| 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 |
/** |
| 19 |
* Sitemap Generator Class |
| 20 |
* |
| 21 |
* Generates and manages XML sitemaps for search engine optimization. |
| 22 |
* Provides dynamic sitemap generation with content filtering, priority |
| 23 |
* calculation, and change frequency optimization. |
| 24 |
* |
| 25 |
* @since 1.0.0 |
| 26 |
*/ |
| 27 |
class Sitemap_Generator extends Abstract_SEO_Manager { |
| 28 |
|
| 29 |
/** |
| 30 |
* Supported sitemap types |
| 31 |
* |
| 32 |
* @since 1.0.0 |
| 33 |
* @var array |
| 34 |
*/ |
| 35 |
private array $sitemap_types = [ |
| 36 |
'posts' => [ |
| 37 |
'name' => 'Posts', |
| 38 |
'post_types' => ['post'], |
| 39 |
'priority' => 0.8, |
| 40 |
'changefreq' => 'weekly' |
| 41 |
], |
| 42 |
'pages' => [ |
| 43 |
'name' => 'Pages', |
| 44 |
'post_types' => ['page'], |
| 45 |
'priority' => 0.9, |
| 46 |
'changefreq' => 'monthly' |
| 47 |
], |
| 48 |
'categories' => [ |
| 49 |
'name' => 'Categories', |
| 50 |
'taxonomy' => 'category', |
| 51 |
'priority' => 0.6, |
| 52 |
'changefreq' => 'weekly' |
| 53 |
], |
| 54 |
'tags' => [ |
| 55 |
'name' => 'Tags', |
| 56 |
'taxonomy' => 'post_tag', |
| 57 |
'priority' => 0.4, |
| 58 |
'changefreq' => 'monthly' |
| 59 |
] |
| 60 |
]; |
| 61 |
|
| 62 |
/** |
| 63 |
* Constructor |
| 64 |
* |
| 65 |
* @since 1.0.0 |
| 66 |
*/ |
| 67 |
public function __construct() { |
| 68 |
parent::__construct('sitemap'); |
| 69 |
|
| 70 |
// Initialize auto-generation hooks |
| 71 |
$this->init_auto_generation_hooks(); |
| 72 |
} |
| 73 |
|
| 74 |
/** |
| 75 |
* Initialize WordPress hooks for auto-generation |
| 76 |
* |
| 77 |
* @since 1.0.0 |
| 78 |
* @return void |
| 79 |
*/ |
| 80 |
private function init_auto_generation_hooks(): void { |
| 81 |
// Content change hooks - use priority 20 to run after other plugins |
| 82 |
add_action('save_post', [$this, 'handle_content_change'], 20, 2); |
| 83 |
add_action('delete_post', [$this, 'handle_content_deletion'], 20); |
| 84 |
add_action('wp_trash_post', [$this, 'handle_content_deletion'], 20); |
| 85 |
add_action('untrash_post', [$this, 'handle_content_change_by_id'], 20); |
| 86 |
|
| 87 |
// Taxonomy change hooks |
| 88 |
add_action('created_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 89 |
add_action('edited_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 90 |
add_action('delete_term', [$this, 'handle_taxonomy_change'], 20, 3); |
| 91 |
|
| 92 |
// Debounced regeneration to prevent multiple rapid-fire generations |
| 93 |
add_action('thinkrank_regenerate_sitemap', [$this, 'auto_regenerate_sitemap']); |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Generate XML sitemap |
| 98 |
* |
| 99 |
* @since 1.0.0 |
| 100 |
* |
| 101 |
* @param array $options Sitemap generation options |
| 102 |
* @return string XML sitemap content |
| 103 |
*/ |
| 104 |
public function generate_sitemap(array $options = []): string { |
| 105 |
$settings = $this->get_settings('site'); |
| 106 |
|
| 107 |
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 108 |
|
| 109 |
// Add XSL stylesheet only if styling is enabled |
| 110 |
if (!empty($settings['enable_styling'])) { |
| 111 |
$xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap.xsl') . '"?>' . "\n"; |
| 112 |
} |
| 113 |
|
| 114 |
// Add image namespace if images are enabled |
| 115 |
if (!empty($settings['include_images'])) { |
| 116 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n"; |
| 117 |
} else { |
| 118 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 119 |
} |
| 120 |
|
| 121 |
// Add homepage |
| 122 |
$xml .= $this->generate_url_entry(home_url('/'), gmdate('c'), 1.0, 'daily'); |
| 123 |
|
| 124 |
// MEMORY-EFFICIENT: Generate posts using chunked processing |
| 125 |
$enabled_post_types = $this->get_enabled_post_types($settings); |
| 126 |
if (!empty($enabled_post_types)) { |
| 127 |
$xml .= $this->generate_posts_chunked($enabled_post_types, $settings); |
| 128 |
} |
| 129 |
|
| 130 |
// OPTIMIZED: Get all enabled taxonomies and fetch in single query |
| 131 |
$enabled_taxonomies = $this->get_enabled_taxonomies($settings); |
| 132 |
if (!empty($enabled_taxonomies)) { |
| 133 |
$all_taxonomy_data = $this->fetch_taxonomies_optimized($enabled_taxonomies, $settings); |
| 134 |
|
| 135 |
// Process each taxonomy's data |
| 136 |
foreach ($enabled_taxonomies as $taxonomy) { |
| 137 |
if (!empty($all_taxonomy_data[$taxonomy])) { |
| 138 |
$xml .= $this->process_taxonomies_to_xml($all_taxonomy_data[$taxonomy], $taxonomy, $settings); |
| 139 |
} |
| 140 |
} |
| 141 |
} |
| 142 |
|
| 143 |
$xml .= '</urlset>'; |
| 144 |
|
| 145 |
return $xml; |
| 146 |
} |
| 147 |
|
| 148 |
/** |
| 149 |
* Validate SEO settings (implements interface) |
| 150 |
* |
| 151 |
* @since 1.0.0 |
| 152 |
* |
| 153 |
* @param array $settings Settings array to validate |
| 154 |
* @return array Validation results |
| 155 |
*/ |
| 156 |
public function validate_settings(array $settings): array { |
| 157 |
$validation = [ |
| 158 |
'valid' => true, |
| 159 |
'errors' => [], |
| 160 |
'warnings' => [], |
| 161 |
'suggestions' => [] |
| 162 |
]; |
| 163 |
|
| 164 |
try { |
| 165 |
// Validate exclude_posts |
| 166 |
if (!empty($settings['exclude_posts'])) { |
| 167 |
$this->validate_exclude_posts($settings['exclude_posts']); |
| 168 |
} |
| 169 |
|
| 170 |
// Validate exclude_terms |
| 171 |
if (!empty($settings['exclude_terms'])) { |
| 172 |
$this->validate_exclude_terms($settings['exclude_terms']); |
| 173 |
} |
| 174 |
|
| 175 |
// Validate custom_url_pattern |
| 176 |
if (!empty($settings['custom_url_pattern'])) { |
| 177 |
$this->validate_custom_url_pattern($settings['custom_url_pattern']); |
| 178 |
} |
| 179 |
|
| 180 |
// Validate sitemap_urls |
| 181 |
if (!empty($settings['sitemap_urls']) && is_array($settings['sitemap_urls'])) { |
| 182 |
foreach ($settings['sitemap_urls'] as $sitemap) { |
| 183 |
if (!empty($sitemap['url'])) { |
| 184 |
$this->validate_sitemap_url($sitemap['url']); |
| 185 |
} |
| 186 |
} |
| 187 |
} |
| 188 |
|
| 189 |
// Validate numeric settings |
| 190 |
if (isset($settings['links_per_sitemap'])) { |
| 191 |
$this->validate_links_per_sitemap($settings['links_per_sitemap']); |
| 192 |
} |
| 193 |
|
| 194 |
} catch (InvalidArgumentException $e) { |
| 195 |
$validation['valid'] = false; |
| 196 |
$validation['errors'][] = $e->getMessage(); |
| 197 |
} |
| 198 |
|
| 199 |
return $validation; |
| 200 |
} |
| 201 |
|
| 202 |
/** |
| 203 |
* Get output data for frontend rendering (implements interface) |
| 204 |
* |
| 205 |
* @since 1.0.0 |
| 206 |
* |
| 207 |
* @param string $context_type The context type |
| 208 |
* @param int|null $context_id Optional. Context ID |
| 209 |
* @return array Output data ready for frontend rendering |
| 210 |
*/ |
| 211 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 212 |
$settings = $this->get_settings($context_type, $context_id); |
| 213 |
|
| 214 |
return [ |
| 215 |
'sitemap_url' => home_url('/sitemap.xml'), |
| 216 |
'enabled' => $settings['enabled'] ?? true, |
| 217 |
'last_generated' => $settings['last_generated'] ?? '', |
| 218 |
'total_urls' => $this->count_sitemap_urls($settings) |
| 219 |
]; |
| 220 |
} |
| 221 |
|
| 222 |
/** |
| 223 |
* Get default settings for a context type (implements interface) |
| 224 |
* |
| 225 |
* @since 1.0.0 |
| 226 |
* |
| 227 |
* @param string $context_type The context type to get defaults for |
| 228 |
