| 1 |
<?php |
| 2 |
/** |
| 3 |
* Sitemap Generator Class |
| 4 |
* |
| 5 |
* XML sitemap generation and management for search engine optimization. |
| 6 |
* Implements 2025 SEO best practices with real sitemap specifications, |
| 7 |
* dynamic content inclusion, and performance optimization. |
| 8 |
* |
| 9 |
* @package ThinkRank |
| 10 |
* @subpackage SEO |
| 11 |
* @since 1.0.0 |
| 12 |
*/ |
| 13 |
|
| 14 |
declare(strict_types=1); |
| 15 |
|
| 16 |
namespace ThinkRank\SEO; |
| 17 |
|
| 18 |
use InvalidArgumentException; |
| 19 |
|
| 20 |
// Prevent direct access. |
| 21 |
if ( ! defined( 'ABSPATH' ) ) { |
| 22 |
exit; |
| 23 |
} |
| 24 |
|
| 25 |
// Filename derivation and web-root removal are shared with the deactivator and |
| 26 |
// with uninstall.php, which runs without an autoloader — so they live in a |
| 27 |
// plain function file both can require. See includes/cleanup-webroot.php. |
| 28 |
require_once __DIR__ . '/../cleanup-webroot.php'; |
| 29 |
|
| 30 |
/** |
| 31 |
* Sitemap Generator Class |
| 32 |
* |
| 33 |
* Generates and manages XML sitemaps for search engine optimization. |
| 34 |
* Provides dynamic sitemap generation with content filtering, priority |
| 35 |
* calculation, and change frequency optimization. |
| 36 |
* |
| 37 |
* @since 1.0.0 |
| 38 |
*/ |
| 39 |
class Sitemap_Generator extends Abstract_SEO_Manager { |
| 40 |
|
| 41 |
/** |
| 42 |
* Memoised WooCommerce page IDs kept out of the sitemap. Null until resolved. |
| 43 |
* |
| 44 |
* @since 2.0.1 |
| 45 |
* @var int[]|null |
| 46 |
*/ |
| 47 |
private ?array $woocommerce_excluded_page_ids = null; |
| 48 |
|
| 49 |
/** |
| 50 |
* Inclusion flag -> the child sitemap it controls. |
| 51 |
* |
| 52 |
* Shared by the child-list builder and by the "did the caller change what |
| 53 |
* the sitemap includes?" check in maybe_promote_to_index(). |
| 54 |
* |
| 55 |
* @since 1.31.0 |
| 56 |
* @var array<string, string> |
| 57 |
*/ |
| 58 |
private const INCLUSION_CHILD_TYPES = [ |
| 59 |
'include_posts' => 'posts', |
| 60 |
'include_pages' => 'pages', |
| 61 |
'include_categories' => 'categories', |
| 62 |
'include_tags' => 'tags', |
| 63 |
]; |
| 64 |
|
| 65 |
/** |
| 66 |
* Supported sitemap types |
| 67 |
* |
| 68 |
* @since 1.0.0 |
| 69 |
* @var array |
| 70 |
*/ |
| 71 |
/** |
| 72 |
* How many post IDs to hydrate at a time while walking the sitemap set. |
| 73 |
* |
| 74 |
* @since 2.0.1 |
| 75 |
* @var int |
| 76 |
*/ |
| 77 |
private const ID_WALK_CHUNK = 500; |
| 78 |
|
| 79 |
/** |
| 80 |
* How many term IDs to hydrate at a time while walking a taxonomy. |
| 81 |
* |
| 82 |
* @since 2.0.1 |
| 83 |
* @var int |
| 84 |
*/ |
| 85 |
private const TERM_WALK_CHUNK = 1000; |
| 86 |
|
| 87 |
private array $sitemap_types = [ |
| 88 |
'posts' => [ |
| 89 |
'name' => 'Posts', |
| 90 |
'post_types' => ['post'], |
| 91 |
'priority' => 0.8, |
| 92 |
'changefreq' => 'weekly' |
| 93 |
], |
| 94 |
'pages' => [ |
| 95 |
'name' => 'Pages', |
| 96 |
'post_types' => ['page'], |
| 97 |
'priority' => 0.9, |
| 98 |
'changefreq' => 'monthly' |
| 99 |
], |
| 100 |
'categories' => [ |
| 101 |
'name' => 'Categories', |
| 102 |
'taxonomy' => 'category', |
| 103 |
'priority' => 0.6, |
| 104 |
'changefreq' => 'weekly' |
| 105 |
], |
| 106 |
'tags' => [ |
| 107 |
'name' => 'Tags', |
| 108 |
'taxonomy' => 'post_tag', |
| 109 |
'priority' => 0.4, |
| 110 |
'changefreq' => 'monthly' |
| 111 |
] |
| 112 |
]; |
| 113 |
|
| 114 |
/** |
| 115 |
* Constructor |
| 116 |
* |
| 117 |
* @since 1.0.0 |
| 118 |
* |
| 119 |
* @param bool $register_hooks Optional. Whether to register the auto-generation |
| 120 |
* hooks. Pass false for a read-only instance built |
| 121 |
* solely to query settings — the hooks are bound to |
| 122 |
* `$this`, so a second hook-registering instance |
| 123 |
* would run `handle_content_change()` twice per save. |
| 124 |
*/ |
| 125 |
public function __construct(bool $register_hooks = true) { |
| 126 |
parent::__construct('sitemap'); |
| 127 |
|
| 128 |
// Initialize auto-generation hooks |
| 129 |
if ($register_hooks) { |
| 130 |
$this->init_auto_generation_hooks(); |
| 131 |
} |
| 132 |
} |
| 133 |
|
| 134 |
/** |
| 135 |
* Filter the args of a sitemap post query. |
| 136 |
* |
| 137 |
* Exists so integrations can widen what the sitemap sees — the multilingual |
| 138 |
* manager uses it to include every language, since these queries otherwise |
| 139 |
* run in whichever language happened to be active at generation time. |
| 140 |
* |
| 141 |
* @since 1.23.0 |
| 142 |
* |
| 143 |
* @param array $args get_posts() arguments. |
| 144 |
* @return array Filtered arguments. |
| 145 |
*/ |
| 146 |
private function filter_query_args(array $args): array { |
| 147 |
/** |
| 148 |
* Filter the arguments of a sitemap post query. |
| 149 |
* |
| 150 |
* @since 1.23.0 |
| 151 |
* |
| 152 |
* @param array $args get_posts() arguments. |
| 153 |
*/ |
| 154 |
return (array) apply_filters('thinkrank_sitemap_query_args', $args); |
| 155 |
} |
| 156 |
|
| 157 |
/** |
| 158 |
* Filter the args of a sitemap term query. |
| 159 |
* |
| 160 |
* @since 1.23.0 |
| 161 |
* |
| 162 |
* @param array $args get_terms() arguments. |
| 163 |
* @return array Filtered arguments. |
| 164 |
*/ |
| 165 |
private function filter_term_query_args(array $args): array { |
| 166 |
/** |
| 167 |
* Filter the arguments of a sitemap term query. |
| 168 |
* |
| 169 |
* @since 1.23.0 |
| 170 |
* |
| 171 |
* @param array $args get_terms() arguments. |
| 172 |
*/ |
| 173 |
return (array) apply_filters('thinkrank_sitemap_term_query_args', $args); |
| 174 |
} |
| 175 |
|
| 176 |
/** |
| 177 |
* Initialize WordPress hooks for auto-generation |
| 178 |
* |
| 179 |
* @since 1.0.0 |
| 180 |
* @return void |
| 181 |
*/ |
| 182 |
private function init_auto_generation_hooks(): void { |
| 183 |
// Content change hooks - use priority 20 to run after other plugins |
| 184 |
add_action('save_post', [$this, 'handle_content_change'], 20, 2); |
| 185 |
add_action('delete_post', [$this, 'handle_content_deletion'], 20); |
| 186 |
add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20); |
| 187 |
add_action('untrash_post', [$this, 'handle_content_change_by_id'], 20); |
| 188 |
|
| 189 |
// Taxonomy change hooks |
| 190 |
add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 191 |
add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 192 |
add_action('delete_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 193 |
|
| 194 |
// NOTE: the WP-Cron regeneration listeners (thinkrank_regenerate_sitemap |
| 195 |
// and thinkrank_regenerate_sitemap_settings) are registered at plugin |
| 196 |
// bootstrap (Plugin::register_sitemap_cron_listeners(), on plugins_loaded) |
| 197 |
// rather than here. A cron run never builds this class via the REST |
| 198 |
// endpoint (no rest_api_init), so registering them in the constructor |
| 199 |
// would leave the scheduled events with no listener at cron time. |
| 200 |
} |
| 201 |
|
| 202 |
/** |
| 203 |
* Generate XML sitemap |
| 204 |
* |
| 205 |
* @since 1.0.0 |
| 206 |
* |
| 207 |
* @param array $options Sitemap generation options |
| 208 |
* @return string XML sitemap content |
| 209 |
*/ |
| 210 |
public function generate_sitemap(array $options = []): string { |
| 211 |
$settings = $this->get_settings('site'); |
| 212 |
|
| 213 |
$xml = $this->xml_prolog($settings, 'sitemap.xsl'); |
| 214 |
|
| 215 |
// Add image namespace if images are enabled |
| 216 |
if (!empty($settings['include_images'])) { |
| 217 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n"; |
| 218 |
} else { |
| 219 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 220 |
} |
| 221 |
|
| 222 |
// Add homepage. Use the static front page's real modified time when set, |
| 223 |
// so lastmod reflects real changes rather than the generation time. |
| 224 |
$xml .= $this->generate_url_entry(home_url('/'), $this->get_homepage_lastmod(), 1.0, 'daily'); |
| 225 |
|
| 226 |
// MEMORY-EFFICIENT: Generate posts using chunked processing. |
| 227 |
// Note: the single "general" sitemap includes every URL (no per-file cap); |
| 228 |
// per-file chunking applies to the segmented/index model where each type |
| 229 |
// gets its own paginated file (see generate_multiple_sitemaps). |
| 230 |
$enabled_post_types = $this->get_enabled_post_types($settings); |
| 231 |
if (!empty($enabled_post_types)) { |
| 232 |
$xml .= implode('', $this->collect_post_entries($enabled_post_types, $settings)); |
| 233 |
} |
| 234 |
|
| 235 |
// OPTIMIZED: Get all enabled taxonomies and fetch in single query |
| 236 |
$enabled_taxonomies = $this->get_enabled_taxonomies($settings); |
| 237 |
if (!empty($enabled_taxonomies)) { |
| 238 |
$all_taxonomy_data = $this->fetch_taxonomies_optimized($enabled_taxonomies, $settings); |
| 239 |
|
| 240 |
// Process each taxonomy's data |
| 241 |
foreach ($enabled_taxonomies as $taxonomy) { |
| 242 |
if (!empty($all_taxonomy_data[$taxonomy])) { |
| 243 |
$xml .= $this->process_taxonomies_to_xml($all_taxonomy_data[$taxonomy], $taxonomy, $settings); |
| 244 |
} |
| 245 |
} |
| 246 |
} |
| 247 |
|
| 248 |
$xml .= '</urlset>'; |
| 249 |
|
| 250 |
return $xml; |
| 251 |
} |
| 252 |
|
| 253 |
/** |
| 254 |
* Validate SEO settings (implements interface) |
| 255 |
* |
| 256 |
* @since 1.0.0 |
| 257 |
* |
| 258 |
* @param array $settings Settings array to validate |
| 259 |
* @return array Validation results |
| 260 |
*/ |
| 261 |
public function validate_settings(array $settings): array { |
| 262 |
$validation = [ |
| 263 |
'valid' => true, |
| 264 |
'errors' => [], |
| 265 |
'warnings' => [], |
| 266 |
'suggestions' => [] |
| 267 |
]; |
| 268 |
|
| 269 |
try { |
| 270 |
// Validate exclude_posts |
| 271 |
if (!empty($settings['exclude_posts'])) { |
| 272 |
$this->validate_exclude_posts($settings['exclude_posts']); |
| 273 |
} |
| 274 |
|
| 275 |
// Validate exclude_terms |
| 276 |
if (!empty($settings['exclude_terms'])) { |
| 277 |
$this->validate_exclude_terms($settings['exclude_terms']); |
| 278 |
} |
| 279 |
|
| 280 |
// Validate custom_url_pattern |
| 281 |
if (!empty($settings['custom_url_pattern'])) { |
| 282 |
$this->validate_custom_url_pattern($settings['custom_url_pattern']); |
| 283 |
} |
| 284 |
|
| 285 |
// Validate sitemap_urls |
| 286 |
if (!empty($settings['sitemap_urls']) && is_array($settings['sitemap_urls'])) { |
| 287 |
foreach ($settings['sitemap_urls'] as $sitemap) { |
| 288 |
if (!empty($sitemap['url'])) { |
| 289 |
$this->validate_sitemap_url($sitemap['url']); |
| 290 |
} |
| 291 |
} |
| 292 |
} |
| 293 |
|
| 294 |
// Validate numeric settings |
| 295 |
if (isset($settings['links_per_sitemap'])) { |
| 296 |
$this->validate_links_per_sitemap($settings['links_per_sitemap']); |
| 297 |
} |
| 298 |
|
| 299 |
} catch (InvalidArgumentException $e) { |
| 300 |
$validation['valid'] = false; |
| 301 |
$validation['errors'][] = $e->getMessage(); |
| 302 |
} |
| 303 |
|
| 304 |
return $validation; |
| 305 |
} |
| 306 |
|
| 307 |
/** |
| 308 |
* Get output data for frontend rendering (implements interface) |
| 309 |
* |
| 310 |
* @since 1.0.0 |
| 311 |
* |
| 312 |
* @param string $context_type The context type |
| 313 |
* @param int|null $context_id Optional. Context ID |
| 314 |
* @return array Output data ready for frontend rendering |
| 315 |
*/ |
| 316 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 317 |
$settings = $this->get_settings($context_type, $context_id); |
| 318 |
|
| 319 |
return [ |
| 320 |
'sitemap_url' => home_url('/sitemap.xml'), |
| 321 |
'enabled' => $settings['enabled'] ?? true, |
| 322 |
'last_generated' => $settings['last_generated'] ?? '', |
| 323 |
'total_urls' => $this->count_sitemap_urls($settings) |
| 324 |
]; |
| 325 |
} |
| 326 |
|
| 327 |
/** |
| 328 |
* Sitemap keys outside the defaults. |
| 329 |
* |
| 330 |
* @since 2.0.1 |
| 331 |
* |
| 332 |
* @return string[] |
| 333 |
*/ |
| 334 |
protected function additional_setting_keys(): array { |
| 335 |
return ['selected_preset']; |
| 336 |
} |
| 337 |
|
| 338 |
/** |
| 339 |
* Inclusion flags are per post type and per taxonomy. |
| 340 |
* |
| 341 |
* A site registering a `product` post type stores `include_product`; an |
| 342 |
* enumerated list would go stale on the next registration, so the family |
| 343 |
* is matched instead. |
| 344 |
* |
| 345 |
* @since 2.0.1 |
| 346 |
* |
| 347 |
* @return string[] |
| 348 |
*/ |
| 349 |
protected function dynamic_setting_key_patterns(): array { |
| 350 |
return ['/^include_[a-z0-9_]+$/', '/^exclude_[a-z0-9_]+$/']; |
| 351 |
} |
| 352 |
|
| 353 |
/** |
| 354 |
* Get default settings for a context type (implements interface) |
| 355 |
* |
| 356 |
* @since 1.0.0 |
| 357 |
* |
| 358 |
* @param string $context_type The context type to get defaults for |
| 359 |
* @return array Default settings array |
| 360 |
*/ |
| 361 |
public function get_default_settings(string $context_type): array { |
| 362 |
return [ |
| 363 |
// Core settings |
| 364 |
'enabled' => true, |
| 365 |
|
| 366 |
// Multiple Sitemap URLs |
| 367 |
'sitemap_urls' => [ |
| 368 |
[ |
| 369 |
'url' => '/sitemap.xml', |
| 370 |
'type' => 'general', |
| 371 |
'enabled' => true, |
| 372 |
'last_checked' => null, |
| 373 |
'status' => 'unknown' |
| 374 |
] |
| 375 |
], |
| 376 |
'use_sitemap_index' => false, |
| 377 |
|
| 378 |
// General Settings |
| 379 |
'links_per_sitemap' => 1000, |
| 380 |
'include_images' => true, |
| 381 |
'include_featured_images' => false, |
| 382 |
'auto_generate' => true, |
| 383 |
'ping_search_engines' => true, |
| 384 |
|
| 385 |
// Content Inclusion (backward compatibility) |
| 386 |
'include_posts' => true, |
| 387 |
'include_pages' => true, |
| 388 |
'include_categories' => true, |
| 389 |
'include_tags' => false, |
| 390 |
|
| 391 |
// Content Filtering |
| 392 |
'exclude_posts' => '', |
| 393 |
'exclude_terms' => '', |
| 394 |
'exclude_password_protected' => true, |
| 395 |
'exclude_private_posts' => true, |
| 396 |
|
| 397 |
// Advanced Options |
| 398 |
'enable_styling' => true, |
| 399 |
'custom_url_pattern' => 'sitemap-{type}.xml', |
| 400 |
|
| 401 |
// Generation tracking |
| 402 |
'last_generated' => '' |
| 403 |
]; |
| 404 |
} |
| 405 |
|
| 406 |
/** |
