| 1 |
<?php |
| 2 |
/** |
| 3 |
* This handles the OTTO SSR |
| 4 |
*/ |
| 5 |
if ( ! defined( 'ABSPATH' ) ) { |
| 6 |
exit; |
| 7 |
} |
| 8 |
|
| 9 |
if (!defined('METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY')) { |
| 10 |
define('METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY', 'metasync_otto_excluded_urls_manual'); |
| 11 |
} |
| 12 |
if (!defined('METASYNC_OTTO_EXCLUDED_TRANSIENT_TTL')) { |
| 13 |
define('METASYNC_OTTO_EXCLUDED_TRANSIENT_TTL', 60); |
| 14 |
} |
| 15 |
|
| 16 |
# uses the simple html dom library |
| 17 |
use simplehtmldom\HtmlDocument; |
| 18 |
|
| 19 |
# include the otto class file |
| 20 |
require_once plugin_dir_path( __FILE__ ) . '/vendor/autoload.php'; |
| 21 |
require_once plugin_dir_path( __FILE__ ) . '/Otto_html_class.php'; |
| 22 |
require_once plugin_dir_path( __FILE__ ) . '/Otto_pixel_class.php'; |
| 23 |
require_once plugin_dir_path( __FILE__ ) . '/metasync-otto-seo-functions.php'; |
| 24 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-transient-cache.php'; |
| 25 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-render-strategy.php'; |
| 26 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-config.php'; |
| 27 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-bot-detector.php'; |
| 28 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-bot-statistics-database.php'; |
| 29 |
|
| 30 |
# OPTIMIZED: get the metasync options (cached in static class) |
| 31 |
$metasync_options = Metasync_Otto_Config::get_options(); |
| 32 |
|
| 33 |
# OTTO SSR is always enabled by default |
| 34 |
$otto_enabled = true; |
| 35 |
|
| 36 |
# add tag to wp head |
| 37 |
add_action('wp_head', function(){ |
| 38 |
# load globals |
| 39 |
global $metasync_options, $otto_enabled; |
| 40 |
|
| 41 |
# OTTO SSR is always enabled |
| 42 |
$string_enabled = 'true'; |
| 43 |
|
| 44 |
# OPTIMIZED: check uuid set using cached config |
| 45 |
if(!Metasync_Otto_Config::is_otto_enabled()){ |
| 46 |
return; |
| 47 |
} |
| 48 |
|
| 49 |
# Performance optimization: Add DNS prefetch and preconnect for OTTO API |
| 50 |
# This improves connection speed by resolving DNS and establishing connections early |
| 51 |
# Use endpoint manager to get the correct domain |
| 52 |
$otto_domain = 'sa.searchatlas.com'; # default |
| 53 |
if (class_exists('Metasync_Endpoint_Manager')) { |
| 54 |
$otto_api_domain = Metasync_Endpoint_Manager::get_endpoint('OTTO_API_DOMAIN'); |
| 55 |
$parsed = parse_url($otto_api_domain); |
| 56 |
if (!empty($parsed['host'])) { |
| 57 |
$otto_domain = $parsed['host']; |
| 58 |
} |
| 59 |
} |
| 60 |
echo '<link rel="dns-prefetch" href="//' . esc_attr($otto_domain) . '">' . "\n"; |
| 61 |
echo '<link rel="preconnect" href="https://' . esc_attr($otto_domain) . '" crossorigin>' . "\n"; |
| 62 |
|
| 63 |
# adding the otto tag to pages |
| 64 |
$plugin_version = defined('METASYNC_VERSION') ? METASYNC_VERSION : 'unknown'; |
| 65 |
# OPTIMIZED: use cached uuid |
| 66 |
$otto_tag = '<meta name="otto" content="uuid='.esc_attr(Metasync_Otto_Config::get_otto_uuid()).'; type=wordpress; enabled='.esc_attr($string_enabled).'; version='.esc_attr($plugin_version).'">'; |
| 67 |
|
| 68 |
# out the otto tag |
| 69 |
echo $otto_tag; |
| 70 |
}, 1); # Priority 1 to output early in head |
| 71 |
|
| 72 |
/** |
| 73 |
* Start end point to handle requests on page updates |
| 74 |
* The Otto Crawler will call this end point once a page is updated |
| 75 |
**/ |
| 76 |
|
| 77 |
# function to register the route |
| 78 |
function metasync_otto_crawl_notify($request){ |
| 79 |
|
| 80 |
# get request data params |
| 81 |
$data = $request->get_json_params(); |
| 82 |
|
| 83 |
# fields |
| 84 |
$fields = ['domain', 'urls']; |
| 85 |
|
| 86 |
# validate the json request |
| 87 |
foreach ($fields as $key => $value) { |
| 88 |
|
| 89 |
# if the value is empty stop tehre |
| 90 |
if(empty($data[$value])){ |
| 91 |
|
| 92 |
# Handle the POST request |
| 93 |
return new WP_REST_Response(array( |
| 94 |
'success' => false, |
| 95 |
'message' => 'Invalid Field : '. $value, |
| 96 |
), 400); |
| 97 |
} |
| 98 |
|
| 99 |
} |
| 100 |
|
| 101 |
# load otto pixel |
| 102 |
$otto_pixel = new Metasync_otto_pixel(false); |
| 103 |
|
| 104 |
# save the otto data first (cheap local DB write — kept synchronous) |
| 105 |
$otto_pixel->save_crawl_data($data); |
| 106 |
|
| 107 |
# Defer the expensive per-URL work (OTTO API fetch, post meta sync, |
| 108 |
# per-URL cache purge) to background jobs. The webhook caller enforces a |
| 109 |
# strict response-time budget; doing this work inline previously blocked |
| 110 |
# the response for up to 30s × N URLs and timed the caller out. |
| 111 |
# Each URL gets its own scheduled event so failures are isolated. |
| 112 |
# |
| 113 |
# WP-299: Cap the number of per-URL cron jobs scheduled per webhook batch. |
| 114 |
# Large crawl batches (100+ URLs) previously created hundreds of cron events |
| 115 |
# that overwhelmed WP-Cron on shared/managed hosts like WP Engine. Excess |
| 116 |
# URLs beyond the cap are silently skipped — OTTO will re-crawl them on the |
| 117 |
# next webhook cycle. |
| 118 |
$max_jobs_per_batch = defined('METASYNC_MAX_JOBS_PER_BATCH') ? METASYNC_MAX_JOBS_PER_BATCH : 25; |
| 119 |
$now = time(); |
| 120 |
$routes_to_process = array(); |
| 121 |
$scheduled_count = 0; |
| 122 |
|
| 123 |
foreach($data['urls'] AS $key => $url){ |
| 124 |
# Enforce per-batch cap to prevent cron overload |
| 125 |
if ($scheduled_count >= $max_jobs_per_batch) { |
| 126 |
break; |
| 127 |
} |
| 128 |
|
| 129 |
# prepare the route |
| 130 |
$route = $data['domain'] . $url; |
| 131 |
|
| 132 |
# validate the route |
| 133 |
$route = rtrim($route, '/'); |
| 134 |
|
| 135 |
# Resolve redirect table: use final destination URL before excluded/404 checks and OTTO processing |
| 136 |
$route = metasync_otto_resolve_redirect_to_final_url($route); |
| 137 |
|
| 138 |
# OTTO has confirmed this URL is crawlable. Remove any auto-exclusion that was |
| 139 |
# previously set (e.g. because url_to_postid() returned 0 for a custom post type |
| 140 |
# and metasync_otto_is_url_available() incorrectly treated it as a 404). |
| 141 |
# Manual exclusions (auto_excluded = 0) are left untouched. |
| 142 |
metasync_otto_remove_auto_exclusion($route); |
| 143 |
|
| 144 |
# Skip manually excluded URLs - don't queue OTTO processing for them |
| 145 |
if (metasync_is_otto_url_excluded($route)) { |
| 146 |
continue; |
| 147 |
} |
| 148 |
|
| 149 |
# Queue the per-URL processing for background execution. The handler |
| 150 |
# (metasync_handle_otto_crawl_url_job) performs transient warming, |
| 151 |
# SEO meta sync, and per-URL host cache purge. |
| 152 |
# Offset each event by $key seconds to avoid wp_schedule_single_event() |
| 153 |
# silently dropping duplicates when timestamp + hook + args collide. |
| 154 |
wp_schedule_single_event($now + $key, 'metasync_process_otto_crawl_url_job', array($route)); |
| 155 |
|
| 156 |
$routes_to_process[] = $route; |
| 157 |
$scheduled_count++; |
| 158 |
} |
| 159 |
|
| 160 |
# Schedule a single batch job for cache warming and edge CDN purge. |
| 161 |
# These operations benefit from batching: warm_urls() uses concurrent |
| 162 |
# fire-and-forget requests, and edge providers (Cloudflare, Fastly) |
| 163 |
# support multi-URL purge in a single API call. |
| 164 |
if (!empty($routes_to_process)) { |
| 165 |
$batch_time = $now + count($data['urls']) + 5; // run after all per-URL jobs |
| 166 |
wp_schedule_single_event($batch_time, 'metasync_process_otto_batch_cache_job', array($routes_to_process)); |
| 167 |
} |
| 168 |
|
| 169 |
# Track OTTO optimization event in GA4 (local, non-blocking) |
| 170 |
try { |
| 171 |
Metasync_GA4::get_instance()->track_otto_optimization($data); |
| 172 |
} catch (Exception $e) { |
| 173 |
// Analytics tracking failed, continue |
| 174 |
} |
| 175 |
|
| 176 |
# Return 200 immediately so the webhook caller does not time out. |
| 177 |
return new WP_REST_Response(array( |
| 178 |
'success' => true, |
| 179 |
'message' => 'OTTO crawl notification received', |
| 180 |
), 200); |
| 181 |
} |
| 182 |
|
| 183 |
/** |
| 184 |
* Background handler for a single OTTO crawl-notify URL. |
| 185 |
* |
| 186 |
* Runs the per-URL sequence that was previously inline in metasync_otto_crawl_notify(): |
| 187 |
* warm the OTTO transient cache, sync SEO post meta, and clear the per-URL host cache. |
| 188 |
* |
| 189 |
* Cache warming and edge CDN purge are handled separately by |
| 190 |
* metasync_handle_otto_batch_cache_job() in a single batched event, because |
| 191 |
* those operations benefit from multi-URL batching (fewer API calls). |
| 192 |
* |
| 193 |
* On failure, retries up to METASYNC_OTTO_JOB_MAX_RETRIES times with 60s backoff. |
| 194 |
* After all retries are exhausted the failure is recorded via |
| 195 |
* metasync_record_failed_action() for surfacing in Site Health. |
| 196 |
* |
| 197 |
* @param string $route Fully-qualified URL to process. |
| 198 |
* @param int $retry_count Current retry attempt (0 = first run). |
| 199 |
*/ |
| 200 |
function metasync_handle_otto_crawl_url_job($route, $retry_count = 0) { |
| 201 |
try { |
| 202 |
# Step 1: Warm OTTO transient cache (fetch fresh suggestions from OTTO API into WP transient) |
| 203 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 204 |
if (!empty($otto_uuid)) { |
| 205 |
$transient_cache = new Metasync_Otto_Transient_Cache($otto_uuid); |
| 206 |
$transient_cache->warm_cache($route); |
| 207 |
} |
| 208 |
|
| 209 |
# Step 2: Write SEO post meta synchronously BEFORE the cache purge. |
| 210 |
# This ensures the DB is fully up-to-date when Kinsta (or any host) re-populates |
| 211 |
# the cache on the very next request — eliminating the race condition where the |
| 212 |
# 1-second scheduled job hadn't run yet and stale meta got cached. |
| 213 |
# allow_defer=false: do NOT reschedule a new metasync_process_seo_job cron |
| 214 |
# event from this synchronous path — the crawl_url_job retry mechanism already |
| 215 |
# handles failures, and rescheduling here caused an unbounded cron pile-up (WP-299). |
| 216 |
metasync_process_otto_seo_data($route, false); |
| 217 |
|
| 218 |
# Step 3: Clear the per-URL cache entry |
| 219 |
$otto_pixel = new Metasync_otto_pixel(false); |
| 220 |
$otto_pixel->refresh_cache($route); |
| 221 |
Metasync_Cache_Purge::purge_single_url($route); |
| 222 |
|
| 223 |
} catch (Exception $e) { |
| 224 |
$max_retries = defined('METASYNC_OTTO_JOB_MAX_RETRIES') ? METASYNC_OTTO_JOB_MAX_RETRIES : 3; |
| 225 |
|
| 226 |
if ($retry_count < $max_retries) { |
| 227 |
# Schedule a retry with exponential backoff: 60s, 120s, 240s |
| 228 |
$delay = 60 * pow(2, $retry_count); |
| 229 |
wp_schedule_single_event(time() + $delay, 'metasync_process_otto_crawl_url_job', array($route, $retry_count + 1)); |
| 230 |
error_log('MetaSync OTTO: retrying crawl-url job for ' . $route . ' (attempt ' . ($retry_count + 1) . '/' . $max_retries . ') in ' . $delay . 's: ' . $e->getMessage()); |
| 231 |
} else { |
| 232 |
metasync_record_failed_action('metasync_process_otto_crawl_url_job'); |
| 233 |
error_log('MetaSync OTTO: background crawl-url job permanently failed for ' . $route . ' after ' . $max_retries . ' retries: ' . $e->getMessage() . ' in ' . $e->getFile() . ':' . $e->getLine()); |
