| 1 |
<?php |
| 2 |
/** |
| 3 |
* Firewall Class |
| 4 |
* |
| 5 |
* WordPress-optimized firewall protection |
| 6 |
* |
| 7 |
* @package Vigilante |
| 8 |
*/ |
| 9 |
|
| 10 |
// Prevent direct access |
| 11 |
if ( ! defined( 'ABSPATH' ) ) { |
| 12 |
exit; |
| 13 |
} |
| 14 |
|
| 15 |
/** |
| 16 |
* Class Vigilante_Firewall |
| 17 |
* |
| 18 |
* Provides firewall protection against common attacks |
| 19 |
*/ |
| 20 |
class Vigilante_Firewall { |
| 21 |
|
| 22 |
/** |
| 23 |
* Rate limiting window, in seconds. |
| 24 |
* |
| 25 |
* The "Requests per Minute" setting is measured over this window. |
| 26 |
* |
| 27 |
* @var int |
| 28 |
*/ |
| 29 |
const RATE_LIMIT_WINDOW = 60; |
| 30 |
|
| 31 |
/** |
| 32 |
* Settings instance |
| 33 |
* |
| 34 |
* @var Vigilante_Settings |
| 35 |
*/ |
| 36 |
private $settings; |
| 37 |
|
| 38 |
/** |
| 39 |
* Activity log instance |
| 40 |
* |
| 41 |
* @var Vigilante_Activity_Log |
| 42 |
*/ |
| 43 |
private $activity_log; |
| 44 |
|
| 45 |
/** |
| 46 |
* Firewall options |
| 47 |
* |
| 48 |
* @var array |
| 49 |
*/ |
| 50 |
private $options; |
| 51 |
|
| 52 |
/** |
| 53 |
* Current request data |
| 54 |
* |
| 55 |
* @var array |
| 56 |
*/ |
| 57 |
private $request_data = array(); |
| 58 |
|
| 59 |
/** |
| 60 |
* Memoized haystack the pattern checks run against |
| 61 |
* |
| 62 |
* @since 2.9.9 |
| 63 |
* |
| 64 |
* @var string|null |
| 65 |
*/ |
| 66 |
private $haystack = null; |
| 67 |
|
| 68 |
/** |
| 69 |
* Constructor |
| 70 |
* |
| 71 |
* @param Vigilante_Settings $settings Settings instance. |
| 72 |
* @param Vigilante_Activity_Log $activity_log Activity log instance. |
| 73 |
*/ |
| 74 |
public function __construct( $settings, $activity_log ) { |
| 75 |
$this->settings = $settings; |
| 76 |
$this->activity_log = $activity_log; |
| 77 |
$this->options = $settings->get_section( 'firewall' ); |
| 78 |
|
| 79 |
// Run firewall checks - must be after plugin init (priority 1) |
| 80 |
add_action( 'init', array( $this, 'run_firewall' ), 2 ); |
| 81 |
|
| 82 |
// Rate limiting |
| 83 |
if ( ! empty( $this->options['rate_limiting']['enabled'] ) ) { |
| 84 |
add_action( 'init', array( $this, 'check_rate_limit' ), 2 ); |
| 85 |
} |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Run all firewall checks |
| 90 |
*/ |
| 91 |
public function run_firewall() { |
| 92 |
// Skip for whitelisted IPs |
| 93 |
if ( $this->is_ip_whitelisted() ) { |
| 94 |
return; |
| 95 |
} |
| 96 |
|
| 97 |
// Skip for whitelisted User-Agents (ManageWP, MainWP, etc.) |
| 98 |
if ( $this->is_ua_whitelisted() ) { |
| 99 |
return; |
| 100 |
} |
| 101 |
|
| 102 |
// Gather request data first: a block is logged with the address it |
| 103 |
// turned away, and until 2.11.1 the blacklist ran before this, so the |
| 104 |
// entry for a blacklisted IP recorded no address at all. |
| 105 |
$this->gather_request_data(); |
| 106 |
|
| 107 |
// Check if IP is blacklisted |
| 108 |
if ( $this->is_ip_blacklisted() ) { |
| 109 |
$this->block_request( 'ip_blacklisted', __( 'IP address is blacklisted', 'vigilante' ) ); |
| 110 |
} |
| 111 |
|
| 112 |
// Check if User-Agent is blacklisted (after gathering request data) |
| 113 |
if ( $this->is_ua_blacklisted() ) { |
| 114 |
$this->block_request( 'ua_blacklisted', __( 'User-Agent is blacklisted', 'vigilante' ) ); |
| 115 |
} |
| 116 |
|
| 117 |
// Run security checks |
| 118 |
// NOTE: These are PHP-based checks that complement htaccess rules |
| 119 |
// Some protections exist in both layers for defense in depth |
| 120 |
$checks = array( |
| 121 |
// PHP request filtering (complements htaccess block_bad_query_strings) |
| 122 |
'block_bad_query_strings' => 'check_query_strings', |
| 123 |
'block_sql_injection' => 'check_sql_injection', |
| 124 |
'block_xss_attacks' => 'check_xss_attacks', |
| 125 |
'block_file_inclusion' => 'check_file_inclusion', |
| 126 |
'block_directory_traversal' => 'check_directory_traversal', |
| 127 |
// Bot protection (complements htaccess block_bad_bots) |
| 128 |
'block_bad_bots' => 'check_bad_bots', |
| 129 |
'block_empty_user_agent' => 'check_empty_user_agent', |
| 130 |
); |
| 131 |
|
| 132 |
foreach ( $checks as $option => $method ) { |
| 133 |
if ( ! empty( $this->options[ $option ] ) && method_exists( $this, $method ) ) { |
| 134 |
$result = $this->$method(); |
| 135 |
if ( is_string( $result ) ) { |
| 136 |
$this->block_request( $option, $result ); |
| 137 |
} |
| 138 |
} |
| 139 |
} |
| 140 |
|
| 141 |
// Check HTTP method if limit_http_methods is enabled |
| 142 |
if ( ! empty( $this->options['limit_http_methods'] ) ) { |
| 143 |
$this->check_http_method(); |
| 144 |
} |
| 145 |
} |
| 146 |
|
| 147 |
/** |
| 148 |
* Gather current request data |
| 149 |
*/ |
| 150 |
private function gather_request_data() { |
| 151 |
$this->haystack = null; |
| 152 |
|
| 153 |
$this->request_data = array( |
| 154 |
'uri' => isset( $_SERVER['REQUEST_URI'] ) ? sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ) ) : '', |
| 155 |
'query_string'=> isset( $_SERVER['QUERY_STRING'] ) ? sanitize_text_field( wp_unslash( $_SERVER['QUERY_STRING'] ) ) : '', |
| 156 |
/* |
| 157 |
* Copies that keep the percent encoding, used only as the haystack |
| 158 |
* of the pattern checks and never logged, printed or stored. |
| 159 |
* |
| 160 |
* They exist because sanitize_text_field() deletes every %XX |
| 161 |
* sequence instead of decoding it: the copies above are the payload |
| 162 |
* with the evidence removed, so an encoded attack was invisible to |
| 163 |
* every rule that reads them. Measured on 22 aug 2026 against 2.9.8, |
| 164 |
* ?x=%3Cscript%3E, javascript%3A, php%3A%2F%2F and GLOBALS%5B all |
| 165 |
* reached the checks as harmless text and went straight through. |
| 166 |
* |
| 167 |
* No sanitizer is applied, and that is the point: every one of them |
| 168 |
* destroys exactly what has to be matched. sanitize_text_field() |
| 169 |
* deletes the %XX sequences and strips tags. esc_url_raw() is worse |
| 170 |
* here: measured on 22 aug 2026, it returns an empty string for a |
| 171 |
* query that carries an unencoded :// , which is precisely the |
| 172 |
* remote inclusion shape, so it would blind the firewall instead of |
| 173 |
* arming it. These two values are never echoed, never stored and |
| 174 |
* never reach a query; they are the haystack of preg_match() and |
| 175 |
* nothing else. |
| 176 |
*/ |
| 177 |
// phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- inspection buffer for the pattern checks, see the note above. Sanitizing it is what hid the attacks. Never output, stored nor queried. |
| 178 |
'uri_raw' => isset( $_SERVER['REQUEST_URI'] ) ? wp_unslash( $_SERVER['REQUEST_URI'] ) : '', |
| 179 |
// phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- same as uri_raw. |
| 180 |
'query_raw' => isset( $_SERVER['QUERY_STRING'] ) ? wp_unslash( $_SERVER['QUERY_STRING'] ) : '', |