| 407 |
* Get settings schema definition (implements interface) |
| 408 |
* |
| 409 |
* @since 1.0.0 |
| 410 |
* |
| 411 |
* @param string $context_type The context type to get schema for |
| 412 |
* @return array Settings schema definition |
| 413 |
*/ |
| 414 |
public function get_settings_schema(string $context_type): array { |
| 415 |
return [ |
| 416 |
'enabled' => [ |
| 417 |
'type' => 'boolean', |
| 418 |
'title' => 'Enable Sitemap', |
| 419 |
'description' => 'Generate XML sitemap for search engines', |
| 420 |
'default' => true |
| 421 |
], |
| 422 |
'include_posts' => [ |
| 423 |
'type' => 'boolean', |
| 424 |
'title' => 'Include Posts', |
| 425 |
'description' => 'Include blog posts in sitemap', |
| 426 |
'default' => true |
| 427 |
], |
| 428 |
'include_pages' => [ |
| 429 |
'type' => 'boolean', |
| 430 |
'title' => 'Include Pages', |
| 431 |
'description' => 'Include static pages in sitemap', |
| 432 |
'default' => true |
| 433 |
], |
| 434 |
'include_categories' => [ |
| 435 |
'type' => 'boolean', |
| 436 |
'title' => 'Include Categories', |
| 437 |
'description' => 'Include category pages in sitemap', |
| 438 |
'default' => true |
| 439 |
], |
| 440 |
'include_tags' => [ |
| 441 |
'type' => 'boolean', |
| 442 |
'title' => 'Include Tags', |
| 443 |
'description' => 'Include tag pages in sitemap', |
| 444 |
'default' => false |
| 445 |
], |
| 446 |
|
| 447 |
'auto_generate' => [ |
| 448 |
'type' => 'boolean', |
| 449 |
'title' => 'Auto Generate', |
| 450 |
'description' => 'Automatically regenerate sitemap when content changes', |
| 451 |
'default' => true |
| 452 |
], |
| 453 |
'ping_search_engines' => [ |
| 454 |
'type' => 'boolean', |
| 455 |
'title' => 'Ping Search Engines', |
| 456 |
'description' => 'Notify Google and Bing when sitemap is updated', |
| 457 |
'default' => true |
| 458 |
], |
| 459 |
'last_generated' => [ |
| 460 |
'type' => 'string', |
| 461 |
'title' => 'Last Generated', |
| 462 |
'description' => 'Timestamp of last sitemap generation', |
| 463 |
'default' => '' |
| 464 |
] |
| 465 |
]; |
| 466 |
} |
| 467 |
|
| 468 |
/** |
| 469 |
* Generate URL entry for sitemap |
| 470 |
* |
| 471 |
* @since 1.0.0 |
| 472 |
* |
| 473 |
* @param string $url URL |
| 474 |
* @param string $lastmod Last modification date |
| 475 |
* @param float $priority Priority (0.0 to 1.0) |
| 476 |
* @param string $changefreq Change frequency |
| 477 |
* @param array $images Optional array of image data |
| 478 |
* @return string XML URL entry |
| 479 |
*/ |
| 480 |
private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string { |
| 481 |
$xml = " <url>\n"; |
| 482 |
$xml .= " <loc>" . esc_url($url) . "</loc>\n"; |
| 483 |
// Omit <lastmod> when unknown (empty) — a fabricated timestamp is worse |
| 484 |
// than no timestamp, and an absent lastmod is valid per the spec. |
| 485 |
if (!empty($lastmod)) { |
| 486 |
$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n"; |
| 487 |
} |
| 488 |
$xml .= " <priority>" . number_format($priority, 1) . "</priority>\n"; |
| 489 |
$xml .= " <changefreq>" . esc_html($changefreq) . "</changefreq>\n"; |
| 490 |
|
| 491 |
// Add image entries if provided |
| 492 |
foreach ($images as $image) { |
| 493 |
$xml .= " <image:image>\n"; |
| 494 |
$xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n"; |
| 495 |
|
| 496 |
if (!empty($image['title'])) { |
| 497 |
$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n"; |
| 498 |
} |
| 499 |
|
| 500 |
if (!empty($image['alt'])) { |
| 501 |
$xml .= " <image:caption>" . esc_html($image['alt']) . "</image:caption>\n"; |
| 502 |
} |
| 503 |
|
| 504 |
$xml .= " </image:image>\n"; |
| 505 |
} |
| 506 |
|
| 507 |
$xml .= " </url>\n"; |
| 508 |
|
| 509 |
return $xml; |
| 510 |
} |
| 511 |
|
| 512 |
/** |
| 513 |
* Collect every post <url> entry for the given post type(s) as an array of |
| 514 |
* XML strings, using memory-efficient chunked fetching. |
| 515 |
* |
| 516 |
* Unlike the old capped generator, this returns ALL matching entries — the |
| 517 |
* links-per-sitemap limit is applied later by paginating this array into |
| 518 |
* separate files (see generate_multiple_sitemaps), matching how Rank Math |
| 519 |
* splits large sitemaps instead of truncating them. |
| 520 |
* |
| 521 |
* @since 1.14.0 |
| 522 |
* |
| 523 |
* @param array $post_types Array of post types to fetch |
| 524 |
* @param array $settings Sitemap settings |
| 525 |
* @return array<string> Individual <url>…</url> entry strings |
| 526 |
*/ |
| 527 |
/** |
| 528 |
* lastmod for the homepage <url> entry: the static front page's real |
| 529 |
* modification time when one is set, otherwise the generation time. |
| 530 |
* |
| 531 |
* @return string ISO-8601 date. |
| 532 |
*/ |
| 533 |
private function get_homepage_lastmod(): string { |
| 534 |
if (get_option('show_on_front') === 'page') { |
| 535 |
$front_id = (int) get_option('page_on_front'); |
| 536 |
if ($front_id) { |
| 537 |
$modified = get_post_field('post_modified_gmt', $front_id); |
| 538 |
if (!empty($modified) && $modified !== '0000-00-00 00:00:00') { |
| 539 |
return gmdate('c', strtotime($modified)); |
| 540 |
} |
| 541 |
} |
| 542 |
} |
| 543 |
return gmdate('c'); |
| 544 |
} |
| 545 |
|
| 546 |
private function collect_post_entries(array $post_types, array $settings): array { |
| 547 |
// Materialize the streaming source for callers that need the full set |
| 548 |
// (e.g. the single un-paginated "general" sitemap). The paginated index |
| 549 |
// path streams collect_post_entries_iter() directly to bound memory. |
| 550 |
return iterator_to_array($this->collect_post_entries_iter($post_types, $settings), false); |
| 551 |
} |
| 552 |
|
| 553 |
/** |
| 554 |
* Stream <url> entries for the given post types, yielding one at a time. |
| 555 |
* |
| 556 |
* Same chunked query, filtering, and ordering as before, but yields each |
| 557 |
* entry instead of accumulating the whole set — so the paginated index |
| 558 |
* generator never holds every URL of a large post type in memory at once. |
| 559 |
* |
| 560 |
* @param array $post_types Post types to include. |
| 561 |
* @param array $settings Sitemap settings. |
| 562 |
* @return \Generator<string> <url> entry strings. |
| 563 |
*/ |
| 564 |
private function collect_post_entries_iter(array $post_types, array $settings): \Generator { |
| 565 |
if (empty($post_types)) { |
| 566 |
return; |
| 567 |
} |
| 568 |
|
| 569 |
// Parse and validate exclude_posts setting |
| 570 |
$exclude_ids = []; |
| 571 |
if (!empty($settings['exclude_posts'])) { |
| 572 |
try { |
| 573 |
$exclude_ids = $this->validate_exclude_posts($settings['exclude_posts']); |
| 574 |
} catch (InvalidArgumentException $e) { |
| 575 |
// Continue with empty array on validation failure - error details available in exception |
| 576 |
} |
| 577 |
} |
| 578 |
|
| 579 |
// The static front page is emitted once as the explicit homepage entry, |
| 580 |
// so exclude it here to avoid a duplicate <loc> (its permalink equals |
| 581 |
// home_url('/')). |
| 582 |
if (get_option('show_on_front') === 'page') { |
| 583 |
$front_id = (int) get_option('page_on_front'); |
| 584 |
if ($front_id) { |
| 585 |
$exclude_ids[] = $front_id; |
| 586 |
} |
| 587 |
} |
| 588 |
|
| 589 |
// Resolve the ordered ID list in one indexed query, then hydrate in |
| 590 |
// chunks via post__in. This avoids large OFFSET windows (which MySQL |
| 591 |
// must scan-and-discard, making a full walk O(n^2)) while still loading |
| 592 |
// only one chunk of full post objects into memory at a time. The |
| 593 |
// original order is preserved (the id query and post__in hydration both |
| 594 |
// use it). |
| 595 |
$all_ids = get_posts($this->filter_query_args([ |
| 596 |
'post_type' => $post_types, |
| 597 |
'post_status' => 'publish', |
| 598 |
'numberposts' => -1, |
| 599 |
'exclude' => $exclude_ids, |
| 600 |
'orderby' => 'post_type post_date', |
| 601 |
'order' => 'ASC DESC', |
| 602 |
'fields' => 'ids', |
| 603 |
])); |
| 604 |
|
| 605 |
if (empty($all_ids)) { |
| 606 |
return; |
| 607 |
} |
| 608 |
|
| 609 |
// Walk the ID list with a moving window rather than array_chunk(). |
| 610 |
// array_chunk() builds a second array holding every element again, so |
| 611 |
// peak memory was twice the ID list — on a 100k-post site that is ~16MB |
| 612 |
// where ~8MB is needed, and this walk is the one part of an otherwise |
| 613 |
// well-bounded routine with no ceiling (#402). |
| 614 |
$total = count($all_ids); |
| 615 |
|
| 616 |
for ($offset = 0; $offset < $total; $offset += self::ID_WALK_CHUNK) { |
| 617 |
$chunk = array_slice($all_ids, $offset, self::ID_WALK_CHUNK); |
| 618 |
|
| 619 |
$posts = get_posts($this->filter_query_args([ |
| 620 |
'post_type' => $post_types, |
| 621 |
'post_status' => 'publish', |
| 622 |
'numberposts' => count($chunk), |
| 623 |
'post__in' => $chunk, |
| 624 |
'orderby' => 'post__in', // preserve the resolved order |
| 625 |
])); |
| 626 |
|
| 627 |
foreach ($posts as $post) { |
| 628 |
if ($this->should_include_in_sitemap($post, $settings)) { |
| 629 |
/** |
| 630 |
* Filter a sitemap entry's permalink. |
| 631 |
* |
| 632 |
* The multilingual manager uses this to generate each |
| 633 |
* translation's URL in its OWN language: the sitemap query |
| 634 |
* deliberately runs with suppress_filters, and the cron |
| 635 |
* rebuild runs with no language context at all, so a bare |
| 636 |
* get_permalink() resolved every translation to the |
| 637 |
* default-language URL — N entries sharing one <loc> (#409). |
| 638 |
* |
| 639 |
* @since 2.0.1 |
| 640 |
* @param string $url Permalink as WordPress resolved it. |
| 641 |
* @param \WP_Post $post Post the entry describes. |
| 642 |
*/ |
| 643 |
$url = apply_filters('thinkrank_sitemap_post_permalink', get_permalink($post), $post); |
| 644 |
$lastmod = gmdate('c', strtotime($post->post_modified_gmt)); |
| 645 |
$priority = $this->calculate_intelligent_priority($post, $post->post_type); |
| 646 |
$changefreq = $this->calculate_change_frequency($post, $post->post_type); |
| 647 |
$images = $this->extract_post_images($post, $settings); |
| 648 |
|
| 649 |
yield $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images); |
| 650 |
} |
| 651 |
} |
| 652 |
|
| 653 |
// Free the hydrated chunk before loading the next one. |
| 654 |
unset($posts); |
| 655 |
} |
| 656 |
} |
| 657 |
|
| 658 |
/** |
| 659 |
* Resolve and bound the configured links-per-sitemap limit. |
| 660 |
* |
| 661 |
* @since 1.14.0 |
| 662 |
* |
| 663 |
* @param array $settings Sitemap settings |
| 664 |
* @return int Links per sitemap file (1–50000) |
| 665 |
*/ |
| 666 |
private function get_links_per_sitemap(array $settings): int { |
| 667 |
$limit = !empty($settings['links_per_sitemap']) ? intval($settings['links_per_sitemap']) : 1000; |
| 668 |
|
| 669 |
return max(1, min(50000, $limit)); |
| 670 |
} |
| 671 |
|
| 672 |
/** |
| 673 |
* Stream <url> entries for a taxonomy's terms, yielding one at a time and |
| 674 |
* fetching terms in bounded chunks (number/offset) — so the paginated index |
| 675 |
* generator never holds every term of a large taxonomy in memory at once. |
| 676 |
* |
| 677 |
* @param string $taxonomy Taxonomy name. |
| 678 |
* @param array $settings Sitemap settings. |
| 679 |
* @return \Generator<string> <url> entry strings. |
| 680 |
*/ |
| 681 |
private function collect_taxonomy_entries_iter(string $taxonomy, array $settings): \Generator { |
| 682 |
$exclude_term_ids = []; |
| 683 |
if (!empty($settings['exclude_terms'])) { |
| 684 |
try { |
| 685 |
$exclude_term_ids = $this->validate_exclude_terms($settings['exclude_terms']); |
| 686 |
} catch (InvalidArgumentException $e) { |
| 687 |
// Continue with an empty exclude list on validation failure. |
| 688 |
} |
| 689 |
} |
| 690 |
|
| 691 |
// product_cat may legitimately have empty terms (products added later); |
| 692 |
// every other taxonomy hides empties — matching fetch_taxonomies_optimized(). |
| 693 |
$hide_empty = $taxonomy !== 'product_cat'; |
| 694 |
$priority = $taxonomy === 'category' ? 0.6 : 0.4; |
| 695 |
|
| 696 |
// Resolve the ordered term IDs in one query, then hydrate in chunks via |
| 697 |
// include. Avoids large OFFSET windows (O(n^2) over a full walk) while |
| 698 |
// holding only one chunk of full term objects at a time. |
| 699 |
$all_ids = get_terms($this->filter_term_query_args([ |
| 700 |
'taxonomy' => $taxonomy, |
| 701 |
'hide_empty' => $hide_empty, |
| 702 |
'exclude' => $exclude_term_ids, |
| 703 |
'orderby' => 'count', |
| 704 |
'order' => 'DESC', |
| 705 |
'fields' => 'ids', |
| 706 |
])); |
| 707 |
|
| 708 |
if (is_wp_error($all_ids) || empty($all_ids)) { |
| 709 |
return; |
| 710 |
} |
| 711 |
|
| 712 |
// Same moving window as the post walk above, for the same reason. |
| 713 |
$total = count($all_ids); |
| 714 |
|
| 715 |
for ($offset = 0; $offset < $total; $offset += self::TERM_WALK_CHUNK) { |
| 716 |
$chunk = array_slice($all_ids, $offset, self::TERM_WALK_CHUNK); |
| 717 |
|
| 718 |
$terms = get_terms($this->filter_term_query_args([ |
| 719 |
'taxonomy' => $taxonomy, |
| 720 |
'include' => $chunk, |
| 721 |
'orderby' => 'include', // preserve the resolved order |
| 722 |
'hide_empty' => false, // already filtered by the id query |
| 723 |
])); |
| 724 |
|
| 725 |
if (is_wp_error($terms) || empty($terms)) { |
| 726 |
continue; |
| 727 |
} |
| 728 |
|
| 729 |
foreach ($terms as $term) { |
| 730 |
// A term the user marked noindex must not be advertised in the |
| 731 |
// sitemap: the robots tag now honours term meta, so listing it |
| 732 |
// here would have the sitemap contradict the page's own tag. |
| 733 |
if ($this->term_is_noindexed((int) $term->term_id)) { |
| 734 |
continue; |
| 735 |
} |
| 736 |
|
| 737 |
$url = get_term_link($term); |
| 738 |
if (!is_wp_error($url)) { |
| 739 |
// Omit lastmod for terms — the generation time is not a real |
| 740 |
// modification time and would mislabel every term as just-changed. |
| 741 |
yield $this->generate_url_entry($url, '', $priority, 'weekly'); |
| 742 |
} |
| 743 |
} |
| 744 |
|
| 745 |
unset($terms); |
| 746 |
} |
| 747 |
} |
| 748 |
|
| 749 |
/** |
| 750 |
* The XML declaration, ownership marker and optional stylesheet every |
