| 1 |
<?php |
| 2 |
if (!defined('ABSPATH')) die('No direct access allowed'); |
| 3 |
|
| 4 |
if (!class_exists('WP_Optimize_404_Detector')) : |
| 5 |
|
| 6 |
class WP_Optimize_404_Detector { |
| 7 |
|
| 8 |
/** |
| 9 |
* Suspicious requests threshold |
| 10 |
* |
| 11 |
* @var int |
| 12 |
*/ |
| 13 |
private $suspicious_request_count_threshold = 50; |
| 14 |
|
| 15 |
/** |
| 16 |
* Remove trivial requests older than hours |
| 17 |
* |
| 18 |
* @var int |
| 19 |
*/ |
| 20 |
private $suspicious_trivial_request_ttl_in_hours = 24; |
| 21 |
|
| 22 |
/** |
| 23 |
* Remove suspicious requests older than hours |
| 24 |
* |
| 25 |
* @var int |
| 26 |
*/ |
| 27 |
private $suspicious_request_ttl_in_hours = 672; |
| 28 |
|
| 29 |
/** |
| 30 |
* How many suspicious requests count in total to show in the dashboard? |
| 31 |
* |
| 32 |
* @var int |
| 33 |
*/ |
| 34 |
private $dashboard_alert_request_count_threshold = 100; |
| 35 |
|
| 36 |
/** |
| 37 |
* Store the total count for each url, to be able to sort it later |
| 38 |
* |
| 39 |
* @var array |
| 40 |
*/ |
| 41 |
private $total_count_per_url = array(); |
| 42 |
|
| 43 |
/** |
| 44 |
* Class constructor |
| 45 |
*/ |
| 46 |
private function __construct() { |
| 47 |
add_action('wpo_prune_404_log', array($this, 'prune_404_log')); |
| 48 |
} |
| 49 |
|
| 50 |
/** |
| 51 |
* Initialize the class as a singleton |
| 52 |
* |
| 53 |
* @return WP_Optimize_404_Detector |
| 54 |
*/ |
| 55 |
public static function get_instance() { |
| 56 |
static $_instance = null; |
| 57 |
if (null === $_instance) { |
| 58 |
$_instance = new self(); |
| 59 |
} |
| 60 |
return $_instance; |
| 61 |
} |
| 62 |
|
| 63 |
/** |
| 64 |
* Handle 404 requests |
| 65 |
* |
| 66 |
* @return void |
| 67 |
*/ |
| 68 |
public function handle_request() { |
| 69 |
$now = time(); |
| 70 |
$request_timestamp = $now - ($now % 3600); |
| 71 |
|
| 72 |
$url_data = isset($_SERVER['REQUEST_URI']) ? $this->parse_url(esc_url_raw(wp_unslash($_SERVER['REQUEST_URI']))) : false; |
| 73 |
|
| 74 |
if (!$url_data || !isset($url_data['path'])) { |
| 75 |
return; |
| 76 |
} |
| 77 |
|
| 78 |
$is_plain_structure = $this->is_plain_permalink_structure(); |
| 79 |
|
| 80 |
if (($is_plain_structure && empty($url_data['query'])) || (!$is_plain_structure && '/' === $url_data['path'])) { |
| 81 |
return; |
| 82 |
} |
| 83 |
|
| 84 |
$url = $is_plain_structure ? home_url('?' . $url_data['query']) : home_url($url_data['path']); |
| 85 |
|
| 86 |
$this->save_request_hour_row($request_timestamp, $url); |
| 87 |
} |
| 88 |
|
| 89 |
/** |
| 90 |
* Log an entry for 404 errors |
| 91 |
* |
| 92 |
* @param string $request_timestamp The moment the request is happening |
| 93 |
* @param string $url Relative url to the URL being analyzed |
| 94 |
* @return void |
| 95 |
*/ |
| 96 |
private function save_request_hour_row($request_timestamp, $url) { |
| 97 |
global $wpdb; |
| 98 |
|
| 99 |
$log_table_name = esc_sql($this->get_table_name()); |
| 100 |
|
| 101 |
$referrer = isset($_SERVER['HTTP_REFERER']) ? esc_url_raw(wp_unslash($_SERVER['HTTP_REFERER'])) : ""; |
| 102 |
|
| 103 |
$safe_referrer = ''; |
| 104 |
if ('' !== $referrer) { |
| 105 |
