| 1 |
<?php |
| 2 |
/** |
| 3 |
* MCP Tool: Post Meta Operations |
| 4 |
* |
| 5 |
* Provides tools for managing WordPress post meta fields, |
| 6 |
* specifically SEO-related metadata. |
| 7 |
* |
| 8 |
* @package MetaSync |
| 9 |
* @subpackage MCP_Server/Tools |
| 10 |
*/ |
| 11 |
|
| 12 |
if (!defined('ABSPATH')) { |
| 13 |
exit; |
| 14 |
} |
| 15 |
|
| 16 |
/** |
| 17 |
* Update Post Meta Tool |
| 18 |
*/ |
| 19 |
class MCP_Tool_Update_Post_Meta extends MCP_Tool_Base { |
| 20 |
|
| 21 |
public function get_name() { |
| 22 |
return 'wordpress_update_post_meta'; |
| 23 |
} |
| 24 |
|
| 25 |
public function get_description() { |
| 26 |
return 'Update a WordPress post meta field (SEO data like title, description, keywords, robots settings, Open Graph/social meta, hreflang/language alternates)'; |
| 27 |
} |
| 28 |
|
| 29 |
public function get_input_schema() { |
| 30 |
return [ |
| 31 |
'type' => 'object', |
| 32 |
'properties' => [ |
| 33 |
'post_id' => [ |
| 34 |
'type' => 'integer', |
| 35 |
'description' => 'WordPress post or page ID', |
| 36 |
'minimum' => 1 |
| 37 |
], |
| 38 |
'meta_key' => [ |
| 39 |
'type' => 'string', |
| 40 |
'description' => 'Meta field to update', |
| 41 |
'enum' => [ |
| 42 |
'_metasync_metatitle', |
| 43 |
'_metasync_metadesc', |
| 44 |
'_metasync_focus_keyword', |
| 45 |
'_metasync_robots_index', |
| 46 |
'_metasync_canonical_url', |
| 47 |
'_metasync_og_enabled', |
| 48 |
'_metasync_og_title', |
| 49 |
'_metasync_og_description', |
| 50 |
'_metasync_og_image', |
| 51 |
'_metasync_og_url', |
| 52 |
'_metasync_og_type', |
| 53 |
'_metasync_twitter_title', |
| 54 |
'_metasync_twitter_description', |
| 55 |
'_metasync_twitter_card', |
| 56 |
'_metasync_primary_category', |
| 57 |
'_metasync_otto_keywords', |
| 58 |
'_metasync_og_article_author', |
| 59 |
'_metasync_hreflang', |
| 60 |
'_metasync_breadcrumb_title', |
| 61 |
'_metasync_robots_advanced' |
| 62 |
] |
| 63 |
], |
| 64 |
'meta_value' => [ |
| 65 |
'type' => 'string', |
| 66 |
'description' => 'Value to set for the meta field' |
| 67 |
] |
| 68 |
], |
| 69 |
'required' => ['post_id', 'meta_key', 'meta_value'] |
| 70 |
]; |
| 71 |
} |
| 72 |
|
| 73 |
public function execute($params) { |
| 74 |
// Validate and sanitize |
| 75 |
$this->validate_params($params); |
| 76 |
$this->require_capability('edit_posts'); |
| 77 |
|
| 78 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 79 |
$meta_key = $this->sanitize_string($params['meta_key']); |
| 80 |
$meta_value = $this->sanitize_textarea($params['meta_value']); |
| 81 |
|
| 82 |
// hreflang / language alternates: value must be a JSON array of |
| 83 |
// {lang, url} objects (with optional region). Parse from the raw |
| 84 |
// param to avoid textarea sanitization mangling the JSON, validate |
| 85 |
// shape AND values (ISO language codes, absolute http(s) URLs), |
| 86 |
// normalise, and re-encode to canonical form before storing. |
| 87 |
if ($meta_key === '_metasync_hreflang') { |
| 88 |
$raw_value = isset($params['meta_value']) ? (string) $params['meta_value'] : ''; |
| 89 |
$decoded = json_decode($raw_value, true); |
| 90 |
if (!is_array($decoded)) { |
| 91 |
throw new Exception("_metasync_hreflang must be a JSON array"); |
| 92 |
} |
| 93 |
$clean = []; |