* @return array Default settings array |
| 229 |
*/ |
| 230 |
public function get_default_settings(string $context_type): array { |
| 231 |
return [ |
| 232 |
// Core settings |
| 233 |
'enabled' => true, |
| 234 |
|
| 235 |
// Multiple Sitemap URLs |
| 236 |
'sitemap_urls' => [ |
| 237 |
[ |
| 238 |
'url' => '/sitemap.xml', |
| 239 |
'type' => 'general', |
| 240 |
'enabled' => true, |
| 241 |
'last_checked' => null, |
| 242 |
'status' => 'unknown' |
| 243 |
] |
| 244 |
], |
| 245 |
'use_sitemap_index' => false, |
| 246 |
|
| 247 |
// General Settings |
| 248 |
'links_per_sitemap' => 1000, |
| 249 |
'include_images' => true, |
| 250 |
'include_featured_images' => false, |
| 251 |
'auto_generate' => true, |
| 252 |
'ping_search_engines' => true, |
| 253 |
|
| 254 |
// Content Inclusion (backward compatibility) |
| 255 |
'include_posts' => true, |
| 256 |
'include_pages' => true, |
| 257 |
'include_categories' => true, |
| 258 |
'include_tags' => false, |
| 259 |
|
| 260 |
// Content Filtering |
| 261 |
'exclude_posts' => '', |
| 262 |
'exclude_terms' => '', |
| 263 |
'exclude_password_protected' => true, |
| 264 |
'exclude_private_posts' => true, |
| 265 |
|
| 266 |
// Advanced Options |
| 267 |
'enable_styling' => true, |
| 268 |
'custom_url_pattern' => 'sitemap-{type}.xml', |
| 269 |
|
| 270 |
// Generation tracking |
| 271 |
'last_generated' => '' |
| 272 |
]; |
| 273 |
} |
| 274 |
|
| 275 |
/** |
| 276 |
* Get settings schema definition (implements interface) |
| 277 |
* |
| 278 |
* @since 1.0.0 |
| 279 |
* |
| 280 |
* @param string $context_type The context type to get schema for |
| 281 |
* @return array Settings schema definition |
| 282 |
*/ |
| 283 |
public function get_settings_schema(string $context_type): array { |
| 284 |
return [ |
| 285 |
'enabled' => [ |
| 286 |
'type' => 'boolean', |
| 287 |
'title' => 'Enable Sitemap', |
| 288 |
'description' => 'Generate XML sitemap for search engines', |
| 289 |
'default' => true |
| 290 |
], |
| 291 |
'include_posts' => [ |
| 292 |
'type' => 'boolean', |
| 293 |
'title' => 'Include Posts', |
| 294 |
'description' => 'Include blog posts in sitemap', |
| 295 |
'default' => true |
| 296 |
], |
| 297 |
'include_pages' => [ |
| 298 |
'type' => 'boolean', |
| 299 |
'title' => 'Include Pages', |
| 300 |
'description' => 'Include static pages in sitemap', |
| 301 |
'default' => true |
| 302 |
], |
| 303 |
'include_categories' => [ |
| 304 |
'type' => 'boolean', |
| 305 |
'title' => 'Include Categories', |
| 306 |
'description' => 'Include category pages in sitemap', |
| 307 |
'default' => true |
| 308 |
], |
| 309 |
'include_tags' => [ |
| 310 |
'type' => 'boolean', |
| 311 |
'title' => 'Include Tags', |
| 312 |
'description' => 'Include tag pages in sitemap', |
| 313 |
'default' => false |
| 314 |
], |
| 315 |
|
| 316 |
'auto_generate' => [ |
| 317 |
'type' => 'boolean', |
| 318 |
'title' => 'Auto Generate', |
| 319 |
'description' => 'Automatically regenerate sitemap when content changes', |
| 320 |
'default' => true |
| 321 |
], |
| 322 |
'ping_search_engines' => [ |
| 323 |
'type' => 'boolean', |
| 324 |
'title' => 'Ping Search Engines', |
| 325 |
'description' => 'Notify Google and Bing when sitemap is updated', |
| 326 |
'default' => true |
| 327 |
], |
| 328 |
'last_generated' => [ |
| 329 |
'type' => 'string', |
| 330 |
'title' => 'Last Generated', |
| 331 |
'description' => 'Timestamp of last sitemap generation', |
| 332 |
'default' => '' |
| 333 |
] |
| 334 |
]; |
| 335 |
} |
| 336 |
|
| 337 |
/** |
| 338 |
* Generate URL entry for sitemap |
| 339 |
* |
| 340 |
* @since 1.0.0 |
| 341 |
* |
| 342 |
* @param string $url URL |
| 343 |
* @param string $lastmod Last modification date |
| 344 |
* @param float $priority Priority (0.0 to 1.0) |
| 345 |
* @param string $changefreq Change frequency |
| 346 |
* @param array $images Optional array of image data |
| 347 |
* @return string XML URL entry |
| 348 |
*/ |
| 349 |
private function generate_url_entry(string $url, string $lastmod, float $priority, string $changefreq, array $images = []): string { |
| 350 |
$xml = " <url>\n"; |
| 351 |
$xml .= " <loc>" . esc_url($url) . "</loc>\n"; |
| 352 |
$xml .= " <lastmod>" . esc_html($lastmod) . "</lastmod>\n"; |
| 353 |
$xml .= " <priority>" . number_format($priority, 1) . "</priority>\n"; |
| 354 |
$xml .= " <changefreq>" . esc_html($changefreq) . "</changefreq>\n"; |
| 355 |
|
| 356 |
// Add image entries if provided |
| 357 |
foreach ($images as $image) { |
| 358 |
$xml .= " <image:image>\n"; |
| 359 |
$xml .= " <image:loc>" . esc_url($image['url']) . "</image:loc>\n"; |
| 360 |
|
| 361 |
if (!empty($image['title'])) { |
| 362 |
$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n"; |
| 363 |
} |
| 364 |
|
| 365 |
if (!empty($image['alt'])) { |
| 366 |
$xml .= " <image:caption>" . esc_html($image['alt']) . "</image:caption>\n"; |
| 367 |
} |
| 368 |
|
| 369 |
$xml .= " </image:image>\n"; |
| 370 |
} |
| 371 |
|
| 372 |
$xml .= " </url>\n"; |
| 373 |
|
| 374 |
return $xml; |
| 375 |
} |
| 376 |
|
| 377 |
/** |
| 378 |
* Generate post entries for sitemap (MEMORY-EFFICIENT VERSION) |
| 379 |
* |
| 380 |
* @since 1.0.0 |
| 381 |
* |
| 382 |
* @param string $post_type Post type |
| 383 |
* @param array $settings Sitemap settings |
| 384 |
* @return string XML entries |
| 385 |
*/ |
| 386 |
private function generate_post_entries(string $post_type, array $settings): string { |
| 387 |
// Use memory-efficient chunked processing for single post type |
| 388 |
return $this->generate_posts_chunked([$post_type], $settings); |
| 389 |
} |
| 390 |
|
| 391 |
/** |
| 392 |
* Generate posts XML using memory-efficient chunked processing |
| 393 |
* |
| 394 |
* @since 1.0.0 |
| 395 |
* |
| 396 |
* @param array $post_types Array of post types to fetch |
| 397 |
* @param array $settings Sitemap settings |
| 398 |
* @return string XML entries for all posts |
| 399 |
*/ |
| 400 |
private function generate_posts_chunked(array $post_types, array $settings): string { |
| 401 |
if (empty($post_types)) { |
| 402 |
return ''; |
| 403 |
} |
| 404 |
|
| 405 |
// Parse and validate exclude_posts setting |
| 406 |
$exclude_ids = []; |
| 407 |
if (!empty($settings['exclude_posts'])) { |
| 408 |
try { |
| 409 |
$exclude_ids = $this->validate_exclude_posts($settings['exclude_posts']); |
| 410 |
} catch (InvalidArgumentException $e) { |
| 411 |
// Continue with empty array on validation failure - error details available in exception |
| 412 |
} |
| 413 |
} |
| 414 |
|
| 415 |
// Use links_per_sitemap setting to limit posts per sitemap |
| 416 |
$links_per_sitemap = !empty($settings['links_per_sitemap']) ? |
| 417 |
intval($settings['links_per_sitemap']) : 1000; |
| 418 |
|
| 419 |
// Ensure value is within reasonable bounds |
| 420 |
$links_per_sitemap = max(1, min(50000, $links_per_sitemap)); |
| 421 |
|
| 422 |
$xml = ''; |
| 423 |
$chunk_size = 500; // Process 500 posts at a time |
| 424 |
$offset = 0; |
| 425 |
$total_processed = 0; |
| 426 |
|
| 427 |
do { |
| 428 |
// Fetch posts in chunks to prevent memory issues |
| 429 |
$posts = get_posts([ |
| 430 |
'post_type' => $post_types, |
| 431 |
'post_status' => 'publish', |
| 432 |
'numberposts' => $chunk_size, |
| 433 |
'offset' => $offset, |
| 434 |
'exclude' => $exclude_ids, |
| 435 |
'orderby' => 'post_type post_date', |
| 436 |
'order' => 'ASC DESC' |
| 437 |
]); |
| 438 |
|
| 439 |
// Process each post in the chunk |
| 440 |
