| 1 |
<?php |
| 2 |
/** |
| 3 |
* MetaSync Internal Link Suggestions |
| 4 |
* |
| 5 |
* Provides a REST API endpoint that returns relevant internal link suggestions |
| 6 |
* based on content similarity for the Gutenberg editor sidebar. |
| 7 |
* |
| 8 |
* @package Metasync |
| 9 |
* @subpackage Metasync/admin |
| 10 |
* @since 2.8.0 |
| 11 |
*/ |
| 12 |
|
| 13 |
if (!defined('ABSPATH')) { |
| 14 |
exit; |
| 15 |
} |
| 16 |
|
| 17 |
class Metasync_Link_Suggestions { |
| 18 |
|
| 19 |
/** |
| 20 |
* Common stop words to filter out from bigram matching |
| 21 |
*/ |
| 22 |
private static $stop_words = array( |
| 23 |
'the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', |
| 24 |
'of', 'with', 'by', 'from', 'is', 'it', 'as', 'be', 'was', 'are', |
| 25 |
'been', 'has', 'have', 'had', 'do', 'does', 'did', 'will', 'would', |
| 26 |
'could', 'should', 'may', 'might', 'can', 'this', 'that', 'these', |
| 27 |
'those', 'not', 'no', 'so', 'if', 'how', 'what', 'when', 'where', |
| 28 |
'who', 'which', 'why', 'all', 'each', 'every', 'both', 'few', 'more', |
| 29 |
'most', 'other', 'some', 'such', 'than', 'too', 'very', 'just', 'about', |
| 30 |
'above', 'after', 'again', 'also', 'any', 'because', 'before', 'between', |
| 31 |
'into', 'its', 'our', 'out', 'own', 'same', 'she', 'he', 'her', 'his', |
| 32 |
'him', 'my', 'your', 'they', 'them', 'their', 'we', 'you', 'up', 'get', |
| 33 |
); |
| 34 |
|
| 35 |
/** |
| 36 |
* Constructor |
| 37 |
*/ |
| 38 |
public function __construct() { |
| 39 |
$this->init(); |
| 40 |
} |
| 41 |
|
| 42 |
/** |
| 43 |
* Initialize hooks |
| 44 |
*/ |
| 45 |
private function init() { |
| 46 |
add_action('rest_api_init', array($this, 'register_rest_routes')); |
| 47 |
add_action('save_post', array($this, 'invalidate_cache'), 10, 2); |
| 48 |
} |
| 49 |
|
| 50 |
/** |
| 51 |
* Register REST API routes |
| 52 |
*/ |
| 53 |
public function register_rest_routes() { |
| 54 |
register_rest_route('metasync/v1', '/link-suggestions', array( |
| 55 |
'methods' => 'GET', |
| 56 |
'callback' => array($this, 'get_suggestions'), |
| 57 |
'permission_callback' => array($this, 'permission_check'), |
| 58 |
'args' => array( |
| 59 |
'post_id' => array( |
| 60 |
'required' => true, |
| 61 |
'type' => 'integer', |
| 62 |
'sanitize_callback' => 'absint', |
| 63 |
), |
| 64 |
'limit' => array( |
| 65 |
'default' => 10, |
| 66 |
'type' => 'integer', |
| 67 |
'sanitize_callback' => 'absint', |
| 68 |
), |
| 69 |
), |
| 70 |
)); |
| 71 |
} |
| 72 |
|
| 73 |
/** |
| 74 |
* Permission callback - requires edit_posts capability |
| 75 |
* |
| 76 |
* @return bool |
| 77 |
*/ |
| 78 |
public function permission_check() { |
| 79 |
return current_user_can('edit_posts'); |
| 80 |
} |
| 81 |
|
| 82 |
/** |
| 83 |
* Check if a bigram contains only stop words |
| 84 |
* |
| 85 |
* @param string $bigram Two-word phrase |
| 86 |
* @return bool True if the bigram is too generic |
| 87 |
*/ |
| 88 |
private function is_generic_bigram($bigram) { |
