| 1 |
<?php |
| 2 |
// If this file is called directly, abort. |
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* The sitemap generator functionality of the plugin. |
| 9 |
* |
| 10 |
* @link https://searchatlas.com |
| 11 |
* @since 1.0.0 |
| 12 |
* |
| 13 |
* @package Metasync |
| 14 |
* @subpackage Metasync/sitemap |
| 15 |
*/ |
| 16 |
|
| 17 |
/** |
| 18 |
* The sitemap generator class. |
| 19 |
* |
| 20 |
* Handles XML sitemap generation and management. |
| 21 |
* |
| 22 |
* @package Metasync |
| 23 |
* @subpackage Metasync/sitemap |
| 24 |
* @author Engineering Team <[email protected]> |
| 25 |
*/ |
| 26 |
class Metasync_Sitemap_Generator |
| 27 |
{ |
| 28 |
/** |
| 29 |
* Prefetched primary-category meta map for the current chunk. |
| 30 |
* |
| 31 |
* Populated by collect_all_urls() only on sites whose permalink structure |
| 32 |
* contains %category%. The map is keyed by `post_id` => [meta_key => value] |
| 33 |
* and holds the three primary-category keys (_metasync_primary_category, |
| 34 |
* _yoast_wpseo_primary_category, rank_math_primary_category). |
| 35 |
* |
| 36 |
* Metasync_Seo_Output::filter_post_link_category_primary() reads this map |
| 37 |
* before falling back to get_post_meta(), so that disabling the WP_Query |
| 38 |
* post-meta cache prime (update_post_meta_cache => false) does not regress |
| 39 |
* primary-category permalink resolution with per-post queries. |
| 40 |
* |
| 41 |
* @var array |
| 42 |
*/ |
| 43 |
public static $primary_category_prefetch = []; |
| 44 |
|
| 45 |
/** |
| 46 |
* The sitemap index file path |
| 47 |
* |
| 48 |
* @var string |
| 49 |
*/ |
| 50 |
private $sitemap_index_path; |
| 51 |
|
| 52 |
/** |
| 53 |
* The sitemap index URL |
| 54 |
* |
| 55 |
* @var string |
| 56 |
*/ |
| 57 |
private $sitemap_index_url; |
| 58 |
|
| 59 |
/** |
| 60 |
* Maximum URLs per sitemap file |
| 61 |
* |
| 62 |
* @var int |
| 63 |
*/ |
| 64 |
private $urls_per_sitemap = 5000; |
| 65 |
|
| 66 |
/** |
| 67 |
* Initialize the class and set its properties. |
| 68 |
*/ |
| 69 |
public function __construct() |
| 70 |
{ |
| 71 |
$this->sitemap_index_path = ABSPATH . 'sitemap_index.xml'; |
| 72 |
$this->sitemap_index_url = home_url('/sitemap_index.xml'); |
| 73 |
|
| 74 |
// MetaSync must only take over the sitemap URLs (/sitemap*.xml) |
| 75 |
// when its own sitemap feature is actually in use. When it is not, the |
| 76 |
// URLs are left untouched so a third-party provider (RankMath, Yoast, |
| 77 |
// AIOSEO, or WordPress core) can serve them. Previously the interception |
| 78 |
// below was registered unconditionally on every front-end request, which |
| 79 |
// silently shadowed another sitemap plugin, could not be turned off from |
| 80 |
// the UI, and even regenerated our sitemap on access after a delete. |
| 81 |
if ($this->is_sitemap_enabled()) { |
| 82 |
// Disable WordPress core sitemap so we don't serve duplicate sitemaps |
| 83 |
// to search engines. |
| 84 |
add_filter('wp_sitemaps_enabled', '__return_false', 10); |
| 85 |
if (!get_option('metasync_disable_wp_sitemap', false)) { |
| 86 |
update_option('metasync_disable_wp_sitemap', true); |
| 87 |
} |
| 88 |
|
| 89 |
// Register hook to serve virtual sitemap files. |
| 90 |
add_action('template_redirect', array($this, 'serve_virtual_sitemap'), 1); |
| 91 |
|
| 92 |
// Route sitemap URLs through WordPress rewrites so requests always reach PHP, |
| 93 |
// even on nginx setups that 403 direct .xml access. |
| 94 |
static $rewrite_hooks_registered = false; |
| 95 |
if (!$rewrite_hooks_registered) { |
| 96 |
$rewrite_hooks_registered = true; |
| 97 |
add_filter('query_vars', array($this, 'add_sitemap_query_var')); |
| 98 |
add_filter('option_rewrite_rules', array($this, 'inject_rewrite_rules')); |
| 99 |
add_filter('sanitize_option_rewrite_rules', array($this, 'strip_dynamic_rewrite_rules')); |
| 100 |
add_filter('redirect_canonical', array($this, 'disable_canonical_redirect_for_sitemaps')); |
| 101 |
} |
| 102 |
|
| 103 |
// Reconcile stray physical sitemap files on any request that runs |
| 104 |
// PHP, not just one that reaches our sitemap route. |
| 105 |
// |
| 106 |
// WordPress' own rewrite block ends with |
| 107 |
// `RewriteCond %{REQUEST_FILENAME} !-f`, so on Apache a physical |
| 108 |
// news-sitemap.xml at ABSPATH is served statically and index.php |
| 109 |
// is never loaded — serve_virtual_sitemap() and the absorption in |
| 110 |
// get_sitemap_document() are both unreachable for exactly the |
| 111 |
// request that needs them. Sweeping from `init` closes that window: |
| 112 |
// any page view on the site absorbs the file, so the next crawler |
| 113 |
// fetch gets the live document. |
| 114 |
static $absorb_hook_registered = false; |
| 115 |
if (!$absorb_hook_registered) { |
| 116 |
$absorb_hook_registered = true; |
| 117 |
add_action('init', array($this, 'absorb_stray_physical_sitemaps'), 20); |
| 118 |
} |
| 119 |
} |
| 120 |
|
| 121 |
$this->register_cache_bust_hooks(); |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Whether MetaSync should own the XML sitemap URLs (/sitemap*.xml). |
| 126 |
* |
| 127 |
* Instance wrapper around {@see is_feature_enabled()}; kept so existing |
| 128 |
* callers (and tests) that hold a generator instance are unchanged. |
| 129 |
* |
| 130 |
* @return bool |
| 131 |
*/ |
| 132 |
public function is_sitemap_enabled() |
| 133 |
{ |
| 134 |
return self::is_feature_enabled(); |
| 135 |
} |
| 136 |
|
| 137 |
/** |
| 138 |
* Side-effect-free, static form of the sitemap-feature gate. |
| 139 |
* |
| 140 |
* MetaSync only intercepts and serves the sitemap URLs when its own sitemap |
| 141 |
* feature is actually in use. When it is not, the URLs are left untouched so |
| 142 |
* a third-party provider (RankMath, Yoast, AIOSEO, or WordPress core) can |
| 143 |
* serve them — this prevents MetaSync from silently shadowing another |
| 144 |
* sitemap plugin. |
| 145 |
* |
| 146 |
* This is exposed as a static so callers outside the generator — notably the |
| 147 |
* robots.txt manager, which must not advertise a `Sitemap:` line when the |
| 148 |
* feature is off — can consult the exact same gate without constructing a |
| 149 |
* generator (whose constructor registers front-end hooks and would |
| 150 |
* re-register them). Single source of truth. |
| 151 |
* |
| 152 |
* Resolution order: |
| 153 |
* 1. Explicit `metasync_sitemap_enabled` option when set to a real boolean. |
| 154 |
* Store it as the string '1' or '0' — never boolean false, which |
| 155 |
* WordPress persists as an empty string and which we (correctly) read |
| 156 |
* back as "not configured". An unset option or empty string falls |
| 157 |
* through to auto-detect. |
| 158 |
* 2. Auto-detect: the general sitemap is tracked, auto-update is on, or the |
| 159 |
* news/video sitemaps are enabled. |
| 160 |
* The result is passed through the `metasync_sitemap_enabled` filter so a |
| 161 |
* host or site owner can force it on or off in code. |
| 162 |
* |
| 163 |
* @return bool |
| 164 |
*/ |
| 165 |
public static function is_feature_enabled() |
| 166 |
{ |
| 167 |
// Only a value that parses to a real boolean counts as an explicit |
| 168 |
// choice; unset, '' or unparseable values fall through to auto-detect |
| 169 |
// (so garbage never silently forces the feature off). |
| 170 |
$explicit = get_option('metasync_sitemap_enabled', null); |
| 171 |
$explicit = (null === $explicit || '' === $explicit) |
| 172 |
? null |
| 173 |
: filter_var($explicit, FILTER_VALIDATE_BOOLEAN, FILTER_NULL_ON_FAILURE); |
| 174 |
|
| 175 |
if (null !== $explicit) { |
| 176 |
$enabled = $explicit; |
| 177 |
} else { |
| 178 |
// Auto-detect: is any MetaSync sitemap feature actually in use? |
| 179 |
$news_settings = get_option('metasync_news_sitemap_settings', []); |
| 180 |
$video_settings = get_option('metasync_video_sitemap_settings', []); |
| 181 |
|
| 182 |
$enabled = (bool) get_option('metasync_sitemap_auto_update', false) |
| 183 |
|| !empty(get_option('metasync_sitemap_files', [])) |
| 184 |
|| (is_array($news_settings) && !empty($news_settings['enabled'])) |
| 185 |
|| (is_array($video_settings) && !empty($video_settings['enabled'])) |
| 186 |
|| self::sitemap_index_exists(); |
| 187 |
} |
| 188 |
|
| 189 |
/** |
| 190 |
* Filter whether MetaSync manages the XML sitemap URLs. |
| 191 |
* |
| 192 |
* Return false to hand /sitemap*.xml back to another sitemap provider |
| 193 |
* (e.g. RankMath, Yoast, AIOSEO, or WordPress core) while keeping the |
| 194 |
* rest of MetaSync active. |
| 195 |
* |
| 196 |
* @param bool $enabled Whether MetaSync currently owns the sitemap URLs. |
| 197 |
*/ |
| 198 |
return (bool) apply_filters('metasync_sitemap_enabled', $enabled); |
| 199 |
} |
| 200 |
|
| 201 |
/** |
| 202 |
* Static, side-effect-free check for the general sitemap index, mirroring |
| 203 |
* the instance {@see sitemap_exists()} without touching $this so it can be |
| 204 |
* used by {@see is_feature_enabled()} from a static context. |
| 205 |
* |
| 206 |
* @return bool |
| 207 |
*/ |
| 208 |
private static function sitemap_index_exists() |
| 209 |
{ |
| 210 |
// Virtual content is the primary storage path (transient). |
| 211 |
if (false !== get_transient('metasync_vsm_' . md5('sitemap_index.xml'))) { |
| 212 |
return true; |
| 213 |
} |
| 214 |
|
| 215 |
// Legacy option storage (pre-transient installs), read-only — no migration |
| 216 |
// side effects here. |
| 217 |
$legacy_virtual = get_option('metasync_sitemap_virtual', []); |
| 218 |
if (is_array($legacy_virtual) && isset($legacy_virtual['sitemap_index.xml'])) { |
| 219 |
return true; |
| 220 |
} |
| 221 |
|
| 222 |
// Tracked sitemap files option (survives transient eviction). |
| 223 |
if (!empty(get_option('metasync_sitemap_files', []))) { |
| 224 |
return true; |
| 225 |
} |
| 226 |
|
| 227 |
// Legacy physical file for installs that haven't migrated yet. |
| 228 |
return file_exists(ABSPATH . 'sitemap_index.xml'); |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Register `metasync_sitemap` as a public WordPress query var. |
| 233 |
* |
| 234 |
* @param array $vars Registered query vars. |
| 235 |
* @return array |
| 236 |
*/ |
| 237 |
public function add_sitemap_query_var($vars) |
| 238 |
{ |
| 239 |
$vars[] = 'metasync_sitemap'; |
| 240 |
return $vars; |
| 241 |
} |
| 242 |
|
| 243 |
/** |
| 244 |
* Dynamically inject sitemap rewrite rules at the top of the rules array so |
| 245 |
* they take effect without requiring a manual flush_rewrite_rules() call |
| 246 |
* (mirrors Yoast's class-yoast-dynamic-rewrites.php pattern). |
| 247 |
* |
| 248 |
* @param mixed $rules Existing rewrite rules array (or empty value before flush). |
| 249 |
* @return array |
| 250 |
*/ |
| 251 |
public function inject_rewrite_rules($rules) |
| 252 |
{ |
| 253 |
// CRITICAL: never inject into an empty/non-array rules value. |
| 254 |
// |
| 255 |
// WordPress self-heals broken routing in WP_Rewrite::wp_rewrite_rules(): |
| 256 |
// "if ( empty( get_option('rewrite_rules') ) ) { regenerate + save; }". |
| 257 |
// Because this filter runs on `option_rewrite_rules`, returning a |
| 258 |
// non-empty array when the stored option is empty makes WordPress think |
| 259 |
// rules already exist, so it NEVER regenerates. The site is then left |
| 260 |
// with only these sitemap rules, and every URL except the home page |
| 261 |
// 404s (pages, posts, WooCommerce products) until the plugin is |
| 262 |
// deactivated (filter removed → WP regenerates) or permalinks are |
| 263 |
// re-saved. Returning the empty value unchanged preserves the self-heal. |
| 264 |
if (empty($rules) || !is_array($rules)) { |
| 265 |
return $rules; |
| 266 |
} |
| 267 |
|
| 268 |
$sitemap_rules = [ |
| 269 |
'^sitemap_index\.xml$' => 'index.php?metasync_sitemap=sitemap_index.xml', |
| 270 |
'^sitemap(\d*)\.xml$' => 'index.php?metasync_sitemap=sitemap$matches[1].xml', |
| 271 |
'^news-sitemap\.xml$' => 'index.php?metasync_sitemap=news-sitemap.xml', |
| 272 |
'^video-sitemap\.xml$' => 'index.php?metasync_sitemap=video-sitemap.xml', |
| 273 |
]; |
| 274 |
|
| 275 |
return $sitemap_rules + $rules; |
| 276 |
} |
| 277 |
|
| 278 |
/** |
| 279 |
* Strip dynamically-injected sitemap rules before they are persisted to the |
| 280 |
* rewrite_rules option in the database. |
| 281 |
* |
| 282 |
* @param mixed $rules Rewrite rules about to be saved. |
| 283 |
* @return array|string Filtered rules without the dynamic sitemap entries. |
| 284 |
*/ |
| 285 |
public function strip_dynamic_rewrite_rules($rules) |
| 286 |
{ |
| 287 |
if (empty($rules) || !is_array($rules)) { |
| 288 |
return $rules; |
| 289 |
} |
| 290 |
|
| 291 |
$dynamic_keys = [ |
| 292 |
'^sitemap_index\.xml$', |
| 293 |
'^sitemap(\d*)\.xml$', |
| 294 |
'^news-sitemap\.xml$', |
| 295 |
'^video-sitemap\.xml$', |
| 296 |
]; |
| 297 |
|
| 298 |
return array_diff_key($rules, array_flip($dynamic_keys)); |
| 299 |
} |
| 300 |
|
| 301 |
/** |
| 302 |
* Suppress WordPress's canonical redirect when a sitemap is being served. |
| 303 |
* |
| 304 |
* @param string $redirect_url The proposed canonical URL. |
| 305 |
* @return string|false The unmodified URL, or false to cancel the redirect. |
| 306 |
*/ |
| 307 |
public function disable_canonical_redirect_for_sitemaps($redirect_url) |
| 308 |
{ |
| 309 |
if (get_query_var('metasync_sitemap')) { |
| 310 |
return false; |
| 311 |
} |
| 312 |
return $redirect_url; |
| 313 |
} |
| 314 |
|
| 315 |
/** |
| 316 |
* Register WordPress content lifecycle hooks that bust the sitemap caches. |
| 317 |
* |
| 318 |
* Bust callbacks are intentionally lightweight (transient deletes only) so they |
| 319 |
* do not add latency to the editor's save action. Optional async warm-up can |
| 320 |
* be enabled via the `metasync_sitemap_async_warmup` filter. |
| 321 |
*/ |
| 322 |
public function register_cache_bust_hooks() |
| 323 |
{ |
| 324 |
// Post lifecycle. |
| 325 |
add_action('save_post', array($this, 'bust_sitemap_cache')); |
| 326 |
add_action('delete_post', array($this, 'bust_sitemap_cache')); |
| 327 |
add_action('trashed_post', array($this, 'bust_sitemap_cache')); |
| 328 |
|
| 329 |
// Taxonomy lifecycle. |
| 330 |
add_action('created_term', array($this, 'bust_sitemap_cache')); |
| 331 |
add_action('edited_term', array($this, 'bust_sitemap_cache')); |