| 751 |
* sitemap document opens with. |
| 752 |
* |
| 753 |
* The marker is written unconditionally, and that is the point: removal on |
| 754 |
* deactivate and uninstall deletes a web-root sitemap only when the file |
| 755 |
* says it is ours, and our filenames are the canonical ones another SEO |
| 756 |
* plugin writes too (#515). Tying the proof to `enable_styling` — the one |
| 757 |
* marker older versions left — would mean a site with styling off either |
| 758 |
* kept a shadowing file behind (#510) or had a competitor's deleted. |
| 759 |
* |
| 760 |
* @since 2.1.1 |
| 761 |
* |
| 762 |
* @param array $settings Sitemap settings (read for `enable_styling`). |
| 763 |
* @param string $stylesheet Stylesheet basename in static/xsl/. |
| 764 |
* @return string Prolog lines, newline-terminated. |
| 765 |
*/ |
| 766 |
private function xml_prolog(array $settings, string $stylesheet): string { |
| 767 |
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 768 |
$xml .= THINKRANK_SITEMAP_MARKER . "\n"; |
| 769 |
|
| 770 |
// The stylesheet is presentation only, so it stays opt-in. |
| 771 |
if (!empty($settings['enable_styling'])) { |
| 772 |
$xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/static/xsl/' . $stylesheet) . '"?>' . "\n"; |
| 773 |
} |
| 774 |
|
| 775 |
return $xml; |
| 776 |
} |
| 777 |
|
| 778 |
/** |
| 779 |
* Wrap a set of <url> entry strings in a complete <urlset> document. |
| 780 |
* |
| 781 |
* @since 1.14.0 |
| 782 |
* |
| 783 |
* @param array<string> $entries Entry strings |
| 784 |
* @param array $settings Sitemap settings |
| 785 |
* @param bool $with_image_ns Include the image sitemap namespace |
| 786 |
* @return string Full sitemap XML |
| 787 |
*/ |
| 788 |
private function wrap_urlset(array $entries, array $settings, bool $with_image_ns): string { |
| 789 |
$xml = $this->xml_prolog($settings, 'sitemap.xsl'); |
| 790 |
|
| 791 |
if ($with_image_ns && !empty($settings['include_images'])) { |
| 792 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n"; |
| 793 |
} else { |
| 794 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 795 |
} |
| 796 |
|
| 797 |
$xml .= implode('', $entries); |
| 798 |
$xml .= '</urlset>'; |
| 799 |
|
| 800 |
return $xml; |
| 801 |
} |
| 802 |
|
| 803 |
/** |
| 804 |
* Derive the filename/URL for a given pagination page. |
| 805 |
* |
| 806 |
* Page 1 keeps the base URL (e.g. /sitemap-posts.xml); pages 2+ insert the |
| 807 |
* page number before the extension (e.g. /sitemap-posts-2.xml). |
| 808 |
* |
| 809 |
* @since 1.14.0 |
| 810 |
* |
| 811 |
* @param string $url Base sitemap URL |
| 812 |
* @param int $page 1-based page number |
| 813 |
* @return string Paginated URL |
| 814 |
*/ |
| 815 |
private function paginate_url(string $url, int $page): string { |
| 816 |
if ($page <= 1) { |
| 817 |
return $url; |
| 818 |
} |
| 819 |
|
| 820 |
return (string) preg_replace('/\.xml$/i', '-' . $page . '.xml', $url); |
| 821 |
} |
| 822 |
|
| 823 |
|
| 824 |
|
| 825 |
/** |
| 826 |
* Extract images from post for sitemap |
| 827 |
* |
| 828 |
* @since 1.0.0 |
| 829 |
* |
| 830 |
* @param \WP_Post $post Post object |
| 831 |
* @param array $settings Sitemap settings |
| 832 |
* @return array Array of image data |
| 833 |
*/ |
| 834 |
private function extract_post_images(\WP_Post $post, array $settings): array { |
| 835 |
$images = []; |
| 836 |
|
| 837 |
// Skip if images are disabled |
| 838 |
if (empty($settings['include_images'])) { |
| 839 |
return $images; |
| 840 |
} |
| 841 |
|
| 842 |
// Get featured image if enabled |
| 843 |
if (!empty($settings['include_featured_images'])) { |
| 844 |
$featured_image = $this->get_featured_image($post->ID); |
| 845 |
if ($featured_image) { |
| 846 |
$images[] = $featured_image; |
| 847 |
} |
| 848 |
} |
| 849 |
|
| 850 |
// Extract images from content |
| 851 |
$content_images = $this->extract_content_images($post->post_content); |
| 852 |
$images = array_merge($images, $content_images); |
| 853 |
|
| 854 |
// Remove duplicates based on URL |
| 855 |
$unique_images = []; |
| 856 |
$seen_urls = []; |
| 857 |
foreach ($images as $image) { |
| 858 |
if (!in_array($image['url'], $seen_urls, true)) { |
| 859 |
$unique_images[] = $image; |
| 860 |
$seen_urls[] = $image['url']; |
| 861 |
} |
| 862 |
} |
| 863 |
|
| 864 |
return $unique_images; |
| 865 |
} |
| 866 |
|
| 867 |
/** |
| 868 |
* Get featured image data |
| 869 |
* |
| 870 |
* @since 1.0.0 |
| 871 |
* |
| 872 |
* @param int $post_id Post ID |
| 873 |
* @return array|null Featured image data or null |
| 874 |
*/ |
| 875 |
private function get_featured_image(int $post_id): ?array { |
| 876 |
$thumbnail_id = get_post_thumbnail_id($post_id); |
| 877 |
if (!$thumbnail_id) { |
| 878 |
return null; |
| 879 |
} |
| 880 |
|
| 881 |
$image_url = wp_get_attachment_image_url($thumbnail_id, 'full'); |
| 882 |
if (!$image_url) { |
| 883 |
return null; |
| 884 |
} |
| 885 |
|
| 886 |
$image_title = get_the_title($thumbnail_id); |
| 887 |
$image_alt = get_post_meta($thumbnail_id, '_wp_attachment_image_alt', true); |
| 888 |
|
| 889 |
return [ |
| 890 |
'url' => $image_url, |
| 891 |
'title' => $image_title ?: '', |
| 892 |
'alt' => $image_alt ?: '' |
| 893 |
]; |
| 894 |
} |
| 895 |
|
| 896 |
/** |
| 897 |
* Extract images from post content |
| 898 |
* |
| 899 |
* @since 1.0.0 |
| 900 |
* |
| 901 |
* @param string $content Post content |
| 902 |
* @return array Array of image data |
| 903 |
*/ |
| 904 |
private function extract_content_images(string $content): array { |
| 905 |
$images = []; |
| 906 |
|
| 907 |
// Find all img tags in content |
| 908 |
preg_match_all('/<img[^>]+>/i', $content, $img_tags); |
| 909 |
|
| 910 |
foreach ($img_tags[0] as $img_tag) { |
| 911 |
// Extract src attribute |
| 912 |
if (preg_match('/src=["\']([^"\']+)["\']/', $img_tag, $src_match)) { |
| 913 |
$image_url = $src_match[1]; |
| 914 |
|
| 915 |
// Skip if not a valid URL or external image |
| 916 |
if (!filter_var($image_url, FILTER_VALIDATE_URL)) { |
| 917 |
continue; |
| 918 |
} |
| 919 |
|
| 920 |
// Extract title and alt attributes |
| 921 |
$title = ''; |
| 922 |
$alt = ''; |
| 923 |
|
| 924 |
if (preg_match('/title=["\']([^"\']*)["\']/', $img_tag, $title_match)) { |
| 925 |
$title = $title_match[1]; |
| 926 |
} |
| 927 |
|
| 928 |
if (preg_match('/alt=["\']([^"\']*)["\']/', $img_tag, $alt_match)) { |
| 929 |
$alt = $alt_match[1]; |
| 930 |
} |
| 931 |
|
| 932 |
$images[] = [ |
| 933 |
'url' => $image_url, |
| 934 |
'title' => $title, |
| 935 |
'alt' => $alt |
| 936 |
]; |
| 937 |
} |
| 938 |
} |
| 939 |
|
| 940 |
return $images; |
| 941 |
} |
| 942 |
|
| 943 |
/** |
| 944 |
* Get enabled taxonomies based on settings |
| 945 |
* |
| 946 |
* @since 1.0.0 |
| 947 |
* @param array $settings Sitemap settings |
| 948 |
* @return array Array of enabled taxonomies |
| 949 |
*/ |
| 950 |
private function get_enabled_taxonomies(array $settings): array { |
| 951 |
$taxonomies = []; |
| 952 |
|
| 953 |
// Core taxonomies based on settings |
| 954 |
if (!empty($settings['include_categories'])) { // � |
| 955 |
Evidence-based field name |
| 956 |
$taxonomies[] = 'category'; |
| 957 |
} |
| 958 |
|
| 959 |
if (!empty($settings['include_tags'])) { // � |
| 960 |
Evidence-based field name |
| 961 |
$taxonomies[] = 'post_tag'; |
| 962 |
} |
| 963 |
|
| 964 |
// Auto-detect public custom taxonomies |
| 965 |
$custom_taxonomies = get_taxonomies([ |
| 966 |
'public' => true, |
| 967 |
'_builtin' => false |
| 968 |
], 'names'); |
| 969 |
|
| 970 |
foreach ($custom_taxonomies as $taxonomy) { |
| 971 |
if ($this->should_include_taxonomy($taxonomy)) { |
| 972 |
$taxonomies[] = $taxonomy; |
| 973 |
} |
| 974 |
} |
| 975 |
|
| 976 |
return array_unique($taxonomies); |
| 977 |
} |
| 978 |
|
| 979 |
|
| 980 |
|
| 981 |
/** |
| 982 |
* Fetch taxonomies using optimized combined query |
| 983 |
* |
| 984 |
* @since 1.0.0 |
| 985 |
* |
| 986 |
* @param array $taxonomies Array of taxonomy names to fetch |
| 987 |
* @param array $settings Sitemap settings |
| 988 |
* @return array Grouped terms by taxonomy |
| 989 |
*/ |
| 990 |
private function fetch_taxonomies_optimized(array $taxonomies, array $settings): array { |
| 991 |
if (empty($taxonomies)) { |
| 992 |
return []; |
| 993 |
} |
| 994 |
|
| 995 |
// Parse and validate exclude_terms setting |
| 996 |
$exclude_term_ids = []; |
| 997 |
if (!empty($settings['exclude_terms'])) { |
| 998 |
try { |
| 999 |
$exclude_term_ids = $this->validate_exclude_terms($settings['exclude_terms']); |
| 1000 |
} catch (InvalidArgumentException $e) { |
| 1001 |
// Continue with empty array on validation failure - error details available in exception |
| 1002 |
} |
| 1003 |
} |
| 1004 |
|
| 1005 |
// Determine hide_empty setting based on taxonomies |
| 1006 |
$hide_empty = true; |
| 1007 |
foreach ($taxonomies as $taxonomy) { |
| 1008 |
// For product categories, don't hide empty categories since they might not have products yet |
| 1009 |
if ($taxonomy === 'product_cat') { |
| 1010 |
$hide_empty = false; |
| 1011 |
break; |
| 1012 |
} |
| 1013 |
} |
| 1014 |
|
| 1015 |
// OPTIMIZED: Single query for multiple taxonomies |
| 1016 |
$all_terms = get_terms($this->filter_term_query_args([ |
| 1017 |
'taxonomy' => $taxonomies, // � |
| 1018 |
Multiple taxonomies in single query |
| 1019 |
'hide_empty' => $hide_empty, |
| 1020 |
'exclude' => $exclude_term_ids, |
| 1021 |
'orderby' => 'taxonomy count', |
| 1022 |
'order' => 'ASC DESC' // Order by taxonomy ASC, then count DESC |
| 1023 |
])); |
| 1024 |
|
| 1025 |
if (is_wp_error($all_terms)) { |
| 1026 |
return []; |
| 1027 |
} |
| 1028 |
|
| 1029 |
// Drop terms the user marked noindex. This path feeds the single general |
| 1030 |
// sitemap while collect_taxonomy_entries_iter() feeds the segmented ones, |
| 1031 |
// so both need the filter or the two disagree about the same term. |
| 1032 |
$all_terms = array_values(array_filter( |
| 1033 |
$all_terms, |
| 1034 |
fn($term) => !$this->term_is_noindexed((int) $term->term_id) |
| 1035 |
)); |
| 1036 |
|
| 1037 |
// Group terms by taxonomy |
| 1038 |
return $this->group_terms_by_taxonomy($all_terms); |
| 1039 |
} |
| 1040 |
|
| 1041 |
/** |
| 1042 |
* Group terms by taxonomy |
| 1043 |
* |
| 1044 |
* @since 1.0.0 |
| 1045 |
* |
| 1046 |
* @param array $terms Array of term objects |
| 1047 |
* @return array Grouped terms by taxonomy |
| 1048 |
*/ |
| 1049 |
private function group_terms_by_taxonomy(array $terms): array { |
| 1050 |
$grouped = []; |
| 1051 |
|
| 1052 |
foreach ($terms as $term) { |
| 1053 |
$taxonomy = $term->taxonomy; // � |
| 1054 |
Evidence-based property name |
| 1055 |
|
| 1056 |
if (!isset($grouped[$taxonomy])) { |
| 1057 |
$grouped[$taxonomy] = []; |
| 1058 |
} |
| 1059 |
|
| 1060 |
$grouped[$taxonomy][] = $term; |
| 1061 |
} |
| 1062 |
|
| 1063 |
return $grouped; |
| 1064 |
} |
| 1065 |
|
| 1066 |
/** |
| 1067 |
* Process taxonomy terms to XML entries |
| 1068 |
* |
| 1069 |
* @since 1.0.0 |
| 1070 |
* |
| 1071 |
* @param array $terms Array of term objects |
| 1072 |
* @param string $taxonomy Taxonomy name |
| 1073 |
* @param array $settings Sitemap settings |
| 1074 |
* @return string XML entries |
| 1075 |
*/ |
| 1076 |
private function process_taxonomies_to_xml(array $terms, string $taxonomy, array $settings): string { |
| 1077 |
$xml = ''; |
| 1078 |
|
| 1079 |
foreach ($terms as $term) { |
| 1080 |
$url = get_term_link($term); |
| 1081 |
if (!is_wp_error($url)) { |
| 1082 |
// Omit lastmod for terms (generation time is not a real |
| 1083 |
// modification time). |
| 1084 |
$lastmod = ''; |
| 1085 |
$priority = $taxonomy === 'category' ? 0.6 : 0.4; |
| 1086 |
$changefreq = 'weekly'; |
| 1087 |
|
| 1088 |
$xml .= $this->generate_url_entry($url, $lastmod, $priority, $changefreq); |
| 1089 |
} |
| 1090 |
} |
| 1091 |
|
| 1092 |
return $xml; |
| 1093 |
} |
| 1094 |
|
| 1095 |
/** |
| 1096 |
* Calculate intelligent priority based on content factors |
| 1097 |
* |
| 1098 |
* @since 1.0.0 |
| 1099 |
* |
| 1100 |
* @param \WP_Post $post Post object |
| 1101 |
* @param string $post_type Post type |
| 1102 |
* @return float Priority value between 0.1 and 1.0 |
| 1103 |
*/ |
| 1104 |
private function calculate_intelligent_priority(\WP_Post $post, string $post_type): float { |
| 1105 |
$base_priority = $post_type === 'page' ? 0.9 : 0.8; |
| 1106 |
|
| 1107 |
// Factors that can adjust priority |
| 1108 |
$adjustments = 0; |
| 1109 |
|
| 1110 |
// Recent content gets higher priority |
| 1111 |
$days_old = (time() - strtotime($post->post_date)) / DAY_IN_SECONDS; |
| 1112 |
if ($days_old < 30) { |
| 1113 |
$adjustments += 0.1; // Recent content boost |
| 1114 |
} elseif ($days_old > 365) { |
| 1115 |
$adjustments -= 0.1; // Older content penalty |
| 1116 |
} |
| 1117 |
|
| 1118 |
// Content length factor |
| 1119 |
$content_length = strlen(wp_strip_all_tags($post->post_content)); |
| 1120 |
if ($content_length > 2000) { |
| 1121 |
$adjustments += 0.05; // Comprehensive content boost |
| 1122 |
} elseif ($content_length < 500) { |
| 1123 |
$adjustments -= 0.1; // Thin content penalty |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Special page types get higher priority |
| 1127 |
if ($post_type === 'page') { |
| 1128 |
$page_template = get_page_template_slug($post->ID); |
| 1129 |
if (in_array($page_template, ['page-home.php', 'front-page.php'], true) || |
| 1130 |
(int) $post->ID === (int) get_option('page_on_front')) { |
| 1131 |
$base_priority = 1.0; // Homepage gets maximum priority |
| 1132 |
} elseif (in_array($page_template, ['page-contact.php', 'page-about.php'], true)) { |
| 1133 |
$adjustments += 0.05; // Important pages boost |
| 1134 |
} |
| 1135 |
} |
| 1136 |
|
| 1137 |
// Ensure priority stays within valid range |
| 1138 |
$final_priority = max(0.1, min(1.0, $base_priority + $adjustments)); |
| 1139 |
|
| 1140 |
return round($final_priority, 1); |
| 1141 |
} |
| 1142 |
|
| 1143 |
/** |
| 1144 |
* Calculate change frequency based on content type and age |
| 1145 |
* |
| 1146 |
* @since 1.0.0 |
| 1147 |
* |
| 1148 |
* @param \WP_Post $post Post object |
| 1149 |
* @param string $post_type Post type |
| 1150 |