| 89 |
$words = preg_split('/\s+/', strtolower(trim($bigram))); |
| 90 |
if (count($words) < 2) { |
| 91 |
return true; |
| 92 |
} |
| 93 |
// Generic if BOTH words are stop words |
| 94 |
$word1_is_stop = in_array($words[0], self::$stop_words, true); |
| 95 |
$word2_is_stop = in_array($words[1], self::$stop_words, true); |
| 96 |
return ($word1_is_stop && $word2_is_stop); |
| 97 |
} |
| 98 |
|
| 99 |
/** |
| 100 |
* Calculate relevance score for a suggestion |
| 101 |
* |
| 102 |
* @param string $match_type Type of match (full_title, focus_keyword, bigram, tag) |
| 103 |
* @param string $matched_phrase The phrase that matched |
| 104 |
* @param string $candidate_title The candidate post title |
| 105 |
* @return float Score between 0 and 1 |
| 106 |
*/ |
| 107 |
private function calculate_relevance_score($match_type, $matched_phrase, $candidate_title) { |
| 108 |
switch ($match_type) { |
| 109 |
case 'full_title': |
| 110 |
return 1.0; |
| 111 |
case 'focus_keyword': |
| 112 |
return 0.9; |
| 113 |
case 'tag': |
| 114 |
// Longer tag matches are more relevant |
| 115 |
$len = strlen($matched_phrase); |
| 116 |
return $len > 10 ? 0.7 : 0.6; |
| 117 |
case 'bigram': |
| 118 |
// Bigrams from longer titles that are more specific score higher |
| 119 |
$title_word_count = str_word_count($candidate_title); |
| 120 |
if ($title_word_count <= 3) { |
| 121 |
return 0.5; // Short title = bigram covers most of it |
| 122 |
} |
| 123 |
return 0.4; |
| 124 |
default: |
| 125 |
return 0.3; |
| 126 |
} |
| 127 |
} |
| 128 |
|
| 129 |
/** |
| 130 |
* Get link suggestions for a post |
| 131 |
* |
| 132 |
* @param WP_REST_Request $request |
| 133 |
* @return WP_REST_Response |
| 134 |
*/ |
| 135 |
public function get_suggestions($request) { |
| 136 |
$post_id = intval($request->get_param('post_id')); |
| 137 |
$limit = min(intval($request->get_param('limit')), 10); |
| 138 |
|
| 139 |
if ($limit < 1) { |
| 140 |
$limit = 10; |
| 141 |
} |
| 142 |
|
| 143 |
// Check transient cache |
| 144 |
$cache_key = 'metasync_link_suggestions_' . $post_id; |
| 145 |
$cached = get_transient($cache_key); |
| 146 |
if ($cached !== false) { |
| 147 |
return new WP_REST_Response($cached, 200); |
| 148 |
} |
| 149 |
|
| 150 |
// Verify the user can read this specific post (prevents IDOR) |
| 151 |
if (!current_user_can('read_post', $post_id)) { |
| 152 |
return new WP_REST_Response(array( |
| 153 |
'suggestions' => array(), |
| 154 |
'yoast_premium_active' => class_exists('WPSEO_Premium'), |
| 155 |
'rankmath_active' => defined('RANK_MATH_VERSION'), |
| 156 |
), 200); |
| 157 |
} |
| 158 |
|
| 159 |
$post = get_post($post_id); |
| 160 |
if (!$post) { |
| 161 |
return new WP_REST_Response(array( |
| 162 |
'suggestions' => array(), |
| 163 |
'yoast_premium_active' => class_exists('WPSEO_Premium'), |
| 164 |
'rankmath_active' => defined('RANK_MATH_VERSION'), |
| 165 |
), 200); |
| 166 |
} |
| 167 |
|
| 168 |
