| 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-job-status.php'; |
| 26 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-render-strategy.php'; |
| 27 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-config.php'; |
| 28 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-bot-detector.php'; |
| 29 |
require_once plugin_dir_path( __FILE__ ) . '/class-metasync-otto-bot-statistics-database.php'; |
| 30 |
|
| 31 |
/** |
| 32 |
* Is $method actually available on $class in this process right now? |
| 33 |
* |
| 34 |
* A partially updated install can leave a newer copy of one plugin file beside |
| 35 |
* an older copy of another — stale opcache bytecode for a single file is enough. |
| 36 |
* class_exists() is satisfied by the older copy, and calling a method it does |
| 37 |
* not declare is a fatal. Everything in this file runs on the front end, so that |
| 38 |
* fatal is a white screen on every page view rather than a degraded feature. |
| 39 |
* |
| 40 |
* Defined here rather than pulled from a shared utility class on purpose: a |
| 41 |
* helper loaded from another file could itself be the stale one. |
| 42 |
* |
| 43 |
* $class and $method are parameters rather than literals at the call site so the |
| 44 |
* check survives static analysis, which would otherwise narrow a literal |
| 45 |
* method_exists() on a known class to a constant true — the skew this guards |
| 46 |
* against exists only at runtime. |
| 47 |
* |
| 48 |
* @param string $class Class about to be called. |
| 49 |
* @param string $method Method about to be called on it. |
| 50 |
* @return bool |
| 51 |
*/ |
| 52 |
function metasync_otto_class_provides($class, $method) { |
| 53 |
return class_exists($class) && method_exists($class, $method); |
| 54 |
} |
| 55 |
|
| 56 |
# OPTIMIZED: get the metasync options (cached in static class) |
| 57 |
$metasync_options = Metasync_Otto_Config::get_options(); |
| 58 |
|
| 59 |
# OTTO SSR is always enabled by default |
| 60 |
$otto_enabled = true; |
| 61 |
|
| 62 |
# add tag to wp head |
| 63 |
add_action('wp_head', function(){ |
| 64 |
# load globals |
| 65 |
global $metasync_options, $otto_enabled; |
| 66 |
|
| 67 |
# OTTO SSR is always enabled |
| 68 |
$string_enabled = 'true'; |
| 69 |
|
| 70 |
# OPTIMIZED: check uuid set using cached config |
| 71 |
if(!Metasync_Otto_Config::is_otto_enabled()){ |
| 72 |
return; |
| 73 |
} |
| 74 |
|
| 75 |
# Performance optimization: Add DNS prefetch and preconnect for OTTO API |
| 76 |
# This improves connection speed by resolving DNS and establishing connections early |
| 77 |
# Use endpoint manager to get the correct domain |
| 78 |
$otto_domain = 'sa.searchatlas.com'; # default |
| 79 |
if (class_exists('Metasync_Endpoint_Manager')) { |
| 80 |
$otto_api_domain = Metasync_Endpoint_Manager::get_endpoint('OTTO_API_DOMAIN'); |
| 81 |
$parsed = parse_url($otto_api_domain); |
| 82 |
if (!empty($parsed['host'])) { |
| 83 |
$otto_domain = $parsed['host']; |
| 84 |
} |
| 85 |
} |
| 86 |
echo '<link rel="dns-prefetch" href="//' . esc_attr($otto_domain) . '">' . "\n"; |
| 87 |
echo '<link rel="preconnect" href="https://' . esc_attr($otto_domain) . '" crossorigin>' . "\n"; |
| 88 |
|
| 89 |
# adding the otto tag to pages |
| 90 |
$plugin_version = defined('METASYNC_VERSION') ? METASYNC_VERSION : 'unknown'; |
| 91 |
# OPTIMIZED: use cached uuid |
| 92 |
$otto_tag = '<meta name="otto" content="uuid='.esc_attr(Metasync_Otto_Config::get_otto_uuid()).'; type=wordpress; enabled='.esc_attr($string_enabled).'; version='.esc_attr($plugin_version).'">'; |
| 93 |
|
| 94 |
# out the otto tag |
| 95 |
echo $otto_tag; |
| 96 |
}, 1); # Priority 1 to output early in head |
| 97 |
|
| 98 |
/** |
| 99 |
* Scheduling seam for the OTTO pipeline. |
| 100 |
* |
| 101 |
* wp_schedule_single_event() returns false on failure and the old pipeline |
| 102 |
* ignored it, so a broken cron table acknowledged URLs that were never |
| 103 |
* queued. Every schedule in this pipeline goes through this seam so the |
| 104 |
* return value is honored (and tests can observe scheduling). |
| 105 |
*/ |
| 106 |
function metasync_otto_schedule_single_event($timestamp, $hook, $args = array()) { |
| 107 |
return wp_schedule_single_event($timestamp, $hook, $args) === true; |
| 108 |
} |
| 109 |
|
| 110 |
/** |
| 111 |
* Dedupe seam: does this URL already have live per-URL work in flight? |
| 112 |
* |
| 113 |
* Combines the durable status store with the raw cron table so a re-delivered |
| 114 |
* webhook cannot stack a second job behind a live one. |
| 115 |
*/ |
| 116 |
function metasync_otto_job_already_scheduled($route) { |
| 117 |
if (class_exists('Metasync_Otto_Job_Status', false) |
| 118 |
&& Metasync_Otto_Job_Status::has_active($route)) { |
| 119 |
return true; |
| 120 |
} |
| 121 |
return wp_next_scheduled('metasync_process_otto_crawl_url_job', array($route)) !== false; |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Cron-table dedupe seam: does a per-URL cron event already exist for this URL? |
| 126 |
* |
| 127 |
* The pending-queue drainer must dedupe against the scheduler table only. |
| 128 |
* Every URL it takes off the durable queue carries a fresh webhook |
| 129 |
* accepted/batch_overflow ledger row; that row exists so a re-delivered |
| 130 |
* webhook cannot stack a second job behind live work, and consulting it |
| 131 |
* from the drainer — after take_pending() has already removed the URL |
| 132 |
* from the queue — silently dropped every overflow URL instead of |
| 133 |
* draining it. Only a real, still-pending cron event may veto here. |
| 134 |
*/ |
| 135 |
function metasync_otto_job_event_pending($route) { |
| 136 |
return wp_next_scheduled('metasync_process_otto_crawl_url_job', array($route)) !== false; |
| 137 |
} |
| 138 |
|
| 139 |
/** |
| 140 |
* Start end point to handle requests on page updates |
| 141 |
* The Otto Crawler will call this end point once a page is updated |
| 142 |
**/ |
| 143 |
|
| 144 |
# function to register the route |
| 145 |
function metasync_otto_crawl_notify($request){ |
| 146 |
|
| 147 |
# get request data params |
| 148 |
$data = $request->get_json_params(); |
| 149 |
|
| 150 |
# fields |
| 151 |
$fields = ['domain', 'urls']; |
| 152 |
|
| 153 |
# validate the json request |
| 154 |
foreach ($fields as $key => $value) { |
| 155 |
|
| 156 |
# if the value is empty stop tehre |
| 157 |
if(empty($data[$value])){ |
| 158 |
|
| 159 |
# Handle the POST request |
| 160 |
return new WP_REST_Response(array( |
| 161 |
'success' => false, |
| 162 |
'message' => 'Invalid Field : '. $value, |
| 163 |
), 400); |
| 164 |
} |
| 165 |
|
| 166 |
} |
| 167 |
|
| 168 |
# load otto pixel |
| 169 |
$otto_pixel = new Metasync_otto_pixel(false); |
| 170 |
|
| 171 |
# save the otto data first (cheap local DB write — kept synchronous) |
| 172 |
$otto_pixel->save_crawl_data($data); |
| 173 |
|
| 174 |
# Defer the expensive per-URL work (OTTO API fetch, post meta sync, |
| 175 |
# per-URL cache purge) to background jobs. The webhook caller enforces a |
| 176 |
# strict response-time budget, so the response acknowledges SCHEDULING, |
| 177 |
# never execution. The per-URL status store records what actually |
| 178 |
# happened to each URL once the jobs run. |
| 179 |
# |
| 180 |
# Cap the number of per-URL cron jobs scheduled per webhook batch so a |
| 181 |
# large crawl cannot overwhelm WP-Cron on shared hosts. URLs past the cap |
| 182 |
# are queued durably and worked off by a self-rescheduling drainer job — |
| 183 |
# never silently dropped. |
| 184 |
$max_jobs_per_batch = defined('METASYNC_MAX_JOBS_PER_BATCH') ? METASYNC_MAX_JOBS_PER_BATCH : 25; |
| 185 |
$now = time(); |
| 186 |
$routes_to_process = array(); |
| 187 |
$accepted = 0; |
| 188 |
$deferred = 0; |
| 189 |
$rejected = 0; |
| 190 |
$scheduled_this_batch = 0; |
| 191 |
|
| 192 |
foreach($data['urls'] AS $key => $url){ |
| 193 |
# prepare the route |
| 194 |
$route = $data['domain'] . $url; |
| 195 |
|
| 196 |
# validate the route |
| 197 |
$route = rtrim($route, '/'); |
| 198 |
|
| 199 |
# Resolve redirect table: use final destination URL before excluded/404 checks and OTTO processing |
| 200 |
$route = metasync_otto_resolve_redirect_to_final_url($route); |
| 201 |
|
| 202 |
# OTTO has confirmed this URL is crawlable. Remove any auto-exclusion that was |
| 203 |
# previously set (e.g. because url_to_postid() returned 0 for a custom post type |
| 204 |
# and metasync_otto_is_url_available() incorrectly treated it as a 404). |
| 205 |
# Manual exclusions (auto_excluded = 0) are left untouched. |
| 206 |
metasync_otto_remove_auto_exclusion($route); |
| 207 |
|
| 208 |
# Skip manually excluded URLs - don't queue OTTO processing for them |
| 209 |
if (metasync_is_otto_url_excluded($route)) { |
| 210 |
continue; |
| 211 |
} |
| 212 |
|
| 213 |
# Every valid URL belongs to the batched cache-purge set even when its |
| 214 |
# per-URL meta-sync job is deduped or deferred: the crawler hit all of |
| 215 |
# them, so the caches for all of them must be refreshed. |
| 216 |
$routes_to_process[] = $route; |
| 217 |
|
| 218 |
# Duplicate delivery: an active job already owns this URL. Idempotent |
| 219 |
# no-op — not re-scheduled, not queued, not counted as new work. |
| 220 |
if (metasync_otto_job_already_scheduled($route)) { |
| 221 |
continue; |
| 222 |
} |
| 223 |
|
| 224 |
if ($scheduled_this_batch >= $max_jobs_per_batch) { |
| 225 |
# Batch overflow: queue durably for the drainer instead of dropping. |
| 226 |
if (Metasync_Otto_Job_Status::enqueue_pending($route)) { |
| 227 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_ACCEPTED, 'queue', 0, 'batch_overflow'); |
| 228 |
$deferred++; |
| 229 |
} else { |
| 230 |
$rejected++; |
| 231 |
} |
| 232 |
continue; |
| 233 |
} |
| 234 |
|
| 235 |
# Queue the per-URL processing for background execution. The handler |
| 236 |
# (metasync_handle_otto_crawl_url_job) performs transient warming, |
| 237 |
# SEO meta sync, and per-URL host cache purge. |
| 238 |
# Offset each event by $key seconds to avoid wp_schedule_single_event() |
| 239 |
# silently dropping duplicates when timestamp + hook + args collide. |
| 240 |
# A false return means cron refused the event — fall back to the |
| 241 |
# durable queue rather than acknowledging a URL nobody will process. |
| 242 |
if (metasync_otto_schedule_single_event($now + $key, 'metasync_process_otto_crawl_url_job', array($route))) { |
| 243 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_ACCEPTED, 'queue', 0, ''); |
| 244 |
$accepted++; |
| 245 |
$scheduled_this_batch++; |
| 246 |
} elseif (Metasync_Otto_Job_Status::enqueue_pending($route)) { |
| 247 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_ACCEPTED, 'queue', 0, 'cron_refused'); |
| 248 |
$deferred++; |
| 249 |
} else { |
| 250 |
$rejected++; |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
# Schedule a single batch job for the edge CDN purge. |
| 255 |
# The batch waits for per-URL stragglers (retries) before purging, so a |
| 256 |
# retried URL's fresh content is what reaches the edge. Multi-URL purge |
| 257 |
# in a single API call keeps edge-provider request counts low. |
| 258 |
if (!empty($routes_to_process)) { |
| 259 |
$batch_time = $now + count($data['urls']) + 5; // run after all per-URL jobs |
| 260 |
metasync_otto_schedule_single_event($batch_time, 'metasync_process_otto_batch_cache_job', array($routes_to_process)); |
| 261 |
} |
| 262 |
|
| 263 |
# Work off any overflow (or cron-refused URLs) from the durable queue. |
| 264 |
if ($deferred > 0) { |
| 265 |
metasync_otto_schedule_single_event($now + 120, 'metasync_otto_pending_drainer', array()); |
| 266 |
} |
| 267 |
|
| 268 |
# Return 200 immediately so the webhook caller does not time out, with an |
| 269 |
# honest accounting of what happened to the batch. |
| 270 |
return new WP_REST_Response(array( |
| 271 |
'success' => true, |
| 272 |
'message' => 'OTTO crawl notification received', |
| 273 |
'accepted' => $accepted, |
| 274 |
'deferred' => $deferred, |
| 275 |
'rejected' => $rejected, |
| 276 |
), 200); |
| 277 |
} |
| 278 |
|
| 279 |
/** |
| 280 |
* Background handler for a single OTTO crawl-notify URL. |
| 281 |
* |
| 282 |
* Pipeline per URL: |
| 283 |
* 1. Warm the OTTO transient cache — this is the ONE live API call the job |
| 284 |
* makes; the warmed suggestions are handed to the meta sync so the |
| 285 |
* endpoint is never fetched twice per URL. |
| 286 |
* 2. Sync SEO post meta (metasync_process_otto_seo_data) and branch on its |
| 287 |
* outcome: only SUCCESS / NO_CHANGE reach the purge step. |
| 288 |
* 3. Clear the per-URL host cache and re-warm it. |
| 289 |
* |
| 290 |
* Retryable outcomes (API timeout / 5xx / 429 / lock contention / CPU deferral) |
| 291 |
* retry with exponential backoff up to METASYNC_OTTO_JOB_MAX_RETRIES; the |
| 292 |
* exhaustion event is recorded via metasync_record_failed_action() for |
| 293 |
* Site Health. Permanent outcomes (not connected, excluded/404 URL, API auth |
| 294 |
* or input error) never retry and never purge. |
| 295 |
* |
| 296 |
* @param string $route Fully-qualified URL to process. |
| 297 |
* @param int $retry_count Current retry attempt (0 = first run). |
| 298 |
*/ |
| 299 |
function metasync_handle_otto_crawl_url_job($route = '', $retry_count = 0) { |
| 300 |
// WP-Cron events persist independently of plugin code. A stale or malformed |
| 301 |
// event may therefore invoke this callback without its required route. |
| 302 |
// Docblock type isn't enforced at runtime; a stale cron record can pass a non-string. |
| 303 |
// @phpstan-ignore-next-line function.alreadyNarrowedType |
| 304 |
if (!is_string($route) || $route === '') { |
| 305 |
error_log('MetaSync OTTO: skipping crawl-url job without a valid route.'); |
| 306 |
return; |
| 307 |
} |
| 308 |
|
| 309 |
$max_retries = defined('METASYNC_OTTO_JOB_MAX_RETRIES') ? METASYNC_OTTO_JOB_MAX_RETRIES : 3; |
| 310 |
|
| 311 |
# Shared retry/terminal bookkeeping so every failure path stays consistent. |
| 312 |
$schedule_retry = function ($reason) use ($route, $retry_count, $max_retries) { |
| 313 |
if ($retry_count < $max_retries) { |
| 314 |
# Exponential backoff (60s, 120s, 240s) plus jitter so a batch of |
| 315 |
# failures cannot stampede the API on the same tick. |
| 316 |
$delay = 60 * pow(2, $retry_count) + wp_rand(0, 15); |
| 317 |
$scheduled = metasync_otto_schedule_single_event(time() + $delay, 'metasync_process_otto_crawl_url_job', array($route, $retry_count + 1)); |
| 318 |
if (!$scheduled && Metasync_Otto_Job_Status::enqueue_pending($route)) { |
| 319 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_ACCEPTED, 'queue', $retry_count + 1, $reason . '/cron_refused'); |
| 320 |
} else { |
| 321 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_RETRYING, 'job', $retry_count + 1, $reason); |
| 322 |
} |
| 323 |
error_log('MetaSync OTTO: retrying crawl-url job for ' . $route . ' (attempt ' . ($retry_count + 1) . '/' . $max_retries . ') reason=' . $reason . ' in ' . $delay . 's'); |
| 324 |
} else { |
| 325 |
metasync_record_failed_action('metasync_process_otto_crawl_url_job'); |
| 326 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_FAILED, 'job', $retry_count, 'exhausted:' . $reason); |
| 327 |
error_log('MetaSync OTTO: background crawl-url job permanently failed for ' . $route . ' after ' . $max_retries . ' retries, last reason=' . $reason); |
| 328 |
} |
| 329 |
}; |
| 330 |
|
| 331 |
try { |
| 332 |
# Not connected: retrying cannot help, and there is nothing to purge. |
| 333 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 334 |
# A corrupted option row can hold a non-string UUID; the docblock |
| 335 |
# contract alone does not enforce that at runtime. |
| 336 |
# @phpstan-ignore-next-line function.alreadyNarrowedType |
| 337 |
if (!is_string($otto_uuid) || $otto_uuid === '') { |
| 338 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_FAILED, 'job', $retry_count, 'not_connected'); |
| 339 |
return; |
| 340 |
} |
| 341 |
|
| 342 |
# Step 1: Warm OTTO transient cache (fetch fresh suggestions from OTTO |
| 343 |
# API into WP transient — the rate limiter, circuit breaker, and |
| 344 |
# request lock all live inside this call). |
| 345 |
$transient_cache = new Metasync_Otto_Transient_Cache($otto_uuid); |
| 346 |
$warmed = $transient_cache->warm_cache($route); |
| 347 |
if ($warmed === false) { |
| 348 |
# A 4xx refusal (bad key, revoked project) is permanent: retrying |
| 349 |
# cannot change the answer, so fail fast instead of burning the |
| 350 |
# attempt budget on warm calls that will keep being rejected. |
| 351 |
if ($transient_cache->last_failure_is_permanent()) { |
| 352 |
metasync_record_failed_action('metasync_process_otto_crawl_url_job'); |
| 353 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_FAILED, 'job', $retry_count, 'permanent:warm_rejected'); |
| 354 |
error_log('MetaSync OTTO: crawl-url job failed permanently for ' . $route . ' (warm request rejected by the API)'); |
| 355 |
return; |
| 356 |
} |
| 357 |
# Transient API failure (timeout / 5xx / breaker open). No meta |
| 358 |
# write, no purge — straight to the retry path. |
| 359 |
$schedule_retry('warm_failed'); |
| 360 |
return; |
| 361 |
} |
| 362 |
|
| 363 |
# Step 2: Write SEO post meta synchronously BEFORE the cache purge. |
| 364 |
# This ensures the DB is fully up-to-date when the host re-populates |
| 365 |
# the cache on the very next request. The warmed suggestions are |
| 366 |
# passed through so the sync makes NO second API call. |
| 367 |
# allow_defer=false: the crawl_url_job retry mechanism owns failure |
| 368 |
# handling here; letting the sync self-reschedule caused an |
| 369 |
# unbounded cron pile-up. |
| 370 |
# warm_cache() may hand back a non-array truthy on contract drift; |
| 371 |