| 234 |
} |
| 235 |
} |
| 236 |
} |
| 237 |
|
| 238 |
/** |
| 239 |
* Batch handler for cache warming and edge CDN purge. |
| 240 |
* |
| 241 |
* Runs after all per-URL jobs have had time to complete. Batching these |
| 242 |
* operations avoids N separate API calls to edge providers (Cloudflare |
| 243 |
* supports up to 30 tags per request) and lets warm_urls() issue concurrent |
| 244 |
* fire-and-forget requests. |
| 245 |
* |
| 246 |
* @param array $routes List of fully-qualified URLs to warm and purge. |
| 247 |
*/ |
| 248 |
function metasync_handle_otto_batch_cache_job($routes) { |
| 249 |
try { |
| 250 |
# Re-populate the cache so OTTO-modified output is what gets stored. |
| 251 |
# Per-URL host cache was already cleared by individual jobs. |
| 252 |
# warm_urls() hits each URL with a non-blocking request so our code — |
| 253 |
# with fresh transients and post meta — is first to populate the host cache. |
| 254 |
Metasync_Cache_Purge::warm_urls($routes); |
| 255 |
|
| 256 |
# Purge edge CDN caches (Cloudflare, Fastly, Akamai, Sucuri, Sevalla, etc.) |
| 257 |
# Tag-based providers purge only the affected posts; full-flush providers fire once per batch. |
| 258 |
Metasync_Edge_Cache_Purge::purge($routes); |
| 259 |
} catch (Exception $e) { |
| 260 |
metasync_record_failed_action('metasync_process_otto_batch_cache_job'); |
| 261 |
error_log('MetaSync OTTO: batch cache job failed for ' . count($routes) . ' URLs: ' . $e->getMessage() . ' in ' . $e->getFile() . ':' . $e->getLine()); |
| 262 |
} |
| 263 |
} |
| 264 |
|
| 265 |
# NOTE: Cache system removed - these functions are no longer needed |
| 266 |
# Kept for backward compatibility in case old cache directories need cleanup |
| 267 |
function metasync_deleteDir($dir) { |
| 268 |
if (!is_dir($dir)) { |
| 269 |
return false; |
| 270 |
} |
| 271 |
$files = array_diff(scandir($dir), array('.', '..')); |
| 272 |
foreach ($files as $file) { |
| 273 |
$filePath = $dir . DIRECTORY_SEPARATOR . $file; |
| 274 |
if (is_dir($filePath)) { |
| 275 |
metasync_deleteDir($filePath); |
| 276 |
} else { |
| 277 |
unlink($filePath); |
| 278 |
} |
| 279 |
} |
| 280 |
return rmdir($dir); |
| 281 |
} |
| 282 |
|
| 283 |
# Cleanup function for removing old cache directories (if they exist) |
| 284 |
function metasync_invalidate_all_caches($folder = ''){ |
| 285 |
# Cache system removed - this function only exists to clean up old cache directories |
| 286 |
if(!defined('WP_CONTENT_DIR')){ |
| 287 |
return false; |
| 288 |
} |
| 289 |
$wp_content_dir = WP_CONTENT_DIR; |
| 290 |
$cache_dir = $wp_content_dir . '/metasync_caches'; |
| 291 |
if(in_array($folder, ['posts', 'pages'])){ |
| 292 |
$cache_dir = $cache_dir . '/' . $folder; |
| 293 |
} |
| 294 |
if(is_dir($cache_dir)){ |
| 295 |
metasync_deleteDir($cache_dir); |
| 296 |
} |
| 297 |
} |
| 298 |
|
| 299 |
function metasync_start_otto(){ |
| 300 |
|
| 301 |
# PERFORMANCE FIX: Cache is now enabled for speed |
| 302 |
# Skip initial cache cleanup to preserve existing cache |
| 303 |
# Cache files are valuable for performance - only clear on OTTO updates |
| 304 |
# Periodic cache clearing can be configured in plugin settings if needed |
| 305 |
|
| 306 |
# exclude AJAX requests and WooCommerce transactional pages from OTTO SSR |
| 307 |
# SSR is now ENABLED for: single products, product categories, product tags |
| 308 |
# SSR is SKIPPED for: cart, checkout |
| 309 |
# Note: title/description filters (pre_get_document_title, wp_head meta desc) |
| 310 |
# run on ALL pages regardless — they are hooked unconditionally in seo-functions.php |
| 311 |
|
| 312 |
# ── WooCommerce-independent cart/checkout protection (WP-374) ────────── |
| 313 |
# The is_cart()/is_checkout() guards below only recognize WooCommerce. Carts |
| 314 |
# from other systems — e.g. the Point of Rental "Catalog" plugin on |
| 315 |
# venturarental.com — are invisible to them, so OTTO would process those |
| 316 |
# pages and let caching layers store them. A cached cart page corrupts the |
| 317 |
# live cart (added items vanish because a stale page is served). The checks |
| 318 |
# here do not depend on WooCommerce being active. |
| 319 |
|
| 320 |
# Never run OTTO on non-GET requests. SSR exists for crawlers, which only |
| 321 |
# issue GET; POST/PUT/etc. are form or cart submissions that must pass |
| 322 |
# through untouched. (OTTO's own internal fetches use GET.) |
| 323 |
if (($_SERVER['REQUEST_METHOD'] ?? 'GET') !== 'GET') { |
| 324 |
return; |
| 325 |
} |
| 326 |
|
| 327 |
# Skip OTTO on cart/checkout-type paths regardless of the cart plugin, and |
| 328 |
# signal caching plugins (WP Rocket/Kinsta/etc.) to never cache them — a |
| 329 |
# cached cart/checkout page is precisely what corrupts cart state. |
| 330 |
# Matched against the FIRST path segment (relative to the WP home path) so |
| 331 |
# subdirectory installs (/shop/cart) still work, while unrelated content pages |
| 332 |
# like /blog/cart or /features/checkout do NOT false-match. |
| 333 |
# Drop the query string with explode() (no shared internal pointer like strtok), |
| 334 |
# then lower-case and strip surrounding slashes. |
| 335 |
$otto_req_path = explode('?', (string) wp_unslash($_SERVER['REQUEST_URI'] ?? ''), 2)[0]; |
| 336 |
$otto_req_path = strtolower(trim($otto_req_path, '/')); |
| 337 |
# Strip the site's base path so first-segment matching works on subdir installs. |
| 338 |
$otto_home_path = trim((string) parse_url(home_url(), PHP_URL_PATH), '/'); |
| 339 |
if ($otto_home_path !== '' && strpos($otto_req_path, $otto_home_path . '/') === 0) { |
| 340 |
$otto_req_path = substr($otto_req_path, strlen($otto_home_path) + 1); |
| 341 |
} |
| 342 |
$otto_first_segment = explode('/', $otto_req_path, 2)[0]; |
| 343 |
# Filterable so sites can add/remove transactional slugs without code changes. |
| 344 |
$otto_cart_segments = apply_filters('metasync_otto_cart_paths', array( |
| 345 |
'cart', 'checkout', 'basket', 'request-a-quote', 'quote-request', |
| 346 |
)); |
| 347 |
$otto_cart_segments = array_map('strtolower', (array) $otto_cart_segments); |
| 348 |
if ($otto_first_segment !== '' && in_array($otto_first_segment, $otto_cart_segments, true)) { |
| 349 |
if (!defined('DONOTCACHEPAGE')) { |
| 350 |
define('DONOTCACHEPAGE', true); |
| 351 |
} |
| 352 |
return; |
| 353 |
} |
| 354 |
# ────────────────────────────────────────────────────────────────────── |
| 355 |
|
| 356 |
if ( |
| 357 |
# disable ajax calls |
| 358 |
isset($_GET['ucfrontajaxaction']) || |
| 359 |
# OTTO Preview mode - skip OTTO when previewing original content |
| 360 |
(isset($_GET['otto_preview']) && $_GET['otto_preview'] === '1') || |
| 361 |
# WooCommerce shop archive page only (products and categories now use SSR) |
| 362 |
//(function_exists('is_shop') && is_shop()) || |
| 363 |
# Cart page |
| 364 |
(function_exists('is_cart') && is_cart()) || |
| 365 |
# Checkout page |
| 366 |
(function_exists('is_checkout') && is_checkout()) || |
| 367 |
# My Account page |
| 368 |
//(function_exists('is_account_page') && is_account_page()) || |
| 369 |
# Standard WordPress AJAX |
| 370 |
(function_exists('wp_doing_ajax') && wp_doing_ajax()) || |
| 371 |
# check by constant |
| 372 |
(defined('DOING_AJAX') && DOING_AJAX) || |
| 373 |
# WooCommerce AJAX endpoint (e.g., ?wc-ajax=update_cart) |
| 374 |
(isset($_REQUEST['wc-ajax']) && !empty($_REQUEST['wc-ajax'])) || |
| 375 |
# AJAX requests via X-Requested-With header |
| 376 |
(isset($_SERVER['HTTP_X_REQUESTED_WITH']) && strtolower($_SERVER['HTTP_X_REQUESTED_WITH']) === 'xmlhttprequest') || |
| 377 |
# Gravity Forms submission detection - skip OTTO to allow form processing |
| 378 |
(isset($_POST['gform_submit']) && ( |
| 379 |
is_array($_POST['gform_submit']) || |
| 380 |
(is_string($_POST['gform_submit']) && isset($_POST['is_submit_' . $_POST['gform_submit']]) && !empty($_POST['gform_submit'])) |
| 381 |
)) || |
| 382 |
# Gravity Forms AJAX submission |
| 383 |
(isset($_POST['gform_ajax']) && isset($_POST['gform_submit']) && ( |
| 384 |
is_array($_POST['gform_submit']) || !empty($_POST['gform_submit']) |
| 385 |
)) || |
| 386 |
# Gravity Forms file upload |
| 387 |
(isset($_POST['gform_uploaded_files'])) || |
| 388 |
# Any Gravity Forms POST parameter |
| 389 |
(isset($_POST['gform_submit']) || isset($_POST['gform_unique_id']) || isset($_POST['gform_field_values'])) || |
| 390 |
# Formidable Forms AJAX submission detection - skip OTTO to allow form processing |
| 391 |
(isset($_POST['action']) && $_POST['action'] === 'frm_entries_create') || |
| 392 |
# Formidable Forms POST parameters |
| 393 |
(isset($_POST['form_id']) && !empty($_POST['form_id'])) || |
| 394 |
# Formidable Forms action parameter |
| 395 |
(isset($_POST['frm_action']) && !empty($_POST['frm_action'])) || |
| 396 |
# Formidable Forms item_key (used in form submissions) |
| 397 |
(isset($_POST['item_key']) && !empty($_POST['item_key'])) |
| 398 |
) { |
| 399 |
return; |
| 400 |
} |
| 401 |
|
| 402 |
# fetch globals |
| 403 |
global $metasync_options, $otto_enabled; |
| 404 |
|
| 405 |
# OPTIMIZED: check for the disable otto for logged in users option using cached config |
| 406 |
if(Metasync_Otto_Config::is_disabled_for_loggedin()){ |
| 407 |
|
| 408 |
# get user |
| 409 |
$current_user = wp_get_current_user(); |
| 410 |
|
| 411 |
# check if user is logged in |
| 412 |
if( !empty($current_user->ID)){ |
| 413 |
|
| 414 |
return; |
| 415 |
} |
| 416 |
} |
| 417 |
|
| 418 |
# check if current URL is manually excluded from OTTO |
| 419 |
# NOTE: auto-exclusions (false-positive 404 detections) are intentionally NOT checked |
| 420 |
# here — they must not block OTTO rendering. Use metasync_is_otto_url_excluded() only |
| 421 |
# in the webhook handler where we gate SEO meta writes. |
| 422 |
$request_uri = sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'] ?? '')); |
| 423 |
$current_url = home_url(strtok($request_uri, '?') ?: $request_uri); |
| 424 |
if (metasync_is_otto_url_manually_excluded($current_url)) { |
| 425 |
return; |
| 426 |
} |
| 427 |
|
| 428 |
# Skip OTTO for XML endpoints (sitemaps, RSS-as-xml, etc.). OTTO has no |
| 429 |
# suggestions for machine-readable XML, and the upstream API returns |
| 430 |
# API_ERROR for these URLs — which then stamps misleading |
| 431 |
# X-MetaSync-OTTO-Cache: API_ERROR / X-MetaSync-OTTO-Method: NONE headers |
| 432 |
# onto otherwise-healthy sitemap responses and causes false-alarm bug |
| 433 |
# reports from customers. See WP-353. |
| 434 |
$request_path = strtok($request_uri, '?'); |
| 435 |
if ($request_path && preg_match('#\.xml$#i', $request_path)) { |
| 436 |
return; |
| 437 |
} |
| 438 |
|
| 439 |
# BOT DETECTION: Always detect bots so crawl data reaches the SA backend. |
| 440 |
# Real bots don't execute JS, so otto-tracker.js never fires for them. |
| 441 |
# push_crawl_log_to_sa() fires a non-blocking wp_remote_post here instead. |
| 442 |
$bot_detector = Metasync_Otto_Bot_Detector::get_instance(); |
| 443 |
$detection = $bot_detector->detect(); |
| 444 |
if ( $detection['is_bot'] ) { |
| 445 |
$bot_detector->push_crawl_log_to_sa( $detection, $current_url ); |
| 446 |
} |
| 447 |
|
| 448 |
# Throttle OTTO rendering for non-search-engine bots: at most one render |
| 449 |
# per URL+bot every 5 minutes. Humans and verified search engines |
| 450 |
