal but has no autoloader to reach this
	2744	+ * class, so the logic moved to includes/cleanup-webroot.php
	2745	+ * and this stays as the in-plugin entry point.
	2746	+ *
	2747	+ * @param array|null $settings Optional. Sitemap settings; defaults to the
	2748	+ * saved site settings.
	2749	+ * @return array{deleted: string[], failed: string[]} Basenames removed, and
	2750	+ * those that existed but could not be removed.
1362	2751	*/
1363		- public function delete_published_sitemaps(): void {
1364		- $settings = $this->get_settings('site');
	2752	+ public function delete_published_sitemaps(?array $settings = null): array {
	2753	+ return thinkrank_webroot_delete_sitemaps($settings ?? $this->get_settings('site'));
	2754	+ }
1365	2755
1366		- // Base filenames to remove. Derive the child/index names from the stored
1367		- // sitemap_urls so a custom custom_url_pattern (e.g. seo-{type}.xml) is
1368		- // honored, not just the default sitemap-*.xml naming. Include the current
1369		- // primary + the default names as a safety net.
1370		- $names = [
1371		- 'sitemap.xml',
1372		- 'sitemap_index.xml',
1373		- 'local-sitemap.xml',
1374		- $this->get_primary_sitemap_filename($settings),
1375		- ];
	2756	+ /**
	2757	+ * Every child-sitemap filename this site could have published.
	2758	+ *
	2759	+ * @since 1.31.0
	2760	+ * @since 2.1.0 Delegates to thinkrank_webroot_segment_filenames().
	2761	+ *
	2762	+ * @param array $settings Sitemap settings (read for `custom_url_pattern`).
	2763	+ * @return string[] Basenames, e.g. ['sitemap-posts.xml', 'sitemap-pages.xml'].
	2764	+ */
	2765	+ private function publishable_segment_filenames(array $settings): array {
	2766	+ return thinkrank_webroot_segment_filenames($settings);
	2767	+ }
1376	2768
1377		- foreach ((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : []) as $config) {
1378		- if (empty($config['url'])) {
1379		- continue;
1380		- }
1381		- $name = basename((string) wp_parse_url($config['url'], PHP_URL_PATH));
1382		- if ($name !== '') {
1383		- $names[] = $name;
1384		- }
1385		- }
1386		-
1387		- foreach (array_unique(array_filter($names)) as $name) {
1388		- // Remove the file itself and any paginated -N variants of its stem
1389		- // (e.g. seo-posts.xml plus seo-posts-2.xml, seo-posts-3.xml…).
1390		- $path = ABSPATH . $name;
1391		- if (file_exists($path)) {
1392		- wp_delete_file($path);
1393		- }
1394		- if (preg_match('/^(.*)\.xml$/i', $name, $m)) {
1395		- foreach (glob(ABSPATH . $m[1] . '-*.xml') ?: [] as $paged) {
1396		- wp_delete_file($paged);
1397		- }
1398		- }
1399		- }
	2769	+ /**
	2770	+ * Can this web-root file be shown to be a sitemap ThinkRank wrote?
	2771	+ *
	2772	+ * The generator deletes as often as cleanup does — a segment that dropped
	2773	+ * out of the set, a pagination page beyond the new count, the local sitemap
	2774	+ * after the business identity was cleared — and until 2.1.1 it did all
	2775	+ * three by filename alone. That is the #515 bug on a far more frequent
	2776	+ * trigger: our names are the canonical ones, so an ordinary regeneration
	2777	+ * (post save, term change, settings save) destroyed RankMath's
	2778	+ * `sitemap-tags.xml` and `local-sitemap.xml` with no deactivation involved.
	2779	+ *
	2780	+ * Every name the generator derives comes from the current settings — the
	2781	+ * url pattern, the configured `sitemap_urls`, `local-sitemap.xml` — but the
	2782	+ * ownership test is still asked with `$name_derived = false`, which switches
	2783	+ * off the legacy fallback for the whole generator side.
	2784	+ *
	2785	+ * The fallback exists to recover a pre-2.1.1 file written with
	2786	+ * `enable_styling` off, which carries neither marker. That recovery belongs
	2787	+ * to the once-off cleanup paths. Here it can only do harm: this method runs
	2788	+ * on every post save, and everything this version writes carries
	2789	+ * THINKRANK_SITEMAP_MARKER, so after the site's first regeneration an
	2790	+ * unmarked file at one of our names is by definition somebody else's — and
	2791	+ * deleting it on an ordinary regeneration is #515 through the more common
	2792	+ * door. The cost is a stale unmarked segment left on disk until deactivation
	2793	+ * picks it up, which is the safe direction to fail in.
	2794	+ *
	2795	+ * @since 2.1.1
	2796	+ *
	2797	+ * @param string $path Absolute path to a file in the web root.
	2798	+ * @param array $settings Sitemap settings.
	2799	+ * @return bool True when the file may be deleted.
	2800	+ */
	2801	+ private function webroot_sitemap_is_ours(string $path, array $settings): bool {
	2802	+ return thinkrank_webroot_sitemap_is_ours($path, $settings, false);
1400	2803	}
1401	2804
1402	2805	/**
1403	2806	* Auto-regenerate sitemap (called by scheduled action)
		@@ -1405,20 +2808,48 @@
1405	2808	* @since 1.0.0
1406	2809	* @return void
1407	2810	*/
1408	2811	public function auto_regenerate_sitemap(): void {
	2812	+ if (!$this->acquire_generation_lock()) {
	2813	+ // A manual generation (or another request's takeover) is already
	2814	+ // writing the files; the pending marker survives so this rebuild is
	2815	+ // retried rather than lost.
	2816	+ return;
	2817	+ }
	2818	+
1409	2819	try {
1410	2820	// Double-check that auto-generation is still enabled
1411	2821	if (!$this->should_auto_generate()) {
	2822	+ // Nothing outstanding can be delivered while the feature is off,
	2823	+ // so drop the marker rather than let the takeover retry forever.
	2824	+ $this->mark_regeneration_complete();
1412	2825	return;
1413	2826	}
1414	2827
1415		- $this->generate_and_save($this->get_settings('site'));
	2828	+ $revision = $this->current_regeneration_revision();
	2829	+ $settings = $this->get_settings('site');
1416	2830
1417		- // Sitemap auto-regenerated successfully
	2831	+ // See regenerate_sitemap_from_settings(): nothing to write.
	2832	+ if ('dynamic' === $this->resolve_delivery_mode($settings)) {
	2833	+ $this->switch_to_dynamic_delivery($settings, $revision, 'content');
	2834	+ return;
	2835	+ }
1418	2836
	2837	+ if ($this->generate_for_regeneration($settings)) {
	2838	+ $this->mark_regeneration_complete($revision);
	2839	+ } else {
	2840	+ // Previously this returned quietly and last_generated simply
	2841	+ // stopped advancing, leaving the site owner with no way to learn
	2842	+ // the sitemap had stopped updating (#629).
	2843	+ $this->record_regeneration_failure(
	2844	+ $this->write_failure_message(),
	2845	+ 'content'
	2846	+ );
	2847	+ }
1419	2848	} catch (\Throwable $e) {
1420		- // Sitemap auto-regeneration failed - error details available in exception
	2849	+ $this->record_thrown_regeneration_failure($e, 'content');
	2850	+ } finally {
	2851	+ $this->release_generation_lock();
1421	2852	}
1422	2853	}
1423	2854
1424	2855	/**
		@@ -1433,8 +2864,26 @@
1433	2864	* @param array $settings Sitemap settings.
1434	2865	* @return bool True when the sitemap files were written.
1435	2866	*/
1436	2867	public function generate_and_save(array $settings): bool {
	2868	+ // Dynamic delivery publishes no files, so writing them here would put a
	2869	+ // static copy back in the web root for the server to serve in place of
	2870	+ // the dynamic route. Guarding at each call site left gaps — the
	2871	+ // snapshot migrator's post-import regeneration had none — so the rule
	2872	+ // lives with the writing instead.
	2873	+ //
	2874	+ // `is_collecting()` is the exception that makes dynamic delivery work
	2875	+ // at all: render_document() and collect_documents() reach this same
	2876	+ // method with the writer swapped for a collector, and that is precisely
	2877	+ // the dynamic build. Only a real write is skipped.
	2878	+ if (!$this->is_collecting() && 'dynamic' === $this->resolve_delivery_mode($settings)) {
	2879	+ // Whatever prompted this call changed the sitemap's content, so the
	2880	+ // rendered copies must not outlive it.