# the sync treats that as "no prefetched suggestions". |
| 372 |
# @phpstan-ignore-next-line function.alreadyNarrowedType |
| 373 |
$prefetched = is_array($warmed) ? $warmed : null; |
| 374 |
$outcome = metasync_process_otto_seo_data($route, false, 0, $prefetched); |
| 375 |
|
| 376 |
if ($outcome === Metasync_Otto_Job_Status::OUTCOME_RETRYABLE |
| 377 |
|| $outcome === Metasync_Otto_Job_Status::OUTCOME_DEFERRED) { |
| 378 |
$schedule_retry('outcome_' . $outcome); |
| 379 |
return; |
| 380 |
} |
| 381 |
|
| 382 |
if ($outcome !== Metasync_Otto_Job_Status::OUTCOME_SUCCESS |
| 383 |
&& $outcome !== Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE) { |
| 384 |
# Permanent failure: no retry, no purge — caches must keep serving |
| 385 |
# the last good content. |
| 386 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_FAILED, 'meta', $retry_count, 'permanent:' . $outcome); |
| 387 |
return; |
| 388 |
} |
| 389 |
|
| 390 |
# Step 3: meta is current (changed or confirmed unchanged) — clear the |
| 391 |
# per-URL host cache entry and re-warm it so OTTO-modified output is |
| 392 |
# what gets stored. |
| 393 |
$otto_pixel = new Metasync_otto_pixel(false); |
| 394 |
$otto_pixel->refresh_cache($route); |
| 395 |
Metasync_Cache_Purge::purge_single_url($route); |
| 396 |
Metasync_Cache_Purge::warm_urls(array($route)); |
| 397 |
|
| 398 |
# A URL that completed on a retry missed its batch's edge purge (the |
| 399 |
# batch only waits so long). Late completers purge the edge themselves. |
| 400 |
if ($retry_count > 0) { |
| 401 |
Metasync_Edge_Cache_Purge::purge(array($route)); |
| 402 |
} |
| 403 |
|
| 404 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_COMPLETED, 'job', $retry_count, $outcome); |
| 405 |
|
| 406 |
} catch (Exception $e) { |
| 407 |
$schedule_retry('exception:' . $e->getMessage()); |
| 408 |
} |
| 409 |
} |
| 410 |
|
| 411 |
/** |
| 412 |
* Batch handler for the edge CDN purge. |
| 413 |
* |
| 414 |
* Runs after all per-URL jobs have had time to complete. Per-URL host caches |
| 415 |
* are cleared and re-warmed by the individual jobs; this batched event only |
| 416 |
* handles the edge CDN purge, which benefits from multi-URL API calls |
| 417 |
* (Cloudflare, Fastly, Akamai, Sucuri, Sevella, etc.). |
| 418 |
* |
| 419 |
* URLs still retrying are stragglers: the batch waits (up to |
| 420 |
* METASYNC_OTTO_BATCH_MAX_WAITS reschedules 60s apart) so their freshly |
| 421 |
* synced content — not stale content — reaches the edge. URLs that complete |
| 422 |
* after the wait budget purge the edge themselves from the retry path. |
| 423 |
* |
| 424 |
* @param array $routes List of fully-qualified URLs in this batch. |
| 425 |
* @param int $wait_count How many times this batch has already waited. |
| 426 |
*/ |
| 427 |
function metasync_handle_otto_batch_cache_job($routes = array(), $wait_count = 0) { |
| 428 |
// WP-Cron events persist independently of plugin code. A stale or malformed |
| 429 |
// event may therefore invoke this callback without its required route list. |
| 430 |
// Docblock type isn't enforced at runtime; a stale cron record can pass a non-array. |
| 431 |
// @phpstan-ignore-next-line function.alreadyNarrowedType |
| 432 |
if (!is_array($routes) || empty($routes)) { |
| 433 |
error_log('MetaSync OTTO: skipping batch cache job without valid routes.'); |
| 434 |
return; |
| 435 |
} |
| 436 |
|
| 437 |
$max_waits = defined('METASYNC_OTTO_BATCH_MAX_WAITS') ? METASYNC_OTTO_BATCH_MAX_WAITS : 3; |
| 438 |
// Docblock type isn't enforced at runtime; a stale cron record can pass a non-numeric. |
| 439 |
// @phpstan-ignore-next-line function.alreadyNarrowedType |
| 440 |
$wait_count = is_numeric($wait_count) ? (int) $wait_count : 0; |
| 441 |
|
| 442 |
# Split the batch: terminally-done URLs vs. URLs with work still in |
| 443 |
# flight. Unknown-state URLs (legacy events scheduled before the status |
| 444 |
# store existed, or deduped duplicates whose entry expired) count as done |
| 445 |
# — their per-URL job has either finished or never will, and both mean |
| 446 |
# the edge purge should not stall behind them. |
| 447 |
$ready = array(); |
| 448 |
$stragglers = 0; |
| 449 |
foreach ($routes as $route) { |
| 450 |
$entry = Metasync_Otto_Job_Status::get($route); |
| 451 |
$state = is_array($entry) && isset($entry['state']) ? $entry['state'] : null; |
| 452 |
if ($state === Metasync_Otto_Job_Status::STATE_RETRYING |
| 453 |
|| $state === Metasync_Otto_Job_Status::STATE_ACCEPTED) { |
| 454 |
$stragglers++; |
| 455 |
} else { |
| 456 |
# completed / failed / unknown: nothing left to wait for. |
| 457 |
$ready[] = $route; |
| 458 |
} |
| 459 |
} |
| 460 |
|
| 461 |
# Still live work in the batch and wait budget left: hold the edge purge. |
| 462 |
if ($stragglers > 0 && $wait_count < $max_waits) { |
| 463 |
metasync_otto_schedule_single_event(time() + 60, 'metasync_process_otto_batch_cache_job', array($routes, $wait_count + 1)); |
| 464 |
return; |
| 465 |
} |
| 466 |
|
| 467 |
if (empty($ready)) { |
| 468 |
return; |
| 469 |
} |
| 470 |
|
| 471 |
try { |
| 472 |
# Purge edge CDN caches (Cloudflare, Fastly, Akamai, Sucuri, Sevalla, etc.) |
| 473 |
# Tag-based providers purge only the affected posts; full-flush providers fire once per batch. |
| 474 |
Metasync_Edge_Cache_Purge::purge($ready); |
| 475 |
} catch (Exception $e) { |
| 476 |
metasync_record_failed_action('metasync_process_otto_batch_cache_job'); |
| 477 |
error_log('MetaSync OTTO: batch cache job failed for ' . count($ready) . ' URLs: ' . $e->getMessage() . ' in ' . $e->getFile() . ':' . $e->getLine()); |
| 478 |
} |
| 479 |
} |
| 480 |
|
| 481 |
/** |
| 482 |
* Drainer: work off URLs that could not be scheduled inline (batch overflow |
| 483 |
* or a cron that refused the event) from the durable pending queue. |
| 484 |
* |
| 485 |
* Processes at most 25 URLs per run to stay friendly to WP-Cron, schedules |
| 486 |
* the same batched edge purge as the webhook path for the chunk it handled, |
| 487 |
* and reschedules itself while the queue is non-empty. URLs it takes off |
| 488 |
* the queue are scheduled unconditionally unless a real cron event already |
| 489 |
* exists for them — the webhook's overflow bookkeeping must never veto, |
| 490 |
* because by then the URL has already left the durable queue. |
| 491 |
*/ |
| 492 |
function metasync_handle_otto_pending_drainer() { |
| 493 |
$chunk = Metasync_Otto_Job_Status::take_pending(25); |
| 494 |
if (empty($chunk)) { |
| 495 |
return; |
| 496 |
} |
| 497 |
|
| 498 |
$now = time(); |
| 499 |
$scheduled = 0; |
| 500 |
foreach ($chunk as $key => $route) { |
| 501 |
# This run owns the chunk: take_pending() already removed these URLs |
| 502 |
# from the durable queue, so skipping here would drop them. Only a |
| 503 |
# real cron event vetoes — the webhook's overflow ledger row must |
| 504 |
# not (see metasync_otto_job_event_pending). |
| 505 |
if (metasync_otto_job_event_pending($route)) { |
| 506 |
continue; |
| 507 |
} |
| 508 |
if (metasync_otto_schedule_single_event($now + $key, 'metasync_process_otto_crawl_url_job', array($route))) { |
| 509 |
Metasync_Otto_Job_Status::record($route, Metasync_Otto_Job_Status::STATE_ACCEPTED, 'drain', 0, ''); |
| 510 |
$scheduled++; |
| 511 |
} else { |
| 512 |
# Cron still refusing: put it back at the end of the queue. |
| 513 |
Metasync_Otto_Job_Status::enqueue_pending($route); |
| 514 |
} |
| 515 |
} |
| 516 |
|
| 517 |
if ($scheduled > 0) { |
| 518 |
metasync_otto_schedule_single_event($now + count($chunk) + 5, 'metasync_process_otto_batch_cache_job', array($chunk)); |
| 519 |
} |
| 520 |
|
| 521 |
if (Metasync_Otto_Job_Status::pending_count() > 0) { |
| 522 |
metasync_otto_schedule_single_event($now + 60, 'metasync_otto_pending_drainer', array()); |
| 523 |
} |
| 524 |
} |
| 525 |
|
| 526 |
# NOTE: Cache system removed - these functions are no longer needed |
| 527 |
# Kept for backward compatibility in case old cache directories need cleanup |
| 528 |
function metasync_deleteDir($dir) { |
| 529 |
if (!is_dir($dir)) { |
| 530 |
return false; |
| 531 |
} |
| 532 |
$files = array_diff(scandir($dir), array('.', '..')); |
| 533 |
foreach ($files as $file) { |
| 534 |
$filePath = $dir . DIRECTORY_SEPARATOR . $file; |
| 535 |
if (is_dir($filePath)) { |
| 536 |
metasync_deleteDir($filePath); |
| 537 |
} else { |
| 538 |
unlink($filePath); |
| 539 |
} |
| 540 |
} |
| 541 |
return rmdir($dir); |
| 542 |
} |
| 543 |
|
| 544 |
# Cleanup function for removing old cache directories (if they exist) |
| 545 |
function metasync_invalidate_all_caches($folder = ''){ |
| 546 |
# Cache system removed - this function only exists to clean up old cache directories |
| 547 |
if(!defined('WP_CONTENT_DIR')){ |
| 548 |
return false; |
| 549 |
} |
| 550 |
$wp_content_dir = WP_CONTENT_DIR; |
| 551 |
$cache_dir = $wp_content_dir . '/metasync_caches'; |
| 552 |
if(in_array($folder, ['posts', 'pages'])){ |
| 553 |
$cache_dir = $cache_dir . '/' . $folder; |
| 554 |
} |
| 555 |
if(is_dir($cache_dir)){ |
| 556 |
metasync_deleteDir($cache_dir); |
| 557 |
} |
| 558 |
} |
| 559 |
|
| 560 |
// metasync_is_custom_or_lps_page() now lives in includes/metasync-helpers.php |
| 561 |
// (loaded unconditionally before this file) so all SEO surfaces share one rule. |
| 562 |
|
| 563 |
/** |
| 564 |
* Detect an Elementor editor / preview request that OTTO must never process. |
| 565 |
* |
| 566 |
* Elementor renders its editing canvas in a front-end <iframe> — a logged-in |
| 567 |
* GET request that is NOT is_admin(), so the admin/ajax/REST guards in |
| 568 |
* metasync_start_otto() do not catch it. If OTTO output-buffers and rewrites |
| 569 |
* that response (SimpleHtmlDom in Otto_html_class), the markup the editor's |
| 570 |
* JavaScript expects is altered and the canvas stays stuck on "loading" — |
| 571 |
* the editor never finishes initialising. The editor surface is for authoring |
| 572 |
* only; it is never crawled, so there is nothing for OTTO to optimise there. |
| 573 |
* |
| 574 |
* Locally this is masked because the OTTO API is unreachable (API_ERROR ⇒ no |
| 575 |
* suggestions ⇒ no rewrite); on a live site with real suggestions the rewrite |
| 576 |
* happens and the editor breaks. See the "Elementor editor fails to load while |
| 577 |
* the plugin is active" report. |
| 578 |
* |
| 579 |
* @return bool True when the current request is an Elementor editor/preview. |
| 580 |
*/ |
| 581 |
function metasync_is_elementor_editor_request() { |
| 582 |
# The preview iframe loads the front end with ?elementor-preview=POST_ID. |
| 583 |
# This is the request OTTO would otherwise buffer and corrupt. |
| 584 |
if (isset($_GET['elementor-preview'])) { |
| 585 |
return true; |
| 586 |
} |
| 587 |
|
| 588 |
# Elementor editor / app entry points carried as an action on the front end. |
| 589 |
if (isset($_REQUEST['action']) |
| 590 |
&& in_array($_REQUEST['action'], array('elementor', 'elementor_ajax'), true) |
| 591 |
) { |
| 592 |
return true; |
| 593 |
} |
| 594 |
|
| 595 |
# Authoritative check when Elementor is loaded: its own preview-mode flag. |
| 596 |
if (class_exists('\Elementor\Plugin') |
| 597 |
&& isset(\Elementor\Plugin::$instance->preview) |
| 598 |
&& is_object(\Elementor\Plugin::$instance->preview) |
| 599 |
&& method_exists(\Elementor\Plugin::$instance->preview, 'is_preview_mode') |
| 600 |
&& \Elementor\Plugin::$instance->preview->is_preview_mode() |
| 601 |
) { |
| 602 |
return true; |
| 603 |
} |
| 604 |
|
| 605 |
return false; |
| 606 |
} |
| 607 |
|
| 608 |
function metasync_start_otto(){ |
| 609 |
|
| 610 |
# PERFORMANCE FIX: Cache is now enabled for speed |
| 611 |
# Skip initial cache cleanup to preserve existing cache |
| 612 |
# Cache files are valuable for performance - only clear on OTTO updates |
| 613 |
# Periodic cache clearing can be configured in plugin settings if needed |
| 614 |
|
| 615 |
# exclude AJAX requests and WooCommerce transactional pages from OTTO SSR |
| 616 |
# SSR is now ENABLED for: single products, product categories, product tags |
| 617 |
# SSR is SKIPPED for: cart, checkout |
| 618 |
# Note: title/description filters (pre_get_document_title, wp_head meta desc) |
| 619 |
# run on ALL pages regardless — they are hooked unconditionally in seo-functions.php |
| 620 |
|
| 621 |
# ── WooCommerce-independent cart/checkout protection ────────── |
| 622 |
# The is_cart()/is_checkout() guards below only recognize WooCommerce. Carts |
| 623 |
# from other systems — e.g. the Point of Rental "Catalog" plugin on |
| 624 |
# venturarental.com — are invisible to them, so OTTO would process those |
| 625 |
# pages and let caching layers store them. A cached cart page corrupts the |
| 626 |
# live cart (added items vanish because a stale page is served). The checks |
| 627 |
# here do not depend on WooCommerce being active. |
| 628 |
|
| 629 |
# Never run OTTO on non-GET requests. SSR exists for crawlers, which only |
| 630 |
# issue GET; POST/PUT/etc. are form or cart submissions that must pass |
| 631 |
# through untouched. (OTTO's own internal fetches use GET.) |
| 632 |
if (($_SERVER['REQUEST_METHOD'] ?? 'GET') !== 'GET') { |
| 633 |
return; |
| 634 |
} |
| 635 |
|
| 636 |
# Never run OTTO on WordPress core draft/pending previews |
| 637 |
# (?p=<ID>&preview=true, with preview_id/preview_nonce on the full link). |
| 638 |
# Previews are per-user, non-cacheable, and routed purely by query string — |
| 639 |
# and OTTO's route identity ignores the query string, so a processed preview |
| 640 |
# is answered with the route's public page (the homepage for /?p=…) instead |
| 641 |
# of the draft the editor asked for. Skip early and let WP render it. |
| 642 |
if ( |
| 643 |
(isset($_GET['preview']) && $_GET['preview'] === 'true') || |
| 644 |
!empty($_GET['preview_id']) |
| 645 |
) { |
| 646 |
if (!defined('DONOTCACHEPAGE')) { |
| 647 |
define('DONOTCACHEPAGE', true); |
| 648 |
} |
| 649 |
return; |
| 650 |
} |
| 651 |
|
| 652 |
# Never run OTTO on the Elementor editor/preview iframe. It is a logged-in |
| 653 |
# front-end GET (not is_admin()), so it slips past the admin guard below; |
| 654 |
# buffering and rewriting it breaks the editor canvas ("fails to load"). |
| 655 |
if (metasync_is_elementor_editor_request()) { |
| 656 |
return; |
| 657 |
} |
| 658 |
|
| 659 |
# WP core file-editor self-check. When an admin saves a THEME file |
| 660 |
# via the Plugin/Theme Editor, wp_edit_theme_plugin_file() fires an internal |
| 661 |
# loopback GET to home_url('/') carrying wp_scrape_key/wp_scrape_nonce to |
| 662 |
# detect a white-screen. That request is NOT is_admin(), so — exactly like |
| 663 |
# the Elementor case above — it slips past the admin/AJAX/REST guards below. |
| 664 |
# OTTO must not output-buffer or rewrite it: the SimpleHtmlDom pass (and any |
| 665 |
# fatal thrown inside the ob_start() callback, e.g. under memory pressure) |
| 666 |
# corrupts the scrape and surfaces as the misleading |
| 667 |
# "preg_match(): Cannot use output buffering in output buffering display |
| 668 |
# handlers" fatal. Skipping OTTO lets WP render the page normally so the |
| 669 |
# scrape works. Returning here also avoids the HTTP-fallback path firing a |
| 670 |
# second loopback for an already-internal self-request. |
| 671 |
if (metasync_is_scrape_request()) { |
| 672 |
return; |
| 673 |
} |
| 674 |
|
| 675 |
# Skip OTTO on cart/checkout-type paths regardless of the cart plugin, and |
| 676 |
# signal caching plugins (WP Rocket/Kinsta/etc.) to never cache them — a |
| 677 |
# cached cart/checkout page is precisely what corrupts cart state. |
| 678 |
# Matched against the FIRST path segment (relative to the WP home path) so |
| 679 |
# subdirectory installs (/shop/cart) still work, while unrelated content pages |
| 680 |
# like /blog/cart or /features/checkout do NOT false-match. |
| 681 |
# Drop the query string with explode() (no shared internal pointer like strtok), |
| 682 |
# then lower-case and strip surrounding slashes. |
| 683 |
$otto_req_path = explode('?', (string) wp_unslash($_SERVER['REQUEST_URI'] ?? ''), 2)[0]; |
| 684 |
$otto_req_path = strtolower(trim($otto_req_path, '/')); |
| 685 |
# Strip the site's base path so first-segment matching works on subdir installs. |
| 686 |
$otto_home_path = trim((string) parse_url(home_url(), PHP_URL_PATH), '/'); |
| 687 |
if ($otto_home_path !== '' && strpos($otto_req_path, $otto_home_path . '/') === 0) { |
| 688 |
$otto_req_path = substr($otto_req_path, strlen($otto_home_path) + 1); |
| 689 |
} |
| 690 |
$otto_first_segment = explode('/', $otto_req_path, 2)[0]; |
| 691 |
# Filterable so sites can add/remove transactional slugs without code changes. |
| 692 |
$otto_cart_segments = apply_filters('metasync_otto_cart_paths', array( |
| 693 |
'cart', 'checkout', 'basket', 'request-a-quote', 'quote-request', |
| 694 |
)); |
| 695 |
$otto_cart_segments = array_map('strtolower', (array) $otto_cart_segments); |
| 696 |
if ($otto_first_segment !== '' && in_array($otto_first_segment, $otto_cart_segments, true)) { |
| 697 |
if (!defined('DONOTCACHEPAGE')) { |
| 698 |
define('DONOTCACHEPAGE', true); |
| 699 |
} |
| 700 |
return; |
| 701 |
} |
| 702 |
# ────────────────────────────────────────────────────────────────────── |
| 703 |
|
| 704 |
if ( |
| 705 |
# disable ajax calls |
| 706 |
isset($_GET['ucfrontajaxaction']) || |
| 707 |
# OTTO Preview mode - skip OTTO when previewing original content |
| 708 |
(isset($_GET['otto_preview']) && $_GET['otto_preview'] === '1') || |
| 709 |
# WooCommerce shop archive page only (products and categories now use SSR) |
| 710 |
//(function_exists('is_shop') && is_shop()) || |
| 711 |
# Cart page |
| 712 |
(function_exists('is_cart') && is_cart()) || |
| 713 |
# Checkout page |
| 714 |