# (Googlebot, Bingbot, etc.) are never throttled — their hits always get |
| 451 |
# full OTTO content so indexing remains current. SEO tools, AI scrapers, |
| 452 |
# uptime monitors, and unverified crawlers are de-duplicated so repeat |
| 453 |
# hits don't re-trigger the expensive DOM-rewriting path. |
| 454 |
if ( $detection['is_bot'] && $detection['bot_type'] !== 'search_engine' ) { |
| 455 |
$normalized_url = strtok( $current_url, '?' ); |
| 456 |
if ( $normalized_url === false ) { |
| 457 |
$normalized_url = $current_url; |
| 458 |
} |
| 459 |
// Lowercase + length cap so attacker-controlled bot_name variations |
| 460 |
// (EvilBota, EvilBotB, EvilBotc, ...) collapse to a bounded key space |
| 461 |
// and cannot flood wp_options with unique transient rows. |
| 462 |
$bot_name = substr( strtolower( $detection['bot_name'] ?? 'unknown' ), 0, 32 ); |
| 463 |
$render_throttle_key = 'metasync_otto_rendered_' . md5( $normalized_url . '|' . $bot_name ); |
| 464 |
if ( get_transient( $render_throttle_key ) ) { |
| 465 |
// Prevent page caches (WP Rocket, W3TC, etc.) from saving the |
| 466 |
// un-OTTO'd response and serving it to subsequent human visitors. |
| 467 |
if ( ! defined( 'DONOTCACHEPAGE' ) ) { |
| 468 |
define( 'DONOTCACHEPAGE', true ); |
| 469 |
} |
| 470 |
return; |
| 471 |
} |
| 472 |
set_transient( $render_throttle_key, 1, 5 * MINUTE_IN_SECONDS ); |
| 473 |
} |
| 474 |
|
| 475 |
# Optionally skip OTTO processing for bot traffic (when the setting is enabled) |
| 476 |
if ($bot_detector->should_skip_otto()) { |
| 477 |
// Log the bot locally and count the saved API call |
| 478 |
$bot_detector->log_detection($detection); |
| 479 |
$bot_stats_db = Metasync_Otto_Bot_Statistics_Database::get_instance(); |
| 480 |
$bot_stats_db->increment_api_calls_saved(); |
| 481 |
|
| 482 |
// Skip OTTO processing for this bot |
| 483 |
return; |
| 484 |
} |
| 485 |
|
| 486 |
# OPTIMIZED: Check if Otto should be disabled for WP Rocket compatibility |
| 487 |
if (class_exists('WP_Rocket')) { |
| 488 |
$wp_rocket_compat_mode = Metasync_Otto_Config::get_wp_rocket_compat_mode(); |
| 489 |
|
| 490 |
if ($wp_rocket_compat_mode === 'disable_otto') { |
| 491 |
return; # Exit early, Otto is disabled when WP Rocket is active |
| 492 |
} |
| 493 |
} |
| 494 |
|
| 495 |
# Skip OTTO for Divi AJAX pagination and paginated archive requests (WP-315). |
| 496 |
# ?et_blog = Divi AJAX pagination callback |
| 497 |
# /page/N/ = paginated blog/archive pages — OTTO's buffer/HTTP render causes |
| 498 |
# module numbering mismatch between page 1 (with TB template) and page N |
| 499 |
# (without TB template), breaking Divi's JS pagination selector matching. |
| 500 |
if (isset($_GET['et_blog']) || (is_paged() && !is_singular())) { |
| 501 |
return; |
| 502 |
} |
| 503 |
|
| 504 |
# Handle cache plugin compatibility early - before any caching happens |
| 505 |
metasync_otto_handle_cache_compatibility(); |
| 506 |
|
| 507 |
# check if OTTO is disabled for this specific page/post |
| 508 |
$post_id = get_the_ID(); |
| 509 |
if ($post_id && class_exists('Metasync_Otto_Frontend_Toolbar')) { |
| 510 |
if (Metasync_Otto_Frontend_Toolbar::is_otto_disabled($post_id)) { |
| 511 |
return; |
| 512 |
} |
| 513 |
} |
| 514 |
|
| 515 |
# check if we are having an otto request |
| 516 |
if(!empty($_GET['is_otto_page_fetch'])){ |
| 517 |
|
| 518 |
# Block SEO plugins NOW for this internal fetch request |
| 519 |
# metasync_otto_block_seo_plugins(); |
| 520 |
# $_SERVER['REQUEST_URI'] = remove_query_arg('is_otto_page_fetch', $_SERVER['REQUEST_URI']); |
| 521 |
$block_title = !empty($_GET['otto_block_title']) && $_GET['otto_block_title'] === '1'; |
| 522 |
$block_description = !empty($_GET['otto_block_desc']) && $_GET['otto_block_desc'] === '1'; |
| 523 |
|
| 524 |
# Block SEO plugins conditionally based on what Otto has |
| 525 |
if ($block_title || $block_description) { |
| 526 |
metasync_otto_block_seo_plugins($block_title, $block_description); |
| 527 |
} |
| 528 |
|
| 529 |
# Remove ALL Otto parameters from REQUEST_URI to prevent them from appearing in pagination, etc. |
| 530 |
$request_uri_raw = sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'] ?? '')); |
| 531 |
$_SERVER['REQUEST_URI'] = remove_query_arg( |
| 532 |
['is_otto_page_fetch', 'otto_block_title', 'otto_block_desc'], |
| 533 |
$request_uri_raw |
| 534 |
); |
| 535 |
|
| 536 |
# Also remove from $_GET to prevent WordPress from using them |
| 537 |
unset($_GET['is_otto_page_fetch']); |
| 538 |
unset($_GET['otto_block_title']); |
| 539 |
unset($_GET['otto_block_desc']); |
| 540 |
return; |
| 541 |
} |
| 542 |
|
| 543 |
# to avoid unnecessary processese |
| 544 |
# OPTIMIZED: check that otto is configured |
| 545 |
# And the UUID is properly set before running OTT |
| 546 |
|
| 547 |
# check that we have the option |
| 548 |
if(!Metasync_Otto_Config::is_otto_enabled()){ |
| 549 |
return; |
| 550 |
} |
| 551 |
|
| 552 |
# check that otto is enabled |
| 553 |
if(!$otto_enabled){ |
| 554 |
return; |
| 555 |
} |
| 556 |
|
| 557 |
# get the otto uuid |
| 558 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 559 |
|
| 560 |
# start the class |
| 561 |
$otto = new Metasync_otto_pixel($otto_uuid); |
| 562 |
|
| 563 |
# call render |
| 564 |
$otto->render_route_html(); |
| 565 |
} |
| 566 |
|
| 567 |
/** |
| 568 |
* Handle cache plugin compatibility with Otto |
| 569 |
* Controls DONOTCACHEPAGE constant based on active plugins and configuration |
| 570 |
* This function is called early in the WordPress lifecycle |
| 571 |
*/ |
| 572 |
function metasync_otto_handle_cache_compatibility() { |
| 573 |
# Detect active plugins |
| 574 |
$brizy_active = class_exists('Brizy_Editor') || defined('BRIZY_VERSION'); |
| 575 |
$wp_rocket_active = class_exists('WP_Rocket'); |
| 576 |
$sg_optimizer_active = is_plugin_active('sg-cachepress/sg-cachepress.php'); |
| 577 |
|
| 578 |
# OPTIMIZED: Get configuration option using cached config |
| 579 |
$wp_rocket_compat_mode = Metasync_Otto_Config::get_wp_rocket_compat_mode(); |
| 580 |
|
| 581 |
# Check for Brizy posts in database |
| 582 |
global $wpdb; |
| 583 |
$has_brizy_posts = false; |
| 584 |
|
| 585 |
if ($brizy_active) { |
| 586 |
# OPTIMIZED: Check cache first (1-hour TTL) to avoid querying on every page load |
| 587 |
$cached = get_transient('metasync_has_brizy_posts'); |
| 588 |
if ($cached !== false) { |
| 589 |
$has_brizy_posts = ($cached === 'yes'); |
| 590 |
} else { |
| 591 |
# Query database only if cache missed |
| 592 |
$has_brizy_posts = $wpdb->get_var( |
| 593 |
"SELECT COUNT(*) FROM {$wpdb->postmeta} |
| 594 |
WHERE meta_key = 'brizy_post_uid' |
| 595 |
AND meta_value != '' |
| 596 |
LIMIT 1" |
| 597 |
); |
| 598 |
|
| 599 |
# Cache result for 1 hour |
| 600 |
$result = !empty($has_brizy_posts) ? 'yes' : 'no'; |
| 601 |
set_transient('metasync_has_brizy_posts', $result, HOUR_IN_SECONDS); |
| 602 |
} |
| 603 |
} |
| 604 |
|
| 605 |
# Determine if DONOTCACHEPAGE should be set |
| 606 |
$should_set_donotcachepage = false; |
| 607 |
|
| 608 |
# Case 1: Brizy is active with posts - always needed |
| 609 |
if ($brizy_active && !empty($has_brizy_posts)) { |
| 610 |
$should_set_donotcachepage = true; |
| 611 |
} |
| 612 |
|
| 613 |
# Case 2: SG Optimizer without WP Rocket - prevent conflicts |
| 614 |
elseif ($sg_optimizer_active && !$wp_rocket_active) { |
| 615 |
$should_set_donotcachepage = true; |
| 616 |
} |
| 617 |
|
| 618 |
# Case 3: User explicitly disabled Otto for WP Rocket compatibility |
| 619 |
elseif ($wp_rocket_active && $wp_rocket_compat_mode === 'disable_otto') { |
| 620 |
$should_set_donotcachepage = true; |
| 621 |
return; # Exit early, Otto won't run |
| 622 |
} |
| 623 |
|
| 624 |
# Case 4: WP Rocket active with auto/buffer mode - DON'T set DONOTCACHEPAGE |
| 625 |
# This allows WP Rocket optimizations to continue working |
| 626 |
|
| 627 |
# Only set DONOTCACHEPAGE if needed |
| 628 |
if ($should_set_donotcachepage && !defined('DONOTCACHEPAGE')) { |
| 629 |
define('DONOTCACHEPAGE', true); |
| 630 |
} |
| 631 |
|
| 632 |
# Apply SG Optimizer specific filters only if SG is active and no WP Rocket |
| 633 |
if ($sg_optimizer_active && !$wp_rocket_active && $should_set_donotcachepage) { |
| 634 |
if (!defined('SG_CachePress_SUPERCACHER')) { |
| 635 |
define('SG_CachePress_SUPERCACHER', false); |
| 636 |
} |
| 637 |
|
| 638 |
add_filter('sgo_html_cache_disable', '__return_true', 999); |
| 639 |
add_filter('sgo_css_combine_exclude', '__return_true', 999); |
| 640 |
add_filter('sgo_js_combine_exclude', '__return_true', 999); |
| 641 |
add_filter('sgo_cache_this_page', '__return_false', 999); |
| 642 |
|
| 643 |
if (!headers_sent()) { |
| 644 |
header('Cache-Control: no-cache, must-revalidate, max-age=0'); |
| 645 |
header('X-Accel-Expires: 0'); |
| 646 |
} |
| 647 |
} |
| 648 |
} |
| 649 |
|
| 650 |
/** |
| 651 |
* Block SEO plugins conditionally based on what Otto is providing |
| 652 |
* Only blocks title if Otto has title, only blocks description if Otto has description |
| 653 |
* This prevents duplicate SEO tags while allowing fallback to SEO plugins when Otto has no data. |
| 654 |
* Supports Yoast SEO, Rank Math, and AIOSEO (free + pro). |
| 655 |
* |
| 656 |
* @param bool $block_title Whether to block title tags. |
| 657 |
* @param bool $block_description Whether to block description tags. |
| 658 |
* @param array $description_tags Optional. Granular list of OTTO-provided description tags |
| 659 |
* (e.g. ['meta[name=description]', 'meta[property=og:description]']). |
| 660 |
* When provided, only matching AIOSEO tags are suppressed. |
| 661 |
* When empty, all AIOSEO description tags are suppressed (legacy behavior). |
| 662 |
*/ |
| 663 |
function metasync_otto_block_seo_plugins($block_title = false, $block_description = false, $description_tags = []) { |
| 664 |
# Disable Yoast SEO (free and premium) |
| 665 |
if (is_plugin_active('wordpress-seo/wp-seo.php') || |
| 666 |
is_plugin_active('wordpress-seo-premium/wp-seo-premium.php')) { |
| 667 |
|
| 668 |
# TITLE: Never block Yoast's title output during SSR fetch. |
| 669 |
# Yoast removes WordPress's native _wp_render_title_tag action and is the sole |
| 670 |
# renderer of the <title> tag. Returning false/empty from wpseo_title or removing |
| 671 |
# Title_Presenter leaves the page with NO <title> tag at all — OTTO's buffer |
| 672 |
# post-processing then has nothing to replace, producing a missing title. |
| 673 |
# Instead, let Yoast render its own title; OTTO's replace_title() will overwrite |
| 674 |
# it in the final HTML buffer. deduplicate_title_tags() cleans up any duplicates. |
| 675 |
|
| 676 |
# Block description only if Otto has description |
| 677 |
if ($block_description) { |
| 678 |
add_filter('wpseo_metadesc', '__return_false', 999); |
| 679 |
add_filter('wpseo_meta_description', '__return_false', 999); |
| 680 |
add_filter('wpseo_metakeywords', '__return_false', 999); |
| 681 |
} |
| 682 |
|
| 683 |
# Block Yoast's modern presenters — description only, never title |
| 684 |
add_filter('wpseo_frontend_presenters', function($presenters) use ($block_description) { |
| 685 |
if (!is_array($presenters)) return $presenters; |
| 686 |
|
| 687 |
$presenters_to_remove = []; |
| 688 |
|
| 689 |
|
| 690 |
