| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* Robots.txt Management Class |
| 5 |
* |
| 6 |
* @package Search Atlas SEO |
| 7 |
* @copyright Copyright (C) 2021-2025, Search Atlas Group - [email protected] |
| 8 |
* @since 2.5.6 |
| 9 |
*/ |
| 10 |
|
| 11 |
// If this file is called directly, abort. |
| 12 |
if (!defined('WPINC')) { |
| 13 |
die; |
| 14 |
} |
| 15 |
|
| 16 |
class Metasync_Robots_Txt |
| 17 |
{ |
| 18 |
/** |
| 19 |
* Instance of this class |
| 20 |
* |
| 21 |
* @var Metasync_Robots_Txt |
| 22 |
*/ |
| 23 |
private static $instance = null; |
| 24 |
|
| 25 |
/** |
| 26 |
* Database instance |
| 27 |
* |
| 28 |
* @var Metasync_Robots_Txt_Database |
| 29 |
*/ |
| 30 |
private $database; |
| 31 |
|
| 32 |
/** |
| 33 |
* Robots.txt file path |
| 34 |
* |
| 35 |
* @var string |
| 36 |
*/ |
| 37 |
private $robots_file_path; |
| 38 |
|
| 39 |
/** |
| 40 |
* Get singleton instance |
| 41 |
*/ |
| 42 |
public static function get_instance() |
| 43 |
{ |
| 44 |
if (null === self::$instance) { |
| 45 |
self::$instance = new self(); |
| 46 |
} |
| 47 |
return self::$instance; |
| 48 |
} |
| 49 |
|
| 50 |
/** |
| 51 |
* Constructor |
| 52 |
*/ |
| 53 |
private function __construct() |
| 54 |
{ |
| 55 |
$this->robots_file_path = ABSPATH . 'robots.txt'; |
| 56 |
|
| 57 |
// Load database class |
| 58 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-robots-txt-database.php'; |
| 59 |
$this->database = Metasync_Robots_Txt_Database::get_instance(); |
| 60 |
|
| 61 |
// Register hook to serve virtual robots.txt |
| 62 |
add_filter('robots_txt', array($this, 'serve_virtual_robots_txt'), 10, 2); |
| 63 |
} |
| 64 |
|
| 65 |
/** |
| 66 |
* Serve virtual robots.txt content via WordPress filter |
| 67 |
* |
| 68 |
* @param string $output The robots.txt output |
| 69 |
* @param bool $public Whether the site is public |
| 70 |
* @return string The robots.txt content |
| 71 |
*/ |
| 72 |
public function serve_virtual_robots_txt($output, $public) |
| 73 |
{ |
| 74 |
// Only serve virtual content if physical file doesn't exist |
| 75 |
if (!file_exists($this->robots_file_path)) { |
| 76 |
// Check if we have virtual content |
| 77 |
if ($this->database->is_virtual_mode()) { |
| 78 |
$virtual_content = $this->database->get_virtual_content(); |
| 79 |
if (false !== $virtual_content) { |
| 80 |
return $virtual_content; |
| 81 |
} |
| 82 |
} |
| 83 |
} |
| 84 |
|
| 85 |
// Return default output (WordPress will handle it) |
| 86 |
return $output; |
| 87 |
} |
| 88 |
|
| 89 |
/** |
| 90 |
* Read robots.txt file |
| 91 |
* |
| 92 |
* @return string|WP_Error File contents or error |
| 93 |
*/ |
| 94 |
public function read_robots_file() |
| 95 |
{ |
| 96 |
// Check virtual content first if virtual mode is active |
| 97 |
if ($this->database->is_virtual_mode()) { |
| 98 |
$virtual_content = $this->database->get_virtual_content(); |
| 99 |
if (false !== $virtual_content) { |
| 100 |
return $virtual_content; |
| 101 |
} |
| 102 |
} |
| 103 |
|
| 104 |
// Check if physical file exists |
| 105 |
if (!file_exists($this->robots_file_path)) { |
| 106 |
// Check if we have virtual content as fallback |
| 107 |
$virtual_content = $this->database->get_virtual_content(); |
| 108 |
if (false !== $virtual_content) { |