(function_exists('is_checkout') && is_checkout()) || |
| 715 |
# My Account page |
| 716 |
//(function_exists('is_account_page') && is_account_page()) || |
| 717 |
# Standard WordPress AJAX |
| 718 |
(function_exists('wp_doing_ajax') && wp_doing_ajax()) || |
| 719 |
# check by constant |
| 720 |
(defined('DOING_AJAX') && DOING_AJAX) || |
| 721 |
# WooCommerce AJAX endpoint (e.g., ?wc-ajax=update_cart) |
| 722 |
(isset($_REQUEST['wc-ajax']) && !empty($_REQUEST['wc-ajax'])) || |
| 723 |
# AJAX requests via X-Requested-With header |
| 724 |
(isset($_SERVER['HTTP_X_REQUESTED_WITH']) && strtolower($_SERVER['HTTP_X_REQUESTED_WITH']) === 'xmlhttprequest') || |
| 725 |
# Gravity Forms submission detection - skip OTTO to allow form processing |
| 726 |
(isset($_POST['gform_submit']) && ( |
| 727 |
is_array($_POST['gform_submit']) || |
| 728 |
(is_string($_POST['gform_submit']) && isset($_POST['is_submit_' . $_POST['gform_submit']]) && !empty($_POST['gform_submit'])) |
| 729 |
)) || |
| 730 |
# Gravity Forms AJAX submission |
| 731 |
(isset($_POST['gform_ajax']) && isset($_POST['gform_submit']) && ( |
| 732 |
is_array($_POST['gform_submit']) || !empty($_POST['gform_submit']) |
| 733 |
)) || |
| 734 |
# Gravity Forms file upload |
| 735 |
(isset($_POST['gform_uploaded_files'])) || |
| 736 |
# Any Gravity Forms POST parameter |
| 737 |
(isset($_POST['gform_submit']) || isset($_POST['gform_unique_id']) || isset($_POST['gform_field_values'])) || |
| 738 |
# Formidable Forms AJAX submission detection - skip OTTO to allow form processing |
| 739 |
(isset($_POST['action']) && $_POST['action'] === 'frm_entries_create') || |
| 740 |
# Formidable Forms POST parameters |
| 741 |
(isset($_POST['form_id']) && !empty($_POST['form_id'])) || |
| 742 |
# Formidable Forms action parameter |
| 743 |
(isset($_POST['frm_action']) && !empty($_POST['frm_action'])) || |
| 744 |
# Formidable Forms item_key (used in form submissions) |
| 745 |
(isset($_POST['item_key']) && !empty($_POST['item_key'])) |
| 746 |
) { |
| 747 |
return; |
| 748 |
} |
| 749 |
|
| 750 |
# fetch globals |
| 751 |
global $metasync_options, $otto_enabled; |
| 752 |
|
| 753 |
# OPTIMIZED: check for the disable otto for logged in users option using cached config |
| 754 |
if(Metasync_Otto_Config::is_disabled_for_loggedin()){ |
| 755 |
|
| 756 |
# get user |
| 757 |
$current_user = wp_get_current_user(); |
| 758 |
|
| 759 |
# check if user is logged in |
| 760 |
if( !empty($current_user->ID)){ |
| 761 |
|
| 762 |
return; |
| 763 |
} |
| 764 |
} |
| 765 |
|
| 766 |
# check if current URL is manually excluded from OTTO |
| 767 |
# NOTE: auto-exclusions (false-positive 404 detections) are intentionally NOT checked |
| 768 |
# here — they must not block OTTO rendering. Use metasync_is_otto_url_excluded() only |
| 769 |
# in the webhook handler where we gate SEO meta writes. |
| 770 |
$request_uri = sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'] ?? '')); |
| 771 |
$current_url = home_url(strtok($request_uri, '?') ?: $request_uri); |
| 772 |
if (metasync_is_otto_url_manually_excluded($current_url)) { |
| 773 |
return; |
| 774 |
} |
| 775 |
|
| 776 |
# Skip OTTO for XML endpoints (sitemaps, RSS-as-xml, etc.). OTTO has no |
| 777 |
# suggestions for machine-readable XML, and the upstream API returns |
| 778 |
# API_ERROR for these URLs — which then stamps misleading |
| 779 |
# X-MetaSync-OTTO-Cache: API_ERROR / X-MetaSync-OTTO-Method: NONE headers |
| 780 |
# onto otherwise-healthy sitemap responses and causes false-alarm bug |
| 781 |
# reports from customers. |
| 782 |
$request_path = strtok($request_uri, '?'); |
| 783 |
if ($request_path && preg_match('#\.xml$#i', $request_path)) { |
| 784 |
return; |
| 785 |
} |
| 786 |
|
| 787 |
# Machine-readable core endpoints. Using WordPress's own conditionals rather |
| 788 |
# than matching the path: they are exact, and a path pattern here would also |
| 789 |
# match ordinary content whose slug merely starts with the same characters |
| 790 |
# (e.g. /robots.txt-explained/, /favicon.ico-vs-svg/). |
| 791 |
# is_favicon() is WP 5.4+; this plugin supports 5.2. |
| 792 |
if (is_robots() || (function_exists('is_favicon') && is_favicon())) { |
| 793 |
return; |
| 794 |
} |
| 795 |
|
| 796 |
# Protocol paths and static assets that are never OTTO-able content. |
| 797 |
# The guard fails open — if the strategy is unavailable we let the request |
| 798 |
# proceed rather than fatal. metasync_otto_class_provides() rather than a |
| 799 |
# bare class_exists(): a partially updated install can leave an older copy of |
| 800 |
# the strategy class loaded that predates is_non_content_path(), which |
| 801 |
# satisfies class_exists() and then fatals on every front-end request. |
| 802 |
if (metasync_otto_class_provides('Metasync_Otto_Render_Strategy', 'is_non_content_path') |
| 803 |
&& Metasync_Otto_Render_Strategy::is_non_content_path($request_path)) { |
| 804 |
return; |
| 805 |
} |
| 806 |
|
| 807 |
# BOT DETECTION: Always detect bots so crawl data reaches the SA backend. |
| 808 |
# Real bots don't execute JS, so otto-tracker.js never fires for them. |
| 809 |
# push_crawl_log_to_sa() fires a non-blocking wp_remote_post here instead. |
| 810 |
$bot_detector = Metasync_Otto_Bot_Detector::get_instance(); |
| 811 |
$detection = $bot_detector->detect(); |
| 812 |
if ( $detection['is_bot'] ) { |
| 813 |
$bot_detector->push_crawl_log_to_sa( $detection, $current_url ); |
| 814 |
} |
| 815 |
|
| 816 |
# Throttle OTTO rendering for bots that are not exempt: at most one render |
| 817 |
# per URL+bot every 5 minutes. Humans, search engines (Googlebot, Bingbot, |
| 818 |
# Yahoo's Slurp, Sogou, Exabot, Applebot, ...), host cache warmers and |
| 819 |
# synthetic performance auditors are never throttled — their hits always get |
| 820 |
# full OTTO content, so indexing stays current, the page cache can actually |
| 821 |
# be filled, and audit tools measure what a real visitor sees. SEO tools, |
| 822 |
# social crawlers, archivers, AI scrapers, uptime monitors and unrecognized |
| 823 |
# crawlers are de-duplicated so repeat hits don't re-trigger the expensive |
| 824 |
# DOM-rewriting path. |
| 825 |
# |
| 826 |
# Two complementary gates, deliberately kept separate: |
| 827 |
# - is_unthrottled_category() consults the classifier's explicit per-bot |
| 828 |
# category map, so the exempt set is defined in exactly one place. |
| 829 |
# - metasync_otto_is_unthrottled_infrastructure_agent() matches the raw |
| 830 |
# user-agent (plus a query marker and a site filter), which catches |
| 831 |
# preload agents that have no entry in the map at all. |
| 832 |
if ( $detection['is_bot'] |
| 833 |
&& ! Metasync_Otto_Bot_Detector::is_unthrottled_category( $detection['bot_type'] ) |
| 834 |
&& ! metasync_otto_is_unthrottled_infrastructure_agent( $detection ) ) { |
| 835 |
$normalized_url = strtok( $current_url, '?' ); |
| 836 |
if ( $normalized_url === false ) { |
| 837 |
$normalized_url = $current_url; |
| 838 |
} |
| 839 |
// Lowercase + length cap so attacker-controlled bot_name variations |
| 840 |
// (EvilBota, EvilBotB, EvilBotc, ...) collapse to a bounded key space |
| 841 |
// and cannot flood wp_options with unique transient rows. |
| 842 |
$bot_name = substr( strtolower( $detection['bot_name'] ?? 'unknown' ), 0, 32 ); |
| 843 |
$render_throttle_key = 'metasync_otto_rendered_' . md5( $normalized_url . '|' . $bot_name ); |
| 844 |
if ( get_transient( $render_throttle_key ) ) { |
| 845 |
// Prevent page caches (WP Rocket, W3TC, etc.) from saving the |
| 846 |
// un-OTTO'd response and serving it to subsequent human visitors. |
| 847 |
if ( ! defined( 'DONOTCACHEPAGE' ) ) { |
| 848 |
define( 'DONOTCACHEPAGE', true ); |
| 849 |
} |
| 850 |
return; |
| 851 |
} |
| 852 |
set_transient( $render_throttle_key, 1, 5 * MINUTE_IN_SECONDS ); |
| 853 |
} |
| 854 |
|
| 855 |
# Skip OTTO on 404s. |
| 856 |
# |
| 857 |
# OTTO suggestions are keyed to real, crawled pages. Asking the API about a URL |
| 858 |
# that returns 404 spends a full API_TIMEOUT of PHP-FPM worker time on a |
| 859 |
# guaranteed-empty answer. Not theoretical: in one production sample, 21 of 54 |
| 860 |
# crawler requests were 404s on URLs that never existed on the site, each |
| 861 |
# costing ~2s of a worker. |
| 862 |
# |
| 863 |
# Deliberately placed AFTER bot detection so push_crawl_log_to_sa() still |
| 864 |
# reports the crawl to SearchAtlas — that telemetry is how the crawler side |
| 865 |
# can learn to stop requesting these URLs. Only the expensive suggestions |
| 866 |
# lookup is skipped. |
| 867 |
# |
| 868 |
# Filterable because "should a 404 page get OTTO treatment" is arguably a site |
| 869 |
# decision, even though there is no known case where it should. |
| 870 |
if (is_404() && apply_filters('metasync_otto_skip_on_404', true)) { |
| 871 |
return; |
| 872 |
} |
| 873 |
|
| 874 |
# Optionally skip OTTO processing for bot traffic (when the setting is enabled) |
| 875 |
if ($bot_detector->should_skip_otto()) { |
| 876 |
// Log the bot locally and count the saved API call |
| 877 |
$bot_detector->log_detection($detection); |
| 878 |
$bot_stats_db = Metasync_Otto_Bot_Statistics_Database::get_instance(); |
| 879 |
$bot_stats_db->increment_api_calls_saved(); |
| 880 |
|
| 881 |
// Skip OTTO processing for this bot |
| 882 |
return; |
| 883 |
} |
| 884 |
|
| 885 |
# OPTIMIZED: Check if Otto should be disabled for WP Rocket compatibility |
| 886 |
if (class_exists('WP_Rocket')) { |
| 887 |
$wp_rocket_compat_mode = Metasync_Otto_Config::get_wp_rocket_compat_mode(); |
| 888 |
|
| 889 |
if ($wp_rocket_compat_mode === 'disable_otto') { |
| 890 |
return; # Exit early, Otto is disabled when WP Rocket is active |
| 891 |
} |
| 892 |
} |
| 893 |
|
| 894 |
# Skip OTTO for Divi AJAX pagination and paginated archive requests. |
| 895 |
# ?et_blog = Divi AJAX pagination callback |
| 896 |
# /page/N/ = paginated blog/archive pages — OTTO's buffer/HTTP render causes |
| 897 |
# module numbering mismatch between page 1 (with TB template) and page N |
| 898 |
# (without TB template), breaking Divi's JS pagination selector matching. |
| 899 |
if (isset($_GET['et_blog']) || (is_paged() && !is_singular())) { |
| 900 |
return; |
| 901 |
} |
| 902 |
|
| 903 |
# Skip OTTO on search-results pages. A search request carries its |
| 904 |
# meaning entirely in the query string (e.g. /?s=term, or FiboSearch's |
| 905 |
# /?s=term&post_type=product&dgwt_wcas=1). get_route() intentionally strips |
| 906 |
# the query string to build a canonical, cache-key-stable route — which |
| 907 |
# collapses a query-only search URL to the site root ('/'). OTTO then finds |
| 908 |
# the HOME page's suggestions for that route and, on the HTTP render path, |
| 909 |
# serves the home page in place of the search results (the address bar keeps |
| 910 |
# the search URL — it is a silent content swap, not a redirect). This only |
| 911 |
# surfaces for logged-in / uncached visitors, since page caches serve the |
| 912 |
# correct pre-rendered search page to everyone else. Search-results pages |
| 913 |
# have no per-canonical-URL OTTO suggestions of their own, so OTTO must never |
| 914 |
# run here. The raw $_GET['s'] check is a defensive fallback for setups where |
| 915 |
# the main query is altered before this point. |
| 916 |
if (is_search() || !empty($_GET['s'])) { |
| 917 |
return; |
| 918 |
} |
| 919 |
|
| 920 |
# Handle cache plugin compatibility early - before any caching happens |
| 921 |
metasync_otto_handle_cache_compatibility(); |
| 922 |
|
| 923 |
# check if OTTO is disabled for this specific page/post |
| 924 |
# Gate on is_singular() for the same reason the custom/LPS check below does: |
| 925 |
# on an archive, search or term query get_the_ID() returns the first post of |
| 926 |
# the loop — WP::register_globals() seeds $GLOBALS['post'] with it — not the |
| 927 |
# page being rendered. Without the gate, one excluded or per-page-disabled |
| 928 |
# post silently takes OTTO off every archive that happens to list it first. |
| 929 |
$post_id = is_singular() ? ( get_queried_object_id() ?: get_the_ID() ) : 0; |
| 930 |
if ($post_id && class_exists('Metasync_Otto_Frontend_Toolbar')) { |
| 931 |
if (Metasync_Otto_Frontend_Toolbar::is_otto_disabled($post_id)) { |
| 932 |
return; |
| 933 |
} |
| 934 |
} |
| 935 |
|
| 936 |
# Skip OTTO on MetaSync custom HTML / LPS-imported pages — they ship their |
| 937 |
# own complete, self-contained SEO and OTTO must not inject or overwrite it |
| 938 |
# with a different/older project's SEO. Resolve the queried object |
| 939 |
# id (with get_the_ID() fallback) so the static-front-page case — an LPS home |
| 940 |
# set as the WP front page, where is_page() is false — is still detected. |
| 941 |
# Applying the skip here, upstream of the single render_route_html() entry, |
| 942 |
# covers all three render paths (Rocket buffer, output buffer, HTTP fallback). |
| 943 |
# Only singular views (posts/pages, incl. a static front page) can be a |
| 944 |
# custom/LPS page; gate on is_singular() so an archive/search/term query |
| 945 |
# can never have its object id mistaken for a custom page's post id. |
| 946 |
$custom_page_id = is_singular() ? ( get_queried_object_id() ?: get_the_ID() ) : 0; |
| 947 |
if (metasync_is_custom_or_lps_page($custom_page_id)) { |
| 948 |
if (!headers_sent() && Metasync_Otto_Render_Strategy::diagnostics_enabled()) { |
| 949 |
header('X-MetaSync-OTTO-Method: EXCLUDED'); |
| 950 |
} |
| 951 |
return; |
| 952 |
} |
| 953 |
|
| 954 |
# check if we are having an otto request |
| 955 |
# |
| 956 |
# Recognise our own internal fetch by header/User-Agent as well as by the |
| 957 |
# query parameter. A redirect that rebuilds the URL strips the parameter, |
| 958 |
# and without this the redirected request re-enters OTTO and fires another |
| 959 |
# internal fetch - an unbounded self-request loop. |
| 960 |
# method_exists() is redundant to static analysis - this MR adds the method, |
| 961 |
# so PHPStan proves the call always true. It is kept for the upgrade window: |
| 962 |
# during a plugin update an opcache can still hold the previous |
| 963 |
# Metasync_Otto_Render_Strategy, where class_exists() passes but the method |
| 964 |
# is absent, and an unguarded static call would fatal the page. |
| 965 |
# @phpstan-ignore-next-line function.alreadyNarrowedType |
| 966 |
$otto_has_fetch_detector = class_exists('Metasync_Otto_Render_Strategy') && method_exists('Metasync_Otto_Render_Strategy', 'is_internal_fetch'); |
| 967 |
|
| 968 |
$otto_is_internal_fetch = $otto_has_fetch_detector |
| 969 |
? Metasync_Otto_Render_Strategy::is_internal_fetch() |
| 970 |
: !empty($_GET['is_otto_page_fetch']); |
| 971 |
|
| 972 |
if($otto_is_internal_fetch){ |
| 973 |
|
| 974 |
# Block SEO plugins NOW for this internal fetch request |
| 975 |
# metasync_otto_block_seo_plugins(); |
| 976 |
# $_SERVER['REQUEST_URI'] = remove_query_arg('is_otto_page_fetch', $_SERVER['REQUEST_URI']); |
| 977 |
$block_title = !empty($_GET['otto_block_title']) && $_GET['otto_block_title'] === '1'; |
| 978 |
$block_description = !empty($_GET['otto_block_desc']) && $_GET['otto_block_desc'] === '1'; |
| 979 |
|
| 980 |
# Block SEO plugins conditionally based on what Otto has |
| 981 |
if ($block_title || $block_description) { |
| 982 |
metasync_otto_block_seo_plugins($block_title, $block_description); |
| 983 |
} |
| 984 |
|
| 985 |
# Remove ALL Otto parameters from REQUEST_URI to prevent them from appearing in pagination, etc. |
| 986 |
$request_uri_raw = sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'] ?? '')); |
| 987 |
$_SERVER['REQUEST_URI'] = remove_query_arg( |
| 988 |
['is_otto_page_fetch', 'otto_block_title', 'otto_block_desc'], |
| 989 |
$request_uri_raw |
| 990 |
); |
| 991 |
|
| 992 |
# Also remove from $_GET to prevent WordPress from using them |
| 993 |
unset($_GET['is_otto_page_fetch']); |
| 994 |
unset($_GET['otto_block_title']); |
| 995 |
unset($_GET['otto_block_desc']); |
| 996 |
return; |
| 997 |
} |
| 998 |
|
| 999 |
# to avoid unnecessary processese |
| 1000 |
# OPTIMIZED: check that otto is configured |
| 1001 |
# And the UUID is properly set before running OTT |
| 1002 |
|
| 1003 |
# check that we have the option |
| 1004 |
if(!Metasync_Otto_Config::is_otto_enabled()){ |
| 1005 |
return; |
| 1006 |
} |
| 1007 |
|
| 1008 |
# check that otto is enabled |
| 1009 |
if(!$otto_enabled){ |
| 1010 |
return; |
| 1011 |
} |
| 1012 |
|
| 1013 |
# get the otto uuid |
| 1014 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 1015 |
|
| 1016 |
# start the class |
| 1017 |
$otto = new Metasync_otto_pixel($otto_uuid); |
| 1018 |
|
| 1019 |
# call render |
| 1020 |
$otto->render_route_html(); |
| 1021 |
} |
| 1022 |
|
| 1023 |
/** |
| 1024 |
* Handle cache plugin compatibility with Otto |
| 1025 |
* Controls DONOTCACHEPAGE constant based on active plugins and configuration |
| 1026 |
* This function is called early in the WordPress lifecycle |
| 1027 |
*/ |
| 1028 |
function metasync_otto_handle_cache_compatibility() { |
| 1029 |
# Detect active plugins. SG Optimizer detection now lives in |
| 1030 |
# metasync_otto_disable_sg_page_cache(), which is invoked later |
| 1031 |
# once OTTO confirms suggestions for the URL. |
| 1032 |
$brizy_active = class_exists('Brizy_Editor') || defined('BRIZY_VERSION'); |
| 1033 |
$wp_rocket_active = class_exists('WP_Rocket'); |
| 1034 |
|
| 1035 |
# OPTIMIZED: Get configuration option using cached config |
| 1036 |
$wp_rocket_compat_mode = Metasync_Otto_Config::get_wp_rocket_compat_mode(); |
| 1037 |
|
| 1038 |
# Check for Brizy posts in database |
| 1039 |
global $wpdb; |
| 1040 |
$has_brizy_posts = false; |
| 1041 |
|
| 1042 |
if ($brizy_active) { |
| 1043 |
# OPTIMIZED: Check cache first (1-hour TTL) to avoid querying on every page load |
| 1044 |