$referrer_parsed = $this->parse_url($referrer); |
| 106 |
$safe_referrer = (isset($referrer_parsed['scheme']) ? $referrer_parsed['scheme'] . '://' : '') . |
| 107 |
($referrer_parsed['host'] ?? '') . |
| 108 |
(isset($referrer_parsed['port']) ? ':' . $referrer_parsed['port'] : '') . |
| 109 |
($referrer_parsed['path'] ?? '') . |
| 110 |
(isset($referrer_parsed['query']) ? '?' . $referrer_parsed['query'] : ''); |
| 111 |
} |
| 112 |
|
| 113 |
$wpdb->query($wpdb->prepare("INSERT INTO `{$log_table_name}` SET `url` = %s, request_timestamp = %d, referrer = %s, request_count = 1 ON DUPLICATE KEY UPDATE request_count = request_count + 1", $url, $request_timestamp, $safe_referrer)); // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 114 |
} |
| 115 |
|
| 116 |
/** |
| 117 |
* Remove trivial requests older than 24 hours, and suspicious requests after 4 weeks |
| 118 |
* |
| 119 |
* @return void |
| 120 |
*/ |
| 121 |
public function prune_404_log() { |
| 122 |
global $wpdb; |
| 123 |
|
| 124 |
$log_table_name = esc_sql($this->get_table_name()); |
| 125 |
|
| 126 |
// Remove old trivial requests |
| 127 |
$hs_to_remove_older = $this->suspicious_trivial_request_ttl_in_hours * 3600; |
| 128 |
$remove_date = time() - $hs_to_remove_older; |
| 129 |
|
| 130 |
|
| 131 |
$wpdb->query($wpdb->prepare("DELETE FROM `$log_table_name` WHERE request_timestamp < %d AND request_count < %d", $remove_date, $this->suspicious_request_count_threshold)); // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 132 |
|
| 133 |
// Remove any type of old request |
| 134 |
$hs_to_remove_older = $this->suspicious_request_ttl_in_hours * 3600; |
| 135 |
$remove_date = time() - $hs_to_remove_older; |
| 136 |
|
| 137 |
$wpdb->query($wpdb->prepare("DELETE FROM `$log_table_name` WHERE request_timestamp < %d", $remove_date)); // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 138 |
} |
| 139 |
|
| 140 |
/** |
| 141 |
* Get suspicious requests from DB grouped by url and referrer |
| 142 |
* |
| 143 |
* @return array |
| 144 |
*/ |
| 145 |
public function get_suspicious_requests() { |
| 146 |
$all_suspicious_referrers = array(); |
| 147 |
$by_referrer = $this->get_single_suspicious_requests_by_referer($all_suspicious_referrers); |
| 148 |
|
| 149 |
$by_url = $this->get_grouped_requests_by_url($all_suspicious_referrers); |
| 150 |
|
| 151 |
$result = array_merge($by_url, $by_referrer); |
| 152 |
|
| 153 |
usort($result, array($this, 'sort_suspicious_requests')); |
| 154 |
|
| 155 |
$per_url = array(); |
| 156 |
|
| 157 |
foreach ($result as $item) { |
| 158 |
$per_url[$item->url][] = $item; |
| 159 |
} |
| 160 |
|
| 161 |
return $per_url; |
| 162 |
} |
| 163 |
|
| 164 |
/** |
| 165 |
* Find requests that by themselves have a request count over the threshold |
| 166 |
* |
| 167 |
* @param array|null $all_suspicious_referrers Optional. By reference, will be populated with hashed referrers |
| 168 |
* @return array |
| 169 |
*/ |
| 170 |
private function get_single_suspicious_requests_by_referer(&$all_suspicious_referrers = null) { |
| 171 |
global $wpdb; |
| 172 |
$log_table_name = esc_sql($this->get_table_name()); |
| 173 |
|
| 174 |