| 109 |
return $virtual_content; |
| 110 |
} |
| 111 |
// Return default robots.txt content |
| 112 |
return $this->get_default_robots_content(); |
| 113 |
} |
| 114 |
|
| 115 |
// Try WP_Filesystem first |
| 116 |
if (function_exists('WP_Filesystem')) { |
| 117 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 118 |
WP_Filesystem(); |
| 119 |
global $wp_filesystem; |
| 120 |
|
| 121 |
if ($wp_filesystem && $wp_filesystem->exists($this->robots_file_path)) { |
| 122 |
$content = $wp_filesystem->get_contents($this->robots_file_path); |
| 123 |
if (false !== $content) { |
| 124 |
return $content; |
| 125 |
} |
| 126 |
} |
| 127 |
} |
| 128 |
|
| 129 |
// Fallback to native PHP file operations |
| 130 |
$content = @file_get_contents($this->robots_file_path); |
| 131 |
|
| 132 |
if (false === $content) { |
| 133 |
// Check virtual content as last resort |
| 134 |
$virtual_content = $this->database->get_virtual_content(); |
| 135 |
if (false !== $virtual_content) { |
| 136 |
return $virtual_content; |
| 137 |
} |
| 138 |
return new WP_Error('read_error', esc_html__('Could not read robots.txt file.', 'metasync')); |
| 139 |
} |
| 140 |
|
| 141 |
return $content; |
| 142 |
} |
| 143 |
|
| 144 |
/** |
| 145 |
* Write robots.txt file |
| 146 |
* |
| 147 |
* @param string $content Content to write |
| 148 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 149 |
*/ |
| 150 |
public function write_robots_file($content) |
| 151 |
{ |
| 152 |
// Create backup before saving |
| 153 |
$current_content = $this->read_robots_file(); |
| 154 |
if (!is_wp_error($current_content) && !empty(trim($current_content))) { |
| 155 |
$this->database->create_backup($current_content); |
| 156 |
} |
| 157 |
|
| 158 |
// Try WP_Filesystem first |
| 159 |
if (function_exists('WP_Filesystem')) { |
| 160 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 161 |
WP_Filesystem(); |
| 162 |
global $wp_filesystem; |
| 163 |
|
| 164 |
if ($wp_filesystem) { |
| 165 |
$result = $wp_filesystem->put_contents( |
| 166 |
$this->robots_file_path, |
| 167 |
$content, |
| 168 |
FS_CHMOD_FILE |
| 169 |
); |
| 170 |
|
| 171 |
if ($result) { |
| 172 |
// Clear virtual mode if file was successfully written |
| 173 |
$this->database->clear_virtual_content(); |
| 174 |
return true; |
| 175 |
} |
| 176 |
} |
| 177 |
} |
| 178 |
|
| 179 |
// Fallback to native PHP file operations |
| 180 |
$result = @file_put_contents($this->robots_file_path, $content); |
| 181 |
|
| 182 |
if (false !== $result) { |
| 183 |
// Set secure permissions: 0644 (rw-r--r--) |
| 184 |
// Owner can read/write, group and others can only read |
| 185 |
// This is the WordPress standard for files |
| 186 |
@chmod($this->robots_file_path, 0644); |
| 187 |
// Clear virtual mode if file was successfully written |
| 188 |
$this->database->clear_virtual_content(); |
| 189 |
return true; |
| 190 |
} |
| 191 |
|
| 192 |
// File write failed - fallback to virtual file storage |
| 193 |
$virtual_stored = $this->database->store_virtual_content($content); |
| 194 |
if ($virtual_stored) { |
| 195 |
$this->database->set_virtual_mode(true); |
| 196 |
// Return success but indicate it's virtual |
| 197 |
return true; |
| 198 |
} |
| 199 |
|
| 200 |