// Custom/LPS pages carry self-managed SEO in their own HTML bundle; |
| 169 |
// bigram-scanning their raw HTML produces noise, so skip them entirely. |
| 170 |
if (metasync_is_custom_or_lps_page($post_id)) { |
| 171 |
return new WP_REST_Response(array( |
| 172 |
'suggestions' => array(), |
| 173 |
'yoast_premium_active' => class_exists('WPSEO_Premium'), |
| 174 |
'rankmath_active' => defined('RANK_MATH_VERSION'), |
| 175 |
), 200); |
| 176 |
} |
| 177 |
|
| 178 |
// Strip block comments and HTML tags to get clean text for matching |
| 179 |
$content = $post->post_content; |
| 180 |
$content_text = wp_strip_all_tags(strip_shortcodes($content)); |
| 181 |
|
| 182 |
// Build list of already-linked URLs |
| 183 |
$already_linked = array(); |
| 184 |
if (preg_match_all('/<a\s[^>]*href=["\']([^"\']+)["\'][^>]*>/i', $content, $matches)) { |
| 185 |
$already_linked = array_map('untrailingslashit', $matches[1]); |
| 186 |
} |
| 187 |
|
| 188 |
// Get public post types |
| 189 |
$public_post_types = get_post_types(array('public' => true), 'names'); |
| 190 |
|
| 191 |
// Extract significant words from content for search query |
| 192 |
$content_words = array_filter( |
| 193 |
str_word_count(strtolower($content_text), 1), |
| 194 |
function($word) { |
| 195 |
return strlen($word) >= 4 && !in_array($word, self::$stop_words, true); |
| 196 |
} |
| 197 |
); |
| 198 |
// Get top frequent words for search |
| 199 |
$word_freq = array_count_values($content_words); |
| 200 |
arsort($word_freq); |
| 201 |
$search_terms = implode(' ', array_slice(array_keys($word_freq), 0, 5)); |
| 202 |
|
| 203 |
// Two-pass query approach: |
| 204 |
// Pass 1: Search by content relevance using WordPress search |
| 205 |
// Pass 2: Recent posts as fallback |
| 206 |
$candidate_ids = array(); |
| 207 |
|
| 208 |
// Pass 1: Relevance-based search |
| 209 |
if (!empty($search_terms)) { |
| 210 |
$search_query = new WP_Query(array( |
| 211 |
'post__not_in' => array($post_id), |
| 212 |
'post_status' => 'publish', |
| 213 |
'posts_per_page' => 100, |
| 214 |
'post_type' => array_values($public_post_types), |
| 215 |
's' => $search_terms, |
| 216 |
'no_found_rows' => true, |
| 217 |
'fields' => 'ids', |
| 218 |
'meta_query' => array( |
| 219 |
'relation' => 'OR', |
| 220 |
array( |
| 221 |
'key' => '_metasync_robots_index', |
| 222 |
'compare' => 'NOT EXISTS', |
| 223 |
), |
| 224 |
array( |
| 225 |
'key' => '_metasync_robots_index', |
| 226 |
'value' => 'noindex', |
| 227 |
'compare' => '!=', |
| 228 |
), |
| 229 |
), |
| 230 |
)); |
| 231 |
$candidate_ids = $search_query->posts; |
| 232 |
} |
| 233 |
|
| 234 |
// Pass 2: Fill up to 200 candidates with recent posts |
| 235 |
$existing_count = count($candidate_ids); |
| 236 |
if ($existing_count < 200) { |
| 237 |
$exclude_ids = array_merge(array($post_id), $candidate_ids); |
| 238 |
$recent_query = new WP_Query(array( |
| 239 |
'post__not_in' => $exclude_ids, |
| 240 |
'post_status' => 'publish', |
| 241 |
'posts_per_page' => 200 - $existing_count, |
