PluginProbe
Jetpack – WP Security, Backup, Speed, & Growth / 16.3
Jetpack – WP Security, Backup, Speed, & Growth v16.3
16.3 16.3-beta 16.3-a.5 16.3-a.7 16.3-a.3 16.3-a.1 16.2 16.2-beta 12.0.3 12.1.3 12.2.3 12.3.2 12.4.2 12.5.2 12.6.4 12.7.3 12.8.3 12.9.5 13.0.2 13.1.5 13.2.4 13.3.3 13.4.5 13.5.2 13.6.2 All 508 releases
← All changes | modules/sitemaps/abilities/class-sitemaps-abilities.php +457 -0 16.2-beta → 16.3 View file →
@@ -1,0 +1,457 @@
1 +<?php
2 +/**
3 + * Jetpack Sitemaps Abilities Registration
4 + *
5 + * Registers Jetpack Sitemaps abilities with the WordPress Abilities API.
6 + *
7 + * @package automattic/jetpack
8 + */
9 +
10 +namespace Automattic\Jetpack\Plugin\Abilities;
11 +
12 +use Automattic\Jetpack\WP_Abilities\Registrar;
13 +use Jetpack;
14 +
15 +if ( ! defined( 'ABSPATH' ) ) {
16 + exit( 0 );
17 +}
18 +
19 +/**
20 + * Registers Jetpack Sitemaps abilities with the WordPress Abilities API.
21 + *
22 + * Exposes a zero-arg sitemap status read (`get-status`) and a zero-arg rebuild
23 + * dispatch (`request-rebuild`) so AI agents can inspect sitemap freshness and
24 + * trigger a regeneration through the standard `wp-abilities/v1` REST surface.
25 + *
26 + * Both abilities only register while the Sitemaps module is active — the
27 + * surrounding `modules/sitemaps.php` is only loaded by Jetpack when the module
28 + * is on, so the `Sitemaps_Abilities::init()` call at the bottom of that file
29 + * is the gate.
30 + */
31 +class Sitemaps_Abilities extends Registrar {
32 +
33 + private const MODULE_SLUG = 'sitemaps';
34 +
35 + /**
36 + * Cron hook name used by the Sitemaps module to drive incremental builds.
37 + *
38 + * Kept as a const here rather than imported from `Jetpack_Sitemap_Manager`
39 + * because the manager registers it as an action name only — there is no
40 + * canonical PHP constant to reference, and the value is part of the
41 + * module's stable public surface (it shows up in `wp cron list`).
42 + */
43 + private const CRON_HOOK = 'jp_sitemap_cron_hook';
44 +
45 + /**
46 + * Transient written by `Jetpack_Sitemap_State::check_out()` while a build
47 + * step is in progress. Presence of this transient is the canonical
48 + * "build currently running" signal; its 15-minute TTL means the signal
49 + * self-clears if a build crashes without unlocking.
50 + */
51 + private const STATE_LOCK_TRANSIENT = 'jetpack-sitemap-state-lock';
52 +
53 + /**
54 + * {@inheritDoc}
55 + *
56 + * Sitemaps abilities live under the WordPress core `site` category — it is
57 + * registered by the Abilities API itself, so we reference it by slug and
58 + * never register it ourselves (see the no-op `register_category()` below).
59 + */
60 + public static function get_category_slug(): string {
61 + return 'site';
62 + }
63 +
64 + /**
65 + * {@inheritDoc}
66 + *
67 + * Unused: the `site` category is owned by WordPress core, so
68 + * `register_category()` is a no-op and this definition is never passed to
69 + * `wp_register_ability_category()`. It remains only to satisfy the abstract
70 + * Registrar contract.
71 + */
72 + public static function get_category_definition(): array {
73 + return array();
74 + }
75 +
76 + /**
77 + * No-op: the `site` ability category is registered by the WordPress core
78 + * Abilities API. Re-registering it here would clobber the core definition,
79 + * so this registrar only references the category by slug.