$cached = get_transient('metasync_has_brizy_posts'); |
| 1045 |
if ($cached !== false) { |
| 1046 |
$has_brizy_posts = ($cached === 'yes'); |
| 1047 |
} else { |
| 1048 |
# Query database only if cache missed |
| 1049 |
$has_brizy_posts = $wpdb->get_var( |
| 1050 |
"SELECT COUNT(*) FROM {$wpdb->postmeta} |
| 1051 |
WHERE meta_key = 'brizy_post_uid' |
| 1052 |
AND meta_value != '' |
| 1053 |
LIMIT 1" |
| 1054 |
); |
| 1055 |
|
| 1056 |
# Cache result for 1 hour |
| 1057 |
$result = !empty($has_brizy_posts) ? 'yes' : 'no'; |
| 1058 |
set_transient('metasync_has_brizy_posts', $result, HOUR_IN_SECONDS); |
| 1059 |
} |
| 1060 |
} |
| 1061 |
|
| 1062 |
# Determine if DONOTCACHEPAGE should be set |
| 1063 |
$should_set_donotcachepage = false; |
| 1064 |
|
| 1065 |
# Case 1: Brizy is active with posts - always needed. Brizy pages are |
| 1066 |
# dynamically rendered regardless of OTTO, so caching must be disabled |
| 1067 |
# whenever Brizy posts exist. |
| 1068 |
if ($brizy_active && !empty($has_brizy_posts)) { |
| 1069 |
$should_set_donotcachepage = true; |
| 1070 |
} |
| 1071 |
|
| 1072 |
# Case 2: User explicitly disabled Otto for WP Rocket compatibility |
| 1073 |
elseif ($wp_rocket_active && $wp_rocket_compat_mode === 'disable_otto') { |
| 1074 |
$should_set_donotcachepage = true; |
| 1075 |
return; # Exit early, Otto won't run |
| 1076 |
} |
| 1077 |
|
| 1078 |
# Case 3: WP Rocket active with auto/buffer mode - DON'T set DONOTCACHEPAGE |
| 1079 |
# This allows WP Rocket optimizations to continue working |
| 1080 |
|
| 1081 |
# NOTE: SiteGround SG Optimizer cache bypass is intentionally NOT |
| 1082 |
# handled here. Emitting no-cache headers / disabling SG cache on this hook |
| 1083 |
# fired on EVERY front-end page — including pages OTTO never modifies — |
| 1084 |
# forcing SG to bypass its page cache site-wide and spiking CPU. The SG |
| 1085 |
# bypass is now deferred to metasync_otto_disable_sg_page_cache(), called |
| 1086 |
# from render_route_html() only after OTTO confirms it has suggestions for |
| 1087 |
# the current URL. |
| 1088 |
|
| 1089 |
# Only set DONOTCACHEPAGE if needed |
| 1090 |
if ($should_set_donotcachepage && !defined('DONOTCACHEPAGE')) { |
| 1091 |
define('DONOTCACHEPAGE', true); |
| 1092 |
} |
| 1093 |
} |
| 1094 |
|
| 1095 |
/** |
| 1096 |
* Disable SiteGround SG Optimizer page caching for the CURRENT request only. |
| 1097 |
* |
| 1098 |
* Called from Metasync_otto_pixel::render_route_html() once OTTO has confirmed |
| 1099 |
* it has suggestions to apply to the requested URL. Scoping the no-cache |
| 1100 |
* override to pages OTTO actually modifies — rather than emitting it on every |
| 1101 |
* front-end page via the unconditional `wp` hook — keeps SG's page cache |
| 1102 |
* working site-wide and avoids the CPU spike reported in. |
| 1103 |
* |
| 1104 |
* Only meaningful on SiteGround sites without WP Rocket; a no-op otherwise. |
| 1105 |
*/ |
| 1106 |
function metasync_otto_disable_sg_page_cache() { |
| 1107 |
# Only relevant when SG Optimizer is active and WP Rocket is not present. |
| 1108 |
if (class_exists('WP_Rocket')) { |
| 1109 |
return; |
| 1110 |
} |
| 1111 |
if (!is_plugin_active('sg-cachepress/sg-cachepress.php')) { |
| 1112 |
return; |
| 1113 |
} |
| 1114 |
|
| 1115 |
if (!defined('DONOTCACHEPAGE')) { |
| 1116 |
define('DONOTCACHEPAGE', true); |
| 1117 |
} |
| 1118 |
if (!defined('SG_CachePress_SUPERCACHER')) { |
| 1119 |
define('SG_CachePress_SUPERCACHER', false); |
| 1120 |
} |
| 1121 |
|
| 1122 |
add_filter('sgo_html_cache_disable', '__return_true', 999); |
| 1123 |
add_filter('sgo_css_combine_exclude', '__return_true', 999); |
| 1124 |
add_filter('sgo_js_combine_exclude', '__return_true', 999); |
| 1125 |
add_filter('sgo_cache_this_page', '__return_false', 999); |
| 1126 |
|
| 1127 |
if (!headers_sent()) { |
| 1128 |
header('Cache-Control: no-cache, must-revalidate, max-age=0'); |
| 1129 |
header('X-Accel-Expires: 0'); |
| 1130 |
} |
| 1131 |
} |
| 1132 |
|
| 1133 |
/** |
| 1134 |
* How long a response that OTTO could not optimize may be cached, in seconds. |
| 1135 |
* |
| 1136 |
* Deliberately short rather than zero — see |
| 1137 |
* metasync_otto_limit_response_cache() for why. |
| 1138 |
*/ |
| 1139 |
if (!defined('METASYNC_OTTO_FALLBACK_CACHE_TTL')) { |
| 1140 |
define('METASYNC_OTTO_FALLBACK_CACHE_TTL', 60); |
| 1141 |
} |
| 1142 |
|
| 1143 |
/** |
| 1144 |
* Cap how long the current response may be cached, because OTTO could not apply |
| 1145 |
* its suggestions to it. |
| 1146 |
* |
| 1147 |
* OTTO fails open: when suggestions are unavailable, or when the HTML rewrite |
| 1148 |
* fails, the site's own un-optimized markup is served instead. That markup is |
| 1149 |
* valid — the harm is that a caching layer then stores it and serves it to |
| 1150 |
* everyone for the rest of its TTL (a host edge cache with s-maxage=86400 pins |
| 1151 |
* it for a day), which surfaces to customers as the OTTO title "reverting". |
| 1152 |
* |
| 1153 |
* A short TTL is used rather than `no-store`, which is the obvious choice but the |
| 1154 |
* wrong one. Some failures look transient and are not: a page that always breaks |
| 1155 |
* the rewriter, or a host that cannot fetch its own pages, fails identically on |
| 1156 |
* every request. `no-store` would make those permanently uncacheable, sending |
| 1157 |
* 100% of their traffic to the origin forever in exchange for markup that will |
| 1158 |
* never improve. A one-minute ceiling fixes the reported bug just as completely — |
| 1159 |
* a day-long pin becomes a minute — while capping the cost of being wrong about |
| 1160 |
* which failures recover. |
| 1161 |
* |
| 1162 |
* Different layers read different signals, so all of them are emitted: |
| 1163 |
* |
| 1164 |
* - Cache-Control CDNs, host edge caches, browsers. `max-age=0` keeps |
| 1165 |
* browsers revalidating while `s-maxage` lets shared |
| 1166 |
* caches absorb a burst for the ceiling's duration. |
| 1167 |
* - X-Accel-Expires Nginx's own directive. Evaluated with higher |
| 1168 |
* precedence than Cache-Control, and not covered by |
| 1169 |
* the `fastcgi_ignore_headers Cache-Control Expires` |
| 1170 |
* recipe common on managed hosts — so it can survive |
| 1171 |
* a config that discards Cache-Control. Nginx strips |
| 1172 |
* X-Accel-* before the response reaches the client, |
| 1173 |
* so there is no visitor-facing side effect. |
| 1174 |
* - Surrogate-Control Varnish and Fastly. Note that on stock Varnish the |
| 1175 |
* mere presence of this header suppresses its |
| 1176 |
* Cache-Control check, so it must carry a real |
| 1177 |
* lifetime rather than be left at some other value. |
| 1178 |
* - DONOTCACHEPAGE WordPress-level page caches (WP Rocket, W3TC, |
| 1179 |
* LiteSpeed, SG Optimizer). These run inside PHP and |
| 1180 |
* never see response headers. They have no notion of a |
| 1181 |
* short TTL — it is cache or do not — so they skip |
| 1182 |
* entirely. That is safe here because callers only |
| 1183 |
* invoke this for failures expected to resolve; a |
| 1184 |
* permanent one (an oversized page, a host that cannot |
| 1185 |
* reach itself) is left fully cacheable by the caller |
| 1186 |
* and never reaches this function. |
| 1187 |
* |
| 1188 |
* Cache-tagging headers (Surrogate-Key, Cache-Tag, Edge-Cache-Tag) are |
| 1189 |
* deliberately left in place. They are purge-targeting labels, not cacheability |
| 1190 |
* directives — no CDN stores a response *because* a tag is present. On a layer |
| 1191 |
* that stores this response anyway, those tags are the only handle a later purge |
| 1192 |
* has on it, so removing them would discard the recovery path exactly where it |
| 1193 |
* is needed most. |
| 1194 |
* |
| 1195 |
* Some layers may honour none of this; a managed host that overrides origin |
| 1196 |
* cache headers will cache the response regardless. This bounds the damage where |
| 1197 |
* it can and is inert where it cannot. |
| 1198 |
* |
| 1199 |
* @param string $reason Short machine-readable reason, surfaced as a response |
| 1200 |
* header for debugging (e.g. 'API_ERROR', 'RENDER_FAILED'). |
| 1201 |
*/ |
| 1202 |
/** |
| 1203 |
* Build the response headers that cap how long an un-optimized response may be |
| 1204 |
* cached. |
| 1205 |
* |
| 1206 |
* Thin wrapper over Metasync_Otto_Render_Strategy::fallback_cache_headers(), which |
| 1207 |
* owns the header set so it can be unit-tested (this file cannot be loaded in a |
| 1208 |
* test context). Returns an empty array if the strategy class is somehow absent, |
| 1209 |
* so a caller emits nothing rather than an untested divergent copy. |
| 1210 |
* |
| 1211 |
* @param int $ttl Cache ceiling in seconds. |
| 1212 |
* @param string $reason Optional machine-readable reason for the debug header. |
| 1213 |
* @return array Header name => value. |
| 1214 |
*/ |
| 1215 |
function metasync_otto_fallback_cache_headers($ttl, $reason = '') { |
| 1216 |
if (!class_exists('Metasync_Otto_Render_Strategy')) { |
| 1217 |
return []; |
| 1218 |
} |
| 1219 |
|
| 1220 |
return Metasync_Otto_Render_Strategy::fallback_cache_headers_for_response($ttl, $reason); |
| 1221 |
} |
| 1222 |
|
| 1223 |
function metasync_otto_limit_response_cache($reason = '') { |
| 1224 |
# This function must only ever make the response LESS cacheable. header() |
| 1225 |
# replaces by default and the capped set carries X-Accel-Expires, which nginx |
| 1226 |
# evaluates ahead of Cache-Control — so emitting it unconditionally could |
| 1227 |
# override an upstream no-store and grant edge caching that nothing had |
| 1228 |
# granted. The strategy reconciles existing cache directives before emitting |
| 1229 |
# a capped set. |
| 1230 |
|
| 1231 |
$strategy_available = class_exists('Metasync_Otto_Render_Strategy'); |
| 1232 |
|
| 1233 |
# 1. Logged-in users. OTTO renders for them unless the "disable on logged in" |
| 1234 |
# setting is switched on, which is off by default — so this is the common |
| 1235 |
# case, not an edge case. Their page may carry session-specific or |
| 1236 |
# membership-gated content, and capping at 60 seconds would still let a |
| 1237 |
# shared cache store and re-serve it to somebody else. |
| 1238 |
if (function_exists('is_user_logged_in') && is_user_logged_in()) { |
| 1239 |
if (!headers_sent() && $strategy_available) { |
| 1240 |
foreach (Metasync_Otto_Render_Strategy::no_store_headers($reason) as $name => $value) { |
| 1241 |
header($name . ': ' . $value); |
| 1242 |
} |
| 1243 |
} |
| 1244 |
if (!defined('DONOTCACHEPAGE')) { |
| 1245 |
define('DONOTCACHEPAGE', true); |
| 1246 |
} |
| 1247 |
return; |
| 1248 |
} |
| 1249 |
|
| 1250 |
# 2. Something upstream already restricted this response. Leave its headers |
| 1251 |
# alone — a page deliberately marked no-store must not be handed a 60 |
| 1252 |
# second shared-cache lifetime by us. |
| 1253 |
# The strategy turns an existing no-store into a consistent no-store set and |
| 1254 |
# uses the strictest known existing lifetime for every cache layer. |
| 1255 |
$ttl = (int) METASYNC_OTTO_FALLBACK_CACHE_TTL; |
| 1256 |
|
| 1257 |
if (!headers_sent()) { |
| 1258 |
foreach (metasync_otto_fallback_cache_headers($ttl, $reason) as $name => $value) { |
| 1259 |
header($name . ': ' . $value); |
| 1260 |
} |
| 1261 |
} |
| 1262 |
|
| 1263 |
# Reaches page-cache plugins, which do not see the headers above. |
| 1264 |
if (!defined('DONOTCACHEPAGE')) { |
| 1265 |
define('DONOTCACHEPAGE', true); |
| 1266 |
} |
| 1267 |
} |
| 1268 |
|
| 1269 |
/** |
| 1270 |
* Report an OTTO render failure to the local error log and to Sentry. |
| 1271 |
* |
| 1272 |
* These failures were invisible: the render path swallowed them and returned the |
| 1273 |
* original HTML, so nobody knew how often they fired. Reporting them is what makes |
| 1274 |
* a persistently-failing URL discoverable instead of silently costing origin work. |
| 1275 |
* |
| 1276 |
* Two constraints shape this: |
| 1277 |
* |
| 1278 |
* 1. It runs on the visitor render path, so it must not add latency. The Sentry |
| 1279 |
* report is therefore queued here and sent by |
| 1280 |
* metasync_otto_flush_render_failure_reports() after the response has been |
| 1281 |
* handed to the visitor, so the cost is invisible rather than merely small. |
| 1282 |
* |
| 1283 |
* Two alternatives were tried and rejected. Sending inline with a blocking |
| 1284 |
* transport — cURL with a 5s connect and 5s read timeout — could add ten |
| 1285 |
* seconds to a page load, precisely when the request is already degraded. |
| 1286 |
* Sending inline non-blocking is far cheaper but still not free: opening the |
| 1287 |
* connection and writing the body costs the visitor something. |
| 1288 |
* |
| 1289 |
* A WP-Cron single event was the other obvious option and is worse than both: |
| 1290 |
* every scheduled event is stored in the `cron` option, which is autoloaded on |
| 1291 |
* every request site-wide, and WP's own 10-minute duplicate suppression keys on |
| 1292 |
* the serialized args — which carry per-request byte counts here, so it would |
| 1293 |
* never engage. A site with many failing URLs would grow that option |
| 1294 |
* unboundedly and pay for it on every page load, which is the cost the deferral |
| 1295 |
* was meant to avoid in the first place. |
| 1296 |
* |
| 1297 |
* The local log stays synchronous. It is cheaper than Sentry but not free — |
| 1298 |
* Metasync_Error_Logger::log() rewrites a summary option on each call — which is |
| 1299 |
* the second reason for the throttle below. |
| 1300 |
* |
| 1301 |
* 2. A failing page can be requested thousands of times an hour, so reports are |
| 1302 |
* throttled per reason+URL. |
| 1303 |
* |
| 1304 |
* @param string $reason Machine-readable failure reason (e.g. 'RENDER_EXCEPTION'). |
| 1305 |
* @param string $message Human-readable description. |
| 1306 |
* @param array $context Extra key/value context. |
| 1307 |
* @param string $level Sentry level: 'error' or 'warning'. |
| 1308 |
*/ |
| 1309 |
function metasync_otto_report_render_failure($reason, $message, $context = [], $level = 'error') { |
| 1310 |
$url = ''; |
| 1311 |
if (!empty($_SERVER['HTTP_HOST']) && !empty($_SERVER['REQUEST_URI'])) { |
| 1312 |
$url = (is_ssl() ? 'https://' : 'http://') |
| 1313 |
. $_SERVER['HTTP_HOST'] |
| 1314 |
. strtok(sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])), '?'); |
| 1315 |
} |
| 1316 |
|
| 1317 |
# Throttle: at most one report per reason+URL per 5 minutes. Requests without a |
| 1318 |
# resolvable URL (CLI, odd SAPIs) collapse onto one key, which is intended — |
| 1319 |
# they cannot be attributed to a page anyway. |
| 1320 |
$throttle_key = 'metasync_otto_fail_' . md5($reason . '|' . $url); |
| 1321 |
if (get_transient($throttle_key)) { |
| 1322 |
return; |
| 1323 |
} |
| 1324 |
set_transient($throttle_key, 1, 5 * MINUTE_IN_SECONDS); |
| 1325 |
|
| 1326 |
$context = array_merge([ |
| 1327 |
'reason' => $reason, |
| 1328 |
'url' => $url, |
| 1329 |
], $context); |
| 1330 |
|
| 1331 |
# Local structured log. Stays on the customer's site, so it is useful to |
| 1332 |
# support but not visible to us — hence the Sentry dispatch below. |
| 1333 |
if (class_exists('Metasync_Error_Logger')) { |
| 1334 |
Metasync_Error_Logger::log( |
| 1335 |
Metasync_Error_Logger::CATEGORY_OTTO_RENDER, |
| 1336 |
$level === 'warning' |
| 1337 |
? Metasync_Error_Logger::SEVERITY_WARNING |
| 1338 |
: Metasync_Error_Logger::SEVERITY_ERROR, |
| 1339 |
$message, |
| 1340 |
$context |
| 1341 |
); |
| 1342 |
} |
| 1343 |
|
| 1344 |
# Sentry. Queued rather than sent here: even a non-blocking dispatch costs the |
| 1345 |
# visitor something (opening the connection and writing the body), and this runs |
| 1346 |
# on a request that is already degraded. The queue is flushed after the response |
| 1347 |
# has been handed to the visitor — see metasync_otto_flush_render_failure_reports(). |
| 1348 |
if (!function_exists('metasync_sentry_capture_message_nonblocking')) { |
| 1349 |
return; |
| 1350 |
} |
| 1351 |
|
| 1352 |
$GLOBALS['metasync_otto_pending_failure_reports'][] = [ |
| 1353 |
'message' => $message, |
| 1354 |
'level' => $level, |
| 1355 |
'context' => $context, |
| 1356 |
]; |
| 1357 |
|
| 1358 |
if (empty($GLOBALS['metasync_otto_failure_flush_hooked'])) { |
| 1359 |
$GLOBALS['metasync_otto_failure_flush_hooked'] = true; |
| 1360 |
# PHP_INT_MAX so this runs after everything else registered on shutdown — |
| 1361 |
# WordPress flushes its own output buffers and closes the object cache at the |
| 1362 |
# default priority, and this handler ends output for the request. |
| 1363 |
add_action('shutdown', 'metasync_otto_flush_render_failure_reports', PHP_INT_MAX); |
| 1364 |
} |
| 1365 |
} |
| 1366 |
|
| 1367 |
/** |
| 1368 |
* Send any queued render-failure reports after the response has been delivered. |
| 1369 |
* |
| 1370 |
* fastcgi_finish_request() (PHP-FPM) and litespeed_finish_request() (LiteSpeed) |
| 1371 |
* flush the response and close the connection to the visitor while leaving PHP |
| 1372 |
* running. Calling one of them here means the page is already on its way before |
| 1373 |
* any telemetry work starts, so the reporting cost is invisible to the visitor |
| 1374 |
* rather than merely small. |
| 1375 |
* |
| 1376 |
* Neither exists under mod_php or the CLI SAPI. There the reports are still sent |
| 1377 |
* at the very end of the request, after all page output has been generated and |
| 1378 |
* flushed — the same ordering, just without the connection being closed first. |
| 1379 |
* |
| 1380 |
* The dispatch itself stays non-blocking. Now that the visitor is served, the only |
| 1381 |
* remaining cost is PHP worker occupancy, and a blocking transport with a 5s |
