| 1 |
<?php |
| 2 |
/** |
| 3 |
* MCP Tools: OTTO Pipeline Tools |
| 4 |
* |
| 5 |
* Provides three MCP tools that close the loop for AI-driven SEO work: |
| 6 |
* 1. Trigger Otto Optimization — runs the full OTTO pipeline (warm |
| 7 |
* transient → SEO DB sync → plugin sync → cache purge → cache warm → |
| 8 |
* CDN purge) for a URL or post. |
| 9 |
* 2. Get Otto Status — returns the current OTTO state for a URL or post |
| 10 |
* (enabled flag, exclusion, transient warmth, last-written meta, |
| 11 |
* persistence settings, plugin sync timestamps). |
| 12 |
* 3. Verify SEO Output — fetches the live rendered HTML for a URL and |
| 13 |
* compares the rendered head tags against stored post meta. |
| 14 |
* |
| 15 |
* @package MetaSync |
| 16 |
* @subpackage MCP_Server/Tools |
| 17 |
*/ |
| 18 |
|
| 19 |
if (!defined('ABSPATH')) { |
| 20 |
exit; |
| 21 |
} |
| 22 |
|
| 23 |
/** |
| 24 |
* Trigger OTTO Optimization Tool |
| 25 |
*/ |
| 26 |
class MCP_Tool_Trigger_Otto_Optimization extends MCP_Tool_Base { |
| 27 |
|
| 28 |
public function get_name() { |
| 29 |
return 'wordpress_trigger_otto_optimization'; |
| 30 |
} |
| 31 |
|
| 32 |
public function get_description() { |
| 33 |
return 'Trigger the full OTTO optimization pipeline for a URL or post ID: warm transient → SEO DB sync → plugin sync → cache purge → cache warm → CDN purge. Requires OTTO to be enabled and is rate-limited to 10 calls per minute per API key.'; |
| 34 |
} |
| 35 |
|
| 36 |
public function get_input_schema() { |
| 37 |
return [ |
| 38 |
'type' => 'object', |
| 39 |
'properties' => [ |
| 40 |
'url' => [ |
| 41 |
'type' => 'string', |
| 42 |
'description' => 'Absolute URL to optimize. Either url or post_id must be provided.', |
| 43 |
], |
| 44 |
'post_id' => [ |
| 45 |
'type' => 'integer', |
| 46 |
'description' => 'Post or page ID to optimize. Either url or post_id must be provided.', |
| 47 |
], |
| 48 |
], |
| 49 |
'required' => [], |
| 50 |
]; |
| 51 |
} |
| 52 |
|
| 53 |
public function execute($params) { |
| 54 |
$this->validate_params($params); |
| 55 |
$this->require_capability('manage_options'); |
| 56 |
|
| 57 |
$has_url = !empty($params['url']); |
| 58 |
$has_post_id = isset($params['post_id']) && (int) $params['post_id'] > 0; |
| 59 |
|
| 60 |
if (!$has_url && !$has_post_id) { |
| 61 |
throw new InvalidArgumentException('Either url or post_id is required'); |
| 62 |
} |
| 63 |
if ($has_url && $has_post_id) { |
| 64 |
throw new InvalidArgumentException('Provide only one of url or post_id, not both'); |
| 65 |
} |
| 66 |
|
| 67 |
// Resolve url <-> post_id |
| 68 |
if ($has_post_id) { |
| 69 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 70 |
$url = get_permalink($post_id); |
| 71 |
if (empty($url)) { |
| 72 |
throw new Exception(sprintf('Unable to resolve permalink for post_id %d', $post_id)); |
| 73 |
} |
| 74 |
} else { |
| 75 |
$url = $this->sanitize_url($params['url']); |
| 76 |
$post_id = (int) url_to_postid($url); |
| 77 |
} |
| 78 |
|
| 79 |
// Post-level permission check (when a specific post is targeted) |
| 80 |
if ($has_post_id && $post_id > 0) { |
| 81 |
$this->verify_post_exists($post_id); |
| 82 |
$this->check_post_permission($post_id); |
| 83 |
} |
| 84 |
|
| 85 |
// Load OTTO config |
| 86 |
$otto_base = dirname(dirname(dirname(__FILE__))) . '/otto/'; |
| 87 |
if (!class_exists('Metasync_Otto_Config')) { |
