| 1 |
<?php |
| 2 |
/** |
| 3 |
* Sitemap API Endpoints Class |
| 4 |
* |
| 5 |
* REST API endpoints for XML sitemap management including generation, |
| 6 |
* validation, status monitoring, and search engine submission with |
| 7 |
* proper authentication and comprehensive error handling. |
| 8 |
* |
| 9 |
* @package ThinkRank |
| 10 |
* @subpackage API |
| 11 |
* @since 1.0.0 |
| 12 |
*/ |
| 13 |
|
| 14 |
declare(strict_types=1); |
| 15 |
|
| 16 |
namespace ThinkRank\API; |
| 17 |
|
| 18 |
// Prevent direct access |
| 19 |
if (!defined('ABSPATH')) { |
| 20 |
exit; |
| 21 |
} |
| 22 |
|
| 23 |
use ThinkRank\SEO\Sitemap_Generator; |
| 24 |
use ThinkRank\API\Traits\CSRF_Protection; |
| 25 |
use ThinkRank\API\Traits\Context_Authorization; |
| 26 |
use WP_REST_Controller; |
| 27 |
use WP_REST_Request; |
| 28 |
use WP_REST_Response; |
| 29 |
use WP_Error; |
| 30 |
|
| 31 |
// Prevent direct access |
| 32 |
if (!defined('ABSPATH')) { |
| 33 |
exit; |
| 34 |
} |
| 35 |
|
| 36 |
// Load CSRF Protection trait |
| 37 |
require_once THINKRANK_PLUGIN_DIR . 'includes/api/traits/trait-csrf-protection.php'; |
| 38 |
require_once THINKRANK_PLUGIN_DIR . 'includes/api/traits/trait-context-authorization.php'; |
| 39 |
|
| 40 |
/** |
| 41 |
* Sitemap API Endpoints Class |
| 42 |
* |
| 43 |
* Provides REST API endpoints for sitemap operations including |
| 44 |
* XML generation, validation, status monitoring, and search engine |
| 45 |
* submission with proper authentication and validation. |
| 46 |
* |
| 47 |
* @since 1.0.0 |
| 48 |
*/ |
| 49 |
class Sitemap_Endpoint extends WP_REST_Controller { |
| 50 |
use CSRF_Protection; |
| 51 |
use Context_Authorization; |
| 52 |
|
| 53 |
/** |
| 54 |
* Sitemap Generator instance |
| 55 |
* |
| 56 |
* @since 1.0.0 |
| 57 |
* @var Sitemap_Generator |
| 58 |
*/ |
| 59 |
private Sitemap_Generator $sitemap_generator; |
| 60 |
|
| 61 |
/** |
| 62 |
* API namespace |
| 63 |
* |
| 64 |
* @since 1.0.0 |
| 65 |
* @var string |
| 66 |
*/ |
| 67 |
protected $namespace = 'thinkrank/v1'; |
| 68 |
|
| 69 |
/** |
| 70 |
* API resource base |
| 71 |
* |
| 72 |
* @since 1.0.0 |
| 73 |
* @var string |
| 74 |
*/ |
| 75 |
protected $rest_base = 'sitemap'; |
| 76 |
|
| 77 |
/** |
| 78 |
* Constructor |
| 79 |
* |
| 80 |
* @since 1.0.0 |
| 81 |
*/ |
| 82 |
public function __construct() { |
| 83 |
$this->sitemap_generator = new Sitemap_Generator(); |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Register API routes |
| 88 |
* |
| 89 |
* @since 1.0.0 |
| 90 |
*/ |
| 91 |
public function register_routes(): void { |
| 92 |
// Generate XML sitemap |
| 93 |
register_rest_route( |
| 94 |
$this->namespace, |
| 95 |
'/' . $this->rest_base . '/generate', |
| 96 |
[ |
| 97 |
[ |
| 98 |
'methods' => 'POST', |
| 99 |
'callback' => [$this, 'generate_sitemap'], |
| 100 |
'permission_callback' => [$this, 'check_manage_permissions'], |
| 101 |
'args' => $this->get_generate_args() |
| 102 |
] |
| 103 |
] |
| 104 |
); |
| 105 |
|
| 106 |
// Validate sitemap (read-only operation, no CSRF needed) |
| 107 |
register_rest_route( |
| 108 |
$this->namespace, |
| 109 |
'/' . $this->rest_base . '/validate', |
| 110 |
[ |
| 111 |
[ |
| 112 |
'methods' => 'POST', |
| 113 |
'callback' => [$this, 'validate_sitemap'], |
| 114 |
'permission_callback' => [$this, 'check_read_permissions'], |
| 115 |
'args' => $this->get_validate_args() |
| 116 |
] |
| 117 |
] |
| 118 |
); |
| 119 |
|
| 120 |
// Get sitemap status |
| 121 |
register_rest_route( |
| 122 |
$this->namespace, |
| 123 |
'/' . $this->rest_base . '/status', |
| 124 |
[ |
| 125 |
[ |
| 126 |
'methods' => 'GET', |
| 127 |
'callback' => [$this, 'get_sitemap_status'], |
| 128 |
'permission_callback' => [$this, 'check_read_permissions'] |
| 129 |
] |
| 130 |
] |
| 131 |
); |
| 132 |
|
| 133 |
// Submit sitemap to search engines |
| 134 |
register_rest_route( |
| 135 |
$this->namespace, |
| 136 |
'/' . $this->rest_base . '/submit', |
| 137 |
[ |
| 138 |
[ |
| 139 |
'methods' => 'POST', |
| 140 |
'callback' => [$this, 'submit_sitemap'], |
| 141 |
'permission_callback' => [$this, 'check_manage_permissions'], |
| 142 |
'args' => $this->get_submit_args() |
| 143 |
] |
| 144 |
] |
| 145 |
); |
| 146 |
|
| 147 |
// Ping search engines (unified endpoint for manual ping button) |
| 148 |
register_rest_route( |
| 149 |
$this->namespace, |
| 150 |
'/' . $this->rest_base . '/ping', |
| 151 |
[ |
| 152 |
[ |
| 153 |
'methods' => 'POST', |
| 154 |
'callback' => [$this, 'ping_search_engines'], |
| 155 |
'permission_callback' => [$this, 'check_manage_permissions'] |
| 156 |
] |
| 157 |
] |
| 158 |
); |
| 159 |
|
| 160 |
// Get sitemap statistics |
| 161 |
register_rest_route( |
| 162 |
$this->namespace, |
| 163 |
'/' . $this->rest_base . '/stats', |
| 164 |
[ |
| 165 |
[ |
| 166 |
'methods' => 'GET', |
| 167 |
'callback' => [$this, 'get_sitemap_stats'], |
| 168 |
'permission_callback' => [$this, 'check_read_permissions'] |
| 169 |
] |
| 170 |
] |
| 171 |
); |
| 172 |
|
| 173 |
// Sitemap settings management (following Site Identity pattern) |
| 174 |
register_rest_route( |
| 175 |
$this->namespace, |
| 176 |
'/' . $this->rest_base . '/settings', |
| 177 |
[ |
| 178 |
[ |
| 179 |
'methods' => 'GET', |
| 180 |
'callback' => [$this, 'get_sitemap_settings'], |
| 181 |
'permission_callback' => [$this, 'check_read_permissions'], |
| 182 |
'args' => $this->get_context_route_args() |
| 183 |
], |
| 184 |
[ |
| 185 |
'methods' => 'POST', |
| 186 |
'callback' => [$this, 'update_sitemap_settings'], |
| 187 |
'permission_callback' => [$this, 'check_manage_permissions'], |
| 188 |
'args' => $this->get_settings_args() |
| 189 |
] |
| 190 |
] |
| 191 |
); |
| 192 |
|
| 193 |
// Get custom post types |
| 194 |
register_rest_route( |
| 195 |
$this->namespace, |
| 196 |
'/' . $this->rest_base . '/custom-post-types', |
| 197 |
[ |
| 198 |
[ |
| 199 |
'methods' => 'GET', |
| 200 |
'callback' => [$this, 'get_custom_post_types'], |
| 201 |
'permission_callback' => [$this, 'check_read_permissions'] |
| 202 |
] |
| 203 |
] |
| 204 |
); |
| 205 |
|
| 206 |
// Get sitemap URLs for robots.txt integration |
| 207 |
register_rest_route( |
| 208 |
$this->namespace, |
| 209 |
'/' . $this->rest_base . '/robots-urls', |
| 210 |
[ |
| 211 |
[ |
| 212 |
'methods' => 'GET', |
| 213 |
'callback' => [$this, 'get_robots_sitemap_urls'], |
| 214 |
'permission_callback' => [$this, 'check_read_permissions'] |
| 215 |
] |
| 216 |
] |
| 217 |
); |
| 218 |
|
| 219 |
// Get WooCommerce status |
| 220 |
register_rest_route( |
| 221 |
$this->namespace, |
| 222 |
'/' . $this->rest_base . '/woocommerce-status', |
| 223 |
[ |
| 224 |
[ |
| 225 |
'methods' => 'GET', |
| 226 |
'callback' => [$this, 'get_woocommerce_status'], |
| 227 |
'permission_callback' => [$this, 'check_read_permissions'] |
| 228 |
] |
| 229 |
] |
| 230 |
); |
| 231 |
|
| 232 |