| 94 |
foreach ($decoded as $entry) { |
| 95 |
if (!is_array($entry) || !isset($entry['lang']) || !isset($entry['url'])) { |
| 96 |
throw new Exception("Each hreflang entry must have 'lang' and 'url' keys"); |
| 97 |
} |
| 98 |
$normalized = Metasync_Hreflang_Output::normalize_entry($entry); |
| 99 |
if (!$normalized['valid']) { |
| 100 |
throw new Exception(sprintf( |
| 101 |
"Invalid hreflang language code '%s' — use an ISO code like 'en', 'en-US' (hyphen, not underscore) or 'x-default'", |
| 102 |
is_scalar($entry['lang']) ? (string) $entry['lang'] : gettype($entry['lang']) |
| 103 |
)); |
| 104 |
} |
| 105 |
// Sanitize each URL to strip javascript: and other unsafe protocols. |
| 106 |
$url = esc_url_raw(trim((string) $entry['url'])); |
| 107 |
if (!Metasync_Hreflang_Output::is_absolute_http_url($url)) { |
| 108 |
throw new Exception(sprintf( |
| 109 |
"hreflang URL '%s' must be an absolute http(s) URL (relative paths are not valid hreflang hrefs)", |
| 110 |
(string) $entry['url'] |
| 111 |
)); |
| 112 |
} |
| 113 |
$clean[] = [ |
| 114 |
'lang' => $normalized['lang'], |
| 115 |
'region' => $normalized['region'], |
| 116 |
'url' => $url, |
| 117 |
]; |
| 118 |
} |
| 119 |
// update_metadata() runs wp_unslash() on the value; combined with |
| 120 |
// the default \uXXXX escaping this corrupted non-ASCII URLs |
| 121 |
// (…/über-uns/ stored as …/u00fcber-uns/). Encoding without the |
| 122 |
// escapes and passing slashed data — the same contract the REST |
| 123 |
// path uses — keeps the stored JSON byte-faithful. |
| 124 |
$meta_value = wp_slash(wp_json_encode($clean, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES)); |
| 125 |
} |
| 126 |
|
| 127 |
// Validate _metasync_robots_advanced JSON |
| 128 |
if ($meta_key === '_metasync_robots_advanced') { |
| 129 |
$raw_value = isset($params['meta_value']) ? (string) $params['meta_value'] : ''; |
| 130 |
$decoded = json_decode($raw_value, true); |
| 131 |
if (!is_array($decoded)) { |
| 132 |
throw new Exception("_metasync_robots_advanced must be a JSON object"); |
| 133 |
} |
| 134 |
$allowed_keys = ['nofollow', 'noarchive', 'nosnippet', 'noimageindex', 'max_snippet', 'max_image_preview', 'max_video_preview']; |
| 135 |
foreach (array_keys($decoded) as $k) { |
| 136 |
if (!in_array($k, $allowed_keys, true)) { |
| 137 |
throw new Exception("Unknown key '{$k}' in _metasync_robots_advanced. Allowed: " . implode(', ', $allowed_keys)); |
| 138 |
} |
| 139 |
} |
| 140 |
// Validate types |
| 141 |
foreach (['nofollow', 'noarchive', 'nosnippet', 'noimageindex'] as $bool_key) { |
| 142 |
if (isset($decoded[$bool_key]) && !is_bool($decoded[$bool_key])) { |
| 143 |
$decoded[$bool_key] = (bool) $decoded[$bool_key]; |
| 144 |
} |
| 145 |
} |
| 146 |
if (isset($decoded['max_snippet'])) { |
| 147 |
$decoded['max_snippet'] = (int) $decoded['max_snippet']; |
| 148 |
} |
| 149 |
if (isset($decoded['max_image_preview'])) { |
| 150 |
$valid = ['none', 'standard', 'large']; |
| 151 |
if (!in_array($decoded['max_image_preview'], $valid, true)) { |
| 152 |
throw new Exception("max_image_preview must be one of: " . implode(', ', $valid)); |
| 153 |
} |
| 154 |
} |
| 155 |
if (isset($decoded['max_video_preview'])) { |
| 156 |