# Remove description presenters only when OTTO has a description |
| 691 |
if ($block_description) { |
| 692 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Meta_Description_Presenter'; |
| 693 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Open_Graph\Description_Presenter'; |
| 694 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Twitter\Description_Presenter'; |
| 695 |
} |
| 696 |
|
| 697 |
foreach ($presenters as $key => $presenter) { |
| 698 |
// Safely get class name, suppressing autoload errors |
| 699 |
// This prevents warnings when Composer autoloader tries to load deprecated Yoast files |
| 700 |
$class_name = is_object($presenter) ? @get_class($presenter) : ''; |
| 701 |
|
| 702 |
if (!empty($class_name) && in_array($class_name, $presenters_to_remove)) { |
| 703 |
unset($presenters[$key]); |
| 704 |
} |
| 705 |
} |
| 706 |
return $presenters; |
| 707 |
}, 999); |
| 708 |
} |
| 709 |
|
| 710 |
# Disable Rank Math |
| 711 |
if (is_plugin_active('seo-by-rank-math/rank-math.php') || |
| 712 |
is_plugin_active('seo-by-rankmath/rank-math.php')) { |
| 713 |
|
| 714 |
if ($block_title) { |
| 715 |
add_filter('rank_math/frontend/title', '__return_empty_string', 999); |
| 716 |
} |
| 717 |
|
| 718 |
if ($block_description) { |
| 719 |
add_filter('rank_math/frontend/description', '__return_false', 999); |
| 720 |
add_filter('rank_math/frontend/show_keywords', '__return_false', 999); |
| 721 |
} |
| 722 |
} |
| 723 |
|
| 724 |
# Disable AIOSEO (free and pro) |
| 725 |
if (is_plugin_active('all-in-one-seo-pack/all_in_one_seo_pack.php') || |
| 726 |
is_plugin_active('all-in-one-seo-pack-pro/all_in_one_seo_pack.php')) { |
| 727 |
|
| 728 |
if ($block_title) { |
| 729 |
add_filter('aioseo_title', '__return_empty_string', 999); |
| 730 |
add_filter('aioseo_facebook_tags', function($meta) { |
| 731 |
if (is_array($meta)) { unset($meta['og:title']); } |
| 732 |
return $meta; |
| 733 |
}, 999); |
| 734 |
add_filter('aioseo_twitter_tags', function($meta) { |
| 735 |
if (is_array($meta)) { unset($meta['twitter:title']); } |
| 736 |
return $meta; |
| 737 |
}, 999); |
| 738 |
} |
| 739 |
|
| 740 |
if ($block_description) { |
| 741 |
# Use granular tag list when available to only block what OTTO provides |
| 742 |
$tags = !empty($description_tags) ? $description_tags : []; |
| 743 |
$block_standard = empty($tags) || in_array('meta[name=description]', $tags); |
| 744 |
$block_og_desc = empty($tags) || in_array('meta[property=og:description]', $tags); |
| 745 |
$block_tw_desc = empty($tags) || in_array('meta[name=twitter:description]', $tags); |
| 746 |
|
| 747 |
if ($block_standard) { |
| 748 |
add_filter('aioseo_description', '__return_empty_string', 999); |
| 749 |
} |
| 750 |
if ($block_og_desc) { |
| 751 |
add_filter('aioseo_facebook_tags', function($meta) { |
| 752 |
if (is_array($meta)) { unset($meta['og:description']); } |
| 753 |
return $meta; |
| 754 |
}, 999); |
| 755 |
} |
| 756 |
if ($block_tw_desc) { |
| 757 |
add_filter('aioseo_twitter_tags', function($meta) { |
| 758 |
if (is_array($meta)) { unset($meta['twitter:description']); } |
| 759 |
return $meta; |
| 760 |
}, 999); |
| 761 |
} |
| 762 |
} |
| 763 |
} |
| 764 |
} |
| 765 |
# check that otto is not added via js to the site |
| 766 |
function metasync_check_otto_js(){ |
| 767 |
|
| 768 |
# Cache result to avoid a blocking HTTP self-request (~2.4s) |
| 769 |
# on every single admin page load. |
| 770 |
$cache_key = 'metasync_otto_js_detected'; |
| 771 |
$cached = get_transient($cache_key); |
| 772 |
|
| 773 |
if ($cached !== false) { |
| 774 |
return $cached === 'yes'; |
| 775 |
} |
| 776 |
|
| 777 |
# get the site url |
| 778 |
$site_url = site_url() . '?is_otto_page_fetch=1'; |
| 779 |
|
| 780 |
# get the html — short timeout so admin doesn't hang |
| 781 |
$page_data = wp_remote_get($site_url, array('timeout' => 5, 'sslverify' => false)); |
| 782 |
|
| 783 |
if (is_wp_error($page_data)) { |
| 784 |
set_transient($cache_key, 'no', HOUR_IN_SECONDS); |
| 785 |
return false; |
| 786 |
} |
| 787 |
|
| 788 |
# now get the html body |
| 789 |
$body = wp_remote_retrieve_body($page_data); |
| 790 |
|
| 791 |
# now load the body into html |
| 792 |
$dom = new HtmlDocument($body); |
| 793 |
|
| 794 |
# now check the dom for a meta tag with |
| 795 |
$script = $dom->find('script#sa-dynamic-optimization', 0); |
| 796 |
|
| 797 |
# check script |
| 798 |
if($script AND !empty($script->getAttribute('data-uuid'))){ |
| 799 |
set_transient($cache_key, 'yes', 12 * HOUR_IN_SECONDS); |
| 800 |
return true; |
| 801 |
} |
| 802 |
|
| 803 |
set_transient($cache_key, 'no', 12 * HOUR_IN_SECONDS); |
| 804 |
return false; |
| 805 |
}; |
| 806 |
|
| 807 |
# Handle AJAX Clear Cache request |
| 808 |
# NOTE: Cache system removed - this is now a no-op |
| 809 |
function metasync_clear_otto_cache_handler() { |
| 810 |
if (!empty($_GET['clear_otto_cache'])) { |
| 811 |
delete_transient('metasync_otto_js_detected'); |
| 812 |
# Cache system has been removed - no cache to clear |
| 813 |
wp_send_json_success(['message' => 'Cache system removed - all pages processed in real-time']); |
| 814 |
} |
| 815 |
else { |
| 816 |
wp_send_json_error(['message' => 'Missing parameter']); |
| 817 |
} |
| 818 |
} |
| 819 |
|
| 820 |
# Clear cache hook |
| 821 |
add_action('wp_ajax_metasync_clear_otto_cache', 'metasync_clear_otto_cache_handler'); |
| 822 |
|
| 823 |
# add admin action to check script |
| 824 |
function metasync_show_otto_ssr_notice() { |
| 825 |
if (!Metasync::current_user_has_plugin_access()) { |
| 826 |
return; // Only show to admins |
| 827 |
} |
| 828 |
|
| 829 |
# Get the plugin name using centralized method |
| 830 |
$plugin_name = Metasync::get_effective_plugin_name(); |
| 831 |
$whitelabel_otto_name = Metasync::get_whitelabel_otto_name(); |
| 832 |
if (metasync_check_otto_js()) { |
| 833 |
|
| 834 |
# Show admin notice with plugin name included in the message |
| 835 |
echo '<div class="notice notice-error"> |
| 836 |
<p><b>Warning from ' . esc_html($plugin_name) . '</b> |
| 837 |
<br> |
| 838 |
' . esc_html($whitelabel_otto_name) . ' JavaScript has been detected on your site. Please remove it and configure ' . esc_html($whitelabel_otto_name) . ' for Wordpress. Contact support for help |
| 839 |
</p> |
| 840 |
</div>'; |
| 841 |
} |
| 842 |
} |
| 843 |
|
| 844 |
add_action('admin_notices', 'metasync_show_otto_ssr_notice'); |
| 845 |
|
| 846 |
|
| 847 |
# staging dummy change |
| 848 |
# load otto in the wp hook |
| 849 |
add_action('wp', 'metasync_start_otto'); |
| 850 |
|
| 851 |
# ENHANCED OTTO SEO INTEGRATION |
| 852 |
# Register async SEO processing hook |
| 853 |
add_action('metasync_process_seo_job', 'metasync_process_otto_seo_data', 10, 3); |
| 854 |
add_action('metasync_process_otto_crawl_url_job', 'metasync_handle_otto_crawl_url_job', 10, 2); |
| 855 |
add_action('metasync_process_otto_batch_cache_job', 'metasync_handle_otto_batch_cache_job'); |
| 856 |
|
| 857 |
# Process OTTO SEO data and update WordPress meta fields for SEO plugins |
| 858 |
# This function now runs asynchronously via WordPress cron system |
| 859 |
# |
| 860 |
# @param string $route Fully-qualified URL to process. |
| 861 |
# @param bool $allow_defer When false, skip CPU-deferral rescheduling (used when |
| 862 |
# called synchronously from crawl_url_job whose own retry |
| 863 |
# mechanism already handles failures). |
| 864 |
# @param int $deferral_count How many times this job has already been deferred for |
| 865 |
# CPU load. Prevents infinite reschedule loops. |
| 866 |
|
| 867 |
function metasync_process_otto_seo_data($route, $allow_defer = true, $deferral_count = 0) { |
| 868 |
# Maximum number of times a job can be deferred before it is dropped. |
| 869 |
$max_deferrals = defined('METASYNC_SEO_JOB_MAX_DEFERRALS') ? METASYNC_SEO_JOB_MAX_DEFERRALS : 5; |
| 870 |
$lock_key = null; |
| 871 |
$lock_acquired = false; |
| 872 |
|
| 873 |
try { |
| 874 |
# Validate input |
| 875 |
if (empty($route) || !is_string($route)) { |
| 876 |
return false; |
| 877 |
} |
| 878 |
|
| 879 |
# CPU load check — defer if server is under load. |
| 880 |
# Only reschedule when called from the cron hook (allow_defer=true) and |
| 881 |
# we haven't exceeded the maximum deferral count. |
| 882 |
# Checked BEFORE acquiring the lock so that deferred jobs don't |
| 883 |
# acquire-then-immediately-release it (which defeats concurrency protection). |
| 884 |
if (!Metasync_CPU_Monitor::is_load_safe()) { |
| 885 |
if ($allow_defer && $deferral_count < $max_deferrals) { |
| 886 |
wp_schedule_single_event( |
| 887 |
time() + 60, |
| 888 |
'metasync_process_seo_job', |
| 889 |
array($route, true, $deferral_count + 1) |
| 890 |
); |
| 891 |
return false; |
| 892 |
} |
| 893 |
# Max deferrals reached or not allowed to defer — drop the job to |
| 894 |
# prevent unbounded cron accumulation that causes server overload. |
| 895 |
if ($deferral_count >= $max_deferrals) { |
| 896 |
metasync_record_failed_action('metasync_process_seo_job'); |
| 897 |
return false; |
| 898 |
} |
| 899 |
# allow_defer=false (sync path): just return false, parent handles retry. |
| 900 |
return false; |
| 901 |
} |
| 902 |
|
| 903 |
# Concurrency lock: prevent multiple SEO jobs from running simultaneously. |
| 904 |
# Uses a per-URL transient lock with a 120s TTL as a safety net (the lock is |
| 905 |
# explicitly deleted on completion). If the lock exists another run is already |
| 906 |
# processing this URL — reschedule once with a short delay instead of stacking. |
| 907 |
$lock_key = 'metasync_seo_lock_' . md5($route); |
| 908 |
if (get_transient($lock_key) !== false) { |
| 909 |
if ($allow_defer && $deferral_count < $max_deferrals) { |
| 910 |
wp_schedule_single_event( |
| 911 |
time() + 30, |
| 912 |
'metasync_process_seo_job', |
| 913 |
array($route, true, $deferral_count + 1) |
| 914 |
); |
| 915 |
} |
| 916 |
return false; |
| 917 |
} |
| 918 |
set_transient($lock_key, true, 120); |
| 919 |
$lock_acquired = true; |
| 920 |
|
| 921 |
# Resolve redirect table: use final destination URL before 404 checks and OTTO processing |
| 922 |
$route = metasync_otto_resolve_redirect_to_final_url($route); |
| 923 |
|
| 924 |
# Skip excluded URLs - don't process SEO data for them |
| 925 |
if (metasync_is_otto_url_excluded($route)) { |
| 926 |
//error_log('MetaSync OTTO: Skipping SEO processing for excluded URL: ' . $route); |
| 927 |
return false; |
| 928 |
} |
| 929 |
|
| 930 |
# Pre-flight 404 check: exclude URLs that would return 404 before making API call |
| 931 |
if (!metasync_otto_is_url_available($route)) { |
| 932 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404: {$route}"); |
| 933 |
metasync_otto_auto_exclude_404_url($route); |
| 934 |
return false; |
| 935 |
} |
| 936 |
|
| 937 |
# Get OTTO UUID from settings |
| 938 |
# OPTIMIZED: Use cached options |
| 939 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 940 |
|
| 941 |
if (empty($otto_uuid)) { |
| 942 |
return false; |
| 943 |
} |
| 944 |
|
| 945 |
# Meta descriptions are always enabled by default - no check needed |
| 946 |
|
| 947 |
# Fetch SEO data from OTTO API |
| 948 |
$seo_data = metasync_fetch_otto_seo_data($route, $otto_uuid); |
| 949 |
|
| 950 |
if (!$seo_data) { |
| 951 |
metasync_record_failed_action( 'metasync_process_seo_job' ); |
| 952 |
return false; |