* @return string Change frequency |
| 1151 |
*/ |
| 1152 |
private function calculate_change_frequency(\WP_Post $post, string $post_type): string { |
| 1153 |
// Pages typically change less frequently |
| 1154 |
if ($post_type === 'page') { |
| 1155 |
$page_template = get_page_template_slug($post->ID); |
| 1156 |
if ((int) $post->ID === (int) get_option('page_on_front')) { |
| 1157 |
return 'daily'; // Homepage changes frequently |
| 1158 |
} elseif (in_array($page_template, ['page-contact.php', 'page-about.php'], true)) { |
| 1159 |
return 'monthly'; // Static pages change monthly |
| 1160 |
} |
| 1161 |
return 'yearly'; // Other pages change rarely |
| 1162 |
} |
| 1163 |
|
| 1164 |
// Posts frequency based on age and type |
| 1165 |
$days_old = (time() - strtotime($post->post_date)) / DAY_IN_SECONDS; |
| 1166 |
|
| 1167 |
if ($days_old < 7) { |
| 1168 |
return 'daily'; // Very recent posts |
| 1169 |
} elseif ($days_old < 30) { |
| 1170 |
return 'weekly'; // Recent posts |
| 1171 |
} elseif ($days_old < 365) { |
| 1172 |
return 'monthly'; // Older posts |
| 1173 |
} |
| 1174 |
|
| 1175 |
return 'yearly'; // Very old posts |
| 1176 |
} |
| 1177 |
|
| 1178 |
/** |
| 1179 |
* Determine if content should be included in sitemap |
| 1180 |
* |
| 1181 |
* @since 1.0.0 |
| 1182 |
* |
| 1183 |
* @param \WP_Post $post Post object |
| 1184 |
* @param array $settings Sitemap settings |
| 1185 |
* @return bool Whether to include in sitemap |
| 1186 |
*/ |
| 1187 |
private function should_include_in_sitemap(\WP_Post $post, array $settings): bool { |
| 1188 |
// The WooCommerce cart, checkout and account pages are transactional, |
| 1189 |
// never indexable, and generate_default_robots_rules() already emits a |
| 1190 |
// Disallow for each of them. Listing them here submitted URLs our own |
| 1191 |
// robots.txt blocks, which Search Console reports as "Submitted URL |
| 1192 |
// blocked by robots.txt". Yoast and Rank Math exclude the same three. |
| 1193 |
if (in_array($post->ID, $this->woocommerce_excluded_page_ids(), true)) { |
| 1194 |
return false; |
| 1195 |
} |
| 1196 |
|
| 1197 |
// Respect user setting for password protected content |
| 1198 |
if (!empty($post->post_password) && !empty($settings['exclude_password_protected'])) { |
| 1199 |
return false; |
| 1200 |
} |
| 1201 |
|
| 1202 |
// Respect user setting for private posts |
| 1203 |
if ($post->post_status === 'private' && !empty($settings['exclude_private_posts'])) { |
| 1204 |
return false; |
| 1205 |
} |
| 1206 |
|
| 1207 |
// Only published (and, per setting, private) content belongs in the |
| 1208 |
// sitemap. Content-quality heuristics (length, "demo"/"test"/"sample" |
| 1209 |
// in the title, "lorem ipsum" text) were intentionally removed: an XML |
| 1210 |
// sitemap should list every indexable published URL. Filtering by |
| 1211 |
// description length silently dropped legitimate WooCommerce products |
| 1212 |
// with short descriptions, and the substring title match excluded real |
| 1213 |
// pages such as "Demo" or "Product Samples". Indexability is governed |
| 1214 |
// by noindex directives below, not by heuristics. |
| 1215 |
if (!in_array($post->post_status, ['publish', 'private'], true)) { |
| 1216 |
return false; |
| 1217 |
} |
| 1218 |
|
| 1219 |
// Check if post overrides robots and sets noindex. |
| 1220 |
if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) { |
| 1221 |
$raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true); |
| 1222 |
if (is_string($raw) && $raw !== '') { |
| 1223 |
$robots = json_decode($raw, true); |
| 1224 |
if (is_array($robots) && !empty($robots['noindex'])) { |
| 1225 |
return false; |
| 1226 |
} |
| 1227 |
} |
| 1228 |
} |
| 1229 |
|
| 1230 |
return true; |
| 1231 |
} |
| 1232 |
|
| 1233 |
/** |
| 1234 |
* WooCommerce pages that must never reach the sitemap. |
| 1235 |
* |
| 1236 |
* Resolved through wc_get_page_id() so a store that moved or renamed its |
| 1237 |
* cart/checkout/account pages is still matched. Returns an empty list when |
| 1238 |
* WooCommerce is not active. Memoised — should_include_in_sitemap() runs |
| 1239 |
* once per post. |
| 1240 |
* |
| 1241 |
* @since 2.0.1 |
| 1242 |
* |
| 1243 |
* @return int[] Page IDs to exclude. |
| 1244 |
*/ |
| 1245 |
private function woocommerce_excluded_page_ids(): array { |
| 1246 |
if ($this->woocommerce_excluded_page_ids !== null) { |
| 1247 |
return $this->woocommerce_excluded_page_ids; |
| 1248 |
} |
| 1249 |
|
| 1250 |
$ids = []; |
| 1251 |
|
| 1252 |
if (function_exists('wc_get_page_id')) { |
| 1253 |
foreach (['cart', 'checkout', 'myaccount'] as $page) { |
| 1254 |
$id = (int) wc_get_page_id($page); |
| 1255 |
// wc_get_page_id() returns -1 when the page is not configured. |
| 1256 |
if ($id > 0) { |
| 1257 |
$ids[] = $id; |
| 1258 |
} |
| 1259 |
} |
| 1260 |
} |
| 1261 |
|
| 1262 |
$this->woocommerce_excluded_page_ids = $ids; |
| 1263 |
|
| 1264 |
return $ids; |
| 1265 |
} |
| 1266 |
|
| 1267 |
/** |
| 1268 |
* Whether a term carries an explicit noindex override. |
| 1269 |
* |
| 1270 |
* Mirrors the post-side check in should_include_post(); terms store the same |
| 1271 |
* `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written |
| 1272 |
* by the update-term-seo ability and by the SEO importer. |
| 1273 |
* |
| 1274 |
* @since 1.31.0 |
| 1275 |
* |
| 1276 |
* @param int $term_id Term to test. |
| 1277 |
* @return bool True when the term is marked noindex. |
| 1278 |
*/ |
| 1279 |
private function term_is_noindexed(int $term_id): bool { |
| 1280 |
if (!(bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true)) { |
| 1281 |
return false; |
| 1282 |
} |
| 1283 |
|
| 1284 |
$raw = get_term_meta($term_id, '_thinkrank_robots_meta', true); |
| 1285 |
if (!is_string($raw) || $raw === '') { |
| 1286 |
return false; |
| 1287 |
} |
| 1288 |
|
| 1289 |
$robots = json_decode($raw, true); |
| 1290 |
|
| 1291 |
return is_array($robots) && !empty($robots['noindex']); |
| 1292 |
} |
| 1293 |
|
| 1294 |
/** |
| 1295 |
* Count total URLs in sitemap |
| 1296 |
* |
| 1297 |
* @since 1.0.0 |
| 1298 |
* |
| 1299 |
* @param array $settings Sitemap settings |
| 1300 |
* @return int Total URL count |
| 1301 |
*/ |
| 1302 |
public function count_sitemap_urls(array $settings): int { |
| 1303 |
$count = 1; // Homepage |
| 1304 |
|
| 1305 |
if (!empty($settings['include_posts'])) { |
| 1306 |
$count += wp_count_posts('post')->publish; |
| 1307 |
} |
| 1308 |
|
| 1309 |
if (!empty($settings['include_pages'])) { |
| 1310 |
$count += wp_count_posts('page')->publish; |
| 1311 |
} |
| 1312 |
|
| 1313 |
if (!empty($settings['include_categories'])) { |
| 1314 |
$count += wp_count_terms(['taxonomy' => 'category', 'hide_empty' => true]); |
| 1315 |
} |
| 1316 |
|
| 1317 |
if (!empty($settings['include_tags'])) { |
| 1318 |
$count += wp_count_terms(['taxonomy' => 'post_tag', 'hide_empty' => true]); |
| 1319 |
} |
| 1320 |
|
| 1321 |
return $count; |
| 1322 |
} |
| 1323 |
|
| 1324 |
/** |
| 1325 |
* Handle content changes for auto-generation |
| 1326 |
* |
| 1327 |
* @since 1.0.0 |
| 1328 |
* @param int $post_id Post ID |
| 1329 |
* @param \WP_Post $post Post object |
| 1330 |
* @return void |
| 1331 |
*/ |
| 1332 |
public function handle_content_change(int $post_id, \WP_Post $post): void { |
| 1333 |
// Skip if auto-generation is disabled |
| 1334 |
if (!$this->should_auto_generate()) { |
| 1335 |
return; |
| 1336 |
} |
| 1337 |
|
| 1338 |
// Skip autosaves and revisions |
| 1339 |
if (wp_is_post_autosave($post_id) || wp_is_post_revision($post_id)) { |
| 1340 |
return; |
| 1341 |
} |
| 1342 |
|
| 1343 |
// Only process published content |
| 1344 |
if ($post->post_status !== 'publish') { |
| 1345 |
return; |
| 1346 |
} |
| 1347 |
|
| 1348 |
// Check if this post type should trigger regeneration |
| 1349 |
if (!$this->should_include_post_type($post->post_type)) { |
| 1350 |
return; |
| 1351 |
} |
| 1352 |
|
| 1353 |
// Schedule debounced regeneration |
| 1354 |
$this->schedule_debounced_regeneration(); |
| 1355 |
} |
| 1356 |
|
| 1357 |
/** |
| 1358 |
* Handle content deletion for auto-generation |
| 1359 |
* |
| 1360 |
* @since 1.0.0 |
| 1361 |
* @param int $post_id Post ID |
| 1362 |
* @return void |
| 1363 |
*/ |
| 1364 |
public function handle_content_deletion(int $post_id): void { |
| 1365 |
// Skip if auto-generation is disabled |
| 1366 |
if (!$this->should_auto_generate()) { |
| 1367 |
return; |
| 1368 |
} |
| 1369 |
|
| 1370 |
$post = get_post($post_id); |
| 1371 |
if (!$post) { |
| 1372 |
return; |
| 1373 |
} |
| 1374 |
|
| 1375 |
// Check if this post type should trigger regeneration |
| 1376 |
if (!$this->should_include_post_type($post->post_type)) { |
| 1377 |
return; |
| 1378 |
} |
| 1379 |
|
| 1380 |
// Schedule debounced regeneration |
| 1381 |
$this->schedule_debounced_regeneration(); |
| 1382 |
} |
| 1383 |
|
| 1384 |
/** |
| 1385 |
* Handle content change by ID (for untrash, etc.) |
| 1386 |
* |
| 1387 |
* @since 1.0.0 |
| 1388 |
* @param int $post_id Post ID |
| 1389 |
* @return void |
| 1390 |
*/ |
| 1391 |
public function handle_content_change_by_id(int $post_id): void { |
| 1392 |
$post = get_post($post_id); |
| 1393 |
if ($post) { |
| 1394 |
$this->handle_content_change($post_id, $post); |
| 1395 |
} |
| 1396 |
} |
| 1397 |
|
| 1398 |
/** |
| 1399 |
* Handle taxonomy changes for auto-generation |
| 1400 |
* |
| 1401 |
* @since 1.0.0 |
| 1402 |
* @param int $term_id Term ID |
| 1403 |
* @param int $tt_id Term taxonomy ID |
| 1404 |
* @param string $taxonomy Taxonomy slug |
| 1405 |
* @return void |
| 1406 |
*/ |
| 1407 |
public function handle_taxonomy_change(int $term_id, int $tt_id, string $taxonomy): void { |
| 1408 |
// Skip if auto-generation is disabled |
| 1409 |
if (!$this->should_auto_generate()) { |
| 1410 |
return; |
| 1411 |
} |
| 1412 |
|
| 1413 |
// Check if this taxonomy should trigger regeneration |
| 1414 |
if (!$this->should_include_taxonomy($taxonomy)) { |
| 1415 |
return; |
| 1416 |
} |
| 1417 |
|
| 1418 |
// Schedule debounced regeneration |
| 1419 |
$this->schedule_debounced_regeneration(); |
| 1420 |
} |
| 1421 |
|
| 1422 |
/** |
| 1423 |
* Check if auto-generation is enabled |
| 1424 |
* |
| 1425 |
* @since 1.0.0 |
| 1426 |
* @return bool True if auto-generation is enabled |
| 1427 |
*/ |
| 1428 |
private function should_auto_generate(): bool { |
| 1429 |
$settings = $this->get_settings('site'); |
| 1430 |
|
| 1431 |
// Check if sitemap is enabled |
| 1432 |
if (empty($settings['enabled'])) { |
| 1433 |
return false; |
| 1434 |
} |
| 1435 |
|
| 1436 |
// Check if auto-generation is enabled |
| 1437 |
return !empty($settings['auto_generate']); |
| 1438 |
} |
| 1439 |
|
| 1440 |
/** |
| 1441 |
* Get enabled post types based on settings |
| 1442 |
* |
| 1443 |
* @since 1.0.0 |
| 1444 |
* @param array $settings Sitemap settings |
| 1445 |
* @return array Array of enabled post types |
| 1446 |
*/ |
| 1447 |
private function get_enabled_post_types(array $settings): array { |
| 1448 |
$post_types = []; |
| 1449 |
|
| 1450 |
// Core post types based on settings |
| 1451 |
if (!empty($settings['include_posts'])) { |
| 1452 |
$post_types[] = 'post'; |
| 1453 |
} |
| 1454 |
|
| 1455 |
if (!empty($settings['include_pages'])) { |
| 1456 |
$post_types[] = 'page'; |
| 1457 |
} |
| 1458 |
|
| 1459 |
// Auto-detect public custom post types that should be included |
| 1460 |
$custom_post_types = get_post_types([ |
| 1461 |
'public' => true, |
| 1462 |
'_builtin' => false |
| 1463 |
], 'names'); |
| 1464 |
|
| 1465 |
foreach ($custom_post_types as $post_type) { |
| 1466 |
if ($this->should_include_post_type($post_type)) { |
| 1467 |
$post_types[] = $post_type; |
| 1468 |
} |
| 1469 |
} |
| 1470 |
|
| 1471 |
return array_unique($post_types); |
| 1472 |
} |
| 1473 |
|
| 1474 |
/** |
| 1475 |
* Check if post type should trigger regeneration |
| 1476 |
* |
| 1477 |
* @since 1.0.0 |
| 1478 |
* @param string $post_type Post type |
| 1479 |
* @return bool True if post type should trigger regeneration |
| 1480 |
*/ |
| 1481 |
private function should_include_post_type(string $post_type): bool { |
| 1482 |
// Match Rank Math: only publicly viewable post types belong in the |
| 1483 |
// sitemap (public && publicly_queryable). Additionally skip post types |
| 1484 |
// that opt out of front-end search (exclude_from_search => true) — e.g. |
| 1485 |
// Templately's internal `templately_library` store — which are template |
| 1486 |
// records, not standalone indexable URLs. BetterDocs `docs` and |
| 1487 |
// WooCommerce `product` register exclude_from_search => false, so they |
| 1488 |
// remain included. |
| 1489 |
if (!is_post_type_viewable($post_type)) { |
| 1490 |
return false; |
| 1491 |
} |
| 1492 |
|
| 1493 |
$post_type_obj = get_post_type_object($post_type); |
| 1494 |
if (!$post_type_obj || !empty($post_type_obj->exclude_from_search)) { |
| 1495 |
return false; |
| 1496 |
} |
| 1497 |
|
| 1498 |
return true; |
| 1499 |
} |
| 1500 |
|
| 1501 |
/** |
| 1502 |
* Check if taxonomy should trigger regeneration |
| 1503 |
* |
| 1504 |
* @since 1.0.0 |
| 1505 |
* @param string $taxonomy Taxonomy slug |
| 1506 |
* @return bool True if taxonomy should trigger regeneration |
| 1507 |
*/ |
| 1508 |
private function should_include_taxonomy(string $taxonomy): bool { |
| 1509 |
// Include all public taxonomies (presets control which sitemaps are created) |
| 1510 |
$taxonomy_obj = get_taxonomy($taxonomy); |
| 1511 |
return $taxonomy_obj && $taxonomy_obj->public; |
| 1512 |
} |
| 1513 |
|
| 1514 |
/** |
| 1515 |
* Schedule debounced sitemap regeneration |
| 1516 |
* |
| 1517 |
* @since 1.0.0 |
| 1518 |
* @return void |
| 1519 |
*/ |
| 1520 |
private function schedule_debounced_regeneration(): void { |
| 1521 |
// Clear any existing scheduled regeneration |
| 1522 |
wp_clear_scheduled_hook('thinkrank_regenerate_sitemap'); |
| 1523 |
|
| 1524 |
// Schedule regeneration in 30 seconds to debounce rapid changes |
| 1525 |
wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap'); |
| 1526 |
} |
| 1527 |
|
| 1528 |
/** |
| 1529 |
* Public entry point to debounce-rebuild the sitemap after a settings change |
| 1530 |
* (e.g. toggling inclusion rules via REST or the MCP ability), so the served |