| 1382 |
* connect and 5s read timeout could tie up a worker for ten seconds per report on |
| 1383 |
* a site that is failing often. |
| 1384 |
*/ |
| 1385 |
function metasync_otto_flush_render_failure_reports() { |
| 1386 |
$reports = $GLOBALS['metasync_otto_pending_failure_reports'] ?? []; |
| 1387 |
$GLOBALS['metasync_otto_pending_failure_reports'] = []; |
| 1388 |
|
| 1389 |
if (empty($reports)) { |
| 1390 |
return; |
| 1391 |
} |
| 1392 |
|
| 1393 |
if (function_exists('fastcgi_finish_request')) { |
| 1394 |
fastcgi_finish_request(); |
| 1395 |
} elseif (function_exists('litespeed_finish_request')) { |
| 1396 |
litespeed_finish_request(); |
| 1397 |
} |
| 1398 |
|
| 1399 |
if (!function_exists('metasync_sentry_capture_message_nonblocking')) { |
| 1400 |
return; |
| 1401 |
} |
| 1402 |
|
| 1403 |
foreach ($reports as $report) { |
| 1404 |
try { |
| 1405 |
metasync_sentry_capture_message_nonblocking( |
| 1406 |
$report['message'], |
| 1407 |
$report['level'], |
| 1408 |
$report['context'] |
| 1409 |
); |
| 1410 |
} catch (Exception $e) { |
| 1411 |
# Telemetry must never be able to break a request. |
| 1412 |
} catch (Error $e) { |
| 1413 |
# Same. |
| 1414 |
} |
| 1415 |
} |
| 1416 |
} |
| 1417 |
|
| 1418 |
/** |
| 1419 |
* Whether MetaSync is still emitting Open Graph / Twitter tags. |
| 1420 |
* |
| 1421 |
* The Social Media & Open Graph switch stops MetaSync writing those tags on |
| 1422 |
* every render path, so the suppression below has to stand down with it — |
| 1423 |
* removing another plugin's og:/twitter: tag while contributing none of our own |
| 1424 |
* would leave the page with no social tags at all. |
| 1425 |
* |
| 1426 |
* The class_exists() guard keeps this usable if the flags file has not loaded |
| 1427 |
* yet; treating that window as "enabled" preserves the previous behaviour. |
| 1428 |
* |
| 1429 |
* @return bool |
| 1430 |
*/ |
| 1431 |
function metasync_otto_social_output_enabled() { |
| 1432 |
if (!class_exists('Metasync_Feature_Flags')) { |
| 1433 |
return true; |
| 1434 |
} |
| 1435 |
|
| 1436 |
return Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::SOCIAL_OG); |
| 1437 |
} |
| 1438 |
|
| 1439 |
/** |
| 1440 |
* Block SEO plugins conditionally based on what Otto is providing |
| 1441 |
* Only blocks title if Otto has title, only blocks description if Otto has description |
| 1442 |
* This prevents duplicate SEO tags while allowing fallback to SEO plugins when Otto has no data. |
| 1443 |
* Supports Yoast SEO, Rank Math, and AIOSEO (free + pro). |
| 1444 |
* |
| 1445 |
* @param bool $block_title Whether to block title tags. |
| 1446 |
* @param bool $block_description Whether to block description tags. |
| 1447 |
* @param array $description_tags Optional. Granular list of OTTO-provided description tags |
| 1448 |
* (e.g. ['meta[name=description]', 'meta[property=og:description]']). |
| 1449 |
* When provided, only matching AIOSEO tags are suppressed. |
| 1450 |
* When empty, all AIOSEO description tags are suppressed (legacy behavior). |
| 1451 |
*/ |
| 1452 |
function metasync_otto_block_seo_plugins($block_title = false, $block_description = false, $description_tags = []) { |
| 1453 |
# Disable Yoast SEO (free and premium) |
| 1454 |
if (is_plugin_active('wordpress-seo/wp-seo.php') || |
| 1455 |
is_plugin_active('wordpress-seo-premium/wp-seo-premium.php')) { |
| 1456 |
|
| 1457 |
# TITLE: Never block Yoast's title output during SSR fetch. |
| 1458 |
# Yoast removes WordPress's native _wp_render_title_tag action and is the sole |
| 1459 |
# renderer of the <title> tag. Returning false/empty from wpseo_title or removing |
| 1460 |
# Title_Presenter leaves the page with NO <title> tag at all — OTTO's buffer |
| 1461 |
# post-processing then has nothing to replace, producing a missing title. |
| 1462 |
# Instead, let Yoast render its own title; OTTO's replace_title() will overwrite |
| 1463 |
# it in the final HTML buffer. deduplicate_title_tags() cleans up any duplicates. |
| 1464 |
|
| 1465 |
# Block description only if Otto has description |
| 1466 |
if ($block_description) { |
| 1467 |
add_filter('wpseo_metadesc', '__return_false', 999); |
| 1468 |
add_filter('wpseo_meta_description', '__return_false', 999); |
| 1469 |
add_filter('wpseo_metakeywords', '__return_false', 999); |
| 1470 |
} |
| 1471 |
|
| 1472 |
# Block Yoast's modern presenters — description only, never title |
| 1473 |
add_filter('wpseo_frontend_presenters', function($presenters) use ($block_description) { |
| 1474 |
if (!is_array($presenters)) return $presenters; |
| 1475 |
|
| 1476 |
$presenters_to_remove = []; |
| 1477 |
|
| 1478 |
|
| 1479 |
# Remove description presenters only when OTTO has a description |
| 1480 |
if ($block_description) { |
| 1481 |
# The plain meta description belongs to the SEO title/description |
| 1482 |
# feature, which has its own setting. |
| 1483 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Meta_Description_Presenter'; |
| 1484 |
|
| 1485 |
# The OG/Twitter descriptions are social tags. With that feature |
| 1486 |
# switched off MetaSync writes none of its own, so Yoast's must be |
| 1487 |
# left alone rather than removed with nothing to replace them. |
| 1488 |
if (metasync_otto_social_output_enabled()) { |
| 1489 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Open_Graph\Description_Presenter'; |
| 1490 |
$presenters_to_remove[] = 'Yoast\WP\SEO\Presenters\Twitter\Description_Presenter'; |
| 1491 |
} |
| 1492 |
} |
| 1493 |
|
| 1494 |
foreach ($presenters as $key => $presenter) { |
| 1495 |
// Safely get class name, suppressing autoload errors |
| 1496 |
// This prevents warnings when Composer autoloader tries to load deprecated Yoast files |
| 1497 |
$class_name = is_object($presenter) ? @get_class($presenter) : ''; |
| 1498 |
|
| 1499 |
if (!empty($class_name) && in_array($class_name, $presenters_to_remove)) { |
| 1500 |
unset($presenters[$key]); |
| 1501 |
} |
| 1502 |
} |
| 1503 |
return $presenters; |
| 1504 |
}, 999); |
| 1505 |
} |
| 1506 |
|
| 1507 |
# Disable Rank Math |
| 1508 |
if (is_plugin_active('seo-by-rank-math/rank-math.php') || |
| 1509 |
is_plugin_active('seo-by-rankmath/rank-math.php')) { |
| 1510 |
|
| 1511 |
if ($block_title) { |
| 1512 |
add_filter('rank_math/frontend/title', '__return_empty_string', 999); |
| 1513 |
} |
| 1514 |
|
| 1515 |
if ($block_description) { |
| 1516 |
add_filter('rank_math/frontend/description', '__return_false', 999); |
| 1517 |
add_filter('rank_math/frontend/show_keywords', '__return_false', 999); |
| 1518 |
} |
| 1519 |
} |
| 1520 |
|
| 1521 |
# Disable AIOSEO (free and pro) |
| 1522 |
if (is_plugin_active('all-in-one-seo-pack/all_in_one_seo_pack.php') || |
| 1523 |
is_plugin_active('all-in-one-seo-pack-pro/all_in_one_seo_pack.php')) { |
| 1524 |
|
| 1525 |
if ($block_title) { |
| 1526 |
# The document title belongs to the SEO title feature, which has its |
| 1527 |
# own setting; only the og:/twitter: unsets below are social. |
| 1528 |
add_filter('aioseo_title', '__return_empty_string', 999); |
| 1529 |
|
| 1530 |
if (metasync_otto_social_output_enabled()) { |
| 1531 |
add_filter('aioseo_facebook_tags', function($meta) { |
| 1532 |
if (is_array($meta)) { unset($meta['og:title']); } |
| 1533 |
return $meta; |
| 1534 |
}, 999); |
| 1535 |
add_filter('aioseo_twitter_tags', function($meta) { |
| 1536 |
if (is_array($meta)) { unset($meta['twitter:title']); } |
| 1537 |
return $meta; |
| 1538 |
}, 999); |
| 1539 |
} |
| 1540 |
} |
| 1541 |
|
| 1542 |
if ($block_description) { |
| 1543 |
# Use granular tag list when available to only block what OTTO provides |
| 1544 |
$tags = !empty($description_tags) ? $description_tags : []; |
| 1545 |
$block_standard = empty($tags) || in_array('meta[name=description]', $tags); |
| 1546 |
# Social tags only get blocked while MetaSync is still emitting them. |
| 1547 |
$social_enabled = metasync_otto_social_output_enabled(); |
| 1548 |
$block_og_desc = $social_enabled && (empty($tags) || in_array('meta[property=og:description]', $tags)); |
| 1549 |
$block_tw_desc = $social_enabled && (empty($tags) || in_array('meta[name=twitter:description]', $tags)); |
| 1550 |
|
| 1551 |
if ($block_standard) { |
| 1552 |
add_filter('aioseo_description', '__return_empty_string', 999); |
| 1553 |
} |
| 1554 |
if ($block_og_desc) { |
| 1555 |
add_filter('aioseo_facebook_tags', function($meta) { |
| 1556 |
if (is_array($meta)) { unset($meta['og:description']); } |
| 1557 |
return $meta; |
| 1558 |
}, 999); |
| 1559 |
} |
| 1560 |
if ($block_tw_desc) { |
| 1561 |
add_filter('aioseo_twitter_tags', function($meta) { |
| 1562 |
if (is_array($meta)) { unset($meta['twitter:description']); } |
| 1563 |
return $meta; |
| 1564 |
}, 999); |
| 1565 |
} |
| 1566 |
} |
| 1567 |
} |
| 1568 |
} |
| 1569 |
# Check whether OTTO is ALSO injected via JavaScript on the site (a misconfiguration |
| 1570 |
# we warn admins about). |
| 1571 |
# |
| 1572 |
# The actual detection makes a loopback HTTP request to the site's OWN url. On hosts |
| 1573 |
# that disallow same-server loopback (e.g. SiteGround), that request blocks for the |
| 1574 |
# full timeout and then fails — and when it runs inline on admin_notices it makes |
| 1575 |
# every wp-admin page hang. It must therefore NEVER run inline on an admin request. |
| 1576 |
# |
| 1577 |
# metasync_check_otto_js() is now read-only: it returns the cached result and, on a |
| 1578 |
# cache miss, schedules a one-off BACKGROUND job (WP-Cron) to compute it. The notice |
| 1579 |
# simply shows nothing until the background result is available. |
| 1580 |
function metasync_check_otto_js(){ |
| 1581 |
|
| 1582 |
$cache_key = 'metasync_otto_js_detected'; |
| 1583 |
$cached = get_transient($cache_key); |
| 1584 |
|
| 1585 |
if ($cached !== false) { |
| 1586 |
return $cached === 'yes'; |
| 1587 |
} |
| 1588 |
|
| 1589 |
# No cached result yet — compute it off the request so we never block admin |
| 1590 |
# with a loopback call. Show nothing this load. |
| 1591 |
metasync_schedule_otto_js_check(); |
| 1592 |
return false; |
| 1593 |
}; |
| 1594 |
|
| 1595 |
# Queue the background loopback detection if it isn't already scheduled. |
| 1596 |
function metasync_schedule_otto_js_check(){ |
| 1597 |
if (!wp_next_scheduled('metasync_otto_js_check_event')) { |
| 1598 |
wp_schedule_single_event(time() + 5, 'metasync_otto_js_check_event'); |
| 1599 |
} |
| 1600 |
} |
| 1601 |
|
| 1602 |
# Background worker (runs in WP-Cron context, NOT on the admin request): performs the |
| 1603 |
# blocking loopback request and caches the result. Hooked to metasync_otto_js_check_event. |
| 1604 |
function metasync_run_otto_js_check(){ |
| 1605 |
$cache_key = 'metasync_otto_js_detected'; |
| 1606 |
|
| 1607 |
# the site url with the internal-fetch marker so OTTO does not re-process it |
| 1608 |
$site_url = site_url() . '?is_otto_page_fetch=1'; |
| 1609 |
|
| 1610 |
$page_data = wp_remote_get($site_url, array('timeout' => 5, 'sslverify' => false)); |
| 1611 |
|
| 1612 |
if (is_wp_error($page_data)) { |
| 1613 |
# Loopback failed (host likely blocks same-server requests). Cache "no" so we |
| 1614 |
# don't keep re-queuing the check on every admin page load. |
| 1615 |
set_transient($cache_key, 'no', HOUR_IN_SECONDS); |
| 1616 |
return; |
| 1617 |
} |
| 1618 |
|
| 1619 |
$body = wp_remote_retrieve_body($page_data); |
| 1620 |
|
| 1621 |
# Detect the OTTO script via a lightweight regex instead of parsing |
| 1622 |
# the entire fetched page into a SimpleHtmlDom tree (which exhausted memory |
| 1623 |
# on large pages — this check runs on admin page loads). We only need to |
| 1624 |
# confirm a <script id="sa-dynamic-optimization" ... data-uuid="..."> exists. |
| 1625 |
$found = preg_match( |
| 1626 |
'/<script\b[^>]*\bid=["\']sa-dynamic-optimization["\'][^>]*\bdata-uuid=["\'][^"\']+["\']/i', |
| 1627 |
(string) $body |
| 1628 |
) |
| 1629 |
|| preg_match( |
| 1630 |
# Attribute order may vary (data-uuid before id) |
| 1631 |
'/<script\b[^>]*\bdata-uuid=["\'][^"\']+["\'][^>]*\bid=["\']sa-dynamic-optimization["\']/i', |
| 1632 |
(string) $body |
| 1633 |
); |
| 1634 |
|
| 1635 |
if ($found) { |
| 1636 |
set_transient($cache_key, 'yes', 12 * HOUR_IN_SECONDS); |
| 1637 |
return; |
| 1638 |
} |
| 1639 |
|
| 1640 |
set_transient($cache_key, 'no', 12 * HOUR_IN_SECONDS); |
| 1641 |
} |
| 1642 |
|
| 1643 |
# Register the background worker hook. |
| 1644 |
add_action('metasync_otto_js_check_event', 'metasync_run_otto_js_check'); |
| 1645 |
|
| 1646 |
# Handle AJAX Clear Cache request |
| 1647 |
# NOTE: Cache system removed - this is now a no-op |
| 1648 |
function metasync_clear_otto_cache_handler() { |
| 1649 |
if (!empty($_GET['clear_otto_cache'])) { |
| 1650 |
delete_transient('metasync_otto_js_detected'); |
| 1651 |
# Re-run the JS detection in the background so the notice refreshes. |
| 1652 |
metasync_schedule_otto_js_check(); |
| 1653 |
# Cache system has been removed - no cache to clear |
| 1654 |
wp_send_json_success(['message' => 'Cache system removed - all pages processed in real-time']); |
| 1655 |
} |
| 1656 |
else { |
| 1657 |
wp_send_json_error(['message' => 'Missing parameter']); |
| 1658 |
} |
| 1659 |
} |
| 1660 |
|
| 1661 |
# Clear cache hook |
| 1662 |
add_action('wp_ajax_metasync_clear_otto_cache', 'metasync_clear_otto_cache_handler'); |
| 1663 |
|
| 1664 |
# add admin action to check script |
| 1665 |
function metasync_show_otto_ssr_notice() { |
| 1666 |
if (!Metasync::current_user_has_plugin_access()) { |
| 1667 |
return; // Only show to admins |
| 1668 |
} |
| 1669 |
|
| 1670 |
# Get the plugin name using centralized method |
| 1671 |
$plugin_name = Metasync::get_effective_plugin_name(); |
| 1672 |
$whitelabel_otto_name = Metasync::get_whitelabel_otto_name(); |
| 1673 |
if (metasync_check_otto_js()) { |
| 1674 |
|
| 1675 |
# Show admin notice with plugin name included in the message |
| 1676 |
echo '<div class="notice notice-error"> |
| 1677 |
<p><b>Warning from ' . esc_html($plugin_name) . '</b> |
| 1678 |
<br> |
| 1679 |
' . 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 |
| 1680 |
</p> |
| 1681 |
</div>'; |
| 1682 |
} |
| 1683 |
} |
| 1684 |
|
| 1685 |
add_action('admin_notices', 'metasync_show_otto_ssr_notice'); |
| 1686 |
|
| 1687 |
|
| 1688 |
# staging dummy change |
| 1689 |
# load otto in the wp hook |
| 1690 |
add_action('wp', 'metasync_start_otto'); |
| 1691 |
|
| 1692 |
# ENHANCED OTTO SEO INTEGRATION |
| 1693 |
# Register async SEO processing hook |
| 1694 |
add_action('metasync_process_seo_job', 'metasync_process_otto_seo_data', 10, 4); |
| 1695 |
add_action('metasync_process_otto_crawl_url_job', 'metasync_handle_otto_crawl_url_job', 10, 2); |
| 1696 |
add_action('metasync_process_otto_batch_cache_job', 'metasync_handle_otto_batch_cache_job', 10, 2); |
| 1697 |
add_action('metasync_otto_pending_drainer', 'metasync_handle_otto_pending_drainer'); |
| 1698 |
|
| 1699 |
# Process OTTO SEO data and update WordPress meta fields for SEO plugins |
| 1700 |
# This function now runs asynchronously via WordPress cron system |
| 1701 |
# |
| 1702 |
# Returns one of the Metasync_Otto_Job_Status::OUTCOME_* strings so callers |
| 1703 |
# can distinguish success, confirmed no-change, retryable failure, permanent |
| 1704 |
# failure, and CPU deferral — a bare boolean cannot drive correct retry and |
| 1705 |
# purge decisions. |
| 1706 |
# |
| 1707 |
# @param string $route Fully-qualified URL to process. |
| 1708 |
# @param bool $allow_defer When false, skip CPU-deferral rescheduling (used when |
| 1709 |
# called synchronously from crawl_url_job whose own retry |
| 1710 |
# mechanism already handles failures). |
| 1711 |
# @param int $deferral_count How many times this job has already been deferred for |
| 1712 |
# CPU load. Prevents infinite reschedule loops. |
| 1713 |
# @param array|null $prefetched_seo_data Suggestions already fetched by the transient |
| 1714 |
# warm step. When an array it is used as-is so the job |
| 1715 |
# makes exactly ONE API call per URL; the endpoint is |
| 1716 |
# only fetched here when no prefetch was provided. |
| 1717 |
|
| 1718 |
function metasync_process_otto_seo_data($route, $allow_defer = true, $deferral_count = 0, $prefetched_seo_data = null) { |
| 1719 |
# Maximum number of times a job can be deferred before it is dropped. |
| 1720 |
$max_deferrals = defined('METASYNC_SEO_JOB_MAX_DEFERRALS') ? METASYNC_SEO_JOB_MAX_DEFERRALS : 5; |
| 1721 |
$lock_key = null; |
| 1722 |
$lock_acquired = false; |
| 1723 |
|
| 1724 |
try { |
| 1725 |
# Validate input |
| 1726 |
if (empty($route) || !is_string($route)) { |
| 1727 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1728 |
} |
| 1729 |
|
| 1730 |
# CPU load check — defer if server is under load. |
| 1731 |
# Only reschedule when called from the cron hook (allow_defer=true) and |
| 1732 |
# we haven't exceeded the maximum deferral count. |
| 1733 |
# Checked BEFORE acquiring the lock so that deferred jobs don't |
| 1734 |
# acquire-then-immediately-release it (which defeats concurrency protection). |
| 1735 |
if (!Metasync_CPU_Monitor::is_load_safe()) { |
| 1736 |
if ($allow_defer) { |
| 1737 |
if ($deferral_count < $max_deferrals) { |
| 1738 |
metasync_otto_schedule_single_event( |
| 1739 |
time() + 60, |
| 1740 |
'metasync_process_seo_job', |
| 1741 |
array($route, true, $deferral_count + 1) |
| 1742 |
); |
| 1743 |
return Metasync_Otto_Job_Status::OUTCOME_DEFERRED; |
| 1744 |
} |
| 1745 |
# deferral budget exhausted — RUN the job late instead |
| 1746 |
# of dropping it. The write is idempotent and bounded, and a |
| 1747 |
# late sync is strictly better than SEO data that never |
| 1748 |
# arrives. No new cron events are created, so the |
| 1749 |
# anti-pile-up guarantee is preserved. Fall through. |
| 1750 |
} else { |
| 1751 |