| 332 |
add_action('delete_term', array($this, 'bust_sitemap_cache')); |
| 333 |
|
| 334 |
// Meta updates that affect indexability. |
| 335 |
add_action('added_post_meta', array($this, 'bust_sitemap_on_meta_update'), 10, 3); |
| 336 |
add_action('updated_post_meta', array($this, 'bust_sitemap_on_meta_update'), 10, 3); |
| 337 |
add_action('deleted_post_meta', array($this, 'bust_sitemap_on_meta_update'), 10, 3); |
| 338 |
|
| 339 |
// Redirect stores: our own manager fires 'metasync_redirections_changed' |
| 340 |
// on every write; Rank Math fires 'rank_math/redirection/deleted' on |
| 341 |
// deletions (its free build exposes no save-side hook). As a coarse |
| 342 |
// save proxy, 'rank_math/redirection/get_redirections_query' fires |
| 343 |
// whenever Rank Math re-renders its redirect list — i.e. immediately |
| 344 |
// after every add/update/toggle. Yoast's per-post redirect meta is |
| 345 |
// covered by the meta listeners above via the '_yoast_wpseo_redirect' |
| 346 |
// bust key. |
| 347 |
add_action('metasync_redirections_changed', array($this, 'bust_sitemap_cache')); |
| 348 |
add_action('rank_math/redirection/deleted', array($this, 'bust_sitemap_cache')); |
| 349 |
add_action('rank_math/redirection/get_redirections_query', array($this, 'bust_sitemap_cache')); |
| 350 |
|
| 351 |
// Async warm-up listener. |
| 352 |
add_action('metasync_sitemap_async_warmup_event', array($this, 'async_warmup_handler')); |
| 353 |
} |
| 354 |
|
| 355 |
/** |
| 356 |
* Bust all sitemap-related transients. |
| 357 |
* |
| 358 |
* Coalesces multiple calls within the same request via a static guard so |
| 359 |
* bulk operations (e.g. a 50-product import) only trigger one bust. |
| 360 |
*/ |
| 361 |
public function bust_sitemap_cache() |
| 362 |
{ |
| 363 |
static $already_busted = false; |
| 364 |
if ($already_busted) { |
| 365 |
return; |
| 366 |
} |
| 367 |
$already_busted = true; |
| 368 |
|
| 369 |
delete_transient('metasync_vsm_' . md5('news-sitemap.xml')); |
| 370 |
delete_transient('metasync_vsm_' . md5('video-sitemap.xml')); |
| 371 |
|
| 372 |
$virtual_index = get_option('metasync_sitemap_virtual_index', []); |
| 373 |
if (is_array($virtual_index)) { |
| 374 |
foreach ($virtual_index as $tkey) { |
| 375 |
if (is_string($tkey) && $tkey !== '') { |
| 376 |
delete_transient($tkey); |
| 377 |
} |
| 378 |
} |
| 379 |
} |
| 380 |
delete_option('metasync_sitemap_virtual_index'); |
| 381 |
|
| 382 |
if (apply_filters('metasync_sitemap_async_warmup', false) |
| 383 |
&& !wp_next_scheduled('metasync_sitemap_async_warmup_event')) { |
| 384 |
wp_schedule_single_event(time() + 30, 'metasync_sitemap_async_warmup_event'); |
| 385 |
} |
| 386 |
} |
| 387 |
|
| 388 |
/** |
| 389 |
* Conditionally bust the sitemap cache when meta affecting indexability changes. |
| 390 |
* |
| 391 |
* Only `_metasync_robots_index`, `_metasync_canonical_url`, and |
| 392 |
* `metasync_common_robots` (the robots metabox array, which carries the |
| 393 |
* per-post noindex toggle the sitemap excludes on) trigger a bust; |
| 394 |
* unrelated keys (e.g. `_edit_lock`) are ignored. |
| 395 |
* |
| 396 |
* @param int|int[] $meta_id ID of the metadata entry (an array of IDs for |
| 397 |
* the deleted_post_meta hook); unused here. |
| 398 |
* @param int $object_id Object the metadata is attached to. |
| 399 |
* @param string $meta_key The meta key being updated. |
| 400 |
*/ |
| 401 |
public function bust_sitemap_on_meta_update($meta_id, $object_id, $meta_key) |
| 402 |
{ |
| 403 |
$bust_keys = array('_metasync_robots_index', '_metasync_canonical_url', 'metasync_common_robots', '_yoast_wpseo_redirect'); |
| 404 |
if (!in_array($meta_key, $bust_keys, true)) { |
| 405 |
return; |
| 406 |
} |
| 407 |
$this->bust_sitemap_cache(); |
| 408 |
} |
| 409 |
|
| 410 |
/** |
| 411 |
* Async warm-up handler invoked by wp_schedule_single_event after a bust. |
| 412 |
*/ |
| 413 |
public function async_warmup_handler() |
| 414 |
{ |
| 415 |
$this->generate_sitemap(); |
| 416 |
} |
| 417 |
|
| 418 |
/** |
| 419 |
* Generate the XML sitemap (split into multiple files if needed) |
| 420 |
* |
| 421 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 422 |
*/ |
| 423 |
public function generate_sitemap( $track_generation = false ) |
| 424 |
{ |
| 425 |
try { |
| 426 |
// Delete existing general sitemap files first (preserve news/video) |
| 427 |
$this->delete_sitemap('general'); |
| 428 |
|
| 429 |
// Clean up any orphaned temp files from previous crashed generations |
| 430 |
$this->cleanup_temp_sitemap_files(); |
| 431 |
|
| 432 |
// Collect all URLs |
| 433 |
$all_urls = $this->collect_all_urls(); |
| 434 |
|
| 435 |
if (empty($all_urls)) { |
| 436 |
return new WP_Error('no_urls', 'No URLs found to include in sitemap.'); |
| 437 |
} |
| 438 |
|
| 439 |
// Split URLs into chunks |
| 440 |
$url_chunks = array_chunk($all_urls, $this->urls_per_sitemap); |
| 441 |
$sitemap_files = []; |
| 442 |
|
| 443 |
// Generate individual sitemap files |
| 444 |
foreach ($url_chunks as $index => $urls) { |
| 445 |
$sitemap_number = $index + 1; |
| 446 |
$sitemap_filename = $this->get_sitemap_filename($sitemap_number); |
| 447 |
$sitemap_path = ABSPATH . $sitemap_filename; |
| 448 |
|
| 449 |
$result = $this->generate_sitemap_file($sitemap_path, $urls); |
| 450 |
|
| 451 |
if ($result === false) { |
| 452 |
// The chunk could not be stored, so it cannot be served. |
| 453 |
// Carrying on would publish an index that points at a |
| 454 |
// sitemap which does not exist, and report success while |
| 455 |
// doing it. |
| 456 |
return false; |
| 457 |
} |
| 458 |
|
| 459 |
if (is_wp_error($result)) { |
| 460 |
return $result; |
| 461 |
} |
| 462 |
|
| 463 |
$sitemap_files[] = [ |
| 464 |
'filename' => $sitemap_filename, |
| 465 |
'url' => home_url('/' . $sitemap_filename), |
| 466 |
'lastmod' => $this->get_chunk_lastmod($urls), |
| 467 |
]; |
| 468 |
} |
| 469 |
|
| 470 |
// Build the index entries: main sitemaps + news/video if they exist |
| 471 |
$index_entries = $sitemap_files; |
| 472 |
|
| 473 |
$news_settings = get_option('metasync_news_sitemap_settings', []); |
| 474 |
$video_settings = get_option('metasync_video_sitemap_settings', []); |
| 475 |
|
| 476 |
if (!empty($news_settings['enabled'])) { |
| 477 |
$index_entries[] = [ |
| 478 |
'filename' => 'news-sitemap.xml', |
| 479 |
'url' => home_url('/news-sitemap.xml'), |
| 480 |
'lastmod' => current_time('mysql', true), |
| 481 |
]; |
| 482 |
} |
| 483 |
|
| 484 |
if (!empty($video_settings['enabled'])) { |
| 485 |
$index_entries[] = [ |
| 486 |
'filename' => 'video-sitemap.xml', |
| 487 |
'url' => home_url('/video-sitemap.xml'), |
| 488 |
'lastmod' => current_time('mysql', true), |
| 489 |
]; |
| 490 |
} |
| 491 |
|
| 492 |
// Generate sitemap index file (includes all sitemaps) |
| 493 |
$index_result = $this->generate_sitemap_index($index_entries); |
| 494 |
|
| 495 |
if ($index_result === false) { |
| 496 |
return false; // Return false instead of WP_Error to prevent Sentry capture |
| 497 |
} |
| 498 |
|
| 499 |
if (is_wp_error($index_result)) { |
| 500 |
return $index_result; |
| 501 |
} |
| 502 |
|
| 503 |
// Store sitemap info for admin display |
| 504 |
update_option('metasync_sitemap_files', $sitemap_files); |
| 505 |
update_option('metasync_sitemap_total_urls', count($all_urls)); |
| 506 |
update_option('metasync_sitemap_last_generated', current_time('mysql')); |
| 507 |
|
| 508 |
// Consent-gated product analytics: only callers passing |
| 509 |
// $track_generation (the manual admin buttons) report; auto |
| 510 |
// regeneration (save_post hooks, warm-up cron, MCP) must not — |
| 511 |
// bulk publishing would otherwise send one event per post. |
| 512 |
if ($track_generation && !wp_doing_cron()) { |
| 513 |
Metasync_PostHog::feature('sitemap_generated', array( |
| 514 |
'url_count' => count($all_urls), |
| 515 |
)); |
| 516 |
} |
| 517 |
|
| 518 |
// Auto-update robots.txt with sitemap index URL |
| 519 |
$robots_result = $this->update_robots_txt_sitemap(); |
| 520 |
if ($robots_result['success'] && $robots_result['action'] !== 'unchanged') { |
| 521 |
// Store the result for admin notice |
| 522 |
set_transient('metasync_sitemap_robots_updated', $robots_result, 60); |
| 523 |
} |
| 524 |
|
| 525 |
return true; |
| 526 |
|
| 527 |
} catch (Exception $e) { |
| 528 |
return new WP_Error('sitemap_generation_failed', $e->getMessage()); |
| 529 |
} |
| 530 |
} |
| 531 |
|
| 532 |
/** |
| 533 |
* Collect all URLs for the sitemap |
| 534 |
* |
| 535 |
* @return array Array of URL data |
| 536 |
*/ |
| 537 |
private function collect_all_urls() |
| 538 |
{ |
| 539 |
global $wpdb; |
| 540 |
|
| 541 |
$urls = []; |
| 542 |
|
| 543 |
// Load user-configured sitemap settings |
| 544 |
$sitemap_settings = get_option('metasync_sitemap_settings', []); |
| 545 |
|
| 546 |
// Get all published posts, pages, and custom post types |
| 547 |
$post_types = get_post_types(['public' => true], 'names'); |
| 548 |
|
| 549 |
// Exclude certain post types |
| 550 |
$excluded_types = [ |
| 551 |
'attachment', |
| 552 |
'revision', |
| 553 |
'nav_menu_item', |
| 554 |
'elementor_library', |
| 555 |
'elementor-hf', |
| 556 |
'e-floating-buttons', |
| 557 |
'ct_template', |
| 558 |
'oxy_user_library', |
| 559 |
'brizy-template', |
| 560 |
'fusion_template', |
| 561 |
'fusion_tb_section', |
| 562 |
'ae_global_templates', |
| 563 |
'custom_css', |
| 564 |
'customize_changeset', |
| 565 |
'oembed_cache', |
| 566 |
'user_request', |
| 567 |
'wp_block', |
| 568 |
'wp_template', |
| 569 |
'wp_template_part', |
| 570 |
'wp_global_styles', |
| 571 |
'wp_navigation', |
| 572 |
'acf-field-group', |
| 573 |
'acf-field', |
| 574 |
'fl-builder-template', |
| 575 |
'fl-theme-layout', |
| 576 |
'wpr_mega_menu', |
| 577 |
'wpr_templates', |
| 578 |
]; |
| 579 |
$excluded_types = apply_filters('metasync_sitemap_excluded_post_types', $excluded_types); |
| 580 |
$post_types = array_diff($post_types, $excluded_types); |
| 581 |
|
| 582 |
// Apply user-configured post type selection (only when settings have been explicitly saved) |
| 583 |
if (!empty($sitemap_settings['_configured'])) { |
| 584 |
if (!empty($sitemap_settings['post_types'])) { |
| 585 |
$post_types = array_intersect($post_types, (array) $sitemap_settings['post_types']); |
| 586 |
} else { |
| 587 |
// Configured but empty = user unchecked all post types |
| 588 |
$post_types = []; |
| 589 |
} |
| 590 |
} |
| 591 |
|
| 592 |
if (empty($post_types)) { |
| 593 |
return $urls; |
| 594 |
} |
| 595 |
|
| 596 |
// Build excluded URLs set for fast lookup |
| 597 |
$excluded_urls = []; |
| 598 |
if (!empty($sitemap_settings['excluded_urls'])) { |
| 599 |
$raw_lines = explode("\n", $sitemap_settings['excluded_urls']); |
| 600 |
foreach ($raw_lines as $line) { |
| 601 |
$line = trim($line); |
| 602 |
if ($line !== '') { |
| 603 |
$excluded_urls[$line] = true; |
| 604 |
} |
| 605 |
} |
| 606 |
} |
| 607 |
|
| 608 |
// Get indexation control settings |
| 609 |
$seo_controls = $this->get_seo_controls(); |
| 610 |
|
| 611 |
// URLs that are the source of an active redirect must not be |
| 612 |
// advertised in the sitemap: a crawler following them never receives |
| 613 |
// the page, only a bounce to the destination, wasting crawl budget |
| 614 |
// and delaying indexing. Sources are collected from our own redirect |
| 615 |
// manager and from active third-party SEO plugins (Rank Math, |
| 616 |
// Yoast), so the sitemap matches what a crawler actually |
| 617 |
// experiences. Redirect destinations themselves are never excluded. |
| 618 |
$redirect_sources = $this->get_redirect_source_paths(); |
| 619 |
|
| 620 |
// Add homepage first |
| 621 |
$home_url = home_url('/'); |
| 622 |
$urls[] = [ |
| 623 |
'loc' => $home_url, |
| 624 |
'lastmod' => get_lastpostmodified('gmt'), |
| 625 |
'priority' => '1.0', |
| 626 |
'changefreq' => 'daily', |
| 627 |
]; |
| 628 |
|
| 629 |
// Track seen URLs to prevent duplicates (e.g. front page appearing again from posts query). |
| 630 |
// Normalize with untrailingslashit() to avoid trailing-slash mismatches between |
| 631 |
// home_url('/') and get_permalink() for static front pages. |
| 632 |
$seen_urls = [untrailingslashit($home_url) => true]; |
| 633 |
|
| 634 |
// Build taxonomy query for category/tag filtering (only when explicitly configured). |
| 635 |
// Each term filter is OR-ed with a NOT EXISTS clause: the query runs across all |
| 636 |
// selected post types in one WP_Query, and content with no terms in the taxonomy |
| 637 |
// (pages, custom post types, untagged posts) must not be excluded by a term |
| 638 |
// selection that cannot apply to it. |
| 639 |
$sitemap_tax_query = []; |
| 640 |
if (!empty($sitemap_settings['_configured']) && !empty($sitemap_settings['categories'])) { |
| 641 |
$sitemap_tax_query[] = [ |
| 642 |
'relation' => 'OR', |
| 643 |
[ |
| 644 |
'taxonomy' => 'category', |
| 645 |
'field' => 'term_id', |
| 646 |
'terms' => array_map('absint', (array) $sitemap_settings['categories']), |
| 647 |
], |
| 648 |
[ |
| 649 |
'taxonomy' => 'category', |
| 650 |
'operator' => 'NOT EXISTS', |
| 651 |
], |
| 652 |
]; |
| 653 |
} |
| 654 |
if (!empty($sitemap_settings['_configured']) && !empty($sitemap_settings['tags'])) { |
| 655 |
$sitemap_tax_query[] = [ |
| 656 |
'relation' => 'OR', |
| 657 |
[ |
| 658 |
'taxonomy' => 'post_tag', |
| 659 |
'field' => 'term_id', |
| 660 |
'terms' => array_map('absint', (array) $sitemap_settings['tags']), |
| 661 |
], |
| 662 |
[ |
| 663 |
'taxonomy' => 'post_tag', |
| 664 |
'operator' => 'NOT EXISTS', |
| 665 |
], |
| 666 |
]; |
| 667 |
} |
| 668 |
if (count($sitemap_tax_query) > 1) { |
| 669 |
$sitemap_tax_query['relation'] = 'AND'; |
| 670 |
} |
| 671 |
|
| 672 |
// Process posts in chunks to bound memory and database load. |
| 673 |
$chunk_size = (int) apply_filters('metasync_sitemap_chunk_size', 200); |
| 674 |
if ($chunk_size < 1) { |
| 675 |
$chunk_size = 200; |
| 676 |
} |
| 677 |
$paged = 1; |
| 678 |
|
| 679 |
while (true) { |
| 680 |
$query_args = [ |
| 681 |
'post_type' => array_values($post_types), |
| 682 |
'post_status' => 'publish', |
| 683 |
'posts_per_page' => $chunk_size, |
| 684 |
'paged' => $paged, |
| 685 |
'orderby' => ['modified' => 'DESC', 'ID' => 'DESC'], |
| 686 |
'no_found_rows' => true, |
| 687 |