| 1531 |
* file reflects the new settings instead of going stale until a content edit. |
| 1532 |
* |
| 1533 |
* @return void |
| 1534 |
*/ |
| 1535 |
public function schedule_regeneration(): void { |
| 1536 |
// Debounce against rapid successive saves, but use the settings-specific |
| 1537 |
// hook so the rebuild runs regardless of the auto_generate toggle (which |
| 1538 |
// only governs content-change-triggered regeneration). |
| 1539 |
wp_clear_scheduled_hook('thinkrank_regenerate_sitemap_settings'); |
| 1540 |
wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap_settings'); |
| 1541 |
} |
| 1542 |
|
| 1543 |
/** |
| 1544 |
* Rebuild the served sitemap after an explicit settings change. |
| 1545 |
* |
| 1546 |
* Unlike {@see auto_regenerate_sitemap()}, this is NOT gated on the |
| 1547 |
* auto_generate setting: the user deliberately changed inclusion rules and |
| 1548 |
* expects the served file to reflect them even if content-triggered |
| 1549 |
* auto-generation is turned off. Still respects the master `enabled` flag. |
| 1550 |
* |
| 1551 |
* @return void |
| 1552 |
*/ |
| 1553 |
public function regenerate_sitemap_from_settings(): void { |
| 1554 |
try { |
| 1555 |
$settings = $this->get_settings('site'); |
| 1556 |
if (empty($settings['enabled'])) { |
| 1557 |
// The sitemap was disabled: remove the previously generated static |
| 1558 |
// files so the web server stops serving a stale sitemap that |
| 1559 |
// crawlers would otherwise keep fetching. |
| 1560 |
$this->delete_published_sitemaps(); |
| 1561 |
return; |
| 1562 |
} |
| 1563 |
$this->generate_and_save($settings); |
| 1564 |
} catch (\Throwable $e) { |
| 1565 |
// Settings-triggered regeneration failed - details in exception. |
| 1566 |
} |
| 1567 |
} |
| 1568 |
|
| 1569 |
/** |
| 1570 |
* Remove every static sitemap file ThinkRank publishes to the web root. |
| 1571 |
* |
| 1572 |
* Called when the sitemap feature is disabled, by the cleanup route, and by |
| 1573 |
* both removal paths, so /sitemap.xml, /sitemap_index.xml, the segmented |
| 1574 |
* children (incl. paginated -N pages), and /local-sitemap.xml stop being |
| 1575 |
* served. Only ThinkRank's own filenames are targeted; WordPress core's |
| 1576 |
* wp-sitemap.xml and any other plugin's sitemap in the web root are left |
| 1577 |
* untouched. |
| 1578 |
* |
| 1579 |
* @since 1.31.0 Returns the filenames removed, and accepts the settings to |
| 1580 |
* derive them from, so a caller that already read them (and |
| 1581 |
* needs to report what went) does not have to re-read or |
| 1582 |
* re-derive the name list. |
| 1583 |
* @since 2.1.0 Delegates to thinkrank_webroot_delete_sitemaps(). Uninstall |
| 1584 |
* needs the same removal but has no autoloader to reach this |
| 1585 |
* class, so the logic moved to includes/cleanup-webroot.php |
| 1586 |
* and this stays as the in-plugin entry point. |
| 1587 |
* |
| 1588 |
* @param array|null $settings Optional. Sitemap settings; defaults to the |
| 1589 |
* saved site settings. |
| 1590 |
* @return array{deleted: string[], failed: string[]} Basenames removed, and |
| 1591 |
* those that existed but could not be removed. |
| 1592 |
*/ |
| 1593 |
public function delete_published_sitemaps(?array $settings = null): array { |
| 1594 |
return thinkrank_webroot_delete_sitemaps($settings ?? $this->get_settings('site')); |
| 1595 |
} |
| 1596 |
|
| 1597 |
/** |
| 1598 |
* Every child-sitemap filename this site could have published. |
| 1599 |
* |
| 1600 |
* @since 1.31.0 |
| 1601 |
* @since 2.1.0 Delegates to thinkrank_webroot_segment_filenames(). |
| 1602 |
* |
| 1603 |
* @param array $settings Sitemap settings (read for `custom_url_pattern`). |
| 1604 |
* @return string[] Basenames, e.g. ['sitemap-posts.xml', 'sitemap-pages.xml']. |
| 1605 |
*/ |
| 1606 |
private function publishable_segment_filenames(array $settings): array { |
| 1607 |
return thinkrank_webroot_segment_filenames($settings); |
| 1608 |
} |
| 1609 |
|
| 1610 |
/** |
| 1611 |
* Can this web-root file be shown to be a sitemap ThinkRank wrote? |
| 1612 |
* |
| 1613 |
* The generator deletes as often as cleanup does — a segment that dropped |
| 1614 |
* out of the set, a pagination page beyond the new count, the local sitemap |
| 1615 |
* after the business identity was cleared — and until 2.1.1 it did all |
| 1616 |
* three by filename alone. That is the #515 bug on a far more frequent |
| 1617 |
* trigger: our names are the canonical ones, so an ordinary regeneration |
| 1618 |
* (post save, term change, settings save) destroyed RankMath's |
| 1619 |
* `sitemap-tags.xml` and `local-sitemap.xml` with no deactivation involved. |
| 1620 |
* |
| 1621 |
* Every name the generator derives comes from the current settings — the |
| 1622 |
* url pattern, the configured `sitemap_urls`, `local-sitemap.xml` — but the |
| 1623 |
* ownership test is still asked with `$name_derived = false`, which switches |
| 1624 |
* off the legacy fallback for the whole generator side. |
| 1625 |
* |
| 1626 |
* The fallback exists to recover a pre-2.1.1 file written with |
| 1627 |
* `enable_styling` off, which carries neither marker. That recovery belongs |
| 1628 |
* to the once-off cleanup paths. Here it can only do harm: this method runs |
| 1629 |
* on every post save, and everything this version writes carries |
| 1630 |
* THINKRANK_SITEMAP_MARKER, so after the site's first regeneration an |
| 1631 |
* unmarked file at one of our names is by definition somebody else's — and |
| 1632 |
* deleting it on an ordinary regeneration is #515 through the more common |
| 1633 |
* door. The cost is a stale unmarked segment left on disk until deactivation |
| 1634 |
* picks it up, which is the safe direction to fail in. |
| 1635 |
* |
| 1636 |
* @since 2.1.1 |
| 1637 |
* |
| 1638 |
* @param string $path Absolute path to a file in the web root. |
| 1639 |
* @param array $settings Sitemap settings. |
| 1640 |
* @return bool True when the file may be deleted. |
| 1641 |
*/ |
| 1642 |
private function webroot_sitemap_is_ours(string $path, array $settings): bool { |
| 1643 |
return thinkrank_webroot_sitemap_is_ours($path, $settings, false); |
| 1644 |
} |
| 1645 |
|
| 1646 |
/** |
| 1647 |
* Auto-regenerate sitemap (called by scheduled action) |
| 1648 |
* |
| 1649 |
* @since 1.0.0 |
| 1650 |
* @return void |
| 1651 |
*/ |
| 1652 |
public function auto_regenerate_sitemap(): void { |
| 1653 |
try { |
| 1654 |
// Double-check that auto-generation is still enabled |
| 1655 |
if (!$this->should_auto_generate()) { |
| 1656 |
return; |
| 1657 |
} |
| 1658 |
|
| 1659 |
$this->generate_and_save($this->get_settings('site')); |
| 1660 |
|
| 1661 |
// Sitemap auto-regenerated successfully |
| 1662 |
|
| 1663 |
} catch (\Throwable $e) { |
| 1664 |
// Sitemap auto-regeneration failed - error details available in exception |
| 1665 |
} |
| 1666 |
} |
| 1667 |
|
| 1668 |
/** |
| 1669 |
* Build and write the sitemap files for the given settings. |
| 1670 |
* |
| 1671 |
* Segmented files plus an index when the settings configure one, a single |
| 1672 |
* file otherwise, and the standalone local business sitemap either way. |
| 1673 |
* Records `last_generated` so the UI's "View Generated Sitemaps" links |
| 1674 |
* unlock, mirroring the manual generate endpoint. |
| 1675 |
* |
| 1676 |
* @since 1.17.0 |
| 1677 |
* @param array $settings Sitemap settings. |
| 1678 |
* @return bool True when the sitemap files were written. |
| 1679 |
*/ |
| 1680 |
public function generate_and_save(array $settings): bool { |
| 1681 |
// Index mode is driven by the use_sitemap_index toggle (not merely by how |
| 1682 |
// many sitemap_urls happen to be configured). When the toggle is on but |
| 1683 |
// no child sitemaps are set up yet, synthesize the per-type segmented set |
| 1684 |
// so we emit a real <sitemapindex> with paginated children instead of a |
| 1685 |
// flat urlset misnamed sitemap_index.xml. |
| 1686 |
$settings = $this->maybe_promote_to_index($settings); |
| 1687 |
|
| 1688 |
if (!empty($settings['use_sitemap_index']) || count((is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : [])) > 1) { |
| 1689 |
$results = $this->generate_multiple_sitemaps($settings); |
| 1690 |
$written = !empty($results['success']); |
| 1691 |
} else { |
| 1692 |
$xml = $this->generate_sitemap($settings); |
| 1693 |
$primary = $this->get_primary_sitemap_filename($settings); |
| 1694 |
$written = $this->save_sitemap_to_file($xml, $primary); |
| 1695 |
|
| 1696 |
// Local business sitemap is a standalone file, regenerated on the |
| 1697 |
// single-sitemap path too (this is the default mode). |
| 1698 |
$this->regenerate_local_sitemap($settings); |
| 1699 |
|
| 1700 |
// Switching out of index mode leaves sitemap_index.xml and every |
| 1701 |
// child on disk, still served and never refreshed again. The index |
| 1702 |
// path already prunes what it no longer owns; this path never did, |
| 1703 |
// so the site kept serving two sitemap trees (#563). Ownership is |
| 1704 |
// still tested per file, so another plugin's sitemap at one of our |
| 1705 |
// names is never touched (#515). |
| 1706 |
$this->prune_orphaned_segments($settings, [['filename' => basename($primary)]]); |
| 1707 |
} |
| 1708 |
|
| 1709 |
if ($written) { |
| 1710 |
$settings['last_generated'] = gmdate('c'); |
| 1711 |
$this->save_settings('site', null, $settings); |
| 1712 |
} |
| 1713 |
|
| 1714 |
return $written; |
| 1715 |
} |
| 1716 |
|
| 1717 |
/** |
| 1718 |
* Resolve index-vs-single mode, synthesizing child sitemaps when needed. |
| 1719 |
* |
| 1720 |
* - When use_sitemap_index is on but no child sitemaps are configured, build |
| 1721 |
* the per-type segmented set so a real <sitemapindex> is produced (#127). |
| 1722 |
* - When the toggle is off but a single flat file would exceed the per-file |
| 1723 |
* URL cap, auto-promote to a paginated index instead of one oversized file |
| 1724 |
* that can cross Google's 50k-URL/50MB limits (#129). |
| 1725 |
* |
| 1726 |
* @param array $settings Sitemap settings. |
| 1727 |
* @return array Possibly-updated settings. |
| 1728 |
*/ |
| 1729 |
public function maybe_promote_to_index(array $settings): array { |
| 1730 |
$saved = $this->get_settings('site'); |
| 1731 |
|
| 1732 |
// Read from the *payload*, before the merge below folds the saved values |
| 1733 |
// in: "the caller named this" and "this has a value" are different |
| 1734 |
// questions, and the mode resolution turns on the former. |
| 1735 |
$mode_supplied = array_key_exists('use_sitemap_index', $settings); |
| 1736 |
$urls_supplied = is_array($settings['sitemap_urls'] ?? null); |
| 1737 |
$has_children = count($urls_supplied ? $settings['sitemap_urls'] : []) > 1; |
| 1738 |
|
| 1739 |
// Which inclusion flags did the caller actually name? The child list is |
| 1740 |
// the only thing that reads them, and inheriting a saved one skipped |
| 1741 |
// that — so on an index-mode site the include_* flags were enforced |
| 1742 |
// nowhere but in the browser, where SitemapGeneration.js recomputes |
| 1743 |
// sitemap_urls itself. Every non-UI client, the shipped |
| 1744 |
// `update-sitemap-settings` ability included, saved the flag and changed |
| 1745 |
// nothing (#398). Read from the payload for the same reason as above: |
| 1746 |
// after the merge every saved flag would look like one the caller named. |
| 1747 |
$named_inclusions = array_intersect( |
| 1748 |
array_keys(self::INCLUSION_CHILD_TYPES), |
| 1749 |
array_keys($settings) |
| 1750 |
); |
| 1751 |
$inclusions_supplied = (bool) $named_inclusions; |
| 1752 |
|
| 1753 |
// Inclusion flags may be absent from a partial payload (e.g. the manual |
| 1754 |
// generate endpoint) — fall back to saved settings so synthesized child |
| 1755 |
// sitemaps reflect the real include_posts/pages/categories choices. |
| 1756 |
$inclusions = array_merge($saved, $settings); |
| 1757 |
|
| 1758 |
// Hand the generators a *complete* settings array. Only the mode was |
| 1759 |
// resolved before, so every other unnamed key reached them missing: a |
| 1760 |
// bare `{}` from a REST/MCP client republished the sitemap with |
| 1761 |
// enable_styling and include_images read as off, overwriting the live |
| 1762 |
// files with output that had lost its XSL stylesheet, its image |
| 1763 |
// namespace and its image entries. Presentation and inclusion settings |
| 1764 |
// are not something a generate call opts into — they are the site's |
| 1765 |
// configuration, and only a value actually present in the payload |
| 1766 |
// overrides them. |
| 1767 |
$settings = $inclusions; |
| 1768 |
|
| 1769 |
// The two keys that drive mode keep their own resolution rules below, |
| 1770 |
// so they must go back to "not specified" when the caller omitted them. |
| 1771 |
if (!$mode_supplied) { |
| 1772 |
unset($settings['use_sitemap_index']); |
| 1773 |
} |
| 1774 |
if (!$urls_supplied) { |
| 1775 |
unset($settings['sitemap_urls']); |
| 1776 |
} |
| 1777 |
|
| 1778 |
// An absent use_sitemap_index means "not specified", which is not the |
| 1779 |
// same as "single file". Reading it as the latter meant a partial payload |
| 1780 |
// — `{}` from a REST/MCP client, or anything short of the full settings |
| 1781 |
// object the admin bundle sends — republished one flat sitemap.xml on an |
| 1782 |
// index-mode site and left sitemap_index.xml and its children stale or |
| 1783 |
// missing. Inherit the saved mode instead; only a value actually present |
| 1784 |
// in the payload decides the mode. |
| 1785 |
if (!array_key_exists('use_sitemap_index', $settings)) { |
| 1786 |
$settings['use_sitemap_index'] = $saved['use_sitemap_index'] ?? ''; |
| 1787 |
|
| 1788 |
// Inheriting the mode means inheriting its children too, unless the |
| 1789 |
// caller named its own set. |
| 1790 |
$saved_children = (is_array($saved['sitemap_urls'] ?? null) ? $saved['sitemap_urls'] : []); |
| 1791 |
if (!empty($settings['use_sitemap_index']) |
| 1792 |
&& !$urls_supplied |
| 1793 |
&& count($saved_children) > 1) { |
| 1794 |
// A named inclusion flag is applied *to* the inherited list, not |
| 1795 |