# allow_defer=false (sync path): report retryable without creating |
| 1752 |
# cron events — the crawl-url job's retry mechanism (or OTTO's |
| 1753 |
# next crawl) re-triggers this URL. |
| 1754 |
return Metasync_Otto_Job_Status::OUTCOME_RETRYABLE; |
| 1755 |
} |
| 1756 |
} |
| 1757 |
|
| 1758 |
# Concurrency lock: prevent multiple SEO jobs from running simultaneously. |
| 1759 |
# Uses a per-URL transient lock with a 120s TTL as a safety net (the lock is |
| 1760 |
# explicitly deleted on completion). If the lock exists another run is already |
| 1761 |
# processing this URL — reschedule once with a short delay instead of stacking. |
| 1762 |
$lock_key = 'metasync_seo_lock_' . md5($route); |
| 1763 |
if (get_transient($lock_key) !== false) { |
| 1764 |
if ($allow_defer && $deferral_count < $max_deferrals) { |
| 1765 |
metasync_otto_schedule_single_event( |
| 1766 |
time() + 30, |
| 1767 |
'metasync_process_seo_job', |
| 1768 |
array($route, true, $deferral_count + 1) |
| 1769 |
); |
| 1770 |
} |
| 1771 |
return Metasync_Otto_Job_Status::OUTCOME_RETRYABLE; |
| 1772 |
} |
| 1773 |
set_transient($lock_key, true, 120); |
| 1774 |
$lock_acquired = true; |
| 1775 |
|
| 1776 |
# Resolve redirect table: use final destination URL before 404 checks and OTTO processing |
| 1777 |
$route = metasync_otto_resolve_redirect_to_final_url($route); |
| 1778 |
|
| 1779 |
# Skip excluded URLs - don't process SEO data for them |
| 1780 |
if (metasync_is_otto_url_excluded($route)) { |
| 1781 |
//error_log('MetaSync OTTO: Skipping SEO processing for excluded URL: ' . $route); |
| 1782 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1783 |
} |
| 1784 |
|
| 1785 |
# Pre-flight 404 check: exclude URLs that would return 404 before making API call |
| 1786 |
if (!metasync_otto_is_url_available($route)) { |
| 1787 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404: {$route}"); |
| 1788 |
metasync_otto_auto_exclude_404_url($route); |
| 1789 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1790 |
} |
| 1791 |
|
| 1792 |
# Get OTTO UUID from settings |
| 1793 |
# OPTIMIZED: Use cached options |
| 1794 |
$otto_uuid = Metasync_Otto_Config::get_otto_uuid(); |
| 1795 |
|
| 1796 |
if (empty($otto_uuid)) { |
| 1797 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1798 |
} |
| 1799 |
|
| 1800 |
# Meta descriptions are always enabled by default - no check needed |
| 1801 |
|
| 1802 |
# Obtain the OTTO suggestions for this URL. When the caller already |
| 1803 |
# warmed the transient cache (crawl-url job), those suggestions are |
| 1804 |
# handed in and NO second API call is made — the endpoint, rate |
| 1805 |
# limiter, and breaker are only hit once per URL. |
| 1806 |
# An empty array is a legitimate answer ("OTTO holds nothing for this |
| 1807 |
# URL") and must not be treated as a fetch failure; only a strict |
| 1808 |
# false from the fetcher is a failure. |
| 1809 |
if (is_array($prefetched_seo_data)) { |
| 1810 |
$seo_data = $prefetched_seo_data; |
| 1811 |
} else { |
| 1812 |
$failure_class = null; |
| 1813 |
$seo_data = metasync_fetch_otto_seo_data($route, $otto_uuid, $failure_class); |
| 1814 |
if ($seo_data === false) { |
| 1815 |
if ($failure_class === Metasync_Otto_Job_Status::OUTCOME_PERMANENT) { |
| 1816 |
# Auth/input-level rejection from the API — retrying the |
| 1817 |
# same request cannot help. |
| 1818 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1819 |
} |
| 1820 |
# Timeout / 5xx / 429 / empty body / bad JSON: retryable. |
| 1821 |
metasync_record_failed_action( 'metasync_process_seo_job' ); |
| 1822 |
return Metasync_Otto_Job_Status::OUTCOME_RETRYABLE; |
| 1823 |
} |
| 1824 |
} |
| 1825 |
|
| 1826 |
# Mark this URL as crawled by OTTO for SSR |
| 1827 |
# Extract domain and path from route |
| 1828 |
$parsed_url = parse_url($route); |
| 1829 |
$domain_with_scheme = ($parsed_url['scheme'] ?? 'https') . '://' . ($parsed_url['host'] ?? ''); |
| 1830 |
$url_path = ($parsed_url['path'] ?? '/'); |
| 1831 |
|
| 1832 |
# Create crawl data structure |
| 1833 |
$crawl_data = array( |
| 1834 |
'domain' => $domain_with_scheme, |
| 1835 |
'urls' => array($url_path) |
| 1836 |
); |
| 1837 |
|
| 1838 |
# Load Otto pixel class and save crawl data |
| 1839 |
$otto_pixel = new Metasync_otto_pixel($otto_uuid); |
| 1840 |
$otto_pixel->save_crawl_data($crawl_data); |
| 1841 |
|
| 1842 |
# Get WordPress post ID from URL |
| 1843 |
$post_id = url_to_postid($route); |
| 1844 |
|
| 1845 |
# Special handling for WooCommerce shop page (url_to_postid doesn't work for it) |
| 1846 |
if ((!$post_id || $post_id <= 0) && function_exists('wc_get_page_id')) { |
| 1847 |
# Check if this URL is the WooCommerce shop page |
| 1848 |
$shop_page_id = wc_get_page_id('shop'); |
| 1849 |
if ($shop_page_id > 0) { |
| 1850 |
$shop_url = get_permalink($shop_page_id); |
| 1851 |
$route_normalized = rtrim($route, '/'); |
| 1852 |
$shop_url_normalized = rtrim($shop_url, '/'); |
| 1853 |
|
| 1854 |
if ($route_normalized === $shop_url_normalized) { |
| 1855 |
$post_id = $shop_page_id; |
| 1856 |
} |
| 1857 |
} |
| 1858 |
} |
| 1859 |
|
| 1860 |
# Try to find WooCommerce product by URL if url_to_postid failed |
| 1861 |
if ((!$post_id || $post_id <= 0) && strpos($route, '/product/') !== false && function_exists('wc_get_products')) { |
| 1862 |
# Extract product slug from URL |
| 1863 |
$product_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1864 |
|
| 1865 |
# Try to get product by slug |
| 1866 |
$products = wc_get_products(array( |
| 1867 |
'name' => $product_slug, |
| 1868 |
'limit' => 1, |
| 1869 |
'status' => 'publish', |
| 1870 |
)); |
| 1871 |
|
| 1872 |
if (empty($products)) { |
| 1873 |
# Fallback: try by slug using WP_Query |
| 1874 |
$args = array( |
| 1875 |
'post_type' => 'product', |
| 1876 |
'name' => $product_slug, |
| 1877 |
'posts_per_page' => 1, |
| 1878 |
'post_status' => 'publish', |
| 1879 |
); |
| 1880 |
$query = new WP_Query($args); |
| 1881 |
|
| 1882 |
if ($query->have_posts()) { |
| 1883 |
$product_post = $query->posts[0]; |
| 1884 |
$post_id = $product_post->ID; |
| 1885 |
} |
| 1886 |
} else { |
| 1887 |
$product = $products[0]; |
| 1888 |
$post_id = $product->get_id(); |
| 1889 |
} |
| 1890 |
} |
| 1891 |
|
| 1892 |
if (!$post_id || $post_id <= 0) { |
| 1893 |
# Check if this is a category page |
| 1894 |
if (strpos($route, '/category/') !== false) { |
| 1895 |
# Extract category slug from URL |
| 1896 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1897 |
$category = get_category_by_slug($category_slug); |
| 1898 |
|
| 1899 |
if ($category) { |
| 1900 |
# Check if category would return 404 before applying OTTO changes |
| 1901 |
if (metasync_would_term_return_404($category->term_id, 'category', $route)) { |
| 1902 |
error_log("MetaSync OTTO: Skipping SEO processing for category that would return 404: {$route} (Category ID: {$category->term_id})"); |
| 1903 |
metasync_otto_auto_exclude_404_url($route); |
| 1904 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 1905 |
} |
| 1906 |
|
| 1907 |
# Update comprehensive category SEO meta fields |
| 1908 |
$update_result = metasync_update_comprehensive_category_seo_fields($category->term_id, $seo_data); |
| 1909 |
|
| 1910 |
if ($update_result['updated']) { |
| 1911 |
# Prepare trimmed values to 30 characters |
| 1912 |
$trim = function($value) { |
| 1913 |
if ($value === null) { return ''; } |
| 1914 |
$value = (string) $value; |
| 1915 |
$value = trim($value); |
| 1916 |
if (mb_strlen($value) > 30) { |
| 1917 |
return mb_substr($value, 0, 30); |
| 1918 |
} |
| 1919 |
return $value; |
| 1920 |
}; |
| 1921 |
|
| 1922 |
# Log individual field updates for category |
| 1923 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 1924 |
$short = ''; |
| 1925 |
$title = ''; |
| 1926 |
|
| 1927 |
switch ($field_type) { |
| 1928 |
case 'meta_title': |
| 1929 |
$short = $trim($field_value); |
| 1930 |
$title = "Category Meta Title Update ({$short}...)"; |
| 1931 |
break; |
| 1932 |
case 'meta_description': |
| 1933 |
$short = $trim($field_value); |
| 1934 |
$title = "Category Meta Description Update ({$short}...)"; |
| 1935 |
break; |
| 1936 |
case 'meta_keywords': |
| 1937 |
$short = $trim($field_value); |
| 1938 |
$title = "Category Meta Keywords Update ({$short}...)"; |
| 1939 |
break; |
| 1940 |
case 'og_title': |
| 1941 |
$short = $trim($field_value); |
| 1942 |
$title = "Category Open Graph Title Update ({$short}...)"; |
| 1943 |
break; |
| 1944 |
case 'og_description': |
| 1945 |
$short = $trim($field_value); |
| 1946 |
$title = "Category Open Graph Description Update ({$short}...)"; |
| 1947 |
break; |
| 1948 |
case 'twitter_title': |
| 1949 |
$short = $trim($field_value); |
| 1950 |
$title = "Category Twitter Title Update ({$short}...)"; |
| 1951 |
break; |
| 1952 |
case 'twitter_description': |
| 1953 |
$short = $trim($field_value); |
| 1954 |
$title = "Category Twitter Description Update ({$short}...)"; |
| 1955 |
break; |
| 1956 |
case 'image_alt_data': |
| 1957 |
$image_count = count($field_value); |
| 1958 |
$title = "Category Image Alt Text Update ({$image_count} images)"; |
| 1959 |
break; |
| 1960 |
case 'headings_data': |
| 1961 |
$heading_count = count($field_value); |
| 1962 |
$title = "Category Headings Update ({$heading_count} headings)"; |
| 1963 |
break; |
| 1964 |
case 'structured_data': |
| 1965 |
$title = 'Category Structured Data Update'; |
| 1966 |
break; |
| 1967 |
} |
| 1968 |
|
| 1969 |
if (!empty($title)) { |
| 1970 |
metasync_log_sync_history([ |
| 1971 |
'title' => $title, |
| 1972 |
'source' => 'OTTO SEO', |
| 1973 |
'status' => 'published', |
| 1974 |
'content_type' => 'Category SEO', |
| 1975 |
'url' => $route, |
| 1976 |
'meta_data' => json_encode([ |
| 1977 |
'field' => $field_type, |
| 1978 |
'field_value' => $field_value, |
| 1979 |
'category_id' => $category->term_id, |
| 1980 |
'category_name' => $category->name |
| 1981 |
]) |
| 1982 |
]); |
| 1983 |
} |
| 1984 |
} |
| 1985 |
|
| 1986 |
return Metasync_Otto_Job_Status::OUTCOME_SUCCESS; |
| 1987 |
} |
| 1988 |
|
| 1989 |
return Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE; |
| 1990 |
} |
| 1991 |
} |
| 1992 |
|
| 1993 |
# Check if this is a WooCommerce product category |
| 1994 |
if (strpos($route, '/product-category/') !== false) { |
| 1995 |
# Extract product category slug from URL |
| 1996 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 1997 |
$term = get_term_by('slug', $category_slug, 'product_cat'); |
| 1998 |
|
| 1999 |
if ($term && !is_wp_error($term)) { |
| 2000 |
# Check if product category would return 404 before applying OTTO changes |
| 2001 |
if (metasync_would_term_return_404($term->term_id, 'product_cat', $route)) { |
| 2002 |
error_log("MetaSync OTTO: Skipping SEO processing for product category that would return 404: {$route} (Term ID: {$term->term_id})"); |
| 2003 |
metasync_otto_auto_exclude_404_url($route); |
| 2004 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2005 |
} |
| 2006 |
|
| 2007 |
# Update comprehensive taxonomy SEO meta fields |
| 2008 |
$update_result = metasync_update_comprehensive_taxonomy_seo_fields($term->term_id, 'product_cat', $seo_data); |
| 2009 |
|
| 2010 |
if ($update_result['updated']) { |
| 2011 |
# Prepare trimmed values to 30 characters |
| 2012 |
$trim = function($value) { |
| 2013 |
if ($value === null) { return ''; } |
| 2014 |
$value = (string) $value; |
| 2015 |
$value = trim($value); |
| 2016 |
if (mb_strlen($value) > 30) { |
| 2017 |
return mb_substr($value, 0, 30); |
| 2018 |
} |
| 2019 |
return $value; |
| 2020 |
}; |
| 2021 |
|
| 2022 |
# Log individual field updates for product category |
| 2023 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 2024 |
$short = ''; |
| 2025 |
$title = ''; |
| 2026 |
|
| 2027 |
switch ($field_type) { |
| 2028 |
case 'meta_title': |
| 2029 |
$short = $trim($field_value); |
| 2030 |
$title = "Product Category Meta Title Update ({$short}...)"; |
| 2031 |
break; |
| 2032 |
case 'meta_description': |
| 2033 |
$short = $trim($field_value); |
| 2034 |
$title = "Product Category Meta Description Update ({$short}...)"; |
| 2035 |
break; |
| 2036 |
case 'meta_keywords': |
| 2037 |
$short = $trim($field_value); |
| 2038 |
$title = "Product Category Meta Keywords Update ({$short}...)"; |
| 2039 |
break; |
| 2040 |
case 'og_title': |
| 2041 |
$short = $trim($field_value); |
| 2042 |
$title = "Product Category Open Graph Title Update ({$short}...)"; |
| 2043 |
break; |
| 2044 |
case 'og_description': |
| 2045 |
$short = $trim($field_value); |
| 2046 |
$title = "Product Category Open Graph Description Update ({$short}...)"; |
| 2047 |
break; |
| 2048 |
case 'twitter_title': |
| 2049 |
$short = $trim($field_value); |
| 2050 |
$title = "Product Category Twitter Title Update ({$short}...)"; |
| 2051 |
break; |
| 2052 |
case 'twitter_description': |
| 2053 |
$short = $trim($field_value); |
| 2054 |
$title = "Product Category Twitter Description Update ({$short}...)"; |
| 2055 |
break; |
| 2056 |
case 'image_alt_data': |
| 2057 |
$image_count = count($field_value); |
| 2058 |
$title = "Product Category Image Alt Text Update ({$image_count} images)"; |
| 2059 |
break; |
| 2060 |
case 'headings_data': |
| 2061 |
$heading_count = count($field_value); |
| 2062 |
$title = "Product Category Headings Update ({$heading_count} headings)"; |
| 2063 |
break; |
| 2064 |
case 'structured_data': |
| 2065 |
$title = 'Product Category Structured Data Update'; |
| 2066 |
break; |
| 2067 |
} |
| 2068 |
|
| 2069 |
if (!empty($title)) { |
| 2070 |
metasync_log_sync_history([ |
| 2071 |
'title' => $title, |
| 2072 |
'source' => 'OTTO SEO', |
| 2073 |
'status' => 'published', |
| 2074 |
'content_type' => 'WooCommerce Product Category SEO', |
| 2075 |
'url' => $route, |
| 2076 |
'meta_data' => json_encode([ |
| 2077 |
'field' => $field_type, |
| 2078 |
'field_value' => $field_value, |
| 2079 |
'term_id' => $term->term_id, |
| 2080 |
'term_name' => $term->name, |
| 2081 |
'taxonomy' => 'product_cat' |
| 2082 |
]) |
| 2083 |
]); |
| 2084 |
} |
| 2085 |
} |
| 2086 |
|
| 2087 |
return Metasync_Otto_Job_Status::OUTCOME_SUCCESS; |
| 2088 |
} |
| 2089 |
|
| 2090 |
return Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE; |
| 2091 |
} |
| 2092 |
} |
| 2093 |
|
| 2094 |
# Check if this is the home page (landing page) |
| 2095 |
$site_url = rtrim(site_url(), '/'); |
| 2096 |
$route_clean = rtrim($route, '/'); |
| 2097 |
|
| 2098 |
if ($route_clean === $site_url) { |
| 2099 |
# Get the home page (front page) |
| 2100 |
$front_page_id = get_option('page_on_front'); |
| 2101 |
$home_page = null; |
| 2102 |
|
| 2103 |
if ($front_page_id && $front_page_id > 0) { |
| 2104 |
$home_page = get_post($front_page_id); |
| 2105 |
} else { |
| 2106 |
# If no static front page is set, get the latest post |
| 2107 |
$home_page = get_posts(['numberposts' => 1, 'post_status' => 'publish'])[0] ?? null; |
| 2108 |
} |
| 2109 |
|
| 2110 |
if ($home_page) { |
| 2111 |
# Check if home page would return 404 before applying OTTO changes |
| 2112 |
if (metasync_would_page_return_404($home_page->ID, $route)) { |
| 2113 |
error_log("MetaSync OTTO: Skipping SEO processing for home page that would return 404: {$route} (Post ID: {$home_page->ID}, Status: {$home_page->post_status})"); |
| 2114 |
metasync_otto_auto_exclude_404_url($route); |
| 2115 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2116 |
} |
| 2117 |
|
| 2118 |
# Update comprehensive home page SEO meta fields |
| 2119 |
$update_result = metasync_update_comprehensive_seo_fields($home_page->ID, $seo_data); |
| 2120 |
|
| 2121 |
if ($update_result['updated']) { |
| 2122 |
# Clear relevant caches |
| 2123 |
metasync_clear_post_seo_caches($home_page->ID); |
| 2124 |
|
| 2125 |
# Prepare trimmed values to 30 characters |
| 2126 |
$trim = function($value) { |
| 2127 |
if ($value === null) { return ''; } |
| 2128 |
$value = (string) $value; |
| 2129 |
$value = trim($value); |
| 2130 |
if (mb_strlen($value) > 30) { |
| 2131 |
return mb_substr($value, 0, 30); |
| 2132 |
} |
| 2133 |
return $value; |
| 2134 |
}; |
| 2135 |
|
| 2136 |
# Log individual field updates for home page |
| 2137 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 2138 |
$short = ''; |
| 2139 |
$title = ''; |
| 2140 |
|
| 2141 |
switch ($field_type) { |
| 2142 |
case 'meta_title': |
| 2143 |
$short = $trim($field_value); |
| 2144 |
$title = "Home Page Meta Title Update ({$short}...)"; |
| 2145 |
break; |
| 2146 |
case 'meta_description': |
| 2147 |
$short = $trim($field_value); |
| 2148 |
$title = "Home Page Meta Description Update ({$short}...)"; |
| 2149 |
break; |
| 2150 |
case 'meta_keywords': |
| 2151 |
$short = $trim($field_value); |
| 2152 |
$title = "Home Page Meta Keywords Update ({$short}...)"; |
| 2153 |
break; |
| 2154 |
case 'og_title': |
| 2155 |
$short = $trim($field_value); |
| 2156 |
$title = "Home Page Open Graph Title Update ({$short}...)"; |
| 2157 |
break; |
| 2158 |
case 'og_description': |
| 2159 |
$short = $trim($field_value); |
| 2160 |
$title = "Home Page Open Graph Description Update ({$short}...)"; |
| 2161 |
break; |
| 2162 |
case 'twitter_title': |
| 2163 |
$short = $trim($field_value); |
| 2164 |
$title = "Home Page Twitter Title Update ({$short}...)"; |
| 2165 |
break; |
| 2166 |
case 'twitter_description': |
| 2167 |
$short = $trim($field_value); |
| 2168 |
$title = "Home Page Twitter Description Update ({$short}...)"; |
| 2169 |
break; |
| 2170 |
case 'image_alt_data': |
| 2171 |
$image_count = count($field_value); |
| 2172 |
$title = "Home Page Image Alt Text Update ({$image_count} images)"; |
| 2173 |
break; |
| 2174 |
case 'headings_data': |
| 2175 |
$heading_count = count($field_value); |
| 2176 |
$title = "Home Page Headings Update ({$heading_count} headings)"; |
| 2177 |
break; |
| 2178 |
case 'structured_data': |
| 2179 |