	2881	+ $this->flush_dynamic_cache();
	2882	+
	2883	+ return true;
	2884	+ }
	2885	+
1437	2886	// Index mode is driven by the use_sitemap_index toggle (not merely by how
1438	2887	// many sitemap_urls happen to be configured). When the toggle is on but
1439	2888	// no child sitemaps are set up yet, synthesize the per-type segmented set
1440	2889	// so we emit a real <sitemapindex> with paginated children instead of a
		@@ -1445,16 +2894,29 @@
1445	2894	$results = $this->generate_multiple_sitemaps($settings);
1446	2895	$written = !empty($results['success']);
1447	2896	} else {
1448	2897	$xml = $this->generate_sitemap($settings);
1449		- $written = $this->save_sitemap_to_file($xml, $this->get_primary_sitemap_filename($settings));
	2898	+ $primary = $this->get_primary_sitemap_filename($settings);
	2899	+ $written = $this->save_sitemap_to_file($xml, $primary);
1450	2900
1451	2901	// Local business sitemap is a standalone file, regenerated on the
1452	2902	// single-sitemap path too (this is the default mode).
1453	2903	$this->regenerate_local_sitemap($settings);
	2904	+
	2905	+ // Switching out of index mode leaves sitemap_index.xml and every
	2906	+ // child on disk, still served and never refreshed again. The index
	2907	+ // path already prunes what it no longer owns; this path never did,
	2908	+ // so the site kept serving two sitemap trees (#563). Ownership is
	2909	+ // still tested per file, so another plugin's sitemap at one of our
	2910	+ // names is never touched (#515).
	2911	+ $this->prune_orphaned_segments($settings, [['filename' => basename($primary)]]);
1454	2912	}
1455	2913
1456		- if ($written) {
	2914	+ // last_generated describes what is on disk. A dynamic render publishes
	2915	+ // nothing, so advancing it would report a static publication that never
	2916	+ // happened and would let primary_sitemap_file_exists() callers believe
	2917	+ // there is a file to serve.
	2918	+ if ($written && !$this->is_collecting()) {
1457	2919	$settings['last_generated'] = gmdate('c');
1458	2920	$this->save_settings('site', null, $settings);
1459	2921	}
1460	2922
		@@ -1461,8 +2923,433 @@
1461	2923	return $written;
1462	2924	}
1463	2925
1464	2926	/**
	2927	+ * Whether this instance is rendering documents rather than publishing them.
	2928	+ *
	2929	+ * @since 2.9.0
	2930	+ *
	2931	+ * @return bool
	2932	+ */
	2933	+ private function is_collecting(): bool {
	2934	+ return $this->document_sink !== null;
	2935	+ }
	2936	+
	2937	+ /**
	2938	+ * Complete a regeneration that delivers dynamically, retiring stale files.
	2939	+ *
	2940	+ * Dynamic delivery renders nothing to disk, but that is only half the job.
	2941	+ * A web server hands back an existing `/sitemap.xml` without ever loading
	2942	+ * WordPress, so any file left over from a previous static generation goes on
	2943	+ * being served forever and {@see \ThinkRank\Frontend\SEO_Manager
	2944	+ * ::maybe_serve_sitemap()} is never reached. Switching to dynamic while
	2945	+ * leaving those files in place would therefore appear to do nothing at all.
	2946	+ *
	2947	+ * Both transitions matter and they differ:
	2948	+ *
	2949	+ * - An explicit switch to `dynamic` happens on a site whose root is usually
	2950	+ * still writable, so the files can simply be removed.
	2951	+ * - An `auto` site that becomes read-only cannot remove them, because
	2952	+ * deleting an entry needs write permission on the directory that holds
	2953	+ * it. There the stale sitemap really is stuck in front of us, and the
	2954	+ * honest outcome is a recorded failure naming it rather than a rebuild
	2955	+ * reported as complete (#754 review).
	2956	+ *
	2957	+ * Ownership is tested per file by the shared helper, so another plugin's
	2958	+ * sitemap at one of our names is never deleted (#515).
	2959	+ *
	2960	+ * @since 2.9.0
	2961	+ *
	2962	+ * @param array $settings Sitemap settings.
	2963	+ * @param int $revision Revision this rebuild is completing.
	2964	+ * @param string $source 'settings' or 'content', for the failure record.
	2965	+ * @return void
	2966	+ */
	2967	+ private function switch_to_dynamic_delivery(array $settings, int $revision, string $source): bool {
	2968	+ $this->flush_dynamic_cache();
	2969	+
	2970	+ $removal = $this->delete_published_sitemaps($settings);
	2971	+ $stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : [];
	2972	+
	2973	+ if (!empty($stuck)) {
	2974	+ $this->record_regeneration_failure($this->stuck_files_message($stuck), $source);
	2975	+
	2976	+ return false;
	2977	+ }
	2978	+
	2979	+ $this->mark_regeneration_complete($revision);
	2980	+
	2981	+ return true;
	2982	+ }
	2983	+
	2984	+ /**
	2985	+ * Why a stale file left in the web root means the change has not landed.
	2986	+ *
	2987	+ * Shared by every path that removes published files, so they cannot
	2988	+ * describe the same situation differently (#764).
	2989	+ *
	2990	+ * The two situations that reach it differ in what WordPress is doing, and
	2991	+ * the message has to say which. After a switch to dynamic delivery
	2992	+ * WordPress IS serving the sitemap and the files shadow it. After the
	2993	+ * sitemap is switched off WordPress serves nothing, so the one message
	2994	+ * used to tell a site owner who had just disabled the sitemap that it was
	2995	+ * "being served from WordPress", which is the opposite of what they did.
	2996	+ *
	2997	+ * @since 2.10.0
	2998	+ * @since 2.10.0 Public, so the REST endpoint uses it rather than a copy;
	2999	+ * takes $sitemap_disabled for the disabled path.
	3000	+ *
	3001	+ * @param string[] $stuck Basenames that could not be removed.
	3002	+ * @param bool $sitemap_disabled True when the files outlived disabling
	3003	+ * the sitemap rather than a switch to
	3004	+ * dynamic delivery.
	3005	+ * @return string
	3006	+ */
	3007	+ public function stuck_files_message(array $stuck, bool $sitemap_disabled = false): string {
	3008	+ if ($sitemap_disabled) {
	3009	+ return sprintf(
	3010	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3011	+ __('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'),
	3012	+ implode(', ', $stuck),
	3013	+ untrailingslashit(ABSPATH)
	3014	+ );
	3015	+ }
	3016	+
	3017	+ return sprintf(
	3018	+ /* translators: 1: comma-separated file names, 2: absolute path to the WordPress root. */
	3019	+ __('The sitemap is being served from WordPress, but these files are still in the site root and your web server will keep serving them instead: %1$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'),
	3020	+ implode(', ', $stuck),
	3021	+ untrailingslashit(ABSPATH)
	3022	+ );
	3023	+ }
	3024	+
	3025	+ /**
	3026	+ * What to tell the site owner when publishing the files failed.
	3027	+ *
	3028	+ * The old wording stated the symptom and stopped there, so the reported
	3029	+ * cause was a guess and this reached support as a plugin fault rather than
	3030	+ * a folder permission (#752, #753). When the root is demonstrably
	3031	+ * unwritable, say that, and say what to do about it.
	3032	+ *
	3033	+ * @since 2.9.0
	3034	+ *
	3035	+ * @return string
	3036	+ */
	3037	+ private function write_failure_message(): string {
	3038	+ if (!wp_is_writable(ABSPATH)) {
	3039	+ return sprintf(
	3040	+ /* translators: %s: absolute path to the WordPress root. */
	3041	+ __('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'),
	3042	+ untrailingslashit(ABSPATH)
	3043	+ );
	3044	+ }
	3045	+
	3046	+ return __('The sitemap files could not be written to the site root.', 'thinkrank');
	3047	+ }
	3048	+
	3049	+ /**
	3050	+ * How this site delivers its sitemap.
	3051	+ *
	3052	+ * `auto` is resolved on whether the web root can be written. That is the
	3053	+ * right signal here (unlike llms.txt, where the question is whether the
	3054	+ * server applies the .htaccess charset block): a site whose root is
	3055	+ * read-only cannot publish a sitemap file at all, and before this existed
	3056	+ * the feature simply failed with "The sitemap files could not be written to
	3057	+ * the site root." and served nothing (#752).
	3058	+ *
	3059	+ * @since 2.9.0
	3060	+ *
	3061	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3062	+ * @return string One of 'static' or 'dynamic'. Never 'auto'.
	3063	+ */
	3064	+ public function resolve_delivery_mode(?array $settings = null): string {
	3065	+ $settings = $settings ?? $this->get_settings('site');
	3066	+ $mode = (string) ($settings['delivery_mode'] ?? 'auto');
	3067	+
	3068	+ if ('static' === $mode || 'dynamic' === $mode) {
	3069	+ return $mode;
	3070	+ }
	3071	+
	3072	+ return wp_is_writable(ABSPATH) ? 'static' : 'dynamic';
	3073	+ }
	3074	+
	3075	+ /**
	3076	+ * Render one published sitemap document without touching the filesystem.