return new WP_Error('write_error', esc_html__('Could not write to robots.txt file. Please check file permissions.', 'metasync')); |
| 201 |
} |
| 202 |
|
| 203 |
/** |
| 204 |
* Validate robots.txt content |
| 205 |
* |
| 206 |
* @param string $content Content to validate |
| 207 |
* @return array Array of validation results |
| 208 |
*/ |
| 209 |
public function validate_content($content) |
| 210 |
{ |
| 211 |
$warnings = array(); |
| 212 |
$errors = array(); |
| 213 |
|
| 214 |
// Normalize content for easier checking |
| 215 |
$normalized_content = preg_replace('/\s+/', ' ', strtolower($content)); |
| 216 |
|
| 217 |
// Check for complete site disallow (blocking everything) |
| 218 |
if (preg_match('/user-agent:\s*\*.*?disallow:\s*\/(\s|$)/i', $normalized_content)) { |
| 219 |
// Check if it's ONLY blocking / without any other paths |
| 220 |
if (!preg_match('/disallow:\s*\/\w+/i', $content)) { |
| 221 |
$warnings[] = esc_html__('Warning: You are blocking all crawlers from your entire site (Disallow: /). This will prevent search engines from indexing your content.', 'metasync'); |
| 222 |
} |
| 223 |
} |
| 224 |
|
| 225 |
// Check for wp-admin disallow without admin-ajax.php exception |
| 226 |
$has_wp_admin_disallow = preg_match('/Disallow:\s*\/wp-admin\/?$/im', $content); |
| 227 |
$has_admin_ajax_allow = preg_match('/Allow:\s*\/wp-admin\/admin-ajax\.php/i', $content); |
| 228 |
|
| 229 |
if ($has_wp_admin_disallow && !$has_admin_ajax_allow) { |
| 230 |
$warnings[] = esc_html__('Warning: Blocking /wp-admin/ without allowing /wp-admin/admin-ajax.php may interfere with AJAX functionality on your site.', 'metasync'); |
| 231 |
} |
| 232 |
|
| 233 |
// Check for basic syntax errors |
| 234 |
$lines = explode("\n", $content); |
| 235 |
$user_agent_found = false; |
| 236 |
|
| 237 |
foreach ($lines as $line_num => $line) { |
| 238 |
$line = trim($line); |
| 239 |
|
| 240 |
// Skip empty lines and comments |
| 241 |
if (empty($line) || strpos($line, '#') === 0) { |
| 242 |
continue; |
| 243 |
} |
| 244 |
|
| 245 |
// Check for valid directives |
| 246 |
if (!preg_match('/^(User-agent|Disallow|Allow|Sitemap|Crawl-delay):/i', $line)) { |
| 247 |
$errors[] = sprintf( |
| 248 |
esc_html__('Line %d: Invalid directive. Valid directives are: User-agent, Disallow, Allow, Sitemap, Crawl-delay', 'metasync'), |
| 249 |
$line_num + 1 |
| 250 |
); |
| 251 |
} |
| 252 |
|
| 253 |
// Track if User-agent is present |
| 254 |
if (preg_match('/^User-agent:/i', $line)) { |
| 255 |
$user_agent_found = true; |
| 256 |
} |
| 257 |
} |
| 258 |
|
| 259 |
// Check if content has at least one User-agent |
| 260 |
if (!empty(trim($content)) && !$user_agent_found) { |
| 261 |
$errors[] = esc_html__('robots.txt must contain at least one User-agent directive.', 'metasync'); |
| 262 |
} |
| 263 |
|
| 264 |
return array( |
| 265 |
'valid' => empty($errors), |
| 266 |
'warnings' => $warnings, |
| 267 |
'errors' => $errors |
| 268 |
); |
| 269 |
} |
| 270 |
|
| 271 |
/** |
| 272 |
* Get default robots.txt content |
| 273 |
* |
| 274 |
* @return string Default content |
| 275 |
*/ |
| 276 |
public function get_default_robots_content() |
| 277 |
{ |
| 278 |
$site_url = get_site_url(); |
| 279 |
|
| 280 |
return "User-agent: *\n" . |