| 181 |
'user_agent' => isset( $_SERVER['HTTP_USER_AGENT'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) : '', |
| 182 |
'referer' => isset( $_SERVER['HTTP_REFERER'] ) ? esc_url_raw( wp_unslash( $_SERVER['HTTP_REFERER'] ) ) : '', |
| 183 |
'method' => isset( $_SERVER['REQUEST_METHOD'] ) ? sanitize_text_field( wp_unslash( $_SERVER['REQUEST_METHOD'] ) ) : 'GET', |
| 184 |
'ip' => $this->get_client_ip(), |
| 185 |
); |
| 186 |
} |
| 187 |
|
| 188 |
/** |
| 189 |
* What the pattern checks run against: the request as it arrived, plus its decoded form |
| 190 |
* |
| 191 |
* Both forms on purpose. Some patterns look for the encoded shape, such as |
| 192 |
* the null byte %00 or the %5b of GLOBALS[, and others for the decoded one, |
| 193 |
* such as <script or ../. Feeding only one of the two leaves half the rules |
| 194 |
* looking at something that cannot match. |
| 195 |
* |
| 196 |
* Decoded once, not twice: a second pass catches a bit more evasion and |
| 197 |
* brings in false positives that are not worth it. |
| 198 |
* |
| 199 |
* @since 2.9.9 |
| 200 |
* |
| 201 |
* @return string |
| 202 |
*/ |
| 203 |
private function inspection_haystack() { |
| 204 |
if ( null !== $this->haystack ) { |
| 205 |
return $this->haystack; |
| 206 |
} |
| 207 |
|
| 208 |
$raw = trim( (string) $this->request_data['uri_raw'] . ' ' . (string) $this->request_data['query_raw'] ); |
| 209 |
$decoded = rawurldecode( $raw ); |
| 210 |
|
| 211 |
$this->haystack = ( $raw === $decoded ) ? $raw : $raw . ' ' . $decoded; |
| 212 |
|
| 213 |
return $this->haystack; |
| 214 |
} |
| 215 |
|
| 216 |
/** |
| 217 |
* Check for malicious query strings |
| 218 |
* |
| 219 |
* @return string|false Error message or false if safe. |
| 220 |
*/ |
| 221 |
private function check_query_strings() { |
| 222 |
$query = $this->request_data['query_raw']; |
| 223 |
|
| 224 |
if ( empty( $query ) ) { |
| 225 |
return false; |
| 226 |
} |
| 227 |
|
| 228 |
// Length is measured on the query alone, the rest of the patterns run |
| 229 |
// against the whole request in both its raw and decoded forms. |
| 230 |
$haystack = $this->inspection_haystack(); |
| 231 |
|
| 232 |
// Dangerous patterns |
| 233 |
$patterns = array( |
| 234 |
// Too long query strings |
| 235 |
'/^.{4000,}$/s' => __( 'Query string too long', 'vigilante' ), |
| 236 |
|
| 237 |
// Null bytes |
| 238 |
'/(\x00|%00)/i' => __( 'Null byte detected', 'vigilante' ), |
| 239 |
|
| 240 |
// PHP wrappers |
| 241 |
'/php:\/\//i' => __( 'PHP wrapper detected', 'vigilante' ), |
| 242 |
'/data:\/\//i' => __( 'Data wrapper detected', 'vigilante' ), |
| 243 |
|
| 244 |
// Globals/Request manipulation |
| 245 |
'/(globals|mosconfig)(\[|\%5b)/i' => __( 'Global manipulation attempt', 'vigilante' ), |
| 246 |
'/_request(\[|\%5b)/i' => __( 'Request manipulation attempt', 'vigilante' ), |
| 247 |
|
| 248 |
// Config file access |
| 249 |
'/wp-config\.php/i' => __( 'Config file access attempt', 'vigilante' ), |
| 250 |
|
| 251 |
// Common attack patterns |
| 252 |
'/(\<|%3c).*script.*(\>|%3e)/i' => __( 'Script tag detected', 'vigilante' ), |
| 253 |
'/document\.(cookie|location|write)/i' => __( 'DOM manipulation attempt', 'vigilante' ), |
| 254 |
); |
| 255 |
|
| 256 |
foreach ( $patterns as $pattern => $message ) { |
| 257 |
// The length rule is anchored, so it has to see the query on its |
| 258 |
// own; every other pattern gets the whole request. |
| 259 |
$subject = ( '/^.{4000,}$/s' === $pattern ) ? $query : $haystack; |
| 260 |
|
| 261 |
if ( preg_match( $pattern, $subject ) ) { |
| 262 |
return $message; |
| 263 |
} |
| 264 |
} |
| 265 |
|
| 266 |
return false; |
| 267 |
} |
| 268 |
|
| 269 |
/** |
| 270 |
* Check for SQL injection attempts |
| 271 |
* |
| 272 |
* @return string|false Error message or false if safe. |
| 273 |
*/ |
| 274 |
private function check_sql_injection() { |
| 275 |
// Skip SQL injection checks for authenticated admin users on admin pages |
| 276 |
// WordPress handles sanitization for these requests |
| 277 |
if ( is_admin() && is_user_logged_in() && current_user_can( 'edit_posts' ) ) { |
| 278 |
return false; |
| 279 |
} |
| 280 |
|
| 281 |
$to_check = array( |
| 282 |
$this->inspection_haystack(), |
| 283 |
); |
| 284 |
|
| 285 |
// Check POST data, but exclude content fields that may contain legitimate code/text |
| 286 |
// phpcs:ignore WordPress.Security.NonceVerification.Missing |
| 287 |
if ( ! empty( $_POST ) ) { |
| 288 |
// phpcs:ignore WordPress.Security.NonceVerification.Missing |
| 289 |
$post_data = $_POST; |
| 290 |
|
| 291 |
// Remove fields that commonly contain user content (posts, comments, etc.) |
| 292 |
// These are sanitized by WordPress core |
| 293 |
$excluded_fields = array( |
| 294 |
'content', |
| 295 |
'post_content', |
| 296 |
'comment', |
| 297 |
'description', |
| 298 |
'excerpt', |
| 299 |
'post_excerpt', |
| 300 |
'message', |
| 301 |
'bio', |
| 302 |
'acf', // Advanced Custom Fields |
| 303 |
'meta', // Post meta |
| 304 |
'tax_input', // Taxonomy input |
| 305 |
'_content', // Various content fields |
| 306 |
); |
| 307 |
|
| 308 |
foreach ( $excluded_fields as $field ) { |
| 309 |
unset( $post_data[ $field ] ); |
| 310 |
} |
| 311 |
|
| 312 |
// Only check remaining POST data if not empty |
| 313 |
if ( ! empty( $post_data ) ) { |
| 314 |
$to_check[] = wp_json_encode( $post_data ); |
| 315 |
} |
| 316 |
} |
| 317 |
|
| 318 |
$combined = implode( ' ', array_filter( $to_check ) ); |
| 319 |
|
| 320 |
if ( empty( $combined ) ) { |
| 321 |
return false; |
| 322 |
} |
| 323 |
|
| 324 |
// SQL injection patterns - focused on actual attack vectors |
| 325 |
$patterns = array( |
| 326 |
// Union based injection - high confidence attack pattern |
| 327 |
'/union\s+(all\s+)?select/i' => __( 'UNION SELECT detected', 'vigilante' ), |
| 328 |
|
| 329 |
// SQL commands in URL/query string context (not in POST body) |
| 330 |
// More specific pattern to reduce false positives |
| 331 |
'/[\'\"]\s*(;|--|#)\s*(select|insert|update|delete|drop|truncate|alter|create)/i' => __( 'SQL command injection attempt', 'vigilante' ), |
| 332 |
|
| 333 |
// Hex encoding of SQL - typically used in attacks |
| 334 |
'/0x[0-9a-f]{16,}/i' => __( 'Hex encoding detected', 'vigilante' ), |
| 335 |
|
| 336 |
// Benchmark/sleep attacks - time-based SQL injection |
| 337 |
'/(benchmark|sleep)\s*\(\s*\d/i' => __( 'Time-based injection attempt', 'vigilante' ), |
| 338 |
|
| 339 |
// Information schema access |
| 340 |
'/information_schema\.(tables|columns|schemata)/i' => __( 'Schema access attempt', 'vigilante' ), |
| 341 |
|
| 342 |
// Load file - file read attempt |
| 343 |
'/load_file\s*\(/i' => __( 'Load file attempt', 'vigilante' ), |
| 344 |
|
| 345 |
// Into outfile - file write attempt |
| 346 |
'/into\s+(out|dump)file/i' => __( 'File write attempt', 'vigilante' ), |