80 + *
81 + * @return void
82 + */
83 + public static function register_category() {}
84 +
85 + /**
86 + * {@inheritDoc}
87 + */
88 + public static function get_abilities(): array {
89 + return array(
90 + 'jetpack-sitemaps/get-status' => array(
91 + 'label' => __( 'Get Jetpack Sitemaps status', 'jetpack' ),
92 + 'description' => __( 'Return the current state of the Jetpack-generated XML sitemaps as { active, url, post_count, page_count, news_sitemap_enabled, sitemaps }. `active` reflects whether the Sitemaps module is on. `url` is the public sitemap.xml entry point. `post_count` / `page_count` are the published `post` / `page` counts (the same baseline the WordPress core sitemap uses). `news_sitemap_enabled` reflects the `jetpack_news_sitemap_include_in_robotstxt` filter (default true). `sitemaps` is the list of child sitemaps actually present in the served sitemap.xml index — each entry is `{ loc, lastmod }`, where `lastmod` is the W3C datetime string the sitemap exposes (or null when that entry omits one). `sitemaps` is an empty array until a master sitemap has been generated. These abilities are only registered while the Sitemaps module is active; if they are absent from wp_get_abilities(), activate the Sitemaps module first.', 'jetpack' ),
93 + 'input_schema' => array(
94 + 'type' => 'object',
95 + 'additionalProperties' => false,
96 + ),
97 + 'output_schema' => array(
98 + 'type' => 'object',
99 + 'properties' => array(
100 + 'active' => array( 'type' => 'boolean' ),
101 + 'url' => array( 'type' => 'string' ),
102 + 'post_count' => array( 'type' => 'integer' ),
103 + 'page_count' => array( 'type' => 'integer' ),
104 + 'news_sitemap_enabled' => array( 'type' => 'boolean' ),
105 + 'sitemaps' => array(
106 + 'type' => 'array',
107 + 'items' => array(
108 + 'type' => 'object',
109 + 'properties' => array(
110 + 'loc' => array( 'type' => 'string' ),
111 + 'lastmod' => array( 'type' => array( 'string', 'null' ) ),
112 + ),
113 + ),
114 + ),
115 + ),
116 + ),
117 + 'execute_callback' => array( __CLASS__, 'get_status' ),
118 + 'permission_callback' => array( __CLASS__, 'can_view_sitemaps' ),
119 + 'meta' => array(
120 + 'annotations' => array(
121 + 'readonly' => true,
122 + 'destructive' => false,
123 + 'idempotent' => true,
124 + ),
125 + 'show_in_rest' => true,
126 + 'mcp' => array(
127 + 'public' => true,
128 + 'type' => 'tool', // default is already "tool", but can be explicit.
129 + ),
130 + ),
131 + ),
132 +
133 + 'jetpack-sitemaps/request-rebuild' => array(
134 + 'label' => __( 'Request a Jetpack Sitemaps rebuild', 'jetpack' ),
135 + 'description' => __( 'Dispatch a full sitemap regeneration by scheduling the existing `jp_sitemap_cron_hook` cron event. Returns { dispatched, status, next_scheduled_at } where status is one of "queued" (a single-event cron tick was just scheduled), "running" (a build is already in flight per the `jetpack-sitemap-state-lock` transient), or "already_running" (alias of "running"; surfaced so callers can branch on either spelling). `next_scheduled_at` is the next `jp_sitemap_cron_hook` tick as an ISO 8601 UTC string with an explicit `Z` zone designator (e.g. `2026-05-19T19:33:20Z`), or null when nothing is scheduled (e.g. status=running with no future tick queued) — it tells the caller when the build they queued (or the one already pending) will actually run. Idempotent — calling this while a build is already in flight or already queued returns dispatched=false and the matching status rather than stacking duplicate cron events.', 'jetpack' ),
136 + 'input_schema' => array(
137 + 'type' => 'object',
138 + 'additionalProperties' => false,
139 + ),
140 + 'output_schema' => array(
141 + 'type' => 'object',
142 + 'properties' => array(
143 + 'dispatched' => array( 'type' => 'boolean' ),
144 + 'status' => array(
145 + 'type' => 'string',
146 + 'enum' => array( 'queued', 'running', 'already_running' ),
147 + ),
148 + 'next_scheduled_at' => array( 'type' => array( 'string', 'null' ) ),
149 + ),
150 + ),
151 + 'execute_callback' => array( __CLASS__, 'request_rebuild' ),
152 + 'permission_callback' => array( __CLASS__, 'can_manage_sitemaps' ),
153 + 'meta' => array(
154 + 'annotations' => array(
155 + 'readonly' => false,
156 + 'destructive' => false,
157 + 'idempotent' => true,
158 + ),
159 + 'show_in_rest' => true,
160 + 'mcp' => array(
161 + 'public' => true,
162 + 'type' => 'tool', // default is already "tool", but can be explicit.