// Cleanup old sitemap files |
| 233 |
register_rest_route( |
| 234 |
$this->namespace, |
| 235 |
'/' . $this->rest_base . '/cleanup', |
| 236 |
[ |
| 237 |
[ |
| 238 |
'methods' => 'POST', |
| 239 |
'callback' => [$this, 'cleanup_sitemap_files'], |
| 240 |
'permission_callback' => [$this, 'check_manage_permissions'], |
| 241 |
'args' => [ |
| 242 |
'sitemap_urls' => [ |
| 243 |
'required' => false, |
| 244 |
'type' => 'array', |
| 245 |
'description' => 'Optional array of specific sitemap URLs to clean up. If not provided, scans filesystem automatically.' |
| 246 |
] |
| 247 |
] |
| 248 |
] |
| 249 |
] |
| 250 |
); |
| 251 |
} |
| 252 |
|
| 253 |
/** |
| 254 |
* Record that a sitemap generation just ran. |
| 255 |
* |
| 256 |
* Persists the `last_generated` timestamp so the admin UI can distinguish |
| 257 |
* generated sitemaps (whose files now exist on disk) from ones that are |
| 258 |
* merely configured — the "View Generated Sitemaps" links stay disabled |
| 259 |
* until this is set. |
| 260 |
* |
| 261 |
* @return string ISO-8601 timestamp stored as the `last_generated` setting. |
| 262 |
*/ |
| 263 |
private function record_generation(): string { |
| 264 |
$timestamp = gmdate('c'); |
| 265 |
$settings = $this->sitemap_generator->get_settings('site'); |
| 266 |
$settings['last_generated'] = $timestamp; |
| 267 |
$this->sitemap_generator->save_settings('site', null, $settings); |
| 268 |
|
| 269 |
// This generation wrote the same files the outstanding automatic rebuild |
| 270 |
// was queued to write, so clear its marker (and any recorded failure) |
| 271 |
// instead of leaving a request-time takeover to repeat the work. |
| 272 |
$this->sitemap_generator->mark_regeneration_complete(); |
| 273 |
|
| 274 |
return $timestamp; |
| 275 |
} |
| 276 |
|
| 277 |
/** |
| 278 |
* Record that the published sitemap files are gone. |
| 279 |
* |
| 280 |
* The inverse of {@see record_generation()}: clears `last_generated` so the |
| 281 |
* admin's "View Generated Sitemaps" links go back to disabled instead of |
| 282 |
* pointing at files that have just been deleted. |
| 283 |
* |
| 284 |
* @since 1.31.0 |
| 285 |
* @return void |
| 286 |
*/ |
| 287 |
private function clear_generation_record(): void { |
| 288 |
$settings = $this->sitemap_generator->get_settings('site'); |
| 289 |
if (empty($settings['last_generated'])) { |
| 290 |
return; |
| 291 |
} |
| 292 |
|
| 293 |
$settings['last_generated'] = ''; |
| 294 |
$this->sitemap_generator->save_settings('site', null, $settings); |
| 295 |
} |
| 296 |
|
| 297 |
/** |
| 298 |
* Stored sitemap settings with this request's options laid over them. |
| 299 |
* |
| 300 |
* The generate payload is partial — the admin screen posts the sitemap |
| 301 |
* shape, not the whole settings record — so reading `delivery_mode` straight |
| 302 |
* off it would resolve to `auto` on every ordinary request and defeat an |
| 303 |
* explicit choice. Merging keeps a mode sent in the payload authoritative |
| 304 |
* while falling back to what is saved. |
| 305 |
* |
| 306 |
* @param array $options Request options. |
| 307 |
* @return array Effective settings for this generation. |
| 308 |
*/ |
| 309 |
private function effective_settings(array $options): array { |
| 310 |
return array_merge($this->sitemap_generator->get_settings('site'), $options); |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* Persist a manual-generation auto-promotion into the stored settings. |
| 315 |
* |
| 316 |
* maybe_promote_to_index() may flip use_sitemap_index on and synthesize the |
| 317 |
* segmented sitemap_urls for the current generation. On the automatic path |
| 318 |
* generate_and_save() saves that resolved state; the manual generate route |
| 319 |
* must do the same, or the next content-/settings-triggered regeneration |
| 320 |
* (which reads stored settings) reverts the site to a single flat file. |
| 321 |
* |
| 322 |
* Only the two mode-defining keys are merged, so this partial generate |
| 323 |
* payload never clobbers unrelated saved settings. |
| 324 |
* |
| 325 |
* @param array $options Options after maybe_promote_to_index(). |
| 326 |
* @return void |
| 327 |
*/ |
| 328 |
private function persist_promoted_mode(array $options): void { |
| 329 |
$saved = $this->sitemap_generator->get_settings('site'); |
| 330 |
|
| 331 |
// Record the mode that was actually written, in both directions, so the |
| 332 |
// stored settings and the files on disk cannot disagree. Persisting a |
| 333 |
// demotion used to be unsafe because an absent use_sitemap_index was |
| 334 |
// indistinguishable from an explicit "off", and treating it as off would |
| 335 |
// clobber a saved index whenever the toggle merely happened to be |
| 336 |
// missing. maybe_promote_to_index() now resolves an absent key from the |
| 337 |
// saved settings before this runs, so whatever arrives here is the |
| 338 |
// resolved decision rather than a gap in the payload. |
| 339 |
$mode = !empty($options['use_sitemap_index']); |
| 340 |
$urls = $options['sitemap_urls'] ?? ($saved['sitemap_urls'] ?? null); |
| 341 |
|
| 342 |
$mode_unchanged = $mode === !empty($saved['use_sitemap_index']); |
| 343 |
$urls_unchanged = $urls === ($saved['sitemap_urls'] ?? null); |
| 344 |
|
| 345 |
if ($mode_unchanged && $urls_unchanged) { |
| 346 |
return; |
| 347 |
} |
| 348 |
|
| 349 |
$saved['use_sitemap_index'] = $mode; |
| 350 |
if (isset($options['sitemap_urls'])) { |
| 351 |
$saved['sitemap_urls'] = $options['sitemap_urls']; |
| 352 |
} |
| 353 |
$this->sitemap_generator->save_settings('site', null, $saved); |
| 354 |
} |
| 355 |
|
| 356 |
/** |
| 357 |
* Write the sitemap files when the site has none yet. |
| 358 |
* |
| 359 |
* The sitemap is served as a static file in the web root, so a site whose |
| 360 |
* sitemap is enabled but never generated serves nothing at /sitemap.xml — |
| 361 |
* WordPress core then claims that URL and redirects to wp-sitemap.xml. |
| 362 |
* Turning the sitemap on therefore has to produce the file, which is what |
| 363 |
* the Setup Wizard's "Save & Continue" relies on for its "View Sitemap" |
| 364 |
* link. Only fills the gap: an existing file is left to the explicit |
| 365 |
* "Generate" action so saving settings stays cheap on large sites. |
| 366 |
* |
| 367 |
* @since 1.17.0 |
| 368 |
* @param string $context_type Settings context type. |
| 369 |
* @param int|null $context_id Settings context id. |
| 370 |
* @return string Sitemap URL, or an empty string when nothing is published. |
| 371 |
*/ |
| 372 |
private function ensure_sitemap_file(string $context_type, ?int $context_id, bool &$generated_now = false): string { |
| 373 |
$generated_now = false; |
| 374 |
|
| 375 |
if ($context_type !== 'site') { |
| 376 |
return ''; |
| 377 |
} |
| 378 |
|
| 379 |
$settings = $this->sitemap_generator->get_settings($context_type, $context_id); |