| 347 |
|
| 348 |
// Stacked queries with dangerous commands |
| 349 |
'/;\s*(drop|truncate|delete\s+from|update\s+\w+\s+set)/i' => __( 'Stacked query injection', 'vigilante' ), |
| 350 |
); |
| 351 |
|
| 352 |
foreach ( $patterns as $pattern => $message ) { |
| 353 |
if ( preg_match( $pattern, $combined ) ) { |
| 354 |
return $message; |
| 355 |
} |
| 356 |
} |
| 357 |
|
| 358 |
return false; |
| 359 |
} |
| 360 |
|
| 361 |
/** |
| 362 |
* Check for XSS attacks |
| 363 |
* |
| 364 |
* @return string|false Error message or false if safe. |
| 365 |
*/ |
| 366 |
private function check_xss_attacks() { |
| 367 |
$combined = $this->inspection_haystack(); |
| 368 |
|
| 369 |
if ( empty( $combined ) ) { |
| 370 |
return false; |
| 371 |
} |
| 372 |
|
| 373 |
// Already carries the decoded form, see inspection_haystack(). |
| 374 |
$decoded = $combined; |
| 375 |
|
| 376 |
// XSS patterns |
| 377 |
$patterns = array( |
| 378 |
// Script tags |
| 379 |
'/<script[^>]*>/i' => __( 'Script tag detected', 'vigilante' ), |
| 380 |
|
| 381 |
/* |
| 382 |
* Event handlers. Two shapes, because the rule used to be a bare |
| 383 |
* \bon\w+\s*= and that matches any parameter whose name starts |
| 384 |
* with "on": only=, once=, online= and onboarding= were all |
| 385 |
* answered with a 403 on every site with the firewall on, and the |
| 386 |
* owner never saw it because it only hits visitors. |
| 387 |
*/ |
| 388 |
'/<[^>]*\bon\w+\s*=/i' => __( 'Event handler detected', 'vigilante' ), |
| 389 |
'/\bon(abort|blur|change|click|contextmenu|copy|cut|dblclick|drag\w*|drop|error|focus\w*|input|invalid|key\w+|load\w*|mouse\w+|paste|pointer\w+|reset|resize|scroll|select|submit|toggle|touch\w+|transitionend|animation\w+|wheel)\s*=\s*["\']?\s*[\w.$]+\s*\(/i' => __( 'Event handler detected', 'vigilante' ), |
| 390 |
|
| 391 |
// JavaScript protocol |
| 392 |
'/javascript\s*:/i' => __( 'JavaScript protocol detected', 'vigilante' ), |
| 393 |
|
| 394 |
// VBScript |
| 395 |
'/vbscript\s*:/i' => __( 'VBScript detected', 'vigilante' ), |
| 396 |
|
| 397 |
// Data URL |
| 398 |
'/data\s*:[^,]*base64/i' => __( 'Base64 data URL detected', 'vigilante' ), |
| 399 |
|
| 400 |
// Expression (IE) |
| 401 |
'/expression\s*\(/i' => __( 'CSS expression detected', 'vigilante' ), |
| 402 |
|
| 403 |
// Iframe injection |
| 404 |
'/<iframe[^>]*>/i' => __( 'Iframe injection detected', 'vigilante' ), |
| 405 |
|
| 406 |
// Object/embed |
| 407 |
'/<(object|embed|applet)[^>]*>/i' => __( 'Object tag detected', 'vigilante' ), |
| 408 |
); |
| 409 |
|
| 410 |
foreach ( $patterns as $pattern => $message ) { |
| 411 |
if ( preg_match( $pattern, $decoded ) ) { |
| 412 |
return $message; |
| 413 |
} |
| 414 |
} |
| 415 |
|
| 416 |
return false; |
| 417 |
} |
| 418 |
|
| 419 |
/** |
| 420 |
* Check for file inclusion attacks |
| 421 |
* |
| 422 |
* @return string|false Error message or false if safe. |
| 423 |
*/ |
| 424 |
private function check_file_inclusion() { |
| 425 |
$combined = $this->inspection_haystack(); |
| 426 |
|
| 427 |
if ( empty( $combined ) ) { |
| 428 |
return false; |
| 429 |
} |
| 430 |
|
| 431 |
// Remote inclusion is decided on the parsed values, not on the raw |
| 432 |
// string. Until 2.9.9 any '=' followed by an absolute URL tripped this |
| 433 |
// rule, and legitimate links carry those all the time: a redirect_to |
| 434 |
// back to the site itself, a return_url, a payment gateway callback. |
| 435 |
// What makes it an inclusion attempt is the target being somewhere |
| 436 |
// else, so a URL pointing at this very site is left alone. |
| 437 |
// |
| 438 |
// The core endpoint that resolves an embed takes an external URL as |
| 439 |
// its whole job, so it is exempted rather than made to look innocent. |
| 440 |
// Only this check is skipped: the PHP wrappers and the system paths |
| 441 |
// below still run on that route, for everyone. |
| 442 |
if ( ! $this->is_core_embed_proxy_request() && $this->has_remote_inclusion() ) { |
| 443 |
return __( 'Remote file inclusion attempt', 'vigilante' ); |
| 444 |
} |
| 445 |
|
| 446 |
// File inclusion patterns |
| 447 |
$patterns = array( |
| 448 |
// PHP wrappers |
| 449 |
'/(php|zip|glob|phar|ssh2|rar|ogg|expect):\/\//i' => __( 'PHP wrapper detected', 'vigilante' ), |
| 450 |
|
| 451 |
// System files |
| 452 |
'/\/etc\/(passwd|shadow|hosts)/i' => __( 'System file access attempt', 'vigilante' ), |
| 453 |
'/\/proc\/self/i' => __( 'Proc access attempt', 'vigilante' ), |
| 454 |
|
| 455 |
// Windows paths |
| 456 |
'/[a-z]:\\\\(windows|winnt)/i' => __( 'Windows path detected', 'vigilante' ), |
| 457 |
); |
| 458 |
|
| 459 |
foreach ( $patterns as $pattern => $message ) { |
| 460 |
if ( preg_match( $pattern, $combined ) ) { |
| 461 |
return $message; |
| 462 |
} |
| 463 |
} |
| 464 |
|
| 465 |
return false; |
| 466 |
} |
| 467 |
|
| 468 |
/** |
| 469 |
* Parameter names a remote inclusion payload travels in |
| 470 |
* |
| 471 |
* An inclusion needs its value to reach an include() or a require(), so it |
| 472 |
* arrives in the parameter a vulnerable script treats as a path. These are |
| 473 |
* the names those scripts use, and the ones every RFI scanner probes. |
| 474 |
* |
| 475 |
* Deliberately absent: url, redirect, redirect_to, return, return_url, |
| 476 |
* callback and the rest of the link-carrying names. Carrying a URL is what |
| 477 |
* those are for. Matching them is what turned WordPress core's own oembed |
| 478 |
* proxy into a 403 for anyone using the block editor, reported on 9 sep |
| 479 |
* 2026 and fixed in 2.11.1. |
| 480 |
* |
| 481 |
* @since 2.11.1 |
| 482 |
* |
| 483 |
* @var string[] |
| 484 |
*/ |
| 485 |
private static $inclusion_param_names = array( |
| 486 |
// The value is read as a file |
| 487 |
'file', 'files', 'filename', 'file_name', 'filepath', 'file_path', |
| 488 |
'archivo', 'arquivo', 'fichier', 'datei', |
| 489 |
// ...or as the place to read it from |
| 490 |
'path', 'paths', 'dir', 'directory', 'folder', 'root', 'base', |
| 491 |
'basepath', 'base_path', 'abs_path', 'absolute_path', |
| 492 |
'mosconfig_absolute_path', |
| 493 |
// ...or as the page a front controller includes |
| 494 |
'page', 'pag', 'pagina', 'pageweb', 'pg', 'seite', |
| 495 |
'include', 'includes', 'inc', 'incl', 'require', 'load', 'loadfile', |
| 496 |
'open', 'read', 'readfile', 'show', 'display', 'view', 'content', |
| 497 |
// ...or as a template, which is a file by another name |
| 498 |
'template', 'templates', 'tpl', 'tmpl', 'theme', 'skin', 'style', |
| 499 |
'layout', 'doc', 'document', 'module', 'mod', 'plugin', 'controller', |
| 500 |
'class', 'func', 'function', 'lang', 'language', |
| 501 |
// ...or as configuration, or straight into a shell |
| 502 |
'conf', 'config', 'cfg', 'src', 'source', 'download', |
| 503 |
'cmd', 'exec', 'shell', 'system', |
| 504 |