| 281 |
"Disallow: /wp-admin/\n" . |
| 282 |
"Allow: /wp-admin/admin-ajax.php\n" . |
| 283 |
"Disallow: /wp-includes/\n\n" . |
| 284 |
"Sitemap: {$site_url}/sitemap_index.xml"; |
| 285 |
} |
| 286 |
|
| 287 |
/** |
| 288 |
* Get backup history |
| 289 |
* |
| 290 |
* @param int $limit Number of backups to retrieve |
| 291 |
* @return array Array of backups |
| 292 |
*/ |
| 293 |
public function get_backup_history($limit = 10) |
| 294 |
{ |
| 295 |
return $this->database->get_backups($limit); |
| 296 |
} |
| 297 |
|
| 298 |
/** |
| 299 |
* Restore from backup |
| 300 |
* |
| 301 |
* @param int $backup_id Backup ID |
| 302 |
* @return bool|WP_Error True on success, WP_Error on failure |
| 303 |
*/ |
| 304 |
public function restore_backup($backup_id) |
| 305 |
{ |
| 306 |
$backup = $this->database->get_backup($backup_id); |
| 307 |
|
| 308 |
if (!$backup) { |
| 309 |
return new WP_Error('backup_not_found', esc_html__('Backup not found.', 'metasync')); |
| 310 |
} |
| 311 |
|
| 312 |
return $this->write_robots_file($backup['content']); |
| 313 |
} |
| 314 |
|
| 315 |
/** |
| 316 |
* Delete a backup |
| 317 |
* |
| 318 |
* @param int $backup_id Backup ID |
| 319 |
* @return bool True on success, false on failure |
| 320 |
*/ |
| 321 |
public function delete_backup($backup_id) |
| 322 |
{ |
| 323 |
return $this->database->delete_backup($backup_id); |
| 324 |
} |
| 325 |
|
| 326 |
/** |
| 327 |
* Check if robots.txt file exists (physical or virtual) |
| 328 |
* |
| 329 |
* @return bool True if exists (physical or virtual), false otherwise |
| 330 |
*/ |
| 331 |
public function file_exists() |
| 332 |
{ |
| 333 |
// Check physical file first |
| 334 |
if (file_exists($this->robots_file_path)) { |
| 335 |
return true; |
| 336 |
} |
| 337 |
|
| 338 |
// Check virtual content |
| 339 |
if ($this->database->is_virtual_mode()) { |
| 340 |
$virtual_content = $this->database->get_virtual_content(); |
| 341 |
if (false !== $virtual_content) { |
| 342 |
return true; |
| 343 |
} |
| 344 |
} |
| 345 |
|
| 346 |
return false; |
| 347 |
} |
| 348 |
|
| 349 |
/** |
| 350 |
* Get file permissions |
| 351 |
* |
| 352 |
* @return string|bool File permissions or false |
| 353 |
*/ |
| 354 |
public function get_file_permissions() |
| 355 |
{ |
| 356 |
if (!function_exists('WP_Filesystem')) { |
| 357 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 358 |
} |
| 359 |
|
| 360 |
WP_Filesystem(); |
| 361 |
global $wp_filesystem; |
| 362 |
|
| 363 |
if (!$wp_filesystem || !$wp_filesystem->exists($this->robots_file_path)) { |
| 364 |
return false; |
| 365 |
} |
| 366 |
|
| 367 |
return substr(sprintf('%o', fileperms($this->robots_file_path)), -4); |
| 368 |
} |
| 369 |
|
| 370 |
/** |
| 371 |
* Check if file is writable |
| 372 |
* |
| 373 |
* @return bool True if writable, false otherwise |
| 374 |
*/ |
| 375 |
public function is_writable() |
| 376 |
{ |
| 377 |
// If file exists, check if it's writable |
| 378 |
if (file_exists($this->robots_file_path)) { |
| 379 |
return is_writable($this->robots_file_path); |
| 380 |
} |
| 381 |
|
| 382 |
// If file doesn't exist, check if parent directory is writable |
| 383 |
// If not writable, virtual mode will be used |
| 384 |
return is_writable(ABSPATH); |