$decoded['max_video_preview'] = (int) $decoded['max_video_preview']; |
| 157 |
} |
| 158 |
$meta_value = wp_json_encode($decoded); |
| 159 |
} |
| 160 |
|
| 161 |
// SECURITY: Apply esc_url_raw() to URL-typed meta keys (prevent javascript: protocol). |
| 162 |
if (in_array($meta_key, ['_metasync_og_image', '_metasync_og_url', '_metasync_canonical_url', '_metasync_og_article_author'])) { |
| 163 |
$meta_value = esc_url_raw($meta_value); |
| 164 |
} |
| 165 |
|
| 166 |
// Canonical must be a usable URL — reject corrupted values like |
| 167 |
// "Array"/"http://Array" so they can't re-enter storage. |
| 168 |
if ($meta_key === '_metasync_canonical_url') { |
| 169 |
$meta_value = Metasync_Canonical_Sanitizer::sanitize_for_save($meta_value); |
| 170 |
if ($meta_value === '') { |
| 171 |
return $this->error('canonical_url must be a valid URL'); |
| 172 |
} |
| 173 |
} |
| 174 |
|
| 175 |
// Sanitize integer meta fields |
| 176 |
if ($meta_key === '_metasync_primary_category') { |
| 177 |
$meta_value = absint($meta_value); |
| 178 |
} |
| 179 |
|
| 180 |
// Verify post exists |
| 181 |
$post = $this->verify_post_exists($post_id); |
| 182 |
|
| 183 |
// SECURITY: Check user has permission to edit this specific post |
| 184 |
$this->check_post_permission($post_id); |
| 185 |
|
| 186 |
// Update meta |
| 187 |
$current_value = get_post_meta($post_id, $meta_key, true); |
| 188 |
// Normalize numeric meta keys: WordPress returns stored values as strings, |
| 189 |
// so cast for type-safe comparison against the integer $meta_value above. |
| 190 |
if ($meta_key === '_metasync_primary_category') { |
| 191 |
$current_value = (int) $current_value; |
| 192 |
} |
| 193 |
// The hreflang payload is stored slashed (wp_slash) so it survives |
| 194 |
// update_metadata()'s wp_unslash(); the stored value is therefore the |
| 195 |
// UNSLASHED form. Compare against that, or an identical re-write would |
| 196 |
// look like a change, run update_post_meta(), come back false ("value |
| 197 |
// unchanged") and wrongly throw below. |
| 198 |
$compare_value = $meta_key === '_metasync_hreflang' ? wp_unslash($meta_value) : $meta_value; |
| 199 |
if ($current_value === $compare_value) { |
| 200 |
// Value already matches — return success without update |
| 201 |
return $this->success([ |
| 202 |
'post_id' => $post_id, |
| 203 |
'meta_key' => $meta_key, |
| 204 |
'meta_value' => $current_value, |
| 205 |
'updated' => false, |
| 206 |
'post_title' => $post->post_title, |
| 207 |
'post_type' => $post->post_type |
| 208 |
], 'Meta value already matches, no update needed'); |
| 209 |
} |
| 210 |
$updated = update_post_meta($post_id, $meta_key, $meta_value); |
| 211 |
|
| 212 |
if ($updated === false) { |
| 213 |
throw new Exception("Failed to update meta key '{$meta_key}'"); |
| 214 |
} |
| 215 |
|
| 216 |
$stored_value = get_post_meta($post_id, $meta_key, true); |
| 217 |
return $this->success([ |
| 218 |
'post_id' => $post_id, |
| 219 |
'meta_key' => $meta_key, |
| 220 |
'meta_value' => $stored_value, |
| 221 |
'updated' => true, |
| 222 |
'post_title' => $post->post_title, |
| 223 |
'post_type' => $post->post_type |
| 224 |
], "Meta field '{$meta_key}' updated successfully"); |
| 225 |
} |
| 226 |
} |
| 227 |
|
| 228 |
/** |
| 229 |
* Get Post Meta Tool |