foreach ($posts as $post) { |
| 441 |
if ($total_processed >= $links_per_sitemap) { |
| 442 |
break 2; // Exit both loops when limit reached |
| 443 |
} |
| 444 |
|
| 445 |
if ($this->should_include_in_sitemap($post, $settings)) { |
| 446 |
$url = get_permalink($post); |
| 447 |
$lastmod = gmdate('c', strtotime($post->post_modified_gmt)); |
| 448 |
$priority = $this->calculate_intelligent_priority($post, $post->post_type); |
| 449 |
$changefreq = $this->calculate_change_frequency($post, $post->post_type); |
| 450 |
$images = $this->extract_post_images($post, $settings); |
| 451 |
|
| 452 |
$xml .= $this->generate_url_entry($url, $lastmod, $priority, $changefreq, $images); |
| 453 |
$total_processed++; |
| 454 |
} |
| 455 |
} |
| 456 |
|
| 457 |
$offset += $chunk_size; |
| 458 |
|
| 459 |
// Store count before freeing memory |
| 460 |
$posts_count = count($posts); |
| 461 |
|
| 462 |
// Free memory after each chunk |
| 463 |
unset($posts); |
| 464 |
|
| 465 |
} while ($posts_count === $chunk_size && $total_processed < $links_per_sitemap); |
| 466 |
|
| 467 |
return $xml; |
| 468 |
} |
| 469 |
|
| 470 |
|
| 471 |
|
| 472 |
/** |
| 473 |
* Extract images from post for sitemap |
| 474 |
* |
| 475 |
* @since 1.0.0 |
| 476 |
* |
| 477 |
* @param \WP_Post $post Post object |
| 478 |
* @param array $settings Sitemap settings |
| 479 |
* @return array Array of image data |
| 480 |
*/ |
| 481 |
private function extract_post_images(\WP_Post $post, array $settings): array { |
| 482 |
$images = []; |
| 483 |
|
| 484 |
// Skip if images are disabled |
| 485 |
if (empty($settings['include_images'])) { |
| 486 |
return $images; |
| 487 |
} |
| 488 |
|
| 489 |
// Get featured image if enabled |
| 490 |
if (!empty($settings['include_featured_images'])) { |
| 491 |
$featured_image = $this->get_featured_image($post->ID); |
| 492 |
if ($featured_image) { |
| 493 |
$images[] = $featured_image; |
| 494 |
} |
| 495 |
} |
| 496 |
|
| 497 |
// Extract images from content |
| 498 |
$content_images = $this->extract_content_images($post->post_content); |
| 499 |
$images = array_merge($images, $content_images); |
| 500 |
|
| 501 |
// Remove duplicates based on URL |
| 502 |
$unique_images = []; |
| 503 |
$seen_urls = []; |
| 504 |
foreach ($images as $image) { |
| 505 |
if (!in_array($image['url'], $seen_urls)) { |
| 506 |
$unique_images[] = $image; |
| 507 |
$seen_urls[] = $image['url']; |
| 508 |
} |
| 509 |
} |
| 510 |
|
| 511 |
return $unique_images; |
| 512 |
} |
| 513 |
|
| 514 |
/** |
| 515 |
* Get featured image data |
| 516 |
* |
| 517 |
* @since 1.0.0 |
| 518 |
* |
| 519 |
* @param int $post_id Post ID |
| 520 |
* @return array|null Featured image data or null |
| 521 |
*/ |
| 522 |
private function get_featured_image(int $post_id): ?array { |
| 523 |
$thumbnail_id = get_post_thumbnail_id($post_id); |
| 524 |
if (!$thumbnail_id) { |
| 525 |
return null; |
| 526 |
} |
| 527 |
|
| 528 |
$image_url = wp_get_attachment_image_url($thumbnail_id, 'full'); |
| 529 |
if (!$image_url) { |
| 530 |
return null; |
| 531 |
} |
| 532 |
|
| 533 |
$image_title = get_the_title($thumbnail_id); |
| 534 |
$image_alt = get_post_meta($thumbnail_id, '_wp_attachment_image_alt', true); |
| 535 |
|
| 536 |
return [ |
| 537 |
'url' => $image_url, |
| 538 |
'title' => $image_title ?: '', |
| 539 |
'alt' => $image_alt ?: '' |
| 540 |
]; |
| 541 |
} |
| 542 |
|
| 543 |
/** |
| 544 |
* Extract images from post content |
| 545 |
* |
| 546 |
* @since 1.0.0 |
| 547 |
* |
| 548 |
* @param string $content Post content |
| 549 |
* @return array Array of image data |
| 550 |
*/ |
| 551 |
private function extract_content_images(string $content): array { |
| 552 |
$images = []; |
| 553 |
|
| 554 |
// Find all img tags in content |
| 555 |
preg_match_all('/<img[^>]+>/i', $content, $img_tags); |
| 556 |
|
| 557 |
foreach ($img_tags[0] as $img_tag) { |
| 558 |
// Extract src attribute |
| 559 |
if (preg_match('/src=["\']([^"\']+)["\']/', $img_tag, $src_match)) { |
| 560 |
$image_url = $src_match[1]; |
| 561 |
|
| 562 |
// Skip if not a valid URL or external image |
| 563 |
if (!filter_var($image_url, FILTER_VALIDATE_URL)) { |
| 564 |
continue; |
| 565 |
} |
| 566 |
|
| 567 |
// Extract title and alt attributes |
| 568 |
$title = ''; |
| 569 |
$alt = ''; |
| 570 |
|
| 571 |
if (preg_match('/title=["\']([^"\']*)["\']/', $img_tag, $title_match)) { |
| 572 |
$title = $title_match[1]; |
| 573 |
} |
| 574 |
|
| 575 |
if (preg_match('/alt=["\']([^"\']*)["\']/', $img_tag, $alt_match)) { |
| 576 |
$alt = $alt_match[1]; |
| 577 |
} |
| 578 |
|
| 579 |
$images[] = [ |
| 580 |
'url' => $image_url, |
| 581 |
'title' => $title, |
| 582 |
'alt' => $alt |
| 583 |
]; |
| 584 |
} |
| 585 |
} |
| 586 |
|
| 587 |
return $images; |
| 588 |
} |
| 589 |
|
| 590 |
/** |
| 591 |
* Generate taxonomy entries for sitemap (OPTIMIZED VERSION) |
| 592 |
* |
| 593 |
* @since 1.0.0 |
| 594 |
* |
| 595 |
* @param string $taxonomy Taxonomy name |
| 596 |
* @param array $settings Sitemap settings |
| 597 |
* @return string XML entries |
| 598 |
*/ |
| 599 |
private function generate_taxonomy_entries(string $taxonomy, array $settings): string { |
| 600 |
// Use optimized batch fetching for single taxonomy |
| 601 |
$taxonomy_data = $this->fetch_taxonomies_optimized([$taxonomy], $settings); |
| 602 |
|
| 603 |
if (empty($taxonomy_data[$taxonomy])) { |
| 604 |
return ''; |
| 605 |
} |
| 606 |
|
| 607 |
return $this->process_taxonomies_to_xml($taxonomy_data[$taxonomy], $taxonomy, $settings); |
| 608 |
} |
| 609 |
|
| 610 |
/** |
| 611 |
* Get enabled taxonomies based on settings |
| 612 |
* |
| 613 |
* @since 1.0.0 |
| 614 |
* @param array $settings Sitemap settings |
| 615 |
* @return array Array of enabled taxonomies |
| 616 |
*/ |
| 617 |
private function get_enabled_taxonomies(array $settings): array { |
| 618 |
$taxonomies = []; |
| 619 |
|
| 620 |
// Core taxonomies based on settings |
| 621 |
if (!empty($settings['include_categories'])) { // � |
| 622 |
Evidence-based field name |
| 623 |
$taxonomies[] = 'category'; |
| 624 |
} |
| 625 |
|
| 626 |
if (!empty($settings['include_tags'])) { // � |
| 627 |
Evidence-based field name |
| 628 |
$taxonomies[] = 'post_tag'; |
| 629 |
} |
| 630 |
|
| 631 |
// Auto-detect public custom taxonomies |
| 632 |
$custom_taxonomies = get_taxonomies([ |
| 633 |
'public' => true, |
| 634 |
'_builtin' => false |
| 635 |
], 'names'); |
| 636 |
|
| 637 |
foreach ($custom_taxonomies as $taxonomy) { |
| 638 |
if ($this->should_include_taxonomy($taxonomy)) { |
| 639 |
$taxonomies[] = $taxonomy; |
| 640 |
} |
| 641 |
} |
| 642 |
|
| 643 |
return array_unique($taxonomies); |
| 644 |
} |
| 645 |
|
| 646 |
|
| 647 |
|
| 648 |
/** |
| 649 |
* Fetch taxonomies using optimized combined query |
| 650 |
* |
| 651 |
* @since 1.0.0 |
| 652 |
* |
| 653 |
* @param array $taxonomies Array of taxonomy names to fetch |
| 654 |
* @param array $settings Sitemap settings |
| 655 |
* @return array Grouped terms by taxonomy |
| 656 |
*/ |
| 657 |
private function fetch_taxonomies_optimized(array $taxonomies, array $settings): array { |
| 658 |
if (empty($taxonomies)) { |
| 659 |
return []; |
| 660 |
} |
| 661 |
|
| 662 |
// Parse and validate exclude_terms setting |
| 663 |
$exclude_term_ids = []; |
| 664 |
if (!empty($settings['exclude_terms'])) { |
| 665 |
try { |
| 666 |
$exclude_term_ids = $this->validate_exclude_terms($settings['exclude_terms']); |
| 667 |
} catch (InvalidArgumentException $e) { |