'update_post_term_cache' => true, |
| 688 |
'update_post_meta_cache' => false, |
| 689 |
'cache_results' => true, |
| 690 |
'ignore_sticky_posts' => true, |
| 691 |
'suppress_filters' => true, // phpcs:ignore WordPressVIPMinimum.Performance.WPQueryParams.SuppressFilters_suppress_filters -- sitemap must enumerate the full site regardless of language/plugin filtering |
| 692 |
]; |
| 693 |
|
| 694 |
if (!empty($sitemap_tax_query)) { |
| 695 |
$query_args['tax_query'] = $sitemap_tax_query; |
| 696 |
} |
| 697 |
|
| 698 |
$query = new WP_Query($query_args); |
| 699 |
|
| 700 |
$posts = $query->posts; |
| 701 |
if (empty($posts)) { |
| 702 |
wp_reset_postdata(); |
| 703 |
unset($query, $posts); |
| 704 |
break; |
| 705 |
} |
| 706 |
|
| 707 |
// Batch-resolve noindex status for the entire chunk in a single query. |
| 708 |
// Mirrors is_post_noindex(): metasync_common_robots is a serialized array |
| 709 |
// and a post is noindex when it contains 'noindex' => 'noindex'. |
| 710 |
$post_ids = wp_list_pluck($posts, 'ID'); |
| 711 |
$post_ids = array_map('intval', $post_ids); |
| 712 |
$noindex_pattern = '%"noindex";s:7:"noindex"%'; |
| 713 |
$noindex_args = array_merge($post_ids, [$noindex_pattern]); |
| 714 |
$noindex_ids = (array) $wpdb->get_col( // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 715 |
$wpdb->prepare( |
| 716 |
'SELECT post_id FROM ' . $wpdb->postmeta . " WHERE meta_key = 'metasync_common_robots' AND post_id IN (" . implode(',', array_fill(0, count($post_ids), '%d')) . ') AND meta_value LIKE %s', |
| 717 |
$noindex_args |
| 718 |
) |
| 719 |
); |
| 720 |
$noindex_ids = array_map('intval', $noindex_ids); |
| 721 |
$noindex_set = array_flip($noindex_ids); |
| 722 |
|
| 723 |
// Yoast SEO Premium stores per-post redirects in post meta: a |
| 724 |
// published post whose own URL redirects elsewhere is in the same |
| 725 |
// position as a redirect-source URL and must not be listed either. |
| 726 |
// Batch-resolved in one query, mirroring the noindex lookup above. |
| 727 |
$yoast_redirect_set = []; |
| 728 |
// Per-post redirects are a Yoast SEO Premium feature; the gate |
| 729 |
// avoids a wasted meta query per chunk on free-Yoast sites. |
| 730 |
if (apply_filters('metasync_sitemap_yoast_redirects_active', defined('WPSEO_VERSION') && defined('WPSEO_PREMIUM_VERSION'))) { |
| 731 |
$yoast_ids = (array) $wpdb->get_col( // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 732 |
$wpdb->prepare( |
| 733 |
'SELECT post_id FROM ' . $wpdb->postmeta . " WHERE meta_key = '_yoast_wpseo_redirect' AND meta_value <> '' AND post_id IN (" . implode(',', array_fill(0, count($post_ids), '%d')) . ')', |
| 734 |
$post_ids |
| 735 |
) |
| 736 |
); |
| 737 |
$yoast_redirect_set = array_flip(array_map('intval', $yoast_ids)); |
| 738 |
} |
| 739 |
|
| 740 |
// Anti-regression for %category% permalink sites: with the post-meta |
| 741 |
// cache prime disabled (update_post_meta_cache => false), the only |
| 742 |
// post-meta consumer in the sitemap path — primary-category |
| 743 |
// resolution in Metasync_Seo_Output::filter_post_link_category_primary() |
| 744 |
// — would otherwise issue one get_post_meta() call per post. Batch- |
| 745 |
// fetch the three primary-category keys for the whole chunk in a |
| 746 |
// single query (mirroring the noindex batch above) and store the |
| 747 |
// result in the static prefetch map that the filter reads first. |
| 748 |
// Only relevant when the permalink structure actually contains |
| 749 |
// %category% (the filter is a no-op otherwise). |
| 750 |
self::$primary_category_prefetch = []; |
| 751 |
if (strpos((string) get_option('permalink_structure', ''), '%category%') !== false) { |
| 752 |
$primary_keys = [ |
| 753 |
'_metasync_primary_category', |
| 754 |
'_yoast_wpseo_primary_category', |
| 755 |
'rank_math_primary_category', |
| 756 |
]; |
| 757 |
$primary_args = array_merge($post_ids, $primary_keys); |
| 758 |
$primary_rows = (array) $wpdb->get_results( // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 759 |
$wpdb->prepare( |
| 760 |
'SELECT post_id, meta_key, meta_value FROM ' . $wpdb->postmeta . ' WHERE post_id IN (' . implode(',', array_fill(0, count($post_ids), '%d')) . ') AND meta_key IN (' . implode(',', array_fill(0, count($primary_keys), '%s')) . ')', |
| 761 |
$primary_args |
| 762 |
), |
| 763 |
ARRAY_A |
| 764 |
); |
| 765 |
$prefetch = []; |
| 766 |
foreach ($primary_rows as $row) { |
| 767 |
$pid = (int) $row['post_id']; |
| 768 |
if (!isset($prefetch[$pid])) { |
| 769 |
$prefetch[$pid] = []; |
| 770 |
} |
| 771 |
$prefetch[$pid][$row['meta_key']] = $row['meta_value']; |
| 772 |
} |
| 773 |
self::$primary_category_prefetch = $prefetch; |
| 774 |
} |
| 775 |
|
| 776 |
foreach ($posts as $post) { |
| 777 |
if (isset($noindex_set[(int) $post->ID])) { |
| 778 |
continue; |
| 779 |
} |
| 780 |
|
| 781 |
// Skip posts carrying a Yoast per-post redirect. |
| 782 |
if (isset($yoast_redirect_set[(int) $post->ID])) { |
| 783 |
continue; |
| 784 |
} |
| 785 |
|
| 786 |
$priority = '0.8'; |
| 787 |
if ($post->post_type === 'page') { |
| 788 |
$priority = '0.9'; |
| 789 |
} elseif ($post->post_type === 'post') { |
| 790 |
$priority = '0.7'; |
| 791 |
} |
| 792 |
|
| 793 |
$changefreq = 'weekly'; |
| 794 |
if ($post->post_type === 'page') { |
| 795 |
$changefreq = 'monthly'; |
| 796 |
} |
| 797 |
|
| 798 |
// Note: get_permalink() respects the post_link_category filter registered |
| 799 |
// in Metasync_Seo_Output, which automatically resolves the primary category. |
| 800 |
$permalink = get_permalink($post); |
| 801 |
|
| 802 |
// Skip URLs with query strings (e.g. ?wpr_mega_menu=... internal items). |
| 803 |
// Only when the site uses pretty permalinks — under the plain permalink |
| 804 |
// structure (?p=123) every legitimate URL contains '?', so applying this |
| 805 |
// unconditionally would empty the sitemap down to the homepage. |
| 806 |
if (get_option('permalink_structure') && strpos($permalink, '?') !== false) { |
| 807 |
continue; |
| 808 |
} |
| 809 |
|
| 810 |
// Skip URLs that are the source of a known redirect — the |
| 811 |
// sitemap must advertise the destination, not the bounce. |
| 812 |
if ($this->is_redirect_source($permalink, $redirect_sources)) { |
| 813 |
continue; |
| 814 |
} |
| 815 |
|
| 816 |
// Skip duplicate URLs (e.g. front page appearing again from posts query) |
| 817 |
$normalized_permalink = untrailingslashit($permalink); |
| 818 |
if (isset($seen_urls[$normalized_permalink])) { |
| 819 |
continue; |
| 820 |
} |
| 821 |
|
| 822 |
// Skip URLs that are in the exclusion list |
| 823 |
if (!empty($excluded_urls) && isset($excluded_urls[$permalink])) { |
| 824 |
continue; |
| 825 |
} |
| 826 |
|
| 827 |
$seen_urls[$normalized_permalink] = true; |
| 828 |
|
| 829 |
$urls[] = [ |
| 830 |
'loc' => $permalink, |
| 831 |
'lastmod' => $post->post_modified_gmt, |
| 832 |
'priority' => $priority, |
| 833 |
'changefreq' => $changefreq, |
| 834 |
]; |
| 835 |
} |
| 836 |
|
| 837 |
wp_reset_postdata(); |
| 838 |
|
| 839 |
// Release the per-chunk object-cache accumulation (post objects, post |
| 840 |
// meta, and term caches) so memory stays flat across chunks regardless |
| 841 |
// of catalog size. clean_post_cache() drops each post's cache entries; |
| 842 |
// with a persistent external object cache (Redis/Memcached) it evicts |
| 843 |
// only those individual post entries (re-fetched on demand) rather than |
| 844 |
// flushing the whole cache, so it never wipes the live site cache. Runs |
| 845 |
// AFTER the foreach above has finished using $post (get_permalink, |
| 846 |
// post_modified_gmt) and BEFORE the next chunk loads its post objects. |
| 847 |
$chunk_post_ids = wp_list_pluck($posts, 'ID'); |
| 848 |
foreach ($chunk_post_ids as $chunk_post_id) { |
| 849 |
clean_post_cache((int) $chunk_post_id); |
| 850 |
} |
| 851 |
|
| 852 |
unset($query, $posts, $noindex_ids, $noindex_set, $post_ids, $noindex_args, $prefetch, $primary_rows, $primary_args, $primary_keys, $chunk_post_ids, $yoast_ids, $yoast_redirect_set); |
| 853 |
$paged++; |
| 854 |
} |
| 855 |
|
| 856 |
// Clear the chunk-scoped prefetch map so it does not leak into subsequent |
| 857 |
// non-sitemap get_permalink() calls during the same request. |
| 858 |
self::$primary_category_prefetch = []; |
| 859 |
|
| 860 |
// Add taxonomies (categories, tags, etc.) |
| 861 |
$taxonomies = get_taxonomies(['public' => true], 'names'); |
| 862 |
|
| 863 |
// Apply user-configured taxonomy selection (only when settings have been explicitly saved) |
| 864 |
if (!empty($sitemap_settings['_configured'])) { |
| 865 |
if (!empty($sitemap_settings['taxonomies'])) { |
| 866 |
$taxonomies = array_intersect($taxonomies, (array) $sitemap_settings['taxonomies']); |
| 867 |
} else { |
| 868 |
// Configured but empty = user unchecked all taxonomies |
| 869 |
$taxonomies = []; |
| 870 |
} |
| 871 |
} |
| 872 |
|
| 873 |
foreach ($taxonomies as $taxonomy) { |
| 874 |
// Skip taxonomies that are set to noindex |
| 875 |
if ($this->is_taxonomy_noindex($taxonomy, $seo_controls)) { |
| 876 |
continue; |
| 877 |
} |
| 878 |
|
| 879 |
// Respect per-term category/tag selections for archive URLs too, so an |
| 880 |
// unchecked term's archive page does not appear in the sitemap. |
| 881 |
$selected_term_ids = null; |
| 882 |
if (!empty($sitemap_settings['_configured'])) { |
| 883 |
if ($taxonomy === 'category' && !empty($sitemap_settings['categories'])) { |
| 884 |
$selected_term_ids = array_map('absint', (array) $sitemap_settings['categories']); |
| 885 |
} elseif ($taxonomy === 'post_tag' && !empty($sitemap_settings['tags'])) { |
| 886 |
$selected_term_ids = array_map('absint', (array) $sitemap_settings['tags']); |
| 887 |
} |
| 888 |
} |
| 889 |
|
| 890 |
$terms = get_terms([ |
| 891 |
'taxonomy' => $taxonomy, |
| 892 |
'hide_empty' => true, |
| 893 |
]); |
| 894 |
|
| 895 |
if (is_wp_error($terms) || empty($terms)) { |
| 896 |
continue; |
| 897 |
} |
| 898 |
|
| 899 |
// Single aggregated query: most-recently-modified post per term in this taxonomy. |
| 900 |
$type_values = array_values($post_types); |
| 901 |
$lastmod_args = array_merge([$taxonomy], $type_values); |
| 902 |
$lastmod_rows = $wpdb->get_results( // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 903 |
$wpdb->prepare( |
| 904 |
'SELECT tt.term_id, MAX(p.post_modified_gmt) AS lastmod |
| 905 |
FROM ' . $wpdb->term_taxonomy . ' tt |
| 906 |
INNER JOIN ' . $wpdb->term_relationships . ' tr ON tr.term_taxonomy_id = tt.term_taxonomy_id |
| 907 |
INNER JOIN ' . $wpdb->posts . " p ON p.ID = tr.object_id |
| 908 |
WHERE tt.taxonomy = %s AND p.post_status = 'publish' AND p.post_type IN (" . implode(',', array_fill(0, count($type_values), '%s')) . ') |
| 909 |
GROUP BY tt.term_id', |
| 910 |
$lastmod_args |
| 911 |
), |
| 912 |
ARRAY_A |
| 913 |
); |
| 914 |
|
| 915 |
$lastmod_map = []; |
| 916 |
if (is_array($lastmod_rows)) { |
| 917 |
foreach ($lastmod_rows as $row) { |
| 918 |
$lastmod_map[(int) $row['term_id']] = $row['lastmod']; |
| 919 |
} |
| 920 |
} |
| 921 |
|
| 922 |
foreach ($terms as $term) { |
| 923 |
if ($selected_term_ids !== null && !in_array((int) $term->term_id, $selected_term_ids, true)) { |
| 924 |
continue; |
| 925 |
} |
| 926 |
|
| 927 |
$term_link = get_term_link($term); |
| 928 |
if (is_wp_error($term_link)) { |
| 929 |
continue; |
| 930 |
} |
| 931 |
|
| 932 |
// Term archives obey the same rule as posts: a redirecting |
| 933 |
// archive URL is not advertised, its destination is. |
| 934 |
if ($this->is_redirect_source($term_link, $redirect_sources)) { |
| 935 |
continue; |
| 936 |
} |
| 937 |
|
| 938 |
// Skip URLs that are in the exclusion list |
| 939 |
if (!empty($excluded_urls) && isset($excluded_urls[$term_link])) { |
| 940 |
continue; |
| 941 |
} |
| 942 |
|
| 943 |
// Skip duplicate URLs |
| 944 |
$normalized_term_link = untrailingslashit($term_link); |
| 945 |
if (isset($seen_urls[$normalized_term_link])) { |
| 946 |
continue; |
| 947 |
} |
| 948 |
|
| 949 |
$seen_urls[$normalized_term_link] = true; |
| 950 |
|
| 951 |
$lastmod = isset($lastmod_map[(int) $term->term_id]) |
| 952 |
? $lastmod_map[(int) $term->term_id] |
| 953 |
: current_time('mysql', 1); |
| 954 |
$urls[] = [ |
| 955 |
'loc' => $term_link, |
| 956 |
'lastmod' => $lastmod, |
| 957 |
'priority' => '0.6', |
| 958 |
'changefreq' => 'weekly', |
| 959 |
]; |
| 960 |
} |
| 961 |
|
| 962 |
unset($lastmod_rows, $lastmod_map, $terms); |
| 963 |
} |
| 964 |
|
| 965 |
return $urls; |
| 966 |
} |
| 967 |
|
| 968 |
/** |
| 969 |
* Get the sitemap filename for a given number |
| 970 |
* |
| 971 |
* @param int $number The sitemap number (1, 2, 3, etc.) |
| 972 |
* @return string The filename |
| 973 |
*/ |
| 974 |
private function get_sitemap_filename($number) |
| 975 |
{ |
| 976 |
if ($number === 1) { |
| 977 |
return 'sitemap.xml'; |
| 978 |
} |
| 979 |
return 'sitemap' . $number . '.xml'; |
| 980 |
} |
| 981 |
|
| 982 |
/** |
| 983 |
* Generate a single sitemap file |
| 984 |
* |
| 985 |
* @param string $path The file path |
| 986 |
* @param array $urls Array of URL data |
| 987 |
* @return bool|WP_Error True on success, false when the sitemap could not be |
| 988 |
* stored, WP_Error on failure |
| 989 |
*/ |
| 990 |
private function generate_sitemap_file($path, $urls) |
| 991 |
{ |
| 992 |
$xml_content = $this->build_sitemap_xml_string($urls); |
| 993 |
return $this->store_virtual_sitemap_file(basename($path), $xml_content); |
| 994 |
} |
| 995 |
|
| 996 |
/** |
| 997 |
* Build the sitemap XML string in memory using DOMDocument. |
| 998 |
* |
| 999 |
* @param array $urls Array of URL data |
| 1000 |
* @return string XML content |
| 1001 |
*/ |
| 1002 |
private function build_sitemap_xml_string($urls) |
| 1003 |
{ |
| 1004 |
if (!class_exists('DOMDocument')) { |
| 1005 |
error_log('MetaSync: sitemap generation skipped — DOMDocument (php-dom extension) is not available on this server.'); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- genuine failure path, bounded, no secrets |
| 1006 |
return ''; |
| 1007 |
} |
| 1008 |
|
| 1009 |
$xml = new DOMDocument('1.0', 'UTF-8'); |
| 1010 |
$xml->formatOutput = true; |
| 1011 |
|
| 1012 |
$urlset = $xml->createElement('urlset'); |
| 1013 |
$urlset->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9'); |
| 1014 |