$title = 'Home Page Structured Data Update'; |
| 2180 |
break; |
| 2181 |
} |
| 2182 |
|
| 2183 |
if (!empty($title)) { |
| 2184 |
metasync_log_sync_history([ |
| 2185 |
'title' => $title, |
| 2186 |
'source' => 'OTTO SEO', |
| 2187 |
'status' => 'published', |
| 2188 |
'content_type' => 'Home Page SEO', |
| 2189 |
'url' => $route, |
| 2190 |
'meta_data' => json_encode([ |
| 2191 |
'field' => $field_type, |
| 2192 |
'field_value' => $field_value, |
| 2193 |
'post_id' => $home_page->ID |
| 2194 |
]) |
| 2195 |
]); |
| 2196 |
} |
| 2197 |
} |
| 2198 |
|
| 2199 |
return Metasync_Otto_Job_Status::OUTCOME_SUCCESS; |
| 2200 |
} |
| 2201 |
|
| 2202 |
return Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE; |
| 2203 |
} |
| 2204 |
} |
| 2205 |
|
| 2206 |
# Check if this is the blog/posts page (page_for_posts) |
| 2207 |
$posts_page_id = intval(get_option('page_for_posts')); |
| 2208 |
if ($posts_page_id > 0) { |
| 2209 |
$posts_page = get_post($posts_page_id); |
| 2210 |
$posts_page_url = rtrim(get_permalink($posts_page_id), '/'); |
| 2211 |
|
| 2212 |
if ($posts_page && $route_clean === $posts_page_url) { |
| 2213 |
# Check if blog page would return 404 |
| 2214 |
if (metasync_would_page_return_404($posts_page->ID, $route)) { |
| 2215 |
error_log("MetaSync OTTO: Skipping SEO processing for blog page that would return 404: {$route} (Post ID: {$posts_page->ID}, Status: {$posts_page->post_status})"); |
| 2216 |
metasync_otto_auto_exclude_404_url($route); |
| 2217 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2218 |
} |
| 2219 |
|
| 2220 |
# Update comprehensive blog page SEO meta fields |
| 2221 |
$update_result = metasync_update_comprehensive_seo_fields($posts_page->ID, $seo_data); |
| 2222 |
|
| 2223 |
if ($update_result['updated']) { |
| 2224 |
# Clear relevant caches |
| 2225 |
metasync_clear_post_seo_caches($posts_page->ID); |
| 2226 |
|
| 2227 |
# Prepare trimmed values to 30 characters |
| 2228 |
$trim = function($value) { |
| 2229 |
if ($value === null) { return ''; } |
| 2230 |
$value = (string) $value; |
| 2231 |
$value = trim($value); |
| 2232 |
if (mb_strlen($value) > 30) { |
| 2233 |
return mb_substr($value, 0, 30); |
| 2234 |
} |
| 2235 |
return $value; |
| 2236 |
}; |
| 2237 |
|
| 2238 |
# Log individual field updates for blog page |
| 2239 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 2240 |
$short = ''; |
| 2241 |
$title = ''; |
| 2242 |
|
| 2243 |
switch ($field_type) { |
| 2244 |
case 'meta_title': |
| 2245 |
$short = $trim($field_value); |
| 2246 |
$title = "Blog Page Meta Title Update ({$short}...)"; |
| 2247 |
break; |
| 2248 |
case 'meta_description': |
| 2249 |
$short = $trim($field_value); |
| 2250 |
$title = "Blog Page Meta Description Update ({$short}...)"; |
| 2251 |
break; |
| 2252 |
case 'meta_keywords': |
| 2253 |
$short = $trim($field_value); |
| 2254 |
$title = "Blog Page Meta Keywords Update ({$short}...)"; |
| 2255 |
break; |
| 2256 |
case 'og_title': |
| 2257 |
$short = $trim($field_value); |
| 2258 |
$title = "Blog Page Open Graph Title Update ({$short}...)"; |
| 2259 |
break; |
| 2260 |
case 'og_description': |
| 2261 |
$short = $trim($field_value); |
| 2262 |
$title = "Blog Page Open Graph Description Update ({$short}...)"; |
| 2263 |
break; |
| 2264 |
case 'twitter_title': |
| 2265 |
$short = $trim($field_value); |
| 2266 |
$title = "Blog Page Twitter Title Update ({$short}...)"; |
| 2267 |
break; |
| 2268 |
case 'twitter_description': |
| 2269 |
$short = $trim($field_value); |
| 2270 |
$title = "Blog Page Twitter Description Update ({$short}...)"; |
| 2271 |
break; |
| 2272 |
case 'image_alt_data': |
| 2273 |
$image_count = count($field_value); |
| 2274 |
$title = "Blog Page Image Alt Text Update ({$image_count} images)"; |
| 2275 |
break; |
| 2276 |
case 'headings_data': |
| 2277 |
$heading_count = count($field_value); |
| 2278 |
$title = "Blog Page Headings Update ({$heading_count} headings)"; |
| 2279 |
break; |
| 2280 |
case 'structured_data': |
| 2281 |
$title = 'Blog Page Structured Data Update'; |
| 2282 |
break; |
| 2283 |
} |
| 2284 |
|
| 2285 |
if (!empty($title)) { |
| 2286 |
metasync_log_sync_history([ |
| 2287 |
'title' => $title, |
| 2288 |
'source' => 'OTTO SEO', |
| 2289 |
'status' => 'published', |
| 2290 |
'content_type' => 'Blog Page SEO', |
| 2291 |
'url' => $route, |
| 2292 |
'meta_data' => json_encode([ |
| 2293 |
'field' => $field_type, |
| 2294 |
'field_value' => $field_value, |
| 2295 |
'post_id' => $posts_page->ID |
| 2296 |
]) |
| 2297 |
]); |
| 2298 |
} |
| 2299 |
} |
| 2300 |
|
| 2301 |
return Metasync_Otto_Job_Status::OUTCOME_SUCCESS; |
| 2302 |
} |
| 2303 |
|
| 2304 |
return Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE; |
| 2305 |
} |
| 2306 |
} |
| 2307 |
|
| 2308 |
# URL didn't resolve to any supported entity (post, category, home page, blog page) |
| 2309 |
# Treat as 404 and auto-exclude (e.g. deleted post, non-existent page) |
| 2310 |
if (!metasync_otto_is_url_available($route)) { |
| 2311 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404 (no matching entity): {$route}"); |
| 2312 |
metasync_otto_auto_exclude_404_url($route); |
| 2313 |
} |
| 2314 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2315 |
} |
| 2316 |
|
| 2317 |
# Verify this is actually a post, page, or WooCommerce product |
| 2318 |
$post = get_post($post_id); |
| 2319 |
|
| 2320 |
# Get supported post types dynamically |
| 2321 |
$supported_post_types = metasync_get_supported_post_types(); |
| 2322 |
|
| 2323 |
if (!$post || !in_array($post->post_type, $supported_post_types)) { |
| 2324 |
# Skip unsupported post types |
| 2325 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2326 |
} |
| 2327 |
|
| 2328 |
# Check if page would return 404 before applying OTTO changes |
| 2329 |
if (metasync_would_page_return_404($post_id, $route)) { |
| 2330 |
error_log("MetaSync OTTO: Skipping SEO processing for URL that would return 404: {$route} (Post ID: {$post_id}, Status: {$post->post_status})"); |
| 2331 |
metasync_otto_auto_exclude_404_url($route); |
| 2332 |
return Metasync_Otto_Job_Status::OUTCOME_PERMANENT; |
| 2333 |
} |
| 2334 |
|
| 2335 |
# Update comprehensive SEO meta fields |
| 2336 |
$update_result = metasync_update_comprehensive_seo_fields($post_id, $seo_data); |
| 2337 |
|
| 2338 |
if ($update_result['updated']) { |
| 2339 |
# Clear relevant caches |
| 2340 |
metasync_clear_post_seo_caches($post_id); |
| 2341 |
|
| 2342 |
# Prepare trimmed values to 30 characters |
| 2343 |
$trim = function($value) { |
| 2344 |
if ($value === null) { return ''; } |
| 2345 |
$value = (string) $value; |
| 2346 |
$value = trim($value); |
| 2347 |
if (mb_strlen($value) > 30) { |
| 2348 |
return mb_substr($value, 0, 30); |
| 2349 |
} |
| 2350 |
return $value; |
| 2351 |
}; |
| 2352 |
|
| 2353 |
# Log individual field updates |
| 2354 |
foreach ($update_result['fields_updated'] as $field_type => $field_value) { |
| 2355 |
$short = ''; |
| 2356 |
$title = ''; |
| 2357 |
|
| 2358 |
switch ($field_type) { |
| 2359 |
case 'meta_title': |
| 2360 |
$short = $trim($field_value); |
| 2361 |
$title = 'Meta Title Update (' . $short . '...)'; |
| 2362 |
break; |
| 2363 |
case 'meta_description': |
| 2364 |
$short = $trim($field_value); |
| 2365 |
$title = 'Meta Description Update (' . $short . '...)'; |
| 2366 |
break; |
| 2367 |
case 'meta_keywords': |
| 2368 |
$short = $trim($field_value); |
| 2369 |
$title = 'Meta Keywords Update (' . $short . '...)'; |
| 2370 |
break; |
| 2371 |
case 'og_title': |
| 2372 |
$short = $trim($field_value); |
| 2373 |
$title = 'Open Graph Title Update (' . $short . '...)'; |
| 2374 |
break; |
| 2375 |
case 'og_description': |
| 2376 |
$short = $trim($field_value); |
| 2377 |
$title = 'Open Graph Description Update (' . $short . '...)'; |
| 2378 |
break; |
| 2379 |
case 'twitter_title': |
| 2380 |
$short = $trim($field_value); |
| 2381 |
$title = 'Twitter Title Update (' . $short . '...)'; |
| 2382 |
break; |
| 2383 |
case 'twitter_description': |
| 2384 |
$short = $trim($field_value); |
| 2385 |
$title = 'Twitter Description Update (' . $short . '...)'; |
| 2386 |
break; |
| 2387 |
case 'image_alt_data': |
| 2388 |
$image_count = count($field_value); |
| 2389 |
$title = "Image Alt Text Update ({$image_count} images)"; |
| 2390 |
break; |
| 2391 |
case 'headings_data': |
| 2392 |
$heading_count = count($field_value); |
| 2393 |
$title = "Headings Update ({$heading_count} headings)"; |
| 2394 |
break; |
| 2395 |
case 'structured_data': |
| 2396 |
$title = 'Structured Data Update'; |
| 2397 |
break; |
| 2398 |
} |
| 2399 |
|
| 2400 |
if (!empty($title)) { |
| 2401 |
metasync_log_sync_history([ |
| 2402 |
'title' => $title, |
| 2403 |
'source' => 'OTTO SEO', |
| 2404 |
'status' => 'published', |
| 2405 |
'content_type' => 'SEO Meta', |
| 2406 |
'url' => $route, |
| 2407 |
'meta_data' => json_encode([ |
| 2408 |
'field' => $field_type, |
| 2409 |
'field_value' => $field_value, |
| 2410 |
'post_id' => $post_id |
| 2411 |
]) |
| 2412 |
]); |
| 2413 |
} |
| 2414 |
} |
| 2415 |
|
| 2416 |
return Metasync_Otto_Job_Status::OUTCOME_SUCCESS; |
| 2417 |
} |
| 2418 |
|
| 2419 |
return Metasync_Otto_Job_Status::OUTCOME_NO_CHANGE; |
| 2420 |
|
| 2421 |
} catch (Exception $e) { |
| 2422 |
metasync_record_failed_action( 'metasync_process_seo_job' ); |
| 2423 |
return Metasync_Otto_Job_Status::OUTCOME_RETRYABLE; |
| 2424 |
} finally { |
| 2425 |
# Release the concurrency lock only if we actually acquired it. |
| 2426 |
# Early returns (CPU deferral, lock contention) must NOT delete a lock |
| 2427 |
# that another process may be holding. |
| 2428 |
if ($lock_acquired && $lock_key) { |
| 2429 |
delete_transient($lock_key); |
| 2430 |
} |
| 2431 |
} |
| 2432 |
} |
| 2433 |
|
| 2434 |
/** |
| 2435 |
* Log sync history entry |
| 2436 |
* @param array $data Sync data to log |
| 2437 |
*/ |
| 2438 |
function metasync_log_sync_history($data) { |
| 2439 |
try { |
| 2440 |
// Classes are now autoloaded, no need for manual require |
| 2441 |
$sync_db = new Metasync_Sync_History_Database(); |
| 2442 |
|
| 2443 |
// Minimal duplicate prevention within short time window |
| 2444 |
if (!empty($data['title']) && !empty($data['source'])) { |
| 2445 |
global $wpdb; |
| 2446 |
$table = $wpdb->prefix . Metasync_Sync_History_Database::$table_name; |
| 2447 |
$recent = $wpdb->get_var($wpdb->prepare( |
| 2448 |
"SELECT COUNT(*) FROM `$table` WHERE title = %s AND source = %s AND created_at >= %s", |
| 2449 |
$data['title'], |
| 2450 |
$data['source'], |
| 2451 |
gmdate('Y-m-d H:i:s', time() - 60) |
| 2452 |
)); |
| 2453 |
if ((int)$recent > 0) { |
| 2454 |
return; // skip duplicate log within 60 seconds |
| 2455 |
} |
| 2456 |
} |
| 2457 |
|
| 2458 |
$sync_db->add($data); |
| 2459 |
|
| 2460 |
} catch (Exception $e) { |
| 2461 |
error_log("MetaSync: Failed to log sync history: " . $e->getMessage()); |
| 2462 |
} |
| 2463 |
} |
| 2464 |
|
| 2465 |
/** |
| 2466 |
* Get supported post types for OTTO SEO optimization |
| 2467 |
* Includes WooCommerce products if WooCommerce is active |
| 2468 |
* |
| 2469 |
* @return array List of supported post types |
| 2470 |
*/ |
| 2471 |
function metasync_get_supported_post_types() { |
| 2472 |
# Start with default post types |
| 2473 |
$post_types = ['post', 'page']; |
| 2474 |
|
| 2475 |
# Add WooCommerce product post type if WooCommerce is active |
| 2476 |
if (class_exists('WooCommerce') || function_exists('is_woocommerce')) { |
| 2477 |
$post_types[] = 'product'; |
| 2478 |
} |
| 2479 |
|
| 2480 |
# Include all public custom post types (e.g. 'location', 'service', 'team', etc.) |
| 2481 |
# so OTTO can write post meta for them during metasync_process_otto_seo_data(). |
| 2482 |
$custom_post_types = get_post_types(['public' => true, '_builtin' => false], 'names'); |
| 2483 |
if (!empty($custom_post_types)) { |
| 2484 |
$post_types = array_merge($post_types, array_values($custom_post_types)); |
| 2485 |
} |
| 2486 |
|
| 2487 |
# Allow developers to filter supported post types |
| 2488 |
$post_types = apply_filters('metasync_otto_supported_post_types', $post_types); |
| 2489 |
|
| 2490 |
return $post_types; |
| 2491 |
} |
| 2492 |
|
| 2493 |
/** |
| 2494 |
* Get supported taxonomies for OTTO SEO optimization |
| 2495 |
* Includes WooCommerce product categories and tags if WooCommerce is active |
| 2496 |
* |
| 2497 |
* @return array List of supported taxonomies |
| 2498 |
*/ |
| 2499 |
function metasync_get_supported_taxonomies() { |
| 2500 |
# Start with default taxonomies |
| 2501 |
$taxonomies = ['category']; |
| 2502 |
|
| 2503 |
# Add WooCommerce taxonomies if WooCommerce is active |
| 2504 |
if (class_exists('WooCommerce') || function_exists('is_woocommerce')) { |
| 2505 |
$taxonomies[] = 'product_cat'; # WooCommerce product categories |
| 2506 |
$taxonomies[] = 'product_tag'; # WooCommerce product tags |
| 2507 |
} |
| 2508 |
|
| 2509 |
# Allow developers to filter supported taxonomies |
| 2510 |
$taxonomies = apply_filters('metasync_otto_supported_taxonomies', $taxonomies); |
| 2511 |
|
| 2512 |
return $taxonomies; |
| 2513 |
} |
| 2514 |
|
| 2515 |
/** |
| 2516 |
* Resolve a URL through the Redirect Manager table to its final destination (follows redirect chains). |
| 2517 |
* Used before 404 checks and OTTO processing so the final canonical URL is used, not intermediate redirects. |
| 2518 |
* |
| 2519 |
* @param string $url Full URL (e.g. https://example.com/old-page) |
| 2520 |
* @return string Final destination URL, or original $url if no redirect matches |
| 2521 |
*/ |
| 2522 |
function metasync_otto_resolve_redirect_to_final_url($url) |
| 2523 |
{ |
| 2524 |
if (empty($url) || !is_string($url)) { |
| 2525 |
return $url; |
| 2526 |
} |
| 2527 |
try { |
| 2528 |
$db_path = plugin_dir_path(dirname(__FILE__)) . 'redirections/class-metasync-redirection-database.php'; |
| 2529 |
$class_path = plugin_dir_path(dirname(__FILE__)) . 'redirections/class-metasync-redirection.php'; |
| 2530 |
if (!file_exists($db_path) || !file_exists($class_path)) { |
| 2531 |
return $url; |
| 2532 |
} |
| 2533 |
require_once $db_path; |
| 2534 |
require_once $class_path; |
| 2535 |
$db = new Metasync_Redirection_Database(); |
| 2536 |
$redirect = new Metasync_Redirection($db); |
| 2537 |
return $redirect->resolve_url_to_final_destination($url, 10); |
| 2538 |
} catch (Exception $e) { |
| 2539 |
error_log('MetaSync OTTO: Redirect resolution failed for ' . $url . ' - ' . $e->getMessage()); |
| 2540 |
return $url; |
| 2541 |
} |
| 2542 |
} |
| 2543 |
|
| 2544 |
/** |
| 2545 |
* Auto-exclude a URL from OTTO with description "Auto-excluded: 404" |
| 2546 |
* Called when a URL is detected as returning 404 so it won't be sent to OTTO again |
| 2547 |
* |
| 2548 |
* @param string $url Full URL to exclude (e.g. https://example.com/404-page) |
| 2549 |
* @return bool|string True on success, false on failure, 'duplicate'/'reactivated' if already exists |
| 2550 |
*/ |
| 2551 |
function metasync_otto_auto_exclude_404_url($url) |
| 2552 |
{ |
| 2553 |
if (empty($url) || !is_string($url)) { |
| 2554 |
return false; |
| 2555 |
} |
| 2556 |
$url = filter_var($url, FILTER_SANITIZE_URL); |
| 2557 |
$url = esc_url_raw($url); |
| 2558 |
if (empty($url) || mb_strlen($url) > 2048) { |
| 2559 |
return false; |
| 2560 |
} |
| 2561 |
try { |
| 2562 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 2563 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 2564 |
return $db->add([ |
| 2565 |
'url_pattern' => $url, |
| 2566 |
'pattern_type' => 'exact', |
| 2567 |
'description' => 'Auto-excluded: 404', |
| 2568 |
'status' => 'active', |
| 2569 |
'auto_excluded' => 1, |
| 2570 |
]); |
| 2571 |
} catch (Exception $e) { |
| 2572 |
error_log('MetaSync OTTO: Failed to auto-exclude 404 URL: ' . $url . ' - ' . $e->getMessage()); |
| 2573 |
return false; |
| 2574 |
} |
| 2575 |
} |
| 2576 |
|
| 2577 |
/** |
| 2578 |
* Remove a URL from the OTTO auto-exclusion list. |
| 2579 |
* Called when OTTO sends a webhook for a URL, confirming it is valid and crawlable. |
| 2580 |
* Only removes records where auto_excluded = 1 (never removes manual exclusions). |
| 2581 |
* |
| 2582 |
* @param string $url Full URL to un-exclude (e.g. https://example.com/location/page) |
| 2583 |
* @return bool True on success |
| 2584 |
*/ |
| 2585 |
function metasync_otto_remove_auto_exclusion($url) |
| 2586 |
{ |
| 2587 |
if (empty($url) || !is_string($url)) { |
| 2588 |
return false; |
| 2589 |
} |
| 2590 |
try { |
| 2591 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 2592 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 2593 |
global $wpdb; |
| 2594 |
$table = $wpdb->prefix . Metasync_Otto_Excluded_URLs_Database::$table_name; |
| 2595 |
# Normalize the same way is_url_excluded() does |
| 2596 |
$url_normalized = rtrim(trim($url), '/'); |
| 2597 |
$records = $wpdb->get_results( |
| 2598 |
$wpdb->prepare( |
| 2599 |
"SELECT id FROM `{$table}` WHERE url_pattern = %s AND auto_excluded = 1 AND status = 'active'", |
| 2600 |
$url_normalized |
| 2601 |
) |
| 2602 |
); |
| 2603 |
if (!empty($records)) { |
| 2604 |
$ids = array_map(function ($r) { return (int) $r->id; }, $records); |
| 2605 |
$db->delete($ids); |
| 2606 |
} |
| 2607 |
return true; |
| 2608 |
} catch (Exception $e) { |
| 2609 |
return false; |
| 2610 |
} |
| 2611 |
} |
| 2612 |
|
| 2613 |
/** |
| 2614 |
* Check if a URL is MANUALLY excluded from OTTO (auto_excluded = 0). |
| 2615 |
* Used at render time (metasync_start_otto) — auto-exclusions must NOT block |
| 2616 |
* rendering because they are often false positives (e.g. custom post types that |
| 2617 |
* url_to_postid() can't resolve). Auto-exclusions are only used to gate the |
| 2618 |
* SEO meta-writing webhook path. |
| 2619 |
* |
| 2620 |
* @param string $url URL to check |
| 2621 |
* @return bool True if URL has a manual exclusion |
| 2622 |
*/ |
| 2623 |