); |
| 505 |
|
| 506 |
/** |
| 507 |
* Extensions a remote inclusion payload is served with |
| 508 |
* |
| 509 |
* The point of the target is that it carries code, or text the vulnerable |
| 510 |
* script will treat as code. Media extensions are not here on purpose: an |
| 511 |
* embedded .mp4 or .jpg from a CDN is what the editor sends all day. |
| 512 |
* |
| 513 |
* @since 2.11.1 |
| 514 |
* |
| 515 |
* @var string[] |
| 516 |
*/ |
| 517 |
private static $inclusion_extensions = array( |
| 518 |
'php', 'php2', 'php3', 'php4', 'php5', 'php6', 'php7', 'php8', |
| 519 |
'phps', 'phtml', 'pht', 'phar', 'inc', 'txt', 'log', 'ini', 'cfg', |
| 520 |
'conf', 'asp', 'aspx', 'jsp', 'jspx', 'cgi', 'pl', 'py', 'rb', 'sh', |
| 521 |
'bash', 'exe', 'dll', 'so', 'bak', 'old', 'env', |
| 522 |
); |
| 523 |
|
| 524 |
/** |
| 525 |
* Whether the request carries a remote file inclusion attempt |
| 526 |
* |
| 527 |
* Works on the parsed parameters rather than on a pattern match over the |
| 528 |
* whole string, for two reasons: a link back to the site itself is not |
| 529 |
* mistaken for an attack, and an encoded payload is seen for what it is. |
| 530 |
* The copy of the query string kept for logging goes through |
| 531 |
* sanitize_text_field(), which strips every %XX sequence instead of |
| 532 |
* decoding it, so the encoded form never looked like a URL there. |
| 533 |
* |
| 534 |
* Until 2.11.0 an external URL in any parameter was the whole signature, |
| 535 |
* and that is not what an inclusion looks like, it is what a link looks |
| 536 |
* like. WordPress core's own /wp-json/oembed/1.0/proxy?url=... is the |
| 537 |
* clearest case: the block editor asks the site to resolve a YouTube URL, |
| 538 |
* the rule read it as RFI, and embedding was dead on every site with the |
| 539 |
* rule on while the classic editor kept working, because it posts the same |
| 540 |
* URL to admin-ajax instead of putting it in a query string. Since 2.11.1 |
| 541 |
* an external URL is an inclusion attempt when it travels in a parameter |
| 542 |
* that is read as a path, or when it points at something includable. |
| 543 |
* |
| 544 |
* @since 2.9.9 |
| 545 |
* |
| 546 |
* @return bool |
| 547 |
*/ |
| 548 |
private function has_remote_inclusion() { |
| 549 |
$query = $this->request_data['query_raw']; |
| 550 |
|
| 551 |
if ( '' === $query ) { |
| 552 |
return false; |
| 553 |
} |
| 554 |
|
| 555 |
// parse_str() decodes as it splits, so this sees the same values PHP |
| 556 |
// would have put in $_GET, without reading the superglobal. |
| 557 |
$params = array(); |
| 558 |
parse_str( $query, $params ); |
| 559 |
|
| 560 |
$home_host = $this->normalize_host( wp_parse_url( home_url(), PHP_URL_HOST ) ); |
| 561 |
|
| 562 |
foreach ( $this->flatten_query_params( $params ) as $pair ) { |
| 563 |
list( $names, $value ) = $pair; |
| 564 |
|
| 565 |
if ( ! preg_match_all( '/(?:https?|ftp):\/\/[^\s\'"<>]+/i', $value, $matches ) ) { |
| 566 |
continue; |
| 567 |
} |
| 568 |
|
| 569 |
foreach ( $matches[0] as $url ) { |
| 570 |
$host = $this->normalize_host( wp_parse_url( $url, PHP_URL_HOST ) ); |
| 571 |
|
| 572 |
// A URL pointing at this very site is not an inclusion. |
| 573 |
if ( '' !== $host && $host === $home_host ) { |
| 574 |
continue; |
| 575 |
} |
| 576 |
|
| 577 |
if ( $this->looks_like_inclusion( $names, $url ) ) { |
| 578 |
return true; |
| 579 |
} |
| 580 |
} |
| 581 |
} |
| 582 |
|
| 583 |
return false; |
| 584 |
} |
| 585 |
|
| 586 |
/** |
| 587 |
* Query parameters as (name segments, value) pairs |
| 588 |
* |
| 589 |
* Keeps every segment of a nested name, so opts[file]=http://... is seen |
| 590 |
* as the inclusion parameter it is and not as an anonymous value. |
| 591 |
* |
| 592 |
* @since 2.11.1 |
| 593 |
* |
| 594 |
* @param array $params Parsed parameters. |
| 595 |
* @param string[] $inherited Name segments of the parent levels. |
| 596 |
* @return array List of array( string[] $names, string $value ). |
| 597 |
*/ |
| 598 |
private function flatten_query_params( $params, $inherited = array() ) { |
| 599 |
$pairs = array(); |
| 600 |
|
| 601 |
foreach ( $params as $key => $value ) { |
| 602 |
$names = array_merge( $inherited, array( strtolower( (string) $key ) ) ); |
| 603 |
|
| 604 |
if ( is_array( $value ) ) { |
| 605 |
$pairs = array_merge( $pairs, $this->flatten_query_params( $value, $names ) ); |
| 606 |
continue; |
| 607 |
} |
| 608 |
|
| 609 |
if ( is_scalar( $value ) ) { |
| 610 |
$pairs[] = array( $names, (string) $value ); |
| 611 |
} |
| 612 |
} |
| 613 |
|
| 614 |
return $pairs; |
| 615 |
} |
| 616 |
|
| 617 |
/** |
| 618 |
* Whether an external URL in this parameter is an inclusion attempt |
| 619 |
* |
| 620 |
* Two independent signals, either is enough: the parameter is one a |
| 621 |
* vulnerable script reads as a path, or the target is something that gets |
| 622 |
* included rather than linked. The trailing '?' and the null byte are the |
| 623 |
* two ways a payload truncates whatever the script appends to it. |
| 624 |
* |
| 625 |
* What this deliberately no longer catches, so nobody reads its silence as |
| 626 |
* coverage: an external URL with no includable extension travelling in a |
| 627 |
* parameter whose name is not on the list. That shape is a link, which is |
| 628 |
* why the site's own search box and every return_url tripped the rule |
| 629 |
* before. A payload in it still has to reach an include() in some other |
| 630 |
* plugin's code to do anything, and the wrappers, the system paths and the |
| 631 |
* traversal rules below are untouched. |
| 632 |
* |
| 633 |
* @since 2.11.1 |
| 634 |
* |
| 635 |
* @param string[] $names Name segments of the parameter. |
| 636 |
* @param string $url The external URL found in its value. |
| 637 |
* @return bool |
| 638 |
*/ |
| 639 |
private function looks_like_inclusion( $names, $url ) { |
| 640 |
foreach ( $names as $name ) { |
| 641 |
if ( in_array( $name, self::$inclusion_param_names, true ) ) { |
| 642 |
return true; |
| 643 |
} |
| 644 |
} |
| 645 |
|
| 646 |
// ftp:// is never how a page links to something; it is how a payload |
| 647 |
// is fetched. |
| 648 |
if ( 0 === stripos( $url, 'ftp://' ) ) { |
| 649 |
return true; |
| 650 |
} |
| 651 |
|
| 652 |
// Truncation of the suffix the vulnerable script appends. |
| 653 |
if ( '?' === substr( $url, -1 ) || false !== stripos( $url, '%00' ) || false !== strpos( $url, "\0" ) ) { |
| 654 |
return true; |
| 655 |
} |
| 656 |
|
| 657 |
$path = (string) wp_parse_url( $url, PHP_URL_PATH ); |
| 658 |
|
| 659 |
if ( '' === $path ) { |
| 660 |
return false; |
| 661 |
} |
| 662 |
|
| 663 |
$extension = strtolower( pathinfo( $path, PATHINFO_EXTENSION ) ); |
| 664 |
|
| 665 |
return ( '' !== $extension && in_array( $extension, self::$inclusion_extensions, true ) ); |
| 666 |
} |
| 667 |
|
| 668 |
/** |