| 88 |
$config_file = $otto_base . 'class-metasync-otto-config.php'; |
| 89 |
if (file_exists($config_file)) { |
| 90 |
require_once $config_file; |
| 91 |
} |
| 92 |
} |
| 93 |
|
| 94 |
if (!class_exists('Metasync_Otto_Config') || !Metasync_Otto_Config::is_otto_enabled()) { |
| 95 |
throw new Exception('OTTO is not enabled for this site.'); |
| 96 |
} |
| 97 |
|
| 98 |
// Rate-limit: 10 calls per 60s per API key (fall back to IP) |
| 99 |
if (class_exists('Metasync_Rate_Limiter')) { |
| 100 |
$options = get_option('metasync_options', []); |
| 101 |
$api_key = isset($options['general']['apikey']) ? (string) $options['general']['apikey'] : ''; |
| 102 |
$rate_source = $api_key !== '' ? $api_key : (isset($_SERVER['REMOTE_ADDR']) ? (string) $_SERVER['REMOTE_ADDR'] : 'anon'); |
| 103 |
$rate_key = hash('sha256', $rate_source); |
| 104 |
$rl_result = Metasync_Rate_Limiter::get_instance()->check_rate_limit($rate_key, 10, 60, 'otto_trigger_'); |
| 105 |
if (is_wp_error($rl_result)) { |
| 106 |
throw new Exception($rl_result->get_error_message()); |
| 107 |
} |
| 108 |
} |
| 109 |
|
| 110 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 111 |
$steps = []; |
| 112 |
|
| 113 |
// Helper: time a callable and return a standardized step result. |
| 114 |
// The callable returns either a boolean success flag or an array |
| 115 |
// overriding fields in the result (e.g. ['success'=>false,'reason'=>'x']). |
| 116 |
$run_step = function (callable $fn) { |
| 117 |
$start = microtime(true); |
| 118 |
try { |
| 119 |
$outcome = $fn(); |
| 120 |
} catch (Exception $e) { |
| 121 |
return [ |
| 122 |
'success' => false, |
| 123 |
'error' => $e->getMessage(), |
| 124 |
'duration_ms' => round((microtime(true) - $start) * 1000, 2), |
| 125 |
]; |
| 126 |
} |
| 127 |
$duration_ms = round((microtime(true) - $start) * 1000, 2); |
| 128 |
if (is_array($outcome)) { |
| 129 |
return array_merge(['duration_ms' => $duration_ms], $outcome); |
| 130 |
} |
| 131 |
return [ |
| 132 |
'success' => (bool) $outcome, |
| 133 |
'duration_ms' => $duration_ms, |
| 134 |
]; |
| 135 |
}; |
| 136 |
|
| 137 |
// Step 1: warm_transient |
| 138 |
$steps['warm_transient'] = $run_step(function () use ($otto_base, $otto_uuid, $url) { |
| 139 |
$transient_file = $otto_base . 'class-metasync-otto-transient-cache.php'; |
| 140 |
if (!class_exists('Metasync_Otto_Transient_Cache') && file_exists($transient_file)) { |
| 141 |
require_once $transient_file; |
| 142 |
} |
| 143 |
if (!class_exists('Metasync_Otto_Transient_Cache') || empty($otto_uuid)) { |
| 144 |
return ['success' => false, 'reason' => 'not_available']; |
| 145 |
} |
| 146 |
$tc = new Metasync_Otto_Transient_Cache($otto_uuid); |
| 147 |
$result = $tc->warm_cache($url); |
| 148 |
// Report success only when real suggestions were cached. A URL OTTO |
| 149 |
// holds no data for returns an empty array rather than false, so a |
| 150 |
// `!== false` check would report a successful warm for a URL with |
| 151 |
// nothing to warm — inverting the signal you would use to debug this. |
| 152 |
if ($result === false) { |
| 153 |
return ['success' => false, 'reason' => 'fetch_failed']; |
| 154 |
} |
| 155 |
if (!$tc->has_payload($result)) { |
| 156 |
return ['success' => false, 'reason' => 'no_suggestions_for_url']; |
| 157 |
} |
| 158 |
return true; |
| 159 |
}); |
| 160 |
|
| 161 |
// Step 2: seo_db_sync |