| 668 |
// Continue with empty array on validation failure - error details available in exception |
| 669 |
} |
| 670 |
} |
| 671 |
|
| 672 |
// Determine hide_empty setting based on taxonomies |
| 673 |
$hide_empty = true; |
| 674 |
foreach ($taxonomies as $taxonomy) { |
| 675 |
// For product categories, don't hide empty categories since they might not have products yet |
| 676 |
if ($taxonomy === 'product_cat') { |
| 677 |
$hide_empty = false; |
| 678 |
break; |
| 679 |
} |
| 680 |
} |
| 681 |
|
| 682 |
// OPTIMIZED: Single query for multiple taxonomies |
| 683 |
$all_terms = get_terms([ |
| 684 |
'taxonomy' => $taxonomies, // � |
| 685 |
Multiple taxonomies in single query |
| 686 |
'hide_empty' => $hide_empty, |
| 687 |
'exclude' => $exclude_term_ids, |
| 688 |
'orderby' => 'taxonomy count', |
| 689 |
'order' => 'ASC DESC' // Order by taxonomy ASC, then count DESC |
| 690 |
]); |
| 691 |
|
| 692 |
if (is_wp_error($all_terms)) { |
| 693 |
return []; |
| 694 |
} |
| 695 |
|
| 696 |
// Group terms by taxonomy |
| 697 |
return $this->group_terms_by_taxonomy($all_terms); |
| 698 |
} |
| 699 |
|
| 700 |
/** |
| 701 |
* Group terms by taxonomy |
| 702 |
* |
| 703 |
* @since 1.0.0 |
| 704 |
* |
| 705 |
* @param array $terms Array of term objects |
| 706 |
* @return array Grouped terms by taxonomy |
| 707 |
*/ |
| 708 |
private function group_terms_by_taxonomy(array $terms): array { |
| 709 |
$grouped = []; |
| 710 |
|
| 711 |
foreach ($terms as $term) { |
| 712 |
$taxonomy = $term->taxonomy; // � |
| 713 |
Evidence-based property name |
| 714 |
|
| 715 |
if (!isset($grouped[$taxonomy])) { |
| 716 |
$grouped[$taxonomy] = []; |
| 717 |
} |
| 718 |
|
| 719 |
$grouped[$taxonomy][] = $term; |
| 720 |
} |
| 721 |
|
| 722 |
return $grouped; |
| 723 |
} |
| 724 |
|
| 725 |
/** |
| 726 |
* Process taxonomy terms to XML entries |
| 727 |
* |
| 728 |
* @since 1.0.0 |
| 729 |
* |
| 730 |
* @param array $terms Array of term objects |
| 731 |
* @param string $taxonomy Taxonomy name |
| 732 |
* @param array $settings Sitemap settings |
| 733 |
* @return string XML entries |
| 734 |
*/ |
| 735 |
private function process_taxonomies_to_xml(array $terms, string $taxonomy, array $settings): string { |
| 736 |
$xml = ''; |
| 737 |
|
| 738 |
foreach ($terms as $term) { |
| 739 |
$url = get_term_link($term); |
| 740 |
if (!is_wp_error($url)) { |
| 741 |
$lastmod = gmdate('c'); |
| 742 |
$priority = $taxonomy === 'category' ? 0.6 : 0.4; |
| 743 |
$changefreq = 'weekly'; |
| 744 |
|
| 745 |
$xml .= $this->generate_url_entry($url, $lastmod, $priority, $changefreq); |
| 746 |
} |
| 747 |
} |
| 748 |
|
| 749 |
return $xml; |
| 750 |
} |
| 751 |
|
| 752 |
/** |
| 753 |
* Calculate intelligent priority based on content factors |
| 754 |
* |
| 755 |
* @since 1.0.0 |
| 756 |
* |
| 757 |
* @param \WP_Post $post Post object |
| 758 |
* @param string $post_type Post type |
| 759 |
* @return float Priority value between 0.1 and 1.0 |
| 760 |
*/ |
| 761 |
private function calculate_intelligent_priority(\WP_Post $post, string $post_type): float { |
| 762 |
$base_priority = $post_type === 'page' ? 0.9 : 0.8; |
| 763 |
|
| 764 |
// Factors that can adjust priority |
| 765 |
$adjustments = 0; |
| 766 |
|
| 767 |
// Recent content gets higher priority |
| 768 |
$days_old = (time() - strtotime($post->post_date)) / DAY_IN_SECONDS; |
| 769 |
if ($days_old < 30) { |
| 770 |
$adjustments += 0.1; // Recent content boost |
| 771 |
} elseif ($days_old > 365) { |
| 772 |
$adjustments -= 0.1; // Older content penalty |
| 773 |
} |
| 774 |
|
| 775 |
// Content length factor |
| 776 |
$content_length = strlen(wp_strip_all_tags($post->post_content)); |
| 777 |
if ($content_length > 2000) { |
| 778 |
$adjustments += 0.05; // Comprehensive content boost |
| 779 |
} elseif ($content_length < 500) { |
| 780 |
$adjustments -= 0.1; // Thin content penalty |
| 781 |
} |
| 782 |
|
| 783 |
// Special page types get higher priority |
| 784 |
if ($post_type === 'page') { |
| 785 |
$page_template = get_page_template_slug($post->ID); |
| 786 |
if (in_array($page_template, ['page-home.php', 'front-page.php']) || |
| 787 |
$post->ID == get_option('page_on_front')) { |
| 788 |
$base_priority = 1.0; // Homepage gets maximum priority |
| 789 |
} elseif (in_array($page_template, ['page-contact.php', 'page-about.php'])) { |
| 790 |
$adjustments += 0.05; // Important pages boost |
| 791 |
} |
| 792 |
} |
| 793 |
|
| 794 |
// Ensure priority stays within valid range |
| 795 |
$final_priority = max(0.1, min(1.0, $base_priority + $adjustments)); |
| 796 |
|
| 797 |
return round($final_priority, 1); |
| 798 |
} |
| 799 |
|
| 800 |
/** |
| 801 |
* Calculate change frequency based on content type and age |
| 802 |
* |
| 803 |
* @since 1.0.0 |
| 804 |
* |
| 805 |
* @param \WP_Post $post Post object |
| 806 |
* @param string $post_type Post type |
| 807 |
* @return string Change frequency |
| 808 |
*/ |
| 809 |
private function calculate_change_frequency(\WP_Post $post, string $post_type): string { |
| 810 |
// Pages typically change less frequently |
| 811 |
if ($post_type === 'page') { |
| 812 |
$page_template = get_page_template_slug($post->ID); |
| 813 |
if ($post->ID == get_option('page_on_front')) { |
| 814 |
return 'daily'; // Homepage changes frequently |
| 815 |
} elseif (in_array($page_template, ['page-contact.php', 'page-about.php'])) { |
| 816 |
return 'monthly'; // Static pages change monthly |
| 817 |
} |
| 818 |
return 'yearly'; // Other pages change rarely |
| 819 |
} |
| 820 |
|
| 821 |
// Posts frequency based on age and type |
| 822 |
$days_old = (time() - strtotime($post->post_date)) / DAY_IN_SECONDS; |
| 823 |
|
| 824 |
if ($days_old < 7) { |
| 825 |
return 'daily'; // Very recent posts |
| 826 |
} elseif ($days_old < 30) { |
| 827 |
return 'weekly'; // Recent posts |
| 828 |
} elseif ($days_old < 365) { |
| 829 |
return 'monthly'; // Older posts |
| 830 |
} |
| 831 |
|
| 832 |
return 'yearly'; // Very old posts |
| 833 |
} |
| 834 |
|
| 835 |
/** |
| 836 |
* Determine if content should be included in sitemap |
| 837 |
* |
| 838 |
* @since 1.0.0 |
| 839 |
* |
| 840 |
* @param \WP_Post $post Post object |
| 841 |
* @param array $settings Sitemap settings |
| 842 |
* @return bool Whether to include in sitemap |
| 843 |
*/ |
| 844 |
private function should_include_in_sitemap(\WP_Post $post, array $settings): bool { |
| 845 |
// Respect user setting for password protected content |
| 846 |
if (!empty($post->post_password) && !empty($settings['exclude_password_protected'])) { |
| 847 |
return false; |
| 848 |
} |
| 849 |
|
| 850 |
// Respect user setting for private posts |
| 851 |
if ($post->post_status === 'private' && !empty($settings['exclude_private_posts'])) { |
| 852 |
return false; |
| 853 |
} |
| 854 |
|
| 855 |
// Skip very short content (likely test/placeholder) |
| 856 |
$content_length = strlen(wp_strip_all_tags($post->post_content)); |
| 857 |
if ($content_length < 100) { |
| 858 |
return false; |
| 859 |
} |
| 860 |
|
| 861 |
// Skip posts with test/demo indicators in title (refined filtering) |
| 862 |
$title_indicators = ['test', 'demo', 'sample', 'placeholder']; |
| 863 |
$title_lower = strtolower($post->post_title); |
| 864 |
|
| 865 |
foreach ($title_indicators as $indicator) { |
| 866 |
if (strpos($title_lower, $indicator) !== false) { |
| 867 |
return false; |
| 868 |
} |
| 869 |
} |
| 870 |
|
| 871 |
// Skip content with obvious placeholder text |