| 669 |
* The REST route of the current request, or '' when it is not a REST call |
| 670 |
* |
| 671 |
* Read from the request itself because REST_REQUEST is not defined yet: |
| 672 |
* the firewall runs on init, and rest_api_loaded() defines it later, on |
| 673 |
* parse_request. Both shapes are covered, the pretty /wp-json/<route> and |
| 674 |
* the plain ?rest_route=<route>, and the prefix is asked for rather than |
| 675 |
* assumed, since rest_url_prefix filters it. |
| 676 |
* |
| 677 |
* @since 2.11.1 |
| 678 |
* |
| 679 |
* @return string Route with a leading slash, or '' when there is none. |
| 680 |
*/ |
| 681 |
private function current_rest_route() { |
| 682 |
$params = array(); |
| 683 |
parse_str( (string) ( $this->request_data['query_raw'] ?? '' ), $params ); |
| 684 |
|
| 685 |
if ( isset( $params['rest_route'] ) && is_string( $params['rest_route'] ) ) { |
| 686 |
return '/' . ltrim( $params['rest_route'], '/' ); |
| 687 |
} |
| 688 |
|
| 689 |
$uri = (string) ( $this->request_data['uri_raw'] ?? '' ); |
| 690 |
$path = (string) wp_parse_url( $uri, PHP_URL_PATH ); |
| 691 |
$needle = '/' . trim( rest_get_url_prefix(), '/' ) . '/'; |
| 692 |
$at = strpos( $path, $needle ); |
| 693 |
|
| 694 |
if ( false === $at ) { |
| 695 |
return ''; |
| 696 |
} |
| 697 |
|
| 698 |
return '/' . ltrim( substr( $path, $at + strlen( $needle ) ), '/' ); |
| 699 |
} |
| 700 |
|
| 701 |
/** |
| 702 |
* Whether this is a logged-in editor asking core to resolve an embed |
| 703 |
* |
| 704 |
* /wp-json/oembed/1.0/proxy is where the block editor sends the URL the |
| 705 |
* author pasted, so an external URL there is the request, not an attack. |
| 706 |
* The exemption is not the route on its own: it asks for the capability |
| 707 |
* that route's own permission_callback asks for, so an anonymous scanner |
| 708 |
* probing it is still blocked and still logged. The other core embed |
| 709 |
* route, /oembed/1.0/embed, only ever answers for this site's own URLs, |
| 710 |
* which the check already leaves alone. |
| 711 |
* |
| 712 |
* @since 2.11.1 |
| 713 |
* |
| 714 |
* @return bool |
| 715 |
*/ |
| 716 |
private function is_core_embed_proxy_request() { |
| 717 |
if ( 0 !== strpos( $this->current_rest_route(), '/oembed/1.0/proxy' ) ) { |
| 718 |
return false; |
| 719 |
} |
| 720 |
|
| 721 |
return ( is_user_logged_in() && current_user_can( 'edit_posts' ) ); |
| 722 |
} |
| 723 |
|
| 724 |
/** |
| 725 |
* Host in a comparable form: lowercase and without a leading www. |
| 726 |
* |
| 727 |
* @since 2.9.9 |
| 728 |
* |
| 729 |
* @param string|null $host Host to normalize. |
| 730 |
* @return string |
| 731 |
*/ |
| 732 |
private function normalize_host( $host ) { |
| 733 |
$host = strtolower( trim( (string) $host ) ); |
| 734 |
|
| 735 |
return ( 0 === strpos( $host, 'www.' ) ) ? substr( $host, 4 ) : $host; |
| 736 |
} |
| 737 |
|
| 738 |
/** |
| 739 |
* Check for directory traversal attacks |
| 740 |
* |
| 741 |
* @return string|false Error message or false if safe. |
| 742 |
*/ |
| 743 |
private function check_directory_traversal() { |
| 744 |
$combined = $this->inspection_haystack(); |
| 745 |
|
| 746 |
if ( empty( $combined ) ) { |
| 747 |
return false; |
| 748 |
} |
| 749 |
|
| 750 |
// Directory traversal patterns |
| 751 |
$patterns = array( |
| 752 |
'/\.\.\//i' => __( 'Directory traversal detected', 'vigilante' ), |
| 753 |
'/\.\.%2f/i' => __( 'Encoded traversal detected', 'vigilante' ), |
| 754 |
'/%2e%2e\//i' => __( 'Double encoded traversal', 'vigilante' ), |
| 755 |
'/\.\.%5c/i' => __( 'Backslash traversal detected', 'vigilante' ), |
| 756 |
); |
| 757 |
|
| 758 |
foreach ( $patterns as $pattern => $message ) { |
| 759 |
if ( preg_match( $pattern, $combined ) ) { |
| 760 |
return $message; |
| 761 |
} |
| 762 |
} |
| 763 |
|
| 764 |
return false; |
| 765 |
} |
| 766 |
|
| 767 |
/** |
| 768 |
* Check for PHP execution in uploads |
| 769 |
* |
| 770 |
* @return string|false Error message or false if safe. |
| 771 |
*/ |
| 772 |
private function check_php_in_uploads() { |
| 773 |
$uri = $this->inspection_haystack(); |
| 774 |
|
| 775 |
// Check if accessing PHP in uploads directory |
| 776 |
if ( preg_match( '/\/wp-content\/uploads\/.*\.ph(p[345s]?|tml)/i', $uri ) ) { |
| 777 |
return __( 'PHP execution in uploads blocked', 'vigilante' ); |
| 778 |
} |
| 779 |
|
| 780 |
return false; |
| 781 |
} |
| 782 |
|
| 783 |
/** |
| 784 |
* Check for bad bots |
| 785 |
* |
| 786 |
* @return string|false Error message or false if safe. |
| 787 |
*/ |
| 788 |
private function check_bad_bots() { |
| 789 |
$user_agent = strtolower( $this->request_data['user_agent'] ); |
| 790 |
|
| 791 |
if ( empty( $user_agent ) ) { |
| 792 |
return false; |
| 793 |
} |
| 794 |
|
| 795 |
// Known malicious bots and scanners |
| 796 |
// NOTE: Matching is done via strpos() on the full User-Agent string, |
| 797 |
// so entries must be specific enough to avoid false positives with |
| 798 |
// legitimate services, plugins, or WordPress loopback requests. |
| 799 |
// Generic short words (e.g. 'scan', 'ninja', 'titan') must stay out |
| 800 |
// of BOTH this list and the htaccess one: the htaccess regex matches |
| 801 |
// bare substrings too, and unlike this layer it runs before PHP, so |
| 802 |
// the ua_whitelist cannot rescue a false positive there. |
| 803 |
$bad_bots = array( |
| 804 |
'ahrefsbot', |
| 805 |
'semrushbot', |
| 806 |
'dotbot', |
| 807 |
'mj12bot', |
| 808 |
'blexbot', |
| 809 |
'linkdexbot', |
| 810 |
'aspiegelbot', |
| 811 |
'alexibot', |
| 812 |
'backlink', |
| 813 |
'bandit', |
| 814 |
'batchftp', |
| 815 |
'bigfoot', |
| 816 |
'blackwidow', |
| 817 |
'blowfish', |
| 818 |
'botalot', |
| 819 |
'builtbottough', |
| 820 |
'bullseye', |
| 821 |
'cheesebot', |
| 822 |
'cherrypicker', |
| 823 |
'chinaclaw', |
| 824 |
'copyrightcheck', |
| 825 |
'crescent', |
| 826 |
'curl/', |
| 827 |
'dittospyder', |
| 828 |
'dragonfly', |
| 829 |
'easydl', |
| 830 |
'ebingbong', |
| 831 |
'ecatch', |
| 832 |
'eirgrabber', |
| 833 |
'emailcollector', |
| 834 |
'emailsiphon', |
| 835 |
'emailwolf', |
| 836 |
'erocrawler', |
| 837 |
'exabot', |
| 838 |
'expressweb', |
| 839 |
'eyenetie', |
| 840 |
'flashget', |
| 841 |
'flunky', |
| 842 |
'frontpage', |
| 843 |
'getright', |
| 844 |
'getweb', |
| 845 |
'go-ahead-got-it', |
| 846 |
'gotit', |
| 847 |
'grabnet', |
| 848 |
'grafula', |
| 849 |
'harvest', |
| 850 |
'hloader', |
| 851 |
'hmview', |
| 852 |
'httplib', |
| 853 |
'httrack', |
| 854 |
'humanlinks', |
| 855 |
'ia_archiver', |
| 856 |
'imagestripper', |
| 857 |
'imagesucker', |
| 858 |
'indy library', |
| 859 |
'infonavirobot', |
| 860 |
'infotekies', |
| 861 |
'intelliseek', |
| 862 |
'interget', |
| 863 |
'intraformant', |
| 864 |
'jakarta', |
| 865 |
'jennybot', |
| 866 |
'jetcar', |
| 867 |
'kenjin', |
| 868 |
'larbin', |
| 869 |
'leechftp', |