163 + ),
164 + ),
165 + ),
166 + );
167 + }
168 +
169 + /**
170 + * Permission check: can the current user read sitemap status?
171 + *
172 + * Sitemap status is metadata about publicly-served XML — anyone who can
173 + * manage content (`edit_posts`) is allowed to see it. Reads do not modify
174 + * state and do not expose secrets.
175 + */
176 + public static function can_view_sitemaps(): bool {
177 + return current_user_can( 'edit_posts' );
178 + }
179 +
180 + /**
181 + * Permission check: can the current user dispatch a sitemap rebuild?
182 + *
183 + * Rebuild scheduling writes to cron + transient state and can run for
184 + * minutes on large sites, so it is gated on `manage_options` (admin only).
185 + */
186 + public static function can_manage_sitemaps(): bool {
187 + return current_user_can( 'manage_options' );
188 + }
189 +
190 + /**
191 + * Execute: status read.
192 + *
193 + * Surfaces an opinionated, agent-friendly projection of the module's state:
194 + * - `active` from `Jetpack::is_module_active`.
195 + * - `url` from `jetpack_sitemap_uri()`, the same helper the public sitemap
196 + * router uses.
197 + * - `post_count` / `page_count` from `wp_count_posts()->publish`, the
198 + * same baseline used by the WordPress core sitemap. Cheap; no joins.
199 + * - `news_sitemap_enabled` from the `jetpack_news_sitemap_include_in_robotstxt`
200 + * filter (the same filter that controls news-sitemap robots.txt inclusion).
201 + * - `sitemaps` from the served master sitemap document itself (see
202 + * `get_sitemap_entries()`) — the real child-sitemap list with each
203 + * entry's own `lastmod`, rather than a synthetic last-build timestamp.
204 + *
205 + * @param array|null $input Ability input (no parameters accepted).
206 + * @return array
207 + */
208 + public static function get_status( $input = null ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable -- Abilities API contract requires execute callbacks to accept the input array even when the schema declares no parameters.
209 + return array(
210 + 'active' => Jetpack::is_module_active( self::MODULE_SLUG ),
211 + 'url' => static::get_master_sitemap_url(),
212 + 'post_count' => static::count_published( 'post' ),
213 + 'page_count' => static::count_published( 'page' ),
214 + 'news_sitemap_enabled' => static::is_news_sitemap_enabled(),
215 + 'sitemaps' => static::get_sitemap_entries(),
216 + );
217 + }
218 +
219 + /**
220 + * Execute: rebuild dispatch.
221 + *
222 + * Three-state idempotent dispatch:
223 + *
224 + * 1. If the state lock transient is set, a build step is currently
225 + * running. Return `dispatched=false`, `status=running`. We also surface
226 + * `already_running` as the alias the plan documents; this function
227 + * returns `running` as the canonical value so callers that branch on
228 + * one or the other both work — the output_schema enum permits both.
229 + * 2. Else if a cron event is already scheduled in the future for our hook,
230 + * a build is queued. Return `dispatched=false`, `status=queued`.
231 + * 3. Otherwise schedule a single-event cron tick to fire immediately and
232 + * return `dispatched=true`, `status=queued`.
233 + *
234 + * Every branch also returns `next_scheduled_at` (see
235 + * `get_next_scheduled_at()`) so the caller learns when the queued/pending
236 + * build will actually run without a follow-up status read.
237 + *
238 + * @param array|null $input Ability input (no parameters accepted).
239 + * @return array
240 + */
241 + public static function request_rebuild( $input = null ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable -- Abilities API contract requires execute callbacks to accept the input array even when the schema declares no parameters.