| 380 |
|
| 381 |
if (empty($settings['enabled'])) { |
| 382 |
return ''; |
| 383 |
} |
| 384 |
|
| 385 |
$sitemap_url = $this->sitemap_generator->get_primary_sitemap_url($settings); |
| 386 |
|
| 387 |
// Dynamic delivery publishes no file, so there is nothing to ensure and |
| 388 |
// nothing to look for on disk. Dropping the rendered documents is what |
| 389 |
// makes the saved settings take effect on the next request (#752). |
| 390 |
if ('dynamic' === $this->sitemap_generator->resolve_delivery_mode($settings)) { |
| 391 |
$this->sitemap_generator->flush_dynamic_cache(); |
| 392 |
|
| 393 |
return $sitemap_url; |
| 394 |
} |
| 395 |
|
| 396 |
if ($this->sitemap_generator->primary_sitemap_file_exists($settings)) { |
| 397 |
return $sitemap_url; |
| 398 |
} |
| 399 |
|
| 400 |
// Never fail the settings save over generation: the settings are already |
| 401 |
// persisted, and content changes or a manual Generate will retry. |
| 402 |
try { |
| 403 |
if (!$this->sitemap_generator->generate_and_save($settings)) { |
| 404 |
return ''; |
| 405 |
} |
| 406 |
$generated_now = true; |
| 407 |
} catch (\Throwable $e) { |
| 408 |
return ''; |
| 409 |
} |
| 410 |
|
| 411 |
return $sitemap_url; |
| 412 |
} |
| 413 |
|
| 414 |
/** |
| 415 |
* Generate XML sitemap |
| 416 |
* |
| 417 |
* @since 1.0.0 |
| 418 |
* |
| 419 |
* @param WP_REST_Request $request Request object |
| 420 |
* @return WP_REST_Response|WP_Error Response object or error |
| 421 |
*/ |
| 422 |
public function generate_sitemap(WP_REST_Request $request) { |
| 423 |
try { |
| 424 |
// Rate limiting: Max 3 generations per 5 minutes per user |
| 425 |
if (!$this->check_rate_limit()) { |
| 426 |
return new WP_Error( |
| 427 |
'rate_limit_exceeded', |
| 428 |
'Too many sitemap generation requests. Please wait before trying again.', |
| 429 |
['status' => 429] |
| 430 |
); |
| 431 |
} |
| 432 |
|
| 433 |
// Concurrent generation protection |
| 434 |
if (!$this->acquire_generation_lock()) { |
| 435 |
return new WP_Error( |
| 436 |
'generation_in_progress', |
| 437 |
'Sitemap generation is already in progress. Please wait.', |
| 438 |
['status' => 409] |
| 439 |
); |
| 440 |
} |
| 441 |
|
| 442 |
$options = $request->get_param('options') ?? []; |
| 443 |
if (!is_array($options)) { |
| 444 |
$options = []; |
| 445 |
} |
| 446 |
// sitemap_urls must be an array wherever it is counted/iterated below |
| 447 |
// (and in the generator); drop a wrong-typed value so a malformed |
| 448 |
// request yields normal output instead of an uncaught TypeError. |
| 449 |
if (isset($options['sitemap_urls']) && !is_array($options['sitemap_urls'])) { |
| 450 |
unset($options['sitemap_urls']); |
| 451 |
} |
| 452 |
|
| 453 |
// Resolve index-vs-single mode from the use_sitemap_index toggle |
| 454 |
// (synthesizing child sitemaps when the toggle is on but none are |
| 455 |
// configured, and auto-promoting an oversized single file), rather |
| 456 |
// than deciding purely by how many sitemap_urls happen to be present. |
| 457 |
$options = $this->sitemap_generator->maybe_promote_to_index($options); |
| 458 |
|
| 459 |
// Persist the resolved index-mode decision so a later content- or |
| 460 |
// settings-triggered regeneration (which reads stored settings) |
| 461 |
// doesn't revert a manual auto-promotion back to a single flat file. |
| 462 |
// generate_and_save() already does this on the automatic path; the |
| 463 |
// manual generate route must match it. |
| 464 |
$this->persist_promoted_mode($options); |
| 465 |
|
| 466 |
// Dynamic delivery answers the sitemap URLs from PHP, so there is |
| 467 |
// nothing to write. Every other sitemap write path already returns |
| 468 |
// early here (class-sitemap-generator.php:2189 and :2313); this one |
| 469 |
// did not, which broke the feature from both directions: on a |
| 470 |
// read-only root — the case dynamic delivery exists for — the button |
| 471 |
// reported 500 "Failed to save sitemap: sitemap.xml" while the URL |
| 472 |
// was serving correctly, and on a writable root with dynamic chosen |
| 473 |
// explicitly it wrote files the web server then served in place of |
| 474 |
// the dynamic route. |
| 475 |
// |
| 476 |
// Regenerating here means dropping the rendered documents so the |
| 477 |
// next request rebuilds them, and clearing any file left behind by |
| 478 |
// an earlier static generation for the same reason. |
| 479 |
if ('dynamic' === $this->sitemap_generator->resolve_delivery_mode($this->effective_settings($options))) { |
| 480 |
$this->sitemap_generator->flush_dynamic_cache(); |
| 481 |
|
| 482 |
$removal = $this->sitemap_generator->delete_published_sitemaps(); |
| 483 |
$stuck = is_array($removal['failed'] ?? null) ? $removal['failed'] : []; |
| 484 |
|
| 485 |
// A stale file shadows the dynamic route, so this is a real |
| 486 |
// failure rather than a tidy-up that did not matter. Worded by |
| 487 |
// the generator so this and its own rebuild paths cannot |
| 488 |
// describe the same stuck files differently (#764). |
| 489 |
if (!empty($stuck)) { |
| 490 |
return new WP_Error( |
| 491 |
'sitemap_stale_files', |
| 492 |
$this->sitemap_generator->stuck_files_message($stuck), |
| 493 |
['status' => 500] |
| 494 |
); |
| 495 |
} |
| 496 |
|
| 497 |
// last_generated deliberately stays untouched: it means "these |
| 498 |
// files are on disk", and primary_sitemap_file_exists() callers |
| 499 |
// rely on that. clear_generation_record() drops a value left |
| 500 |
// over from a previous static generation, so the admin stops |
| 501 |
// linking to files that no longer exist. |
| 502 |
$this->clear_generation_record(); |
| 503 |
$this->sitemap_generator->mark_regeneration_complete(); |
| 504 |
|
| 505 |
return new WP_REST_Response([ |
| 506 |
'success' => true, |
| 507 |
'data' => [ |
| 508 |
'delivery_mode' => 'dynamic', |
| 509 |
'sitemap_url' => $this->sitemap_generator->get_primary_sitemap_url(), |
| 510 |
'generated_at' => gmdate('c'), |
| 511 |
'last_generated' => '', |
| 512 |
], |
| 513 |
'message' => __('Sitemap refreshed. WordPress serves it directly, so no files were written.', 'thinkrank'), |
| 514 |
]); |
| 515 |
} |
| 516 |
|
| 517 |
// Check if an index (multiple sitemaps) is configured |
| 518 |
if (!empty($options['use_sitemap_index']) || (!empty($options['sitemap_urls']) && count($options['sitemap_urls']) > 1)) { |
| 519 |
// Generate multiple sitemaps |
| 520 |
$results = $this->sitemap_generator->generate_multiple_sitemaps($options); |
| 521 |
|
| 522 |
if (!$results['success']) { |
| 523 |
return new WP_Error( |
| 524 |
'sitemap_generation_failed', |
| 525 |
'Failed to generate sitemaps: ' . implode(', ', $results['errors']), |
| 526 |
['status' => 500] |
| 527 |
); |
| 528 |
} |
| 529 |
|
| 530 |
return new WP_REST_Response([ |
| 531 |
'success' => true, |