| 872 |
$content_indicators = ['lorem ipsum']; |
| 873 |
$content_lower = strtolower($post->post_content); |
| 874 |
|
| 875 |
foreach ($content_indicators as $indicator) { |
| 876 |
if (strpos($content_lower, $indicator) !== false) { |
| 877 |
return false; |
| 878 |
} |
| 879 |
} |
| 880 |
|
| 881 |
// Skip drafts (but allow private posts if user setting permits) |
| 882 |
if (!in_array($post->post_status, ['publish', 'private'])) { |
| 883 |
return false; |
| 884 |
} |
| 885 |
|
| 886 |
// Check if post is set to noindex via meta |
| 887 |
$noindex = get_post_meta($post->ID, '_thinkrank_noindex', true); |
| 888 |
if ($noindex === '1' || $noindex === 'true') { |
| 889 |
return false; |
| 890 |
} |
| 891 |
|
| 892 |
return true; |
| 893 |
} |
| 894 |
|
| 895 |
/** |
| 896 |
* Count total URLs in sitemap |
| 897 |
* |
| 898 |
* @since 1.0.0 |
| 899 |
* |
| 900 |
* @param array $settings Sitemap settings |
| 901 |
* @return int Total URL count |
| 902 |
*/ |
| 903 |
private function count_sitemap_urls(array $settings): int { |
| 904 |
$count = 1; // Homepage |
| 905 |
|
| 906 |
if (!empty($settings['include_posts'])) { |
| 907 |
$count += wp_count_posts('post')->publish; |
| 908 |
} |
| 909 |
|
| 910 |
if (!empty($settings['include_pages'])) { |
| 911 |
$count += wp_count_posts('page')->publish; |
| 912 |
} |
| 913 |
|
| 914 |
if (!empty($settings['include_categories'])) { |
| 915 |
$count += wp_count_terms(['taxonomy' => 'category', 'hide_empty' => true]); |
| 916 |
} |
| 917 |
|
| 918 |
if (!empty($settings['include_tags'])) { |
| 919 |
$count += wp_count_terms(['taxonomy' => 'post_tag', 'hide_empty' => true]); |
| 920 |
} |
| 921 |
|
| 922 |
return $count; |
| 923 |
} |
| 924 |
|
| 925 |
/** |
| 926 |
* Handle content changes for auto-generation |
| 927 |
* |
| 928 |
* @since 1.0.0 |
| 929 |
* @param int $post_id Post ID |
| 930 |
* @param \WP_Post $post Post object |
| 931 |
* @return void |
| 932 |
*/ |
| 933 |
public function handle_content_change(int $post_id, \WP_Post $post): void { |
| 934 |
// Skip if auto-generation is disabled |
| 935 |
if (!$this->should_auto_generate()) { |
| 936 |
return; |
| 937 |
} |
| 938 |
|
| 939 |
// Skip autosaves and revisions |
| 940 |
if (wp_is_post_autosave($post_id) || wp_is_post_revision($post_id)) { |
| 941 |
return; |
| 942 |
} |
| 943 |
|
| 944 |
// Only process published content |
| 945 |
if ($post->post_status !== 'publish') { |
| 946 |
return; |
| 947 |
} |
| 948 |
|
| 949 |
// Check if this post type should trigger regeneration |
| 950 |
if (!$this->should_include_post_type($post->post_type)) { |
| 951 |
return; |
| 952 |
} |
| 953 |
|
| 954 |
// Schedule debounced regeneration |
| 955 |
$this->schedule_debounced_regeneration(); |
| 956 |
} |
| 957 |
|
| 958 |
/** |
| 959 |
* Handle content deletion for auto-generation |
| 960 |
* |
| 961 |
* @since 1.0.0 |
| 962 |
* @param int $post_id Post ID |
| 963 |
* @return void |
| 964 |
*/ |
| 965 |
public function handle_content_deletion(int $post_id): void { |
| 966 |
// Skip if auto-generation is disabled |
| 967 |
if (!$this->should_auto_generate()) { |
| 968 |
return; |
| 969 |
} |
| 970 |
|
| 971 |
$post = get_post($post_id); |
| 972 |
if (!$post) { |
| 973 |
return; |
| 974 |
} |
| 975 |
|
| 976 |
// Check if this post type should trigger regeneration |
| 977 |
if (!$this->should_include_post_type($post->post_type)) { |
| 978 |
return; |
| 979 |
} |
| 980 |
|
| 981 |
// Schedule debounced regeneration |
| 982 |
$this->schedule_debounced_regeneration(); |
| 983 |
} |
| 984 |
|
| 985 |
/** |
| 986 |
* Handle content change by ID (for untrash, etc.) |
| 987 |
* |
| 988 |
* @since 1.0.0 |
| 989 |
* @param int $post_id Post ID |
| 990 |
* @return void |
| 991 |
*/ |
| 992 |
public function handle_content_change_by_id(int $post_id): void { |
| 993 |
$post = get_post($post_id); |
| 994 |
if ($post) { |
| 995 |
$this->handle_content_change($post_id, $post); |
| 996 |
} |
| 997 |
} |
| 998 |
|
| 999 |
/** |
| 1000 |
* Handle taxonomy changes for auto-generation |
| 1001 |
* |
| 1002 |
* @since 1.0.0 |
| 1003 |
* @param int $term_id Term ID |
| 1004 |
* @param int $tt_id Term taxonomy ID |
| 1005 |
* @param string $taxonomy Taxonomy slug |
| 1006 |
* @return void |
| 1007 |
*/ |
| 1008 |
public function handle_taxonomy_change(int $term_id, int $tt_id, string $taxonomy): void { |
| 1009 |
// Skip if auto-generation is disabled |
| 1010 |
if (!$this->should_auto_generate()) { |
| 1011 |
return; |
| 1012 |
} |
| 1013 |
|
| 1014 |
// Check if this taxonomy should trigger regeneration |
| 1015 |
if (!$this->should_include_taxonomy($taxonomy)) { |
| 1016 |
return; |
| 1017 |
} |
| 1018 |
|
| 1019 |
// Schedule debounced regeneration |
| 1020 |
$this->schedule_debounced_regeneration(); |
| 1021 |
} |
| 1022 |
|
| 1023 |
/** |
| 1024 |
* Check if auto-generation is enabled |
| 1025 |
* |
| 1026 |
* @since 1.0.0 |
| 1027 |
* @return bool True if auto-generation is enabled |
| 1028 |
*/ |
| 1029 |
private function should_auto_generate(): bool { |
| 1030 |
$settings = $this->get_settings('site'); |
| 1031 |
|
| 1032 |
// Check if sitemap is enabled |
| 1033 |
if (empty($settings['enabled'])) { |
| 1034 |
return false; |
| 1035 |
} |
| 1036 |
|
| 1037 |
// Check if auto-generation is enabled |
| 1038 |
return !empty($settings['auto_generate']); |
| 1039 |
} |
| 1040 |
|
| 1041 |
/** |
| 1042 |
* Get enabled post types based on settings |
| 1043 |
* |
| 1044 |
* @since 1.0.0 |
| 1045 |
* @param array $settings Sitemap settings |
| 1046 |
* @return array Array of enabled post types |
| 1047 |
*/ |
| 1048 |
private function get_enabled_post_types(array $settings): array { |
| 1049 |
$post_types = []; |
| 1050 |
|
| 1051 |
// Core post types based on settings |
| 1052 |
if (!empty($settings['include_posts'])) { |
| 1053 |
$post_types[] = 'post'; |
| 1054 |
} |
| 1055 |
|
| 1056 |
if (!empty($settings['include_pages'])) { |
| 1057 |
$post_types[] = 'page'; |
| 1058 |
} |
| 1059 |
|
| 1060 |
// Auto-detect public custom post types that should be included |
| 1061 |
$custom_post_types = get_post_types([ |
| 1062 |
'public' => true, |
| 1063 |
'_builtin' => false |
| 1064 |
], 'names'); |
| 1065 |
|
| 1066 |
foreach ($custom_post_types as $post_type) { |
| 1067 |
if ($this->should_include_post_type($post_type)) { |
| 1068 |
$post_types[] = $post_type; |
| 1069 |
} |
| 1070 |
} |
| 1071 |
|
| 1072 |
return array_unique($post_types); |
| 1073 |
} |
| 1074 |
|
| 1075 |
/** |
| 1076 |
* Check if post type should trigger regeneration |
| 1077 |
* |
| 1078 |
* @since 1.0.0 |
| 1079 |
* @param string $post_type Post type |
| 1080 |
* @return bool True if post type should trigger regeneration |
| 1081 |
*/ |
| 1082 |
private function should_include_post_type(string $post_type): bool { |
| 1083 |
// Include all public post types (presets control which sitemaps are created) |
| 1084 |
$post_type_obj = get_post_type_object($post_type); |
| 1085 |
return $post_type_obj && $post_type_obj->public; |
| 1086 |
} |
| 1087 |
|
| 1088 |
/** |
| 1089 |
* Check if taxonomy should trigger regeneration |
| 1090 |
* |
| 1091 |
* @since 1.0.0 |
| 1092 |
* @param string $taxonomy Taxonomy slug |
| 1093 |
* @return bool True if taxonomy should trigger regeneration |
| 1094 |
*/ |
| 1095 |
private function should_include_taxonomy(string $taxonomy): bool { |
| 1096 |
// Include all public taxonomies (presets control which sitemaps are created) |