| 162 |
$steps['seo_db_sync'] = $run_step(function () use ($url) { |
| 163 |
if (!function_exists('metasync_process_otto_seo_data')) { |
| 164 |
return ['success' => false, 'reason' => 'not_available']; |
| 165 |
} |
| 166 |
// The sync returns an outcome string (success / no_change / |
| 167 |
// retryable / permanent / deferred); a no-change result is a |
| 168 |
// successful sync, and anything else carries its reason through. |
| 169 |
$outcome = metasync_process_otto_seo_data($url); |
| 170 |
if ($outcome === Metasync_Otto_Job_Status::OUTCOME_SUCCESS |
| 171 |
|| $outcome === Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE) { |
| 172 |
return true; |
| 173 |
} |
| 174 |
return ['success' => false, 'reason' => is_string($outcome) ? $outcome : 'unknown']; |
| 175 |
}); |
| 176 |
|
| 177 |
// Step 3: plugin_sync (may not be available yet) |
| 178 |
$steps['plugin_sync'] = $run_step(function () use ($post_id) { |
| 179 |
if (!class_exists('Metasync_Plugin_Sync')) { |
| 180 |
return ['success' => false, 'reason' => 'not_available']; |
| 181 |
} |
| 182 |
$sync_result = null; |
| 183 |
if (method_exists('Metasync_Plugin_Sync', 'sync_post')) { |
| 184 |
$sync_result = Metasync_Plugin_Sync::sync_post($post_id); |
| 185 |
} elseif (method_exists('Metasync_Plugin_Sync', 'sync')) { |
| 186 |
$sync_result = Metasync_Plugin_Sync::sync($post_id); |
| 187 |
} else { |
| 188 |
return ['success' => false, 'reason' => 'not_available']; |
| 189 |
} |
| 190 |
return $sync_result !== false; |
| 191 |
}); |
| 192 |
|
| 193 |
// Step 4: cache_purge |
| 194 |
$steps['cache_purge'] = $run_step(function () use ($url) { |
| 195 |
if (!class_exists('Metasync_Cache_Purge')) { |
| 196 |
return ['success' => false, 'reason' => 'not_available']; |
| 197 |
} |
| 198 |
Metasync_Cache_Purge::purge_single_url($url); |
| 199 |
return true; |
| 200 |
}); |
| 201 |
|
| 202 |
// Step 5: cache_warm |
| 203 |
$steps['cache_warm'] = $run_step(function () use ($url) { |
| 204 |
if (!class_exists('Metasync_Cache_Purge')) { |
| 205 |
return ['success' => false, 'reason' => 'not_available']; |
| 206 |
} |
| 207 |
Metasync_Cache_Purge::warm_urls([$url]); |
| 208 |
return true; |
| 209 |
}); |
| 210 |
|
| 211 |
// Step 6: cdn_purge |
| 212 |
$steps['cdn_purge'] = $run_step(function () use ($url) { |
| 213 |
if (!class_exists('Metasync_Edge_Cache_Purge')) { |
| 214 |
return ['success' => false, 'reason' => 'not_available']; |
| 215 |
} |
| 216 |
Metasync_Edge_Cache_Purge::purge([$url]); |
| 217 |
return true; |
| 218 |
}); |
| 219 |
|
| 220 |
return $this->success( |
| 221 |
[ |
| 222 |
'url' => $url, |
| 223 |
'post_id' => $post_id ?: null, |
| 224 |
'steps' => $steps, |
| 225 |
], |
| 226 |
'OTTO optimization pipeline completed' |
| 227 |
); |
| 228 |
} |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Get OTTO Status Tool |
| 233 |
*/ |
| 234 |
class MCP_Tool_Get_Otto_Status extends MCP_Tool_Base { |
| 235 |
|
| 236 |
public function get_name() { |
| 237 |
return 'wordpress_get_otto_status'; |
| 238 |
} |
| 239 |
|
| 240 |
public function get_description() { |
| 241 |
return 'Get the current OTTO state for a URL or post: enabled flag, exclusion status, transient warmth, last-written OTTO meta values, persistence settings, and plugin sync timestamps. Works regardless of whether OTTO is enabled.'; |
| 242 |
} |
| 243 |
|
| 244 |