$urlset->setAttribute('xmlns:xsi', 'http://www.w3.org/2001/XMLSchema-instance'); |
| 1015 |
$urlset->setAttribute('xsi:schemaLocation', 'http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd'); |
| 1016 |
$xml->appendChild($urlset); |
| 1017 |
|
| 1018 |
foreach ($urls as $url_data) { |
| 1019 |
$this->add_url_to_sitemap($xml, $urlset, $url_data['loc'], $url_data['lastmod'], $url_data['priority'], $url_data['changefreq']); |
| 1020 |
} |
| 1021 |
|
| 1022 |
return $xml->saveXML(); |
| 1023 |
} |
| 1024 |
|
| 1025 |
/** |
| 1026 |
* Build the sitemap index XML string in memory using DOMDocument. |
| 1027 |
* |
| 1028 |
* @param array $sitemap_files Array of sitemap file info |
| 1029 |
* @return string XML content |
| 1030 |
*/ |
| 1031 |
private function build_sitemap_index_xml_string($sitemap_files) |
| 1032 |
{ |
| 1033 |
if (!class_exists('DOMDocument')) { |
| 1034 |
error_log('MetaSync: sitemap index generation skipped — DOMDocument (php-dom extension) is not available on this server.'); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- genuine failure path, bounded, no secrets |
| 1035 |
return ''; |
| 1036 |
} |
| 1037 |
|
| 1038 |
$xml = new DOMDocument('1.0', 'UTF-8'); |
| 1039 |
$xml->formatOutput = true; |
| 1040 |
|
| 1041 |
$sitemapindex = $xml->createElement('sitemapindex'); |
| 1042 |
$sitemapindex->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9'); |
| 1043 |
$xml->appendChild($sitemapindex); |
| 1044 |
|
| 1045 |
foreach ($sitemap_files as $sitemap) { |
| 1046 |
$sitemap_element = $xml->createElement('sitemap'); |
| 1047 |
|
| 1048 |
$loc = $xml->createElement('loc', esc_url($sitemap['url'])); |
| 1049 |
$sitemap_element->appendChild($loc); |
| 1050 |
|
| 1051 |
if (!empty($sitemap['lastmod'])) { |
| 1052 |
$lastmod_formatted = gmdate('Y-m-d\TH:i:s+00:00', strtotime($sitemap['lastmod'])); |
| 1053 |
$lastmod = $xml->createElement('lastmod', $lastmod_formatted); |
| 1054 |
$sitemap_element->appendChild($lastmod); |
| 1055 |
} |
| 1056 |
|
| 1057 |
$sitemapindex->appendChild($sitemap_element); |
| 1058 |
} |
| 1059 |
|
| 1060 |
return $xml->saveXML(); |
| 1061 |
} |
| 1062 |
|
| 1063 |
/** |
| 1064 |
* Write content to a file atomically using temp file + rename. |
| 1065 |
* |
| 1066 |
* Both files must reside on the same filesystem/mount for rename() to be atomic. |
| 1067 |
* |
| 1068 |
* @param string $dir Directory for the temp file (same mount as $final_path) |
| 1069 |
* @param string $final_path Final destination path |
| 1070 |
* @param string $content File content to write |
| 1071 |
* @return bool True on success, false on failure |
| 1072 |
*/ |
| 1073 |
private function atomic_write($dir, $final_path, $content) |
| 1074 |
{ |
| 1075 |
$tmp_path = tempnam($dir, 'metasync-sitemap-tmp-'); |
| 1076 |
if ($tmp_path === false) { |
| 1077 |
return false; |
| 1078 |
} |
| 1079 |
|
| 1080 |
if (file_put_contents($tmp_path, $content) === false) { |
| 1081 |
$this->safe_unlink($tmp_path); |
| 1082 |
return false; |
| 1083 |
} |
| 1084 |
|
| 1085 |
if (rename($tmp_path, $final_path)) { // phpcs:ignore WordPress.WP.AlternativeFunctions.rename_rename -- atomic same-filesystem replace; WP_Filesystem::move() needs credential bootstrapping on non-direct hosts |
| 1086 |
return true; |
| 1087 |
} |
| 1088 |
|
| 1089 |
$this->safe_unlink($tmp_path); |
| 1090 |
return false; |
| 1091 |
} |
| 1092 |
|
| 1093 |
/** |
| 1094 |
* Unlink a file with WP_DEBUG-gated error logging instead of @ suppression. |
| 1095 |
* |
| 1096 |
* @param string $path File path to delete |
| 1097 |
*/ |
| 1098 |
private function safe_unlink($path) |
| 1099 |
{ |
| 1100 |
if (file_exists($path) && !unlink($path)) { // phpcs:ignore WordPress.WP.AlternativeFunctions.unlink_unlink -- boolean return value is consumed |
| 1101 |
if (defined('WP_DEBUG') && WP_DEBUG) { |
| 1102 |
error_log('MetaSync: failed to remove temp file: ' . $path); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- debug-gated, no secrets |
| 1103 |
} |
| 1104 |
} |
| 1105 |
} |
| 1106 |
|
| 1107 |
/** |
| 1108 |
* Remove orphaned temp files from previous crashed/timed-out generations. |
| 1109 |
*/ |
| 1110 |
private function cleanup_temp_sitemap_files() |
| 1111 |
{ |
| 1112 |
$dirs = array_unique([dirname($this->sitemap_index_path), ABSPATH]); |
| 1113 |
foreach ($dirs as $dir) { |
| 1114 |
$pattern = $dir . '/metasync-sitemap-tmp-*'; |
| 1115 |
$stale_files = glob($pattern); |
| 1116 |
if (!is_array($stale_files)) { |
| 1117 |
continue; |
| 1118 |
} |
| 1119 |
foreach ($stale_files as $file) { |
| 1120 |
$this->safe_unlink($file); |
| 1121 |
} |
| 1122 |
} |
| 1123 |
} |
| 1124 |
|
| 1125 |
/** |
| 1126 |
* Generate the sitemap index file |
| 1127 |
* |
| 1128 |
* @param array $sitemap_files Array of sitemap file info |
| 1129 |
* @return bool|WP_Error True on success, false when the index could not be |
| 1130 |
* stored, WP_Error on failure |
| 1131 |
*/ |
| 1132 |
private function generate_sitemap_index($sitemap_files) |
| 1133 |
{ |
| 1134 |
$xml_content = $this->build_sitemap_index_xml_string($sitemap_files); |
| 1135 |
return $this->store_virtual_sitemap_file('sitemap_index.xml', $xml_content); |
| 1136 |
} |
| 1137 |
|
| 1138 |
/** |
| 1139 |
* Get the most recent lastmod date from a chunk of URLs |
| 1140 |
* |
| 1141 |
* @param array $urls Array of URL data |
| 1142 |
* @return string The most recent lastmod date |
| 1143 |
*/ |
| 1144 |
private function get_chunk_lastmod($urls) |
| 1145 |
{ |
| 1146 |
$latest = ''; |
| 1147 |
foreach ($urls as $url) { |
| 1148 |
if (!empty($url['lastmod']) && $url['lastmod'] > $latest) { |
| 1149 |
$latest = $url['lastmod']; |
| 1150 |
} |
| 1151 |
} |
| 1152 |
return $latest ?: current_time('mysql', 1); |
| 1153 |
} |
| 1154 |
|
| 1155 |
/** |
| 1156 |
* Collect normalized paths that are the source of an active redirect. |
| 1157 |
* |
| 1158 |
* The sitemap must not advertise a URL that immediately redirects: the |
| 1159 |
* crawler never receives the page, only a bounce. Sources are read from |
| 1160 |
* every redirect store that can influence this site's responses: |
| 1161 |
* |
| 1162 |
* - our own redirect manager (metasync_redirections), honoring the |
| 1163 |
* 'exact' and 'start' pattern types; |
| 1164 |
* - Rank Math's Redirections module (rank_math_redirections) when both |
| 1165 |
* the plugin and its Redirections module are active, honoring its |
| 1166 |
* 'exact' and 'start' comparisons — a common pairing where redirects |
| 1167 |
* are managed in Rank Math while we generate the sitemap; |
| 1168 |
* - Yoast SEO Premium's per-post redirect, resolved per chunk in |
| 1169 |
* collect_all_urls() via the '_yoast_wpseo_redirect' post meta. |
| 1170 |
* (Yoast Premium's redirect-manager table — its UI-level redirects — |
| 1171 |
* is a separate store and is deliberately out of scope here.) |
| 1172 |
* |
| 1173 |
* Only deterministic pattern types are honored (exact and prefix). |
| 1174 |
* 'contain', 'end', 'wildcard' and 'regex' patterns are deliberately |
| 1175 |
* skipped: projecting them onto generated URLs risks false exclusions, |
| 1176 |
* and their destinations are still listed. |
| 1177 |
* |
| 1178 |
* Return false from the 'metasync_sitemap_exclude_redirect_sources' |
| 1179 |
* filter to disable the behaviour entirely (e.g. a host that wants the |
| 1180 |
* raw publish-list back). |
| 1181 |
* |
| 1182 |
* @return array{exact: array<string, true>, prefix: array<string, string>} 'exact' maps |
| 1183 |
* normalized path => true; 'prefix' maps normalized path => itself. |
| 1184 |
*/ |
| 1185 |
private function get_redirect_source_paths() |
| 1186 |
{ |
| 1187 |
$sources = ['exact' => [], 'prefix' => []]; |
| 1188 |
|
| 1189 |
if (!apply_filters('metasync_sitemap_exclude_redirect_sources', true)) { |
| 1190 |
return $sources; |
| 1191 |
} |
| 1192 |
|
| 1193 |
global $wpdb; |
| 1194 |
|
| 1195 |
// Our own redirect manager. |
| 1196 |
$metasync_table = $wpdb->prefix . 'metasync_redirections'; |
| 1197 |
if ($wpdb->get_var($wpdb->prepare('SHOW TABLES LIKE %s', $metasync_table)) === $metasync_table) { // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 1198 |
$rows = (array) $wpdb->get_results( |
| 1199 |
"SELECT sources_from, pattern_type FROM {$wpdb->prefix}metasync_redirections WHERE status = 'active'", |
| 1200 |
ARRAY_A |
| 1201 |
); |
| 1202 |
foreach ($rows as $row) { |
| 1203 |
$source_list = maybe_unserialize($row['sources_from']); |
| 1204 |
if (!is_array($source_list)) { |
| 1205 |
continue; |
| 1206 |
} |
| 1207 |
$global_type = isset($row['pattern_type']) ? $row['pattern_type'] : null; |
| 1208 |
foreach ($source_list as $source_key => $source_value) { |
| 1209 |
$source_key = (string) $source_key; |
| 1210 |
// Legacy list-format rows store the URL as the value under |
| 1211 |
// a numeric key ('0' => '/old'); remap for matching. |
| 1212 |
if (ctype_digit($source_key) && is_string($source_value) && '' !== $source_value) { |
| 1213 |
$source_key = $source_value; |
| 1214 |
} |
| 1215 |
$type = in_array($source_value, ['exact', 'contain', 'start', 'end', 'wildcard', 'regex'], true) |
| 1216 |
? $source_value |
| 1217 |
: ($global_type ?: 'exact'); |
| 1218 |
$path = $this->normalize_redirect_path($source_key); |
| 1219 |
if ('' === $path) { |
| 1220 |
continue; |
| 1221 |
} |
| 1222 |
if ('exact' === $type) { |
| 1223 |
// A bare-root exact source ('https://example.com' or '/') |
| 1224 |
// would collide with every query-string permalink on |
| 1225 |
// plain-permalink sites ('?p=N' normalizes to '/'), |
| 1226 |
// collapsing the sitemap to the homepage — never |
| 1227 |
// register it. A '*' in an 'exact' source is wildcard |
| 1228 |
// input the redirect engine treats loosely; it is not |
| 1229 |
// deterministic, so it is skipped like the other |
| 1230 |
// non-deterministic pattern types. |
| 1231 |
if ('/' !== $path && strpos($path, '*') === false) { |
| 1232 |
$sources['exact'][$path] = true; |
| 1233 |
} |
| 1234 |
} elseif ('start' === $type) { |
| 1235 |
$sources['prefix'][$path] = $path; |
| 1236 |
} |
| 1237 |
} |
| 1238 |
} |
| 1239 |
} |
| 1240 |
|
| 1241 |
// Rank Math's Redirections module (free tier), when both the plugin |
| 1242 |
// and its Redirections module are active. Rows can outlive the module |
| 1243 |
// being switched off in Rank Math's settings, and a disabled module |
| 1244 |
// serves no redirects — only honor the table while it is on. |
| 1245 |
$rank_math_active = class_exists('RankMath') || defined('RANK_MATH_VERSION'); |
| 1246 |
if ($rank_math_active) { |
| 1247 |
// Mirror Rank Math's own is_module_active(): rows can outlive the |
| 1248 |
// module being switched off in its settings, and a disabled |
| 1249 |
// module serves no redirects. |
| 1250 |
$rank_math_modules = get_option('rank_math_modules', []); |
| 1251 |
$rank_math_active = is_array($rank_math_modules) && in_array('redirections', $rank_math_modules, true); |
| 1252 |
} |
| 1253 |
if (apply_filters('metasync_sitemap_rank_math_active', $rank_math_active)) { |
| 1254 |
$rank_math_table = $wpdb->prefix . 'rank_math_redirections'; |
| 1255 |
if ($wpdb->get_var($wpdb->prepare('SHOW TABLES LIKE %s', $rank_math_table)) === $rank_math_table) { // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- sitemap build read — output is transient-cached with lifecycle bust hooks; these queries run only on cache rebuilds |
| 1256 |
$rows = (array) $wpdb->get_results( |
| 1257 |
"SELECT sources FROM {$wpdb->prefix}rank_math_redirections WHERE status = 'active'", |
| 1258 |
ARRAY_A |
| 1259 |
); |
| 1260 |
foreach ($rows as $row) { |
| 1261 |
$source_list = isset($row['sources']) ? maybe_unserialize($row['sources']) : null; |
| 1262 |
if (is_string($source_list)) { |
| 1263 |
// Defensive: some Rank Math builds stored JSON here. |
| 1264 |
$decoded = json_decode($source_list, true); |
| 1265 |
$source_list = is_array($decoded) ? $decoded : null; |
| 1266 |
} |
| 1267 |
if (!is_array($source_list)) { |
| 1268 |
continue; |
| 1269 |
} |
| 1270 |
foreach ($source_list as $entry) { |
| 1271 |
if (!is_array($entry) || empty($entry['pattern'])) { |
| 1272 |
continue; |
| 1273 |
} |
| 1274 |
$comparison = isset($entry['comparison']) ? $entry['comparison'] : 'exact'; |
| 1275 |
$path = $this->normalize_redirect_path($entry['pattern']); |
| 1276 |
if ('' === $path) { |
| 1277 |
continue; |
| 1278 |
} |
| 1279 |
if ('exact' === $comparison) { |
| 1280 |
// Same bare-root guard as our own table (query- |
| 1281 |
// string permalinks all normalize to '/'). |
| 1282 |
if ('/' !== $path) { |
| 1283 |
$sources['exact'][$path] = true; |
| 1284 |
} |
| 1285 |
} elseif ('start' === $comparison || 'start with' === $comparison) { |
| 1286 |
// Rank Math stores the comparison token 'start'; |
| 1287 |
// 'start with' is tolerated defensively for forks. |
| 1288 |
// Prefix rules are live-matched against a slash- |
| 1289 |
// trimmed URI with the pattern compared as stored |
| 1290 |
// (Str::starts_with), so a leading slash in the |
| 1291 |
// stored pattern permanently disables the rule at |
| 1292 |
// runtime — mirror that and skip it here too, or |
| 1293 |
// URLs that still serve 200 would be excluded. |
| 1294 |
// A scheme-bearing pattern (a pasted full URL) is |
| 1295 |
// dead for the same reason: the runtime URI never |
| 1296 |
// contains the scheme or host. |
| 1297 |
$raw_pattern = (string) $entry['pattern']; |
| 1298 |
if (0 === strpos($raw_pattern, '/') || strpos($raw_pattern, '://') !== false) { |
| 1299 |
continue; |
| 1300 |
} |
| 1301 |
$sources['prefix'][$path] = $path; |
| 1302 |
} |
| 1303 |
} |
| 1304 |
} |
| 1305 |
} |
| 1306 |
} |
| 1307 |
|
| 1308 |
return $sources; |
| 1309 |
} |
| 1310 |
|
| 1311 |
/** |
| 1312 |
* Normalize a URL or path for redirect-source comparison. |
| 1313 |
* |
| 1314 |
* Redirection stores hold paths with or without a leading slash, and |
| 1315 |
* sometimes full URLs; generated sitemap entries are always full |
| 1316 |
* permalinks. Reducing both sides to a bare path — leading slash, no |
| 1317 |
* trailing slash — makes the comparison host- and scheme-agnostic. |
| 1318 |
* |
| 1319 |
* @param mixed $url_or_path Absolute URL or path; serialized third-party redirect data is not guaranteed to be a string, so non-strings are cast. |
| 1320 |