242 + if ( static::is_build_running() ) {
243 + return array(
244 + 'dispatched' => false,
245 + 'status' => 'running',
246 + 'next_scheduled_at' => static::get_next_scheduled_at(),
247 + );
248 + }
249 +
250 + if ( static::is_build_queued() ) {
251 + return array(
252 + 'dispatched' => false,
253 + 'status' => 'queued',
254 + 'next_scheduled_at' => static::get_next_scheduled_at(),
255 + );
256 + }
257 +
258 + static::schedule_rebuild();
259 +
260 + return array(
261 + 'dispatched' => true,
262 + 'status' => 'queued',
263 + 'next_scheduled_at' => static::get_next_scheduled_at(),
264 + );
265 + }
266 +
267 + /**
268 + * Public sitemap URL for the master sitemap.
269 + *
270 + * Extracted as a protected seam so tests can override without booting the
271 + * rewrite/permalink stack.
272 + */
273 + protected static function get_master_sitemap_url(): string {
274 + if ( function_exists( 'jetpack_sitemap_uri' ) ) {
275 + return (string) jetpack_sitemap_uri( 'sitemap.xml' );
276 + }
277 + // Defensive: when the sitemaps module file is loaded the helper
278 + // exists. This branch only runs if a caller invokes the ability
279 + // outside the normal bootstrap path.
280 + return (string) home_url( '/sitemap.xml' );
281 + }
282 +
283 + /**
284 + * Raw master-sitemap XML — the exact document the public `sitemap.xml`
285 + * router serves, read straight from storage via the librarian (no HTTP
286 + * loopback). Returns an empty string when no master sitemap has been
287 + * generated yet, or when the Sitemaps module helpers are unavailable.
288 + *
289 + * Extracted as a protected seam so tests can feed a known document without
290 + * a librarian / wp_posts.
291 + */
292 + protected static function get_master_sitemap_xml(): string {
293 + if (
294 + ! class_exists( 'Jetpack_Sitemap_Librarian' )
295 + || ! function_exists( 'jp_sitemap_filename' )
296 + || ! defined( 'JP_MASTER_SITEMAP_TYPE' )
297 + ) {
298 + return '';
299 + }
300 +
301 + $librarian = new \Jetpack_Sitemap_Librarian();
302 +
303 + // jp_sitemap_filename() is documented `@param string $number`; for the
304 + // master type it returns 'sitemap.xml' and ignores the number, but it
305 + // must be non-null and string-typed to satisfy the contract (the
306 + // int-`0` router call site predates this and is Phan-baselined).
307 + return (string) $librarian->get_sitemap_text(
308 + \jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, '0' ),
309 + JP_MASTER_SITEMAP_TYPE
310 + );
311 + }
312 +
313 + /**
314 + * The child-sitemap entries actually present in the served master
315 + * sitemap, as a list of `[ 'loc' => string, 'lastmod' => string|null ]`.
316 + *
317 + * Parses the same `<sitemapindex>` document `sitemap.xml` serves rather
318 + * than deriving freshness from the `jetpack-sitemap-state` option: that
319 + * option can read its initial/reset shape (no `max` projection) even while
320 + * a fully-built sitemap.xml is being served, so it is not a reliable
321 + * "what does the sitemap actually contain" source.
322 + *
323 + * Returns an empty array when no master sitemap exists yet or the stored
324 + * document does not parse.
325 + *
326 + * @return array<int, array{loc:string, lastmod:string|null}>
327 + */
328 + protected static function get_sitemap_entries(): array {
329 + $xml = static::get_master_sitemap_xml();
330 + if ( '' === $xml ) {
331 + return array();
332 + }
333 +
334 + $previous = libxml_use_internal_errors( true );
335 + $document = new \DOMDocument();
336 + // Source is Jetpack's own stored sitemap (not user input) and PHP 8+
337 + // disables external-entity loading by default; LIBXML_NONET is belt-
338 + // and-suspenders against any network/entity fetch during parsing.