$threshold = $this->suspicious_request_count_threshold; |
| 175 |
|
| 176 |
$by_referrer = $wpdb->get_results( |
| 177 |
$wpdb->prepare("SELECT `url`, |
| 178 |
SUM(IF(request_count < %d, 0, request_count)) AS total_count, |
| 179 |
referrer, |
| 180 |
MIN(request_timestamp) AS first_access, |
| 181 |
MAX(request_timestamp) AS last_access, |
| 182 |
COUNT(1) AS occurrences, |
| 183 |
1 AS total_referrers, |
| 184 |
'singles' AS row_type |
| 185 |
FROM `$log_table_name` GROUP BY `url`, referrer HAVING total_count >= %d ORDER BY request_timestamp DESC", $threshold, $threshold) // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 186 |
); |
| 187 |
|
| 188 |
foreach ($by_referrer as &$item) { |
| 189 |
$item->referrer = esc_html($item->referrer); |
| 190 |
$item->suspicious_referrers = 1; |
| 191 |
$item->non_suspicious_referrers = 0; |
| 192 |
|
| 193 |
$this->total_count_per_url[$item->url] = $item->total_count; |
| 194 |
|
| 195 |
if (is_array($all_suspicious_referrers)) { |
| 196 |
$all_suspicious_referrers[] = substr(md5($item->referrer), 0, 6); |
| 197 |
} |
| 198 |
} |
| 199 |
|
| 200 |
return $by_referrer; |
| 201 |
} |
| 202 |
|
| 203 |
/** |
| 204 |
* Find urls that if we sum the requests for all referrers, the result is greater than the threshold |
| 205 |
* |
| 206 |
* @param array $known_suspicious_referrers Hashed referrers to filter out from non-suspicious count |
| 207 |
* @return array |
| 208 |
*/ |
| 209 |
private function get_grouped_requests_by_url($known_suspicious_referrers) { |
| 210 |
global $wpdb; |
| 211 |
$log_table_name = esc_sql($this->get_table_name()); |
| 212 |
|
| 213 |
$threshold = $this->suspicious_request_count_threshold; |
| 214 |
|
| 215 |
$known_suspicious_referrers = implode(',', array_unique($known_suspicious_referrers)); |
| 216 |
|
| 217 |
$by_url = $wpdb->get_results( |
| 218 |
$wpdb->prepare("SELECT `url`, |
| 219 |
SUM(request_count) AS total_count, |
| 220 |
'' AS referrer, |
| 221 |
MIN(request_timestamp) AS first_access, |
| 222 |
MAX(request_timestamp) AS last_access, |
| 223 |
COUNT(1) AS occurrences, |
| 224 |
(COUNT(DISTINCT(IF(%d < request_count, '--nonsuspcious--', referrer)))) AS suspicious_referrers, |
| 225 |
(SUM(IF(%d < request_count AND LOCATE(MD5(SUBSTRING(referrer,1,6)), %s) = 0, 1, 0))) AS non_suspicious_referrers, |
| 226 |
COUNT(DISTINCT(referrer)) AS total_referrers, |
| 227 |
'grouped' AS row_type |
| 228 |
FROM `$log_table_name` GROUP BY `url` HAVING 1 < occurrences AND %d <= total_count ORDER BY request_timestamp DESC", $threshold, $threshold, $known_suspicious_referrers, $threshold) // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 229 |
); |
| 230 |
|
| 231 |
foreach ($by_url as &$item) { |
| 232 |
$item->referrer = esc_html__('(any)', 'wp-optimize'); |
| 233 |
|
| 234 |
if (0 < $item->non_suspicious_referrers) { |
| 235 |
// Some non-suspicious referrers exist in the group, all those are grouped under `--nonsuspicious--`, so remove it from the suspicious count |
| 236 |
$item->suspicious_referrers = $item->suspicious_referrers - 1; |
| 237 |
} |
| 238 |
|
| 239 |
$this->total_count_per_url[$item->url] = $item->total_count; |
| 240 |
} |
| 241 |
|
| 242 |