| 230 |
*/ |
| 231 |
class MCP_Tool_Get_Post_Meta extends MCP_Tool_Base { |
| 232 |
|
| 233 |
public function get_name() { |
| 234 |
return 'wordpress_get_post_meta'; |
| 235 |
} |
| 236 |
|
| 237 |
public function get_description() { |
| 238 |
return 'Get WordPress post meta field value(s)'; |
| 239 |
} |
| 240 |
|
| 241 |
public function get_input_schema() { |
| 242 |
return [ |
| 243 |
'type' => 'object', |
| 244 |
'properties' => [ |
| 245 |
'post_id' => [ |
| 246 |
'type' => 'integer', |
| 247 |
'description' => 'WordPress post or page ID', |
| 248 |
'minimum' => 1 |
| 249 |
], |
| 250 |
'meta_key' => [ |
| 251 |
'type' => 'string', |
| 252 |
'description' => 'Specific meta key to retrieve (optional - omit to get all SEO meta)', |
| 253 |
] |
| 254 |
], |
| 255 |
'required' => ['post_id'] |
| 256 |
]; |
| 257 |
} |
| 258 |
|
| 259 |
public function execute($params) { |
| 260 |
$this->validate_params($params); |
| 261 |
$this->require_capability('read'); |
| 262 |
|
| 263 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 264 |
|
| 265 |
// Verify post exists |
| 266 |
$post = $this->verify_post_exists($post_id); |
| 267 |
|
| 268 |
// SECURITY: Check user has permission to read this specific post |
| 269 |
$this->check_post_permission($post_id); |
| 270 |
|
| 271 |
// Get meta |
| 272 |
if (isset($params['meta_key'])) { |
| 273 |
$meta_key = $this->sanitize_string($params['meta_key']); |
| 274 |
$meta_value = get_post_meta($post_id, $meta_key, true); |
| 275 |
|
| 276 |
return $this->success([ |
| 277 |
'post_id' => $post_id, |
| 278 |
'post_title' => $post->post_title, |
| 279 |
'meta_key' => $meta_key, |
| 280 |
'meta_value' => $meta_value |
| 281 |
]); |
| 282 |
} else { |
| 283 |
// Get all SEO meta |
| 284 |
$seo_meta = [ |
| 285 |
'metatitle' => get_post_meta($post_id, '_metasync_metatitle', true), |
| 286 |
'metadesc' => get_post_meta($post_id, '_metasync_metadesc', true), |
| 287 |
'focus_keyword' => get_post_meta($post_id, '_metasync_focus_keyword', true), |
| 288 |
'robots_index' => get_post_meta($post_id, '_metasync_robots_index', true), |
| 289 |
'canonical_url' => get_post_meta($post_id, '_metasync_canonical_url', true) |
| 290 |
]; |
| 291 |
|
| 292 |
// Get Open Graph meta |
| 293 |
$opengraph_meta = [ |
| 294 |
'og_enabled' => get_post_meta($post_id, '_metasync_og_enabled', true), |
| 295 |
'og_title' => get_post_meta($post_id, '_metasync_og_title', true), |
| 296 |
'og_description' => get_post_meta($post_id, '_metasync_og_description', true), |
| 297 |
'og_image' => get_post_meta($post_id, '_metasync_og_image', true), |
| 298 |
'og_url' => get_post_meta($post_id, '_metasync_og_url', true), |
| 299 |
'og_type' => get_post_meta($post_id, '_metasync_og_type', true) |
| 300 |
]; |
| 301 |
|
| 302 |
// Get Twitter Card meta |
| 303 |
$twitter_meta = [ |
| 304 |
'twitter_card' => get_post_meta($post_id, '_metasync_twitter_card', true), |
| 305 |
'twitter_title' => get_post_meta($post_id, '_metasync_twitter_title', true), |
| 306 |
'twitter_description' => get_post_meta($post_id, '_metasync_twitter_description', true), |
| 307 |
]; |
| 308 |
|
| 309 |
return $this->success([ |
| 310 |
'post_id' => $post_id, |
| 311 |
'post_title' => $post->post_title, |
| 312 |