// used to regenerate it. build_segmented_sitemap_urls() also adds |
| 1796 |
// a child for every public custom post type, so rebuilding here |
| 1797 |
// would make `{include_pages: false}` — one thing off — silently |
| 1798 |
// switch on children the saved list never had (an Elementor |
| 1799 |
// internal CPT, a WooCommerce product feed). Only the flags the |
| 1800 |
// caller actually named change anything. |
| 1801 |
$settings['sitemap_urls'] = $inclusions_supplied |
| 1802 |
? $this->apply_inclusion_flags_to_children($saved_children, $named_inclusions, $inclusions) |
| 1803 |
: $saved_children; |
| 1804 |
|
| 1805 |
// The children are resolved either way — including when the |
| 1806 |
// caller switched the last one off, which leaves a bare index and |
| 1807 |
// is what they asked for. |
| 1808 |
$has_children = true; |
| 1809 |
} |
| 1810 |
} |
| 1811 |
|
| 1812 |
if (!empty($settings['use_sitemap_index'])) { |
| 1813 |
if (!$has_children) { |
| 1814 |
$settings['sitemap_urls'] = $this->build_segmented_sitemap_urls($inclusions); |
| 1815 |
} |
| 1816 |
return $settings; |
| 1817 |
} |
| 1818 |
|
| 1819 |
if (!$has_children) { |
| 1820 |
$limit = $this->get_links_per_sitemap($settings); |
| 1821 |
if ($limit > 0 && $this->estimate_total_sitemap_urls($inclusions) > $limit) { |
| 1822 |
$settings['use_sitemap_index'] = true; |
| 1823 |
$settings['sitemap_urls'] = $this->build_segmented_sitemap_urls($inclusions); |
| 1824 |
} |
| 1825 |
} |
| 1826 |
|
| 1827 |
return $settings; |
| 1828 |
} |
| 1829 |
|
| 1830 |
/** |
| 1831 |
* Rough count of URLs a single-file sitemap would contain, used only to |
| 1832 |
* decide whether to auto-promote to a paginated index. Cheap COUNT queries; |
| 1833 |
* intentionally approximate (homepage + published posts of enabled types + |
| 1834 |
* terms of enabled taxonomies). |
| 1835 |
* |
| 1836 |
* @param array $settings Sitemap settings. |
| 1837 |
* @return int |
| 1838 |
*/ |
| 1839 |
private function estimate_total_sitemap_urls(array $settings): int { |
| 1840 |
$total = 1; // homepage |
| 1841 |
|
| 1842 |
foreach ($this->get_enabled_post_types($settings) as $post_type) { |
| 1843 |
$counts = wp_count_posts($post_type); |
| 1844 |
$total += isset($counts->publish) ? (int) $counts->publish : 0; |
| 1845 |
} |
| 1846 |
|
| 1847 |
foreach ($this->get_enabled_taxonomies($settings) as $taxonomy) { |
| 1848 |
$count = wp_count_terms(['taxonomy' => $taxonomy, 'hide_empty' => true]); |
| 1849 |
if (!is_wp_error($count)) { |
| 1850 |
$total += (int) $count; |
| 1851 |
} |
| 1852 |
} |
| 1853 |
|
| 1854 |
return $total; |
| 1855 |
} |
| 1856 |
|
| 1857 |
/** |
| 1858 |
* Resolve the sitemap file the site publishes for the given settings. |
| 1859 |
* |
| 1860 |
* Index mode serves the index (`sitemap_index.xml`); the default single-file |
| 1861 |
* mode serves `sitemap.xml`. Callers use this to link to — or check for — |
| 1862 |
* the file the site actually serves, since ThinkRank's sitemap is a static |
| 1863 |
* file in the web root rather than a route. |
| 1864 |
* |
| 1865 |
* @since 1.17.0 |
| 1866 |
* @param array $settings Sitemap settings. |
| 1867 |
* @return string Sitemap filename. |
| 1868 |
*/ |
| 1869 |
public function get_primary_sitemap_filename(array $settings): string { |
| 1870 |
return thinkrank_webroot_primary_sitemap_filename($settings); |
| 1871 |
} |
| 1872 |
|
| 1873 |
/** |
| 1874 |
* Public URL of the sitemap the site serves. |
| 1875 |
* |
| 1876 |
* @since 1.17.0 |
| 1877 |
* @param array|null $settings Sitemap settings (falls back to saved site settings). |
| 1878 |
* @return string Absolute sitemap URL. |
| 1879 |
*/ |
| 1880 |
public function get_primary_sitemap_url(?array $settings = null): string { |
| 1881 |
$settings = $settings ?? $this->get_settings('site'); |
| 1882 |
|
| 1883 |
return home_url('/' . $this->get_primary_sitemap_filename($settings)); |
| 1884 |
} |
| 1885 |
|
| 1886 |
/** |
| 1887 |
* Whether the sitemap file this site publishes exists on disk. |
| 1888 |
* |
| 1889 |
* @since 1.17.0 |
| 1890 |
* @param array $settings Sitemap settings. |
| 1891 |
* @return bool True when the file is present. |
| 1892 |
*/ |
| 1893 |
public function primary_sitemap_file_exists(array $settings): bool { |
| 1894 |
return file_exists(ABSPATH . $this->get_primary_sitemap_filename($settings)); |
| 1895 |
} |
| 1896 |
|
| 1897 |
/** |
| 1898 |
* Save sitemap XML to file |
| 1899 |
* |
| 1900 |
* @since 1.0.0 |
| 1901 |
* @param string $sitemap_xml Sitemap XML content |
| 1902 |
* @param string $filename Optional. Filename to save (defaults to 'sitemap.xml') |
| 1903 |
* @return bool True on success, false on failure |
| 1904 |
*/ |
| 1905 |
private function save_sitemap_to_file(string $sitemap_xml, string $filename = 'sitemap.xml'): bool { |
| 1906 |
// Validate and sanitize filename for security |
| 1907 |
try { |
| 1908 |
$filename = $this->validate_sitemap_filename($filename); |
| 1909 |
} catch (InvalidArgumentException $e) { |
| 1910 |
// File validation failed - error details available in exception |
| 1911 |
return false; |
| 1912 |
} |
| 1913 |
|
| 1914 |
$sitemap_path = ABSPATH . $filename; |
| 1915 |
|
| 1916 |
// Use WordPress filesystem API for better security |
| 1917 |
global $wp_filesystem; |
| 1918 |
if (!$wp_filesystem) { |
| 1919 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 1920 |
WP_Filesystem(); |
| 1921 |
} |
| 1922 |
|
| 1923 |
if ($wp_filesystem) { |
| 1924 |
$written = $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE); |
| 1925 |
|
| 1926 |
if ($written) { |
| 1927 |
// Every sitemap this version writes carries the ownership |
| 1928 |
// marker, so once one has been written an unmarked file at one |
| 1929 |
// of our names cannot be ours. Recording that retires the |
| 1930 |
// legacy fallback for this install — see |
| 1931 |
// thinkrank_webroot_sitemap_is_ours(). |
| 1932 |
if (get_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION) !== '1') { |
| 1933 |
update_option(THINKRANK_SITEMAP_MARKED_WRITE_OPTION, '1', false); |
| 1934 |
} |
| 1935 |
} |
| 1936 |
|
| 1937 |
return $written; |
| 1938 |
} |
| 1939 |
|
| 1940 |
// WP_Filesystem initialization failed |
| 1941 |
return false; |
| 1942 |
} |
| 1943 |
|
| 1944 |
/** |
| 1945 |
* Build a segmented, index-based sitemap_urls list from inclusion settings. |
| 1946 |
* |
| 1947 |
* Produces the index entry plus one child sitemap per enabled content type |
| 1948 |
* (posts/pages/categories) and one per public custom post type — the shape |
| 1949 |
* the "complete" preset creates and that generate_multiple_sitemaps() expects. |
| 1950 |
* Used when enabling the index during import so the index has real children |
| 1951 |
* instead of being empty. |
| 1952 |
* |
| 1953 |
* @since 1.14.0 |
| 1954 |
* |
| 1955 |
* @param array $inclusions Inclusion flags (include_posts/pages/categories) |
| 1956 |
* and optionally custom_url_pattern. |
| 1957 |
* @return array Sitemap URL configs |
| 1958 |
*/ |
| 1959 |
public function build_segmented_sitemap_urls(array $inclusions): array { |
| 1960 |
$pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml'; |
| 1961 |
|
| 1962 |
$urls = [$this->sitemap_child_entry('/sitemap_index.xml', 'index')]; |
| 1963 |
|
| 1964 |
foreach (self::INCLUSION_CHILD_TYPES as $flag => $type) { |
| 1965 |
if (!empty($inclusions[$flag])) { |
| 1966 |
$urls[] = $this->build_child_sitemap_entry($type, $pattern); |
| 1967 |
} |
| 1968 |
} |
| 1969 |
|
| 1970 |
// Public custom post types each get a child sitemap (parity with the |
| 1971 |
// "complete" preset and with Rank Math, which lists every public CPT). |
| 1972 |
foreach (get_post_types(['public' => true, '_builtin' => false], 'names') as $cpt) { |
| 1973 |
if ($this->should_include_post_type($cpt)) { |
| 1974 |
$urls[] = $this->build_child_sitemap_entry($cpt, $pattern); |
| 1975 |
} |
| 1976 |
} |
| 1977 |
|
| 1978 |
return $urls; |
| 1979 |
} |
| 1980 |
|
| 1981 |
/** |
| 1982 |
* Apply only the inclusion flags the caller named to an existing child list. |
| 1983 |
* |
| 1984 |
* The narrow counterpart to build_segmented_sitemap_urls(): that one |
| 1985 |
* regenerates the whole set from scratch, which is right when there is no set |
| 1986 |
* yet and wrong when there is. Rebuilding an existing list would add a child |
| 1987 |
* for every public custom post type it had never contained, so a payload that |
| 1988 |
* switches one thing off would switch others on. Here a flag adds or removes |
| 1989 |
* exactly its own child and leaves every other entry — custom post types, |
| 1990 |
* hand-added URLs, per-child enabled/status state — untouched (#398). |
| 1991 |
* |
| 1992 |
* @since 1.31.0 |
| 1993 |
* |
| 1994 |
* @param array $children Existing child sitemap entries. |
| 1995 |
* @param string[] $named_inclusions Inclusion flag keys present in the payload. |
| 1996 |
* @param array $inclusions Merged settings, for the resolved flag |
| 1997 |
* values and custom_url_pattern. |
| 1998 |
* @return array Updated child sitemap entries. |
| 1999 |
*/ |
| 2000 |
private function apply_inclusion_flags_to_children( |
| 2001 |
array $children, |
| 2002 |
array $named_inclusions, |
| 2003 |
array $inclusions |
| 2004 |
): array { |
| 2005 |
$pattern = $inclusions['custom_url_pattern'] ?? 'sitemap-{type}.xml'; |
| 2006 |
|
| 2007 |
foreach ($named_inclusions as $flag) { |
| 2008 |
$type = self::INCLUSION_CHILD_TYPES[$flag]; |
| 2009 |
|
| 2010 |
$present = false; |
| 2011 |
foreach ($children as $entry) { |
| 2012 |
if (($entry['type'] ?? '') === $type) { |
| 2013 |
$present = true; |
| 2014 |
break; |
| 2015 |
} |
| 2016 |
} |
| 2017 |
|
| 2018 |
if (empty($inclusions[$flag])) { |
| 2019 |
if ($present) { |
| 2020 |
$children = array_values(array_filter( |
| 2021 |
$children, |
| 2022 |
static function ($entry) use ($type): bool { |
| 2023 |
return (is_array($entry) ? ($entry['type'] ?? '') : '') !== $type; |
| 2024 |
} |
| 2025 |
)); |
| 2026 |
} |
| 2027 |
continue; |
| 2028 |
} |
| 2029 |
|
| 2030 |
if (!$present) { |
| 2031 |
$children[] = $this->build_child_sitemap_entry($type, $pattern); |
| 2032 |
} |
| 2033 |
} |
| 2034 |
|
| 2035 |
return $children; |
| 2036 |
} |
| 2037 |
|
| 2038 |
/** |
| 2039 |
* Build one child sitemap entry, resolving its filename from the url pattern. |
| 2040 |
* |
| 2041 |
* @since 1.31.0 |
| 2042 |
* |
| 2043 |
* @param string $type Child sitemap type (posts, pages, a post type name). |
| 2044 |
* @param string $pattern Filename pattern containing {type}. |
| 2045 |
* @return array Sitemap URL config. |
| 2046 |
*/ |
| 2047 |
private function build_child_sitemap_entry(string $type, string $pattern): array { |
| 2048 |
$file = str_replace('{type}', $type, $pattern); |
| 2049 |
if (strpos($file, '/') !== 0) { |
| 2050 |
$file = '/' . $file; |
| 2051 |
} |
| 2052 |
|
| 2053 |
return $this->sitemap_child_entry($file, $type); |
| 2054 |
} |
| 2055 |
|
| 2056 |
/** |
| 2057 |
* The shape generate_multiple_sitemaps() expects of a sitemap_urls entry. |
| 2058 |
* |
| 2059 |
* @since 1.31.0 |
| 2060 |
* |
| 2061 |
* @param string $url Sitemap path. |
| 2062 |
* @param string $type Entry type. |
| 2063 |
* @return array Sitemap URL config. |
| 2064 |
*/ |
| 2065 |
private function sitemap_child_entry(string $url, string $type): array { |
| 2066 |
return [ |
| 2067 |
'url' => $url, |
| 2068 |
'type' => $type, |
| 2069 |
'enabled' => true, |
| 2070 |
'last_checked' => null, |
| 2071 |
'status' => 'unknown', |
| 2072 |
]; |
| 2073 |
} |
| 2074 |
|
| 2075 |
/** |
| 2076 |
* Generate multiple sitemaps based on settings |
| 2077 |
* |
| 2078 |
* @since 1.0.0 |
| 2079 |
* @param array $settings Sitemap settings |
| 2080 |
* @return array Results of sitemap generation |
| 2081 |
*/ |
| 2082 |
public function generate_multiple_sitemaps(array $settings): array { |
| 2083 |
$results = [ |
| 2084 |
'success' => true, |
| 2085 |
'sitemaps_generated' => [], |
| 2086 |
'errors' => [], |
| 2087 |
'total_urls' => 0 |
| 2088 |
]; |
| 2089 |
|
| 2090 |
$sitemap_urls = (is_array($settings['sitemap_urls'] ?? null) ? $settings['sitemap_urls'] : []); |
| 2091 |
|
| 2092 |
if (empty($sitemap_urls)) { |
| 2093 |
$results['success'] = false; |
| 2094 |
$results['errors'][] = 'No sitemap URLs configured'; |
| 2095 |
return $results; |
| 2096 |
} |
| 2097 |
|
| 2098 |
$limit = $this->get_links_per_sitemap($settings); |
| 2099 |
|
| 2100 |
// Defer the index until every child sitemap has been generated, so it can |
| 2101 |
// list the actual files produced (including pagination pages). |
| 2102 |
$index_config = null; |
| 2103 |
$index_children = []; |
| 2104 |
|
| 2105 |
foreach ($sitemap_urls as $sitemap_config) { |
| 2106 |
if (empty($sitemap_config['enabled'])) { |
| 2107 |
continue; |
| 2108 |
} |
| 2109 |
|
| 2110 |
$type = $sitemap_config['type']; |
| 2111 |
|
| 2112 |
if ($type === 'index') { |
| 2113 |
$index_config = $sitemap_config; |
| 2114 |
continue; |
| 2115 |
} |
| 2116 |
|
| 2117 |
// Skip CPT children that no longer qualify (e.g. an internal store |
| 2118 |
// like Templately's `templately_library`) even when a previously |
| 2119 |
// saved config still lists them. Built-in aggregate types ('posts', |
| 2120 |
// 'pages', 'general', etc.) are not post type names, so this only |
| 2121 |
// affects real custom post types. |
| 2122 |
if (post_type_exists($type) && !$this->should_include_post_type($type)) { |
| 2123 |
continue; |
| 2124 |
} |
| 2125 |
|
| 2126 |
try { |
| 2127 |
// The single "general"/"WordPress" sitemap is one un-paginated file |
| 2128 |
// (there is no index to reference extra pages); it no longer drops |
| 2129 |
// overflow URLs. |
| 2130 |
if ($type === 'general' || $type === 'wordpress') { // phpcs:ignore WordPress.WP.CapitalPDangit.MisspelledInText -- lowercase on purpose: this is the stored type slug. |
| 2131 |
$xml = $this->generate_sitemap($settings); |
| 2132 |
$this->write_sitemap_page($sitemap_config['url'], $xml, $type, $this->count_urls_in_xml($xml), $results, $index_children); |
| 2133 |
continue; |
| 2134 |
} |
| 2135 |
|
| 2136 |
$source = $this->stream_type_entries($type, $settings); |
| 2137 |
|
| 2138 |