| 1097 |
$taxonomy_obj = get_taxonomy($taxonomy); |
| 1098 |
return $taxonomy_obj && $taxonomy_obj->public; |
| 1099 |
} |
| 1100 |
|
| 1101 |
/** |
| 1102 |
* Schedule debounced sitemap regeneration |
| 1103 |
* |
| 1104 |
* @since 1.0.0 |
| 1105 |
* @return void |
| 1106 |
*/ |
| 1107 |
private function schedule_debounced_regeneration(): void { |
| 1108 |
// Clear any existing scheduled regeneration |
| 1109 |
wp_clear_scheduled_hook('thinkrank_regenerate_sitemap'); |
| 1110 |
|
| 1111 |
// Schedule regeneration in 30 seconds to debounce rapid changes |
| 1112 |
wp_schedule_single_event(time() + 30, 'thinkrank_regenerate_sitemap'); |
| 1113 |
} |
| 1114 |
|
| 1115 |
/** |
| 1116 |
* Auto-regenerate sitemap (called by scheduled action) |
| 1117 |
* |
| 1118 |
* @since 1.0.0 |
| 1119 |
* @return void |
| 1120 |
*/ |
| 1121 |
public function auto_regenerate_sitemap(): void { |
| 1122 |
try { |
| 1123 |
// Double-check that auto-generation is still enabled |
| 1124 |
if (!$this->should_auto_generate()) { |
| 1125 |
return; |
| 1126 |
} |
| 1127 |
|
| 1128 |
// Generate sitemap with current settings |
| 1129 |
$settings = $this->get_settings('site'); |
| 1130 |
$sitemap_xml = $this->generate_sitemap($settings); |
| 1131 |
|
| 1132 |
// Save sitemap to file |
| 1133 |
$this->save_sitemap_to_file($sitemap_xml); |
| 1134 |
|
| 1135 |
// Update last generated timestamp |
| 1136 |
$this->update_setting('last_generated', gmdate('c'), 'site'); |
| 1137 |
|
| 1138 |
// Ping search engines if enabled |
| 1139 |
if (!empty($settings['ping_search_engines'])) { |
| 1140 |
$this->ping_search_engines(); |
| 1141 |
} |
| 1142 |
|
| 1143 |
// Sitemap auto-regenerated successfully |
| 1144 |
|
| 1145 |
} catch (\Exception $e) { |
| 1146 |
// Sitemap auto-regeneration failed - error details available in exception |
| 1147 |
} |
| 1148 |
} |
| 1149 |
|
| 1150 |
/** |
| 1151 |
* Save sitemap XML to file |
| 1152 |
* |
| 1153 |
* @since 1.0.0 |
| 1154 |
* @param string $sitemap_xml Sitemap XML content |
| 1155 |
* @param string $filename Optional. Filename to save (defaults to 'sitemap.xml') |
| 1156 |
* @return bool True on success, false on failure |
| 1157 |
*/ |
| 1158 |
private function save_sitemap_to_file(string $sitemap_xml, string $filename = 'sitemap.xml'): bool { |
| 1159 |
// Validate and sanitize filename for security |
| 1160 |
try { |
| 1161 |
$filename = $this->validate_sitemap_filename($filename); |
| 1162 |
} catch (InvalidArgumentException $e) { |
| 1163 |
// File validation failed - error details available in exception |
| 1164 |
return false; |
| 1165 |
} |
| 1166 |
|
| 1167 |
$sitemap_path = ABSPATH . $filename; |
| 1168 |
|
| 1169 |
// Use WordPress filesystem API for better security |
| 1170 |
global $wp_filesystem; |
| 1171 |
if (!$wp_filesystem) { |
| 1172 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 1173 |
WP_Filesystem(); |
| 1174 |
} |
| 1175 |
|
| 1176 |
if ($wp_filesystem) { |
| 1177 |
return $wp_filesystem->put_contents($sitemap_path, $sitemap_xml, FS_CHMOD_FILE); |
| 1178 |
} |
| 1179 |
|
| 1180 |
// Fallback to file_put_contents if WP_Filesystem fails |
| 1181 |
return file_put_contents($sitemap_path, $sitemap_xml) !== false; |
| 1182 |
} |
| 1183 |
|
| 1184 |
/** |
| 1185 |
* Generate multiple sitemaps based on settings |
| 1186 |
* |
| 1187 |
* @since 1.0.0 |
| 1188 |
* @param array $settings Sitemap settings |
| 1189 |
* @return array Results of sitemap generation |
| 1190 |
*/ |
| 1191 |
public function generate_multiple_sitemaps(array $settings): array { |
| 1192 |
$results = [ |
| 1193 |
'success' => true, |
| 1194 |
'sitemaps_generated' => [], |
| 1195 |
'errors' => [], |
| 1196 |
'total_urls' => 0 |
| 1197 |
]; |
| 1198 |
|
| 1199 |
$sitemap_urls = $settings['sitemap_urls'] ?? []; |
| 1200 |
|
| 1201 |
if (empty($sitemap_urls)) { |
| 1202 |
$results['success'] = false; |
| 1203 |
$results['errors'][] = 'No sitemap URLs configured'; |
| 1204 |
return $results; |
| 1205 |
} |
| 1206 |
|
| 1207 |
foreach ($sitemap_urls as $sitemap_config) { |
| 1208 |
if (!$sitemap_config['enabled']) { |
| 1209 |
continue; |
| 1210 |
} |
| 1211 |
|
| 1212 |
try { |
| 1213 |
$sitemap_xml = ''; |
| 1214 |
$url_count = 0; |
| 1215 |
|
| 1216 |
// Generate sitemap based on type |
| 1217 |
switch ($sitemap_config['type']) { |
| 1218 |
case 'index': |
| 1219 |
$sitemap_xml = $this->generate_sitemap_index($sitemap_urls); |
| 1220 |
$url_count = count(array_filter($sitemap_urls, fn($s) => $s['enabled'] && $s['type'] !== 'index')); |
| 1221 |
break; |
| 1222 |
case 'general': |
| 1223 |
case 'wordpress': |
| 1224 |
$sitemap_xml = $this->generate_sitemap($settings); |
| 1225 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1226 |
break; |
| 1227 |
case 'posts': |
| 1228 |
$sitemap_xml = $this->generate_post_sitemap('post', $settings); |
| 1229 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1230 |
break; |
| 1231 |
case 'pages': |
| 1232 |
$sitemap_xml = $this->generate_post_sitemap('page', $settings); |
| 1233 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1234 |
break; |
| 1235 |
case 'categories': |
| 1236 |
$sitemap_xml = $this->generate_taxonomy_sitemap('category', $settings); |
| 1237 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1238 |
break; |
| 1239 |
case 'products': |
| 1240 |
// Generate product-specific sitemap |
| 1241 |
if (post_type_exists('product')) { |
| 1242 |
$sitemap_xml = $this->generate_post_sitemap('product', $settings); |
| 1243 |
} else { |
| 1244 |
// Create empty sitemap if WooCommerce not installed |
| 1245 |
$sitemap_xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 1246 |
|
| 1247 |
// Add XSL stylesheet only if styling is enabled |
| 1248 |
if (!empty($settings['enable_styling'])) { |
| 1249 |
$sitemap_xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap.xsl') . '"?>' . "\n"; |
| 1250 |
} |
| 1251 |
|
| 1252 |
$sitemap_xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 1253 |
$sitemap_xml .= '</urlset>'; |
| 1254 |
} |
| 1255 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1256 |
break; |
| 1257 |
case 'product_categories': |
| 1258 |
// Generate product category sitemap |
| 1259 |
if (taxonomy_exists('product_cat')) { |
| 1260 |
$sitemap_xml = $this->generate_taxonomy_sitemap('product_cat', $settings); |
| 1261 |
} else { |
| 1262 |
// Create empty sitemap if WooCommerce not installed |
| 1263 |
$sitemap_xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 1264 |
|
| 1265 |
// Add XSL stylesheet only if styling is enabled |
| 1266 |
if (!empty($settings['enable_styling'])) { |
| 1267 |
$sitemap_xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap.xsl') . '"?>' . "\n"; |
| 1268 |
} |
| 1269 |
|
| 1270 |
$sitemap_xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 1271 |
$sitemap_xml .= '</urlset>'; |
| 1272 |
} |
| 1273 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1274 |
break; |
| 1275 |
default: |
| 1276 |
// Check if it's a custom post type |
| 1277 |
$post_types = get_post_types(['public' => true, '_builtin' => false], 'names'); |
| 1278 |
if (in_array($sitemap_config['type'], $post_types)) { |
| 1279 |
$sitemap_xml = $this->generate_post_sitemap($sitemap_config['type'], $settings); |
| 1280 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1281 |
} else { |
| 1282 |
// Fallback to general sitemap |
| 1283 |
$sitemap_xml = $this->generate_sitemap($settings); |
| 1284 |