'post_type' => $post->post_type, |
| 313 |
'post_status' => $post->post_status, |
| 314 |
'seo_meta' => $seo_meta, |
| 315 |
'opengraph_meta' => $opengraph_meta, |
| 316 |
'twitter_meta' => $twitter_meta |
| 317 |
]); |
| 318 |
} |
| 319 |
} |
| 320 |
} |
| 321 |
|
| 322 |
/** |
| 323 |
* Get SEO Meta Tool |
| 324 |
*/ |
| 325 |
class MCP_Tool_Get_SEO_Meta extends MCP_Tool_Base { |
| 326 |
|
| 327 |
public function get_name() { |
| 328 |
return 'wordpress_get_seo_meta'; |
| 329 |
} |
| 330 |
|
| 331 |
public function get_description() { |
| 332 |
return 'Get all SEO-related metadata for a post including title, description, keywords, indexing settings, and Open Graph/social meta'; |
| 333 |
} |
| 334 |
|
| 335 |
public function get_input_schema() { |
| 336 |
return [ |
| 337 |
'type' => 'object', |
| 338 |
'properties' => [ |
| 339 |
'post_id' => [ |
| 340 |
'type' => 'integer', |
| 341 |
'description' => 'WordPress post or page ID', |
| 342 |
'minimum' => 1 |
| 343 |
] |
| 344 |
], |
| 345 |
'required' => ['post_id'] |
| 346 |
]; |
| 347 |
} |
| 348 |
|
| 349 |
public function execute($params) { |
| 350 |
$this->validate_params($params); |
| 351 |
$this->require_capability('read'); |
| 352 |
|
| 353 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 354 |
|
| 355 |
// Verify post exists |
| 356 |
$post = $this->verify_post_exists($post_id); |
| 357 |
|
| 358 |
// Get all SEO meta |
| 359 |
$seo_data = [ |
| 360 |
'post_info' => [ |
| 361 |
'id' => $post_id, |
| 362 |
'title' => $post->post_title, |
| 363 |
'type' => $post->post_type, |
| 364 |
'status' => $post->post_status, |
| 365 |
'url' => get_permalink($post_id) |
| 366 |
], |
| 367 |
'seo_meta' => [ |
| 368 |
'meta_title' => get_post_meta($post_id, '_metasync_metatitle', true), |
| 369 |
'meta_description' => get_post_meta($post_id, '_metasync_metadesc', true), |
| 370 |
'focus_keyword' => get_post_meta($post_id, '_metasync_focus_keyword', true), |
| 371 |
'robots_index' => get_post_meta($post_id, '_metasync_robots_index', true), |
| 372 |
'canonical_url' => get_post_meta($post_id, '_metasync_canonical_url', true), |
| 373 |
'robots_advanced' => json_decode(get_post_meta($post_id, '_metasync_robots_advanced', true) ?: '{}', true), |
| 374 |
'primary_category' => $this->get_primary_category_data($post_id), |
| 375 |
], |
| 376 |
'opengraph_meta' => [ |
| 377 |
'enabled' => get_post_meta($post_id, '_metasync_og_enabled', true), |
| 378 |
'title' => get_post_meta($post_id, '_metasync_og_title', true), |
| 379 |
'description' => get_post_meta($post_id, '_metasync_og_description', true), |
| 380 |
'image' => get_post_meta($post_id, '_metasync_og_image', true), |
| 381 |
'url' => get_post_meta($post_id, '_metasync_og_url', true), |
| 382 |
'type' => get_post_meta($post_id, '_metasync_og_type', true) |
| 383 |
], |
| 384 |
'twitter_meta' => [ |
| 385 |
'twitter_card' => get_post_meta($post_id, '_metasync_twitter_card', true), |
| 386 |
'twitter_title' => get_post_meta($post_id, '_metasync_twitter_title', true), |
| 387 |
'twitter_description' => get_post_meta($post_id, '_metasync_twitter_description', true), |
| 388 |
], |
| 389 |
'analysis' => [ |
| 390 |
'meta_title_length' => mb_strlen(get_post_meta($post_id, '_metasync_metatitle', true)), |