// Unknown type — fall back to a single general sitemap file. |
| 2139 |
if ($source === null) { |
| 2140 |
$xml = $this->generate_sitemap($settings); |
| 2141 |
$this->write_sitemap_page($sitemap_config['url'], $xml, $type, $this->count_urls_in_xml($xml), $results, $index_children); |
| 2142 |
continue; |
| 2143 |
} |
| 2144 |
|
| 2145 |
// Stream the entries into files of at most $limit URLs, matching |
| 2146 |
// Rank Math: page 1 keeps the base filename, pages 2+ get a -N |
| 2147 |
// suffix, and every page is listed in the index. Buffering only one |
| 2148 |
// page at a time keeps peak memory bounded to $limit entries rather |
| 2149 |
// than every URL of the type. |
| 2150 |
$image_ns = $source['image_ns']; |
| 2151 |
$buffer = []; |
| 2152 |
$page = 0; |
| 2153 |
|
| 2154 |
foreach ($source['entries'] as $entry) { |
| 2155 |
$buffer[] = $entry; |
| 2156 |
if (count($buffer) >= $limit) { |
| 2157 |
$page++; |
| 2158 |
$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children); |
| 2159 |
$buffer = []; |
| 2160 |
} |
| 2161 |
} |
| 2162 |
|
| 2163 |
// Flush the trailing partial page, or a single empty page when the |
| 2164 |
// type had no entries at all (parity with the previous behavior of |
| 2165 |
// always writing at least one page per configured child). |
| 2166 |
if (!empty($buffer) || $page === 0) { |
| 2167 |
$page++; |
| 2168 |
$this->write_sitemap_page($this->paginate_url($sitemap_config['url'], $page), $this->wrap_urlset($buffer, $settings, $image_ns), $type, count($buffer), $results, $index_children); |
| 2169 |
} |
| 2170 |
|
| 2171 |
// Remove pages left over from a previous, larger generation. |
| 2172 |
$this->cleanup_stale_pages($sitemap_config['url'], $page, $settings); |
| 2173 |
|
| 2174 |
} catch (\Exception $e) { |
| 2175 |
$results['errors'][] = "Error generating {$type} sitemap: " . $e->getMessage(); |
| 2176 |
$results['success'] = false; |
| 2177 |
} |
| 2178 |
} |
| 2179 |
|
| 2180 |
// Local business sitemap (parity with Rank Math's local-sitemap.xml). |
| 2181 |
// The physical file is written whenever a business identity is |
| 2182 |
// configured — independent of segmented vs single mode — and is also |
| 2183 |
// listed in the sitemap index when one exists. |
| 2184 |
try { |
| 2185 |
if ($this->regenerate_local_sitemap($settings) && $index_config !== null) { |
| 2186 |
$index_children[] = ['url' => '/local-sitemap.xml']; |
| 2187 |
} |
| 2188 |
} catch (\Exception $e) { |
| 2189 |
$results['errors'][] = 'Error generating local sitemap: ' . $e->getMessage(); |
| 2190 |
$results['success'] = false; |
| 2191 |
} |
| 2192 |
|
| 2193 |
// Build the index last, from the child files actually generated. |
| 2194 |
if ($index_config !== null) { |
| 2195 |
try { |
| 2196 |
$index_xml = $this->generate_sitemap_index($index_children, $settings); |
| 2197 |
$filename = basename(wp_parse_url($index_config['url'], PHP_URL_PATH)); |
| 2198 |
|
| 2199 |
if ($this->save_sitemap_to_file($index_xml, $filename)) { |
| 2200 |
$results['sitemaps_generated'][] = [ |
| 2201 |
'url' => $index_config['url'], |
| 2202 |
'type' => 'index', |
| 2203 |
'filename' => $filename, |
| 2204 |
'url_count' => count($index_children), |
| 2205 |
]; |
| 2206 |
} else { |
| 2207 |
$results['errors'][] = "Failed to save sitemap: {$filename}"; |
| 2208 |
$results['success'] = false; |
| 2209 |
} |
| 2210 |
} catch (\Exception $e) { |
| 2211 |
$results['errors'][] = 'Error generating index sitemap: ' . $e->getMessage(); |
| 2212 |
$results['success'] = false; |
| 2213 |
} |
| 2214 |
} |
| 2215 |
|
| 2216 |
// Remove segments that are no longer part of the set. Publishing was |
| 2217 |
// purely additive: a type that dropped out (Categories unticked, a CPT |
| 2218 |
// that stopped qualifying) simply stopped being overwritten, so its file |
| 2219 |
// kept serving and — until the index happened to be rebuilt — kept being |
| 2220 |
// listed in it. The index above is built from the children actually |
| 2221 |
// generated, so pruning here leaves disk and index agreeing. |
| 2222 |
$this->prune_orphaned_segments($settings, $results['sitemaps_generated']); |
| 2223 |
|
| 2224 |
return $results; |
| 2225 |
} |
| 2226 |
|
| 2227 |
/** |
| 2228 |
* Delete published segment files that this run did not write. |
| 2229 |
* |
| 2230 |
* Only filenames this site could have published under its own url pattern |
| 2231 |
* are considered, so another plugin's or core's sitemap in the web root is |
| 2232 |
* never a candidate — the same reason cleanup does not glob 'sitemap-*.xml'. |
| 2233 |
* |
| 2234 |
* @since 1.31.0 |
| 2235 |
* |
| 2236 |
* @param array $settings Sitemap settings (read for `custom_url_pattern`). |
| 2237 |
* @param array $generated Entries from $results['sitemaps_generated']. |
| 2238 |
* @return string[] Basenames removed. |
| 2239 |
*/ |
| 2240 |
private function prune_orphaned_segments(array $settings, array $generated): array { |
| 2241 |
$kept = []; |
| 2242 |
foreach ($generated as $entry) { |
| 2243 |
if (!empty($entry['filename'])) { |
| 2244 |
$kept[strtolower((string) $entry['filename'])] = true; |
| 2245 |
} |
| 2246 |
} |
| 2247 |
|
| 2248 |
// The current mode's primary and the local business sitemap are written |
| 2249 |
// by their own paths and are never orphans here. |
| 2250 |
$primary = strtolower(basename($this->get_primary_sitemap_filename($settings))); |
| 2251 |
$kept[$primary] = true; |
| 2252 |
$kept['local-sitemap.xml'] = true; |
| 2253 |
|
| 2254 |
// The OTHER mode's primary is an orphan the moment the mode changes: |
| 2255 |
// index mode leaves sitemap.xml behind, flat mode leaves |
| 2256 |
// sitemap_index.xml and its children. Both used to be kept |
| 2257 |
// unconditionally, so the site served two sitemap trees and only ever |
| 2258 |
// refreshed one (#563). The children are already covered by the segment |
| 2259 |
// sweep below, which now sees them because the index is no longer kept. |
| 2260 |
$stale_primaries = array_diff(['sitemap.xml', 'sitemap_index.xml'], [$primary]); |
| 2261 |
|
| 2262 |
$removed = []; |
| 2263 |
|
| 2264 |
global $wp_filesystem; |
| 2265 |
if (!$wp_filesystem) { |
| 2266 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 2267 |
WP_Filesystem(); |
| 2268 |
} |
| 2269 |
if (!$wp_filesystem) { |
| 2270 |
return $removed; |
| 2271 |
} |
| 2272 |
|
| 2273 |
$candidates = array_merge($this->publishable_segment_filenames($settings), $stale_primaries); |
| 2274 |
|
| 2275 |
foreach ($candidates as $candidate) { |
| 2276 |
if (isset($kept[strtolower($candidate)])) { |
| 2277 |
continue; |
| 2278 |
} |
| 2279 |
|
| 2280 |
if (!preg_match('/^(.*)\.xml$/i', $candidate, $m)) { |
| 2281 |
continue; |
| 2282 |
} |
| 2283 |
|
| 2284 |
// The base file plus its numeric pagination pages. |
| 2285 |
$paths = [ABSPATH . $candidate]; |
| 2286 |
foreach (glob(ABSPATH . $m[1] . '-*.xml') ?: [] as $paged) { |
| 2287 |
if (preg_match('/^' . preg_quote($m[1], '/') . '-\d+\.xml$/i', basename($paged))) { |
| 2288 |
$paths[] = $paged; |
| 2289 |
} |
| 2290 |
} |
| 2291 |
|
| 2292 |
foreach ($paths as $path) { |
| 2293 |
if (!file_exists($path)) { |
| 2294 |
continue; |
| 2295 |
} |
| 2296 |
// A name we could have published is not proof we published |
| 2297 |
// this file: RankMath and core write at the same paths (#515). |
| 2298 |
if (!$this->webroot_sitemap_is_ours($path, $settings)) { |
| 2299 |
continue; |
| 2300 |
} |
| 2301 |
if ($wp_filesystem->delete($path)) { |
| 2302 |
$removed[] = basename($path); |
| 2303 |
} |
| 2304 |
} |
| 2305 |
} |
| 2306 |
|
| 2307 |
return $removed; |
| 2308 |
} |
| 2309 |
|
| 2310 |
|
| 2311 |
/** |
| 2312 |
* Collect the full (un-paginated) entry list for a content sitemap type. |
| 2313 |
* |
| 2314 |
* @since 1.14.0 |
| 2315 |
* |
| 2316 |
* @param string $type Sitemap config type (posts, pages, categories, …) |
| 2317 |
* @param array $settings Sitemap settings |
| 2318 |
* @return array{entries: array<string>, image_ns: bool}|null Null for an |
| 2319 |
* unknown type (caller falls back to a general sitemap). |
| 2320 |
*/ |
| 2321 |
/** |
| 2322 |
* Write (or remove) the physical local-sitemap.xml file. |
| 2323 |
* |
| 2324 |
* Called from every sitemap regeneration path — segmented generation, the |
| 2325 |
* single-sitemap auto-regeneration, and the manual generate endpoint — so |
| 2326 |
* the local sitemap works regardless of whether the site uses a sitemap |
| 2327 |
* index. When no business identity is configured, any stale file is removed. |
| 2328 |
* |
| 2329 |
* @since 1.15.x |
| 2330 |
* @param array|null $settings Sitemap settings (falls back to saved site settings). |
| 2331 |
* @return bool True when the file was written, false when nothing was written. |
| 2332 |
*/ |
| 2333 |
public function regenerate_local_sitemap(?array $settings = null): bool { |
| 2334 |
$settings = $settings ?? $this->get_settings('site'); |
| 2335 |
$entries = $this->collect_local_entries(); |
| 2336 |
|
| 2337 |
if (empty($entries)) { |
| 2338 |
// Business identity was cleared — drop the file we left from |
| 2339 |
// before, but only ours. `local-sitemap.xml` is the name Rank Math |
| 2340 |
// publishes under too (this method mirrors it deliberately), so on |
| 2341 |
// a migrated site the file at that path may never have been ours |
| 2342 |
// to delete (#515). |
| 2343 |
$path = ABSPATH . 'local-sitemap.xml'; |
| 2344 |
if (file_exists($path) && $this->webroot_sitemap_is_ours($path, $settings)) { |
| 2345 |
wp_delete_file($path); |
| 2346 |
} |
| 2347 |
return false; |
| 2348 |
} |
| 2349 |
|
| 2350 |
$xml = $this->wrap_urlset($entries, $settings, false); |
| 2351 |
return $this->save_sitemap_to_file($xml, 'local-sitemap.xml'); |
| 2352 |
} |
| 2353 |
|
| 2354 |
/** |
| 2355 |
* Build the local business sitemap entries. |
| 2356 |
* |
| 2357 |
* Returns a single URL entry pointing at the on-site page that carries the |
| 2358 |
* business's LocalBusiness/Organization schema (the configured business URL |
| 2359 |
* when it's on this site, otherwise the homepage). Gated — like Rank Math's |
| 2360 |
* `local-sitemap.xml` — on a business (non-person) type being configured |
| 2361 |
* with an actual location (address or geo coordinates). Returns an empty |
| 2362 |
* array when no business location is set, so no empty local sitemap is |
| 2363 |
* written or added to the index. |
| 2364 |
* |
| 2365 |
* @since 1.15.x |
| 2366 |
* @return array Zero or one URL entry |
| 2367 |
*/ |
| 2368 |
private function collect_local_entries(): array { |
| 2369 |
if (!class_exists('ThinkRank\\SEO\\Site_Identity_Manager')) { |
| 2370 |
require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-site-identity-manager.php'; |
| 2371 |
} |
| 2372 |
|
| 2373 |
$identity = (new \ThinkRank\SEO\Site_Identity_Manager())->get_settings('site'); |
| 2374 |
|
| 2375 |
// Gate like Rank Math's local sitemap: a business (non-person) identity |
| 2376 |
// is configured. We only emit a single URL (the location page), so a |
| 2377 |
// full postal address is NOT required — requiring one wrongly excluded |
| 2378 |
// migrated sites, since the Rank Math importer carries over business |
| 2379 |
// type/name/phone but not the address. Personal sites, and sites with no |
| 2380 |
// business identity at all, get no local sitemap. |
| 2381 |
$business_type = strtolower((string) ($identity['business_type'] ?? '')); |
| 2382 |
if ($business_type === 'person') { |
| 2383 |
return []; |
| 2384 |
} |
| 2385 |
if ($business_type === '' && empty($identity['business_name'])) { |
| 2386 |
return []; |
| 2387 |
} |
| 2388 |
|
| 2389 |
// Prefer the configured business URL when it points at this site; |
| 2390 |
// otherwise fall back to the homepage (which outputs the schema). |
| 2391 |
$location_url = home_url('/'); |
| 2392 |
if (!empty($identity['business_url'])) { |
| 2393 |
$candidate = esc_url_raw((string) $identity['business_url']); |
| 2394 |
if ($candidate && wp_parse_url($candidate, PHP_URL_HOST) === wp_parse_url(home_url(), PHP_URL_HOST)) { |
| 2395 |
$location_url = $candidate; |
| 2396 |
} |
| 2397 |
} |
| 2398 |
|
| 2399 |
// Omit lastmod — no real modification source for the local business URL. |
| 2400 |
return [$this->generate_url_entry($location_url, '', 0.8, 'weekly')]; |
| 2401 |
} |
| 2402 |
|
| 2403 |
/** |
| 2404 |
* Resolve a streaming entry source for a sitemap type. |
| 2405 |
* |
| 2406 |
* Returns an iterable that yields the type's <url> entries one at a time |
| 2407 |
* (so the paginator never materializes the whole type), the image-namespace |
| 2408 |
* flag, or null when the type isn't one we build (caller falls back to a |
| 2409 |
* single general sitemap). Entry order matches the previous array-based |
| 2410 |
* collect_type_entries() exactly. |
| 2411 |
* |
| 2412 |
* @param string $type Sitemap type. |
| 2413 |
* @param array $settings Sitemap settings. |
| 2414 |
* @return array{entries: iterable<string>, image_ns: bool}|null |
| 2415 |
*/ |
| 2416 |
private function stream_type_entries(string $type, array $settings): ?array { |
| 2417 |
switch ($type) { |
| 2418 |
case 'posts': |
| 2419 |
return ['entries' => $this->collect_post_entries_iter(['post'], $settings), 'image_ns' => true]; |
| 2420 |
|
| 2421 |
case 'pages': |
| 2422 |
// The homepage lives in the pages sitemap (parity with prior |
| 2423 |
// output) and is emitted first; collect_post_entries_iter() |
| 2424 |
// already excludes the static front page, so there's no duplicate. |
| 2425 |
$entries = (function () use ($settings) { |
| 2426 |
yield $this->generate_url_entry(home_url('/'), $this->get_homepage_lastmod(), 1.0, 'daily'); |
| 2427 |
yield from $this->collect_post_entries_iter(['page'], $settings); |
| 2428 |
})(); |
| 2429 |
return ['entries' => $entries, 'image_ns' => true]; |
| 2430 |
|
| 2431 |
case 'categories': |
| 2432 |
return ['entries' => $this->collect_taxonomy_entries_iter('category', $settings), 'image_ns' => false]; |
| 2433 |
|
| 2434 |
case 'tags': |
| 2435 |
return ['entries' => $this->collect_taxonomy_entries_iter('post_tag', $settings), 'image_ns' => false]; |
| 2436 |
|
| 2437 |
case 'products': |
| 2438 |