return $by_url; |
| 243 |
} |
| 244 |
|
| 245 |
/** |
| 246 |
* Get suspicious requests from DB for a single request url, grouped by under/over threshold |
| 247 |
* |
| 248 |
* @param string $url The actual url we are fetching for |
| 249 |
* @return array |
| 250 |
*/ |
| 251 |
public function get_url_requests_by_referrer($url) { |
| 252 |
global $wpdb; |
| 253 |
|
| 254 |
$log_table_name = esc_sql($this->get_table_name()); |
| 255 |
|
| 256 |
$return = array('over' => array(), 'under' => array()); |
| 257 |
|
| 258 |
$threshold = $this->suspicious_request_count_threshold; |
| 259 |
|
| 260 |
$requests = $wpdb->get_results( |
| 261 |
$wpdb->prepare("SELECT SUM(request_count) AS total_count, |
| 262 |
referrer, |
| 263 |
MIN(request_timestamp) AS first_access, |
| 264 |
MAX(request_timestamp) AS last_access |
| 265 |
FROM `$log_table_name` " . // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $log_table_name uses esc_sql, %i not supported till WP 6.2 |
| 266 |
"WHERE `url` = %s GROUP BY referrer, (%d < request_count) |
| 267 |
ORDER BY request_count DESC", |
| 268 |
$url, |
| 269 |
$threshold |
| 270 |
) |
| 271 |
); |
| 272 |
|
| 273 |
foreach ($requests as $request) { |
| 274 |
$group = $request->total_count >= $threshold ? 'over' : 'under'; |
| 275 |
$return[$group][] = $request; |
| 276 |
} |
| 277 |
|
| 278 |
return $return; |
| 279 |
} |
| 280 |
|
| 281 |
/** |
| 282 |
* Check if there are any suspicious requests logged, then return the count of unique URLs and total 404 requests |
| 283 |
* |
| 284 |
* @return array |
| 285 |
*/ |
| 286 |
public function get_suspicious_requests_count() { |
| 287 |
$requests = $this->get_single_suspicious_requests_by_referer(); |
| 288 |
|
| 289 |
$result = array( |
| 290 |
'unique_urls' => 0, |
| 291 |
'total_requests' => 0, |
| 292 |
); |
| 293 |
|
| 294 |
if (empty($requests)) { |
| 295 |
return $result; |
| 296 |
} |
| 297 |
|
| 298 |
if (!is_multisite() || (is_multisite() && is_network_admin())) { |
| 299 |
$result = $this->handle_single_site_or_network_admin_requests($requests, $result); |
| 300 |
} elseif (is_multisite() && defined('SUBDOMAIN_INSTALL') && SUBDOMAIN_INSTALL) { |
| 301 |
$result = $this->handle_subdomain_multisite_requests($requests, $result); |
| 302 |
} elseif (is_multisite()) { |
| 303 |
$result = $this->handle_subdirectory_multisite_requests($requests, $result); |
| 304 |
} |
| 305 |
return $result; |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* This method run to count total 404 requests and unique URLs for single site and multi site network admin. |
| 310 |
* |
| 311 |
* @param array $requests requests save in db |
| 312 |
* @param array $result result array containing unique urls and total request count default values (zero) |
| 313 |
* |
| 314 |
* @return array |
| 315 |
*/ |
| 316 |
private function handle_single_site_or_network_admin_requests($requests, $result) { |
| 317 |
|
| 318 |
foreach ($requests as $req) { |
| 319 |
$result = $this->increment_result($req->total_count, $result); |
| 320 |
} |
| 321 |
|
| 322 |
return $result; |
| 323 |
} |
| 324 |
|
| 325 |
/** |
| 326 |