| 391 |
'meta_desc_length' => mb_strlen(get_post_meta($post_id, '_metasync_metadesc', true)), |
| 392 |
'has_focus_keyword' => !empty(get_post_meta($post_id, '_metasync_focus_keyword', true)), |
| 393 |
'is_indexable' => get_post_meta($post_id, '_metasync_robots_index', true) !== 'noindex', |
| 394 |
'og_enabled' => get_post_meta($post_id, '_metasync_og_enabled', true) === '1', |
| 395 |
'has_og_image' => !empty(get_post_meta($post_id, '_metasync_og_image', true)), |
| 396 |
'has_robots_advanced' => !empty(get_post_meta($post_id, '_metasync_robots_advanced', true)) |
| 397 |
] |
| 398 |
]; |
| 399 |
|
| 400 |
return $this->success($seo_data); |
| 401 |
} |
| 402 |
|
| 403 |
/** |
| 404 |
* Get primary category data for a post |
| 405 |
* |
| 406 |
* @param int $post_id The post ID |
| 407 |
* @return array|null Primary category data or null if not set |
| 408 |
*/ |
| 409 |
private function get_primary_category_data($post_id) { |
| 410 |
$primary_cat_id = (int) get_post_meta($post_id, '_metasync_primary_category', true); |
| 411 |
|
| 412 |
if ($primary_cat_id === 0) { |
| 413 |
return null; |
| 414 |
} |
| 415 |
|
| 416 |
$term = get_term($primary_cat_id, 'category'); |
| 417 |
if (!$term || is_wp_error($term)) { |
| 418 |
return null; |
| 419 |
} |
| 420 |
|
| 421 |
return [ |
| 422 |
'term_id' => $primary_cat_id, |
| 423 |
'name' => $term->name, |
| 424 |
'slug' => $term->slug, |
| 425 |
]; |
| 426 |
} |
| 427 |
} |
| 428 |
|
| 429 |
/** |
| 430 |
* Get Hreflang Links Tool |
| 431 |
* |
| 432 |
* Returns all hreflang entries for a post (manual + WPML auto-detected), |
| 433 |
* validates that each referenced URL returns HTTP 200 and links back to |
| 434 |
* this post (hreflang reciprocity), and flags a missing x-default entry |
| 435 |
* and a missing self-reference. |
| 436 |
*/ |
| 437 |
class MCP_Tool_Get_Hreflang_Links extends MCP_Tool_Base { |
| 438 |
|
| 439 |
public function get_name() { |
| 440 |
return 'wordpress_get_hreflang_links'; |
| 441 |
} |
| 442 |
|
| 443 |
public function get_description() { |
| 444 |
return 'Get all hreflang entries for a post (manual + WPML auto-detected), validate each URL returns HTTP 200 and links back to this post (reciprocity), and flag missing x-default / self-reference entries'; |
| 445 |
} |
| 446 |
|
| 447 |
public function get_input_schema() { |
| 448 |
return [ |
| 449 |
'type' => 'object', |
| 450 |
'properties' => [ |
| 451 |
'post_id' => [ |
| 452 |
'type' => 'integer', |
| 453 |
'description' => 'WordPress post or page ID', |
| 454 |
'minimum' => 1 |
| 455 |
] |
| 456 |
], |
| 457 |
'required' => ['post_id'] |
| 458 |
]; |
| 459 |
} |
| 460 |
|
| 461 |
public function execute($params) { |
| 462 |
$this->validate_params($params); |
| 463 |
$this->require_capability('read'); |
| 464 |
|
| 465 |
$post_id = $this->sanitize_integer($params['post_id']); |
| 466 |
$post = $this->verify_post_exists($post_id); |
| 467 |
$this->check_post_permission($post_id); |
| 468 |
|
| 469 |
$manual_entries = $this->get_manual_entries($post_id); |
| 470 |
$wpml_entries = $this->get_wpml_entries($post_id, $post); |
| 471 |
|
| 472 |
// Merge: auto-detected first, then manual entries override by the |
| 473 |
// normalised language-code collision key. Rows with an invalid code |