339 + $loaded = $document->loadXML( $xml, LIBXML_NONET );
340 + libxml_clear_errors();
341 + libxml_use_internal_errors( $previous );
342 +
343 + if ( ! $loaded ) {
344 + return array();
345 + }
346 +
347 + $entries = array();
348 + foreach ( $document->getElementsByTagName( 'sitemap' ) as $sitemap_node ) {
349 + $loc_nodes = $sitemap_node->getElementsByTagName( 'loc' );
350 + if ( 0 === $loc_nodes->length ) {
351 + continue;
352 + }
353 +
354 + $loc = trim( $loc_nodes->item( 0 )->textContent );
355 + if ( '' === $loc ) {
356 + continue;
357 + }
358 +
359 + $lastmod_nodes = $sitemap_node->getElementsByTagName( 'lastmod' );
360 + $lastmod = $lastmod_nodes->length > 0
361 + ? trim( $lastmod_nodes->item( 0 )->textContent )
362 + : '';
363 +
364 + $entries[] = array(
365 + 'loc' => $loc,
366 + 'lastmod' => '' === $lastmod ? null : $lastmod,
367 + );
368 + }
369 +
370 + return $entries;
371 + }
372 +
373 + /**
374 + * Whether news-sitemap inclusion is enabled.
375 + *
376 + * Mirrors the filter chain in `Jetpack_Sitemap_Manager::callback_action_do_robotstxt`
377 + * but only resolves the modern filter — the deprecated 7.4.0 alias is
378 + * already merged into the modern filter by the time it runs in production.
379 + */
380 + protected static function is_news_sitemap_enabled(): bool {
381 + /** This filter is documented in modules/sitemaps/sitemaps.php */
382 + return (bool) apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', true );
383 + }
384 +
385 + /**
386 + * Count published posts of a given post type.
387 + *
388 + * Wraps `wp_count_posts()` so tests can override without a real WP_Posts
389 + * factory.
390 + *
391 + * @param string $post_type Post type slug.
392 + */
393 + protected static function count_published( string $post_type ): int {
394 + $counts = wp_count_posts( $post_type );
395 + if ( ! is_object( $counts ) || ! isset( $counts->publish ) ) {
396 + return 0;
397 + }
398 + return (int) $counts->publish;
399 + }
400 +
401 + /**
402 + * Whether a sitemap build step is currently running.
403 + *
404 + * The Sitemaps module sets a 15-minute transient lock at the start of
405 + * `Jetpack_Sitemap_State::check_out()` and deletes it on `unlock()` /
406 + * `reset()`. Presence of the transient is the canonical "in flight" signal.
407 + */
408 + protected static function is_build_running(): bool {
409 + return true === get_transient( self::STATE_LOCK_TRANSIENT );
410 + }
411 +
412 + /**
413 + * Whether a sitemap build is already scheduled for a future cron tick.
414 + *
415 + * Uses `wp_next_scheduled` so we don't stack duplicate single-event cron
416 + * entries when the recurring `jp_sitemap_cron_hook` is already pending.
417 + */
418 + protected static function is_build_queued(): bool {
419 + return false !== wp_next_scheduled( self::CRON_HOOK );
420 + }
421 +
422 + /**
423 + * Schedule a single-event cron tick to drive the next build step.
424 + *
425 + * Matches the dispatch pattern used by
426 + * `Jetpack_Sitemap_Manager::callback_action_purge_data` — `wp_schedule_single_event`
427 + * with an immediate execution time. The recurring `sitemap-interval`
428 + * schedule still fires on its normal cadence; this just front-runs the
429 + * next tick.
430 + */
431 + protected static function schedule_rebuild(): void {
432 + wp_schedule_single_event( time(), self::CRON_HOOK );
433 + }
434 +
435 + /**
436 + * When the next `jp_sitemap_cron_hook` build tick is scheduled, as an
437 + * ISO 8601 UTC string (e.g. `2026-05-19T19:33:20Z`), or null when nothing
438 + * is scheduled.
439 + *
440 + * Returned alongside the dispatch result so callers immediately know when
441 + * the build they queued (or the one already pending) will actually run,
442 + * without a second round-trip. Null in the `running` case when the lock is
443 + * held but no future tick is queued.
444 + *
445 + * ISO 8601 with the explicit `Z` zone designator (not `human_time_diff()`,
446 + * not a bare "Y-m-d H:i:s") so the timezone is unambiguous and the value is
447 + * locale-stable and machine-parseable — the same format the `sitemaps[]`
448 + * `lastmod` values use in `get-status`.
449 + */
450 + protected static function get_next_scheduled_at(): ?string {
451 + $timestamp = wp_next_scheduled( self::CRON_HOOK );
452 + if ( false === $timestamp ) {
453 + return null;
454 + }
455 + return gmdate( 'Y-m-d\TH:i:s\Z', $timestamp );
456 + }
457 +}