* This method run to count total 404 requests and unique URLs for multi site sub-domain setup only (not for network admin). |
| 327 |
* |
| 328 |
* @param array $requests requests save in db |
| 329 |
* @param array $result result array containing unique urls and total request count default values (zero) |
| 330 |
* |
| 331 |
* @return array |
| 332 |
*/ |
| 333 |
private function handle_subdomain_multisite_requests($requests, $result) { |
| 334 |
|
| 335 |
$current_blog_id = get_current_blog_id(); |
| 336 |
$current_site_url = get_site_url($current_blog_id); |
| 337 |
$current_host = $this->parse_url($current_site_url, PHP_URL_HOST); |
| 338 |
|
| 339 |
foreach ($requests as $req) { |
| 340 |
$req_host = $this->parse_url($req->url, PHP_URL_HOST); |
| 341 |
|
| 342 |
if ($req_host === $current_host) { |
| 343 |
$result = $this->increment_result($req->total_count, $result); |
| 344 |
} |
| 345 |
} |
| 346 |
|
| 347 |
return $result; |
| 348 |
} |
| 349 |
|
| 350 |
/** |
| 351 |
* This method run to count total 404 requests and unique URLs for multi site sub-directory setup only. |
| 352 |
* |
| 353 |
* @param array $requests requests save in db |
| 354 |
* @param array $result result array containing unique urls and total request count default values (zero) |
| 355 |
* |
| 356 |
* @return array |
| 357 |
*/ |
| 358 |
private function handle_subdirectory_multisite_requests($requests, $result) { |
| 359 |
|
| 360 |
$current_blog_id = get_current_blog_id(); |
| 361 |
$current_site_url = get_site_url($current_blog_id); |
| 362 |
$current_host = $this->parse_url($current_site_url, PHP_URL_HOST); |
| 363 |
$current_path = trim((string) $this->parse_url($current_site_url, PHP_URL_PATH), '/'); |
| 364 |
$other_paths = array(); |
| 365 |
$all_sites = get_sites(array('fields' => 'ids')); |
| 366 |
|
| 367 |
foreach ($all_sites as $site_id) { |
| 368 |
if ($site_id === $current_blog_id) { |
| 369 |
continue; |
| 370 |
} |
| 371 |
$site_path = trim((string) $this->parse_url(get_site_url($site_id), PHP_URL_PATH), '/'); |
| 372 |
if (!empty($site_path)) { |
| 373 |
$other_paths[] = '/' . $site_path; |
| 374 |
} |
| 375 |
} |
| 376 |
|
| 377 |
foreach ($requests as $req) { |
| 378 |
|
| 379 |
$req_host = $this->parse_url($req->url, PHP_URL_HOST); |
| 380 |
$req_path = trim((string) $this->parse_url($req->url, PHP_URL_PATH), '/'); |
| 381 |
|
| 382 |
if ($req_host !== $current_host) { |
| 383 |
continue; |
| 384 |
} |
| 385 |
|
| 386 |
if (empty($current_path)) { |
| 387 |
$is_sub_site_request = false; |
| 388 |
foreach ($other_paths as $sub_path) { |
| 389 |
if (0 === strpos('/' . $req_path, $sub_path)) { |
| 390 |
$is_sub_site_request = true; |
| 391 |
break; |
| 392 |
} |
| 393 |
} |
| 394 |
if ($is_sub_site_request) { |
| 395 |
continue; |
| 396 |
} |
| 397 |
} else { |
| 398 |
if (0 !== strpos('/' . $req_path, '/' . $current_path)) { |
| 399 |
continue; |
| 400 |
} |
| 401 |
} |
| 402 |
$result = $this->increment_result($req->total_count, $result); |
| 403 |
} |
| 404 |
|
| 405 |
return $result; |
| 406 |
} |
| 407 |
|
| 408 |
/** |
| 409 |
* Returns the incremented result for unique urls and total requests |
| 410 |
* |
| 411 |
* @param int $total_count total requests count (didn't type cast as $wpdb->get_results() doesn't guarantee it) |