* @return string Normalized path ('' when nothing usable remains, '/' for root). |
| 1321 |
*/ |
| 1322 |
private function normalize_redirect_path($url_or_path) |
| 1323 |
{ |
| 1324 |
$path = $url_or_path; |
| 1325 |
if (is_string($url_or_path) && strpos($url_or_path, 'http') === 0) { |
| 1326 |
$parsed = wp_parse_url($url_or_path); |
| 1327 |
// A bare domain ('https://example.com') is the site root. |
| 1328 |
$path = isset($parsed['path']) ? $parsed['path'] : '/'; |
| 1329 |
} |
| 1330 |
$path = (string) $path; |
| 1331 |
if ('' === $path) { |
| 1332 |
return ''; |
| 1333 |
} |
| 1334 |
$path = '/' . ltrim($path, '/'); |
| 1335 |
$path = rtrim($path, '/'); |
| 1336 |
return ('' === $path) ? '/' : $path; |
| 1337 |
} |
| 1338 |
|
| 1339 |
/** |
| 1340 |
* Whether a URL is the source of a known redirect. |
| 1341 |
* |
| 1342 |
* @param string $url URL as it would appear in the sitemap. |
| 1343 |
* @param array $sources Redirect-source sets from get_redirect_source_paths(). |
| 1344 |
* @return bool True when the URL redirects (exact match or known prefix). |
| 1345 |
*/ |
| 1346 |
private function is_redirect_source($url, $sources) |
| 1347 |
{ |
| 1348 |
if (empty($sources['exact']) && empty($sources['prefix'])) { |
| 1349 |
return false; |
| 1350 |
} |
| 1351 |
$path = $this->normalize_redirect_path($url); |
| 1352 |
if ('' === $path) { |
| 1353 |
return false; |
| 1354 |
} |
| 1355 |
if (isset($sources['exact'][$path])) { |
| 1356 |
return true; |
| 1357 |
} |
| 1358 |
foreach ($sources['prefix'] as $prefix) { |
| 1359 |
// A root prefix would match every URL and empty the sitemap — |
| 1360 |
// treat it as unusable rather than honoring it literally. |
| 1361 |
if ('/' !== $prefix && 0 === strpos($path, $prefix)) { |
| 1362 |
return true; |
| 1363 |
} |
| 1364 |
} |
| 1365 |
return false; |
| 1366 |
} |
| 1367 |
|
| 1368 |
/** |
| 1369 |
* Get SEO controls/indexation settings |
| 1370 |
* |
| 1371 |
* @return array The SEO controls settings |
| 1372 |
*/ |
| 1373 |
private function get_seo_controls() |
| 1374 |
{ |
| 1375 |
if (class_exists('Metasync')) { |
| 1376 |
return Metasync::get_option('seo_controls', []); |
| 1377 |
} |
| 1378 |
return get_option('metasync_seo_controls', []); |
| 1379 |
} |
| 1380 |
|
| 1381 |
/** |
| 1382 |
* Check if a taxonomy is set to noindex in indexation controls |
| 1383 |
* |
| 1384 |
* @param string $taxonomy The taxonomy name |
| 1385 |
* @param array $seo_controls The SEO controls settings |
| 1386 |
* @return bool True if the taxonomy should be excluded from sitemap |
| 1387 |
*/ |
| 1388 |
private function is_taxonomy_noindex($taxonomy, $seo_controls) |
| 1389 |
{ |
| 1390 |
// Map taxonomy names to their indexation control settings |
| 1391 |
$taxonomy_settings_map = [ |
| 1392 |
'post_tag' => 'index_tag_archives', |
| 1393 |
'category' => 'index_category_archives', |
| 1394 |
'post_format' => 'index_format_archives', |
| 1395 |
]; |
| 1396 |
|
| 1397 |
// Check if this taxonomy has an indexation control setting |
| 1398 |
if (isset($taxonomy_settings_map[$taxonomy])) { |
| 1399 |
$setting_key = $taxonomy_settings_map[$taxonomy]; |
| 1400 |
$setting_value = $seo_controls[$setting_key] ?? false; |
| 1401 |
|
| 1402 |
// If the setting is 'true' or true, it means noindex is enabled |
| 1403 |
// (The setting name is "index_*" but when true, it means "disable indexing") |
| 1404 |
if ($setting_value === 'true' || $setting_value === true) { |
| 1405 |
return true; |
| 1406 |
} |
| 1407 |
} |
| 1408 |
|
| 1409 |
return false; |
| 1410 |
} |
| 1411 |
|
| 1412 |
/** |
| 1413 |
* Add URL to sitemap |
| 1414 |
* |
| 1415 |
* @param DOMDocument $xml The XML document |
| 1416 |
* @param DOMElement $urlset The urlset element |
| 1417 |
* @param string $loc The URL location |
| 1418 |
* @param string $lastmod The last modified date |
| 1419 |
* @param string $priority The priority |
| 1420 |
* @param string $changefreq The change frequency |
| 1421 |
*/ |
| 1422 |
private function add_url_to_sitemap($xml, $urlset, $loc, $lastmod, $priority, $changefreq) |
| 1423 |
{ |
| 1424 |
$url = $xml->createElement('url'); |
| 1425 |
|
| 1426 |
$loc_element = $xml->createElement('loc', esc_url($loc)); |
| 1427 |
$url->appendChild($loc_element); |
| 1428 |
|
| 1429 |
if (!empty($lastmod)) { |
| 1430 |
$lastmod_timestamp = strtotime($lastmod); |
| 1431 |
// Legacy zero dates ("0000-00-00 00:00:00") make strtotime() |
| 1432 |
// return a pre-epoch timestamp, which would serialise as a |
| 1433 |
// year -0001 lastmod. lastmod is optional — skip it instead. |
| 1434 |
if ($lastmod_timestamp !== false && $lastmod_timestamp >= 0) { |
| 1435 |
$lastmod_formatted = gmdate('Y-m-d\TH:i:s+00:00', $lastmod_timestamp); |
| 1436 |
$lastmod_element = $xml->createElement('lastmod', $lastmod_formatted); |
| 1437 |
$url->appendChild($lastmod_element); |
| 1438 |
} |
| 1439 |
} |
| 1440 |
|
| 1441 |
$changefreq_element = $xml->createElement('changefreq', $changefreq); |
| 1442 |
$url->appendChild($changefreq_element); |
| 1443 |
|
| 1444 |
$priority_element = $xml->createElement('priority', $priority); |
| 1445 |
$url->appendChild($priority_element); |
| 1446 |
|
| 1447 |
$urlset->appendChild($url); |
| 1448 |
} |
| 1449 |
|
| 1450 |
/** |
| 1451 |
* Check if sitemap index exists |
| 1452 |
* |
| 1453 |
* @return bool |
| 1454 |
*/ |
| 1455 |
public function sitemap_exists() |
| 1456 |
{ |
| 1457 |
// Check virtual content first (now the primary storage path). |
| 1458 |
$virtual_content = $this->get_virtual_sitemap_file('sitemap_index.xml'); |
| 1459 |
if (false !== $virtual_content) { |
| 1460 |
return true; |
| 1461 |
} |
| 1462 |
|
| 1463 |
// Check tracked sitemap files option (survives transient eviction). |
| 1464 |
$tracked = get_option('metasync_sitemap_files', []); |
| 1465 |
if (!empty($tracked)) { |
| 1466 |
return true; |
| 1467 |
} |
| 1468 |
|
| 1469 |
// Fall back to physical file for legacy installs that haven't migrated yet. |
| 1470 |
return file_exists($this->sitemap_index_path); |
| 1471 |
} |
| 1472 |
|
| 1473 |
/** |
| 1474 |
* Get sitemap index content |
| 1475 |
* |
| 1476 |
* @return string|false |
| 1477 |
*/ |
| 1478 |
public function get_sitemap_content() |
| 1479 |
{ |
| 1480 |
// Check virtual content first (now the primary storage path). |
| 1481 |
$virtual_content = $this->get_virtual_sitemap_file('sitemap_index.xml'); |
| 1482 |
if (false !== $virtual_content) { |
| 1483 |
return $virtual_content; |
| 1484 |
} |
| 1485 |
|
| 1486 |
// Fall back to physical file for legacy installs that haven't migrated yet. |
| 1487 |
if (file_exists($this->sitemap_index_path)) { |
| 1488 |
return file_get_contents($this->sitemap_index_path); |
| 1489 |
} |
| 1490 |
|
| 1491 |
return false; |
| 1492 |
} |
| 1493 |
|
| 1494 |
/** |
| 1495 |
* Get sitemap index URL |
| 1496 |
* |
| 1497 |
* @return string |
| 1498 |
*/ |
| 1499 |
public function get_sitemap_url() |
| 1500 |
{ |
| 1501 |
return $this->sitemap_index_url; |
| 1502 |
} |
| 1503 |
|
| 1504 |
/** |
| 1505 |
* Get total size of all sitemap files |
| 1506 |
* |
| 1507 |
* @return int|false |
| 1508 |
*/ |
| 1509 |
public function get_sitemap_size() |
| 1510 |
{ |
| 1511 |
$sitemap_files = get_option('metasync_sitemap_files', []); |
| 1512 |
$filenames = ['sitemap_index.xml']; |
| 1513 |
foreach (is_array($sitemap_files) ? $sitemap_files : [] as $sitemap) { |
| 1514 |
if (!empty($sitemap['filename']) && is_string($sitemap['filename'])) { |
| 1515 |
$filenames[] = $sitemap['filename']; |
| 1516 |
} |
| 1517 |
} |
| 1518 |
|
| 1519 |
$total_size = 0; |
| 1520 |
$found_document = false; |
| 1521 |
foreach (array_unique($filenames) as $filename) { |
| 1522 |
// The document accessor prefers virtual/transient content and only |
| 1523 |
// falls back to legacy physical files when no virtual copy exists. |
| 1524 |
$content = $this->get_sitemap_document($filename); |
| 1525 |
if (false === $content) { |
| 1526 |
continue; |
| 1527 |
} |
| 1528 |
$found_document = true; |
| 1529 |
$total_size += strlen($content); |
| 1530 |
} |
| 1531 |
|
| 1532 |
return $found_document ? $total_size : false; |
| 1533 |
} |
| 1534 |
|
| 1535 |
/** |
| 1536 |
* Get last generation time |
| 1537 |
* |
| 1538 |
* @return string|false |
| 1539 |
*/ |
| 1540 |
public function get_last_generated_time() |
| 1541 |
{ |
| 1542 |
return get_option('metasync_sitemap_last_generated', false); |
| 1543 |
} |
| 1544 |
|
| 1545 |
/** |
| 1546 |
* Count total URLs across all sitemaps |
| 1547 |
* |
| 1548 |
* @return int |
| 1549 |
*/ |
| 1550 |
public function count_urls() |
| 1551 |
{ |
| 1552 |
return (int) get_option('metasync_sitemap_total_urls', 0); |
| 1553 |
} |
| 1554 |
|
| 1555 |
/** |
| 1556 |
* Get the list of generated sitemap files |
| 1557 |
* |
| 1558 |
* @return array |
| 1559 |
*/ |
| 1560 |
public function get_sitemap_files() |
| 1561 |
{ |
| 1562 |
return get_option('metasync_sitemap_files', []); |
| 1563 |
} |
| 1564 |
|
| 1565 |
/** |
| 1566 |
* Get the number of sitemap files |
| 1567 |
* |
| 1568 |
* @return int |
| 1569 |
*/ |
| 1570 |
public function get_sitemap_count() |
| 1571 |
{ |
| 1572 |
$files = $this->get_sitemap_files(); |
| 1573 |
return count($files); |
| 1574 |
} |
| 1575 |
|
| 1576 |
/** |
| 1577 |
* Get the maximum URLs per sitemap file |
| 1578 |
* |
| 1579 |
* @return int |
| 1580 |
*/ |
| 1581 |
public function get_urls_per_sitemap() |
| 1582 |
{ |
| 1583 |
return $this->urls_per_sitemap; |
| 1584 |
} |
| 1585 |
|
| 1586 |
/** |
| 1587 |
* Delete sitemap files (physical and virtual) |
| 1588 |
* |
| 1589 |
* @param string $type One of 'all', 'general', 'news', 'video'. Defaults to 'all'. |
| 1590 |
* @return bool |
| 1591 |
*/ |
| 1592 |
public function delete_sitemap($type = 'all') |
| 1593 |
{ |
| 1594 |
if (!in_array($type, ['all', 'general', 'news', 'video'], true)) { |
| 1595 |
return false; |
| 1596 |
} |
| 1597 |
|
| 1598 |
$deleted = false; |
| 1599 |
|
| 1600 |
if ('general' === $type || 'all' === $type) { |
| 1601 |
// Delete physical sitemap index |
| 1602 |
if (file_exists($this->sitemap_index_path)) { |
| 1603 |
@wp_delete_file($this->sitemap_index_path); |
| 1604 |
$deleted = true; |
| 1605 |
} |
| 1606 |
|
| 1607 |
// Delete the virtual sitemap index transient by name. sitemap_exists() |
| 1608 |
// reads this transient directly, so it must be cleared here even when |
| 1609 |
// metasync_sitemap_virtual_index no longer tracks it (an out-of-sync |
| 1610 |
// index leaves the transient orphaned, which made "Delete All Sitemaps" |
| 1611 |
// report failure while the sitemap kept being served. |
| 1612 |
$index_tkey = 'metasync_vsm_' . md5('sitemap_index.xml'); |
| 1613 |
if (false !== get_transient($index_tkey)) { |
| 1614 |
delete_transient($index_tkey); |
| 1615 |
$deleted = true; |
| 1616 |
} |
| 1617 |
|
| 1618 |
// Delete all physical sitemap files (sitemap.xml, sitemap2.xml, etc.) |
| 1619 |
// and their virtual transients (chunks stored in memory are tracked |
| 1620 |
// only by filename in metasync_sitemap_files). |
| 1621 |
$sitemap_files = get_option('metasync_sitemap_files', []); |
| 1622 |
foreach ($sitemap_files as $sitemap) { |
| 1623 |
$filename = is_array($sitemap) && isset($sitemap['filename']) ? $sitemap['filename'] : null; |
| 1624 |
if (null === $filename) { |
| 1625 |
continue; |
| 1626 |
} |
| 1627 |
$path = ABSPATH . $filename; |
| 1628 |
if (file_exists($path)) { |
| 1629 |
@wp_delete_file($path); |
| 1630 |
$deleted = true; |
| 1631 |
} |
| 1632 |
$chunk_tkey = 'metasync_vsm_' . md5($filename); |
| 1633 |
if (false !== get_transient($chunk_tkey)) { |
| 1634 |
delete_transient($chunk_tkey); |
| 1635 |
$deleted = true; |
| 1636 |
} |
| 1637 |
} |
| 1638 |
|
| 1639 |
// Delete only physical sitemap files tracked as MetaSync-owned. |
| 1640 |
$sitemap_files = get_option('metasync_sitemap_files', []); |
| 1641 |
$tracked_names = []; |
| 1642 |
foreach ($sitemap_files as $sitemap) { |
| 1643 |
if (is_array($sitemap) && !empty($sitemap['filename'])) { |
| 1644 |
$tracked_names[] = basename($sitemap['filename']); |
| 1645 |
} |
| 1646 |
} |
| 1647 |
foreach (array_unique($tracked_names) as $filename) { |
| 1648 |
$file = ABSPATH . $filename; |
| 1649 |
if (preg_match('/^sitemap\d*\.xml$/', $filename) && file_exists($file)) { |
| 1650 |
@wp_delete_file($file); |
| 1651 |
$deleted = true; |
| 1652 |
} |
| 1653 |
} |
| 1654 |
} |
| 1655 |
|
| 1656 |
if ('news' === $type || 'all' === $type) { |
| 1657 |
$news_file = 'news-sitemap.xml'; |
| 1658 |
$tkey = 'metasync_vsm_' . md5($news_file); |
| 1659 |
if (false !== get_transient($tkey)) { |
| 1660 |
delete_transient($tkey); |
| 1661 |
$deleted = true; |
| 1662 |
} |
| 1663 |
$physical = ABSPATH . $news_file; |
| 1664 |
if (file_exists($physical)) { |
| 1665 |
@wp_delete_file($physical); |
| 1666 |
$deleted = true; |
| 1667 |
} |
| 1668 |
|
| 1669 |
// Clear the news sitemap "enabled" setting so the feature gate no |
| 1670 |
// longer reports the sitemap feature as active once the file is gone. |
| 1671 |
// Without this, is_feature_enabled() keeps returning true (because |
| 1672 |
// metasync_news_sitemap_settings['enabled'] persists), so the |
| 1673 |
// "Default Behavior" robots.txt template keeps re-adding a stale |
| 1674 |
// Sitemap: line for a sitemap that no longer serves (WP-574). |
| 1675 |
$news_settings = get_option('metasync_news_sitemap_settings', []); |
| 1676 |
if (is_array($news_settings) && !empty($news_settings['enabled'])) { |
| 1677 |
$news_settings['enabled'] = false; |
| 1678 |
update_option('metasync_news_sitemap_settings', $news_settings); |
| 1679 |
} |
| 1680 |
} |
| 1681 |
|
| 1682 |
if ('video' === $type || 'all' === $type) { |
| 1683 |
$video_file = 'video-sitemap.xml'; |
| 1684 |
$tkey = 'metasync_vsm_' . md5($video_file); |
| 1685 |
if (false !== get_transient($tkey)) { |
| 1686 |
delete_transient($tkey); |
| 1687 |
$deleted = true; |
| 1688 |
} |
| 1689 |
$physical = ABSPATH . $video_file; |
| 1690 |
if (file_exists($physical)) { |
| 1691 |
@wp_delete_file($physical); |
| 1692 |
$deleted = true; |
| 1693 |
} |
| 1694 |
|
| 1695 |
// Clear the video sitemap "enabled" setting (see the news block above). |
| 1696 |
$video_settings = get_option('metasync_video_sitemap_settings', []); |
| 1697 |
if (is_array($video_settings) && !empty($video_settings['enabled'])) { |
| 1698 |
$video_settings['enabled'] = false; |