$url_count = $this->count_urls_in_xml($sitemap_xml); |
| 1285 |
} |
| 1286 |
break; |
| 1287 |
} |
| 1288 |
|
| 1289 |
// Extract filename from URL |
| 1290 |
$filename = basename(wp_parse_url($sitemap_config['url'], PHP_URL_PATH)); |
| 1291 |
|
| 1292 |
// Save sitemap to file |
| 1293 |
if ($this->save_sitemap_to_file($sitemap_xml, $filename)) { |
| 1294 |
$results['sitemaps_generated'][] = [ |
| 1295 |
'url' => $sitemap_config['url'], |
| 1296 |
'type' => $sitemap_config['type'], |
| 1297 |
'filename' => $filename, |
| 1298 |
'url_count' => $url_count |
| 1299 |
]; |
| 1300 |
$results['total_urls'] += $url_count; |
| 1301 |
} else { |
| 1302 |
$error_message = "Failed to save sitemap: {$filename}"; |
| 1303 |
$results['errors'][] = $error_message; |
| 1304 |
$results['success'] = false; |
| 1305 |
|
| 1306 |
// File save error details available in results array |
| 1307 |
} |
| 1308 |
|
| 1309 |
} catch (\Exception $e) { |
| 1310 |
$results['errors'][] = "Error generating {$sitemap_config['type']} sitemap: " . $e->getMessage(); |
| 1311 |
$results['success'] = false; |
| 1312 |
} |
| 1313 |
} |
| 1314 |
|
| 1315 |
return $results; |
| 1316 |
} |
| 1317 |
|
| 1318 |
/** |
| 1319 |
* Generate sitemap index XML |
| 1320 |
* |
| 1321 |
* @since 1.0.0 |
| 1322 |
* @param array $sitemap_urls Array of sitemap configurations |
| 1323 |
* @return string Sitemap index XML |
| 1324 |
*/ |
| 1325 |
private function generate_sitemap_index(array $sitemap_urls): string { |
| 1326 |
$settings = $this->get_settings('site'); |
| 1327 |
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 1328 |
|
| 1329 |
// Add XSL stylesheet only if styling is enabled |
| 1330 |
if (!empty($settings['enable_styling'])) { |
| 1331 |
$xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap-index.xsl') . '"?>' . "\n"; |
| 1332 |
} |
| 1333 |
$xml .= '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 1334 |
|
| 1335 |
$site_url = home_url(); |
| 1336 |
|
| 1337 |
foreach ($sitemap_urls as $sitemap_config) { |
| 1338 |
if (!$sitemap_config['enabled'] || $sitemap_config['type'] === 'index') { |
| 1339 |
continue; |
| 1340 |
} |
| 1341 |
|
| 1342 |
$sitemap_url = $sitemap_config['url']; |
| 1343 |
if (!str_starts_with($sitemap_url, 'http')) { |
| 1344 |
$sitemap_url = $site_url . $sitemap_url; |
| 1345 |
} |
| 1346 |
|
| 1347 |
$xml .= " <sitemap>\n"; |
| 1348 |
$xml .= " <loc>" . esc_url($sitemap_url) . "</loc>\n"; |
| 1349 |
$xml .= " <lastmod>" . gmdate('c') . "</lastmod>\n"; |
| 1350 |
$xml .= " </sitemap>\n"; |
| 1351 |
} |
| 1352 |
|
| 1353 |
$xml .= '</sitemapindex>'; |
| 1354 |
|
| 1355 |
return $xml; |
| 1356 |
} |
| 1357 |
|
| 1358 |
/** |
| 1359 |
* Generate sitemap for specific post type |
| 1360 |
* |
| 1361 |
* @since 1.0.0 |
| 1362 |
* @param string $post_type Post type to generate sitemap for |
| 1363 |
* @param array $settings Sitemap settings |
| 1364 |
* @return string Sitemap XML |
| 1365 |
*/ |
| 1366 |
private function generate_post_sitemap(string $post_type, array $settings): string { |
| 1367 |
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 1368 |
|
| 1369 |
// Add XSL stylesheet only if styling is enabled |
| 1370 |
if (!empty($settings['enable_styling'])) { |
| 1371 |
$xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap.xsl') . '"?>' . "\n"; |
| 1372 |
} |
| 1373 |
|
| 1374 |
// Add image namespace if images are enabled |
| 1375 |
if (!empty($settings['include_images'])) { |
| 1376 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n"; |
| 1377 |
} else { |
| 1378 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 1379 |
} |
| 1380 |
|
| 1381 |
// Add homepage only for general sitemap |
| 1382 |
if ($post_type === 'page') { |
| 1383 |
$xml .= $this->generate_url_entry(home_url('/'), gmdate('c'), 1.0, 'daily'); |
| 1384 |
} |
| 1385 |
|
| 1386 |
// Add posts/pages of specific type |
| 1387 |
$xml .= $this->generate_post_entries($post_type, $settings); |
| 1388 |
|
| 1389 |
$xml .= '</urlset>'; |
| 1390 |
return $xml; |
| 1391 |
} |
| 1392 |
|
| 1393 |
/** |
| 1394 |
* Generate sitemap for specific taxonomy |
| 1395 |
* |
| 1396 |
* @since 1.0.0 |
| 1397 |
* @param string $taxonomy Taxonomy to generate sitemap for |
| 1398 |
* @param array $settings Sitemap settings |
| 1399 |
* @return string Sitemap XML |
| 1400 |
*/ |
| 1401 |
private function generate_taxonomy_sitemap(string $taxonomy, array $settings): string { |
| 1402 |
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
| 1403 |
|
| 1404 |
// Add XSL stylesheet only if styling is enabled |
| 1405 |
if (!empty($settings['enable_styling'])) { |
| 1406 |
$xml .= '<?xml-stylesheet type="text/xsl" href="' . home_url('/wp-content/plugins/thinkrank/assets/sitemap.xsl') . '"?>' . "\n"; |
| 1407 |
} |
| 1408 |
$xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n"; |
| 1409 |
|
| 1410 |
// Add taxonomy terms |
| 1411 |
$xml .= $this->generate_taxonomy_entries($taxonomy, $settings); |
| 1412 |
|
| 1413 |
$xml .= '</urlset>'; |
| 1414 |
return $xml; |
| 1415 |
} |
| 1416 |
|
| 1417 |
/** |
| 1418 |
* Count URLs in sitemap XML content |
| 1419 |
* |
| 1420 |
* @since 1.0.0 |
| 1421 |
* @param string $sitemap_xml Sitemap XML content |
| 1422 |
* @return int Number of URLs |
| 1423 |
*/ |
| 1424 |
private function count_urls_in_xml(string $sitemap_xml): int { |
| 1425 |
return substr_count($sitemap_xml, '<url>'); |
| 1426 |
} |
| 1427 |
|
| 1428 |
/** |
| 1429 |
* Ping search engines about sitemap update |
| 1430 |
* |
| 1431 |
* @since 1.0.0 |
| 1432 |
* @return void |
| 1433 |
*/ |
| 1434 |
private function ping_search_engines(): void { |
| 1435 |
$sitemap_url = home_url('/sitemap.xml'); |
| 1436 |
|
| 1437 |
$ping_urls = [ |
| 1438 |
'google' => 'https://www.google.com/ping?sitemap=' . urlencode($sitemap_url), |
| 1439 |
'bing' => 'https://www.bing.com/ping?sitemap=' . urlencode($sitemap_url) |
| 1440 |
]; |
| 1441 |
|
| 1442 |
foreach ($ping_urls as $engine => $ping_url) { |
| 1443 |
// Use wp_remote_get with timeout to prevent hanging |
| 1444 |
$response = wp_remote_get($ping_url, [ |
| 1445 |
'timeout' => 10, |
| 1446 |
'blocking' => false // Non-blocking to prevent delays |
| 1447 |
]); |
| 1448 |
|
| 1449 |
if (is_wp_error($response)) { |
| 1450 |
// Search engine ping failed - error details available in WP_Error response |
| 1451 |
} |
| 1452 |
} |
| 1453 |
} |
| 1454 |
|
| 1455 |
/** |
| 1456 |
* Validate and sanitize exclude posts input |
| 1457 |
* |
| 1458 |
* @since 1.0.0 |
| 1459 |
* @param string $exclude_posts Comma-separated post IDs |
| 1460 |
* @return array Validated post IDs |
| 1461 |
* @throws InvalidArgumentException If validation fails |
| 1462 |
*/ |
| 1463 |
private function validate_exclude_posts(string $exclude_posts): array { |
| 1464 |
if (empty($exclude_posts)) { |
| 1465 |
return []; |
| 1466 |
} |
| 1467 |
|
| 1468 |
// Limit input length to prevent DoS |
| 1469 |
if (strlen($exclude_posts) > 1000) { |
| 1470 |
throw new InvalidArgumentException('Exclude posts list too long (max 1000 characters)'); |
| 1471 |
} |
| 1472 |
|
| 1473 |
// Validate comma-separated integers only |
| 1474 |
if (!preg_match('/^[\d,\s]+$/', $exclude_posts)) { |
| 1475 |
throw new InvalidArgumentException('Exclude posts must contain only numbers and commas'); |
| 1476 |
} |
| 1477 |
|
| 1478 |
$ids = array_map('intval', array_filter(explode(',', $exclude_posts))); |
| 1479 |
|
| 1480 |
// Limit number of exclusions to prevent performance issues |