| 870 |
'lexibot', |
| 871 |
'libweb', |
| 872 |
'likse', |
| 873 |
'linkscan', |
| 874 |
'linkwalker', |
| 875 |
'lnspiderguy', |
| 876 |
'lwp', |
| 877 |
'magnet', |
| 878 |
'mag-net', |
| 879 |
'markwatch', |
| 880 |
'mass downloader', |
| 881 |
'masscan', |
| 882 |
'microsoft.url', |
| 883 |
'midown', |
| 884 |
'miixpc', |
| 885 |
'missigua', |
| 886 |
'moget', |
| 887 |
'nameprotect', |
| 888 |
'navroad', |
| 889 |
'nearsite', |
| 890 |
'net vampire', |
| 891 |
'netants', |
| 892 |
'netcraft', |
| 893 |
'netmechanic', |
| 894 |
'netspider', |
| 895 |
'nextgensearchbot', |
| 896 |
'nibbler', |
| 897 |
'nicerspro', |
| 898 |
'niki-bot', |
| 899 |
'npbot', |
| 900 |
'offline explorer', |
| 901 |
'offline navigator', |
| 902 |
'openfind', |
| 903 |
'outfoxbot', |
| 904 |
'pagegrabber', |
| 905 |
'pavuk', |
| 906 |
'pcbrowser', |
| 907 |
'php/', |
| 908 |
'pockey', |
| 909 |
'prowebwalker', |
| 910 |
'psycheclone', |
| 911 |
'python-urllib', |
| 912 |
'python-requests', |
| 913 |
'python/', |
| 914 |
'queryn', |
| 915 |
'reget', |
| 916 |
'repomonkey', |
| 917 |
'siphon', |
| 918 |
'siteexplorer', |
| 919 |
'sitesnagger', |
| 920 |
'slurp', |
| 921 |
'smartdownload', |
| 922 |
'snapbot', |
| 923 |
'snoopy', |
| 924 |
'sogou', |
| 925 |
'spacebison', |
| 926 |
'spankbot', |
| 927 |
'sqworm', |
| 928 |
'superbot', |
| 929 |
'superhttp', |
| 930 |
'surfbot', |
| 931 |
'suzuran', |
| 932 |
'szukacz', |
| 933 |
'takeout', |
| 934 |
'teleport', |
| 935 |
'telesoft', |
| 936 |
'thenomad', |
| 937 |
'tighttwatbot', |
| 938 |
'true_robot', |
| 939 |
'turingos', |
| 940 |
'turnitinbot', |
| 941 |
'voideye', |
| 942 |
'webalta', |
| 943 |
'webbandit', |
| 944 |
'webcollector', |
| 945 |
'webcopier', |
| 946 |
'webdup', |
| 947 |
'webenhancer', |
| 948 |
'webfetch', |
| 949 |
'webgo', |
| 950 |
'webmasterworldforumbot', |
| 951 |
'webpictures', |
| 952 |
'webreaper', |
| 953 |
'websauger', |
| 954 |
'webspider', |
| 955 |
'webstripper', |
| 956 |
'websucker', |
| 957 |
'webwhacker', |
| 958 |
'webzip', |
| 959 |
'widow', |
| 960 |
'wisenut', |
| 961 |
'wwwoffle', |
| 962 |
'xaldon', |
| 963 |
'xxxyy', |
| 964 |
'zeus', |
| 965 |
'zermelo', |
| 966 |
'zyborg', |
| 967 |
); |
| 968 |
|
| 969 |
foreach ( $bad_bots as $bot ) { |
| 970 |
if ( strpos( $user_agent, $bot ) !== false ) { |
| 971 |
return sprintf( |
| 972 |
/* translators: %s: Bot name */ |
| 973 |
__( 'Bad bot blocked: %s', 'vigilante' ), |
| 974 |
$bot |
| 975 |
); |
| 976 |
} |
| 977 |
} |
| 978 |
|
| 979 |
return false; |
| 980 |
} |
| 981 |
|
| 982 |
/** |
| 983 |
* Check for empty user agent |
| 984 |
* |
| 985 |
* @return string|false Error message or false if safe. |
| 986 |
*/ |
| 987 |
private function check_empty_user_agent() { |
| 988 |
if ( empty( $this->request_data['user_agent'] ) ) { |
| 989 |
return __( 'Empty user agent blocked', 'vigilante' ); |
| 990 |
} |
| 991 |
return false; |
| 992 |
} |
| 993 |
|
| 994 |
/** |
| 995 |
* Check HTTP method |
| 996 |
* |
| 997 |
* Logged-in users with edit capabilities are excluded to ensure |
| 998 |
* Gutenberg, REST API, and page builders work correctly. |
| 999 |
*/ |
| 1000 |
private function check_http_method() { |
| 1001 |
// Skip for authenticated users who can edit content |
| 1002 |
// They need OPTIONS, PUT, PATCH, DELETE for Gutenberg, REST API, and page builders |
| 1003 |
if ( is_user_logged_in() && current_user_can( 'edit_posts' ) ) { |
| 1004 |
return; |
| 1005 |
} |
| 1006 |
|
| 1007 |
// Skip for WordPress REST API requests |
| 1008 |
// The REST API uses PUT, DELETE, PATCH for legitimate operations and has its own |
| 1009 |
// authentication and authorization layer — no need to filter methods here |
| 1010 |
$rest_prefix = rest_get_url_prefix(); // Typically 'wp-json' |
| 1011 |
if ( false !== strpos( $this->request_data['uri'], '/' . $rest_prefix . '/' ) ) { |
| 1012 |
return; |
| 1013 |
} |
| 1014 |
|
| 1015 |
$method = strtoupper( $this->request_data['method'] ); |
| 1016 |
$allowed_methods = isset( $this->options['allowed_http_methods'] ) |
| 1017 |
? $this->options['allowed_http_methods'] |
| 1018 |
: array( 'GET', 'POST', 'HEAD', 'OPTIONS', 'PUT', 'PATCH', 'DELETE' ); |
| 1019 |
$allowed = array_map( 'strtoupper', $allowed_methods ); |
| 1020 |
|
| 1021 |
if ( ! in_array( $method, $allowed, true ) ) { |
| 1022 |
$this->block_request( |
| 1023 |
'http_method', |
| 1024 |
sprintf( |
| 1025 |
/* translators: %s: HTTP method */ |
| 1026 |
__( 'HTTP method %s not allowed', 'vigilante' ), |
| 1027 |
$method |
| 1028 |
) |
| 1029 |
); |
| 1030 |
} |
| 1031 |
} |
| 1032 |
|
| 1033 |
/** |
| 1034 |
* Upper bound of the vigilante_firewall_blocks index |
| 1035 |
* |
| 1036 |
* The index only feeds the admin screen; enforcement reads a transient per |
| 1037 |
* IP. Under a distributed attack the oldest entries are dropped first, so |
| 1038 |
* the option cannot grow without limit (S6). |
| 1039 |
* |
| 1040 |
* @since 2.11.0 |
| 1041 |
*/ |
| 1042 |
const MAX_TRACKED_BLOCKS = 500; |
| 1043 |
|
| 1044 |
/** |
| 1045 |
* Add a block to the bounded admin index |
| 1046 |
* |
| 1047 |
* Prunes expired entries on every write, not only when an administrator |
| 1048 |
* opens the Firewall tab, and keeps at most MAX_TRACKED_BLOCKS entries, |
| 1049 |
* dropping the oldest by blocked_at. |
| 1050 |
* |
| 1051 |
* @since 2.11.0 |
| 1052 |
* |
| 1053 |
* @param string $ip Blocked address. |
| 1054 |
* @param array $block Block data (expires, blocked_at, duration, reason, strikes). |
| 1055 |
*/ |
| 1056 |
private static function index_block( $ip, $block ) { |
| 1057 |
$blocks = get_option( 'vigilante_firewall_blocks', array() ); |
| 1058 |
$now = time(); |
| 1059 |
|
| 1060 |
if ( ! is_array( $blocks ) ) { |
| 1061 |
$blocks = array(); |
| 1062 |
} |
| 1063 |
|
| 1064 |
foreach ( $blocks as $blocked_ip => $data ) { |
| 1065 |
if ( ! is_array( $data ) || ! isset( $data['expires'] ) || $now >= (int) $data['expires'] ) { |
| 1066 |
unset( $blocks[ $blocked_ip ] ); |
| 1067 |
} |
| 1068 |
} |
| 1069 |
|
| 1070 |
$blocks[ $ip ] = $block; |
| 1071 |
|
| 1072 |
if ( count( $blocks ) > self::MAX_TRACKED_BLOCKS ) { |
| 1073 |
uasort( |
| 1074 |
$blocks, |
| 1075 |
static function ( $a, $b ) { |
| 1076 |
return (int) ( $a['blocked_at'] ?? 0 ) <=> (int) ( $b['blocked_at'] ?? 0 ); |
| 1077 |
} |
| 1078 |
); |
| 1079 |
$blocks = array_slice( $blocks, count( $blocks ) - self::MAX_TRACKED_BLOCKS, null, true ); |
| 1080 |
} |
| 1081 |
|
| 1082 |
update_option( 'vigilante_firewall_blocks', $blocks, false ); |
| 1083 |
} |
| 1084 |
|
| 1085 |
/** |
| 1086 |
* Check rate limiting |
| 1087 |
*/ |
| 1088 |
public function check_rate_limit() { |
| 1089 |
// Skip rate limiting for whitelisted IPs |
| 1090 |
if ( $this->is_ip_whitelisted() ) { |
| 1091 |
return; |
| 1092 |
} |
| 1093 |
|
| 1094 |
// Skip rate limiting for logged-in administrators |