| 953 |
} |
| 954 |
|
| 955 |
# Mark this URL as crawled by OTTO for SSR |
| 956 |
# Extract domain and path from route |
| 957 |
$parsed_url = parse_url($route); |
| 958 |
$domain_with_scheme = ($parsed_url['scheme'] ?? 'https') . '://' . ($parsed_url['host'] ?? ''); |
| 959 |
$url_path = ($parsed_url['path'] ?? '/'); |
| 960 |
|
| 961 |
# Create crawl data structure |
| 962 |
$crawl_data = array( |
| 963 |
'domain' => $domain_with_scheme, |
| 964 |
'urls' => array($url_path) |
| 965 |
); |
| 966 |
|
| 967 |
# Load Otto pixel class and save crawl data |
| 968 |
$otto_pixel = new Metasync_otto_pixel($otto_uuid); |
| 969 |
$otto_pixel->save_crawl_data($crawl_data); |
| 970 |
|
| 971 |
# Get WordPress post ID from URL |
| 972 |
$post_id = url_to_postid($route); |
| 973 |
|
| 974 |
# Special handling for WooCommerce shop page (url_to_postid doesn't work for it) |
| 975 |
if ((!$post_id || $post_id <= 0) && function_exists('wc_get_page_id')) { |
| 976 |
# Check if this URL is the WooCommerce shop page |
| 977 |
$shop_page_id = wc_get_page_id('shop'); |
| 978 |
if ($shop_page_id > 0) { |
| 979 |
$shop_url = get_permalink($shop_page_id); |
| 980 |
$route_normalized = rtrim($route, '/'); |
| 981 |
$shop_url_normalized = rtrim($shop_url, '/'); |
| 982 |
|
| 983 |
if ($route_normalized === $shop_url_normalized) { |
| 984 |
$post_id = $shop_page_id; |
| 985 |
} |
| 986 |
} |
| 987 |
} |
| 988 |
|
| 989 |
# Try to find WooCommerce product by URL if url_to_postid failed |
| 990 |
if ((!$post_id || $post_id <= 0) && strpos($route, '/product/') !== false && function_exists('wc_get_products')) { |
| 991 |
# Extract product slug from URL |
| 992 |
$product_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 993 |
|
| 994 |
# Try to get product by slug |
| 995 |
$products = wc_get_products(array( |
| 996 |
'name' => $product_slug, |
| 997 |
'limit' => 1, |
| 998 |
'status' => 'publish', |
| 999 |
)); |
| 1000 |
|
| 1001 |
if (empty($products)) { |
| 1002 |
# Fallback: try by slug using WP_Query |
| 1003 |
$args = array( |
| 1004 |
'post_type' => 'product', |
| 1005 |
'name' => $product_slug, |
| 1006 |
'posts_per_page' => 1, |
| 1007 |
'post_status' => 'publish', |
| 1008 |
); |
| 1009 |
$query = new WP_Query($args); |
| 1010 |
|
| 1011 |
if ($query->have_posts()) { |
| 1012 |
$product_post = $query->posts[0]; |
| 1013 |
$post_id = $product_post->ID; |
| 1014 |
} |
| 1015 |
} else { |
| 1016 |
$product = $products[0]; |
| 1017 |
$post_id = $product->get_id(); |
| 1018 |
} |
| 1019 |
} |
| 1020 |
|
| 1021 |
if (!$post_id || $post_id <= 0) { |
| 1022 |
# Check if this is a category page |
| 1023 |
if (strpos($route, '/category/') !== false) { |
| 1024 |
# Extract category slug from URL |
| 1025 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1026 |
$category = get_category_by_slug($category_slug); |
| 1027 |
|
| 1028 |
if ($category) { |
| 1029 |
# Check if category would return 404 before applying OTTO changes |
| 1030 |
if (metasync_would_term_return_404($category->term_id, 'category', $route)) { |
| 1031 |
error_log("MetaSync OTTO: Skipping SEO processing for category that would return 404: {$route} (Category ID: {$category->term_id})"); |
| 1032 |
metasync_otto_auto_exclude_404_url($route); |
| 1033 |
return false; |
| 1034 |
} |
| 1035 |
|
| 1036 |
# Update comprehensive category SEO meta fields |
| 1037 |
$update_result = metasync_update_comprehensive_category_seo_fields($category->term_id, $seo_data); |
| 1038 |
|
| 1039 |
if ($update_result['updated']) { |
| 1040 |
# Prepare trimmed values to 30 characters |
| 1041 |
$trim = function($value) { |
| 1042 |
if ($value === null) { return ''; } |
| 1043 |
$value = (string) $value; |
| 1044 |
$value = trim($value); |
| 1045 |
if (mb_strlen($value) > 30) { |
| 1046 |
return mb_substr($value, 0, 30); |
| 1047 |
} |
| 1048 |
return $value; |
| 1049 |
}; |
| 1050 |
|
| 1051 |
# Log individual field updates for category |
| 1052 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1053 |
$short = ''; |
| 1054 |
$title = ''; |
| 1055 |
|
| 1056 |
switch ($field_type) { |
| 1057 |
case 'meta_title': |
| 1058 |
$short = $trim($field_value); |
| 1059 |
$title = "Category Meta Title Update ({$short}...)"; |
| 1060 |
break; |
| 1061 |
case 'meta_description': |
| 1062 |
$short = $trim($field_value); |
| 1063 |
$title = "Category Meta Description Update ({$short}...)"; |
| 1064 |
break; |
| 1065 |
case 'meta_keywords': |
| 1066 |
$short = $trim($field_value); |
| 1067 |
$title = "Category Meta Keywords Update ({$short}...)"; |
| 1068 |
break; |
| 1069 |
case 'og_title': |
| 1070 |
$short = $trim($field_value); |
| 1071 |
$title = "Category Open Graph Title Update ({$short}...)"; |
| 1072 |
break; |
| 1073 |
case 'og_description': |
| 1074 |
$short = $trim($field_value); |
| 1075 |
$title = "Category Open Graph Description Update ({$short}...)"; |
| 1076 |
break; |
| 1077 |
case 'twitter_title': |
| 1078 |
$short = $trim($field_value); |
| 1079 |
$title = "Category Twitter Title Update ({$short}...)"; |
| 1080 |
break; |
| 1081 |
case 'twitter_description': |
| 1082 |
$short = $trim($field_value); |
| 1083 |
$title = "Category Twitter Description Update ({$short}...)"; |
| 1084 |
break; |
| 1085 |
case 'image_alt_data': |
| 1086 |
$image_count = count($field_value); |
| 1087 |
$title = "Category Image Alt Text Update ({$image_count} images)"; |
| 1088 |
break; |
| 1089 |
case 'headings_data': |
| 1090 |
$heading_count = count($field_value); |
| 1091 |
$title = "Category Headings Update ({$heading_count} headings)"; |
| 1092 |
break; |
| 1093 |
case 'structured_data': |
| 1094 |
$title = 'Category Structured Data Update'; |
| 1095 |
break; |
| 1096 |
} |
| 1097 |
|
| 1098 |
if (!empty($title)) { |
| 1099 |
metasync_log_sync_history([ |
| 1100 |
'title' => $title, |
| 1101 |
'source' => 'OTTO SEO', |
| 1102 |
'status' => 'published', |
| 1103 |
'content_type' => 'Category SEO', |
| 1104 |
'url' => $route, |
| 1105 |
'meta_data' => json_encode([ |
| 1106 |
'field' => $field_type, |
| 1107 |
'field_value' => $field_value, |
| 1108 |
'category_id' => $category->term_id, |
| 1109 |
'category_name' => $category->name |
| 1110 |
]) |
| 1111 |
]); |
| 1112 |
} |
| 1113 |
} |
| 1114 |
|
| 1115 |
return true; |
| 1116 |
} |
| 1117 |
|
| 1118 |
return false; |
| 1119 |
} |
| 1120 |
} |
| 1121 |
|
| 1122 |
# Check if this is a WooCommerce product category |
| 1123 |
if (strpos($route, '/product-category/') !== false) { |
| 1124 |
# Extract product category slug from URL |
| 1125 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1126 |
$term = get_term_by('slug', $category_slug, 'product_cat'); |
| 1127 |
|
| 1128 |
if ($term && !is_wp_error($term)) { |
| 1129 |
# Check if product category would return 404 before applying OTTO changes |
| 1130 |
if (metasync_would_term_return_404($term->term_id, 'product_cat', $route)) { |
| 1131 |
error_log("MetaSync OTTO: Skipping SEO processing for product category that would return 404: {$route} (Term ID: {$term->term_id})"); |
| 1132 |
metasync_otto_auto_exclude_404_url($route); |
| 1133 |
return false; |
| 1134 |
} |
| 1135 |
|
| 1136 |
# Update comprehensive taxonomy SEO meta fields |
| 1137 |
$update_result = metasync_update_comprehensive_taxonomy_seo_fields($term->term_id, 'product_cat', $seo_data); |
| 1138 |
|
| 1139 |
if ($update_result['updated']) { |
| 1140 |
# Prepare trimmed values to 30 characters |
| 1141 |
$trim = function($value) { |
| 1142 |
if ($value === null) { return ''; } |
| 1143 |
$value = (string) $value; |
| 1144 |
$value = trim($value); |
| 1145 |
if (mb_strlen($value) > 30) { |
| 1146 |
return mb_substr($value, 0, 30); |
| 1147 |
} |
| 1148 |
return $value; |
| 1149 |
}; |
| 1150 |
|
| 1151 |
# Log individual field updates for product category |
| 1152 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1153 |
$short = ''; |
| 1154 |
$title = ''; |
| 1155 |
|
| 1156 |
switch ($field_type) { |
| 1157 |
case 'meta_title': |
| 1158 |
$short = $trim($field_value); |
| 1159 |
$title = "Product Category Meta Title Update ({$short}...)"; |
| 1160 |
break; |
| 1161 |
case 'meta_description': |
| 1162 |
$short = $trim($field_value); |
| 1163 |
$title = "Product Category Meta Description Update ({$short}...)"; |
| 1164 |
break; |
| 1165 |
case 'meta_keywords': |
| 1166 |
$short = $trim($field_value); |
| 1167 |
$title = "Product Category Meta Keywords Update ({$short}...)"; |
| 1168 |
break; |
| 1169 |
case 'og_title': |
| 1170 |
$short = $trim($field_value); |
| 1171 |
$title = "Product Category Open Graph Title Update ({$short}...)"; |
| 1172 |
break; |
| 1173 |
case 'og_description': |
| 1174 |
$short = $trim($field_value); |
| 1175 |
$title = "Product Category Open Graph Description Update ({$short}...)"; |
| 1176 |
break; |
| 1177 |
case 'twitter_title': |
| 1178 |
$short = $trim($field_value); |
| 1179 |
$title = "Product Category Twitter Title Update ({$short}...)"; |
| 1180 |
break; |
| 1181 |
case 'twitter_description': |
| 1182 |
$short = $trim($field_value); |
| 1183 |
$title = "Product Category Twitter Description Update ({$short}...)"; |
| 1184 |
break; |
| 1185 |
case 'image_alt_data': |
| 1186 |
$image_count = count($field_value); |
| 1187 |
$title = "Product Category Image Alt Text Update ({$image_count} images)"; |
| 1188 |
break; |
| 1189 |
case 'headings_data': |
| 1190 |
$heading_count = count($field_value); |
| 1191 |
$title = "Product Category Headings Update ({$heading_count} headings)"; |
| 1192 |
break; |
| 1193 |
case 'structured_data': |
| 1194 |
$title = 'Product Category Structured Data Update'; |
| 1195 |
break; |
| 1196 |
} |
| 1197 |
|
| 1198 |
if (!empty($title)) { |
| 1199 |
metasync_log_sync_history([ |
| 1200 |
'title' => $title, |
| 1201 |
'source' => 'OTTO SEO', |
| 1202 |
'status' => 'published', |
| 1203 |
'content_type' => 'WooCommerce Product Category SEO', |
| 1204 |
'url' => $route, |
| 1205 |
'meta_data' => json_encode([ |
| 1206 |
'field' => $field_type, |
| 1207 |
'field_value' => $field_value, |
| 1208 |
'term_id' => $term->term_id, |
| 1209 |
'term_name' => $term->name, |
| 1210 |
'taxonomy' => 'product_cat' |
| 1211 |
]) |
| 1212 |
]); |
| 1213 |
} |
| 1214 |
} |
| 1215 |
|
| 1216 |
return true; |
| 1217 |
} |
| 1218 |
|
| 1219 |
return false; |
| 1220 |
} |
| 1221 |
} |
| 1222 |
|
| 1223 |
# Check if this is the home page (landing page) |
| 1224 |
$site_url = rtrim(site_url(), '/'); |
| 1225 |
$route_clean = rtrim($route, '/'); |
| 1226 |
|
| 1227 |
if ($route_clean === $site_url) { |
| 1228 |
# Get the home page (front page) |
| 1229 |
$front_page_id = get_option('page_on_front'); |
| 1230 |
$home_page = null; |
| 1231 |
|
| 1232 |
if ($front_page_id && $front_page_id > 0) { |
| 1233 |
$home_page = get_post($front_page_id); |
| 1234 |
} else { |
| 1235 |
# If no static front page is set, get the latest post |
| 1236 |
$home_page = get_posts(['numberposts' => 1, 'post_status' => 'publish'])[0] ?? null; |
| 1237 |
} |
| 1238 |
|
| 1239 |
if ($home_page) { |
| 1240 |
# Check if home page would return 404 before applying OTTO changes |
| 1241 |
if (metasync_would_page_return_404($home_page->ID, $route)) { |
| 1242 |
error_log("MetaSync OTTO: Skipping SEO processing for home page that would return 404: {$route} (Post ID: {$home_page->ID}, Status: {$home_page->post_status})"); |
| 1243 |