function metasync_is_otto_url_manually_excluded($url) |
| 2624 |
{ |
| 2625 |
if (empty($url) || !is_string($url)) { |
| 2626 |
return false; |
| 2627 |
} |
| 2628 |
try { |
| 2629 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 2630 |
global $wpdb; |
| 2631 |
$table = $wpdb->prefix . Metasync_Otto_Excluded_URLs_Database::$table_name; |
| 2632 |
$url_normalized = rtrim(trim($url), '/'); |
| 2633 |
|
| 2634 |
$records = get_transient(METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY); |
| 2635 |
|
| 2636 |
if ($records === false) { |
| 2637 |
// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- no user input, table name from $wpdb->prefix |
| 2638 |
$records = $wpdb->get_results( |
| 2639 |
"SELECT url_pattern, pattern_type FROM `{$table}` WHERE status = 'active' AND (auto_excluded = 0 OR auto_excluded IS NULL) ORDER BY created_at DESC" |
| 2640 |
); |
| 2641 |
|
| 2642 |
// Graceful recovery: auto_excluded column missing on pre-v2.7.4 installs. |
| 2643 |
// Run ALTER TABLE to add it and treat URL as not excluded so OTTO continues rendering. |
| 2644 |
if ($wpdb->last_error && strpos($wpdb->last_error, 'auto_excluded') !== false) { |
| 2645 |
// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared |
| 2646 |
$wpdb->query("ALTER TABLE `{$table}` ADD COLUMN `auto_excluded` TINYINT(1) NOT NULL DEFAULT 0"); |
| 2647 |
return false; |
| 2648 |
} |
| 2649 |
|
| 2650 |
set_transient(METASYNC_OTTO_EXCLUDED_TRANSIENT_KEY, $records ?: [], METASYNC_OTTO_EXCLUDED_TRANSIENT_TTL); |
| 2651 |
} |
| 2652 |
|
| 2653 |
if (empty($records)) { |
| 2654 |
return false; |
| 2655 |
} |
| 2656 |
|
| 2657 |
foreach ($records as $excluded) { |
| 2658 |
# Delegate to the shared matcher so this gate supports every pattern type the |
| 2659 |
# add handler accepts (exact|contain|start|end|regex). Keeping a second switch |
| 2660 |
# here is what let `end` and `regex` exclusions be silently ignored on the |
| 2661 |
# visitor render path while still applying on the queue/SEO-write path. |
| 2662 |
# The raw pattern is passed through - the matcher normalizes it per type, |
| 2663 |
# because a regex must keep the trailing slash that closes its delimiters. |
| 2664 |
if (Metasync_Otto_Excluded_URLs_Database::pattern_matches($url_normalized, $excluded->url_pattern, $excluded->pattern_type)) { |
| 2665 |
return true; |
| 2666 |
} |
| 2667 |
} |
| 2668 |
return false; |
| 2669 |
} catch (Exception $e) { |
| 2670 |
return false; |
| 2671 |
} |
| 2672 |
} |
| 2673 |
|
| 2674 |
/** |
| 2675 |
* Check if a URL is excluded from OTTO |
| 2676 |
* @param string $url URL to check |
| 2677 |
* @return bool True if URL is excluded, false otherwise |
| 2678 |
*/ |
| 2679 |
function metasync_is_otto_url_excluded($url) |
| 2680 |
{ |
| 2681 |
try { |
| 2682 |
// Load database class |
| 2683 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 2684 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 2685 |
|
| 2686 |
// Check if URL is excluded |
| 2687 |
return $db->is_url_excluded($url); |
| 2688 |
|
| 2689 |
} catch (Exception $e) { |
| 2690 |
return false; |
| 2691 |
} |
| 2692 |
} |
| 2693 |
|
| 2694 |
/** |
| 2695 |
* Check if a URL is now available (would NOT return 404) |
| 2696 |
* Uses same resolution logic as metasync_process_otto_seo_data |
| 2697 |
* Used when rechecking auto-excluded 404 URLs after 7 days |
| 2698 |
* |
| 2699 |
* @param string $url Full URL to check (e.g. https://example.com/page) |
| 2700 |
* @return bool True if URL is accessible, false if it would return 404 |
| 2701 |
*/ |
| 2702 |
function metasync_otto_is_url_available($url) |
| 2703 |
{ |
| 2704 |
if (empty($url) || !is_string($url)) { |
| 2705 |
return false; |
| 2706 |
} |
| 2707 |
|
| 2708 |
$route = metasync_otto_resolve_redirect_to_final_url($url); |
| 2709 |
$post_id = url_to_postid($route); |
| 2710 |
|
| 2711 |
# WooCommerce shop page |
| 2712 |
if ((!$post_id || $post_id <= 0) && function_exists('wc_get_page_id')) { |
| 2713 |
$shop_page_id = wc_get_page_id('shop'); |
| 2714 |
if ($shop_page_id > 0) { |
| 2715 |
$shop_url = get_permalink($shop_page_id); |
| 2716 |
if (rtrim($route, '/') === rtrim($shop_url, '/')) { |
| 2717 |
$post_id = $shop_page_id; |
| 2718 |
} |
| 2719 |
} |
| 2720 |
} |
| 2721 |
|
| 2722 |
# WooCommerce product by slug |
| 2723 |
if ((!$post_id || $post_id <= 0) && strpos($route, '/product/') !== false && function_exists('wc_get_products')) { |
| 2724 |
$product_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 2725 |
$products = wc_get_products(array('name' => $product_slug, 'limit' => 1, 'status' => 'publish')); |
| 2726 |
if (!empty($products)) { |
| 2727 |
$post_id = $products[0]->get_id(); |
| 2728 |
} else { |
| 2729 |
$query = new WP_Query(array( |
| 2730 |
'post_type' => 'product', |
| 2731 |
'name' => $product_slug, |
| 2732 |
'posts_per_page' => 1, |
| 2733 |
'post_status' => 'publish', |
| 2734 |
)); |
| 2735 |
if ($query->have_posts()) { |
| 2736 |
$post_id = $query->posts[0]->ID; |
| 2737 |
} |
| 2738 |
} |
| 2739 |
} |
| 2740 |
|
| 2741 |
if ($post_id && $post_id > 0) { |
| 2742 |
$post = get_post($post_id); |
| 2743 |
# Accept any post type — custom post types (e.g. 'location', 'service') are valid URLs. |
| 2744 |
# The old in_array check against metasync_get_supported_post_types() caused CPT URLs |
| 2745 |
# to be wrongly auto-excluded as "404" pages. |
| 2746 |
if ($post) { |
| 2747 |
return !metasync_would_page_return_404($post_id, $route); |
| 2748 |
} |
| 2749 |
} |
| 2750 |
|
| 2751 |
# Category |
| 2752 |
if (strpos($route, '/category/') !== false) { |
| 2753 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 2754 |
$category = get_category_by_slug($category_slug); |
| 2755 |
if ($category) { |
| 2756 |
return !metasync_would_term_return_404($category->term_id, 'category', $route); |
| 2757 |
} |
| 2758 |
} |
| 2759 |
|
| 2760 |
# WooCommerce product category |
| 2761 |
if (strpos($route, '/product-category/') !== false) { |
| 2762 |
$category_slug = basename(parse_url($route, PHP_URL_PATH)); |
| 2763 |
$term = get_term_by('slug', $category_slug, 'product_cat'); |
| 2764 |
if ($term && !is_wp_error($term)) { |
| 2765 |
return !metasync_would_term_return_404($term->term_id, 'product_cat', $route); |
| 2766 |
} |
| 2767 |
} |
| 2768 |
|
| 2769 |
# Home page |
| 2770 |
if (rtrim($route, '/') === rtrim(site_url(), '/')) { |
| 2771 |
$front_page_id = get_option('page_on_front'); |
| 2772 |
$home_page = ($front_page_id && $front_page_id > 0) |
| 2773 |
? get_post($front_page_id) |
| 2774 |
: (get_posts(['numberposts' => 1, 'post_status' => 'publish'])[0] ?? null); |
| 2775 |
if ($home_page) { |
| 2776 |
return !metasync_would_page_return_404($home_page->ID, $route); |
| 2777 |
} |
| 2778 |
} |
| 2779 |
|
| 2780 |
# Could not verify availability from local data (custom archive, paginated page, etc.). |
| 2781 |
# Assume the URL IS available — OTTO only crawls reachable URLs, so if we can't |
| 2782 |
# prove it's a 404, we should not auto-exclude it. |
| 2783 |
return true; |
| 2784 |
} |
| 2785 |
|
| 2786 |
/** |
| 2787 |
* Recheck auto-excluded 404 URLs when recheck_after has passed; remove from exclusion if now available |
| 2788 |
* Uses recheck_after timestamp (default 7 days from exclusion) to decide when to recheck |
| 2789 |
* Mark as permanent after 30 days if still 404 (no further rechecks) |
| 2790 |
* Called by daily cron job |
| 2791 |
*/ |
| 2792 |
function metasync_otto_recheck_404_exclusions() |
| 2793 |
{ |
| 2794 |
try { |
| 2795 |
require_once plugin_dir_path(__FILE__) . 'class-metasync-otto-excluded-urls-database.php'; |
| 2796 |
$db = new Metasync_Otto_Excluded_URLs_Database(); |
| 2797 |
$records = $db->get_auto_excluded_404_urls_due_for_recheck(); |
| 2798 |
|
| 2799 |
if (empty($records)) { |
| 2800 |
return; |
| 2801 |
} |
| 2802 |
|
| 2803 |
$removed = 0; |
| 2804 |
$marked_permanent = 0; |
| 2805 |
$thirty_days_ago = strtotime('-30 days'); |
| 2806 |
$next_recheck = date('Y-m-d H:i:s', current_time('timestamp') + 7 * DAY_IN_SECONDS); |
| 2807 |
|
| 2808 |
foreach ($records as $record) { |
| 2809 |
$url = trim($record->url_pattern); |
| 2810 |
if (empty($url)) { |
| 2811 |
continue; |
| 2812 |
} |
| 2813 |
if (metasync_otto_is_url_available($url)) { |
| 2814 |
$db->delete([$record->id]); |
| 2815 |
$removed++; |
| 2816 |
} else { |
| 2817 |
# Still 404: if excluded 30+ days ago, mark as permanent (no more rechecks) |
| 2818 |
$created_ts = strtotime($record->created_at); |
| 2819 |
if ($created_ts <= $thirty_days_ago) { |
| 2820 |
$db->update(['is_permanent' => 1], $record->id); |
| 2821 |
$marked_permanent++; |
| 2822 |
} else { |
| 2823 |
# Schedule next recheck in 7 days |
| 2824 |
$db->update(['recheck_after' => $next_recheck], $record->id); |
| 2825 |
} |
| 2826 |
} |
| 2827 |
} |
| 2828 |
|
| 2829 |
if ($removed > 0) { |
| 2830 |
error_log("MetaSync OTTO: Recheck 404 exclusions - removed {$removed} URL(s) that are now available"); |
| 2831 |
} |
| 2832 |
if ($marked_permanent > 0) { |
| 2833 |
error_log("MetaSync OTTO: Recheck 404 exclusions - marked {$marked_permanent} URL(s) as permanent (still 404 after 30 days)"); |
| 2834 |
} |
| 2835 |
} catch (Exception $e) { |
| 2836 |
error_log('MetaSync OTTO: Recheck 404 exclusions failed - ' . $e->getMessage()); |
| 2837 |
} |
| 2838 |
} |
| 2839 |
|
| 2840 |
add_action('metasync_otto_recheck_404_exclusions', 'metasync_otto_recheck_404_exclusions'); |
| 2841 |
|
| 2842 |
/** |
| 2843 |
* Check if a post/page would return 404 without making HTTP request |
| 2844 |
* Uses WordPress database checks for fast validation |
| 2845 |
* |
| 2846 |
* @param int $post_id WordPress post ID |
| 2847 |
* @param string $url The URL being checked (optional, for logging) |
| 2848 |
* @return bool True if page would return 404, false if accessible |
| 2849 |
*/ |
| 2850 |
function metasync_would_page_return_404($post_id, $url = '') { |
| 2851 |
if (!$post_id || $post_id <= 0) { |
| 2852 |
return true; // No post ID = 404 |
| 2853 |
} |
| 2854 |
|
| 2855 |
# Get the post object |
| 2856 |
$post = get_post($post_id); |
| 2857 |
if (!$post) { |
| 2858 |
return true; // Post doesn't exist = 404 |
| 2859 |
} |
| 2860 |
|
| 2861 |
# 1. Check post status - must be 'publish' to be publicly accessible |
| 2862 |
if ($post->post_status !== 'publish') { |
| 2863 |
return true; // Draft, pending, private, etc. = 404 |
| 2864 |
} |
| 2865 |
|
| 2866 |
# 2. Check if post is password protected (requires password to view) |
| 2867 |
if (!empty($post->post_password)) { |
| 2868 |
# Password protected posts are not publicly accessible without password |
| 2869 |
return true; // Password protected = effectively 404 for public |
| 2870 |
} |
| 2871 |
|
| 2872 |
# 3. Check if post is in trash |
| 2873 |
if ($post->post_status === 'trash') { |
| 2874 |
return true; // Trashed = 404 |
| 2875 |
} |
| 2876 |
|
| 2877 |
# 4. Check if post type is publicly queryable |
| 2878 |
$post_type_object = get_post_type_object($post->post_type); |
| 2879 |
if ($post_type_object && !$post_type_object->publicly_queryable) { |
| 2880 |
# Some post types might not be publicly accessible |
| 2881 |
# But we allow if it's in our supported types |
| 2882 |
$supported_post_types = metasync_get_supported_post_types(); |
| 2883 |
if (!in_array($post->post_type, $supported_post_types)) { |
| 2884 |
return true; // Not publicly queryable = 404 |
| 2885 |
} |
| 2886 |
} |
| 2887 |
|
| 2888 |
# 5. WordPress 5.7+ has a built-in function for this |
| 2889 |
if (function_exists('is_post_publicly_viewable')) { |
| 2890 |
if (!is_post_publicly_viewable($post)) { |
| 2891 |
return true; // Not publicly viewable = 404 |
| 2892 |
} |
| 2893 |
} |
| 2894 |
|
| 2895 |
# 6. Check if post is scheduled for future (not yet published) |
| 2896 |
if ($post->post_date > current_time('mysql')) { |
| 2897 |
return true; // Future post = 404 until publish date |
| 2898 |
} |
| 2899 |
|
| 2900 |
# All checks passed - page should be accessible |
| 2901 |
return false; |
| 2902 |
} |
| 2903 |
|
| 2904 |
/** |
| 2905 |
* Check if a taxonomy term (category, tag, etc.) would return 404 |
| 2906 |
* Uses WordPress database checks for fast validation |
| 2907 |
* |
| 2908 |
* @param int $term_id Term ID |
| 2909 |
* @param string $taxonomy Taxonomy name (e.g., 'category', 'product_cat') |
| 2910 |
* @param string $url The URL being checked (optional, for logging) |
| 2911 |
* @return bool True if term would return 404, false if accessible |
| 2912 |
*/ |
| 2913 |
function metasync_would_term_return_404($term_id, $taxonomy, $url = '') { |
| 2914 |
if (!$term_id || $term_id <= 0 || empty($taxonomy)) { |
| 2915 |
return true; // Invalid term = 404 |
| 2916 |
} |
| 2917 |
|
| 2918 |
# Get the term object |
| 2919 |
$term = get_term($term_id, $taxonomy); |
| 2920 |
if (is_wp_error($term) || !$term) { |
| 2921 |
return true; // Term doesn't exist = 404 |
| 2922 |
} |
| 2923 |
|
| 2924 |
# Check if taxonomy is publicly queryable |
| 2925 |
$taxonomy_object = get_taxonomy($taxonomy); |
| 2926 |
if (!$taxonomy_object || !$taxonomy_object->public) { |
| 2927 |
# Check if it's in our supported taxonomies |
| 2928 |
$supported_taxonomies = metasync_get_supported_taxonomies(); |
| 2929 |
if (!in_array($taxonomy, $supported_taxonomies)) { |
| 2930 |
return true; // Not publicly queryable = 404 |
| 2931 |
} |
| 2932 |
} |
| 2933 |
|
| 2934 |
# Terms are generally always accessible if they exist and taxonomy is public |
| 2935 |
# WordPress doesn't have a "draft" status for terms like posts do |
| 2936 |
# But we can check if the term has a count (has posts assigned) |
| 2937 |
# Empty terms might not be useful, but they're still accessible |
| 2938 |
|
| 2939 |
# All checks passed - term should be accessible |
| 2940 |
return false; |
| 2941 |
} |
| 2942 |
|
| 2943 |
/** |
| 2944 |
* Invalidate Brizy posts cache when posts are saved |
| 2945 |
* OPTIMIZATION: Clears transient cache to ensure accurate detection |
| 2946 |
*/ |
| 2947 |
add_action('save_post', function($post_id) { |
| 2948 |
# Check if this post has Brizy metadata |
| 2949 |
if (get_post_meta($post_id, 'brizy_post_uid', true)) { |
| 2950 |
delete_transient('metasync_has_brizy_posts'); |
| 2951 |
} |
| 2952 |
}, 10, 1); |
| 2953 |
|
| 2954 |
/** |
| 2955 |
* Identify infrastructure agents that must never hit the render throttle. |
| 2956 |
* |
| 2957 |
* Two classes of agent are covered: |
| 2958 |
* |
| 2959 |
* - Page-cache warmers. A host preloader (WP Cloud PageCacheBot, WP Rocket preload, |
| 2960 |
* LiteSpeed, SG Optimizer, ...) exists purely to populate the page cache. Throttling |
| 2961 |
* one returns un-OTTO'd HTML *and* defines DONOTCACHEPAGE, so the cache can never be |
| 2962 |
* filled and every subsequent human visitor pays a full uncached render. Measured on |
| 2963 |
* the reporting site: desktop cache MISS on 17/17 requests at ~2.8s TTFB, versus |
| 2964 |
* ~0.56s with OTTO disabled. |
| 2965 |
* |
| 2966 |
* - Synthetic performance auditors (Lighthouse / PageSpeed Insights / GTmetrix). |
| 2967 |
* These are what customers measure with, so they must receive the same fully |
| 2968 |
* rendered OTTO output a real visitor gets. Currently `/lighthouse/i` and |
| 2969 |
* `/pagespeed/i` are generic-bot patterns, so a second PSI run inside the 5-minute |
| 2970 |
* window is served the throttled, un-OTTO'd, uncacheable page. |
| 2971 |
* |
| 2972 |
* Matching is done on the raw user-agent, deliberately independent of the bot-name |
| 2973 |
* categorisation in Metasync_Otto_Bot_Detector. |
| 2974 |
* |
| 2975 |
* @param mixed $detection Result of Metasync_Otto_Bot_Detector::detect(), or malformed input. |
| 2976 |
* @return bool True when this request must bypass the render throttle. |
| 2977 |
*/ |
| 2978 |
function metasync_otto_is_unthrottled_infrastructure_agent( $detection = array() ) { |
| 2979 |
|
| 2980 |
$ua = ''; |
| 2981 |
if ( is_array( $detection ) && ! empty( $detection['user_agent'] ) ) { |
| 2982 |
$ua = (string) $detection['user_agent']; |
| 2983 |
} elseif ( isset( $_SERVER['HTTP_USER_AGENT'] ) ) { |
| 2984 |
$ua = (string) $_SERVER['HTTP_USER_AGENT']; |
| 2985 |
} |
| 2986 |
|
| 2987 |
$needles = array( |
| 2988 |
// Page-cache preloaders. |
| 2989 |
'pagecachebot', // WP Cloud / Automattic (PageCacheBotDesktop|Mobile) |
| 2990 |
'wp rocket', // WP Rocket preload |
| 2991 |
'wprocket', |
| 2992 |
'lscache_runner', // LiteSpeed Cache crawler |
| 2993 |
'sg-optimizer', // SiteGround |
| 2994 |
'sgoptimizer', |
| 2995 |
'nitropack', |
| 2996 |
'cache-warmer', |
| 2997 |
'cachewarmer', |
| 2998 |
'cache warmer', |
| 2999 |
// Synthetic performance auditors. |
| 3000 |
'lighthouse', |
| 3001 |
'pagespeed', |
| 3002 |
'gtmetrix', |
| 3003 |
'pingdom', |
| 3004 |
'webpagetest', |
| 3005 |
); |
| 3006 |
|
| 3007 |
$is_infra = false; |
| 3008 |
|
| 3009 |
if ( '' !== $ua ) { |
| 3010 |
$ua_lower = strtolower( $ua ); |
| 3011 |
foreach ( $needles as $needle ) { |
| 3012 |
if ( false !== strpos( $ua_lower, $needle ) ) { |
| 3013 |
$is_infra = true; |
| 3014 |
break; |
| 3015 |
} |
| 3016 |
} |
| 3017 |
} |
| 3018 |
|
| 3019 |
// WP Cloud's warmer also tags its requests with a query marker. |
| 3020 |
if ( ! $is_infra && isset( $_GET['x-cache-engine'] ) ) { |
| 3021 |
$is_infra = true; |
| 3022 |
} |
| 3023 |
|
| 3024 |
/** |
| 3025 |
* Allow a site or host to declare additional preload agents. |
| 3026 |
* |
| 3027 |
* @param bool $is_infra Whether this request bypasses the OTTO render throttle. |
| 3028 |
* @param string $ua The request user-agent. |
| 3029 |
*/ |
| 3030 |
if ( function_exists( 'apply_filters' ) ) { |
| 3031 |
$is_infra = (bool) apply_filters( 'metasync_otto_unthrottled_agent', $is_infra, $ua ); |
| 3032 |
} |
| 3033 |
|
| 3034 |
return $is_infra; |
| 3035 |
} |
| 3036 |
|