	3077	+ *
	3078	+ * Runs the ordinary build pipeline with the writer swapped for a collector,
	3079	+ * so the bytes returned here are the bytes the static path would have
	3080	+ * written. `SitemapDeliveryParityTest` asserts that equivalence rather than
	3081	+ * trusting it.
	3082	+ *
	3083	+ * The whole set is built to answer for one file, because the index can only
	3084	+ * be assembled from the children that were actually produced. The result is
	3085	+ * cached per document, so that cost is paid once per change and not once
	3086	+ * per crawler request.
	3087	+ *
	3088	+ * @since 2.9.0
	3089	+ *
	3090	+ * @param string $filename Published file name, e.g. 'sitemap.xml'.
	3091	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3092	+ * @return string|null XML, or null when this site does not publish that name.
	3093	+ */
	3094	+ public function render_document(string $filename, ?array $settings = null): ?string {
	3095	+ $filename = basename($filename);
	3096	+ $settings = $settings ?? $this->get_settings('site');
	3097	+
	3098	+ if (empty($settings['enabled'])) {
	3099	+ return null;
	3100	+ }
	3101	+
	3102	+ $cached = get_transient($this->dynamic_cache_key($filename));
	3103	+ if (self::ABSENT_MARKER === $cached) {
	3104	+ return null;
	3105	+ }
	3106	+ if (is_string($cached) && '' !== $cached) {
	3107	+ return $cached;
	3108	+ }
	3109	+
	3110	+ // A miss builds the whole set, because the index can only be assembled
	3111	+ // from the children that were actually produced. Caching only the
	3112	+ // requested document therefore made a crawler walking the index and its
	3113	+ // children rebuild the entire site's sitemap once per file — every post
	3114	+ // and taxonomy query repeated N times on a public endpoint (#754
	3115	+ // review). The set is built once and stored in full.
	3116	+ return $this->stream_documents($settings, $filename);
	3117	+ }
	3118	+
	3119	+ /**
	3120	+ * Build every document, caching each as it is produced, keeping one.
	3121	+ *
	3122	+ * A miss has to build the whole set, because the index can only be
	3123	+ * assembled from the children that were actually produced. It does not have
	3124	+ * to *hold* the whole set: the static path never keeps more than one page
	3125	+ * in memory, writing each to disk as it goes, and buffering every
	3126	+ * document's XML to return one of them undid that on the request path,
	3127	+ * where a large site's entire sitemap corpus would sit in a single PHP
	3128	+ * process (#754 review).
	3129	+ *
	3130	+ * So the sink writes each document straight to its cache entry and lets it
	3131	+ * go, retaining only the one this request is answering. Peak retention is
	3132	+ * one document, whatever the site's size.
	3133	+ *
	3134	+ * Concurrency: the first request through takes a short lock and does the
	3135	+ * work. One that finds the lock held waits a bounded moment for the winner
	3136	+ * to publish, then builds anyway, because serving a correct sitemap late
	3137	+ * beats serving none.
	3138	+ *
	3139	+ * @since 2.9.0
	3140	+ *
	3141	+ * @param array $settings Sitemap settings.
	3142	+ * @param string $wanted Document this request is answering.
	3143	+ * @return string|null XML for $wanted, or null when the site does not publish it.
	3144	+ */
	3145	+ private function stream_documents(array $settings, string $wanted): ?string {
	3146	+ $lock = self::DYNAMIC_CACHE_PREFIX . 'lock';
	3147	+
	3148	+ if (!$this->acquire_render_lock($lock)) {
	3149	+ for ($attempt = 0; $attempt < self::RENDER_LOCK_WAIT_ATTEMPTS; $attempt++) {
	3150	+ usleep(self::RENDER_LOCK_WAIT_MICROSECONDS);
	3151	+
	3152	+ $cached = get_transient($this->dynamic_cache_key($wanted));
	3153	+ if (self::ABSENT_MARKER === $cached) {
	3154	+ return null;
	3155	+ }
	3156	+ if (is_string($cached) && '' !== $cached) {
	3157	+ return $cached;
	3158	+ }
	3159	+ }
	3160	+ }
	3161	+
	3162	+ $kept = null;
	3163	+ // Names only. Keeping the bodies here would be the very retention this
	3164	+ // method exists to avoid.
	3165	+ $produced = [];
	3166	+
	3167	+ $previous = $this->document_sink;
	3168	+ $this->document_sink = function (string $name, string $xml) use (&$kept, &$produced, $wanted): void {
	3169	+ $produced[$name] = true;
	3170	+ set_transient($this->dynamic_cache_key($name), $xml, self::DYNAMIC_CACHE_TTL);
	3171	+
	3172	+ if ($name === $wanted) {
	3173	+ $kept = $xml;
	3174	+ }
	3175	+ };
	3176	+
	3177	+ try {
	3178	+ $this->generate_and_save($settings);
	3179	+
	3180	+ // Names the configuration lists but this build did not produce get
	3181	+ // a negative entry, so asking for one again is a cache hit rather
	3182	+ // than another full rebuild.
	3183	+ $absent = $this->published_document_names($settings);
	3184	+
	3185	+ // Also the exact name this request asked for: a paginated page past
	3186	+ // the end of a stem is a legitimate request shape that the base
	3187	+ // list cannot enumerate, and without an entry it would rebuild on
	3188	+ // every hit.
	3189	+ $absent[] = $wanted;
	3190	+
	3191	+ foreach (array_unique($absent) as $name) {
	3192	+ if (!isset($produced[$name])) {
	3193	+ set_transient($this->dynamic_cache_key($name), self::ABSENT_MARKER, self::DYNAMIC_CACHE_TTL);
	3194	+ }
	3195	+ }
	3196	+ } finally {
	3197	+ $this->document_sink = $previous;
	3198	+ delete_transient($lock);
	3199	+ }
	3200	+
	3201	+ return $kept;
	3202	+ }
	3203	+
	3204	+ /**
	3205	+ * Take the render lock, if it is free.
	3206	+ *
	3207	+ * Not atomic across processes, and deliberately so: the fallback for losing
	3208	+ * a race is duplicated work, never a wrong or missing sitemap, so a
	3209	+ * heavier primitive would buy nothing here.
	3210	+ *
	3211	+ * @since 2.9.0
	3212	+ *
	3213	+ * @param string $lock Lock transient name.
	3214	+ * @return bool True when this request holds the lock.
	3215	+ */
	3216	+ private function acquire_render_lock(string $lock): bool {
	3217	+ if (false !== get_transient($lock)) {
	3218	+ return false;
	3219	+ }
	3220	+
	3221	+ set_transient($lock, time(), self::RENDER_LOCK_TTL);
	3222	+
	3223	+ return true;
	3224	+ }
	3225	+
	3226	+ /**
	3227	+ * Build every document this site publishes and return them all.
	3228	+ *
	3229	+ * Verification and tooling only. This retains the whole set in memory, so
	3230	+ * it must never be used to answer a request: {@see self::stream_documents()}
	3231	+ * is the serving path and keeps one document at a time regardless of site
	3232	+ * size (#754 review). `SitemapDeliveryParityTest` enforces that separation
	3233	+ * by failing if the request path routes back through here.
	3234	+ *
	3235	+ * @since 2.9.0
	3236	+ *
	3237	+ * @param array $settings Sitemap settings.
	3238	+ * @return array<string,string> Filename => XML.
	3239	+ */
	3240	+ public function collect_documents(array $settings): array {
	3241	+ $documents = [];
	3242	+
	3243	+ $previous = $this->document_sink;
	3244	+ $this->document_sink = static function (string $name, string $xml) use (&$documents): void {
	3245	+ $documents[$name] = $xml;
	3246	+ };
	3247	+
	3248	+ try {
	3249	+ $this->generate_and_save($settings);
	3250	+ } finally {
	3251	+ $this->document_sink = $previous;
	3252	+ }
	3253	+
	3254	+ return $documents;
	3255	+ }
	3256	+
	3257	+ /**
	3258	+ * The file names this site publishes, without building their contents.
	3259	+ *
	3260	+ * Used by the request router to decide whether a URL is ours before doing
	3261	+ * any work. Cheap: it reads the configured child list rather than querying
	3262	+ * for entries.
	3263	+ *
	3264	+ * @since 2.9.0
	3265	+ *
	3266	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3267	+ * @return string[] File names, including paginated pages that may exist.