metasync_otto_auto_exclude_404_url($route); |
| 1244 |
return false; |
| 1245 |
} |
| 1246 |
|
| 1247 |
# Update comprehensive home page SEO meta fields |
| 1248 |
$update_result = metasync_update_comprehensive_seo_fields($home_page->ID, $seo_data); |
| 1249 |
|
| 1250 |
if ($update_result['updated']) { |
| 1251 |
# Clear relevant caches |
| 1252 |
metasync_clear_post_seo_caches($home_page->ID); |
| 1253 |
|
| 1254 |
# Prepare trimmed values to 30 characters |
| 1255 |
$trim = function($value) { |
| 1256 |
if ($value === null) { return ''; } |
| 1257 |
$value = (string) $value; |
| 1258 |
$value = trim($value); |
| 1259 |
if (mb_strlen($value) > 30) { |
| 1260 |
return mb_substr($value, 0, 30); |
| 1261 |
} |
| 1262 |
return $value; |
| 1263 |
}; |
| 1264 |
|
| 1265 |
# Log individual field updates for home page |
| 1266 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1267 |
$short = ''; |
| 1268 |
$title = ''; |
| 1269 |
|
| 1270 |
switch ($field_type) { |
| 1271 |
case 'meta_title': |
| 1272 |
$short = $trim($field_value); |
| 1273 |
$title = "Home Page Meta Title Update ({$short}...)"; |
| 1274 |
break; |
| 1275 |
case 'meta_description': |
| 1276 |
$short = $trim($field_value); |
| 1277 |
$title = "Home Page Meta Description Update ({$short}...)"; |
| 1278 |
break; |
| 1279 |
case 'meta_keywords': |
| 1280 |
$short = $trim($field_value); |
| 1281 |
$title = "Home Page Meta Keywords Update ({$short}...)"; |
| 1282 |
break; |
| 1283 |
case 'og_title': |
| 1284 |
$short = $trim($field_value); |
| 1285 |
$title = "Home Page Open Graph Title Update ({$short}...)"; |
| 1286 |
break; |
| 1287 |
case 'og_description': |
| 1288 |
$short = $trim($field_value); |
| 1289 |
$title = "Home Page Open Graph Description Update ({$short}...)"; |
| 1290 |
break; |
| 1291 |
case 'twitter_title': |
| 1292 |
$short = $trim($field_value); |
| 1293 |
$title = "Home Page Twitter Title Update ({$short}...)"; |
| 1294 |
break; |
| 1295 |
case 'twitter_description': |
| 1296 |
$short = $trim($field_value); |
| 1297 |
$title = "Home Page Twitter Description Update ({$short}...)"; |
| 1298 |
break; |
| 1299 |
case 'image_alt_data': |
| 1300 |
$image_count = count($field_value); |
| 1301 |
$title = "Home Page Image Alt Text Update ({$image_count} images)"; |
| 1302 |
break; |
| 1303 |
case 'headings_data': |
| 1304 |
$heading_count = count($field_value); |
| 1305 |
$title = "Home Page Headings Update ({$heading_count} headings)"; |
| 1306 |
break; |
| 1307 |
case 'structured_data': |
| 1308 |
$title = 'Home Page Structured Data Update'; |
| 1309 |
break; |
| 1310 |
} |
| 1311 |
|
| 1312 |
if (!empty($title)) { |
| 1313 |
metasync_log_sync_history([ |
| 1314 |
'title' => $title, |
| 1315 |
'source' => 'OTTO SEO', |
| 1316 |
'status' => 'published', |
| 1317 |
'content_type' => 'Home Page SEO', |
| 1318 |
'url' => $route, |
| 1319 |
'meta_data' => json_encode([ |
| 1320 |
'field' => $field_type, |
| 1321 |
'field_value' => $field_value, |
| 1322 |
'post_id' => $home_page->ID |
| 1323 |
]) |
| 1324 |
]); |
| 1325 |
} |
| 1326 |
} |
| 1327 |
|
| 1328 |
return true; |
| 1329 |
} |
| 1330 |
|
| 1331 |
return false; |
| 1332 |
} |
| 1333 |
} |
| 1334 |
|
| 1335 |
# Check if this is the blog/posts page (page_for_posts) |
| 1336 |
$posts_page_id = intval(get_option('page_for_posts')); |
| 1337 |
if ($posts_page_id > 0) { |
| 1338 |
$posts_page = get_post($posts_page_id); |
| 1339 |
$posts_page_url = rtrim(get_permalink($posts_page_id), '/'); |
| 1340 |
|
| 1341 |
if ($posts_page && $route_clean === $posts_page_url) { |
| 1342 |
# Check if blog page would return 404 |
| 1343 |
if (metasync_would_page_return_404($posts_page->ID, $route)) { |
| 1344 |
error_log("MetaSync OTTO: Skipping SEO processing for blog page that would return 404: {$route} (Post ID: {$posts_page->ID}, Status: {$posts_page->post_status})"); |
| 1345 |
metasync_otto_auto_exclude_404_url($route); |
| 1346 |
return false; |
| 1347 |
} |
| 1348 |
|
| 1349 |
# Update comprehensive blog page SEO meta fields |
| 1350 |
$update_result = metasync_update_comprehensive_seo_fields($posts_page->ID, $seo_data); |
| 1351 |
|
| 1352 |
if ($update_result['updated']) { |
| 1353 |
# Clear relevant caches |
| 1354 |
metasync_clear_post_seo_caches($posts_page->ID); |
| 1355 |
|
| 1356 |
# Prepare trimmed values to 30 characters |
| 1357 |
$trim = function($value) { |
| 1358 |
if ($value === null) { return ''; } |
| 1359 |
$value = (string) $value; |
| 1360 |
$value = trim($value); |
| 1361 |
if (mb_strlen($value) > 30) { |
| 1362 |
return mb_substr($value, 0, 30); |
| 1363 |
} |
| 1364 |
return $value; |
| 1365 |
}; |
| 1366 |
|
| 1367 |
# Log individual field updates for blog page |
| 1368 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1369 |
$short = ''; |
| 1370 |
$title = ''; |
| 1371 |
|
| 1372 |
switch ($field_type) { |
| 1373 |
case 'meta_title': |
| 1374 |
$short = $trim($field_value); |
| 1375 |
$title = "Blog Page Meta Title Update ({$short}...)"; |
| 1376 |
break; |
| 1377 |
case 'meta_description': |
| 1378 |
$short = $trim($field_value); |
| 1379 |
$title = "Blog Page Meta Description Update ({$short}...)"; |
| 1380 |
break; |
| 1381 |
case 'meta_keywords': |
| 1382 |
$short = $trim($field_value); |
| 1383 |
$title = "Blog Page Meta Keywords Update ({$short}...)"; |
| 1384 |
break; |
| 1385 |
case 'og_title': |
| 1386 |
$short = $trim($field_value); |
| 1387 |
$title = "Blog Page Open Graph Title Update ({$short}...)"; |
| 1388 |
break; |
| 1389 |
case 'og_description': |
| 1390 |
$short = $trim($field_value); |
| 1391 |
$title = "Blog Page Open Graph Description Update ({$short}...)"; |
| 1392 |
break; |
| 1393 |
case 'twitter_title': |
| 1394 |
$short = $trim($field_value); |
| 1395 |
$title = "Blog Page Twitter Title Update ({$short}...)"; |
| 1396 |
break; |
| 1397 |
case 'twitter_description': |
| 1398 |
$short = $trim($field_value); |
| 1399 |
$title = "Blog Page Twitter Description Update ({$short}...)"; |
| 1400 |
break; |
| 1401 |
case 'image_alt_data': |
| 1402 |
$image_count = count($field_value); |
| 1403 |
$title = "Blog Page Image Alt Text Update ({$image_count} images)"; |
| 1404 |
break; |
| 1405 |
case 'headings_data': |
| 1406 |
$heading_count = count($field_value); |
| 1407 |
$title = "Blog Page Headings Update ({$heading_count} headings)"; |
| 1408 |
break; |
| 1409 |
case 'structured_data': |
| 1410 |
$title = 'Blog Page Structured Data Update'; |
| 1411 |
break; |
| 1412 |
} |
| 1413 |
|
| 1414 |
if (!empty($title)) { |
| 1415 |
metasync_log_sync_history([ |
| 1416 |
'title' => $title, |
| 1417 |
'source' => 'OTTO SEO', |
| 1418 |
'status' => 'published', |
| 1419 |
'content_type' => 'Blog Page SEO', |
| 1420 |
'url' => $route, |
| 1421 |
'meta_data' => json_encode([ |
| 1422 |
'field' => $field_type, |
| 1423 |
'field_value' => $field_value, |
| 1424 |
'post_id' => $posts_page->ID |
| 1425 |
]) |
| 1426 |
]); |
| 1427 |
} |
| 1428 |
} |
| 1429 |
|
| 1430 |
return true; |
| 1431 |
} |
| 1432 |
|
| 1433 |
return false; |
| 1434 |
} |
| 1435 |
} |
| 1436 |
|
| 1437 |
# URL didn't resolve to any supported entity (post, category, home page, blog page) |
| 1438 |
# Treat as 404 and auto-exclude (e.g. deleted post, non-existent page) |
| 1439 |
if (!metasync_otto_is_url_available($route)) { |
| 1440 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404 (no matching entity): {$route}"); |
| 1441 |
metasync_otto_auto_exclude_404_url($route); |
| 1442 |
} |
| 1443 |
return false; |
| 1444 |
} |
| 1445 |
|
| 1446 |
# Verify this is actually a post, page, or WooCommerce product |
| 1447 |
$post = get_post($post_id); |
| 1448 |
|
| 1449 |
# Get supported post types dynamically |
| 1450 |
$supported_post_types = metasync_get_supported_post_types(); |
| 1451 |
|
| 1452 |
if (!$post || !in_array($post->post_type, $supported_post_types)) { |
| 1453 |
# Skip unsupported post types |
| 1454 |
return false; |
| 1455 |
} |
| 1456 |
|
| 1457 |
# Check if page would return 404 before applying OTTO changes |
| 1458 |
if (metasync_would_page_return_404($post_id, $route)) { |
| 1459 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404: {$route} (Post ID: {$post_id}, Status: {$post->post_status})"); |
| 1460 |
metasync_otto_auto_exclude_404_url($route); |
| 1461 |
return false; |
| 1462 |
} |
| 1463 |
|
| 1464 |
# Update comprehensive SEO meta fields |
| 1465 |
$update_result = metasync_update_comprehensive_seo_fields($post_id, $seo_data); |
| 1466 |
|
| 1467 |
if ($update_result['updated']) { |
| 1468 |
# Clear relevant caches |
| 1469 |
metasync_clear_post_seo_caches($post_id); |
| 1470 |
|
| 1471 |
# Prepare trimmed values to 30 characters |
| 1472 |
$trim = function($value) { |
| 1473 |
if ($value === null) { return ''; } |
| 1474 |
$value = (string) $value; |
| 1475 |
$value = trim($value); |
| 1476 |
if (mb_strlen($value) > 30) { |
| 1477 |
return mb_substr($value, 0, 30); |
| 1478 |
} |
| 1479 |
return $value; |
| 1480 |
}; |
| 1481 |
|
| 1482 |
# Log individual field updates |
| 1483 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1484 |
$short = ''; |
| 1485 |
$title = ''; |
| 1486 |
|
| 1487 |
switch ($field_type) { |
| 1488 |
case 'meta_title': |
| 1489 |
$short = $trim($field_value); |
| 1490 |
$title = 'Meta Title Update (' . $short . '...)'; |
| 1491 |
break; |
| 1492 |
case 'meta_description': |
| 1493 |
$short = $trim($field_value); |
| 1494 |
$title = 'Meta Description Update (' . $short . '...)'; |
| 1495 |
break; |
| 1496 |
case 'meta_keywords': |
| 1497 |
$short = $trim($field_value); |
| 1498 |
$title = 'Meta Keywords Update (' . $short . '...)'; |
| 1499 |
break; |
| 1500 |
case 'og_title': |
| 1501 |
$short = $trim($field_value); |
| 1502 |
$title = 'Open Graph Title Update (' . $short . '...)'; |
| 1503 |
break; |
| 1504 |
case 'og_description': |
| 1505 |
$short = $trim($field_value); |
| 1506 |
$title = 'Open Graph Description Update (' . $short . '...)'; |
| 1507 |
break; |
| 1508 |
case 'twitter_title': |
| 1509 |
$short = $trim($field_value); |
| 1510 |
$title = 'Twitter Title Update (' . $short . '...)'; |
| 1511 |
break; |
| 1512 |
case 'twitter_description': |
| 1513 |
$short = $trim($field_value); |
| 1514 |
$title = 'Twitter Description Update (' . $short . '...)'; |
| 1515 |
break; |
| 1516 |
case 'image_alt_data': |
| 1517 |
$image_count = count($field_value); |
| 1518 |
$title = "Image Alt Text Update ({$image_count} images)"; |
| 1519 |
break; |
| 1520 |
case 'headings_data': |
| 1521 |
$heading_count = count($field_value); |
| 1522 |
$title = "Headings Update ({$heading_count} headings)"; |
| 1523 |
break; |
| 1524 |
case 'structured_data': |
| 1525 |
$title = 'Structured Data Update'; |
| 1526 |
break; |
| 1527 |
} |
| 1528 |
|
| 1529 |
if (!empty($title)) { |
| 1530 |
metasync_log_sync_history([ |
| 1531 |
'title' => $title, |
| 1532 |
'source' => 'OTTO SEO', |
| 1533 |
'status' => 'published', |
| 1534 |
'content_type' => 'SEO Meta', |
| 1535 |
'url' => $route, |
| 1536 |
'meta_data' => json_encode([ |
| 1537 |
'field' => $field_type, |
| 1538 |
'field_value' => $field_value, |
| 1539 |
'post_id' => $post_id |
| 1540 |
]) |
| 1541 |
]); |
| 1542 |
} |
| 1543 |
} |
| 1544 |
|
| 1545 |
return true; |
| 1546 |
} |
| 1547 |
|
| 1548 |
return false; |
| 1549 |
|
| 1550 |
} catch (Exception $e) { |
| 1551 |