public function get_input_schema() { |
| 245 |
return [ |
| 246 |
'type' => 'object', |
| 247 |
'properties' => [ |
| 248 |
'url' => [ |
| 249 |
'type' => 'string', |
| 250 |
'description' => 'Absolute URL to inspect. Either url or post_id must be provided.', |
| 251 |
], |
| 252 |
'post_id' => [ |
| 253 |
'type' => 'integer', |
| 254 |
'description' => 'Post or page ID to inspect. Either url or post_id must be provided.', |
| 255 |
], |
| 256 |
], |
| 257 |
'required' => [], |
| 258 |
]; |
| 259 |
} |
| 260 |
|
| 261 |
public function execute($params) { |
| 262 |
$this->validate_params($params); |
| 263 |
$this->require_capability('manage_options'); |
| 264 |
|
| 265 |
$has_url = !empty($params['url']); |
| 266 |
$has_post_id = isset($params['post_id']) && (int) $params['post_id'] > 0; |
| 267 |
|
| 268 |
if (!$has_url && !$has_post_id) { |
| 269 |
throw new InvalidArgumentException('Either url or post_id is required'); |
| 270 |
} |
| 271 |
if ($has_url && $has_post_id) { |
| 272 |
throw new InvalidArgumentException('Provide only one of url or post_id, not both'); |
| 273 |
} |
| 274 |
|
| 275 |
if ($has_post_id) { |
| 276 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 277 |
$this->verify_post_exists($post_id); |
| 278 |
$this->check_post_permission($post_id); |
| 279 |
$url = get_permalink($post_id); |
| 280 |
} else { |
| 281 |
$url = $this->sanitize_url($params['url']); |
| 282 |
$post_id = (int) url_to_postid($url); |
| 283 |
if ($post_id > 0) { |
| 284 |
$this->check_post_permission($post_id); |
| 285 |
} |
| 286 |
} |
| 287 |
|
| 288 |
$otto_base = dirname(dirname(dirname(__FILE__))) . '/otto/'; |
| 289 |
if (!class_exists('Metasync_Otto_Config')) { |
| 290 |
$config_file = $otto_base . 'class-metasync-otto-config.php'; |
| 291 |
if (file_exists($config_file)) { |
| 292 |
require_once $config_file; |
| 293 |
} |
| 294 |
} |
| 295 |
|
| 296 |
$otto_enabled = class_exists('Metasync_Otto_Config') ? Metasync_Otto_Config::is_otto_enabled() : false; |
| 297 |
$otto_uuid = class_exists('Metasync_Otto_Config') ? Metasync_Otto_Config::get_otto_uuid() : ''; |
| 298 |
|
| 299 |
// Exclusion status (manual only — reflects user-configured exclusions) |
| 300 |
$url_excluded = null; |
| 301 |
if (!empty($url)) { |
| 302 |
if (!function_exists('metasync_is_otto_url_manually_excluded')) { |
| 303 |
$pixel_file = $otto_base . 'otto_pixel.php'; |
| 304 |
if (file_exists($pixel_file)) { |
| 305 |
require_once $pixel_file; |
| 306 |
} |
| 307 |
} |
| 308 |
if (function_exists('metasync_is_otto_url_manually_excluded')) { |
| 309 |
$url_excluded = (bool) metasync_is_otto_url_manually_excluded($url); |
| 310 |
} |
| 311 |
} |
| 312 |
|
| 313 |
// Transient state |
| 314 |
$transient = [ |
| 315 |
'warm' => false, |
| 316 |
'has_suggestions' => false, |
| 317 |
]; |
| 318 |
if (!empty($url) && !empty($otto_uuid)) { |
| 319 |
if (!class_exists('Metasync_Otto_Transient_Cache')) { |
| 320 |
$transient_file = $otto_base . 'class-metasync-otto-transient-cache.php'; |
| 321 |
if (file_exists($transient_file)) { |
| 322 |
require_once $transient_file; |
| 323 |
} |
| 324 |
} |
| 325 |
if (class_exists('Metasync_Otto_Transient_Cache')) { |
| 326 |
$tc = new Metasync_Otto_Transient_Cache($otto_uuid); |
| 327 |
$stats = $tc->get_stats($url); |
| 328 |
$transient['warm'] = !empty($stats['has_cache']); |
| 329 |