| 242 |
'post_type' => array_values($public_post_types), |
| 243 |
'orderby' => 'modified', |
| 244 |
'order' => 'DESC', |
| 245 |
'no_found_rows' => true, |
| 246 |
'fields' => 'ids', |
| 247 |
'meta_query' => array( |
| 248 |
'relation' => 'OR', |
| 249 |
array( |
| 250 |
'key' => '_metasync_robots_index', |
| 251 |
'compare' => 'NOT EXISTS', |
| 252 |
), |
| 253 |
array( |
| 254 |
'key' => '_metasync_robots_index', |
| 255 |
'value' => 'noindex', |
| 256 |
'compare' => '!=', |
| 257 |
), |
| 258 |
), |
| 259 |
)); |
| 260 |
$candidate_ids = array_merge($candidate_ids, $recent_query->posts); |
| 261 |
} |
| 262 |
|
| 263 |
if (empty($candidate_ids)) { |
| 264 |
$data = array( |
| 265 |
'suggestions' => array(), |
| 266 |
'yoast_premium_active' => class_exists('WPSEO_Premium'), |
| 267 |
'rankmath_active' => defined('RANK_MATH_VERSION'), |
| 268 |
); |
| 269 |
set_transient($cache_key, $data, HOUR_IN_SECONDS); |
| 270 |
return new WP_REST_Response($data, 200); |
| 271 |
} |
| 272 |
|
| 273 |
// Batch pre-fetch: prime the post meta cache for all candidates at once |
| 274 |
update_meta_cache('post', $candidate_ids); |
| 275 |
|
| 276 |
// Batch pre-fetch: prime the term cache for all candidates (tags) |
| 277 |
update_object_term_cache($candidate_ids, array_values($public_post_types)); |
| 278 |
|
| 279 |
// Now loop through candidates - meta and term queries will hit cache, not DB |
| 280 |
$suggestions = array(); |
| 281 |
|
| 282 |
foreach ($candidate_ids as $candidate_id) { |
| 283 |
$candidate_post = get_post($candidate_id); |
| 284 |
if (!$candidate_post) { |
| 285 |
continue; |
| 286 |
} |
| 287 |
|
| 288 |
$candidate_title = $candidate_post->post_title; |
| 289 |
$candidate_url = get_permalink($candidate_id); |
| 290 |
|
| 291 |
// Skip if already linked |
| 292 |
if (in_array(untrailingslashit($candidate_url), $already_linked, true)) { |
| 293 |
continue; |
| 294 |
} |
| 295 |
|
| 296 |
// Build match phrases with types for relevance scoring |
| 297 |
// Priority: full title > focus keyword > tags > bigrams |
| 298 |
$match_entries = array(); |
| 299 |
|
| 300 |
// Full title match (highest priority) |
| 301 |
if (strlen($candidate_title) >= 3) { |
| 302 |
$match_entries[] = array('phrase' => $candidate_title, 'type' => 'full_title'); |
| 303 |
} |
| 304 |
|
| 305 |
// Focus keyword match |
| 306 |
$candidate_focus = get_post_meta($candidate_id, '_metasync_focus_keyword', true); |
| 307 |
if (!empty($candidate_focus) && strlen(trim($candidate_focus)) >= 3) { |
| 308 |
$match_entries[] = array('phrase' => trim($candidate_focus), 'type' => 'focus_keyword'); |
| 309 |
} |
| 310 |
|
| 311 |
// Tag matches |
| 312 |
$candidate_tags = wp_get_post_tags($candidate_id, array('fields' => 'names')); |
| 313 |
if (!empty($candidate_tags) && !is_wp_error($candidate_tags)) { |
| 314 |
foreach ($candidate_tags as $tag_name) { |
| 315 |
if (strlen($tag_name) >= 3) { |
| 316 |
$match_entries[] = array('phrase' => $tag_name, 'type' => 'tag'); |