metasync_record_failed_action( 'metasync_process_seo_job' ); |
| 1552 |
return false; |
| 1553 |
} finally { |
| 1554 |
# Release the concurrency lock only if we actually acquired it. |
| 1555 |
# Early returns (CPU deferral, lock contention) must NOT delete a lock |
| 1556 |
# that another process may be holding. |
| 1557 |
if ($lock_acquired && $lock_key) { |
| 1558 |
delete_transient($lock_key); |
| 1559 |
} |
| 1560 |
} |
| 1561 |
} |
| 1562 |
|
| 1563 |
/** |
| 1564 |
* Log sync history entry |
| 1565 |
* @param array $data Sync data to log |
| 1566 |
*/ |
| 1567 |
function metasync_log_sync_history($data) { |
| 1568 |
try { |
| 1569 |
// Classes are now autoloaded, no need for manual require |
| 1570 |
$sync_db = new Metasync_Sync_History_Database(); |
| 1571 |
|
| 1572 |
// Minimal duplicate prevention within short time window |
| 1573 |
if (!empty($data['title']) && !empty($data['source'])) { |
| 1574 |
global $wpdb; |
| 1575 |
$table = $wpdb->prefix . Metasync_Sync_History_Database::$table_name; |
| 1576 |
$recent = $wpdb->get_var($wpdb->prepare( |
| 1577 |
"SELECT COUNT(*) FROM `$table` WHERE title = %s AND source = %s AND created_at >= %s", |
| 1578 |
$data['title'], |
| 1579 |
$data['source'], |
| 1580 |
gmdate('Y-m-d H:i:s', time() - 60) |
| 1581 |
)); |
| 1582 |
if ((int)$recent > 0) { |
| 1583 |
return; // skip duplicate log within 60 seconds |
| 1584 |
} |
| 1585 |
} |
| 1586 |
|
| 1587 |
$sync_db->add($data); |
| 1588 |
|
| 1589 |
} catch (Exception $e) { |
| 1590 |
error_log("MetaSync: Failed to log sync history: " . $e->getMessage()); |
| 1591 |
} |
| 1592 |
} |
| 1593 |
|
| 1594 |
/** |
| 1595 |
* Get supported post types for OTTO SEO optimization |
| 1596 |
* Includes WooCommerce products if WooCommerce is active |
| 1597 |
* |
| 1598 |
* @return array List of supported post types |
| 1599 |
*/ |
| 1600 |
function metasync_get_supported_post_types() { |
| 1601 |
# Start with default post types |
| 1602 |
$post_types = ['post', 'page']; |
| 1603 |
|
| 1604 |
# Add WooCommerce product post type if WooCommerce is active |
| 1605 |
if (class_exists('WooCommerce') || function_exists('is_woocommerce')) { |
| 1606 |
$post_types[] = 'product'; |
| 1607 |
} |
| 1608 |
|
| 1609 |
# Include all public custom post types (e.g. 'location', 'service', 'team', etc.) |
| 1610 |
# so OTTO can write post meta for them during metasync_process_otto_seo_data(). |
| 1611 |
$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names'); |
| 1612 |
if (!empty($custom_post_types)) { |
| 1613 |
$post_types = array_merge($post_types, array_values($custom_post_types)); |
| 1614 |
} |
| 1615 |
|
| 1616 |
# Allow developers to filter supported post types |
| 1617 |
$post_types = apply_filters('metasync_otto_supported_post_types', $post_types); |
| 1618 |
|
| 1619 |
return $post_types; |
| 1620 |
} |
| 1621 |
|
| 1622 |
/** |
| 1623 |
* Get supported taxonomies for OTTO SEO optimization |
| 1624 |
* Includes WooCommerce product categories and tags if WooCommerce is active |
| 1625 |
* |
| 1626 |
* @return array List of supported taxonomies |
| 1627 |
*/ |
| 1628 |
function metasync_get_supported_taxonomies() { |
| 1629 |
# Start with default taxonomies |
| 1630 |
$taxonomies = ['category']; |
| 1631 |
|
| 1632 |
# Add WooCommerce taxonomies if WooCommerce is active |
| 1633 |
if (class_exists('WooCommerce') || function_exists('is_woocommerce')) { |
| 1634 |
$taxonomies[] = 'product_cat'; # WooCommerce product categories |
| 1635 |
$taxonomies[] = 'product_tag'; # WooCommerce product tags |
| 1636 |
} |
| 1637 |
|
| 1638 |
# Allow developers to filter supported taxonomies |
| 1639 |
$taxonomies = apply_filters('metasync_otto_supported_taxonomies', $taxonomies); |
| 1640 |
|
| 1641 |
return $taxonomies; |
| 1642 |
} |
| 1643 |
|
| 1644 |
/** |
| 1645 |
* Resolve a URL through the Redirect Manager table to its final destination (follows redirect chains). |
| 1646 |
* Used before 404 checks and OTTO processing so the final canonical URL is used, not intermediate redirects. |
| 1647 |
* |
| 1648 |
* @param string $url Full URL (e.g. https://example.com/old-page) |
| 1649 |
* @return string Final destination URL, or original $url if no redirect matches |
| 1650 |
*/ |
| 1651 |
function metasync_otto_resolve_redirect_to_final_url($url) |
| 1652 |
{ |
| 1653 |
if (empty($url) || !is_string($url)) { |
| 1654 |
return $url; |
| 1655 |
} |
| 1656 |
try { |
| 1657 |
$db_path = plugin_dir_path(dirname(__FILE__)) . 'redirections/class-metasync-redirection-database.php'; |
| 1658 |
$class_path = plugin_dir_path(dirname(__FILE__)) . 'redirections/class-metasync-redirection.php'; |
| 1659 |
if (!file_exists($db_path) || !file_exists($class_path)) { |
| 1660 |
return $url; |
| 1661 |
} |
| 1662 |
require_once $db_path; |
| 1663 |
require_once $class_path; |
| 1664 |
$db = new Metasync_Redirection_Database(); |
| 1665 |
$redirect = new Metasync_Redirection($db); |
| 1666 |
return $redirect->resolve_url_to_final_destination($url, 10); |
| 1667 |
} catch (Exception $e) { |
| 1668 |
error_log('MetaSync OTTO: Redirect resolution failed for ' . $url . ' - ' . $e->getMessage()); |
| 1669 |
return $url; |
| 1670 |
} |
| 1671 |
} |
| 1672 |
|
| 1673 |
/** |
| 1674 |
* Auto-exclude a URL from OTTO with description "Auto-excluded: 404" |
| 1675 |
* Called when a URL is detected as returning 404 so it won't be sent to OTTO again |
| 1676 |
* |
| 1677 |
* @param string $url Full URL to exclude (e.g. https://example.com/404-page) |
| 1678 |
* @return bool|string True on success, false on failure, 'duplicate'/'reactivated' if already exists |
| 1679 |
*/ |
| 1680 |
function metasync_otto_auto_exclude_404_url($url) |
| 1681 |
{ |
| 1682 |
if (empty($url) || !is_string($url)) { |
| 1683 |
return false; |
| 1684 |
} |
| 1685 |
$url = filter_var($url, FILTER_SANITIZE_URL); |
| 1686 |
$url = esc_url_raw($url); |
| 1687 |
if (empty($url) || mb_strlen($url) > 2048) { |
| 1688 |
return false; |
| 1689 |
} |
| 1690 |
try { |
| 1691 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 1692 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 1693 |
return $db->add([ |
| 1694 |
'url_pattern' => $url, |
| 1695 |
'pattern_type' => 'exact', |
| 1696 |
'description' => 'Auto-excluded: 404', |
| 1697 |
'status' => 'active', |
| 1698 |
'auto_excluded' => 1, |
| 1699 |
]); |
| 1700 |
} catch (Exception $e) { |
| 1701 |
error_log('MetaSync OTTO: Failed to auto-exclude 404 URL: ' . $url . ' - ' . $e->getMessage()); |
| 1702 |
return false; |
| 1703 |
} |
| 1704 |
} |
| 1705 |
|
| 1706 |
/** |
| 1707 |
* Remove a URL from the OTTO auto-exclusion list. |
| 1708 |
* Called when OTTO sends a webhook for a URL, confirming it is valid and crawlable. |
| 1709 |
* Only removes records where auto_excluded = 1 (never removes manual exclusions). |
| 1710 |
* |
| 1711 |
* @param string $url Full URL to un-exclude (e.g. https://example.com/location/page) |
| 1712 |
* @return bool True on success |
| 1713 |
*/ |
| 1714 |
function metasync_otto_remove_auto_exclusion($url) |
| 1715 |
{ |
| 1716 |
if (empty($url) || !is_string($url)) { |
| 1717 |
return false; |
| 1718 |
} |
| 1719 |
try { |
| 1720 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 1721 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 1722 |
global $wpdb; |
| 1723 |
$table = $wpdb->prefix . Metasync_Otto_Excluded_URLs_Database::$table_name; |
| 1724 |
# Normalize the same way is_url_excluded() does |
| 1725 |
$url_normalized = rtrim(trim($url), '/'); |
| 1726 |
$records = $wpdb->get_results( |
| 1727 |
$wpdb->prepare( |
| 1728 |
"SELECT id FROM `{$table}` WHERE url_pattern = %s AND auto_excluded = 1 AND status = 'active'", |
| 1729 |
$url_normalized |
| 1730 |
) |
| 1731 |
); |
| 1732 |
if (!empty($records)) { |
| 1733 |
$ids = array_map(function ($r) { return (int) $r->id; }, $records); |
| 1734 |
$db->delete($ids); |
| 1735 |
} |
| 1736 |
return true; |
| 1737 |
} catch (Exception $e) { |
| 1738 |
return false; |
| 1739 |
} |
| 1740 |
} |
| 1741 |
|
| 1742 |
/** |
| 1743 |
* Check if a URL is MANUALLY excluded from OTTO (auto_excluded = 0). |
| 1744 |
* Used at render time (metasync_start_otto) — auto-exclusions must NOT block |
| 1745 |
* rendering because they are often false positives (e.g. custom post types that |
| 1746 |
* url_to_postid() can't resolve). Auto-exclusions are only used to gate the |
| 1747 |
* SEO meta-writing webhook path. |
| 1748 |
* |
| 1749 |
* @param string $url URL to check |
| 1750 |
* @return bool True if URL has a manual exclusion |
| 1751 |
*/ |
| 1752 |
function metasync_is_otto_url_manually_excluded($url) |
| 1753 |
{ |
| 1754 |
if (empty($url) || !is_string($url)) { |
| 1755 |
return false; |
| 1756 |
} |
| 1757 |
try { |
| 1758 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 1759 |
global $wpdb; |
| 1760 |
$table = $wpdb->prefix . Metasync_Otto_Excluded_URLs_Database::$table_name; |
| 1761 |
$url_normalized = rtrim(trim($url), '/'); |
| 1762 |
|
| 1763 |
$records = get_transient(METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY); |
| 1764 |
|
| 1765 |
if ($records === false) { |
| 1766 |
// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- no user input, table name from $wpdb->prefix |
| 1767 |
$records = $wpdb->get_results( |
| 1768 |
"SELECT url_pattern, pattern_type FROM `{$table}` WHERE status = 'active' AND (auto_excluded = 0 OR auto_excluded IS NULL) ORDER BY created_at DESC" |
| 1769 |
); |
| 1770 |
|
| 1771 |
// Graceful recovery: auto_excluded column missing on pre-v2.7.4 installs. |
| 1772 |
// Run ALTER TABLE to add it and treat URL as not excluded so OTTO continues rendering. |
| 1773 |
if ($wpdb->last_error && strpos($wpdb->last_error, 'auto_excluded') !== false) { |
| 1774 |
// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared |
| 1775 |
$wpdb->query("ALTER TABLE `{$table}` ADD COLUMN `auto_excluded` TINYINT(1) NOT NULL DEFAULT 0"); |
| 1776 |
return false; |
| 1777 |
} |
| 1778 |
|
| 1779 |
set_transient(METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY, $records ?: [], METASYNC_OTTO_EXCLUDED_TRANSIENT_TTL); |
| 1780 |
} |
| 1781 |
|
| 1782 |
if (empty($records)) { |
| 1783 |
return false; |
| 1784 |
} |
| 1785 |
|
| 1786 |
foreach ($records as $excluded) { |
| 1787 |
$pattern = rtrim(trim($excluded->url_pattern), '/'); |
| 1788 |
$pattern_type = $excluded->pattern_type; |
| 1789 |
|
| 1790 |
switch ($pattern_type) { |
| 1791 |
case 'exact': |
| 1792 |
if ($url_normalized === $pattern) { |
| 1793 |
return true; |
| 1794 |
} |
| 1795 |
break; |
| 1796 |
case 'contain': |
| 1797 |
if (strpos($url_normalized, $pattern) !== false) { |
| 1798 |
return true; |
| 1799 |
} |
| 1800 |
break; |
| 1801 |
case 'start': |
| 1802 |
if (strpos($url_normalized, $pattern) === 0) { |
| 1803 |
return true; |
| 1804 |
} |
| 1805 |
break; |
| 1806 |
} |
| 1807 |
} |
| 1808 |
return false; |
| 1809 |
} catch (Exception $e) { |
| 1810 |
return false; |
| 1811 |
} |
| 1812 |
} |
| 1813 |
|
| 1814 |
/** |
| 1815 |
* Check if a URL is excluded from OTTO |
| 1816 |
* @param string $url URL to check |
| 1817 |
* @return bool True if URL is excluded, false otherwise |
| 1818 |
*/ |
| 1819 |
function metasync_is_otto_url_excluded($url) |
| 1820 |
{ |
| 1821 |
try { |
| 1822 |
// Load database class |
