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