| 532 |
'message' => 'Multiple sitemaps generated successfully', |
| 533 |
'data' => [ |
| 534 |
'sitemaps_generated' => $results['sitemaps_generated'], |
| 535 |
'total_sitemaps' => count($results['sitemaps_generated']), |
| 536 |
'url_count' => $results['total_urls'], |
| 537 |
'last_generated' => $this->record_generation() |
| 538 |
] |
| 539 |
]); |
| 540 |
} else { |
| 541 |
// Generate single sitemap (backward compatibility) |
| 542 |
$sitemap_xml = $this->sitemap_generator->generate_sitemap($options); |
| 543 |
|
| 544 |
// Save sitemap to file (optional) |
| 545 |
$save_to_file = $request->get_param('save_to_file') ?? true; |
| 546 |
$last_generated = ''; |
| 547 |
if ($save_to_file) { |
| 548 |
$filename = 'sitemap.xml'; |
| 549 |
if (!empty($options['sitemap_urls'][0]['url'])) { |
| 550 |
$filename = basename(wp_parse_url($options['sitemap_urls'][0]['url'], PHP_URL_PATH)); |
| 551 |
} |
| 552 |
// A failed write has to surface here the way the index |
| 553 |
// branch surfaces one. Discarding it let record_generation() |
| 554 |
// advance last_generated and clear the pending marker and |
| 555 |
// the recorded failure, so an unwritable site root — the |
| 556 |
// exact case this endpoint reports health for — came back |
| 557 |
// as a healthy "Generated successfully". |
| 558 |
if (!$this->save_sitemap_file($sitemap_xml, $filename)) { |
| 559 |
return new WP_Error( |
| 560 |
'sitemap_generation_failed', |
| 561 |
'Failed to save sitemap: ' . $filename, |
| 562 |
['status' => 500] |
| 563 |
); |
| 564 |
} |
| 565 |
|
| 566 |
// Regenerate the standalone local business sitemap on the |
| 567 |
// single-sitemap path too (parity with Rank Math). |
| 568 |
$this->sitemap_generator->regenerate_local_sitemap($options); |
| 569 |
|
| 570 |
// Only record generation when the files were actually |
| 571 |
// written — a preview (save_to_file=false) must not enable |
| 572 |
// the "View Generated Sitemaps" links. |
| 573 |
$last_generated = $this->record_generation(); |
| 574 |
} |
| 575 |
|
| 576 |
return new WP_REST_Response([ |
| 577 |
'success' => true, |
| 578 |
'data' => [ |
| 579 |
'sitemap_xml' => $sitemap_xml, |
| 580 |
'sitemap_url' => home_url('/sitemap.xml'), |
| 581 |
'generated_at' => gmdate('c'), |
| 582 |
'url_count' => $this->count_urls_in_xml($sitemap_xml), |
| 583 |
'last_generated' => $last_generated |
| 584 |
], |
| 585 |
'message' => 'Sitemap generated successfully' |
| 586 |
]); |
| 587 |
} |
| 588 |
|
| 589 |
} catch (\Exception $e) { |
| 590 |
$this->release_generation_lock(); |
| 591 |
return new WP_Error( |
| 592 |
'generation_failed', |
| 593 |
'Sitemap generation failed: ' . $e->getMessage(), |
| 594 |
['status' => 500] |
| 595 |
); |
| 596 |
} finally { |
| 597 |
$this->release_generation_lock(); |
| 598 |
} |
| 599 |
} |
| 600 |
|
| 601 |
/** |
| 602 |
* Validate sitemap |
| 603 |
* |
| 604 |
* @since 1.0.0 |
| 605 |
* |
| 606 |
* @param WP_REST_Request $request Request object |
| 607 |
* @return WP_REST_Response|WP_Error Response object or error |
| 608 |
*/ |
| 609 |
public function validate_sitemap(WP_REST_Request $request) { |
| 610 |
try { |
| 611 |
$sitemap_url = $request->get_param('sitemap_url') ?? home_url('/sitemap.xml'); |
| 612 |
|
| 613 |
// Validate sitemap URL. Url_Validator accepts an internationalised |
| 614 |
// domain or a non-ASCII path (home_url() on an IDN site is one). |
| 615 |
if (!\ThinkRank\Core\Url_Validator::is_valid($sitemap_url)) { |
| 616 |
return new WP_Error( |
| 617 |
'invalid_url', |
| 618 |
'Invalid sitemap URL provided', |
| 619 |
['status' => 400] |
| 620 |
); |
| 621 |
} |
| 622 |
|
| 623 |
// Everything from here on works on the ASCII form (punycode host, |
| 624 |
// percent-encoded path). wp_http_validate_url() resolves the host |
| 625 |
// with gethostbyname(), which cannot resolve a Unicode name, and |
| 626 |
// the SSRF check must see exactly the URL that is fetched. |
| 627 |
$sitemap_url = \ThinkRank\Core\Url_Validator::to_ascii((string) $sitemap_url); |
| 628 |
|
| 629 |
// Block SSRF: this endpoint fetches the URL server-side, so reject |
| 630 |
// loopback/link-local/private hosts and non-http(s) schemes via |
| 631 |
// WordPress's own validator (same guard used in class-schema-endpoint). |
| 632 |
if (!wp_http_validate_url($sitemap_url)) { |
| 633 |
return new WP_Error( |
| 634 |
'invalid_url', |
| 635 |
'The sitemap URL is not allowed.', |
| 636 |
['status' => 400] |
| 637 |
); |
| 638 |
} |
| 639 |
|
| 640 |
// Perform validation |
| 641 |
$validation_result = $this->perform_sitemap_validation($sitemap_url); |
| 642 |
|
| 643 |
return new WP_REST_Response([ |
| 644 |
'success' => true, |
| 645 |
'data' => $validation_result, |
| 646 |
'message' => 'Sitemap validation completed' |
| 647 |
], 200); |
| 648 |
|
| 649 |
} catch (\Exception $e) { |
| 650 |
return new WP_Error( |
| 651 |
'validation_failed', |
| 652 |
'Sitemap validation failed: ' . $e->getMessage(), |
| 653 |
['status' => 500] |
| 654 |
); |
| 655 |
} |
| 656 |
} |
| 657 |
|
| 658 |
/** |
| 659 |
* Get sitemap status |
| 660 |
* |
| 661 |
* @since 1.0.0 |
| 662 |
* |
| 663 |
* @param WP_REST_Request $request Request object |
| 664 |
* @return WP_REST_Response|WP_Error Response object or error |
| 665 |
*/ |
| 666 |
public function get_sitemap_status(WP_REST_Request $request) { |
| 667 |
try { |
| 668 |
// Get sitemap output data from generator |
| 669 |
$status_data = $this->sitemap_generator->get_output_data('site', null); |
| 670 |
|
| 671 |
// Add additional status information |
| 672 |
$sitemap_file_path = ABSPATH . 'sitemap.xml'; |
| 673 |
$status_data['file_exists'] = file_exists($sitemap_file_path); |
| 674 |
$status_data['file_size'] = $status_data['file_exists'] ? filesize($sitemap_file_path) : 0; |
| 675 |
$status_data['file_modified'] = $status_data['file_exists'] ? gmdate('c', filemtime($sitemap_file_path)) : null; |
| 676 |
|
| 677 |
return new WP_REST_Response([ |
| 678 |
'success' => true, |
| 679 |
'data' => $status_data, |
| 680 |
'message' => 'Sitemap status retrieved successfully' |
| 681 |
], 200); |
| 682 |
|
| 683 |
} catch (\Exception $e) { |
| 684 |
return new WP_Error( |
| 685 |
'status_failed', |
| 686 |
'Failed to get sitemap status: ' . $e->getMessage(), |
| 687 |
['status' => 500] |
| 688 |
); |
| 689 |
} |
| 690 |
} |
| 691 |
|
| 692 |
/** |
| 693 |
* Submit sitemap to search engines |
| 694 |
* |
| 695 |
* @since 1.0.0 |
| 696 |
* |
| 697 |
* @param WP_REST_Request $request Request object |
| 698 |
* @return WP_REST_Response|WP_Error Response object or error |
| 699 |
*/ |
| 700 |
public function submit_sitemap(WP_REST_Request $request) { |
| 701 |
// Google removed its sitemap-ping endpoint in 2023 and Bing followed suit; |
| 702 |
// both now discover sitemaps via robots.txt on their own schedule. There |
| 703 |