	3268	+ */
	3269	+ public function published_document_names(?array $settings = null): array {
	3270	+ $settings = $settings ?? $this->get_settings('site');
	3271	+ $resolved = $this->maybe_promote_to_index($settings);
	3272	+
	3273	+ $names = [$this->get_primary_sitemap_filename($settings), 'local-sitemap.xml'];
	3274	+
	3275	+ foreach ((array) ($resolved['sitemap_urls'] ?? []) as $child) {
	3276	+ if (!is_array($child) || empty($child['enabled'])) {
	3277	+ continue;
	3278	+ }
	3279	+
	3280	+ $path = (string) wp_parse_url((string) ($child['url'] ?? ''), PHP_URL_PATH);
	3281	+ if ('' !== $path) {
	3282	+ $names[] = basename($path);
	3283	+ }
	3284	+ }
	3285	+
	3286	+ return array_values(array_unique(array_filter($names)));
	3287	+ }
	3288	+
	3289	+ /**
	3290	+ * Does this site publish a document under that name?
	3291	+ *
	3292	+ * Not a plain membership test against {@see self::published_document_names()}:
	3293	+ * that lists the configured children, and a child over the per-file URL cap
	3294	+ * is split into `<stem>-2.xml`, `<stem>-3.xml` and so on, with every page
	3295	+ * listed in the index. Gating the request router on the base list alone
	3296	+ * therefore 404'd exactly the pages the index points at, which is worse than
	3297	+ * not serving them at all.
	3298	+ *
	3299	+ * Page counts are not knowable without building, so the stem is what is
	3300	+ * matched; a page that does not exist is answered by the build finding
	3301	+ * nothing for it, and is then cached as absent.
	3302	+ *
	3303	+ * @since 2.9.0
	3304	+ *
	3305	+ * @param string $name Requested file name.
	3306	+ * @param array|null $settings Sitemap settings (falls back to saved ones).
	3307	+ * @return bool
	3308	+ */
	3309	+ public function publishes_document_name(string $name, ?array $settings = null): bool {
	3310	+ $names = $this->published_document_names($settings);
	3311	+
	3312	+ if (in_array($name, $names, true)) {
	3313	+ return true;
	3314	+ }
	3315	+
	3316	+ if (!preg_match('/^(.*)-\d+\.xml$/i', $name, $m)) {
	3317	+ return false;
	3318	+ }
	3319	+
	3320	+ return in_array($m[1] . '.xml', $names, true);
	3321	+ }
	3322	+
	3323	+ /**
	3324	+ * Transient key for a rendered document.
	3325	+ *
	3326	+ * @since 2.9.0
	3327	+ *
	3328	+ * @param string $filename Published file name.
	3329	+ * @return string
	3330	+ */
	3331	+ private function dynamic_cache_key(string $filename): string {
	3332	+ return self::DYNAMIC_CACHE_PREFIX . md5($filename);
	3333	+ }
	3334	+
	3335	+ /**
	3336	+ * Drop every cached dynamic document.
	3337	+ *
	3338	+ * Called from the same places that mark the static files stale, so the two
	3339	+ * delivery modes invalidate on identical triggers.
	3340	+ *
	3341	+ * @since 2.9.0
	3342	+ *
	3343	+ * @return void
	3344	+ */
	3345	+ public function flush_dynamic_cache(): void {
	3346	+ foreach ($this->published_document_names() as $name) {
	3347	+ delete_transient($this->dynamic_cache_key($name));
	3348	+ }
	3349	+ }
	3350	+
	3351	+ /**
1465	3352	* Resolve index-vs-single mode, synthesizing child sitemaps when needed.
1466	3353	*
1467	3354	* - When use_sitemap_index is on but no child sitemaps are configured, build
1468	3355	* the per-type segmented set so a real <sitemapindex> is produced (#127).
		@@ -1473,15 +3360,90 @@
1473	3360	* @param array $settings Sitemap settings.
1474	3361	* @return array Possibly-updated settings.
1475	3362	*/
1476	3363	public function maybe_promote_to_index(array $settings): array {
1477		- $has_children = count((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : [])) > 1;
	3364	+ $saved = $this->get_settings('site');
1478	3365
	3366	+ // Read from the *payload*, before the merge below folds the saved values
	3367	+ // in: "the caller named this" and "this has a value" are different
	3368	+ // questions, and the mode resolution turns on the former.
	3369	+ $mode_supplied = array_key_exists('use_sitemap_index', $settings);
	3370	+ $urls_supplied = is_array($settings['sitemap_urls'] ?? null);
	3371	+ $has_children = count($urls_supplied ? $settings['sitemap_urls'] : []) > 1;
	3372	+
	3373	+ // Which inclusion flags did the caller actually name? The child list is
	3374	+ // the only thing that reads them, and inheriting a saved one skipped
	3375	+ // that — so on an index-mode site the include_* flags were enforced
	3376	+ // nowhere but in the browser, where SitemapGeneration.js recomputes
	3377	+ // sitemap_urls itself. Every non-UI client, the shipped
	3378	+ // `update-sitemap-settings` ability included, saved the flag and changed
	3379	+ // nothing (#398). Read from the payload for the same reason as above:
	3380	+ // after the merge every saved flag would look like one the caller named.
	3381	+ $named_inclusions = array_intersect(
	3382	+ array_keys(self::INCLUSION_CHILD_TYPES),
	3383	+ array_keys($settings)
	3384	+ );
	3385	+ $inclusions_supplied = (bool) $named_inclusions;
	3386	+
1479	3387	// Inclusion flags may be absent from a partial payload (e.g. the manual
1480	3388	// generate endpoint) — fall back to saved settings so synthesized child
1481	3389	// sitemaps reflect the real include_posts/pages/categories choices.
1482		- $inclusions = array_merge($this->get_settings('site'), $settings);
	3390	+ $inclusions = array_merge($saved, $settings);
1483	3391
	3392	+ // Hand the generators a *complete* settings array. Only the mode was
	3393	+ // resolved before, so every other unnamed key reached them missing: a
	3394	+ // bare `{}` from a REST/MCP client republished the sitemap with
	3395	+ // enable_styling and include_images read as off, overwriting the live
	3396	+ // files with output that had lost its XSL stylesheet, its image
	3397	+ // namespace and its image entries. Presentation and inclusion settings
	3398	+ // are not something a generate call opts into — they are the site's
	3399	+ // configuration, and only a value actually present in the payload
	3400	+ // overrides them.
	3401	+ $settings = $inclusions;
	3402	+
	3403	+ // The two keys that drive mode keep their own resolution rules below,
	3404	+ // so they must go back to "not specified" when the caller omitted them.
	3405	+ if (!$mode_supplied) {
	3406	+ unset($settings['use_sitemap_index']);
	3407	+ }
	3408	+ if (!$urls_supplied) {
	3409	+ unset($settings['sitemap_urls']);
	3410	+ }
	3411	+
	3412	+ // An absent use_sitemap_index means "not specified", which is not the
	3413	+ // same as "single file". Reading it as the latter meant a partial payload
	3414	+ // — `{}` from a REST/MCP client, or anything short of the full settings
	3415	+ // object the admin bundle sends — republished one flat sitemap.xml on an
	3416	+ // index-mode site and left sitemap_index.xml and its children stale or
	3417	+ // missing. Inherit the saved mode instead; only a value actually present
	3418	+ // in the payload decides the mode.
	3419	+ if (!array_key_exists('use_sitemap_index', $settings)) {
	3420	+ $settings['use_sitemap_index'] = $saved['use_sitemap_index'] ?? '';
	3421	+
	3422	+ // Inheriting the mode means inheriting its children too, unless the
	3423	+ // caller named its own set.
	3424	+ $saved_children = (is_array($saved['sitemap_urls'] ?? null) ? $saved['sitemap_urls'] : []);
	3425	+ if (!empty($settings['use_sitemap_index'])
	3426	+ && !$urls_supplied
	3427	+ && count($saved_children) > 1) {
	3428	+ // A named inclusion flag is applied *to* the inherited list, not
	3429	+ // used to regenerate it. build_segmented_sitemap_urls() also adds
	3430	+ // a child for every public custom post type, so rebuilding here
	3431	+ // would make `{include_pages: false}` — one thing off — silently
	3432	+ // switch on children the saved list never had (an Elementor
	3433	+ // internal CPT, a WooCommerce product feed). Only the flags the
	3434	+ // caller actually named change anything.
	3435	+ $settings['sitemap_urls'] = $inclusions_supplied
	3436	+ ? $this->apply_inclusion_flags_to_children($saved_children, $named_inclusions, $inclusions)
	3437	+ : $saved_children;
	3438	+
	3439	+ // The children are resolved either way — including when the
	3440	+ // caller switched the last one off, which leaves a bare index and
	3441	+ // is what they asked for.