| 317 |
} |
| 318 |
} |
| 319 |
} |
| 320 |
|
| 321 |
// Bigram matches from title (lowest priority, filtered) |
| 322 |
$title_words = preg_split('/\s+/', $candidate_title); |
| 323 |
if (count($title_words) >= 2) { |
| 324 |
for ($i = 0; $i < count($title_words) - 1; $i++) { |
| 325 |
$bigram = $title_words[$i] . ' ' . $title_words[$i + 1]; |
| 326 |
// Skip bigrams that are too short or too generic |
| 327 |
if (strlen($bigram) < 6 || $this->is_generic_bigram($bigram)) { |
| 328 |
continue; |
| 329 |
} |
| 330 |
$match_entries[] = array('phrase' => $bigram, 'type' => 'bigram'); |
| 331 |
} |
| 332 |
} |
| 333 |
|
| 334 |
// Check if any phrase appears in the current post content |
| 335 |
$matched_phrase = null; |
| 336 |
$match_type = null; |
| 337 |
foreach ($match_entries as $entry) { |
| 338 |
if (stripos($content_text, $entry['phrase']) !== false) { |
| 339 |
$matched_phrase = $entry['phrase']; |
| 340 |
$match_type = $entry['type']; |
| 341 |
break; |
| 342 |
} |
| 343 |
} |
| 344 |
|
| 345 |
if ($matched_phrase !== null) { |
| 346 |
$relevance = $this->calculate_relevance_score($match_type, $matched_phrase, $candidate_title); |
| 347 |
$suggestions[] = array( |
| 348 |
'post_id' => $candidate_id, |
| 349 |
'title' => $candidate_title, |
| 350 |
'url' => $candidate_url, |
| 351 |
'matched_phrase' => $matched_phrase, |
| 352 |
'relevance_score' => $relevance, |
| 353 |
); |
| 354 |
} |
| 355 |
} |
| 356 |
|
| 357 |
// Sort by relevance score descending |
| 358 |
usort($suggestions, function($a, $b) { |
| 359 |
return $b['relevance_score'] <=> $a['relevance_score']; |
| 360 |
}); |
| 361 |
|
| 362 |
// Trim to limit |
| 363 |
$suggestions = array_slice($suggestions, 0, $limit); |
| 364 |
|
| 365 |
// Cross-plugin detection |
| 366 |
$yoast_active = class_exists('WPSEO_Premium'); |
| 367 |
$rankmath_active = defined('RANK_MATH_VERSION'); |
| 368 |
|
| 369 |
$data = array( |
| 370 |
'suggestions' => $suggestions, |
| 371 |
'yoast_premium_active' => $yoast_active, |
| 372 |
'rankmath_active' => $rankmath_active, |
| 373 |
); |
| 374 |
|
| 375 |
set_transient($cache_key, $data, HOUR_IN_SECONDS); |
| 376 |
|
| 377 |
return new WP_REST_Response($data, 200); |
| 378 |
} |
| 379 |
|
| 380 |
/** |
| 381 |
* Invalidate link suggestions cache on post save |
| 382 |
* Only for actual published posts, not revisions or autosaves |
| 383 |
* |
| 384 |
* @param int $post_id |
| 385 |
* @param WP_Post $post |
| 386 |
*/ |
| 387 |
public function invalidate_cache($post_id, $post = null) { |
| 388 |
// Skip revisions and autosaves |
| 389 |
if (wp_is_post_revision($post_id) || wp_is_post_autosave($post_id)) { |
| 390 |
return; |
| 391 |
} |
| 392 |
|
| 393 |
// Only invalidate for published posts |
| 394 |
if ($post && $post->post_status !== 'publish') { |
| 395 |
return; |
| 396 |
} |
| 397 |
|
| 398 |
delete_transient('metasync_link_suggestions_' . $post_id); |
| 399 |
} |
| 400 |
} |
| 401 |
|