| 1481 |
if (count($ids) > 100) { |
| 1482 |
throw new InvalidArgumentException('Too many posts to exclude (max 100)'); |
| 1483 |
} |
| 1484 |
|
| 1485 |
return array_filter($ids, function($id) { |
| 1486 |
return $id > 0; // Only positive integers |
| 1487 |
}); |
| 1488 |
} |
| 1489 |
|
| 1490 |
/** |
| 1491 |
* Validate and sanitize exclude terms input |
| 1492 |
* |
| 1493 |
* @since 1.0.0 |
| 1494 |
* @param string $exclude_terms Comma-separated term IDs |
| 1495 |
* @return array Validated term IDs |
| 1496 |
* @throws InvalidArgumentException If validation fails |
| 1497 |
*/ |
| 1498 |
private function validate_exclude_terms(string $exclude_terms): array { |
| 1499 |
if (empty($exclude_terms)) { |
| 1500 |
return []; |
| 1501 |
} |
| 1502 |
|
| 1503 |
// Limit input length to prevent DoS |
| 1504 |
if (strlen($exclude_terms) > 1000) { |
| 1505 |
throw new InvalidArgumentException('Exclude terms list too long (max 1000 characters)'); |
| 1506 |
} |
| 1507 |
|
| 1508 |
// Validate comma-separated integers only |
| 1509 |
if (!preg_match('/^[\d,\s]+$/', $exclude_terms)) { |
| 1510 |
throw new InvalidArgumentException('Exclude terms must contain only numbers and commas'); |
| 1511 |
} |
| 1512 |
|
| 1513 |
$ids = array_map('intval', array_filter(explode(',', $exclude_terms))); |
| 1514 |
|
| 1515 |
// Limit number of exclusions to prevent performance issues |
| 1516 |
if (count($ids) > 100) { |
| 1517 |
throw new InvalidArgumentException('Too many terms to exclude (max 100)'); |
| 1518 |
} |
| 1519 |
|
| 1520 |
return array_filter($ids, function($id) { |
| 1521 |
return $id > 0; // Only positive integers |
| 1522 |
}); |
| 1523 |
} |
| 1524 |
|
| 1525 |
/** |
| 1526 |
* Validate and sanitize custom URL pattern |
| 1527 |
* |
| 1528 |
* @since 1.0.0 |
| 1529 |
* @param string $pattern Custom URL pattern |
| 1530 |
* @return string Validated and sanitized pattern |
| 1531 |
* @throws InvalidArgumentException If validation fails |
| 1532 |
*/ |
| 1533 |
private function validate_custom_url_pattern(string $pattern): string { |
| 1534 |
if (empty($pattern)) { |
| 1535 |
return 'sitemap-{type}.xml'; |
| 1536 |
} |
| 1537 |
|
| 1538 |
// Remove any HTML/script tags to prevent XSS |
| 1539 |
$pattern = wp_strip_all_tags($pattern); |
| 1540 |
|
| 1541 |
// Limit pattern length |
| 1542 |
if (strlen($pattern) > 100) { |
| 1543 |
throw new InvalidArgumentException('URL pattern too long (max 100 characters)'); |
| 1544 |
} |
| 1545 |
|
| 1546 |
// Validate pattern format - only allow safe characters |
| 1547 |
if (!preg_match('/^[a-zA-Z0-9\-_{}\.]+$/', $pattern)) { |
| 1548 |
throw new InvalidArgumentException('URL pattern contains invalid characters. Only letters, numbers, hyphens, underscores, dots, and {type} are allowed'); |
| 1549 |
} |
| 1550 |
|
| 1551 |
// Ensure it contains {type} placeholder |
| 1552 |
if (strpos($pattern, '{type}') === false) { |
| 1553 |
throw new InvalidArgumentException('URL pattern must contain {type} placeholder'); |
| 1554 |
} |
| 1555 |
|
| 1556 |
// Ensure it ends with .xml |
| 1557 |
if (!str_ends_with($pattern, '.xml')) { |
| 1558 |
$pattern .= '.xml'; |
| 1559 |
} |
| 1560 |
|
| 1561 |
return sanitize_file_name($pattern); |
| 1562 |
} |
| 1563 |
|
| 1564 |
/** |
| 1565 |
* Validate sitemap URL |
| 1566 |
* |
| 1567 |
* @since 1.0.0 |
| 1568 |
* @param string $url Sitemap URL |
| 1569 |
* @return string Validated and sanitized URL |
| 1570 |
* @throws InvalidArgumentException If validation fails |
| 1571 |
*/ |
| 1572 |
private function validate_sitemap_url(string $url): string { |
| 1573 |
if (empty($url)) { |
| 1574 |
throw new InvalidArgumentException('Sitemap URL cannot be empty'); |
| 1575 |
} |
| 1576 |
|
| 1577 |
// Remove leading/trailing whitespace |
| 1578 |
$url = trim($url); |
| 1579 |
|
| 1580 |
// Limit URL length |
| 1581 |
if (strlen($url) > 200) { |
| 1582 |
throw new InvalidArgumentException('Sitemap URL too long (max 200 characters)'); |
| 1583 |
} |
| 1584 |
|
| 1585 |
// Ensure it starts with / |
| 1586 |
if (!str_starts_with($url, '/')) { |
| 1587 |
$url = '/' . $url; |
| 1588 |
} |
| 1589 |
|
| 1590 |
// Validate URL path format |
| 1591 |
if (!preg_match('/^\/[a-zA-Z0-9\-_\/\.]+\.xml$/', $url)) { |
| 1592 |
throw new InvalidArgumentException('Invalid sitemap URL format. Must be a valid path ending with .xml'); |
| 1593 |
} |
| 1594 |
|
| 1595 |
// Prevent directory traversal |
| 1596 |
if (strpos($url, '..') !== false) { |
| 1597 |
throw new InvalidArgumentException('Directory traversal not allowed in sitemap URL'); |
| 1598 |
} |
| 1599 |
|
| 1600 |
// Prevent multiple slashes |
| 1601 |
$url = preg_replace('/\/+/', '/', $url); |
| 1602 |
|
| 1603 |
return sanitize_url($url); |
| 1604 |
} |
| 1605 |
|
| 1606 |
/** |
| 1607 |
* Validate links per sitemap setting |
| 1608 |
* |
| 1609 |
* @since 1.0.0 |
| 1610 |
* @param mixed $links_per_sitemap Links per sitemap value |
| 1611 |
* @return int Validated links per sitemap |
| 1612 |
* @throws InvalidArgumentException If validation fails |
| 1613 |
*/ |
| 1614 |
private function validate_links_per_sitemap($links_per_sitemap): int { |
| 1615 |
$links = intval($links_per_sitemap); |
| 1616 |
|
| 1617 |
if ($links < 1) { |
| 1618 |
throw new InvalidArgumentException('Links per sitemap must be at least 1'); |
| 1619 |
} |
| 1620 |
|
| 1621 |
if ($links > 50000) { |
| 1622 |
throw new InvalidArgumentException('Links per sitemap cannot exceed 50,000'); |
| 1623 |
} |
| 1624 |
|
| 1625 |
return $links; |
| 1626 |
} |
| 1627 |
|
| 1628 |
/** |
| 1629 |
* Validate sitemap filename for security |
| 1630 |
* |
| 1631 |
* @since 1.0.0 |
| 1632 |
* @param string $filename Filename to validate |
| 1633 |
* @return string Validated and sanitized filename |
| 1634 |
* @throws InvalidArgumentException If validation fails |
| 1635 |
*/ |
| 1636 |
private function validate_sitemap_filename(string $filename): string { |
| 1637 |
if (empty($filename)) { |
| 1638 |
return 'sitemap.xml'; |
| 1639 |
} |
| 1640 |
|
| 1641 |
// Remove any path components to prevent directory traversal |
| 1642 |
$filename = basename($filename); |
| 1643 |
|
| 1644 |
// Limit filename length |
| 1645 |
if (strlen($filename) > 100) { |
| 1646 |
throw new InvalidArgumentException('Filename too long (max 100 characters)'); |
| 1647 |
} |
| 1648 |
|
| 1649 |
// Validate filename format - only allow safe characters |
| 1650 |
if (!preg_match('/^[a-zA-Z0-9\-_\.]+$/', $filename)) { |
| 1651 |
throw new InvalidArgumentException('Filename contains invalid characters. Only letters, numbers, hyphens, underscores, and dots are allowed'); |
| 1652 |
} |
| 1653 |
|
| 1654 |
// Prevent directory traversal attempts |
| 1655 |
if (strpos($filename, '..') !== false) { |
| 1656 |
throw new InvalidArgumentException('Directory traversal not allowed in filename'); |
| 1657 |
} |
| 1658 |
|
| 1659 |
// Ensure it ends with .xml |
| 1660 |
if (!str_ends_with($filename, '.xml')) { |
| 1661 |
$filename .= '.xml'; |
| 1662 |
} |
| 1663 |
|
| 1664 |
// Additional sanitization |
| 1665 |
$filename = sanitize_file_name($filename); |
| 1666 |
|
| 1667 |
// Final security check - ensure it's still a valid XML filename |
| 1668 |
if (!preg_match('/^[a-zA-Z0-9\-_]+\.xml$/', $filename)) { |
| 1669 |
throw new InvalidArgumentException('Invalid XML filename after sanitization'); |
| 1670 |
} |
| 1671 |
|
| 1672 |
return $filename; |
| 1673 |
} |
| 1674 |
} |
| 1675 |
|