$transient['has_suggestions'] = !empty($stats['has_suggestions']); |
| 330 |
$transient['cache_key'] = isset($stats['cache_key']) ? $stats['cache_key'] : null; |
| 331 |
|
| 332 |
// Get transient expiry for the OTTO cache |
| 333 |
global $wpdb; |
| 334 |
$site_id = is_multisite() ? get_current_blog_id() : 0; |
| 335 |
$normalized_url = rtrim(strtolower($url), '/'); |
| 336 |
$url_hash = md5($normalized_url); |
| 337 |
$transient_key = 'otto_suggestions_' . $site_id . '_' . $url_hash; |
| 338 |
$timeout_key = '_transient_timeout_' . $transient_key; |
| 339 |
$expiry_raw = $wpdb->get_var( |
| 340 |
$wpdb->prepare( |
| 341 |
"SELECT option_value FROM {$wpdb->options} WHERE option_name = %s LIMIT 1", |
| 342 |
$timeout_key |
| 343 |
) |
| 344 |
); |
| 345 |
$cache_expires_at = $expiry_raw ? (int) $expiry_raw : null; |
| 346 |
$cache_expires_in = $cache_expires_at ? max(0, $cache_expires_at - time()) : null; |
| 347 |
|
| 348 |
$transient['cache_expires_at'] = $cache_expires_at; |
| 349 |
$transient['cache_expires_in_seconds'] = $cache_expires_in; |
| 350 |
} |
| 351 |
} |
| 352 |
|
| 353 |
// Last-written OTTO meta values |
| 354 |
$otto_meta_keys = [ |
| 355 |
'_metasync_otto_title', |
| 356 |
'_metasync_otto_description', |
| 357 |
'_metasync_otto_keywords', |
| 358 |
'_metasync_otto_og_title', |
| 359 |
'_metasync_otto_og_description', |
| 360 |
'_metasync_otto_twitter_title', |
| 361 |
'_metasync_otto_twitter_description', |
| 362 |
'_metasync_canonical_url', |
| 363 |
'_metasync_otto_structured_data', |
| 364 |
'_metasync_otto_image_alt_data', |
| 365 |
'_metasync_otto_headings_data', |
| 366 |
'_metasync_otto_last_update', |
| 367 |
]; |
| 368 |
$last_written_meta = []; |
| 369 |
if ($post_id > 0) { |
| 370 |
foreach ($otto_meta_keys as $key) { |
| 371 |
$val = get_post_meta($post_id, $key, true); |
| 372 |
if ($val !== '' && $val !== false && $val !== null) { |
| 373 |
$last_written_meta[$key] = $val; |
| 374 |
} |
| 375 |
} |
| 376 |
} |
| 377 |
|
| 378 |
// Persistence settings |
| 379 |
$persistence_settings = null; |
| 380 |
if (!class_exists('Metasync_Otto_Persistence_Settings')) { |
| 381 |
$persist_file = $otto_base . 'class-metasync-otto-persistence-settings.php'; |
| 382 |
if (file_exists($persist_file)) { |
| 383 |
require_once $persist_file; |
| 384 |
} |
| 385 |
} |
| 386 |
if (class_exists('Metasync_Otto_Persistence_Settings')) { |
| 387 |
$persistence_settings = Metasync_Otto_Persistence_Settings::get_settings(); |
| 388 |
} |
| 389 |
|
| 390 |
// Plugin sync timestamp (written by when available) |
| 391 |
$plugin_sync_timestamp = null; |
| 392 |
if ($post_id > 0) { |
| 393 |
$ts = get_post_meta($post_id, '_metasync_plugin_sync_ts', true); |
| 394 |
$plugin_sync_timestamp = $ts !== '' ? $ts : null; |
| 395 |
} |
| 396 |
|
| 397 |
return $this->success( |
| 398 |
[ |
| 399 |
'otto_enabled' => (bool) $otto_enabled, |
| 400 |
'url' => $url, |
| 401 |
'post_id' => $post_id ?: null, |
| 402 |
'url_excluded' => $url_excluded, |
| 403 |
'transient' => $transient, |
| 404 |
'last_written_meta' => $last_written_meta, |
| 405 |
'persistence_settings' => $persistence_settings, |
| 406 |
'plugin_sync_timestamp' => $plugin_sync_timestamp, |
| 407 |
], |
| 408 |
'OTTO status retrieved' |
| 409 |
); |
| 410 |
} |
| 411 |
} |
| 412 |
|
| 413 |
/** |
| 414 |
* Verify SEO Output Tool |
| 415 |
*/ |
| 416 |