	3442	+ $has_children = true;
	3443	+ }
	3444	+ }
	3445	+
1484	3446	if (!empty($settings['use_sitemap_index'])) {
1485	3447	if (!$has_children) {
1486	3448	$settings['sitemap_urls'] = $this->build_segmented_sitemap_urls($inclusions);
1487	3449	}
		@@ -1538,26 +3500,9 @@
1538	3500	* @param array $settings Sitemap settings.
1539	3501	* @return string Sitemap filename.
1540	3502	*/
1541	3503	public function get_primary_sitemap_filename(array $settings): string {
1542		- $use_index = !empty($settings['use_sitemap_index']);
1543		-
1544		- foreach ((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : []) as $config) {
1545		- if (empty($config['enabled']) || empty($config['url'])) {
1546		- continue;
1547		- }
1548		-
1549		- if ($use_index !== (($config['type'] ?? '') === 'index')) {
1550		- continue;
1551		- }
1552		-
1553		- $path = wp_parse_url($config['url'], PHP_URL_PATH);
1554		- if (!empty($path)) {
1555		- return basename($path);
1556		- }
1557		- }
1558		-
1559		- return $use_index ? 'sitemap_index.xml' : 'sitemap.xml';
	3504	+ return thinkrank_webroot_primary_sitemap_filename($settings);
1560	3505	}
1561	3506
1562	3507	/**
1563	3508	* Public URL of the sitemap the site serves.
		@@ -1599,8 +3544,18 @@
1599	3544	// File validation failed - error details available in exception
1600	3545	return false;
1601	3546	}
1602	3547
	3548	+ // Dynamic delivery: hand the document to the collector instead of the
	3549	+ // filesystem. Reported as published, because for this run it is — the
	3550	+ // caller's success/failure bookkeeping and the index assembly both key
	3551	+ // off this return value.
	3552	+ if ($this->document_sink !== null) {
	3553	+ ($this->document_sink)($filename, $sitemap_xml);
	3554	+
	3555	+ return true;
	3556	+ }
	3557	+
1603	3558	$sitemap_path = ABSPATH . $filename;
1604	3559
1605	3560	// Use WordPress filesystem API for better security
1606	3561	global $wp_filesystem;
		@@ -1609,9 +3564,22 @@
1609	3564	WP_Filesystem();
1610	3565	}
1611	3566
1612	3567	if ($wp_filesystem) {
1613		- return $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3568	+ $written = $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE);
	3569	+
	3570	+ if ($written) {
	3571	+ // Every sitemap this version writes carries the ownership
	3572	+ // marker, so once one has been written an unmarked file at one
	3573	+ // of our names cannot be ours. Recording that retires the
	3574	+ // legacy fallback for this install — see
	3575	+ // thinkrank_webroot_sitemap_is_ours().
	3576	+ if (get_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION) !== '1') {
	3577	+ update_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION, '1', false);
	3578	+ }
	3579	+ }
	3580	+
	3581	+ return $written;
1614	3582	}
1615	3583
1616	3584	// WP_Filesystem initialization failed
1617	3585	return false;
		@@ -1634,52 +3602,164 @@
1634	3602	*/
1635	3603	public function build_segmented_sitemap_urls(array $inclusions): array {
1636	3604	$pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml';
1637	3605
1638		- $entry = static function (string $url, string $type): array {
1639		- return [
1640		- 'url' => $url,
1641		- 'type' => $type,
1642		- 'enabled' => true,
1643		- 'last_checked' => null,
1644		- 'status' => 'unknown',
1645		- ];
1646		- };
1647		- $child = static function (string $type) use ($pattern, $entry): array {
1648		- $file = str_replace('{type}', $type, $pattern);
1649		- if (strpos($file, '/') !== 0) {
1650		- $file = '/' . $file;
	3606	+ $urls = [$this->sitemap_child_entry('/sitemap_index.xml', 'index')];
	3607	+
	3608	+ foreach (self::INCLUSION_CHILD_TYPES as $flag => $type) {
	3609	+ if (!empty($inclusions[$flag])) {
	3610	+ $urls[] = $this->build_child_sitemap_entry($type, $pattern);
1651	3611	}
1652		- return $entry($file, $type);
1653		- };
1654		-
1655		- $urls = [$entry('/sitemap_index.xml', 'index')];
1656		-
1657		- if (!empty($inclusions['include_posts'])) {
1658		- $urls[] = $child('posts');
1659	3612	}
1660		- if (!empty($inclusions['include_pages'])) {
1661		- $urls[] = $child('pages');
1662		- }
1663		- if (!empty($inclusions['include_categories'])) {
1664		- $urls[] = $child('categories');
1665		- }
1666		- if (!empty($inclusions['include_tags'])) {
1667		- $urls[] = $child('tags');
1668		- }
1669	3613
1670	3614	// Public custom post types each get a child sitemap (parity with the
1671	3615	// "complete" preset and with Rank Math, which lists every public CPT).
1672	3616	foreach (get_post_types(['public' => true, '_builtin' => false], 'names') as $cpt) {
1673		- if ($this->should_include_post_type($cpt)) {
1674		- $urls[] = $child($cpt);
	3617	+ if (!$this->should_include_post_type($cpt)) {
	3618	+ continue;
1675	3619	}
	3620	+
	3621	+ // ...and the per-content-type sitemap switch (#660).
	3622	+ // get_enabled_post_types() already honours it, but this list is what
	3623	+ // index mode builds its children from — so without the same test a
	3624	+ // CPT the user had switched off still got its own child sitemap,
	3625	+ // created and streamed in full. The flags live in $inclusions, which
	3626	+ // is the settings array these children are derived from.
	3627	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $cpt, $inclusions)) {
	3628	+ continue;
	3629	+ }
	3630	+
	3631	+ $urls[] = $this->build_child_sitemap_entry($cpt, $pattern);
1676	3632	}
1677	3633
	3634	+ // Public custom taxonomies get the same treatment (#690). Flat mode has
	3635	+ // always walked them through get_enabled_taxonomies(); index mode built
	3636	+ // its children from the list above and never consulted a taxonomy at
	3637	+ // all, so every custom-taxonomy archive silently vanished from the
	3638	+ // sitemap the moment a site switched modes — and the per-taxonomy switch
	3639	+ // the matrix writes had nothing to act on. Same two tests the post-type
	3640	+ // walk applies, in the same order.
	3641	+ $taken = array_column($urls, 'type');
	3642	+
	3643	+ foreach (get_taxonomies(['public' => true, '_builtin' => false], 'names') as $taxonomy) {
	3644	+ if (!$this->should_include_taxonomy($taxonomy)) {
	3645	+ continue;
	3646	+ }
	3647	+
	3648	+ if (!\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $taxonomy, $inclusions)) {
	3649	+ continue;
	3650	+ }
	3651	+
	3652	+ // Post types and taxonomies are separate registries, so a site can
	3653	+ // hold both a `foo` post type and a `foo` taxonomy. They would
	3654	+ // resolve to one filename, and stream_type_entries() answers post
	3655	+ // types first, so the second child would list the first one's file
	3656	+ // twice in the index rather than adding anything.
	3657	+ if (in_array($taxonomy, $taken, true)) {
	3658	+ continue;
	3659	+ }
	3660	+
	3661	+ $urls[] = $this->build_child_sitemap_entry($taxonomy, $pattern);
	3662	+ }
	3663	+
1678	3664	return $urls;
1679	3665	}
1680	3666
1681	3667	/**
	3668	+ * Apply only the inclusion flags the caller named to an existing child list.
	3669	+ *
	3670	+ * The narrow counterpart to build_segmented_sitemap_urls(): that one
	3671	+ * regenerates the whole set from scratch, which is right when there is no set
	3672	+ * yet and wrong when there is. Rebuilding an existing list would add a child
	3673	+ * for every public custom post type it had never contained, so a payload that
	3674	+ * switches one thing off would switch others on. Here a flag adds or removes
	3675	+ * exactly its own child and leaves every other entry — custom post types,
	3676	+ * hand-added URLs, per-child enabled/status state — untouched (#398).
	3677	+ *
	3678	+ * @since 1.31.0
	3679	+ *
	3680	+ * @param array $children Existing child sitemap entries.
	3681	+ * @param string[] $named_inclusions Inclusion flag keys present in the payload.
	3682	+ * @param array $inclusions Merged settings, for the resolved flag
	3683	+ * values and custom_url_pattern.
	3684	+ * @return array Updated child sitemap entries.