| 385 |
} |
| 386 |
|
| 387 |
/** |
| 388 |
* Check if virtual mode is active |
| 389 |
* |
| 390 |
* @return bool True if virtual mode is active |
| 391 |
*/ |
| 392 |
public function is_virtual_mode() |
| 393 |
{ |
| 394 |
return $this->database->is_virtual_mode(); |
| 395 |
} |
| 396 |
|
| 397 |
/** |
| 398 |
* Update or add sitemap URL in robots.txt |
| 399 |
* |
| 400 |
* @param string $sitemap_url The sitemap URL to add/update |
| 401 |
* @return array Result with 'success' boolean and 'action' string ('added', 'updated', 'unchanged', 'created', or 'error') |
| 402 |
*/ |
| 403 |
public function update_sitemap_url($sitemap_url) |
| 404 |
{ |
| 405 |
// Check if we can write to robots.txt |
| 406 |
if (!$this->is_writable()) { |
| 407 |
return [ |
| 408 |
'success' => false, |
| 409 |
'action' => 'error', |
| 410 |
'message' => esc_html__('Cannot write to robots.txt. Please check file permissions.', 'metasync') |
| 411 |
]; |
| 412 |
} |
| 413 |
|
| 414 |
// Check if file exists |
| 415 |
$file_exists = $this->file_exists(); |
| 416 |
|
| 417 |
// Read current content |
| 418 |
$current_content = $this->read_robots_file(); |
| 419 |
|
| 420 |
if (is_wp_error($current_content)) { |
| 421 |
// If error reading, use default content |
| 422 |
$current_content = $this->get_default_robots_content(); |
| 423 |
} |
| 424 |
|
| 425 |
// Normalize the sitemap URL |
| 426 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 427 |
$sitemap_line = "Sitemap: {$sitemap_url}"; |
| 428 |
|
| 429 |
// Check if there's already a Sitemap line |
| 430 |
$has_sitemap = preg_match('/^Sitemap:\s*.+$/im', $current_content); |
| 431 |
|
| 432 |
if ($has_sitemap) { |
| 433 |
// Check if the sitemap URL is already correct |
| 434 |
if (preg_match('/^Sitemap:\s*' . preg_quote($sitemap_url, '/') . '\s*$/im', $current_content)) { |
| 435 |
// If file doesn't exist, we still need to create it even if content matches |
| 436 |
if (!$file_exists) { |
| 437 |
$result = $this->write_robots_file($current_content); |
| 438 |
if (is_wp_error($result)) { |
| 439 |
return [ |
| 440 |
'success' => false, |
| 441 |
'action' => 'error', |
| 442 |
'message' => $result->get_error_message() |
| 443 |
]; |
| 444 |
} |
| 445 |
return [ |
| 446 |
'success' => true, |
| 447 |
'action' => 'created', |
| 448 |
'message' => esc_html__('robots.txt file has been created with sitemap URL.', 'metasync') |
| 449 |
]; |
| 450 |
} |
| 451 |
return [ |
| 452 |
'success' => true, |
| 453 |
'action' => 'unchanged', |
| 454 |
'message' => esc_html__('Sitemap URL already exists in robots.txt.', 'metasync') |
| 455 |
]; |
| 456 |
} |
| 457 |
|
| 458 |
// Update existing Sitemap line(s) - replace all with the new one |
| 459 |
$new_content = preg_replace('/^Sitemap:\s*.+$/im', $sitemap_line, $current_content); |
| 460 |
|
| 461 |
// Remove duplicate sitemap lines (keep only the first one) |
| 462 |
$lines = explode("\n", $new_content); |
| 463 |
$sitemap_found = false; |
| 464 |
$filtered_lines = []; |
| 465 |
foreach ($lines as $line) { |
| 466 |
if (preg_match('/^Sitemap:/i', trim($line))) { |
| 467 |
if (!$sitemap_found) { |
| 468 |
$filtered_lines[] = $sitemap_line; |
| 469 |
$sitemap_found = true; |
| 470 |
} |
| 471 |