class MCP_Tool_Verify_SEO_Output extends MCP_Tool_Base { |
| 417 |
|
| 418 |
public function get_name() { |
| 419 |
return 'wordpress_verify_seo_output'; |
| 420 |
} |
| 421 |
|
| 422 |
public function get_description() { |
| 423 |
return 'Fetch the live rendered HTML for a URL and extract what Google actually sees (title, description, canonical, robots, OG/Twitter tags, JSON-LD schema, hreflang). Compares rendered values against stored post meta and reports duplicate head tags.'; |
| 424 |
} |
| 425 |
|
| 426 |
public function get_input_schema() { |
| 427 |
return [ |
| 428 |
'type' => 'object', |
| 429 |
'properties' => [ |
| 430 |
'url' => [ |
| 431 |
'type' => 'string', |
| 432 |
'description' => 'Absolute URL to fetch and verify.', |
| 433 |
], |
| 434 |
'fields' => [ |
| 435 |
'type' => 'array', |
| 436 |
'description' => 'Optional list of field categories to include. If omitted, all are returned.', |
| 437 |
'items' => [ |
| 438 |
'type' => 'string', |
| 439 |
'enum' => ['title', 'description', 'canonical', 'robots', 'og', 'twitter', 'hreflang', 'schema'], |
| 440 |
], |
| 441 |
], |
| 442 |
], |
| 443 |
'required' => ['url'], |
| 444 |
]; |
| 445 |
} |
| 446 |
|
| 447 |
public function execute($params) { |
| 448 |
$this->validate_params($params); |
| 449 |
$this->require_capability('manage_options'); |
| 450 |
|
| 451 |
$url = $this->sanitize_url($params['url']); |
| 452 |
if (empty($url)) { |
| 453 |
throw new InvalidArgumentException('url is required'); |
| 454 |
} |
| 455 |
|
| 456 |
$fields_filter = []; |
| 457 |
if (!empty($params['fields']) && is_array($params['fields'])) { |
| 458 |
foreach ($params['fields'] as $f) { |
| 459 |
if (is_string($f)) { |
| 460 |
$fields_filter[] = strtolower($f); |
| 461 |
} |
| 462 |
} |
| 463 |
} |
| 464 |
$include = function ($category) use ($fields_filter) { |
| 465 |
return empty($fields_filter) || in_array($category, $fields_filter, true); |
| 466 |
}; |
| 467 |
|
| 468 |
$post_id = (int) url_to_postid($url); |
| 469 |
|
| 470 |
// Fetch the live HTML |
| 471 |
$response = wp_remote_get($url, [ |
| 472 |
'timeout' => 15, |
| 473 |
'redirection' => 5, |
| 474 |
'user-agent' => 'MetaSync-Verify/1.0', |
| 475 |
'sslverify' => false, |
| 476 |
]); |
| 477 |
|
| 478 |
if (is_wp_error($response)) { |
| 479 |
throw new Exception('HTTP fetch failed: ' . $response->get_error_message()); |
| 480 |
} |
| 481 |
|
| 482 |
$status = (int) wp_remote_retrieve_response_code($response); |
| 483 |
$html = (string) wp_remote_retrieve_body($response); |
| 484 |
|
| 485 |
// Load simplehtmldom via the OTTO HTML class which owns the vendor lib |
| 486 |
$html_class_file = dirname(dirname(dirname(__FILE__))) . '/otto/Otto_html_class.php'; |
| 487 |
if (file_exists($html_class_file)) { |
| 488 |
require_once $html_class_file; |
| 489 |
} |
| 490 |
if (!class_exists('simplehtmldom\\HtmlDocument')) { |
| 491 |
throw new Exception('simplehtmldom HtmlDocument is not available'); |
| 492 |
} |
| 493 |
|
| 494 |
$dom = new simplehtmldom\HtmlDocument($html, true, true, 'UTF-8', false); |
| 495 |
|
| 496 |
// --- Extract rendered values --- |
| 497 |
$rendered = []; |
| 498 |
$duplicates = []; |
| 499 |
|
| 500 |
// title |
| 501 |
$title_nodes = $dom->find('title'); |
| 502 |
$title_val = null; |
| 503 |
if (!empty($title_nodes)) { |
| 504 |
$title_val = trim($title_nodes[0]->plaintext); |
| 505 |