| 1095 |
if ( is_user_logged_in() && current_user_can( 'manage_options' ) ) { |
| 1096 |
return; |
| 1097 |
} |
| 1098 |
|
| 1099 |
// Allow other modules to opt out — Under Attack mode uses this so that |
| 1100 |
// visitors who already passed the JS challenge don't burn the |
| 1101 |
// aggressive 30 req/min cap loading a normal page's assets. |
| 1102 |
if ( apply_filters( 'vigilante_skip_rate_limit', false ) ) { |
| 1103 |
return; |
| 1104 |
} |
| 1105 |
|
| 1106 |
$ip = $this->get_client_ip(); |
| 1107 |
$rate_limit = $this->options['rate_limiting']; |
| 1108 |
|
| 1109 |
// Check if already blocked (fast path). The active block lives in a |
| 1110 |
// transient keyed by IP, so this path, which runs on every |
| 1111 |
// unauthenticated request, reads one row and not the whole index of |
| 1112 |
// blocked addresses. Until 2.11.0 it loaded vigilante_firewall_blocks |
| 1113 |
// entire, an array with no upper bound that a distributed attack grew |
| 1114 |
// by one entry per new address, so the firewall amplified the attack it |
| 1115 |
// was blocking (S6). The transient expires with the block itself. |
| 1116 |
$block = get_transient( 'vigilante_rate_block_' . md5( $ip ) ); |
| 1117 |
if ( is_array( $block ) && isset( $block['expires'] ) && time() < (int) $block['expires'] ) { |
| 1118 |
if ( ! headers_sent() ) { |
| 1119 |
status_header( 429 ); |
| 1120 |
nocache_headers(); |
| 1121 |
} |
| 1122 |
wp_die( |
| 1123 |
esc_html__( 'Rate limit exceeded. Please try again later.', 'vigilante' ), |
| 1124 |
esc_html__( 'Too Many Requests', 'vigilante' ), |
| 1125 |
array( 'response' => 429 ) |
| 1126 |
); |
| 1127 |
} |
| 1128 |
|
| 1129 |
$max_requests = absint( $rate_limit['requests_per_minute'] ); |
| 1130 |
|
| 1131 |
// Allow Under Attack mode (or other filters) to override threshold |
| 1132 |
$max_requests = absint( apply_filters( 'vigilante_rate_limit_requests', $max_requests ) ); |
| 1133 |
|
| 1134 |
// Fixed window, anchored to the timestamp of its first request. |
| 1135 |
// |
| 1136 |
// The count used to live in a transient whose TTL was renewed on every |
| 1137 |
// hit, which is a window that never closes: any IP going less than 60 s |
| 1138 |
// between requests kept accumulating, so the effective limit was not |
| 1139 |
// "requests per minute" but "requests since the last full minute of |
| 1140 |
// silence". A logged-in editor publishing several posts in a row could |
| 1141 |
// pile up 150+ requests while never exceeding 60 in any single minute, |
| 1142 |
// and got a 429. Storing the window start makes the reset explicit |
| 1143 |
// instead of relying on the transient expiring. |
| 1144 |
$transient_key = 'vigilante_rate_' . md5( $ip ); |
| 1145 |
$window = get_transient( $transient_key ); |
| 1146 |
$now = time(); |
| 1147 |
|
| 1148 |
// Counts stored before 2.9.5 were a bare integer with no window start. |
| 1149 |
// There is no way to tell how old such a count is, so open a new window. |
| 1150 |
if ( ! is_array( $window ) || ! isset( $window['start'], $window['count'] ) ) { |
| 1151 |
$window = array( |
| 1152 |
'start' => $now, |
| 1153 |
'count' => 0, |
| 1154 |
); |
| 1155 |
} |
| 1156 |
|
| 1157 |
// Window elapsed: start counting again, even under continuous traffic. |
| 1158 |
if ( ( $now - absint( $window['start'] ) ) >= self::RATE_LIMIT_WINDOW ) { |
| 1159 |
$window = array( |
| 1160 |
'start' => $now, |
| 1161 |
'count' => 0, |
| 1162 |
); |
| 1163 |
} |
| 1164 |
|
| 1165 |
// Count this request, then allow up to $max_requests per window. |
| 1166 |
$window['count'] = absint( $window['count'] ) + 1; |
| 1167 |
$request_count = $window['count']; |
| 1168 |
|
| 1169 |
if ( $request_count > $max_requests ) { |
| 1170 |
$base_duration = absint( $rate_limit['block_duration'] ); |
| 1171 |
|
| 1172 |
// Allow Under Attack mode (or other filters) to override duration |
| 1173 |
$base_duration = absint( apply_filters( 'vigilante_rate_limit_duration', $base_duration ) ); |
| 1174 |
|
| 1175 |
$duration = $base_duration; |
| 1176 |
$strikes = 1; |
| 1177 |
|
| 1178 |
// Progressive blocking: double duration on each repeat offense |
| 1179 |
if ( ! empty( $rate_limit['progressive'] ) ) { |
| 1180 |
$strikes_key = 'vigilante_strikes_' . md5( $ip ); |
| 1181 |
$strikes = absint( get_transient( $strikes_key ) ) + 1; |
| 1182 |
|
| 1183 |
$max_duration = absint( $rate_limit['max_block_duration'] ?? 86400 ); |
| 1184 |
$duration = min( |
| 1185 |
$base_duration * pow( 2, $strikes - 1 ), |
| 1186 |
$max_duration |
| 1187 |
); |
| 1188 |
|
| 1189 |
// Persist strikes for 24h so they accumulate across blocks |
| 1190 |
set_transient( $strikes_key, $strikes, 86400 ); |
| 1191 |
} |
| 1192 |
|
| 1193 |
$block = array( |
| 1194 |
'expires' => time() + $duration, |
| 1195 |
'blocked_at' => time(), |
| 1196 |
'duration' => $duration, |
| 1197 |
'reason' => 'rate_limit', |
| 1198 |
'strikes' => $strikes, |
| 1199 |
); |
| 1200 |
|
| 1201 |
// The block itself, read by the fast path above on every request. |
| 1202 |
set_transient( 'vigilante_rate_block_' . md5( $ip ), $block, $duration ); |
| 1203 |
|
| 1204 |
// The bounded index the admin screen lists. |
| 1205 |
self::index_block( $ip, $block ); |
| 1206 |
|
| 1207 |
$this->block_request( 'rate_limit', __( 'Rate limit exceeded. Please try again later.', 'vigilante' ), 429 ); |
| 1208 |
} |
| 1209 |
|
| 1210 |
// The TTL only garbage-collects the payload once the IP goes quiet; what |
| 1211 |
// bounds the count is the window reset above, not the expiry. |
| 1212 |
set_transient( $transient_key, $window, self::RATE_LIMIT_WINDOW ); |
| 1213 |
} |
| 1214 |
|
| 1215 |
/** |
| 1216 |
* Block a request |
| 1217 |
* |
| 1218 |
* @param string $reason Reason code for blocking. |
| 1219 |
* @param string $message Message to log. |
| 1220 |
* @param int $status_code HTTP status code. |
| 1221 |
*/ |
| 1222 |
private function block_request( $reason, $message, $status_code = 403 ) { |
| 1223 |
// Log the block |
| 1224 |
if ( $this->activity_log ) { |
| 1225 |
$this->activity_log->log( |
| 1226 |
'firewall', |
| 1227 |
'blocked', |
| 1228 |
$message, |
| 1229 |
array( |
| 1230 |
'reason' => $reason, |
| 1231 |
'request_uri' => $this->loggable_uri(), |
| 1232 |
'ip' => $this->get_client_ip(), |
| 1233 |
'user_agent' => $this->request_data['user_agent'] ?? '', |
| 1234 |
), |
| 1235 |
'warning' |
| 1236 |
); |
| 1237 |
} |
| 1238 |
|
| 1239 |
// Set response headers |
| 1240 |
if ( ! headers_sent() ) { |
| 1241 |
status_header( $status_code ); |
| 1242 |
nocache_headers(); |
| 1243 |
} |
| 1244 |
|
| 1245 |
// A REST client gets the refusal in the shape it can read. Until |
| 1246 |
// 2.11.0 every block answered with the HTML "Forbidden" page, so the |
| 1247 |
// block editor could only show its own generic message and the reason |
| 1248 |