| 1699 |
update_option('metasync_video_sitemap_settings', $video_settings); |
| 1700 |
} |
| 1701 |
} |
| 1702 |
|
| 1703 |
if ('general' === $type || 'all' === $type) { |
| 1704 |
// Clear tracked virtual sitemap transients, preserving news/video |
| 1705 |
// when deleting only the general sitemap. |
| 1706 |
$preserved_filenames = ['news-sitemap.xml', 'video-sitemap.xml']; |
| 1707 |
$virtual_index = get_option('metasync_sitemap_virtual_index', []); |
| 1708 |
if (is_array($virtual_index)) { |
| 1709 |
$remaining_index = []; |
| 1710 |
foreach ($virtual_index as $filename => $tkey) { |
| 1711 |
if ('general' === $type && in_array($filename, $preserved_filenames, true)) { |
| 1712 |
$remaining_index[$filename] = $tkey; |
| 1713 |
continue; |
| 1714 |
} |
| 1715 |
if (is_string($tkey) && $tkey !== '') { |
| 1716 |
delete_transient($tkey); |
| 1717 |
$deleted = true; |
| 1718 |
} |
| 1719 |
} |
| 1720 |
if (!empty($remaining_index)) { |
| 1721 |
update_option('metasync_sitemap_virtual_index', $remaining_index, false); |
| 1722 |
} else { |
| 1723 |
delete_option('metasync_sitemap_virtual_index'); |
| 1724 |
} |
| 1725 |
} |
| 1726 |
|
| 1727 |
// Clear virtual content if it exists. |
| 1728 |
// Only fully disable virtual mode when deleting all — news/video |
| 1729 |
// still need it when only the general sitemap is removed. |
| 1730 |
if ('all' === $type) { |
| 1731 |
$virtual_mode = get_option('metasync_sitemap_virtual_mode', false); |
| 1732 |
$virtual_files = get_option('metasync_sitemap_virtual', array()); |
| 1733 |
if ($virtual_mode || !empty($virtual_files)) { |
| 1734 |
delete_option('metasync_sitemap_virtual'); |
| 1735 |
delete_option('metasync_sitemap_virtual_mode'); |
| 1736 |
$deleted = true; |
| 1737 |
} |
| 1738 |
} |
| 1739 |
|
| 1740 |
// Clear stored options. Clearing a tracking option that was actually |
| 1741 |
// present counts as a successful deletion — otherwise the caller shows |
| 1742 |
// a false-negative "Failed to delete sitemaps" after the state was in |
| 1743 |
// fact cleared. |
| 1744 |
foreach (['metasync_sitemap_files', 'metasync_sitemap_total_urls', 'metasync_sitemap_last_generated'] as $opt_name) { |
| 1745 |
if (false !== get_option($opt_name, false)) { |
| 1746 |
$deleted = true; |
| 1747 |
} |
| 1748 |
delete_option($opt_name); |
| 1749 |
} |
| 1750 |
} |
| 1751 |
|
| 1752 |
// Keep robots.txt in sync with the sitemap state. Deleting a sitemap is |
| 1753 |
// the inverse of generating one, so undo the robots.txt `Sitemap:` line |
| 1754 |
// the generator added at generation time (see generate_news_sitemap(), |
| 1755 |
// generate_video_sitemap() and update_robots_txt_sitemap()). Routing the |
| 1756 |
// change back through the same Metasync_Robots_Txt functions means the |
| 1757 |
// regenerated robots.txt is captured in the backup history exactly like a |
| 1758 |
// manual save (WP-574). |
| 1759 |
if ('general' === $type || 'all' === $type) { |
| 1760 |
$this->remove_robots_sitemap_line($this->sitemap_index_url); |
| 1761 |
} |
| 1762 |
if ('news' === $type || 'all' === $type) { |
| 1763 |
$this->remove_robots_sitemap_line(home_url('/news-sitemap.xml')); |
| 1764 |
} |
| 1765 |
if ('video' === $type || 'all' === $type) { |
| 1766 |
$this->remove_robots_sitemap_line(home_url('/video-sitemap.xml')); |
| 1767 |
} |
| 1768 |
|
| 1769 |
return $deleted; |
| 1770 |
} |
| 1771 |
|
| 1772 |
/** |
| 1773 |
* Remove a single MetaSync `Sitemap:` line from robots.txt, reusing the |
| 1774 |
* robots.txt manager's own removal path. |
| 1775 |
* |
| 1776 |
* This is the inverse of the add/update performed when a sitemap is |
| 1777 |
* generated. It delegates to {@see Metasync_Robots_Txt::remove_sitemap_url()}, |
| 1778 |
* which persists via write_robots_file() — so the current robots.txt is |
| 1779 |
* backed up before the stale line is dropped, and the change shows up in the |
| 1780 |
* backup history. A no-op (no write, no backup) when the line is not present |
| 1781 |
* or the robots.txt manager class is unavailable. |
| 1782 |
* |
| 1783 |
* @param string $sitemap_url Absolute sitemap URL to remove. |
| 1784 |
* @return void |
| 1785 |
*/ |
| 1786 |
private function remove_robots_sitemap_line($sitemap_url) |
| 1787 |
{ |
| 1788 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 1789 |
$robots_txt_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 1790 |
if (!file_exists($robots_txt_file)) { |
| 1791 |
return; |
| 1792 |
} |
| 1793 |
require_once $robots_txt_file; |
| 1794 |
} |
| 1795 |
|
| 1796 |
Metasync_Robots_Txt::get_instance()->remove_sitemap_url($sitemap_url); |
| 1797 |
} |
| 1798 |
|
| 1799 |
/** |
| 1800 |
* Check if other sitemap plugins are active |
| 1801 |
* |
| 1802 |
* @return array Array of active sitemap plugins |
| 1803 |
*/ |
| 1804 |
public function check_active_sitemap_plugins() |
| 1805 |
{ |
| 1806 |
// Ensure plugin.php is loaded for is_plugin_active() |
| 1807 |
if (!function_exists('is_plugin_active')) { |
| 1808 |
require_once ABSPATH . 'wp-admin/includes/plugin.php'; |
| 1809 |
} |
| 1810 |
|
| 1811 |
$sitemap_plugins = [ |
| 1812 |
'wordpress-seo/wp-seo.php' => 'Yoast SEO', |
| 1813 |
'all-in-one-seo-pack/all_in_one_seo_pack.php' => 'All in One SEO', |
| 1814 |
'google-sitemap-generator/sitemap.php' => 'Google XML Sitemaps', |
| 1815 |
'better-wp-google-xml-sitemaps/bwp-simple-gxs.php' => 'Better WordPress Google XML Sitemaps', |
| 1816 |
'xml-sitemap-feed/xml-sitemap.php' => 'XML Sitemap & Google News', |
| 1817 |
'sitemap/sitemap.php' => 'Google Sitemap Plugin', |
| 1818 |
'rank-math/rank-math.php' => 'Rank Math SEO', |
| 1819 |
'seo-by-rank-math/rank-math.php' => 'Rank Math SEO (Free)', |
| 1820 |
'wordpress-seo-premium/wp-seo-premium.php' => 'Yoast SEO Premium', |
| 1821 |
]; |
| 1822 |
|
| 1823 |
$active_plugins = []; |
| 1824 |
foreach ($sitemap_plugins as $plugin_path => $plugin_name) { |
| 1825 |
if (is_plugin_active($plugin_path)) { |
| 1826 |
$active_plugins[$plugin_path] = $plugin_name; |
| 1827 |
} |
| 1828 |
} |
| 1829 |
|
| 1830 |
return $active_plugins; |
| 1831 |
} |
| 1832 |
|
| 1833 |
/** |
| 1834 |
* Disable sitemap generation from other plugins |
| 1835 |
* |
| 1836 |
* @return bool True if any changes were made |
| 1837 |
*/ |
| 1838 |
public function disable_other_sitemap_generators() |
| 1839 |
{ |
| 1840 |
$changes_made = false; |
| 1841 |
|
| 1842 |
// Disable Yoast SEO sitemap |
| 1843 |
if (class_exists('WPSEO_Options')) { |
| 1844 |
$yoast_options = get_option('wpseo'); |
| 1845 |
if (isset($yoast_options['enable_xml_sitemap']) && $yoast_options['enable_xml_sitemap'] === true) { |
| 1846 |
$yoast_options['enable_xml_sitemap'] = false; |
| 1847 |
update_option('wpseo', $yoast_options); |
| 1848 |
$changes_made = true; |
| 1849 |
} |
| 1850 |
} |
| 1851 |
|
| 1852 |
// Disable All in One SEO sitemap (options stored as JSON string) |
| 1853 |
if (function_exists('aioseo')) { |
| 1854 |
$aioseo_raw = get_option('aioseo_options'); |
| 1855 |
$aioseo_options = is_string($aioseo_raw) ? json_decode($aioseo_raw, true) : $aioseo_raw; |
| 1856 |
if (is_array($aioseo_options) && !empty($aioseo_options['sitemap']['general']['enable'])) { |
| 1857 |
$aioseo_options['sitemap']['general']['enable'] = false; |
| 1858 |
update_option('aioseo_options', is_string($aioseo_raw) ? wp_json_encode($aioseo_options) : $aioseo_options); |
| 1859 |
$changes_made = true; |
| 1860 |
} |
| 1861 |
} |
| 1862 |
|
| 1863 |
// Disable Rank Math sitemap |
| 1864 |
if (class_exists('RankMath')) { |
| 1865 |
$rank_math_modules = get_option('rank_math_modules', []); |
| 1866 |
if (is_array($rank_math_modules) && in_array('sitemap', $rank_math_modules, true)) { |
| 1867 |
$rank_math_modules = array_values(array_diff($rank_math_modules, ['sitemap'])); |
| 1868 |
update_option('rank_math_modules', $rank_math_modules); |
| 1869 |
$changes_made = true; |
| 1870 |
} |
| 1871 |
} |
| 1872 |
|
| 1873 |
// Disable WordPress core sitemap (WP 5.5+) - Store in option for persistence |
| 1874 |
$wp_core_disabled = get_option('metasync_disable_wp_sitemap', false); |
| 1875 |
if (!$wp_core_disabled) { |
| 1876 |
update_option('metasync_disable_wp_sitemap', true); |
| 1877 |
$changes_made = true; |
| 1878 |
} |
| 1879 |
|
| 1880 |
return $changes_made; |
| 1881 |
} |
| 1882 |
|
| 1883 |
/** |
| 1884 |
* Re-enable sitemap generation from other plugins |
| 1885 |
* |
| 1886 |
* @return bool True if any changes were made |
| 1887 |
*/ |
| 1888 |
public function enable_other_sitemap_generators() |
| 1889 |
{ |
| 1890 |
$changes_made = false; |
| 1891 |
|
| 1892 |
// Re-enable Yoast SEO sitemap |
| 1893 |
if (class_exists('WPSEO_Options')) { |
| 1894 |
$yoast_options = get_option('wpseo'); |
| 1895 |
if (isset($yoast_options['enable_xml_sitemap']) && $yoast_options['enable_xml_sitemap'] === false) { |
| 1896 |
$yoast_options['enable_xml_sitemap'] = true; |
| 1897 |
update_option('wpseo', $yoast_options); |
| 1898 |
$changes_made = true; |
| 1899 |
} |
| 1900 |
} |
| 1901 |
|
| 1902 |
// Re-enable All in One SEO sitemap (options stored as JSON string) |
| 1903 |
if (function_exists('aioseo')) { |
| 1904 |
$aioseo_raw = get_option('aioseo_options'); |
| 1905 |
$aioseo_options = is_string($aioseo_raw) ? json_decode($aioseo_raw, true) : $aioseo_raw; |
| 1906 |
if (is_array($aioseo_options) && isset($aioseo_options['sitemap']['general']['enable']) && !$aioseo_options['sitemap']['general']['enable']) { |
| 1907 |
$aioseo_options['sitemap']['general']['enable'] = true; |
| 1908 |
update_option('aioseo_options', is_string($aioseo_raw) ? wp_json_encode($aioseo_options) : $aioseo_options); |
| 1909 |
$changes_made = true; |
| 1910 |
} |
| 1911 |
} |
| 1912 |
|
| 1913 |
// Re-enable Rank Math sitemap |
| 1914 |
if (class_exists('RankMath')) { |
| 1915 |
$rank_math_modules = get_option('rank_math_modules', []); |
| 1916 |
if (is_array($rank_math_modules) && !in_array('sitemap', $rank_math_modules, true)) { |
| 1917 |
$rank_math_modules[] = 'sitemap'; |
| 1918 |
update_option('rank_math_modules', $rank_math_modules); |
| 1919 |
$changes_made = true; |
| 1920 |
} |
| 1921 |
} |
| 1922 |
|
| 1923 |
// Re-enable WordPress core sitemap (WP 5.5+) |
| 1924 |
$wp_core_disabled = get_option('metasync_disable_wp_sitemap', false); |
| 1925 |
if ($wp_core_disabled) { |
| 1926 |
delete_option('metasync_disable_wp_sitemap'); |
| 1927 |
$changes_made = true; |
| 1928 |
} |
| 1929 |
|
| 1930 |
return $changes_made; |
| 1931 |
} |
| 1932 |
|
| 1933 |
/** |
| 1934 |
* Setup automatic sitemap updates on post changes |
| 1935 |
*/ |
| 1936 |
public function setup_auto_update_hooks() |
| 1937 |
{ |
| 1938 |
// Post hooks |
| 1939 |
add_action('save_post', [$this, 'auto_update_sitemap'], 10, 2); |
| 1940 |
add_action('delete_post', [$this, 'auto_update_sitemap_simple']); |
| 1941 |
add_action('trashed_post', [$this, 'auto_update_sitemap_simple']); |
| 1942 |
add_action('untrashed_post', [$this, 'auto_update_sitemap_simple']); |
| 1943 |
|
| 1944 |
// Term hooks |
| 1945 |
add_action('created_term', [$this, 'auto_update_sitemap_simple']); |
| 1946 |
add_action('edited_term', [$this, 'auto_update_sitemap_simple']); |
| 1947 |
add_action('delete_term', [$this, 'auto_update_sitemap_simple']); |
| 1948 |
|
| 1949 |
// News sitemap auto-update on post save |
| 1950 |
add_action('save_post', [$this, 'auto_update_news_sitemap'], 10, 2); |
| 1951 |
|
| 1952 |
// Video sitemap auto-update on post save |
| 1953 |
add_action('save_post', [$this, 'auto_update_video_sitemap'], 10, 2); |
| 1954 |
} |
| 1955 |
|
| 1956 |
/** |
| 1957 |
* Auto update sitemap on post save |
| 1958 |
* |
| 1959 |
* @param int $post_id The post ID |
| 1960 |
* @param WP_Post $post The post object |
| 1961 |
*/ |
| 1962 |
public function auto_update_sitemap($post_id, $post) |
| 1963 |
{ |
| 1964 |
// Check if auto-update is enabled |
| 1965 |
if (!get_option('metasync_sitemap_auto_update', false)) { |
| 1966 |
return; |
| 1967 |
} |
| 1968 |
|
| 1969 |
// Avoid auto-updates during autosave, revisions, or drafts |
| 1970 |
if (wp_is_post_autosave($post_id) || wp_is_post_revision($post_id)) { |
| 1971 |
return; |
| 1972 |
} |
| 1973 |
|
| 1974 |
// Get post object if not provided (some hooks don't pass it) |
| 1975 |
if (empty($post)) { |
| 1976 |
$post = get_post($post_id); |
| 1977 |
} |
| 1978 |
|
| 1979 |
// Validate post object exists |
| 1980 |
if (!$post || !is_object($post)) { |
| 1981 |
return; |
| 1982 |
} |
| 1983 |
|
| 1984 |
// Only update for published posts |
| 1985 |
if ($post->post_status !== 'publish') { |
| 1986 |
return; |
| 1987 |
} |
| 1988 |
|
| 1989 |
// Regenerate sitemap |
| 1990 |
$this->generate_sitemap(); |
| 1991 |
} |
| 1992 |
|
| 1993 |
/** |
| 1994 |
* Auto update sitemap (simple version for hooks without post object) |
| 1995 |
*/ |
| 1996 |
public function auto_update_sitemap_simple() |
| 1997 |
{ |
| 1998 |
// Check if auto-update is enabled |
| 1999 |
if (!get_option('metasync_sitemap_auto_update', false)) { |
| 2000 |
return; |
| 2001 |
} |
| 2002 |
|
| 2003 |
// Regenerate sitemap |
| 2004 |
$this->generate_sitemap(); |
| 2005 |
} |
| 2006 |
|
| 2007 |
/** |
| 2008 |
* Update robots.txt with the sitemap index URL |
| 2009 |
* |
| 2010 |
* @return array Result with 'success' boolean and 'action' string |
| 2011 |
*/ |
| 2012 |
public function update_robots_txt_sitemap() |
| 2013 |
{ |
| 2014 |
// Load the robots.txt class if not already loaded |
| 2015 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 2016 |
$robots_txt_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 2017 |
if (file_exists($robots_txt_file)) { |
| 2018 |
require_once $robots_txt_file; |
| 2019 |
} else { |
| 2020 |
return [ |
| 2021 |
'success' => false, |
| 2022 |
'action' => 'error', |
| 2023 |
'message' => esc_html__('Robots.txt management class not found.', 'metasync') |
| 2024 |
]; |
| 2025 |
} |
| 2026 |
} |
| 2027 |
|
| 2028 |
// Get the robots.txt instance and update the sitemap index URL |
| 2029 |
$robots_txt = Metasync_Robots_Txt::get_instance(); |
| 2030 |
return $robots_txt->update_sitemap_url($this->sitemap_index_url); |
| 2031 |
} |
| 2032 |
|
| 2033 |
/** |
| 2034 |
* Check if robots.txt contains the correct sitemap URL |
| 2035 |
* |
| 2036 |
* @return bool True if robots.txt has the correct sitemap URL |
| 2037 |
*/ |
| 2038 |