| 474 |
// (empty key) are kept verbatim instead of collapsing into one — |
| 475 |
// the audit report is where they become visible. |
| 476 |
$by_key = []; |
| 477 |
$invalid = []; |
| 478 |
foreach ($wpml_entries as $entry) { |
| 479 |
$key = Metasync_Hreflang_Output::lang_code($entry); |
| 480 |
if ($key === '') { |
| 481 |
$invalid[] = $entry; |
| 482 |
} else { |
| 483 |
$by_key[$key] = $entry; |
| 484 |
} |
| 485 |
} |
| 486 |
foreach ($manual_entries as $entry) { |
| 487 |
$key = Metasync_Hreflang_Output::lang_code($entry); |
| 488 |
if ($key === '') { |
| 489 |
$invalid[] = $entry; |
| 490 |
} else { |
| 491 |
$by_key[$key] = $entry; |
| 492 |
} |
| 493 |
} |
| 494 |
$entries = array_merge(array_values($by_key), $invalid); |
| 495 |
|
| 496 |
$self_url = get_permalink($post_id); |
| 497 |
|
| 498 |
// Validate each URL: HTTP 200 (HEAD first, GET on 405) and — when |
| 499 |
// the target is an external alternate — a return link to this post |
| 500 |
// (hreflang must be bidirectional; Google ignores one-way clusters). |
| 501 |
$has_x_default = false; |
| 502 |
$has_self_reference = false; |
| 503 |
foreach ($entries as &$entry) { |
| 504 |
$url = isset($entry['url']) ? $entry['url'] : ''; |
| 505 |
$status = null; |
| 506 |
$error = null; |
| 507 |
$return_link = null; |
| 508 |
|
| 509 |
if (!empty($url)) { |
| 510 |
$response = wp_remote_head($url, ['timeout' => 5, 'sslverify' => false]); |
| 511 |
if (is_wp_error($response)) { |
| 512 |
$error = $response->get_error_message(); |
| 513 |
} else { |
| 514 |
$status = (int) wp_remote_retrieve_response_code($response); |
| 515 |
if ($status === 405) { |
| 516 |
$response = wp_remote_get($url, ['timeout' => 5, 'sslverify' => false]); |
| 517 |
if (is_wp_error($response)) { |
| 518 |
$error = $response->get_error_message(); |
| 519 |
$status = null; |
| 520 |
} else { |
| 521 |
$status = (int) wp_remote_retrieve_response_code($response); |
| 522 |
} |
| 523 |
} |
| 524 |
} |
| 525 |
|
| 526 |
// Reciprocity: fetch the alternate page and check whether |
| 527 |
// its hreflang set contains this post's URL. |
| 528 |
if ($error === null && $status === 200 && !empty($self_url) && $url !== $self_url) { |
| 529 |
$response = wp_remote_get($url, ['timeout' => 5, 'sslverify' => false]); |
| 530 |
if (!is_wp_error($response)) { |
| 531 |
$body = (string) wp_remote_retrieve_body($response); |
| 532 |
$return_link = self::body_links_back($body, $self_url); |
| 533 |
} |
| 534 |
} |
| 535 |
} |
| 536 |
|
| 537 |
$entry['http_status'] = $status; |
| 538 |
$entry['http_ok'] = ($status === 200); |
| 539 |
if ($error !== null) { |
| 540 |
$entry['http_error'] = $error; |
| 541 |
} |
| 542 |
if ($return_link !== null) { |
| 543 |
$entry['return_link'] = $return_link; |
| 544 |
} |
| 545 |
|
| 546 |
if (Metasync_Hreflang_Output::lang_code($entry) === 'x-default') { |
| 547 |
$has_x_default = true; |
| 548 |
} |
| 549 |
if (!empty($self_url) && $url === $self_url) { |
| 550 |
$has_self_reference = true; |
| 551 |
} |
| 552 |
} |
| 553 |
unset($entry); |
| 554 |
|
| 555 |
return $this->success([ |
| 556 |
'post_id' => $post_id, |
| 557 |
'post_title' => $post->post_title, |
| 558 |
'self_url' => $self_url, |
| 559 |
'entries' => $entries, |