if (count($title_nodes) > 1) { |
| 506 |
$duplicates['title'] = count($title_nodes); |
| 507 |
} |
| 508 |
} |
| 509 |
if ($include('title')) { |
| 510 |
$rendered['title'] = $title_val; |
| 511 |
} |
| 512 |
|
| 513 |
// meta description |
| 514 |
$desc_nodes = $dom->find('meta[name=description]'); |
| 515 |
$desc_val = null; |
| 516 |
if (!empty($desc_nodes)) { |
| 517 |
$desc_val = isset($desc_nodes[0]->content) ? trim($desc_nodes[0]->content) : null; |
| 518 |
if (count($desc_nodes) > 1) { |
| 519 |
$duplicates['meta[name=description]'] = count($desc_nodes); |
| 520 |
} |
| 521 |
} |
| 522 |
if ($include('description')) { |
| 523 |
$rendered['description'] = $desc_val; |
| 524 |
} |
| 525 |
|
| 526 |
// canonical |
| 527 |
$canon_nodes = $dom->find('link[rel=canonical]'); |
| 528 |
$canon_val = null; |
| 529 |
if (!empty($canon_nodes)) { |
| 530 |
$canon_val = isset($canon_nodes[0]->href) ? trim($canon_nodes[0]->href) : null; |
| 531 |
if (count($canon_nodes) > 1) { |
| 532 |
$duplicates['link[rel=canonical]'] = count($canon_nodes); |
| 533 |
} |
| 534 |
} |
| 535 |
if ($include('canonical')) { |
| 536 |
$rendered['canonical'] = $canon_val; |
| 537 |
} |
| 538 |
|
| 539 |
// robots |
| 540 |
$robots_nodes = $dom->find('meta[name=robots]'); |
| 541 |
$robots_val = null; |
| 542 |
if (!empty($robots_nodes)) { |
| 543 |
$robots_val = isset($robots_nodes[0]->content) ? trim($robots_nodes[0]->content) : null; |
| 544 |
if (count($robots_nodes) > 1) { |
| 545 |
$duplicates['meta[name=robots]'] = count($robots_nodes); |
| 546 |
} |
| 547 |
} |
| 548 |
if ($include('robots')) { |
| 549 |
$rendered['robots'] = $robots_val; |
| 550 |
} |
| 551 |
|
| 552 |
// OG tags (property starts with og:) |
| 553 |
if ($include('og')) { |
| 554 |
$og_tags = []; |
| 555 |
foreach ($dom->find('meta') as $m) { |
| 556 |
$property = isset($m->property) ? (string) $m->property : ''; |
| 557 |
if ($property !== '' && strpos($property, 'og:') === 0) { |
| 558 |
$og_tags[$property] = isset($m->content) ? (string) $m->content : ''; |
| 559 |
} |
| 560 |
} |
| 561 |
$rendered['og'] = $og_tags; |
| 562 |
} |
| 563 |
|
| 564 |
// Twitter tags |
| 565 |
if ($include('twitter')) { |
| 566 |
$tw_tags = []; |
| 567 |
foreach ($dom->find('meta') as $m) { |
| 568 |
$name = isset($m->name) ? (string) $m->name : ''; |
| 569 |
if ($name !== '' && strpos($name, 'twitter:') === 0) { |
| 570 |
$tw_tags[$name] = isset($m->content) ? (string) $m->content : ''; |
| 571 |
} |
| 572 |
} |
| 573 |
$rendered['twitter'] = $tw_tags; |
| 574 |
} |
| 575 |
|
| 576 |
// hreflang |
| 577 |
if ($include('hreflang')) { |
| 578 |
$hreflang = []; |
| 579 |
foreach ($dom->find('link[rel=alternate]') as $lnk) { |
| 580 |
$hl = isset($lnk->hreflang) ? (string) $lnk->hreflang : ''; |
| 581 |
if ($hl !== '') { |
| 582 |
$hreflang[] = [ |
| 583 |
'hreflang' => $hl, |
| 584 |
'href' => isset($lnk->href) ? (string) $lnk->href : '', |
| 585 |
]; |
| 586 |
} |
| 587 |
} |
| 588 |
$rendered['hreflang'] = $hreflang; |
| 589 |
} |
| 590 |
|
| 591 |
// JSON-LD schema |
| 592 |
if ($include('schema')) { |
| 593 |
$schemas = []; |
| 594 |
foreach ($dom->find('script[type=application/ld+json]') as $s) { |
| 595 |
$raw = trim($s->innertext); |
| 596 |
$decoded = json_decode($raw, true); |
| 597 |
if (json_last_error() === JSON_ERROR_NONE) { |
| 598 |
$schemas[] = $decoded; |