| 412 |
* @param array $result result array containing unique urls and total request count |
| 413 |
* |
| 414 |
* @return array |
| 415 |
*/ |
| 416 |
private function increment_result($total_count, $result) { |
| 417 |
$result['unique_urls']++; |
| 418 |
$result['total_requests'] += (int) $total_count; |
| 419 |
return $result; |
| 420 |
} |
| 421 |
|
| 422 |
/** |
| 423 |
* Returns the dashboard alert request count |
| 424 |
* |
| 425 |
* @return int |
| 426 |
*/ |
| 427 |
public function get_dashboard_alert_request_count_threshold() { |
| 428 |
return $this->dashboard_alert_request_count_threshold; |
| 429 |
} |
| 430 |
|
| 431 |
/** |
| 432 |
* Returns the suspicious request count threshold |
| 433 |
* |
| 434 |
* @return int |
| 435 |
*/ |
| 436 |
public function get_suspicious_request_count_threshold() { |
| 437 |
return $this->suspicious_request_count_threshold; |
| 438 |
} |
| 439 |
|
| 440 |
/** |
| 441 |
* Returns the table name |
| 442 |
* |
| 443 |
* @return string |
| 444 |
*/ |
| 445 |
private function get_table_name() { |
| 446 |
return WP_Optimize_Table_404_Detector::get_instance()->get_table_name(); |
| 447 |
} |
| 448 |
|
| 449 |
/** |
| 450 |
* Sorts the array by `url total count` DESC, `url` ASC, and `request total_count` DESC |
| 451 |
* |
| 452 |
* @param object $result_a |
| 453 |
* @param object $result_b |
| 454 |
* |
| 455 |
* @return int |
| 456 |
*/ |
| 457 |
private function sort_suspicious_requests($result_a, $result_b) { |
| 458 |
$a_url_total = $this->total_count_per_url[$result_a->url] ?? 0; |
| 459 |
$b_url_total = $this->total_count_per_url[$result_b->url] ?? 0; |
| 460 |
|
| 461 |
$a = (PHP_INT_MAX - $a_url_total) . ' ' . $result_a->url . ' ' . (PHP_INT_MAX - $result_a->total_count); |
| 462 |
$b = (PHP_INT_MAX - $b_url_total) . ' ' . $result_b->url . ' ' . (PHP_INT_MAX - $result_b->total_count); |
| 463 |
|
| 464 |
return $a < $b ? -1 : 1; |
| 465 |
} |
| 466 |
|
| 467 |
/** |
| 468 |
* Wrapper over `wp_parse_url` to handle a `false` response, in such case if component is -1 then we return empty array else empty string |
| 469 |
* This wrapper always requests `wp_parse_url` the default return an array with all components found. |
| 470 |
* |
| 471 |
* @param string $url The URL to be parsed |
| 472 |
* @param int $component The specific component to retrieve. Use one of the PHP predefined constants to specify which one. Defaults to -1 (= return all parts as an array). |
| 473 |
* |
| 474 |
* @return string|array |
| 475 |
*/ |
| 476 |
private function parse_url($url, $component = -1) { |
| 477 |
$parsed = wp_parse_url($url, $component); |
| 478 |
|
| 479 |
if (false !== $parsed && null !== $parsed) { |
| 480 |
return $parsed; |
| 481 |
} |
| 482 |
return (-1 === $component) ? array() : ''; |
| 483 |
} |
| 484 |
|
| 485 |
/** |
| 486 |
* Check if the plain permalink structure is currently set |
| 487 |
* |
| 488 |
* @return bool |
| 489 |
*/ |
| 490 |
private function is_plain_permalink_structure(): bool { |
| 491 |
return "" === get_option('permalink_structure'); |
| 492 |
} |
| 493 |
} |
| 494 |
|
| 495 |
endif; |
| 496 |
|