public function robots_has_sitemap_url() |
| 2039 |
{ |
| 2040 |
// Load the robots.txt class if not already loaded |
| 2041 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 2042 |
$robots_txt_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 2043 |
if (file_exists($robots_txt_file)) { |
| 2044 |
require_once $robots_txt_file; |
| 2045 |
} else { |
| 2046 |
return false; |
| 2047 |
} |
| 2048 |
} |
| 2049 |
|
| 2050 |
$robots_txt = Metasync_Robots_Txt::get_instance(); |
| 2051 |
return $robots_txt->has_sitemap_url($this->sitemap_index_url); |
| 2052 |
} |
| 2053 |
|
| 2054 |
/** |
| 2055 |
* Generate the news sitemap. |
| 2056 |
* |
| 2057 |
* @return bool True on success, false on failure. |
| 2058 |
*/ |
| 2059 |
public function generate_news_sitemap() |
| 2060 |
{ |
| 2061 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-sitemap-news.php'; |
| 2062 |
|
| 2063 |
$news = new Metasync_Sitemap_News(); |
| 2064 |
$xml = $news->generate(); |
| 2065 |
|
| 2066 |
if (false === $xml) { |
| 2067 |
return false; |
| 2068 |
} |
| 2069 |
|
| 2070 |
// Only advertise the sitemap once it is actually stored — otherwise |
| 2071 |
// robots.txt and the Google ping point at content that cannot be served. |
| 2072 |
if (false === $this->store_virtual_sitemap_file('news-sitemap.xml', $xml)) { |
| 2073 |
return false; |
| 2074 |
} |
| 2075 |
|
| 2076 |
// Add to robots.txt |
| 2077 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 2078 |
$robots_txt_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 2079 |
if (file_exists($robots_txt_file)) { |
| 2080 |
require_once $robots_txt_file; |
| 2081 |
} |
| 2082 |
} |
| 2083 |
|
| 2084 |
if (class_exists('Metasync_Robots_Txt')) { |
| 2085 |
$robots_txt = Metasync_Robots_Txt::get_instance(); |
| 2086 |
$robots_txt->add_sitemap_url(home_url('/news-sitemap.xml')); |
| 2087 |
} |
| 2088 |
|
| 2089 |
// Ping Google |
| 2090 |
$news->ping_google(home_url('/news-sitemap.xml')); |
| 2091 |
|
| 2092 |
return true; |
| 2093 |
} |
| 2094 |
|
| 2095 |
/** |
| 2096 |
* Generate the video sitemap. |
| 2097 |
* |
| 2098 |
* @return bool True on success, false on failure. |
| 2099 |
*/ |
| 2100 |
public function generate_video_sitemap() |
| 2101 |
{ |
| 2102 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-sitemap-video.php'; |
| 2103 |
|
| 2104 |
$video = new Metasync_Sitemap_Video(); |
| 2105 |
$xml = $video->generate(); |
| 2106 |
|
| 2107 |
if (false === $xml) { |
| 2108 |
return false; |
| 2109 |
} |
| 2110 |
|
| 2111 |
// Same as the news sitemap: only advertise it once it is actually stored. |
| 2112 |
if (false === $this->store_virtual_sitemap_file('video-sitemap.xml', $xml)) { |
| 2113 |
return false; |
| 2114 |
} |
| 2115 |
|
| 2116 |
// Record why entries were left out, so the admin screen can name the |
| 2117 |
// actual cause instead of reporting a bare "0 URLs", which reads as |
| 2118 |
// "this is normal" even on a site full of videos. |
| 2119 |
// |
| 2120 |
// Written AFTER the store succeeds: diagnostics that describe a |
| 2121 |
// sitemap which was never published would be reported against the |
| 2122 |
// older document still being served. |
| 2123 |
update_option( |
| 2124 |
'metasync_video_sitemap_diagnostics', |
| 2125 |
[ |
| 2126 |
'skipped_no_thumbnail' => $video->get_skipped_no_thumbnail_count(), |
| 2127 |
'skipped_no_thumbnail_ids' => $video->get_skipped_no_thumbnail_post_ids(), |
| 2128 |
'skipped_no_thumbnail_videos' => $video->get_skipped_no_thumbnail_video_count(), |
| 2129 |
'skipped_duplicate_thumbnail' => $video->get_skipped_duplicate_thumbnail_count(), |
| 2130 |
'generated_at' => current_time('mysql'), |
| 2131 |
], |
| 2132 |
false |
| 2133 |
); |
| 2134 |
|
| 2135 |
// Add to robots.txt |
| 2136 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 2137 |
$robots_txt_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 2138 |
if (file_exists($robots_txt_file)) { |
| 2139 |
require_once $robots_txt_file; |
| 2140 |
} |
| 2141 |
} |
| 2142 |
|
| 2143 |
if (class_exists('Metasync_Robots_Txt')) { |
| 2144 |
$robots_txt = Metasync_Robots_Txt::get_instance(); |
| 2145 |
$robots_txt->add_sitemap_url(home_url('/video-sitemap.xml')); |
| 2146 |
} |
| 2147 |
|
| 2148 |
return true; |
| 2149 |
} |
| 2150 |
|
| 2151 |
/** |
| 2152 |
* Auto-update news sitemap on post save. |
| 2153 |
* |
| 2154 |
* @param int $post_id The post ID. |
| 2155 |
* @param WP_Post $post The post object. |
| 2156 |
*/ |
| 2157 |
public function auto_update_news_sitemap($post_id, $post) |
| 2158 |
{ |
| 2159 |
$settings = get_option('metasync_news_sitemap_settings', []); |
| 2160 |
if (empty($settings['enabled'])) { |
| 2161 |
return; |
| 2162 |
} |
| 2163 |
|
| 2164 |
// Skip autosave and revisions |
| 2165 |
if (wp_is_post_autosave($post_id) || wp_is_post_revision($post_id)) { |
| 2166 |
return; |
| 2167 |
} |
| 2168 |
|
| 2169 |
if (empty($post)) { |
| 2170 |
$post = get_post($post_id); |
| 2171 |
} |
| 2172 |
|
| 2173 |
if (!$post || !is_object($post)) { |
| 2174 |
return; |
| 2175 |
} |
| 2176 |
|
| 2177 |
if ($post->post_status !== 'publish') { |
| 2178 |
return; |
| 2179 |
} |
| 2180 |
|
| 2181 |
// Only regenerate if the post was modified within the last 2 days |
| 2182 |
$post_time = strtotime($post->post_modified_gmt); |
| 2183 |
$cutoff = strtotime('-2 days'); |
| 2184 |
if (false === $post_time || false === $cutoff || $post_time < $cutoff) { |
| 2185 |
return; |
| 2186 |
} |
| 2187 |
|
| 2188 |
$this->generate_news_sitemap(); |
| 2189 |
} |
| 2190 |
|
| 2191 |
/** |
| 2192 |
* Auto-update video sitemap on post save. |
| 2193 |
* |
| 2194 |
* @param int $post_id The post ID. |
| 2195 |
* @param WP_Post $post The post object. |
| 2196 |
*/ |
| 2197 |
public function auto_update_video_sitemap($post_id, $post) |
| 2198 |
{ |
| 2199 |
$settings = get_option('metasync_video_sitemap_settings', []); |
| 2200 |
if (empty($settings['enabled'])) { |
| 2201 |
return; |
| 2202 |
} |
| 2203 |
|
| 2204 |
// Skip autosave and revisions |
| 2205 |
if (wp_is_post_autosave($post_id) || wp_is_post_revision($post_id)) { |
| 2206 |
return; |
| 2207 |
} |
| 2208 |
|
| 2209 |
if (empty($post)) { |
| 2210 |
$post = get_post($post_id); |
| 2211 |
} |
| 2212 |
|
| 2213 |
if (!$post || !is_object($post)) { |
| 2214 |
return; |
| 2215 |
} |
| 2216 |
|
| 2217 |
if ($post->post_status !== 'publish') { |
| 2218 |
return; |
| 2219 |
} |
| 2220 |
|
| 2221 |
// Only regenerate if the post type is included in video sitemap settings |
| 2222 |
$post_types = !empty($settings['post_types']) ? (array) $settings['post_types'] : ['post', 'page']; |
| 2223 |
if (!in_array($post->post_type, $post_types, true)) { |
| 2224 |
return; |
| 2225 |
} |
| 2226 |
|
| 2227 |
$this->generate_video_sitemap(); |
| 2228 |
} |
| 2229 |
|
| 2230 |
/** |
| 2231 |
* How long a stored sitemap document stays valid. |
| 2232 |
* |
| 2233 |
* Most sitemaps describe content that only changes when the site changes, |
| 2234 |
* and bust_sitemap_cache() already clears them on any post or term edit, |
| 2235 |
* so a long TTL is right for them. |
| 2236 |
* |
| 2237 |
* The news sitemap is different: Google News only accepts entries from the |
| 2238 |
* last 48 hours, and the generator applies that window once, at build |
| 2239 |
* time, via a `date_query`. Nothing re-evaluates it as the clock advances. |
| 2240 |
* On a quiet site — a publisher pausing over a weekend, or any stretch |
| 2241 |
* with no content edits to trigger a bust — a 30-day cache kept |
| 2242 |
* advertising Friday's articles as news well past the window Google |
| 2243 |
* accepts. Expiring inside that window hands the next request to the |
| 2244 |
* existing regenerate-on-miss path, which rebuilds with a fresh |
| 2245 |
* `date_query`. No cron needed. |
| 2246 |
* |
| 2247 |
* @param string $filename The sitemap filename. |
| 2248 |
* @return int TTL in seconds. |
| 2249 |
*/ |
| 2250 |
private function get_sitemap_cache_ttl($filename) |
| 2251 |
{ |
| 2252 |
if ('news-sitemap.xml' === $filename) { |
| 2253 |
// Comfortably inside the 48-hour window, so a served document is |
| 2254 |
// always rebuilt while its own entries are still eligible. |
| 2255 |
// Expressed in DAY_IN_SECONDS because that is the only time |
| 2256 |
// constant this class relies on being defined. |
| 2257 |
return (int) (DAY_IN_SECONDS / 4); |
| 2258 |
} |
| 2259 |
|
| 2260 |
return 30 * DAY_IN_SECONDS; |
| 2261 |
} |
| 2262 |
|
| 2263 |
/** |
| 2264 |
* Store virtual sitemap file content using individual transients. |
| 2265 |
* |
| 2266 |
* The transient is the only durable copy of a virtual sitemap, so a failed |
| 2267 |
* write means the site has no sitemap to serve. Report that to the caller |
| 2268 |
* instead of claiming success. |
| 2269 |
* |
| 2270 |
* @param string $filename The sitemap filename (e.g., 'sitemap.xml', 'sitemap_index.xml') |
| 2271 |
* @param string $content The XML content |
| 2272 |
* @return bool True on success, false when the content could not be stored |
| 2273 |
*/ |
| 2274 |
public function store_virtual_sitemap_file($filename, $content) |
| 2275 |
{ |
| 2276 |
$cache_key = 'metasync_vsm_' . md5($filename); |
| 2277 |
$stored = set_transient($cache_key, $content, $this->get_sitemap_cache_ttl($filename)); |
| 2278 |
|
| 2279 |
if (false === $stored) { |
| 2280 |
// set_transient() also returns false when the stored value is |
| 2281 |
// unchanged, because update_option() short-circuits identical |
| 2282 |
// values. Only treat this as a failure when the content cannot be |
| 2283 |
// read back. |
| 2284 |
$stored = (get_transient($cache_key) === $content); |
| 2285 |
} |
| 2286 |
|
| 2287 |
if (false === $stored) { |
| 2288 |
error_log('MetaSync: failed to store virtual sitemap "' . $filename . '" — the sitemap will not be served.'); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- genuine failure path, bounded, no secrets |
| 2289 |
return false; |
| 2290 |
} |
| 2291 |
|
| 2292 |
// Track which virtual sitemaps exist |
| 2293 |
$index = get_option('metasync_sitemap_virtual_index', []); |
| 2294 |
$index[$filename] = $cache_key; |
| 2295 |
update_option('metasync_sitemap_virtual_index', $index, false); |
| 2296 |
update_option('metasync_sitemap_virtual_mode', true, false); |
| 2297 |
return true; |
| 2298 |
} |
| 2299 |
|
| 2300 |
/** |
| 2301 |
* Get virtual sitemap file content from transient. |
| 2302 |
* |
| 2303 |
* @param string $filename The sitemap filename |
| 2304 |
* @return string|false Content or false if not found |
| 2305 |
*/ |
| 2306 |
private function get_virtual_sitemap_file($filename) |
| 2307 |
{ |
| 2308 |
$cache_key = 'metasync_vsm_' . md5($filename); |
| 2309 |
$content = get_transient($cache_key); |
| 2310 |
|
| 2311 |
if (false !== $content) { |
| 2312 |
return $content; |
| 2313 |
} |
| 2314 |
|
| 2315 |
// Fallback: check legacy option storage for backward compatibility |
| 2316 |
$virtual_sitemaps = get_option('metasync_sitemap_virtual', []); |
| 2317 |
if (isset($virtual_sitemaps[$filename])) { |
| 2318 |
// Migrate to transient |
| 2319 |
$this->store_virtual_sitemap_file($filename, $virtual_sitemaps[$filename]); |
| 2320 |
unset($virtual_sitemaps[$filename]); |
| 2321 |
if (empty($virtual_sitemaps)) { |
| 2322 |
delete_option('metasync_sitemap_virtual'); |
| 2323 |
} else { |
| 2324 |
update_option('metasync_sitemap_virtual', $virtual_sitemaps, false); |
| 2325 |
} |
| 2326 |
return get_transient($cache_key); |
| 2327 |
} |
| 2328 |
|
| 2329 |
return false; |
| 2330 |
} |
| 2331 |
|
| 2332 |
/** |
| 2333 |
* Single source of truth for a sitemap document's content. |
| 2334 |
* |
| 2335 |
* Every consumer — the served route, the admin status tiles, the MCP |
| 2336 |
* tools — must read through this accessor. They used to each implement |
| 2337 |
* their own precedence and disagreed: the served route preferred the |
| 2338 |
* physical file while the admin screen preferred the transient, so a site |
| 2339 |
* holding both counted one document in the dashboard and handed crawlers |
| 2340 |
* the other. No amount of regeneration reconciles that, because each side |
| 2341 |
* keeps reading its own preferred copy. |
| 2342 |
* |
| 2343 |
* Transients win. Virtual mode is the supported storage model — the |
| 2344 |
* generator has written nothing to disk since the physical writer was |
| 2345 |
* retired — so a file at ABSPATH is always a foreign artifact: a host |
| 2346 |
* backup/restore, a rollback to an older build, or a hand-placed copy. |
| 2347 |
* Rather than let it shadow live content, absorb it the way activation's |
| 2348 |
* migration would have: adopt it only when we have no transient, and |
| 2349 |
* remove it once its content is safely stored. |
| 2350 |
* |
| 2351 |
* @param string $filename The sitemap filename (e.g. 'news-sitemap.xml'). |
| 2352 |
* @return string|false Content, or false when no source holds this file. |
| 2353 |
*/ |
| 2354 |
public function get_sitemap_document($filename) |
| 2355 |
{ |
| 2356 |
$content = $this->get_virtual_sitemap_file($filename); |
| 2357 |
if (false !== $content) { |
| 2358 |
// A physical copy alongside live virtual content is exactly the |
| 2359 |
// divergence that made the dashboard and crawlers disagree. The |
| 2360 |
// transient is authoritative, so retire the file. |
| 2361 |
$this->absorb_physical_sitemap($filename, false); |
| 2362 |
return $content; |
| 2363 |
} |
| 2364 |
|
| 2365 |
// No transient: a physical file is the only copy left, so adopt it |
| 2366 |
// instead of reporting the sitemap missing. |
| 2367 |
return $this->absorb_physical_sitemap($filename, true); |
| 2368 |
} |
| 2369 |
|
| 2370 |
/** |
| 2371 |
* Reconcile a stray physical sitemap file with virtual storage. |
| 2372 |
* |
| 2373 |
* Activation runs metasync_migrate_physical_sitemaps() once, so any file |
| 2374 |
* appearing afterwards used to shadow the transient forever. Calling this |
| 2375 |
* from the read path closes that window: the file is either adopted (when |
| 2376 |
* it is the only copy) or removed (when a transient already holds the |
| 2377 |
* live document). |
| 2378 |
* |
| 2379 |
* @param string $filename The sitemap filename. |
| 2380 |
* @param bool $adopt Store the file's content before removing it. |
| 2381 |
* @return string|false Adopted content, or false when nothing was adopted. |