// is nothing to submit, so this is a no-op kept only so existing clients |
| 704 |
// don't 404 (mirrors ping_search_engines()). |
| 705 |
return new WP_REST_Response([ |
| 706 |
'success' => true, |
| 707 |
'data' => [], |
| 708 |
'message' => 'Search engines no longer accept sitemap submission; sitemaps are discovered automatically via robots.txt.', |
| 709 |
], 200); |
| 710 |
} |
| 711 |
|
| 712 |
/** |
| 713 |
* Ping search engines about sitemap updates (unified method) |
| 714 |
* |
| 715 |
* @since 1.0.0 |
| 716 |
* |
| 717 |
* @param WP_REST_Request $request Request object |
| 718 |
* @return WP_REST_Response|WP_Error Response object or error |
| 719 |
*/ |
| 720 |
public function ping_search_engines(WP_REST_Request $request) { |
| 721 |
// Google removed its sitemap-ping endpoint in 2023 and Bing followed suit; |
| 722 |
// both now rely on the sitemap being referenced from robots.txt and pulled |
| 723 |
// on their own schedule. There is nothing left to ping, so this endpoint is |
| 724 |
// a no-op kept only so existing clients don't 404. |
| 725 |
return new WP_REST_Response([ |
| 726 |
'success' => true, |
| 727 |
'message' => 'Search engines no longer support sitemap ping; sitemaps are discovered automatically via robots.txt.', |
| 728 |
'engines' => [], |
| 729 |
'timestamp' => gmdate('c') |
| 730 |
], 200); |
| 731 |
} |
| 732 |
|
| 733 |
/** |
| 734 |
* Get sitemap statistics |
| 735 |
* |
| 736 |
* @since 1.0.0 |
| 737 |
* |
| 738 |
* @param WP_REST_Request $request Request object |
| 739 |
* @return WP_REST_Response|WP_Error Response object or error |
| 740 |
*/ |
| 741 |
public function get_sitemap_stats(WP_REST_Request $request) { |
| 742 |
try { |
| 743 |
$settings = $this->sitemap_generator->get_settings('site'); |
| 744 |
|
| 745 |
$stats = [ |
| 746 |
'total_urls' => $this->sitemap_generator->count_sitemap_urls($settings), |
| 747 |
'post_count' => $settings['include_posts'] ? wp_count_posts('post')->publish : 0, |
| 748 |
'page_count' => $settings['include_pages'] ? wp_count_posts('page')->publish : 0, |
| 749 |
'category_count' => $settings['include_categories'] ? wp_count_terms('category') : 0, |
| 750 |
'tag_count' => $settings['include_tags'] ? wp_count_terms('post_tag') : 0, |
| 751 |
'last_generated' => $settings['last_generated'] ?? null, |
| 752 |
'sitemap_enabled' => $settings['enabled'] ?? true |
| 753 |
]; |
| 754 |
|
| 755 |
return new WP_REST_Response([ |
| 756 |
'success' => true, |
| 757 |
'data' => $stats, |
| 758 |
'message' => 'Sitemap statistics retrieved successfully' |
| 759 |
], 200); |
| 760 |
|
| 761 |
} catch (\Throwable $e) { |
| 762 |
return new WP_Error( |
| 763 |
'stats_failed', |
| 764 |
'Failed to get sitemap statistics: ' . $e->getMessage(), |
| 765 |
['status' => 500] |
| 766 |
); |
| 767 |
} |
| 768 |
} |
| 769 |
|
| 770 |
/** |
| 771 |
* Check read permissions |
| 772 |
* |
| 773 |
* @since 1.0.0 |
| 774 |
* |
| 775 |
* @return bool Permission status |
| 776 |
*/ |
| 777 |
public function check_read_permissions(): bool { |
| 778 |
return current_user_can('edit_posts'); |
| 779 |
} |
| 780 |
|
| 781 |
/** |
| 782 |
* Check manage permissions for the state-changing routes. |
| 783 |
* |
| 784 |
* Every route using this callback is a POST that writes something — |
| 785 |
* /generate, /submit, /ping, /settings, /cleanup — so it is nonce-gated as |
| 786 |
* well as capability-gated, matching Schema_Endpoint, Setup_Wizard_Endpoint |
| 787 |
* and Email_Report_Endpoint. The class already `use`d CSRF_Protection but |
| 788 |
* never called it, leaving this controller the odd one out. |
| 789 |
* |
| 790 |
* @since 1.0.0 |
| 791 |
* |
| 792 |
* @param WP_REST_Request $request Request object |
| 793 |
* @return bool|WP_Error Permission status |
| 794 |
*/ |
| 795 |
public function check_manage_permissions(WP_REST_Request $request) { |
| 796 |
if (!\ThinkRank\Core\Capability_Manager::current_user_can('thinkrank_crawling')) { |
| 797 |
return new WP_Error( |
| 798 |
'rest_forbidden', |
| 799 |
__('You do not have permission to manage sitemaps.', 'thinkrank'), |
| 800 |
['status' => 403] |
| 801 |
); |
| 802 |
} |
| 803 |
|
| 804 |
if (!$this->verify_request_nonce($request)) { |
| 805 |
return new WP_Error( |
| 806 |
'rest_forbidden', |
| 807 |
__('Invalid security token. Please refresh the page and try again.', 'thinkrank'), |
| 808 |
['status' => 403] |
| 809 |
); |
| 810 |
} |
| 811 |
|
| 812 |
return true; |
| 813 |
} |
| 814 |
|
| 815 |
/** |
| 816 |
* Save sitemap to file |
| 817 |
* |
| 818 |
* @since 1.0.0 |
| 819 |
* |
| 820 |
* @param string $sitemap_xml Sitemap XML content |
| 821 |
* @param string $filename Optional. Filename to save (defaults to 'sitemap.xml') |
| 822 |
* @return bool Success status |
| 823 |
*/ |
| 824 |
private function save_sitemap_file(string $sitemap_xml, string $filename = 'sitemap.xml'): bool { |
| 825 |
// Clean filename and ensure it ends with .xml |
| 826 |
$filename = sanitize_file_name($filename); |
| 827 |
if (!str_ends_with($filename, '.xml')) { |
| 828 |
$filename .= '.xml'; |
| 829 |
} |
| 830 |
|
| 831 |
$sitemap_file_path = ABSPATH . $filename; |
| 832 |
|
| 833 |
// Use WordPress filesystem API |
| 834 |
global $wp_filesystem; |
| 835 |
if (empty($wp_filesystem)) { |
| 836 |
require_once ABSPATH . '/wp-admin/includes/file.php'; |
| 837 |
WP_Filesystem(); |
| 838 |
} |
| 839 |
|
| 840 |
return $wp_filesystem->put_contents($sitemap_file_path, $sitemap_xml, FS_CHMOD_FILE); |
| 841 |
} |
| 842 |
|
| 843 |
/** |
| 844 |
* Count URLs in sitemap XML content |
| 845 |
* |
| 846 |
* @since 1.0.0 |
| 847 |
* |
| 848 |
* @param string $sitemap_xml Sitemap XML content |
| 849 |
* @return int URL count |
| 850 |
*/ |
| 851 |
private function count_urls_in_xml(string $sitemap_xml): int { |
| 852 |
return substr_count($sitemap_xml, '<url>'); |
| 853 |
} |
| 854 |
|
| 855 |
/** |
| 856 |
* Perform sitemap validation |
| 857 |
* |
| 858 |
* @since 1.0.0 |
| 859 |
* |
| 860 |
* @param string $sitemap_url Sitemap URL to validate |
| 861 |
* @return array Validation results |
| 862 |
*/ |
| 863 |
private function perform_sitemap_validation(string $sitemap_url): array { |
| 864 |
$validation_result = [ |
| 865 |
'valid' => true, |
| 866 |
'errors' => [], |
| 867 |
'warnings' => [], |
| 868 |
'url_count' => 0, |
| 869 |
'file_size' => 0 |
| 870 |
]; |
| 871 |
|
| 872 |
// Check if sitemap is accessible. wp_safe_remote_get() re-applies the |
| 873 |
// reject-unsafe-URLs / external-host filters (incl. on redirects) so an |
| 874 |
// internal host can't be reached even if it slipped past validation. |
| 875 |
$response = wp_safe_remote_get($sitemap_url, ['timeout' => 30]); |
| 876 |
|
| 877 |
if (is_wp_error($response)) { |
| 878 |
$validation_result['valid'] = false; |
| 879 |
$validation_result['errors'][] = 'Sitemap is not accessible: ' . $response->get_error_message(); |