$entries = post_type_exists('product') ? $this->collect_post_entries_iter(['product'], $settings) : []; |
| 2439 |
return ['entries' => $entries, 'image_ns' => true]; |
| 2440 |
|
| 2441 |
case 'product_categories': |
| 2442 |
$entries = taxonomy_exists('product_cat') ? $this->collect_taxonomy_entries_iter('product_cat', $settings) : []; |
| 2443 |
return ['entries' => $entries, 'image_ns' => false]; |
| 2444 |
|
| 2445 |
default: |
| 2446 |
$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names'); |
| 2447 |
if (in_array($type, $custom_post_types, true)) { |
| 2448 |
return ['entries' => $this->collect_post_entries_iter([$type], $settings), 'image_ns' => true]; |
| 2449 |
} |
| 2450 |
return null; |
| 2451 |
} |
| 2452 |
} |
| 2453 |
|
| 2454 |
/** |
| 2455 |
* Write one sitemap page file and record it in the results + index child list. |
| 2456 |
* |
| 2457 |
* @since 1.14.0 |
| 2458 |
* |
| 2459 |
* @param string $url Sitemap page URL |
| 2460 |
* @param string $xml Sitemap XML |
| 2461 |
* @param string $type Sitemap type (for reporting) |
| 2462 |
* @param int $url_count Number of URLs in this page |
| 2463 |
* @param array $results Results accumulator (by reference) |
| 2464 |
* @param array $index_children Index child list (by reference) |
| 2465 |
* @return void |
| 2466 |
*/ |
| 2467 |
private function write_sitemap_page(string $url, string $xml, string $type, int $url_count, array &$results, array &$index_children): void { |
| 2468 |
$filename = basename(wp_parse_url($url, PHP_URL_PATH)); |
| 2469 |
|
| 2470 |
if ($this->save_sitemap_to_file($xml, $filename)) { |
| 2471 |
$results['sitemaps_generated'][] = [ |
| 2472 |
'url' => $url, |
| 2473 |
'type' => $type, |
| 2474 |
'filename' => $filename, |
| 2475 |
'url_count' => $url_count, |
| 2476 |
]; |
| 2477 |
$results['total_urls'] += $url_count; |
| 2478 |
$index_children[] = ['url' => $url]; |
| 2479 |
} else { |
| 2480 |
$results['errors'][] = "Failed to save sitemap: {$filename}"; |
| 2481 |
$results['success'] = false; |
| 2482 |
} |
| 2483 |
} |
| 2484 |
|
| 2485 |
/** |
| 2486 |
* Delete pagination files left over when a type shrinks to fewer pages. |
| 2487 |
* |
| 2488 |
* For a base URL like /sitemap-posts.xml, removes sitemap-posts-N.xml files |
| 2489 |
* whose page number N exceeds the current page count. Page 1 (the base file, |
| 2490 |
* which has no -N suffix) is never touched. |
| 2491 |
* |
| 2492 |
* @since 1.14.0 |
| 2493 |
* |
| 2494 |
* @since 2.1.1 Each candidate must pass the content ownership test — a |
| 2495 |
* `-N.xml` page of another plugin's sitemap paginates our |
| 2496 |
* stem exactly as ours does (#515). |
| 2497 |
* |
| 2498 |
* @param string $base_url Base (page 1) sitemap URL |
| 2499 |
* @param int $current_pages Number of pages generated this run |
| 2500 |
* @param array $settings Sitemap settings, for the ownership test. |
| 2501 |
* @return void |
| 2502 |
*/ |
| 2503 |
private function cleanup_stale_pages(string $base_url, int $current_pages, array $settings): void { |
| 2504 |
$filename = basename(wp_parse_url($base_url, PHP_URL_PATH)); |
| 2505 |
if (!preg_match('/^(.*)\.xml$/i', $filename, $m)) { |
| 2506 |
return; |
| 2507 |
} |
| 2508 |
$stem = $m[1]; |
| 2509 |
|
| 2510 |
global $wp_filesystem; |
| 2511 |
if (!$wp_filesystem) { |
| 2512 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 2513 |
WP_Filesystem(); |
| 2514 |
} |
| 2515 |
if (!$wp_filesystem) { |
| 2516 |
return; |
| 2517 |
} |
| 2518 |
|
| 2519 |
$candidates = glob(ABSPATH . $stem . '-*.xml') ?: []; |
| 2520 |
foreach ($candidates as $path) { |
| 2521 |
// Only delete numeric-suffixed pages beyond the current count. |
| 2522 |
if (preg_match('/-(\d+)\.xml$/', basename($path), $mm) |
| 2523 |
&& (int) $mm[1] > $current_pages |
| 2524 |
&& $this->webroot_sitemap_is_ours($path, $settings)) { |
| 2525 |
$wp_filesystem->delete($path); |
| 2526 |
} |
| 2527 |
} |
| 2528 |
} |
| 2529 |
|
| 2530 |
/** |
| 2531 |
* Generate sitemap index XML from the list of child sitemap files produced |
| 2532 |
* during generation (each already resolved to its final, possibly paginated, |
| 2533 |
* URL). |
| 2534 |
* |
| 2535 |
* @since 1.0.0 (signature updated 1.14.0) |
| 2536 |
* @param array $children Array of ['url' => string] child sitemap entries |
| 2537 |
* @param array $settings Sitemap settings |
| 2538 |
* @return string Sitemap index XML |
| 2539 |
*/ |
| 2540 |
private function generate_sitemap_index(array $children, array $settings): string { |
| 2541 |
$xml = $this->xml_prolog($settings, 'sitemap-index.xsl'); |
| 2542 |
$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 2543 |
|
| 2544 |
$site_url = home_url(); |
| 2545 |
|
| 2546 |
foreach ($children as $child) { |
| 2547 |
$sitemap_url = $child['url'] ?? ''; |
| 2548 |
if ($sitemap_url === '') { |
| 2549 |
continue; |
| 2550 |
} |
| 2551 |
if (!str_starts_with($sitemap_url, 'http')) { |
| 2552 |
$sitemap_url = $site_url . $sitemap_url; |
| 2553 |
} |
| 2554 |
|
| 2555 |
$xml .= " <sitemap>\n"; |
| 2556 |
$xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n"; |
| 2557 |
$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n"; |
| 2558 |
$xml .= " </sitemap>\n"; |
| 2559 |
} |
| 2560 |
|
| 2561 |
$xml .= '</sitemapindex>'; |
| 2562 |
|
| 2563 |
return $xml; |
| 2564 |
} |
| 2565 |
|
| 2566 |
/** |
| 2567 |
* Count URLs in sitemap XML content |
| 2568 |
* |
| 2569 |
* @since 1.0.0 |
| 2570 |
* @param string $sitemap_xml Sitemap XML content |
| 2571 |
* @return int Number of URLs |
| 2572 |
*/ |
| 2573 |
private function count_urls_in_xml(string $sitemap_xml): int { |
| 2574 |
return substr_count($sitemap_xml, '<url>'); |
| 2575 |
} |
| 2576 |
|
| 2577 |
/** |
| 2578 |
* Validate and sanitize exclude posts input |
| 2579 |
* |
| 2580 |
* @since 1.0.0 |
| 2581 |
* @param string $exclude_posts Comma-separated post IDs |
| 2582 |
* @return array Validated post IDs |
| 2583 |
* @throws InvalidArgumentException If validation fails |
| 2584 |
*/ |
| 2585 |
private function validate_exclude_posts(string $exclude_posts): array { |
| 2586 |
if (empty($exclude_posts)) { |
| 2587 |
return []; |
| 2588 |
} |
| 2589 |
|
| 2590 |
// Limit input length to prevent DoS |
| 2591 |
if (strlen($exclude_posts) > 1000) { |
| 2592 |
throw new InvalidArgumentException('Exclude posts list too long (max 1000 characters)'); |
| 2593 |
} |
| 2594 |
|
| 2595 |
// Validate comma-separated integers only |
| 2596 |
if (!preg_match('/^[\d,\s]+$/', $exclude_posts)) { |
| 2597 |
throw new InvalidArgumentException('Exclude posts must contain only numbers and commas'); |
| 2598 |
} |
| 2599 |
|
| 2600 |
$ids = array_map('intval', array_filter(explode(',', $exclude_posts))); |
| 2601 |
|
| 2602 |
// Limit number of exclusions to prevent performance issues |
| 2603 |
if (count($ids) > 100) { |
| 2604 |
throw new InvalidArgumentException('Too many posts to exclude (max 100)'); |
| 2605 |
} |
| 2606 |
|
| 2607 |
return array_filter($ids, function($id) { |
| 2608 |
return $id > 0; // Only positive integers |
| 2609 |
}); |
| 2610 |
} |
| 2611 |
|
| 2612 |
/** |
| 2613 |
* Validate and sanitize exclude terms input |
| 2614 |
* |
| 2615 |
* @since 1.0.0 |
| 2616 |
* @param string $exclude_terms Comma-separated term IDs |
| 2617 |
* @return array Validated term IDs |
| 2618 |
* @throws InvalidArgumentException If validation fails |
| 2619 |
*/ |
| 2620 |
private function validate_exclude_terms(string $exclude_terms): array { |
| 2621 |
if (empty($exclude_terms)) { |
| 2622 |
return []; |
| 2623 |
} |
| 2624 |
|
| 2625 |
// Limit input length to prevent DoS |
| 2626 |
if (strlen($exclude_terms) > 1000) { |
| 2627 |
throw new InvalidArgumentException('Exclude terms list too long (max 1000 characters)'); |
| 2628 |
} |
| 2629 |
|
| 2630 |
// Validate comma-separated integers only |
| 2631 |
if (!preg_match('/^[\d,\s]+$/', $exclude_terms)) { |
| 2632 |
throw new InvalidArgumentException('Exclude terms must contain only numbers and commas'); |
| 2633 |
} |
| 2634 |
|
| 2635 |
$ids = array_map('intval', array_filter(explode(',', $exclude_terms))); |
| 2636 |
|
| 2637 |
// Limit number of exclusions to prevent performance issues |
| 2638 |
if (count($ids) > 100) { |
| 2639 |
throw new InvalidArgumentException('Too many terms to exclude (max 100)'); |
| 2640 |
} |
| 2641 |
|
| 2642 |
return array_filter($ids, function($id) { |
| 2643 |
return $id > 0; // Only positive integers |
| 2644 |
}); |
| 2645 |
} |
| 2646 |
|
| 2647 |
/** |
| 2648 |
* Validate and sanitize custom URL pattern |
| 2649 |
* |
| 2650 |
* @since 1.0.0 |
| 2651 |
* @param string $pattern Custom URL pattern |
| 2652 |
* @return string Validated and sanitized pattern |
| 2653 |
* @throws InvalidArgumentException If validation fails |
| 2654 |
*/ |
| 2655 |
private function validate_custom_url_pattern(string $pattern): string { |
| 2656 |
if (empty($pattern)) { |
| 2657 |
return 'sitemap-{type}.xml'; |
| 2658 |
} |
| 2659 |
|
| 2660 |
// Remove any HTML/script tags to prevent XSS |
| 2661 |
$pattern = wp_strip_all_tags($pattern); |
| 2662 |
|
| 2663 |
// Limit pattern length |
| 2664 |
if (strlen($pattern) > 100) { |
| 2665 |
throw new InvalidArgumentException('URL pattern too long (max 100 characters)'); |
| 2666 |
} |
| 2667 |
|
| 2668 |
// Validate pattern format - only allow safe characters |
| 2669 |
if (!preg_match('/^[a-zA-Z0-9\-_{}\.]+$/', $pattern)) { |
| 2670 |
throw new InvalidArgumentException('URL pattern contains invalid characters. Only letters, numbers, hyphens, underscores, dots, and {type} are allowed'); |
| 2671 |
} |
| 2672 |
|
| 2673 |
// Ensure it contains {type} placeholder |
| 2674 |
if (strpos($pattern, '{type}') === false) { |
| 2675 |
throw new InvalidArgumentException('URL pattern must contain {type} placeholder'); |
| 2676 |
} |
| 2677 |
|
| 2678 |
// Ensure it ends with .xml |
| 2679 |
if (!str_ends_with($pattern, '.xml')) { |
| 2680 |
$pattern .= '.xml'; |
| 2681 |
} |
| 2682 |
|
| 2683 |
return sanitize_file_name($pattern); |
| 2684 |
} |
| 2685 |
|
| 2686 |
/** |
| 2687 |
* Validate sitemap URL |
| 2688 |
* |
| 2689 |
* @since 1.0.0 |
| 2690 |
* @param string $url Sitemap URL |
| 2691 |
* @return string Validated and sanitized URL |
| 2692 |
* @throws InvalidArgumentException If validation fails |
| 2693 |
*/ |
| 2694 |
private function validate_sitemap_url(string $url): string { |
| 2695 |
if (empty($url)) { |
| 2696 |
throw new InvalidArgumentException('Sitemap URL cannot be empty'); |
| 2697 |
} |
| 2698 |
|
| 2699 |
// Remove leading/trailing whitespace |
| 2700 |
$url = trim($url); |
| 2701 |
|
| 2702 |
// Limit URL length |
| 2703 |
if (strlen($url) > 200) { |
| 2704 |
throw new InvalidArgumentException('Sitemap URL too long (max 200 characters)'); |
| 2705 |
} |
| 2706 |
|
| 2707 |
// Ensure it starts with / |
| 2708 |
if (!str_starts_with($url, '/')) { |
| 2709 |
$url = '/' . $url; |
| 2710 |
} |
| 2711 |
|
| 2712 |
// Validate URL path format |
| 2713 |
if (!preg_match('/^\/[a-zA-Z0-9\-_\/\.]+\.xml$/', $url)) { |
| 2714 |
throw new InvalidArgumentException('Invalid sitemap URL format. Must be a valid path ending with .xml'); |
| 2715 |
} |
| 2716 |
|
| 2717 |
// Prevent directory traversal |
| 2718 |
if (strpos($url, '..') !== false) { |
| 2719 |
throw new InvalidArgumentException('Directory traversal not allowed in sitemap URL'); |
| 2720 |
} |
| 2721 |
|
| 2722 |
// Prevent multiple slashes |
| 2723 |
$url = preg_replace('/\/+/', '/', $url); |
| 2724 |
|
| 2725 |
return sanitize_url($url); |
| 2726 |
} |
| 2727 |
|
| 2728 |
/** |
| 2729 |
* Validate links per sitemap setting |
| 2730 |
* |
| 2731 |
* @since 1.0.0 |
| 2732 |
* @param mixed $links_per_sitemap Links per sitemap value |
| 2733 |
* @return int Validated links per sitemap |
| 2734 |
* @throws InvalidArgumentException If validation fails |
| 2735 |
*/ |
| 2736 |
private function validate_links_per_sitemap($links_per_sitemap): int { |
| 2737 |
$links = intval($links_per_sitemap); |
| 2738 |
|
| 2739 |
if ($links < 1) { |
| 2740 |
throw new InvalidArgumentException('Links per sitemap must be at least 1'); |
| 2741 |
} |
| 2742 |
|
| 2743 |
if ($links > 50000) { |
| 2744 |
throw new InvalidArgumentException('Links per sitemap cannot exceed 50,000'); |
| 2745 |
} |
| 2746 |
|
| 2747 |
return $links; |
| 2748 |
} |
| 2749 |
|
| 2750 |
/** |
| 2751 |
* Validate sitemap filename for security |
| 2752 |
* |
| 2753 |
* @since 1.0.0 |
| 2754 |
* @param string $filename Filename to validate |
| 2755 |
* @return string Validated and sanitized filename |
| 2756 |
* @throws InvalidArgumentException If validation fails |
| 2757 |
*/ |
| 2758 |
private function validate_sitemap_filename(string $filename): string { |
| 2759 |
if (empty($filename)) { |
| 2760 |
return 'sitemap.xml'; |
| 2761 |
} |
| 2762 |
|
| 2763 |
// Remove any path components to prevent directory traversal |
| 2764 |
$filename = basename($filename); |
| 2765 |
|
| 2766 |
// Limit filename length |
| 2767 |
if (strlen($filename) > 100) { |
| 2768 |
throw new InvalidArgumentException('Filename too long (max 100 characters)'); |
| 2769 |
} |
| 2770 |
|
| 2771 |
// Validate filename format - only allow safe characters |
| 2772 |
if (!preg_match('/^[a-zA-Z0-9\-_\.]+$/', $filename)) { |
| 2773 |
throw new InvalidArgumentException('Filename contains invalid characters. Only letters, numbers, hyphens, underscores, and dots are allowed'); |
| 2774 |
} |
| 2775 |
|
| 2776 |
// Prevent directory traversal attempts |
| 2777 |
if (strpos($filename, '..') !== false) { |
| 2778 |
throw new InvalidArgumentException('Directory traversal not allowed in filename'); |
| 2779 |
} |
| 2780 |
|
| 2781 |
// Ensure it ends with .xml |
| 2782 |
if (!str_ends_with($filename, '.xml')) { |
| 2783 |
$filename .= '.xml'; |
| 2784 |
} |
| 2785 |
|
| 2786 |
// Additional sanitization |
| 2787 |
$filename = sanitize_file_name($filename); |
| 2788 |
|
| 2789 |
// Final security check - ensure it's still a valid XML filename |
| 2790 |
if (!preg_match('/^[a-zA-Z0-9\-_]+\.xml$/', $filename)) { |
| 2791 |
throw new InvalidArgumentException('Invalid XML filename after sanitization'); |
| 2792 |
} |
| 2793 |
|
| 2794 |
return $filename; |
| 2795 |
} |
| 2796 |
} |
| 2797 |
|