| 599 |
} else { |
| 600 |
$schemas[] = ['_raw' => $raw, '_parse_error' => json_last_error_msg()]; |
| 601 |
} |
| 602 |
} |
| 603 |
$rendered['schema'] = $schemas; |
| 604 |
} |
| 605 |
|
| 606 |
// --- Stored meta for comparison --- |
| 607 |
$stored_meta_keys = [ |
| 608 |
'title' => '_metasync_otto_title', |
| 609 |
'description' => '_metasync_otto_description', |
| 610 |
'canonical' => '_metasync_canonical_url', |
| 611 |
'og_title' => '_metasync_otto_og_title', |
| 612 |
'og_description' => '_metasync_otto_og_description', |
| 613 |
'twitter_title' => '_metasync_otto_twitter_title', |
| 614 |
'twitter_description' => '_metasync_otto_twitter_description', |
| 615 |
]; |
| 616 |
$stored_meta = []; |
| 617 |
if ($post_id > 0) { |
| 618 |
foreach ($stored_meta_keys as $logical => $meta_key) { |
| 619 |
$val = get_post_meta($post_id, $meta_key, true); |
| 620 |
$stored_meta[$logical] = $val !== '' ? $val : null; |
| 621 |
} |
| 622 |
} else { |
| 623 |
foreach ($stored_meta_keys as $logical => $meta_key) { |
| 624 |
$stored_meta[$logical] = null; |
| 625 |
} |
| 626 |
} |
| 627 |
|
| 628 |
// --- Field-by-field comparison --- |
| 629 |
$comparison = []; |
| 630 |
|
| 631 |
$add_comparison = function ($category, $field, $stored_val, $rendered_val) use (&$comparison, $include) { |
| 632 |
if (!$include($category)) { |
| 633 |
return; |
| 634 |
} |
| 635 |
$comparison[] = [ |
| 636 |
'field' => $field, |
| 637 |
'match' => ($stored_val === $rendered_val) || ($stored_val === null && $rendered_val === null) || ((string) $stored_val === (string) $rendered_val), |
| 638 |
'stored' => $stored_val, |
| 639 |
'rendered' => $rendered_val, |
| 640 |
]; |
| 641 |
}; |
| 642 |
|
| 643 |
$add_comparison('title', 'title', $stored_meta['title'], $title_val); |
| 644 |
$add_comparison('description', 'description', $stored_meta['description'], $desc_val); |
| 645 |
$add_comparison('canonical', 'canonical', $stored_meta['canonical'], $canon_val); |
| 646 |
|
| 647 |
if ($include('og')) { |
| 648 |
$og_rendered = isset($rendered['og']) ? $rendered['og'] : []; |
| 649 |
$add_comparison('og', 'og:title', $stored_meta['og_title'], isset($og_rendered['og:title']) ? $og_rendered['og:title'] : null); |
| 650 |
$add_comparison('og', 'og:description', $stored_meta['og_description'], isset($og_rendered['og:description']) ? $og_rendered['og:description'] : null); |
| 651 |
} |
| 652 |
if ($include('twitter')) { |
| 653 |
$tw_rendered = isset($rendered['twitter']) ? $rendered['twitter'] : []; |
| 654 |
$add_comparison('twitter', 'twitter:title', $stored_meta['twitter_title'], isset($tw_rendered['twitter:title']) ? $tw_rendered['twitter:title'] : null); |
| 655 |
$add_comparison('twitter', 'twitter:description', $stored_meta['twitter_description'], isset($tw_rendered['twitter:description']) ? $tw_rendered['twitter:description'] : null); |
| 656 |
} |
| 657 |
|
| 658 |
return $this->success( |
| 659 |
[ |
| 660 |
'url' => $url, |
| 661 |
'post_id' => $post_id ?: null, |
| 662 |
'http_status' => $status, |
| 663 |
'fetch_status' => $status >= 200 && $status < 400 ? 'ok' : 'error', |
| 664 |
'rendered' => $rendered, |
| 665 |
'stored_meta' => $stored_meta, |
| 666 |
'comparison' => $comparison, |
| 667 |
'duplicate_tags' => $duplicates, |
| 668 |
], |
| 669 |
'SEO output verification complete' |
| 670 |
); |
| 671 |
} |
| 672 |
} |
| 673 |
|