| 880 |
return $validation_result; |
| 881 |
} |
| 882 |
|
| 883 |
$status_code = wp_remote_retrieve_response_code($response); |
| 884 |
if ($status_code !== 200) { |
| 885 |
$validation_result['valid'] = false; |
| 886 |
$validation_result['errors'][] = "Sitemap returned HTTP status code: {$status_code}"; |
| 887 |
return $validation_result; |
| 888 |
} |
| 889 |
|
| 890 |
$sitemap_content = wp_remote_retrieve_body($response); |
| 891 |
$validation_result['file_size'] = strlen($sitemap_content); |
| 892 |
$validation_result['url_count'] = $this->count_urls_in_xml($sitemap_content); |
| 893 |
|
| 894 |
// Basic XML validation |
| 895 |
libxml_use_internal_errors(true); |
| 896 |
$xml = simplexml_load_string($sitemap_content); |
| 897 |
|
| 898 |
if (false === $xml) { |
| 899 |
$validation_result['valid'] = false; |
| 900 |
$validation_result['errors'][] = 'Invalid XML format'; |
| 901 |
|
| 902 |
foreach (libxml_get_errors() as $error) { |
| 903 |
$validation_result['errors'][] = trim($error->message); |
| 904 |
} |
| 905 |
} |
| 906 |
|
| 907 |
// Check file size (should be under 50MB) |
| 908 |
if ($validation_result['file_size'] > 50 * 1024 * 1024) { |
| 909 |
$validation_result['warnings'][] = 'Sitemap file size exceeds 50MB limit'; |
| 910 |
} |
| 911 |
|
| 912 |
// Check URL count (should be under 50,000) |
| 913 |
if ($validation_result['url_count'] > 50000) { |
| 914 |
$validation_result['warnings'][] = 'Sitemap contains more than 50,000 URLs'; |
| 915 |
} |
| 916 |
|
| 917 |
return $validation_result; |
| 918 |
} |
| 919 |
|
| 920 |
/** |
| 921 |
* Get arguments for generate endpoint |
| 922 |
* |
| 923 |
* @since 1.0.0 |
| 924 |
* |
| 925 |
* @return array Arguments array |
| 926 |
*/ |
| 927 |
private function get_generate_args(): array { |
| 928 |
return [ |
| 929 |
'options' => [ |
| 930 |
'required' => false, |
| 931 |
'type' => 'object', |
| 932 |
'description' => 'Sitemap generation options' |
| 933 |
], |
| 934 |
'save_to_file' => [ |
| 935 |
'required' => false, |
| 936 |
'type' => 'boolean', |
| 937 |
'default' => true, |
| 938 |
'description' => 'Save sitemap to file' |
| 939 |
] |
| 940 |
]; |
| 941 |
} |
| 942 |
|
| 943 |
/** |
| 944 |
* Get arguments for validate endpoint |
| 945 |
* |
| 946 |
* @since 1.0.0 |
| 947 |
* |
| 948 |
* @return array Arguments array |
| 949 |
*/ |
| 950 |
private function get_validate_args(): array { |
| 951 |
return [ |
| 952 |
'sitemap_url' => [ |
| 953 |
'required' => false, |
| 954 |
'type' => 'string', |
| 955 |
'format' => 'uri', |
| 956 |
'default' => home_url('/sitemap.xml'), |
| 957 |
'description' => 'Sitemap URL to validate' |
| 958 |
] |
| 959 |
]; |
| 960 |
} |
| 961 |
|
| 962 |
/** |
| 963 |
* Get arguments for submit endpoint |
| 964 |
* |
| 965 |
* @since 1.0.0 |
| 966 |
* |
| 967 |
* @return array Arguments array |
| 968 |
*/ |
| 969 |
private function get_submit_args(): array { |
| 970 |
return [ |
| 971 |
'search_engines' => [ |
| 972 |
'required' => false, |
| 973 |
'type' => 'array', |
| 974 |
'items' => [ |
| 975 |
'type' => 'string', |
| 976 |
'enum' => ['google', 'bing'] |
| 977 |
], |
| 978 |
'default' => ['google', 'bing'], |
| 979 |
'description' => 'Search engines to submit to' |
| 980 |
], |
| 981 |
'sitemap_url' => [ |
| 982 |
'required' => false, |
| 983 |
'type' => 'string', |
| 984 |
'format' => 'uri', |
| 985 |
'default' => home_url('/sitemap.xml'), |
| 986 |
'description' => 'Sitemap URL to submit' |
| 987 |
] |
| 988 |
]; |
| 989 |
} |
| 990 |
|
| 991 |
/** |
| 992 |
* Get sitemap settings |
| 993 |
* |
| 994 |
* @since 1.0.0 |
| 995 |
* |
| 996 |
* @param WP_REST_Request $request Request object |
| 997 |
* @return WP_REST_Response|WP_Error Response object or error |
| 998 |
*/ |
| 999 |
public function get_sitemap_settings(WP_REST_Request $request) { |
| 1000 |
try { |
| 1001 |
// SECURITY: the settings are stored per context, so the object has |
| 1002 |
// to be authorised before it is read (#385). |
| 1003 |
$context = $this->resolve_request_context($request); |
| 1004 |
if (is_wp_error($context)) { |
| 1005 |
return $context; |
| 1006 |
} |
| 1007 |
[$context_type, $context_id] = $context; |
| 1008 |
|
| 1009 |
// Get settings from Sitemap_Generator |
| 1010 |
$settings = $this->sitemap_generator->get_settings($context_type, $context_id); |
| 1011 |
|
| 1012 |
return new WP_REST_Response([ |
| 1013 |
'success' => true, |
| 1014 |
'data' => [ |
| 1015 |
'settings' => $settings, |
| 1016 |
'context_type' => $context_type, |
| 1017 |
'context_id' => $context_id, |
| 1018 |
// Kept out of `settings` on purpose: this is generator state, |
| 1019 |
// not something the settings POST round-trips. |
| 1020 |
'health' => $context_type === 'site' |
| 1021 |
? $this->sitemap_generator->get_regeneration_health() |
| 1022 |
: null, |
| 1023 |
// `delivery_mode` in `settings` may still be 'auto', which |
| 1024 |
// only the server can resolve (it depends on whether the web |
| 1025 |
// root is writable). The admin screen needs the answer, not |
| 1026 |
// the question: a dynamic site publishes no file, so gating |
| 1027 |
// its sitemap links on `last_generated` — which dynamic |
| 1028 |
// delivery deliberately never sets — left every link |
| 1029 |
// permanently disabled. |
| 1030 |
'resolved_delivery_mode' => $context_type === 'site' |
| 1031 |
? $this->sitemap_generator->resolve_delivery_mode($settings) |
| 1032 |
: null |
| 1033 |
], |
| 1034 |
'message' => 'Sitemap settings retrieved successfully' |
| 1035 |
], 200); |
| 1036 |
|
| 1037 |
} catch (\Exception $e) { |
| 1038 |
return new WP_Error( |
| 1039 |
'settings_retrieval_failed', |
| 1040 |
'Failed to retrieve sitemap settings: ' . $e->getMessage(), |
| 1041 |
['status' => 500] |
| 1042 |
); |
| 1043 |
} |
| 1044 |
} |
| 1045 |
|
| 1046 |
/** |
| 1047 |
* Update sitemap settings |
| 1048 |
* |
| 1049 |
* @since 1.0.0 |
| 1050 |
* |
| 1051 |
* @param WP_REST_Request $request Request object |
| 1052 |
* @return WP_REST_Response|WP_Error Response object or error |
| 1053 |
*/ |
| 1054 |
public function update_sitemap_settings(WP_REST_Request $request) { |
| 1055 |
try { |
| 1056 |
$settings = $request->get_param('settings') ?? []; |
| 1057 |
|
| 1058 |
// SECURITY: this write is keyed by the context, so the object has to |
| 1059 |
// be authorised before anything is persisted (#385). |
| 1060 |
$context = $this->resolve_request_context($request); |
| 1061 |
if (is_wp_error($context)) { |
| 1062 |
return $context; |
| 1063 |
} |
| 1064 |
[$context_type, $context_id] = $context; |
| 1065 |
|
| 1066 |
if (empty($settings)) { |
| 1067 |
return new WP_Error( |
| 1068 |
'missing_settings', |
| 1069 |
'Settings data is required', |
| 1070 |
['status' => 400] |
| 1071 |
); |
| 1072 |
} |
| 1073 |
|
| 1074 |
// Save settings using Sitemap_Generator |