// Skip duplicate sitemap lines |
| 472 |
} else { |
| 473 |
$filtered_lines[] = $line; |
| 474 |
} |
| 475 |
} |
| 476 |
$new_content = implode("\n", $filtered_lines); |
| 477 |
$action = 'updated'; |
| 478 |
} else { |
| 479 |
// Add sitemap line at the end |
| 480 |
$new_content = rtrim($current_content) . "\n\n" . $sitemap_line; |
| 481 |
$action = 'added'; |
| 482 |
} |
| 483 |
|
| 484 |
// Write the updated content |
| 485 |
$result = $this->write_robots_file($new_content); |
| 486 |
|
| 487 |
if (is_wp_error($result)) { |
| 488 |
return [ |
| 489 |
'success' => false, |
| 490 |
'action' => 'error', |
| 491 |
'message' => $result->get_error_message() |
| 492 |
]; |
| 493 |
} |
| 494 |
|
| 495 |
$messages = [ |
| 496 |
'added' => esc_html__('Sitemap URL has been added to robots.txt.', 'metasync'), |
| 497 |
'updated' => esc_html__('Sitemap URL has been updated in robots.txt.', 'metasync'), |
| 498 |
'created' => esc_html__('robots.txt file has been created with sitemap URL.', 'metasync'), |
| 499 |
]; |
| 500 |
|
| 501 |
return [ |
| 502 |
'success' => true, |
| 503 |
'action' => $action, |
| 504 |
'message' => $messages[$action] |
| 505 |
]; |
| 506 |
} |
| 507 |
|
| 508 |
/** |
| 509 |
* Check if robots.txt contains the sitemap URL |
| 510 |
* |
| 511 |
* @param string $sitemap_url The sitemap URL to check for |
| 512 |
* @return bool True if the sitemap URL exists in robots.txt |
| 513 |
*/ |
| 514 |
public function has_sitemap_url($sitemap_url) |
| 515 |
{ |
| 516 |
$content = $this->read_robots_file(); |
| 517 |
|
| 518 |
if (is_wp_error($content)) { |
| 519 |
return false; |
| 520 |
} |
| 521 |
|
| 522 |
return (bool) preg_match('/^Sitemap:\s*' . preg_quote($sitemap_url, '/') . '\s*$/im', $content); |
| 523 |
} |
| 524 |
|
| 525 |
/** |
| 526 |
* Add a sitemap URL to robots.txt without replacing existing ones. |
| 527 |
* |
| 528 |
* @param string $sitemap_url The sitemap URL to add. |
| 529 |
* @return array Result with 'success' boolean and 'action' string ('added' or 'unchanged'). |
| 530 |
*/ |
| 531 |
public function add_sitemap_url($sitemap_url) |
| 532 |
{ |
| 533 |
$content = $this->read_robots_file(); |
| 534 |
|
| 535 |
if (is_wp_error($content)) { |
| 536 |
return [ |
| 537 |
'success' => false, |
| 538 |
'action' => 'error', |
| 539 |
'message' => $content->get_error_message(), |
| 540 |
]; |
| 541 |
} |
| 542 |
|
| 543 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 544 |
|
| 545 |
// Check if the specific sitemap line already exists |
| 546 |
if (preg_match('/^Sitemap:\s*' . preg_quote($sitemap_url, '/') . '\s*$/im', $content)) { |
| 547 |
return [ |
| 548 |
'success' => true, |
| 549 |
'action' => 'unchanged', |
| 550 |
'message' => esc_html__('Sitemap URL already exists in robots.txt.', 'metasync'), |
| 551 |
]; |
| 552 |
} |
| 553 |
|
| 554 |
// Append the new Sitemap line |
| 555 |
$content = rtrim($content) . "\nSitemap: " . $sitemap_url; |
| 556 |
|
| 557 |
$result = $this->write_robots_file($content); |
| 558 |
|
| 559 |
if (is_wp_error($result)) { |
| 560 |
return [ |
| 561 |
'success' => false, |
| 562 |
'action' => 'error', |
| 563 |
'message' => $result->get_error_message(), |
| 564 |
]; |
| 565 |
} |
| 566 |
|
| 567 |
return [ |
| 568 |
'success' => true, |