| 2382 |
*/ |
| 2383 |
private function absorb_physical_sitemap($filename, $adopt) |
| 2384 |
{ |
| 2385 |
// Reject traversal before touching the filesystem: $filename reaches |
| 2386 |
// the served route from a query var. |
| 2387 |
$basename = basename($filename); |
| 2388 |
if ($basename !== $filename || '' === $basename) { |
| 2389 |
return false; |
| 2390 |
} |
| 2391 |
|
| 2392 |
$physical_path = ABSPATH . $basename; |
| 2393 |
if (!file_exists($physical_path) || !is_readable($physical_path)) { |
| 2394 |
return false; |
| 2395 |
} |
| 2396 |
|
| 2397 |
if (!$adopt) { |
| 2398 |
$this->safe_unlink($physical_path); |
| 2399 |
return false; |
| 2400 |
} |
| 2401 |
|
| 2402 |
$content = file_get_contents($physical_path); |
| 2403 |
if (false === $content) { |
| 2404 |
return false; |
| 2405 |
} |
| 2406 |
|
| 2407 |
// Only drop the file once its content is durably stored, so a failed |
| 2408 |
// write can never destroy the site's last copy of the sitemap. |
| 2409 |
if ($this->store_virtual_sitemap_file($basename, $content)) { |
| 2410 |
$this->safe_unlink($physical_path); |
| 2411 |
} |
| 2412 |
|
| 2413 |
return $content; |
| 2414 |
} |
| 2415 |
|
| 2416 |
/** |
| 2417 |
* Absorb stray physical sitemap files, from a hook that always runs PHP. |
| 2418 |
* |
| 2419 |
* get_sitemap_document() reconciles a stray file whenever something reads |
| 2420 |
* that document, but on Apache the one request that most needs it never |
| 2421 |
* arrives: WordPress' rewrite block ends with |
| 2422 |
* `RewriteCond %{REQUEST_FILENAME} !-f`, so a physical news-sitemap.xml at |
| 2423 |
* ABSPATH is served straight off disk and index.php is never loaded. The |
| 2424 |
* file would then shadow live content until an admin happened to open the |
| 2425 |
* XML Sitemap screen. Running from `init` means any page view on the site |
| 2426 |
* closes the window instead. |
| 2427 |
* |
| 2428 |
* Kept deliberately cheap, because this runs on ordinary requests: |
| 2429 |
* |
| 2430 |
* - throttled to once per hour per site via a transient, so the common |
| 2431 |
* case costs one cache read and nothing else; |
| 2432 |
* - only ever looks at the handful of filenames MetaSync itself owns — |
| 2433 |
* never a glob, and never a third-party provider's sitemap; |
| 2434 |
* - stats the filesystem before reading any transient, so a site with no |
| 2435 |
* stray files (the normal state) does a few file_exists() calls an hour |
| 2436 |
* and no document reads at all; |
| 2437 |
* - self-limiting, since absorbing a file removes it. |
| 2438 |
* |
| 2439 |
* @return void |
| 2440 |
*/ |
| 2441 |
public function absorb_stray_physical_sitemaps() |
| 2442 |
{ |
| 2443 |
// This sweep is deliberately read/reconcile-only: it only handles |
| 2444 |
// filenames already owned by MetaSync and is throttled below. |
| 2445 |
// It must remain available on frontend requests because Apache may |
| 2446 |
// otherwise serve a stale physical sitemap before WordPress routing. |
| 2447 |
|
| 2448 |
// Cheap cross-request throttle. A miss here is the only path that |
| 2449 |
// touches the filesystem. Expressed in DAY_IN_SECONDS because that is |
| 2450 |
// the only time constant this class relies on being defined. |
| 2451 |
$throttle_key = 'metasync_sitemap_absorb_sweep'; |
| 2452 |
if (false !== get_transient($throttle_key)) { |
| 2453 |
return; |
| 2454 |
} |
| 2455 |
set_transient($throttle_key, 1, (int) (DAY_IN_SECONDS / 24)); |
| 2456 |
|
| 2457 |
foreach ($this->get_owned_sitemap_filenames() as $filename) { |
| 2458 |
// Stat first: the overwhelmingly common case is that no stray file |
| 2459 |
// exists, and that must not cost a transient read per filename. |
| 2460 |
if (!file_exists(ABSPATH . $filename)) { |
| 2461 |
continue; |
| 2462 |
} |
| 2463 |
|
| 2464 |
// Reading through the accessor is what performs the reconciliation: |
| 2465 |
// it adopts the file when it is the only copy and removes it when a |
| 2466 |
// transient already holds the live document. |
| 2467 |
$this->get_sitemap_document($filename); |
| 2468 |
} |
| 2469 |
} |
| 2470 |
|
| 2471 |
/** |
| 2472 |
* The sitemap filenames MetaSync itself serves. |
| 2473 |
* |
| 2474 |
* Used by the absorption sweep so it can never disturb a file belonging to |
| 2475 |
* another sitemap provider. News and video are fixed; the general sitemap's |
| 2476 |
* chunk names come from our own tracking option and the virtual index, |
| 2477 |
* which together describe every document this plugin has produced. |
| 2478 |
* |
| 2479 |
* @return string[] Unique basenames, no paths. |
| 2480 |
*/ |
| 2481 |
private function get_owned_sitemap_filenames() |
| 2482 |
{ |
| 2483 |
$names = array('news-sitemap.xml', 'video-sitemap.xml', 'sitemap.xml', 'sitemap_index.xml'); |
| 2484 |
|
| 2485 |
$tracked = get_option('metasync_sitemap_files', array()); |
| 2486 |
if (is_array($tracked)) { |
| 2487 |
foreach ($tracked as $file) { |
| 2488 |
if (is_array($file) && !empty($file['filename']) && is_string($file['filename'])) { |
| 2489 |
$names[] = $file['filename']; |
| 2490 |
} |
| 2491 |
} |
| 2492 |
} |
| 2493 |
|
| 2494 |
$virtual_index = get_option('metasync_sitemap_virtual_index', array()); |
| 2495 |
if (is_array($virtual_index)) { |
| 2496 |
foreach (array_keys($virtual_index) as $filename) { |
| 2497 |
if (is_string($filename) && '' !== $filename) { |
| 2498 |
$names[] = $filename; |
| 2499 |
} |
| 2500 |
} |
| 2501 |
} |
| 2502 |
|
| 2503 |
// basename() guards against a tracking option that somehow carries a |
| 2504 |
// path; absorb_physical_sitemap() rejects those anyway. |
| 2505 |
return array_unique(array_map('basename', $names)); |
| 2506 |
} |
| 2507 |
|
| 2508 |
/** |
| 2509 |
* Check if virtual mode is active |
| 2510 |
* |
| 2511 |
* @return bool True if virtual mode is active |
| 2512 |
*/ |
| 2513 |
public function is_virtual_mode() |
| 2514 |
{ |
| 2515 |
return get_option('metasync_sitemap_virtual_mode', false) === true; |
| 2516 |
} |
| 2517 |
|
| 2518 |
/** |
| 2519 |
* Serve sitemap files via template_redirect. |
| 2520 |
* |
| 2521 |
* Resolves the requested sitemap filename in two ways: first via the |
| 2522 |
* `metasync_sitemap` query var populated by our rewrite rules (the primary |
| 2523 |
* path), and second via REQUEST_URI parsing (kept as a fallback for sites |
| 2524 |
* with rewrites disabled). Content is read through get_sitemap_document(), |
| 2525 |
* the shared accessor the admin screen and MCP tools also use, so every |
| 2526 |
* consumer reports the same document. It streams from PHP rather than |
| 2527 |
* leaving static files to the web server, which is unsafe because some |
| 2528 |
* nginx configurations 403 static .xml files. |
| 2529 |
*/ |
| 2530 |
public function serve_virtual_sitemap() |
| 2531 |
{ |
| 2532 |
if (is_admin()) { |
| 2533 |
return; |
| 2534 |
} |
| 2535 |
|
| 2536 |
$filename = ''; |
| 2537 |
|
| 2538 |
// Whether our own rewrite rule claimed this URL. Only those requests may |
| 2539 |
// be hard-404'd when no sitemap exists — see the miss handling below. |
| 2540 |
$matched_rewrite = false; |
| 2541 |
|
| 2542 |
// Primary path: query var set by our rewrite rule. |
| 2543 |
$qv = get_query_var('metasync_sitemap'); |
| 2544 |
if (!empty($qv)) { |
| 2545 |
$filename = sanitize_file_name($qv); |
| 2546 |
if (!preg_match('/^(news-|video-)?sitemap(_index)?\d*\.xml$/', $filename)) { |
| 2547 |
return; |
| 2548 |
} |
| 2549 |
$matched_rewrite = true; |
| 2550 |
} else { |
| 2551 |
// Fallback path: REQUEST_URI regex for sites where rewrite rules are |
| 2552 |
// not engaged (e.g. pretty permalinks disabled or pre-flush state). |
| 2553 |
$request_uri = isset($_SERVER['REQUEST_URI']) ? esc_url_raw(wp_unslash($_SERVER['REQUEST_URI'])) : ''; |
| 2554 |
$request_uri = strtok($request_uri, '?'); |
| 2555 |
|
| 2556 |
$sitemap_pattern = '/\/((news-|video-)?sitemap(_index)?\d*\.xml)$/'; |
| 2557 |
if (!preg_match($sitemap_pattern, $request_uri, $matches)) { |
| 2558 |
return; |
| 2559 |
} |
| 2560 |
|
| 2561 |
$filename = $matches[1]; |
| 2562 |
} |
| 2563 |
|
| 2564 |
if (empty($filename)) { |
| 2565 |
return; |
| 2566 |
} |
| 2567 |
|
| 2568 |
// Read through the shared accessor so the served document and the |
| 2569 |
// admin status tiles can never derive from different sources. It also |
| 2570 |
// absorbs a stray physical file at ABSPATH — we cannot rely on the web |
| 2571 |
// server to serve one anyway: on some nginx configurations direct .xml |
| 2572 |
// requests are 403'd before PHP is involved. |
| 2573 |
$virtual_content = $this->get_sitemap_document($filename); |
| 2574 |
|
| 2575 |
// Regenerate-on-miss: if the transient expired, rebuild on the spot |
| 2576 |
// instead of returning 404. Only regenerate for known sitemaps to |
| 2577 |
// prevent DoS via requests to non-existent sitemap numbers. |
| 2578 |
// |
| 2579 |
// The rebuild walks the whole content base, so a crawler fetching the |
| 2580 |
// index and its children in parallel right after the transients |
| 2581 |
// expire used to trigger one full regeneration per concurrent |
| 2582 |
// request. A cross-request lock lets a single request rebuild while |
| 2583 |
// the others briefly poll for its result instead of racing it. |
| 2584 |
if (false === $virtual_content) { |
| 2585 |
$lock_name = 'metasync_vsm_regenlock_' . md5($filename); |
| 2586 |
if ($this->acquire_regen_lock($lock_name)) { |
| 2587 |
// Double-check inside the lock: another request may have |
| 2588 |
// rebuilt this file between our miss and the acquire. |
| 2589 |
$virtual_content = $this->get_virtual_sitemap_file($filename); |
| 2590 |
if (false === $virtual_content) { |
| 2591 |
if ($filename === 'news-sitemap.xml') { |
| 2592 |
$this->generate_news_sitemap(); |
| 2593 |
} elseif ($filename === 'video-sitemap.xml') { |
| 2594 |
$this->generate_video_sitemap(); |
| 2595 |
} elseif ($this->is_sitemap_enabled()) { |
| 2596 |
// only rebuild the general sitemap when the feature is |
| 2597 |
// actually in use. Never fabricate a sitemap the site has not |
| 2598 |
// opted into — the previous `empty($known_files)` fallback |
| 2599 |
// regenerated our sitemap on ANY request, which silently undid |
| 2600 |
// "Delete All Sitemaps" and shadowed a third-party sitemap. |
| 2601 |
$known_files = get_option('metasync_sitemap_files', []); |
| 2602 |
$known_names = array_column($known_files, 'filename'); |
| 2603 |
$known_names[] = 'sitemap_index.xml'; |
| 2604 |
if (in_array($filename, $known_names, true)) { |
| 2605 |
$this->generate_sitemap(); |
| 2606 |
} |
| 2607 |
} |
| 2608 |
$virtual_content = $this->get_virtual_sitemap_file($filename); |
| 2609 |
} |
| 2610 |
delete_option($lock_name); |
| 2611 |
} else { |
| 2612 |
// Another request is rebuilding this sitemap. Poll briefly |
| 2613 |
// for its result instead of starting a duplicate full |
| 2614 |
// rebuild; on timeout fall through to the normal miss |
| 2615 |
// handling below. |
| 2616 |
$deadline = microtime(true) + 5; |
| 2617 |
do { |
| 2618 |
usleep(250000); |
| 2619 |
$virtual_content = $this->get_virtual_sitemap_file($filename); |
| 2620 |
} while (false === $virtual_content && microtime(true) < $deadline); |
| 2621 |
} |
| 2622 |
} |
| 2623 |
|
| 2624 |
if (false === $virtual_content) { |
| 2625 |
// Our rewrite rule claimed this URL but the site has no sitemap for |
| 2626 |
// it — e.g. /sitemap7.xml on a site with a single sitemap. Returning |
| 2627 |
// here hands the request back to the template loader, which renders |
| 2628 |
// the home page under a 200 because `metasync_sitemap` is not a |
| 2629 |
// content query var. That turns `^sitemap(\d*)\.xml$` into an |
| 2630 |
// unlimited supply of soft-404 duplicate-content URLs. Send a real |
| 2631 |
// 404 with no body instead. |
| 2632 |
// |
| 2633 |
// Only the rewrite-matched path is 404'd. The REQUEST_URI fallback |
| 2634 |
// above is unanchored and also matches nested paths such as |
| 2635 |
// /blog/sitemap.xml, plus sitemaps owned by another provider on |
| 2636 |
// installs where our rules are not engaged; those must keep falling |
| 2637 |
// through to whoever can serve them. |
| 2638 |
if ($matched_rewrite) { |
| 2639 |
metasync_discard_buffered_output(); |
| 2640 |
status_header(404); |
| 2641 |
nocache_headers(); |
| 2642 |
header('X-Robots-Tag: noindex'); |
| 2643 |
exit; |
| 2644 |
} |
| 2645 |
|
| 2646 |
return; |
| 2647 |
} |
| 2648 |
|
| 2649 |
metasync_discard_buffered_output(); |
| 2650 |
header('Content-Type: application/xml; charset=utf-8'); |
| 2651 |
header('X-Robots-Tag: noindex'); |
| 2652 |
status_header(200); |
| 2653 |
// phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- XML sitemap document served as application/xml; entries are escaped upstream by the sitemap builder |
| 2654 |
echo $virtual_content; |
| 2655 |
exit; |
| 2656 |
} |
| 2657 |
|
| 2658 |
/** |
| 2659 |
* Atomically acquire the regenerate-on-miss lock for a sitemap file. |
| 2660 |
* |
| 2661 |
* A plain set_transient() cannot arbitrate a race: whether it adds or |
| 2662 |
* updates is not atomic, so simultaneous requests can all believe they |
| 2663 |
* won. add_option() performs a unique-key INSERT, which the database |
| 2664 |
* serializes, so exactly one request acquires the lock even on installs |
| 2665 |
* without a persistent object cache. |
| 2666 |
* |
| 2667 |
* The option value is the acquisition timestamp. A rebuild that crashes |
| 2668 |
* without releasing would otherwise pin the sitemap forever, so an |
| 2669 |
* attempt against a lock older than the TTL clears it and retries. |
| 2670 |
* |
| 2671 |
* @param string $lock_name Full option name for the lock. |
| 2672 |
* @return bool True when this request acquired the lock. |
| 2673 |
*/ |
| 2674 |
private function acquire_regen_lock($lock_name) |
| 2675 |
{ |
| 2676 |
$existing = get_option($lock_name); |
| 2677 |
if (false !== $existing) { |
| 2678 |
if ((int) $existing >= time() - 300) { |
| 2679 |
return false; |
| 2680 |
} |
| 2681 |
// Expired lock from a crashed rebuild: clear it and try to win |
| 2682 |
// the INSERT below. |
| 2683 |
delete_option($lock_name); |
| 2684 |
} |
| 2685 |
return (bool) add_option($lock_name, time(), '', false); |
| 2686 |
} |
| 2687 |
} |
| 2688 |
|