| 560 |
'missing_x_default' => !$has_x_default, |
| 561 |
'missing_self_reference' => !$has_self_reference, |
| 562 |
]); |
| 563 |
} |
| 564 |
|
| 565 |
/** |
| 566 |
* True when the page body contains a hreflang link pointing back at the |
| 567 |
* given URL. |
| 568 |
* |
| 569 |
* @param string $body HTML body of the alternate page. |
| 570 |
* @param string $self_url This post's permalink. |
| 571 |
* @return bool |
| 572 |
*/ |
| 573 |
private static function body_links_back($body, $self_url) { |
| 574 |
if (!preg_match_all('/<link[^>]+hreflang\s*=\s*[\'"]([^\'"]*)[\'"][^>]*>/i', (string) $body, $links)) { |
| 575 |
return false; |
| 576 |
} |
| 577 |
foreach ($links[0] as $tag) { |
| 578 |
if (stripos($tag, 'rel="alternate"') === false && stripos($tag, "rel='alternate'") === false) { |
| 579 |
continue; |
| 580 |
} |
| 581 |
if (strpos($tag, $self_url) !== false) { |
| 582 |
return true; |
| 583 |
} |
| 584 |
} |
| 585 |
return false; |
| 586 |
} |
| 587 |
|
| 588 |
/** |
| 589 |
* Read manual hreflang entries from `_metasync_hreflang` post meta. |
| 590 |
* |
| 591 |
* Rows are normalised and carry their final lang_code plus a valid flag, |
| 592 |
* so an invalid stored row is visible in the report (the emitter skips |
| 593 |
* it silently — this is the tool that surfaces it). |
| 594 |
* |
| 595 |
* @param int $post_id Post ID. |
| 596 |
* @return array |
| 597 |
*/ |
| 598 |
private function get_manual_entries($post_id) { |
| 599 |
$raw = get_post_meta($post_id, '_metasync_hreflang', true); |
| 600 |
if (empty($raw) || !is_string($raw)) { |
| 601 |
return []; |
| 602 |
} |
| 603 |
$decoded = json_decode($raw, true); |
| 604 |
if (!is_array($decoded)) { |
| 605 |
return []; |
| 606 |
} |
| 607 |
$entries = []; |
| 608 |
foreach ($decoded as $entry) { |
| 609 |
if (!is_array($entry)) { |
| 610 |
continue; |
| 611 |
} |
| 612 |
$normalized = Metasync_Hreflang_Output::normalize_entry($entry); |
| 613 |
$entries[] = [ |
| 614 |
'lang' => $normalized['lang'], |
| 615 |
'region' => $normalized['region'], |
| 616 |
'url' => $normalized['url'], |
| 617 |
'lang_code' => $normalized['lang_code'], |
| 618 |
'valid' => $normalized['valid'], |
| 619 |
'source' => 'manual', |
| 620 |
]; |
| 621 |
} |
| 622 |
return $entries; |
| 623 |
} |
| 624 |
|
| 625 |
/** |
| 626 |
* Build WPML auto-detected entries via the shared implementation in |
| 627 |
* Metasync_Hreflang_Output::get_wpml_entries() (publish-only rows, |
| 628 |
* x-default synthesis) so the audit reports exactly what the front end |
| 629 |
* emits. |
| 630 |
* |
| 631 |
* @param int $post_id Post ID. |
| 632 |
* @param WP_Post $post Post object (already verified). |
| 633 |
* @return array |
| 634 |
*/ |
| 635 |
private function get_wpml_entries($post_id, $post) { |
| 636 |
if (!defined('ICL_SITEPRESS_VERSION')) { |
| 637 |
return []; |
| 638 |
} |
| 639 |
|
| 640 |
$entries = (new Metasync_Hreflang_Output())->get_wpml_entries($post_id); |
| 641 |
|
| 642 |
$out = []; |
| 643 |
foreach ($entries as $entry) { |
| 644 |
$normalized = Metasync_Hreflang_Output::normalize_entry($entry); |
| 645 |
$normalized['source'] = 'wpml'; |
| 646 |
$out[] = $normalized; |
| 647 |
} |
| 648 |
return $out; |
| 649 |
} |
| 650 |
} |
| 651 |
|