| 569 |
'action' => 'added', |
| 570 |
'message' => esc_html__('Sitemap URL has been added to robots.txt.', 'metasync'), |
| 571 |
]; |
| 572 |
} |
| 573 |
|
| 574 |
/** |
| 575 |
* Remove a specific sitemap URL from robots.txt. |
| 576 |
* |
| 577 |
* @param string $sitemap_url The sitemap URL to remove. |
| 578 |
* @return array Result with 'success' boolean and 'action' string ('removed' or 'not_found'). |
| 579 |
*/ |
| 580 |
public function remove_sitemap_url($sitemap_url) |
| 581 |
{ |
| 582 |
$content = $this->read_robots_file(); |
| 583 |
|
| 584 |
if (is_wp_error($content)) { |
| 585 |
return [ |
| 586 |
'success' => false, |
| 587 |
'action' => 'error', |
| 588 |
'message' => $content->get_error_message(), |
| 589 |
]; |
| 590 |
} |
| 591 |
|
| 592 |
$sitemap_url = esc_url_raw($sitemap_url); |
| 593 |
$pattern = '/^Sitemap:\s*' . preg_quote($sitemap_url, '/') . '\s*$/im'; |
| 594 |
|
| 595 |
if (!preg_match($pattern, $content)) { |
| 596 |
return [ |
| 597 |
'success' => true, |
| 598 |
'action' => 'not_found', |
| 599 |
'message' => esc_html__('Sitemap URL not found in robots.txt.', 'metasync'), |
| 600 |
]; |
| 601 |
} |
| 602 |
|
| 603 |
$new_content = preg_replace($pattern, '', $content); |
| 604 |
// Clean up extra blank lines |
| 605 |
$new_content = preg_replace("/\n{3,}/", "\n\n", $new_content); |
| 606 |
$new_content = rtrim($new_content) . "\n"; |
| 607 |
|
| 608 |
$result = $this->write_robots_file($new_content); |
| 609 |
|
| 610 |
if (is_wp_error($result)) { |
| 611 |
return [ |
| 612 |
'success' => false, |
| 613 |
'action' => 'error', |
| 614 |
'message' => $result->get_error_message(), |
| 615 |
]; |
| 616 |
} |
| 617 |
|
| 618 |
return [ |
| 619 |
'success' => true, |
| 620 |
'action' => 'removed', |
| 621 |
'message' => esc_html__('Sitemap URL has been removed from robots.txt.', 'metasync'), |
| 622 |
]; |
| 623 |
} |
| 624 |
|
| 625 |
/** |
| 626 |
* Add a reference to /llms.txt in robots.txt. |
| 627 |
* |
| 628 |
* robots.txt has no official directive for LLMs.txt, so the reference is |
| 629 |
* written as a comment (ignored by crawlers but visible to humans and |
| 630 |
* tools that scan robots.txt for related assets). |
| 631 |
* |
| 632 |
* @param string $llms_url Absolute URL to the LLMs.txt file. |
| 633 |
* @return array Result with 'success' and 'action' keys. |
| 634 |
*/ |
| 635 |
public function add_llms_txt_url($llms_url) |
| 636 |
{ |
| 637 |
$content = $this->read_robots_file(); |
| 638 |
|
| 639 |
if (is_wp_error($content)) { |
| 640 |
return [ |
| 641 |
'success' => false, |
| 642 |
'action' => 'error', |
| 643 |
'message' => $content->get_error_message(), |
| 644 |
]; |
| 645 |
} |
| 646 |
|
| 647 |
$llms_url = esc_url_raw($llms_url); |
| 648 |
$line = '# LLMs.txt: ' . $llms_url; |
| 649 |
|
| 650 |
if (false !== strpos($content, $line)) { |
| 651 |
return [ |
| 652 |
'success' => true, |
| 653 |
'action' => 'unchanged', |
| 654 |
'message' => esc_html__('LLMs.txt reference already exists in robots.txt.', 'metasync'), |
| 655 |
]; |
| 656 |
} |
| 657 |
|
| 658 |
$content = rtrim($content) . "\n" . $line . "\n"; |
| 659 |
|
| 660 |
$result = $this->write_robots_file($content); |
| 661 |
|
| 662 |
if (is_wp_error($result)) { |