	3685	+ */
	3686	+ private function apply_inclusion_flags_to_children(
	3687	+ array $children,
	3688	+ array $named_inclusions,
	3689	+ array $inclusions
	3690	+ ): array {
	3691	+ $pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml';
	3692	+
	3693	+ foreach ($named_inclusions as $flag) {
	3694	+ $type = self::INCLUSION_CHILD_TYPES[$flag];
	3695	+
	3696	+ $present = false;
	3697	+ foreach ($children as $entry) {
	3698	+ if (($entry['type'] ?? '') === $type) {
	3699	+ $present = true;
	3700	+ break;
	3701	+ }
	3702	+ }
	3703	+
	3704	+ if (empty($inclusions[$flag])) {
	3705	+ if ($present) {
	3706	+ $children = array_values(array_filter(
	3707	+ $children,
	3708	+ static function ($entry) use ($type): bool {
	3709	+ return (is_array($entry) ? ($entry['type'] ?? '') : '') !== $type;
	3710	+ }
	3711	+ ));
	3712	+ }
	3713	+ continue;
	3714	+ }
	3715	+
	3716	+ if (!$present) {
	3717	+ $children[] = $this->build_child_sitemap_entry($type, $pattern);
	3718	+ }
	3719	+ }
	3720	+
	3721	+ return $children;
	3722	+ }
	3723	+
	3724	+ /**
	3725	+ * Build one child sitemap entry, resolving its filename from the url pattern.
	3726	+ *
	3727	+ * @since 1.31.0
	3728	+ *
	3729	+ * @param string $type Child sitemap type (posts, pages, a post type name).
	3730	+ * @param string $pattern Filename pattern containing {type}.
	3731	+ * @return array Sitemap URL config.
	3732	+ */
	3733	+ private function build_child_sitemap_entry(string $type, string $pattern): array {
	3734	+ $file = str_replace('{type}', $type, $pattern);
	3735	+ if (strpos($file, '/') !== 0) {
	3736	+ $file = '/' . $file;
	3737	+ }
	3738	+
	3739	+ return $this->sitemap_child_entry($file, $type);
	3740	+ }
	3741	+
	3742	+ /**
	3743	+ * The shape generate_multiple_sitemaps() expects of a sitemap_urls entry.
	3744	+ *
	3745	+ * @since 1.31.0
	3746	+ *
	3747	+ * @param string $url Sitemap path.
	3748	+ * @param string $type Entry type.
	3749	+ * @return array Sitemap URL config.
	3750	+ */
	3751	+ private function sitemap_child_entry(string $url, string $type): array {
	3752	+ return [
	3753	+ 'url' => $url,
	3754	+ 'type' => $type,
	3755	+ 'enabled' => true,
	3756	+ 'last_checked' => null,
	3757	+ 'status' => 'unknown',
	3758	+ ];
	3759	+ }
	3760	+
	3761	+ /**
1682	3762	* Generate multiple sitemaps based on settings
1683	3763	*
1684	3764	* @since 1.0.0
1685	3765	* @param array $settings Sitemap settings
		@@ -1728,8 +3808,48 @@
1728	3808	if (post_type_exists($type) && !$this->should_include_post_type($type)) {
1729	3809	continue;
1730	3810	}
1731	3811
	3812	+ // Same for the matrix switch: a child list saved before the user
	3813	+ // excluded this content type still names it, and regenerating from
	3814	+ // that list would rewrite the file they asked not to have. Built-in
	3815	+ // aggregates ('posts', 'pages', ...) are not post type names, so
	3816	+ // post_type_exists() keeps this to real custom post types.
	3817	+ if (post_type_exists($type)
	3818	+ && !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('post_type', $type, $settings)) {
	3819	+ continue;
	3820	+ }
	3821	+
	3822	+ // The taxonomy counterpart of the two guards above (#690). A child
	3823	+ // list saved while a taxonomy was still included keeps naming it, so
	3824	+ // without this, excluding one in the matrix would still rewrite and
	3825	+ // re-list the file the user asked not to have. The built-in
	3826	+ // aggregates are named 'categories'/'tags' rather than
	3827	+ // 'category'/'post_tag', so taxonomy_exists() leaves them to the
	3828	+ // inclusion-flag check below.
	3829	+ $child_taxonomy = self::CHILD_TYPE_ALIASES[$type] ?? $type;
	3830	+ if (taxonomy_exists($child_taxonomy)
	3831	+ && (!$this->should_include_taxonomy($child_taxonomy)
	3832	+ || !\ThinkRank\SEO\Content_Type_Settings::is_included_in_sitemap('taxonomy', $child_taxonomy, $settings))) {
	3833	+ continue;
	3834	+ }
	3835	+
	3836	+ // The built-in aggregates carry their switch in the inclusion flag
	3837	+ // rather than under a post type name, and they are the four most
	3838	+ // people actually use. build_segmented_sitemap_urls() drops a child
	3839	+ // whose flag is empty; regenerating from a list saved while it was
	3840	+ // still on has to make the same decision, or turning Posts, Pages,
	3841	+ // Categories or Tags off in the matrix rewrites and re-lists the
	3842	+ // very file it was asked to remove.
	3843	+ // An absent flag means "not configured", which every other reader
	3844	+ // treats as included; only a flag that is present and off excludes.
	3845	+ $aggregate_flag = array_search($type, self::INCLUSION_CHILD_TYPES, true);
	3846	+ if ($aggregate_flag !== false
	3847	+ && array_key_exists($aggregate_flag, $settings)
	3848	+ && empty($settings[$aggregate_flag])) {
	3849	+ continue;
	3850	+ }
	3851	+
1732	3852	try {
1733	3853	// The single "general"/"WordPress" sitemap is one un-paginated file
1734	3854	// (there is no index to reference extra pages); it no longer drops
1735	3855	// overflow URLs.
		@@ -1765,18 +3885,36 @@
1765	3885	$buffer = [];
1766	3886	}
1767	3887	}
1768	3888
1769		- // Flush the trailing partial page, or a single empty page when the
1770		- // type had no entries at all (parity with the previous behavior of
1771		- // always writing at least one page per configured child).
1772		- if (!empty($buffer) || $page === 0) {
	3889	+ // Flush the trailing partial page.
	3890	+ //
	3891	+ // A type that produced nothing writes no page at all in index
	3892	+ // mode (#836). It used to write one empty urlset and list it in
	3893	+ // the index, so a crawler was asked to fetch a file that
	3894	+ // answers with no URLs — Search Console reports an empty
	3895	+ // sitemap referenced from an index as a warning, and the fetch
	3896	+ // is wasted on every pass. On this site three of eleven
	3897	+ // children were empty: a post type with nothing published and
	3898	+ // two taxonomies with no terms.
	3899	+ //
	3900	+ // Single-file mode still writes its one page even when empty,
	3901	+ // because that file IS the site's /sitemap.xml and a 404 there
	3902	+ // is worse than an empty urlset. Nothing lists it, so it costs
	3903	+ // no crawl budget.
	3904	+ //
	3905	+ // Skipping the write also keeps the filename out of
	3906	+ // $results['sitemaps_generated'], which is what
	3907	+ // prune_orphaned_segments() treats as "written this run" — so
	3908	+ // a type that empties after previously publishing has its
	3909	+ // stale file deleted rather than left serving.
	3910	+ if (!empty($buffer) || ($page === 0 && $index_config === null)) {
1773	3911	$page++;
1774	3912	$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children);
1775	3913	}
1776	3914
1777	3915	// Remove pages left over from a previous, larger generation.
1778		- $this->cleanup_stale_pages($sitemap_config['url'], $page);
	3916	+ $this->cleanup_stale_pages($sitemap_config['url'], $page, $settings);
1779	3917
1780	3918	} catch (\Exception $e) {
1781	3919	$results['errors'][] = "Error generating {$type} sitemap: " . $e->getMessage();
1782	3920	$results['success'] = false;
		@@ -1795,10 +3933,16 @@
1795	3933	$results['errors'][] = 'Error generating local sitemap: ' . $e->getMessage();
1796	3934	$results['success'] = false;
1797	3935	}
1798	3936
1799		- // Build the index last, from the child files actually generated.
	3937	+ // Build the index last, from the child files actually generated, plus the
	3938	+ // sitemaps other plugins own: those serve their own URLs and write no
	3939	+ // file here, so they are appended to the index only (#104).
1800	3940	if ($index_config !== null) {
	3941	+ foreach (self::additional_sitemaps() as $extra) {
	3942	+ $index_children[] = ['url' => $extra];
	3943	+ }
	3944	+
1801	3945	try {
1802	3946	$index_xml = $this->generate_sitemap_index($index_children, $settings);
1803	3947	$filename = basename(wp_parse_url($index_config['url'], PHP_URL_PATH));
1804	3948
		@@ -1818,12 +3962,111 @@
1818	3962	$results['success'] = false;
1819	3963	}
1820	3964	}
1821	3965
	3966	+ // Remove segments that are no longer part of the set. Publishing was
	3967	+ // purely additive: a type that dropped out (Categories unticked, a CPT
	3968	+ // that stopped qualifying) simply stopped being overwritten, so its file
	3969	+ // kept serving and — until the index happened to be rebuilt — kept being
	3970	+ // listed in it. The index above is built from the children actually
	3971	+ // generated, so pruning here leaves disk and index agreeing.