| 1823 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 1824 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 1825 |
|
| 1826 |
// Check if URL is excluded |
| 1827 |
return $db->is_url_excluded($url); |
| 1828 |
|
| 1829 |
} catch (Exception $e) { |
| 1830 |
return false; |
| 1831 |
} |
| 1832 |
} |
| 1833 |
|
| 1834 |
/** |
| 1835 |
* Check if a URL is now available (would NOT return 404) |
| 1836 |
* Uses same resolution logic as metasync_process_otto_seo_data |
| 1837 |
* Used when rechecking auto-excluded 404 URLs after 7 days |
| 1838 |
* |
| 1839 |
* @param string $url Full URL to check (e.g. https://example.com/page) |
| 1840 |
* @return bool True if URL is accessible, false if it would return 404 |
| 1841 |
*/ |
| 1842 |
function metasync_otto_is_url_available($url) |
| 1843 |
{ |
| 1844 |
if (empty($url) || !is_string($url)) { |
| 1845 |
return false; |
| 1846 |
} |
| 1847 |
|
| 1848 |
$route = metasync_otto_resolve_redirect_to_final_url($url); |
| 1849 |
$post_id = url_to_postid($route); |
| 1850 |
|
| 1851 |
# WooCommerce shop page |
| 1852 |
if ((!$post_id || $post_id <= 0) && function_exists('wc_get_page_id')) { |
| 1853 |
$shop_page_id = wc_get_page_id('shop'); |
| 1854 |
if ($shop_page_id > 0) { |
| 1855 |
$shop_url = get_permalink($shop_page_id); |
| 1856 |
if (rtrim($route, '/') === rtrim($shop_url, '/')) { |
| 1857 |
$post_id = $shop_page_id; |
| 1858 |
} |
| 1859 |
} |
| 1860 |
} |
| 1861 |
|
| 1862 |
# WooCommerce product by slug |
| 1863 |
if ((!$post_id || $post_id <= 0) && strpos($route, '/product/') !== false && function_exists('wc_get_products')) { |
| 1864 |
$product_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1865 |
$products = wc_get_products(array('name' => $product_slug, 'limit' => 1, 'status' => 'publish')); |
| 1866 |
if (!empty($products)) { |
| 1867 |
$post_id = $products[0]->get_id(); |
| 1868 |
} else { |
| 1869 |
$query = new WP_Query(array( |
| 1870 |
'post_type' => 'product', |
| 1871 |
'name' => $product_slug, |
| 1872 |
'posts_per_page' => 1, |
| 1873 |
'post_status' => 'publish', |
| 1874 |
)); |
| 1875 |
if ($query->have_posts()) { |
| 1876 |
$post_id = $query->posts[0]->ID; |
| 1877 |
} |
| 1878 |
} |
| 1879 |
} |
| 1880 |
|
| 1881 |
if ($post_id && $post_id > 0) { |
| 1882 |
$post = get_post($post_id); |
| 1883 |
# Accept any post type — custom post types (e.g. 'location', 'service') are valid URLs. |
| 1884 |
# The old in_array check against metasync_get_supported_post_types() caused CPT URLs |
| 1885 |
# to be wrongly auto-excluded as "404" pages. |
| 1886 |
if ($post) { |
| 1887 |
return !metasync_would_page_return_404($post_id, $route); |
| 1888 |
} |
| 1889 |
} |
| 1890 |
|
| 1891 |
# Category |
| 1892 |
if (strpos($route, '/category/') !== false) { |
| 1893 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1894 |
$category = get_category_by_slug($category_slug); |
| 1895 |
if ($category) { |
| 1896 |
return !metasync_would_term_return_404($category->term_id, 'category', $route); |
| 1897 |
} |
| 1898 |
} |
| 1899 |
|
| 1900 |
# WooCommerce product category |
| 1901 |
if (strpos($route, '/product-category/') !== false) { |
| 1902 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1903 |
$term = get_term_by('slug', $category_slug, 'product_cat'); |
| 1904 |
if ($term && !is_wp_error($term)) { |
| 1905 |
return !metasync_would_term_return_404($term->term_id, 'product_cat', $route); |
| 1906 |
} |
| 1907 |
} |
| 1908 |
|
| 1909 |
# Home page |
| 1910 |
if (rtrim($route, '/') === rtrim(site_url(), '/')) { |
| 1911 |
$front_page_id = get_option('page_on_front'); |
| 1912 |
$home_page = ($front_page_id && $front_page_id > 0) |
| 1913 |
? get_post($front_page_id) |
| 1914 |
: (get_posts(['numberposts' => 1, 'post_status' => 'publish'])[0] ?? null); |
| 1915 |
if ($home_page) { |
| 1916 |
return !metasync_would_page_return_404($home_page->ID, $route); |
| 1917 |
} |
| 1918 |
} |
| 1919 |
|
| 1920 |
# Could not verify availability from local data (custom archive, paginated page, etc.). |
| 1921 |
# Assume the URL IS available — OTTO only crawls reachable URLs, so if we can't |
| 1922 |
# prove it's a 404, we should not auto-exclude it. |
| 1923 |
return true; |
| 1924 |
} |
| 1925 |
|
| 1926 |
/** |
| 1927 |
* Recheck auto-excluded 404 URLs when recheck_after has passed; remove from exclusion if now available |
| 1928 |
* Uses recheck_after timestamp (default 7 days from exclusion) to decide when to recheck |
| 1929 |
* Mark as permanent after 30 days if still 404 (no further rechecks) |
| 1930 |
* Called by daily cron job |
| 1931 |
*/ |
| 1932 |
function metasync_otto_recheck_404_exclusions() |
| 1933 |
{ |
| 1934 |
try { |
| 1935 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 1936 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 1937 |
$records = $db->get_auto_excluded_404_urls_due_for_recheck(); |
| 1938 |
|
| 1939 |
if (empty($records)) { |
| 1940 |
return; |
| 1941 |
} |
| 1942 |
|
| 1943 |
$removed = 0; |
| 1944 |
$marked_permanent = 0; |
| 1945 |
$thirty_days_ago = strtotime('-30 days'); |
| 1946 |
$next_recheck = date('Y-m-d H:i:s', current_time('timestamp') + 7 * DAY_IN_SECONDS); |
| 1947 |
|
| 1948 |
foreach ($records as $record) { |
| 1949 |
$url = trim($record->url_pattern); |
| 1950 |
if (empty($url)) { |
| 1951 |
continue; |
| 1952 |
} |
| 1953 |
if (metasync_otto_is_url_available($url)) { |
| 1954 |
$db->delete([$record->id]); |
| 1955 |
$removed++; |
| 1956 |
} else { |
| 1957 |
# Still 404: if excluded 30+ days ago, mark as permanent (no more rechecks) |
| 1958 |
$created_ts = strtotime($record->created_at); |
| 1959 |
if ($created_ts <= $thirty_days_ago) { |
| 1960 |
$db->update(['is_permanent' => 1], $record->id); |
| 1961 |
$marked_permanent++; |
| 1962 |
} else { |
| 1963 |
# Schedule next recheck in 7 days |
| 1964 |
$db->update(['recheck_after' => $next_recheck], $record->id); |
| 1965 |
} |
| 1966 |
} |
| 1967 |
} |
| 1968 |
|
| 1969 |
if ($removed > 0) { |
| 1970 |
error_log("MetaSync OTTO: Recheck 404 exclusions - removed {$removed} URL(s) that are now available"); |
| 1971 |
} |
| 1972 |
if ($marked_permanent > 0) { |
| 1973 |
error_log("MetaSync OTTO: Recheck 404 exclusions - marked {$marked_permanent} URL(s) as permanent (still 404 after 30 days)"); |
| 1974 |
} |
| 1975 |
} catch (Exception $e) { |
| 1976 |
error_log('MetaSync OTTO: Recheck 404 exclusions failed - ' . $e->getMessage()); |
| 1977 |
} |
| 1978 |
} |
| 1979 |
|
| 1980 |
add_action('metasync_otto_recheck_404_exclusions', 'metasync_otto_recheck_404_exclusions'); |
| 1981 |
|
| 1982 |
/** |
| 1983 |
* Check if a post/page would return 404 without making HTTP request |
| 1984 |
* Uses WordPress database checks for fast validation |
| 1985 |
* |
| 1986 |
* @param int $post_id WordPress post ID |
| 1987 |
* @param string $url The URL being checked (optional, for logging) |
| 1988 |
* @return bool True if page would return 404, false if accessible |
| 1989 |
*/ |
| 1990 |
function metasync_would_page_return_404($post_id, $url = '') { |
| 1991 |
if (!$post_id || $post_id <= 0) { |
| 1992 |
return true; // No post ID = 404 |
| 1993 |
} |
| 1994 |
|
| 1995 |
# Get the post object |
| 1996 |
$post = get_post($post_id); |
| 1997 |
if (!$post) { |
| 1998 |
return true; // Post doesn't exist = 404 |
| 1999 |
} |
| 2000 |
|
| 2001 |
# 1. Check post status - must be 'publish' to be publicly accessible |
| 2002 |
if ($post->post_status !== 'publish') { |
| 2003 |
return true; // Draft, pending, private, etc. = 404 |
| 2004 |
} |
| 2005 |
|
| 2006 |
# 2. Check if post is password protected (requires password to view) |
| 2007 |
if (!empty($post->post_password)) { |
| 2008 |
# Password protected posts are not publicly accessible without password |
| 2009 |
return true; // Password protected = effectively 404 for public |
| 2010 |
} |
| 2011 |
|
| 2012 |
# 3. Check if post is in trash |
| 2013 |
if ($post->post_status === 'trash') { |
| 2014 |
return true; // Trashed = 404 |
| 2015 |
} |
| 2016 |
|
| 2017 |
# 4. Check if post type is publicly queryable |
| 2018 |
$post_type_object = get_post_type_object($post->post_type); |
| 2019 |
if ($post_type_object && !$post_type_object->publicly_queryable) { |
| 2020 |
# Some post types might not be publicly accessible |
| 2021 |
# But we allow if it's in our supported types |
| 2022 |
$supported_post_types = metasync_get_supported_post_types(); |
| 2023 |
if (!in_array($post->post_type, $supported_post_types)) { |
| 2024 |
return true; // Not publicly queryable = 404 |
| 2025 |
} |
| 2026 |
} |
| 2027 |
|
| 2028 |
# 5. WordPress 5.7+ has a built-in function for this |
| 2029 |
if (function_exists('is_post_publicly_viewable')) { |
| 2030 |
if (!is_post_publicly_viewable($post)) { |
| 2031 |
return true; // Not publicly viewable = 404 |
| 2032 |
} |
| 2033 |
} |
| 2034 |
|
| 2035 |
# 6. Check if post is scheduled for future (not yet published) |
| 2036 |
if ($post->post_date > current_time('mysql')) { |
| 2037 |
return true; // Future post = 404 until publish date |
| 2038 |
} |
| 2039 |
|
| 2040 |
# All checks passed - page should be accessible |
| 2041 |
return false; |
| 2042 |
} |
| 2043 |
|
| 2044 |
/** |
| 2045 |
* Check if a taxonomy term (category, tag, etc.) would return 404 |
| 2046 |
* Uses WordPress database checks for fast validation |
| 2047 |
* |
| 2048 |
* @param int $term_id Term ID |
| 2049 |
* @param string $taxonomy Taxonomy name (e.g., 'category', 'product_cat') |
| 2050 |
* @param string $url The URL being checked (optional, for logging) |
| 2051 |
* @return bool True if term would return 404, false if accessible |
| 2052 |
*/ |
| 2053 |
function metasync_would_term_return_404($term_id, $taxonomy, $url = '') { |
| 2054 |
if (!$term_id || $term_id <= 0 || empty($taxonomy)) { |
| 2055 |
return true; // Invalid term = 404 |
| 2056 |
} |
| 2057 |
|
| 2058 |
# Get the term object |
| 2059 |
$term = get_term($term_id, $taxonomy); |
| 2060 |
if (is_wp_error($term) || !$term) { |
| 2061 |
return true; // Term doesn't exist = 404 |
| 2062 |
} |
| 2063 |
|
| 2064 |
# Check if taxonomy is publicly queryable |
| 2065 |
$taxonomy_object = get_taxonomy($taxonomy); |
| 2066 |
if (!$taxonomy_object || !$taxonomy_object->public) { |
| 2067 |
# Check if it's in our supported taxonomies |
| 2068 |
$supported_taxonomies = metasync_get_supported_taxonomies(); |
| 2069 |
if (!in_array($taxonomy, $supported_taxonomies)) { |
| 2070 |
return true; // Not publicly queryable = 404 |
| 2071 |
} |
| 2072 |
} |
| 2073 |
|
| 2074 |
# Terms are generally always accessible if they exist and taxonomy is public |
| 2075 |
# WordPress doesn't have a "draft" status for terms like posts do |
| 2076 |
# But we can check if the term has a count (has posts assigned) |
| 2077 |
# Empty terms might not be useful, but they're still accessible |
| 2078 |
|
| 2079 |
# All checks passed - term should be accessible |
| 2080 |
return false; |
| 2081 |
} |
| 2082 |
|
| 2083 |
/** |
| 2084 |
* Invalidate Brizy posts cache when posts are saved |
| 2085 |
* OPTIMIZATION: Clears transient cache to ensure accurate detection |
| 2086 |
*/ |
| 2087 |
add_action('save_post', function($post_id) { |
| 2088 |
# Check if this post has Brizy metadata |
| 2089 |
if (get_post_meta($post_id, 'brizy_post_uid', true)) { |
| 2090 |
delete_transient('metasync_has_brizy_posts'); |
| 2091 |
} |
| 2092 |
}, 10, 1); |