| 663 |
return [ |
| 664 |
'success' => false, |
| 665 |
'action' => 'error', |
| 666 |
'message' => $result->get_error_message(), |
| 667 |
]; |
| 668 |
} |
| 669 |
|
| 670 |
return [ |
| 671 |
'success' => true, |
| 672 |
'action' => 'added', |
| 673 |
'message' => esc_html__('LLMs.txt reference has been added to robots.txt.', 'metasync'), |
| 674 |
]; |
| 675 |
} |
| 676 |
|
| 677 |
/** |
| 678 |
* Remove a /llms.txt reference from robots.txt. |
| 679 |
* |
| 680 |
* @param string $llms_url Absolute URL of the LLMs.txt file to remove. |
| 681 |
* @return array Result with 'success' and 'action' keys. |
| 682 |
*/ |
| 683 |
public function remove_llms_txt_url($llms_url) |
| 684 |
{ |
| 685 |
$content = $this->read_robots_file(); |
| 686 |
|
| 687 |
if (is_wp_error($content)) { |
| 688 |
return [ |
| 689 |
'success' => false, |
| 690 |
'action' => 'error', |
| 691 |
'message' => $content->get_error_message(), |
| 692 |
]; |
| 693 |
} |
| 694 |
|
| 695 |
$llms_url = esc_url_raw($llms_url); |
| 696 |
$pattern = '/^#\s*LLMs\.txt:\s*' . preg_quote($llms_url, '/') . '\s*$/im'; |
| 697 |
|
| 698 |
if (!preg_match($pattern, $content)) { |
| 699 |
return [ |
| 700 |
'success' => true, |
| 701 |
'action' => 'not_found', |
| 702 |
'message' => esc_html__('LLMs.txt reference not found in robots.txt.', 'metasync'), |
| 703 |
]; |
| 704 |
} |
| 705 |
|
| 706 |
$new_content = preg_replace($pattern, '', $content); |
| 707 |
$new_content = preg_replace("/\n{3,}/", "\n\n", $new_content); |
| 708 |
$new_content = rtrim($new_content) . "\n"; |
| 709 |
|
| 710 |
$result = $this->write_robots_file($new_content); |
| 711 |
|
| 712 |
if (is_wp_error($result)) { |
| 713 |
return [ |
| 714 |
'success' => false, |
| 715 |
'action' => 'error', |
| 716 |
'message' => $result->get_error_message(), |
| 717 |
]; |
| 718 |
} |
| 719 |
|
| 720 |
return [ |
| 721 |
'success' => true, |
| 722 |
'action' => 'removed', |
| 723 |
'message' => esc_html__('LLMs.txt reference has been removed from robots.txt.', 'metasync'), |
| 724 |
]; |
| 725 |
} |
| 726 |
|
| 727 |
/** |
| 728 |
* Get the current sitemap URL from robots.txt |
| 729 |
* |
| 730 |
* @return string|false The sitemap URL or false if not found |
| 731 |
*/ |
| 732 |
public function get_sitemap_url_from_robots() |
| 733 |
{ |
| 734 |
$content = $this->read_robots_file(); |
| 735 |
|
| 736 |
if (is_wp_error($content)) { |
| 737 |
return false; |
| 738 |
} |
| 739 |
|
| 740 |
if (preg_match('/^Sitemap:\s*(.+)$/im', $content, $matches)) { |
| 741 |
return trim($matches[1]); |
| 742 |
} |
| 743 |
|
| 744 |
return false; |
| 745 |
} |
| 746 |
|
| 747 |
/** |
| 748 |
* Render the admin page |
| 749 |
* |
| 750 |
* @param object $admin Admin class instance |
| 751 |
* @param string $current_content Current robots.txt content |
| 752 |
* @param array $backups Backup history |
| 753 |
* @param bool $file_exists Whether the file exists |
| 754 |
* @param bool $is_writable Whether the file is writable |
| 755 |
*/ |
| 756 |
public function render($admin, $current_content, $backups, $file_exists, $is_writable) |
| 757 |
{ |
| 758 |
// Load the view template |
| 759 |
require_once plugin_dir_path(__FILE__) . 'views/admin-page.php'; |
| 760 |
} |
| 761 |
} |
| 762 |
|