	3972	+ $this->prune_orphaned_segments($settings, $results['sitemaps_generated']);
	3973	+
1822	3974	return $results;
1823	3975	}
1824	3976
1825	3977	/**
	3978	+ * Delete published segment files that this run did not write.
	3979	+ *
	3980	+ * Only filenames this site could have published under its own url pattern
	3981	+ * are considered, so another plugin's or core's sitemap in the web root is
	3982	+ * never a candidate — the same reason cleanup does not glob 'sitemap-*.xml'.
	3983	+ *
	3984	+ * @since 1.31.0
	3985	+ *
	3986	+ * @param array $settings Sitemap settings (read for `custom_url_pattern`).
	3987	+ * @param array $generated Entries from $results['sitemaps_generated'].
	3988	+ * @return string[] Basenames removed.
	3989	+ */
	3990	+ private function prune_orphaned_segments(array $settings, array $generated): array {
	3991	+ // Rendering for a request, not publishing: there is nothing on disk
	3992	+ // this run owns, and a dynamic render must never delete the files a
	3993	+ // site's previous static mode left behind.
	3994	+ if ($this->is_collecting()) {
	3995	+ return [];
	3996	+ }
	3997	+
	3998	+ $kept = [];
	3999	+ foreach ($generated as $entry) {
	4000	+ if (!empty($entry['filename'])) {
	4001	+ $kept[strtolower((string) $entry['filename'])] = true;
	4002	+ }
	4003	+ }
	4004	+
	4005	+ // The current mode's primary and the local business sitemap are written
	4006	+ // by their own paths and are never orphans here.
	4007	+ $primary = strtolower(basename($this->get_primary_sitemap_filename($settings)));
	4008	+ $kept[$primary] = true;
	4009	+ $kept['local-sitemap.xml'] = true;
	4010	+
	4011	+ // The OTHER mode's primary is an orphan the moment the mode changes:
	4012	+ // index mode leaves sitemap.xml behind, flat mode leaves
	4013	+ // sitemap_index.xml and its children. Both used to be kept
	4014	+ // unconditionally, so the site served two sitemap trees and only ever
	4015	+ // refreshed one (#563). The children are already covered by the segment
	4016	+ // sweep below, which now sees them because the index is no longer kept.
	4017	+ $stale_primaries = array_diff(['sitemap.xml', 'sitemap_index.xml'], [$primary]);
	4018	+
	4019	+ $removed = [];
	4020	+
	4021	+ global $wp_filesystem;
	4022	+ if (!$wp_filesystem) {
	4023	+ require_once ABSPATH . 'wp-admin/includes/file.php';
	4024	+ WP_Filesystem();
	4025	+ }
	4026	+ if (!$wp_filesystem) {
	4027	+ return $removed;
	4028	+ }
	4029	+
	4030	+ $candidates = array_merge($this->publishable_segment_filenames($settings), $stale_primaries);
	4031	+
	4032	+ foreach ($candidates as $candidate) {
	4033	+ if (isset($kept[strtolower($candidate)])) {
	4034	+ continue;
	4035	+ }
	4036	+
	4037	+ if (!preg_match('/^(.*)\.xml$/i', $candidate, $m)) {
	4038	+ continue;
	4039	+ }
	4040	+
	4041	+ // The base file plus its numeric pagination pages.
	4042	+ $paths = [ABSPATH . $candidate];
	4043	+ foreach (glob(ABSPATH . $m[1] . '-*.xml') ?: [] as $paged) {
	4044	+ if (preg_match('/^' . preg_quote($m[1], '/') . '-\d+\.xml$/i', basename($paged))) {
	4045	+ $paths[] = $paged;
	4046	+ }
	4047	+ }
	4048	+
	4049	+ foreach ($paths as $path) {
	4050	+ if (!file_exists($path)) {
	4051	+ continue;
	4052	+ }
	4053	+ // A name we could have published is not proof we published
	4054	+ // this file: RankMath and core write at the same paths (#515).
	4055	+ if (!$this->webroot_sitemap_is_ours($path, $settings)) {
	4056	+ continue;
	4057	+ }
	4058	+ if ($wp_filesystem->delete($path)) {
	4059	+ $removed[] = basename($path);
	4060	+ }
	4061	+ }
	4062	+ }
	4063	+
	4064	+ return $removed;
	4065	+ }
	4066	+
	4067	+
	4068	+ /**
1826	4069	* Collect the full (un-paginated) entry list for a content sitemap type.
1827	4070	*
1828	4071	* @since 1.14.0
1829	4072	*
		@@ -1848,11 +4091,15 @@
1848	4091	$settings = $settings ?? $this->get_settings('site');
1849	4092	$entries = $this->collect_local_entries();
1850	4093
1851	4094	if (empty($entries)) {
1852		- // Business identity was cleared — drop any file left from before.
	4095	+ // Business identity was cleared — drop the file we left from
	4096	+ // before, but only ours. `local-sitemap.xml` is the name Rank Math
	4097	+ // publishes under too (this method mirrors it deliberately), so on
	4098	+ // a migrated site the file at that path may never have been ours
	4099	+ // to delete (#515).
1853	4100	$path = ABSPATH . 'local-sitemap.xml';
1854		- if (file_exists($path)) {
	4101	+ if (!$this->is_collecting() && file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) {
1855	4102	wp_delete_file($path);
1856	4103	}
1857	4104	return false;
1858	4105	}
		@@ -1874,8 +4121,25 @@
1874	4121	*
1875	4122	* @since 1.15.x
1876	4123	* @return array Zero or one URL entry
1877	4124	*/
	4125	+ /**
	4126	+ * Does this site publish a local business sitemap right now?
	4127	+ *
	4128	+ * The same gate {@see self::regenerate_local_sitemap()} applies, asked
	4129	+ * without writing anything. Callers that need to know whether the document
	4130	+ * exists must not test the filesystem: under dynamic delivery it is served
	4131	+ * from PHP and there is no file, which is how `local-sitemap.xml` came to be
	4132	+ * dropped from robots.txt on exactly those sites (#752).
	4133	+ *
	4134	+ * @since 2.9.0
	4135	+ *
	4136	+ * @return bool True when the local sitemap has content to publish.
	4137	+ */
	4138	+ public function publishes_local_sitemap(): bool {
	4139	+ return !empty($this->collect_local_entries());
	4140	+ }
	4141	+
1878	4142	private function collect_local_entries(): array {
1879	4143	if (!class_exists('ThinkRank\\SEO\\Site_Identity_Manager')) {
1880	4144	require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-site-identity-manager.php';
1881	4145	}
		@@ -1947,17 +4211,43 @@
1947	4211	case 'products':
1948	4212	$entries = post_type_exists('product') ? $this->collect_post_entries_iter(['product'], $settings) : [];
1949	4213	return ['entries' => $entries, 'image_ns' => true];
1950	4214
1951		- case 'product_categories':
1952		- $entries = taxonomy_exists('product_cat') ? $this->collect_taxonomy_entries_iter('product_cat', $settings) : [];
1953		- return ['entries' => $entries, 'image_ns' => false];
	4215	+ default:
	4216	+ // Resolve a preset's display name to the object it streams, so
	4217	+ // 'product_categories' is an ordinary taxonomy child rather than
	4218	+ // a case of its own (#690).
	4219	+ $alias = self::CHILD_TYPE_ALIASES[$type] ?? null;
	4220	+ $object = $alias ?? $type;
1954	4221
1955		- default:
1956	4222	$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names');
1957		- if (in_array($type, $custom_post_types, true)) {
1958		- return ['entries' => $this->collect_post_entries_iter([$type], $settings), 'image_ns' => true];
	4223	+ if (in_array($object, $custom_post_types, true)) {
	4224	+ return ['entries' => $this->collect_post_entries_iter([$object], $settings), 'image_ns' => true];
1959	4225	}
	4226	+
	4227	+ // Custom taxonomies reach index mode here, streamed through the
	4228	+ // same iterator flat mode uses so the two modes emit identical
	4229	+ // URLs for the same settings.