| 1075 |
$success = $this->sitemap_generator->save_settings($context_type, $context_id, $settings); |
| 1076 |
|
| 1077 |
if (!$success) { |
| 1078 |
return new WP_Error( |
| 1079 |
'settings_save_failed', |
| 1080 |
'Failed to save sitemap settings', |
| 1081 |
['status' => 500] |
| 1082 |
); |
| 1083 |
} |
| 1084 |
|
| 1085 |
$generated_now = false; |
| 1086 |
$sitemap_url = $this->ensure_sitemap_file($context_type, $context_id, $generated_now); |
| 1087 |
|
| 1088 |
// Rebuild the served sitemap so inclusion-rule changes take effect |
| 1089 |
// instead of waiting for a content edit (debounced against rapid |
| 1090 |
// successive saves). Skip when ensure_sitemap_file() just built a |
| 1091 |
// fresh file synchronously — otherwise we'd immediately schedule a |
| 1092 |
// second full generation of the same content. |
| 1093 |
if (!$generated_now) { |
| 1094 |
$this->sitemap_generator->schedule_regeneration(); |
| 1095 |
} |
| 1096 |
|
| 1097 |
return new WP_REST_Response([ |
| 1098 |
'success' => true, |
| 1099 |
'data' => [ |
| 1100 |
'settings' => $settings, |
| 1101 |
'context_type' => $context_type, |
| 1102 |
'context_id' => $context_id, |
| 1103 |
'sitemap_url' => $sitemap_url |
| 1104 |
], |
| 1105 |
'message' => 'Sitemap settings saved successfully' |
| 1106 |
], 200); |
| 1107 |
|
| 1108 |
} catch (\Exception $e) { |
| 1109 |
return new WP_Error( |
| 1110 |
'settings_update_failed', |
| 1111 |
'Failed to update sitemap settings: ' . $e->getMessage(), |
| 1112 |
['status' => 500] |
| 1113 |
); |
| 1114 |
} |
| 1115 |
} |
| 1116 |
|
| 1117 |
/** |
| 1118 |
* Get arguments for settings endpoints |
| 1119 |
* |
| 1120 |
* @since 1.0.0 |
| 1121 |
* |
| 1122 |
* @return array Arguments array |
| 1123 |
*/ |
| 1124 |
private function get_settings_args(): array { |
| 1125 |
return [ |
| 1126 |
'settings' => [ |
| 1127 |
'required' => true, |
| 1128 |
'type' => 'object', |
| 1129 |
'description' => 'Sitemap settings to save' |
| 1130 |
], |
| 1131 |
'context_type' => [ |
| 1132 |
'required' => false, |
| 1133 |
'type' => 'string', |
| 1134 |
'default' => 'site', |
| 1135 |
'description' => 'Context type for settings' |
| 1136 |
], |
| 1137 |
'context_id' => [ |
| 1138 |
'required' => false, |
| 1139 |
'type' => 'integer', |
| 1140 |
'description' => 'Context ID for settings' |
| 1141 |
] |
| 1142 |
]; |
| 1143 |
} |
| 1144 |
|
| 1145 |
/** |
| 1146 |
* Get custom post types for sitemap generation |
| 1147 |
* |
| 1148 |
* @since 1.0.0 |
| 1149 |
* |
| 1150 |
* @param WP_REST_Request $request Request object |
| 1151 |
* @return WP_REST_Response|WP_Error Response object or error |
| 1152 |
*/ |
| 1153 |
public function get_custom_post_types(WP_REST_Request $request) { |
| 1154 |
try { |
| 1155 |
// Get all public custom post types (excluding built-in types) |
| 1156 |
$post_types = get_post_types([ |
| 1157 |
'public' => true, |
| 1158 |
'_builtin' => false |
| 1159 |
], 'objects'); |
| 1160 |
|
| 1161 |
$custom_post_types = []; |
| 1162 |
foreach ($post_types as $post_type) { |
| 1163 |
// Skip if it's a WooCommerce product (handled separately) |
| 1164 |
if ($post_type->name === 'product') { |
| 1165 |
continue; |
| 1166 |
} |
| 1167 |
|
| 1168 |
$custom_post_types[] = [ |
| 1169 |
'name' => $post_type->name, |
| 1170 |
'label' => $post_type->label, |
| 1171 |
'singular_name' => $post_type->labels->singular_name ?? $post_type->label, |
| 1172 |
'public' => $post_type->public, |
| 1173 |
'has_archive' => $post_type->has_archive, |
| 1174 |
'count' => wp_count_posts($post_type->name)->publish ?? 0 |
| 1175 |
]; |
| 1176 |
} |
| 1177 |
|
| 1178 |
return new WP_REST_Response([ |
| 1179 |
'success' => true, |
| 1180 |
'data' => $custom_post_types, |
| 1181 |
'message' => 'Custom post types retrieved successfully' |
| 1182 |
], 200); |
| 1183 |
|
| 1184 |
} catch (\Exception $e) { |
| 1185 |
return new WP_Error( |
| 1186 |
'custom_post_types_failed', |
| 1187 |
'Failed to get custom post types: ' . $e->getMessage(), |
| 1188 |
['status' => 500] |
| 1189 |
); |
| 1190 |
} |
| 1191 |
} |
| 1192 |
|
| 1193 |
/** |
| 1194 |
* Get WooCommerce status for sitemap generation |
| 1195 |
* |
| 1196 |
* @since 1.0.0 |
| 1197 |
* |
| 1198 |
* @param WP_REST_Request $request Request object |
| 1199 |
* @return WP_REST_Response|WP_Error Response object or error |
| 1200 |
*/ |
| 1201 |
public function get_woocommerce_status(WP_REST_Request $request) { |
| 1202 |
try { |
| 1203 |
// Check if WooCommerce is active |
| 1204 |
$is_woocommerce_active = class_exists('WooCommerce') && function_exists('WC'); |
| 1205 |
|
| 1206 |
// Check if product post type exists |
| 1207 |
$product_post_type_exists = post_type_exists('product'); |
| 1208 |
|
| 1209 |
// Check if product category taxonomy exists |
| 1210 |
$product_cat_taxonomy_exists = taxonomy_exists('product_cat'); |
| 1211 |
|
| 1212 |
$status = [ |
| 1213 |
'is_active' => $is_woocommerce_active, |
| 1214 |
'product_post_type_exists' => $product_post_type_exists, |
| 1215 |
'product_cat_taxonomy_exists' => $product_cat_taxonomy_exists, |
| 1216 |
'product_count' => $product_post_type_exists ? wp_count_posts('product')->publish ?? 0 : 0, |
| 1217 |
'product_category_count' => $product_cat_taxonomy_exists ? wp_count_terms('product_cat') : 0 |
| 1218 |
]; |
| 1219 |
|
| 1220 |
return new WP_REST_Response([ |
| 1221 |
'success' => true, |
| 1222 |
'data' => $status, |
| 1223 |
'message' => 'WooCommerce status retrieved successfully' |
| 1224 |
], 200); |
| 1225 |
|
| 1226 |
} catch (\Exception $e) { |
| 1227 |
return new WP_Error( |
| 1228 |
'woocommerce_status_failed', |
| 1229 |
'Failed to get WooCommerce status: ' . $e->getMessage(), |
| 1230 |
['status' => 500] |
| 1231 |
); |
| 1232 |
} |
| 1233 |
} |
| 1234 |
|
| 1235 |
/** |
| 1236 |
* Clean up old sitemap files |
| 1237 |
* |
| 1238 |
* @since 1.0.0 |
| 1239 |
* |
| 1240 |
* @param WP_REST_Request $request Request object |
| 1241 |
* @return WP_REST_Response|WP_Error Response object or error |
| 1242 |
*/ |
| 1243 |
public function cleanup_sitemap_files(WP_REST_Request $request) { |
| 1244 |
try { |
| 1245 |
$settings = $this->sitemap_generator->get_settings('site'); |
| 1246 |
|
| 1247 |
// Delete only the files ThinkRank published. This used to glob |
| 1248 |
// ABSPATH for 'sitemap*.xml' and '*sitemap*.xml' and delete anything |
| 1249 |
// whose name contained "sitemap", which also swept up a physical |
| 1250 |
// core wp-sitemap.xml and any other plugin's sitemap sitting in the |
| 1251 |
// web root. delete_published_sitemaps() derives the name list from |
| 1252 |
// our own stored sitemap_urls (honouring a custom url pattern) plus |
| 1253 |
// the default names, and covers the -N pagination pages. |
| 1254 |
$removed = $this->sitemap_generator->delete_published_sitemaps($settings); |
| 1255 |
$cleaned_files = $removed['deleted']; |