	4230	+ if (taxonomy_exists($object) && $this->should_include_taxonomy($object)) {
	4231	+ return ['entries' => $this->collect_taxonomy_entries_iter($object, $settings), 'image_ns' => false];
	4232	+ }
	4233	+
	4234	+ // An aliased child whose object is gone (WooCommerce deactivated)
	4235	+ // keeps writing the empty file it always wrote. Returning null
	4236	+ // here would hand it the whole-site fallback instead, dumping
	4237	+ // every URL on the site into a file named for products.
	4238	+ //
	4239	+ // A registered taxonomy this generator will not emit
	4240	+ // (`post_format`, `nav_menu`, a non-public one) needs the same
	4241	+ // answer for the same reason. generate_multiple_sitemaps()
	4242	+ // skips those before they reach here, so nothing takes this
	4243	+ // path today — but it is the one branch where falling through
	4244	+ // to null is silently catastrophic rather than merely wrong,
	4245	+ // and the guard keeping it unreachable lives in another method.
	4246	+ if ($alias !== null || taxonomy_exists($object)) {
	4247	+ return ['entries' => [], 'image_ns' => false];
	4248	+ }
	4249	+
1960	4250	return null;
1961	4251	}
1962	4252	}
1963	4253
		@@ -2000,13 +4290,23 @@
2000	4290	* which has no -N suffix) is never touched.
2001	4291	*
2002	4292	* @since 1.14.0
2003	4293	*
	4294	+ * @since 2.1.1 Each candidate must pass the content ownership test — a
	4295	+ * `-N.xml` page of another plugin's sitemap paginates our
	4296	+ * stem exactly as ours does (#515).
	4297	+ *
2004	4298	* @param string $base_url Base (page 1) sitemap URL
2005	4299	* @param int $current_pages Number of pages generated this run
	4300	+ * @param array $settings Sitemap settings, for the ownership test.
2006	4301	* @return void
2007	4302	*/
2008		- private function cleanup_stale_pages(string $base_url, int $current_pages): void {
	4303	+ private function cleanup_stale_pages(string $base_url, int $current_pages, array $settings): void {
	4304	+ // See prune_orphaned_segments(): a dynamic render deletes nothing.
	4305	+ if ($this->is_collecting()) {
	4306	+ return;
	4307	+ }
	4308	+
2009	4309	$filename = basename(wp_parse_url($base_url, PHP_URL_PATH));
2010	4310	if (!preg_match('/^(.*)\.xml$/i', $filename, $m)) {
2011	4311	return;
2012	4312	}
		@@ -2023,9 +4323,11 @@
2023	4323
2024	4324	$candidates = glob(ABSPATH . $stem . '-*.xml') ?: [];
2025	4325	foreach ($candidates as $path) {
2026	4326	// Only delete numeric-suffixed pages beyond the current count.
2027		- if (preg_match('/-(\d+)\.xml$/', basename($path), $mm) && (int) $mm[1] > $current_pages) {
	4327	+ if (preg_match('/-(\d+)\.xml$/', basename($path), $mm)
	4328	+ && (int) $mm[1] > $current_pages
	4329	+ && $this->webroot_sitemap_is_ours($path, $settings)) {
2028	4330	$wp_filesystem->delete($path);
2029	4331	}
2030	4332	}
2031	4333	}
		@@ -2030,8 +4332,97 @@
2030	4332	}
2031	4333	}
2032	4334
2033	4335	/**
	4336	+ * Sitemap URLs contributed by other plugins.
	4337	+ *
	4338	+ * ThinkRank owns the sitemap index and the robots.txt `Sitemap:` lines, so a
	4339	+ * companion plugin that serves its own sitemap — Pro's news and video
	4340	+ * sitemaps, for instance — had no way to be discovered: it appeared in
	4341	+ * neither, leaving manual Search Console submission as the only route in
	4342	+ * (#104). Registering here puts a sitemap in the index when one exists, and
	4343	+ * in robots.txt when it does not.
	4344	+ *
	4345	+ * Callers get root-relative paths. Entries are normalised to a leading
	4346	+ * slash, de-duplicated, and anything that is not a non-empty string is
	4347	+ * dropped, so one badly-behaved callback cannot produce a malformed index.
	4348	+ *
	4349	+ * @since 2.3.1
	4350	+ *
	4351	+ * @return string[] Root-relative sitemap paths, e.g. ['/news-sitemap.xml'].
	4352	+ */
	4353	+ public static function additional_sitemaps(): array {
	4354	+ /**
	4355	+ * Filters the sitemaps contributed by other plugins.
	4356	+ *
	4357	+ * @since 2.3.1
	4358	+ *
	4359	+ * @param string[] $sitemaps Root-relative sitemap paths.
	4360	+ */
	4361	+ $sitemaps = apply_filters('thinkrank_additional_sitemaps', []);
	4362	+
	4363	+ if (!is_array($sitemaps)) {
	4364	+ return [];
	4365	+ }
	4366	+
	4367	+ // Both consumers resolve an entry with home_url(), which prefixes the
	4368	+ // install's own directory. Everything below is measured against that so
	4369	+ // an absolute URL is reduced to what home_url() will put back.
	4370	+ $home = wp_parse_url(home_url('/'));
	4371	+ $home_host = strtolower((string) ($home['host'] ?? ''));
	4372	+ $home_path = '/' . trim((string) ($home['path'] ?? ''), '/');
	4373	+
	4374	+ $clean = [];
	4375	+ foreach ($sitemaps as $sitemap) {
	4376	+ if (!is_string($sitemap)) {
	4377	+ continue;
	4378	+ }
	4379	+
	4380	+ $sitemap = trim($sitemap);
	4381	+ if ('' === $sitemap) {
	4382	+ continue;
	4383	+ }
	4384	+
	4385	+ // A full URL on this site is accepted and reduced to the part
	4386	+ // home_url() does not already supply, so a caller that reached for
	4387	+ // home_url() still lands in the right place — including on a
	4388	+ // subdirectory install, where keeping the whole path would repeat
	4389	+ // the directory. A URL on another host is dropped rather than
	4390	+ // rewritten: the sitemaps protocol will not accept a cross-host
	4391	+ // child anyway, and reusing its path would advertise a URL on this
	4392	+ // site that does not exist.
	4393	+ if (preg_match('#^(https?:)?//#i', $sitemap)) {
	4394	+ $parts = wp_parse_url('//' === substr($sitemap, 0, 2) ? 'https:' . $sitemap : $sitemap);
	4395	+ if (!is_array($parts)) {
	4396	+ continue;
	4397	+ }
	4398	+
	4399	+ if (strtolower((string) ($parts['host'] ?? '')) !== $home_host) {
	4400	+ continue;
	4401	+ }
	4402	+
	4403	+ $path = (string) ($parts['path'] ?? '');
	4404	+ if ('' === $path) {
	4405	+ continue;
	4406	+ }
	4407	+
	4408	+ if ('/' !== $home_path && ($path === $home_path || 0 === strpos($path, $home_path . '/'))) {
	4409	+ $path = substr($path, strlen($home_path));
	4410	+ }
	4411	+
	4412	+ // A sitemap served from a query string keeps it; dropping the
	4413	+ // query would point at a different document.
	4414	+ $query = (string) ($parts['query'] ?? '');
	4415	+ $sitemap = $path . ('' !== $query ? '?' . $query : '');
	4416	+ }
	4417	+
	4418	+ $clean[] = '/' . ltrim($sitemap, '/');
	4419	+ }
	4420	+
	4421	+ return array_values(array_unique($clean));
	4422	+ }
	4423	+
	4424	+ /**
2034	4425	* Generate sitemap index XML from the list of child sitemap files produced
2035	4426	* during generation (each already resolved to its final, possibly paginated,
2036	4427	* URL).
2037	4428	*
		@@ -2040,14 +4431,9 @@
2040	4431	* @param array $settings Sitemap settings
2041	4432	* @return string Sitemap index XML
2042	4433	*/
2043	4434	private function generate_sitemap_index(array $children, array $settings): string {
2044		- $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
2045		-
2046		- // Add XSL stylesheet only if styling is enabled
2047		- if (!empty($settings['enable_styling'])) {
2048		- $xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/sitemap-index.xsl') . '"?>' . "\n";
2049		- }
	4435	+ $xml = $this->xml_prolog($settings, 'index');
2050	4436	$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
2051	4437
2052	4438	$site_url = home_url();
2053	4439
		@@ -2060,9 +4446,9 @@
2060	4446	$sitemap_url = $site_url . $sitemap_url;
2061	4447	}
2062	4448
2063	4449	$xml .= " <sitemap>\n";
2064		- $xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n";
	4450	+ $xml .= " <loc>" . esc_url(Url_Scheme::apply($sitemap_url)) . "</loc>\n";
2065	4451	$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n";
2066	4452	$xml .= " </sitemap>\n";
2067	4453	}
2068	4454