| 1256 |
$failed_files = $removed['failed']; |
| 1257 |
|
| 1258 |
// Cleanup on its own used to leave the site with no sitemap at all |
| 1259 |
// and nothing scheduled to rebuild one: the regeneration that is |
| 1260 |
// meant to follow lives in the admin bundle, so a bare REST/MCP call |
| 1261 |
// — or a generate that then hit the rate limit or lost the |
| 1262 |
// generation lock — published nothing and 404'd indefinitely. Queue |
| 1263 |
// the rebuild here so the recovery does not depend on the caller. |
| 1264 |
$regeneration_scheduled = false; |
| 1265 |
if (!empty($settings['enabled']) && $cleaned_files) { |
| 1266 |
$this->sitemap_generator->schedule_regeneration(); |
| 1267 |
$regeneration_scheduled = true; |
| 1268 |
} |
| 1269 |
|
| 1270 |
// The files are gone, so stop reporting them as generated — |
| 1271 |
// otherwise the admin keeps offering "View Generated Sitemaps" |
| 1272 |
// links to files that no longer exist. |
| 1273 |
if ($cleaned_files) { |
| 1274 |
$this->clear_generation_record(); |
| 1275 |
} |
| 1276 |
|
| 1277 |
return new WP_REST_Response([ |
| 1278 |
'success' => true, |
| 1279 |
'data' => [ |
| 1280 |
'cleaned_files' => $cleaned_files, |
| 1281 |
'failed_files' => $failed_files, |
| 1282 |
'total_cleaned' => count($cleaned_files), |
| 1283 |
'regeneration_scheduled' => $regeneration_scheduled |
| 1284 |
], |
| 1285 |
'message' => sprintf( |
| 1286 |
'Cleaned up %d sitemap file(s) successfully', |
| 1287 |
count($cleaned_files) |
| 1288 |
) |
| 1289 |
], 200); |
| 1290 |
|
| 1291 |
} catch (\Exception $e) { |
| 1292 |
return new WP_Error( |
| 1293 |
'cleanup_failed', |
| 1294 |
'Failed to clean up sitemap files: ' . $e->getMessage(), |
| 1295 |
['status' => 500] |
| 1296 |
); |
| 1297 |
} |
| 1298 |
} |
| 1299 |
|
| 1300 |
/** |
| 1301 |
* Get sitemap URLs for robots.txt integration |
| 1302 |
* |
| 1303 |
* Returns enabled sitemap URLs from sitemap settings for automatic |
| 1304 |
* inclusion in robots.txt file. This eliminates the need for manual |
| 1305 |
* sitemap URL configuration in robots.txt settings. |
| 1306 |
* |
| 1307 |
* @since 1.0.0 |
| 1308 |
* |
| 1309 |
* @param WP_REST_Request $request Request object |
| 1310 |
* @return WP_REST_Response Response object |
| 1311 |
*/ |
| 1312 |
public function get_robots_sitemap_urls(WP_REST_Request $request): WP_REST_Response { |
| 1313 |
try { |
| 1314 |
// Get sitemap settings |
| 1315 |
$settings = $this->sitemap_generator->get_settings('site'); |
| 1316 |
|
| 1317 |
// If sitemap is disabled, return empty array |
| 1318 |
if (empty($settings['enabled'])) { |
| 1319 |
return new WP_REST_Response([ |
| 1320 |
'success' => true, |
| 1321 |
'data' => [ |
| 1322 |
'sitemap_urls' => [], |
| 1323 |
'enabled' => false, |
| 1324 |
'message' => __('Sitemap generation is disabled', 'thinkrank') |
| 1325 |
] |
| 1326 |
], 200); |
| 1327 |
} |
| 1328 |
|
| 1329 |
// Extract enabled sitemap URLs |
| 1330 |
$sitemap_urls = []; |
| 1331 |
$site_url = home_url(); |
| 1332 |
|
| 1333 |
if (!empty($settings['sitemap_urls']) && is_array($settings['sitemap_urls'])) { |
| 1334 |
foreach ($settings['sitemap_urls'] as $sitemap) { |
| 1335 |
if (!empty($sitemap['enabled']) && !empty($sitemap['url'])) { |
| 1336 |
$sitemap_urls[] = [ |
| 1337 |
'url' => $sitemap['url'], |
| 1338 |
'full_url' => $site_url . $sitemap['url'], |
| 1339 |
'type' => $sitemap['type'] ?? 'general', |
| 1340 |
'type_label' => $this->get_sitemap_type_label($sitemap['type'] ?? 'general') |
| 1341 |
]; |
| 1342 |
} |
| 1343 |
} |
| 1344 |
} |
| 1345 |
|
| 1346 |
// Fallback to default sitemap if no URLs configured |
| 1347 |
if (empty($sitemap_urls)) { |
| 1348 |
$sitemap_urls[] = [ |
| 1349 |
'url' => '/sitemap.xml', |
| 1350 |
'full_url' => $site_url . '/sitemap.xml', |
| 1351 |
'type' => 'general', |
| 1352 |
'type_label' => __('General', 'thinkrank') |
| 1353 |
]; |
| 1354 |
} |
| 1355 |
|
| 1356 |
return new WP_REST_Response([ |
| 1357 |
'success' => true, |
| 1358 |
'data' => [ |
| 1359 |
'sitemap_urls' => $sitemap_urls, |
| 1360 |
'enabled' => true, |
| 1361 |
'count' => count($sitemap_urls) |
| 1362 |
] |
| 1363 |
], 200); |
| 1364 |
|
| 1365 |
} catch (\Exception $e) { |
| 1366 |
return new WP_REST_Response([ |
| 1367 |
'success' => false, |
| 1368 |
'error' => 'Failed to retrieve sitemap URLs: ' . $e->getMessage() |
| 1369 |
], 500); |
| 1370 |
} |
| 1371 |
} |
| 1372 |
|
| 1373 |
/** |
| 1374 |
* Get human-readable label for sitemap type |
| 1375 |
* |
| 1376 |
* @since 1.0.0 |
| 1377 |
* |
| 1378 |
* @param string $type Sitemap type |
| 1379 |
* @return string Human-readable label |
| 1380 |
*/ |
| 1381 |
private function get_sitemap_type_label(string $type): string { |
| 1382 |
$labels = [ |
| 1383 |
'index' => __('Index', 'thinkrank'), |
| 1384 |
'general' => __('General', 'thinkrank'), |
| 1385 |
'posts' => __('Posts', 'thinkrank'), |
| 1386 |
'pages' => __('Pages', 'thinkrank'), |
| 1387 |
'categories' => __('Categories', 'thinkrank'), |
| 1388 |
'tags' => __('Tags', 'thinkrank'), |
| 1389 |
'products' => __('Products', 'thinkrank'), |
| 1390 |
'wordpress' => __('WordPress Core', 'thinkrank'), |
| 1391 |
'custom' => __('Custom', 'thinkrank') |
| 1392 |
]; |
| 1393 |
|
| 1394 |
return $labels[$type] ?? ucfirst($type); |
| 1395 |
} |
| 1396 |
|
| 1397 |
/** |
| 1398 |
* Check rate limit for sitemap generation |
| 1399 |
* |
| 1400 |
* @since 1.0.0 |
| 1401 |
* @return bool True if within rate limit |
| 1402 |
*/ |
| 1403 |
private function check_rate_limit(): bool { |
| 1404 |
$user_id = get_current_user_id(); |
| 1405 |
$rate_key = "thinkrank_sitemap_rate_{$user_id}"; |
| 1406 |
|
| 1407 |
$requests = get_transient($rate_key) ?: 0; |
| 1408 |
|
| 1409 |
if ($requests >= 3) { // Max 3 requests per 5 minutes |
| 1410 |
return false; |
| 1411 |
} |
| 1412 |
|
| 1413 |
set_transient($rate_key, $requests + 1, 5 * MINUTE_IN_SECONDS); |
| 1414 |
return true; |
| 1415 |
} |
| 1416 |
|
| 1417 |
/** |
| 1418 |
* Acquire generation lock to prevent concurrent generation |
| 1419 |
* |
| 1420 |
* @since 1.0.0 |
| 1421 |
* @return bool True if lock acquired |
| 1422 |
*/ |
| 1423 |
private function acquire_generation_lock(): bool { |
| 1424 |
$lock_key = Sitemap_Generator::GENERATION_LOCK_TRANSIENT; |
| 1425 |
|
| 1426 |
if (get_transient($lock_key)) { |
| 1427 |
return false; // Generation already in progress |
| 1428 |
} |
| 1429 |
|
| 1430 |
set_transient($lock_key, time(), 5 * MINUTE_IN_SECONDS); |
| 1431 |
return true; |
| 1432 |
} |
| 1433 |
|
| 1434 |
/** |
| 1435 |
* Release generation lock |
| 1436 |
* |
| 1437 |
* @since 1.0.0 |
| 1438 |
* @return void |
| 1439 |
*/ |
| 1440 |
private function release_generation_lock(): void { |
| 1441 |
delete_transient(Sitemap_Generator::GENERATION_LOCK_TRANSIENT); |
| 1442 |
} |
| 1443 |
} |
| 1444 |
|