// was reachable only by opening the activity log. Same status code, |
| 1249 |
// same message, the envelope core uses for an error. |
| 1250 |
if ( '' !== $this->current_rest_route() ) { |
| 1251 |
wp_send_json( |
| 1252 |
array( |
| 1253 |
'code' => 'vigilante_firewall_blocked', |
| 1254 |
'message' => $message, |
| 1255 |
'data' => array( 'status' => $status_code ), |
| 1256 |
), |
| 1257 |
$status_code |
| 1258 |
); |
| 1259 |
} |
| 1260 |
|
| 1261 |
// Return appropriate response |
| 1262 |
if ( 429 === $status_code ) { |
| 1263 |
wp_die( |
| 1264 |
esc_html( $message ), |
| 1265 |
esc_html__( 'Too Many Requests', 'vigilante' ), |
| 1266 |
array( 'response' => 429 ) |
| 1267 |
); |
| 1268 |
} |
| 1269 |
|
| 1270 |
wp_die( |
| 1271 |
esc_html( $message ), |
| 1272 |
esc_html__( 'Forbidden', 'vigilante' ), |
| 1273 |
array( 'response' => 403 ) |
| 1274 |
); |
| 1275 |
} |
| 1276 |
|
| 1277 |
/** |
| 1278 |
* Upper bound of the address stored with a logged block |
| 1279 |
* |
| 1280 |
* @since 2.11.1 |
| 1281 |
*/ |
| 1282 |
const MAX_LOGGED_URI = 512; |
| 1283 |
|
| 1284 |
/** |
| 1285 |
* The blocked address, in a form that still says what was blocked |
| 1286 |
* |
| 1287 |
* The copy kept for logging goes through sanitize_text_field(), which |
| 1288 |
* deletes every %XX sequence instead of decoding it. A browser percent |
| 1289 |
* encodes the URL it puts in a parameter, so a blocked embed was recorded |
| 1290 |
* as "/wp-json/oembed/1.0/proxy?url=httpswww.youtube.comwatchv..." and the |
| 1291 |
* owner could not tell what the request had been. Reported on 9 sep 2026. |
| 1292 |
* |
| 1293 |
* Nothing is sanitized away here beyond control characters, and that is |
| 1294 |
* the point. The value is stored, never executed: it is escaped where it |
| 1295 |
* is shown, by escapeHtml() in the log detail and by csvCell() in the |
| 1296 |
* export. Invalid UTF-8 is stripped because wp_json_encode() returns false |
| 1297 |
* on it, which would have thrown away the whole entry's context. |
| 1298 |
* |
| 1299 |
* @since 2.11.1 |
| 1300 |
* |
| 1301 |
* @return string |
| 1302 |
*/ |
| 1303 |
private function loggable_uri() { |
| 1304 |
$uri = (string) ( $this->request_data['uri_raw'] ?? '' ); |
| 1305 |
$uri = (string) preg_replace( '/[\x00-\x1F\x7F]/', '', $uri ); |
| 1306 |
$uri = wp_check_invalid_utf8( $uri, true ); |
| 1307 |
|
| 1308 |
if ( strlen( $uri ) > self::MAX_LOGGED_URI ) { |
| 1309 |
$uri = substr( $uri, 0, self::MAX_LOGGED_URI ) . '...'; |
| 1310 |
} |
| 1311 |
|
| 1312 |
return $uri; |
| 1313 |
} |
| 1314 |
|
| 1315 |
/** |
| 1316 |
* Check if current IP is whitelisted |
| 1317 |
* |
| 1318 |
* @return bool |
| 1319 |
*/ |
| 1320 |
private function is_ip_whitelisted() { |
| 1321 |
$whitelist = $this->options['ip_whitelist'] ?? array(); |
| 1322 |
|
| 1323 |
return Vigilante_IP_Utils::in_list( $this->get_client_ip(), $whitelist ); |
| 1324 |
} |
| 1325 |
|
| 1326 |
/** |
| 1327 |
* Check if current IP is blacklisted |
| 1328 |
* |
| 1329 |
* @return bool |
| 1330 |
*/ |
| 1331 |
private function is_ip_blacklisted() { |
| 1332 |
$blacklist = $this->options['ip_blacklist'] ?? array(); |
| 1333 |
|
| 1334 |
return Vigilante_IP_Utils::in_list( $this->get_client_ip(), $blacklist ); |
| 1335 |
} |
| 1336 |
|
| 1337 |
/** |
| 1338 |
* Check if current User-Agent is whitelisted |
| 1339 |
* |
| 1340 |
* Partial matching: if the request UA contains any whitelisted string, |
| 1341 |
* it bypasses all firewall checks. Useful for services like ManageWP, MainWP, etc. |
| 1342 |
* |
| 1343 |
* @return bool |
| 1344 |
*/ |
| 1345 |
private function is_ua_whitelisted() { |
| 1346 |
$whitelist = $this->options['ua_whitelist'] ?? array(); |
| 1347 |
|
| 1348 |
if ( empty( $whitelist ) ) { |
| 1349 |
return false; |
| 1350 |
} |
| 1351 |
|
| 1352 |
$user_agent = isset( $_SERVER['HTTP_USER_AGENT'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) : ''; |
| 1353 |
|
| 1354 |
if ( empty( $user_agent ) ) { |
| 1355 |
return false; |
| 1356 |
} |
| 1357 |
|
| 1358 |
$ua_lower = strtolower( $user_agent ); |
| 1359 |
|
| 1360 |
foreach ( $whitelist as $allowed ) { |
| 1361 |
$allowed = trim( $allowed ); |
| 1362 |
if ( ! empty( $allowed ) && false !== strpos( $ua_lower, strtolower( $allowed ) ) ) { |
| 1363 |
return true; |
| 1364 |
} |
| 1365 |
} |
| 1366 |
|
| 1367 |
return false; |
| 1368 |
} |
| 1369 |
|
| 1370 |
/** |
| 1371 |
* Check if current User-Agent is blacklisted |
| 1372 |
* |
| 1373 |
* Partial matching: if the request UA contains any blacklisted string, block it. |
| 1374 |
* |
| 1375 |
* @return bool |
| 1376 |
*/ |
| 1377 |
private function is_ua_blacklisted() { |
| 1378 |
$blacklist = $this->options['ua_blacklist'] ?? array(); |
| 1379 |
|
| 1380 |
if ( empty( $blacklist ) ) { |
| 1381 |
return false; |
| 1382 |
} |
| 1383 |
|
| 1384 |
$user_agent = $this->request_data['user_agent'] ?? ''; |
| 1385 |
|
| 1386 |
if ( empty( $user_agent ) ) { |
| 1387 |
return false; |
| 1388 |
} |
| 1389 |
|
| 1390 |
$ua_lower = strtolower( $user_agent ); |
| 1391 |
|
| 1392 |
foreach ( $blacklist as $blocked ) { |
| 1393 |
$blocked = trim( $blocked ); |
| 1394 |
if ( ! empty( $blocked ) && false !== strpos( $ua_lower, strtolower( $blocked ) ) ) { |
| 1395 |
return true; |
| 1396 |
} |
| 1397 |
} |
| 1398 |
|
| 1399 |
return false; |
| 1400 |
} |
| 1401 |
|
| 1402 |
/** |
| 1403 |
* Get client IP address |
| 1404 |
* |
| 1405 |
* Delegates to the shared resolver, which only trusts REMOTE_ADDR unless a |
| 1406 |
* proxy header has been explicitly declared in settings. |
| 1407 |
* |
| 1408 |
* @return string |
| 1409 |
*/ |
| 1410 |
private function get_client_ip() { |
| 1411 |
return Vigilante_IP_Utils::get_client_ip(); |
| 1412 |
} |
| 1413 |
|
| 1414 |
// ========================================================================= |
| 1415 |
// BLOCK MANAGEMENT (static, for admin UI) |
| 1416 |
// ========================================================================= |
| 1417 |
|
| 1418 |
/** |
| 1419 |
* Get currently active firewall blocks |
| 1420 |
* |
| 1421 |
* Cleans expired entries on each call. |
| 1422 |
* |
| 1423 |
* @return array Active blocks keyed by IP address. |
| 1424 |
*/ |
| 1425 |
public static function get_active_blocks() { |
| 1426 |
$blocks = get_option( 'vigilante_firewall_blocks', array() ); |
| 1427 |
$now = time(); |
| 1428 |
$dirty = false; |
| 1429 |
|
| 1430 |
foreach ( $blocks as $ip => $data ) { |
| 1431 |
if ( $now >= $data['expires'] ) { |
| 1432 |
unset( $blocks[ $ip ] ); |
| 1433 |
$dirty = true; |
| 1434 |
} |
| 1435 |
} |
| 1436 |
|
| 1437 |
if ( $dirty ) { |
| 1438 |
update_option( 'vigilante_firewall_blocks', $blocks, false ); |
| 1439 |
} |
| 1440 |
|
| 1441 |
return $blocks; |
| 1442 |
} |
| 1443 |
|
| 1444 |
/** |
| 1445 |
* Manually unblock an IP from rate limit blocks |
| 1446 |
* |
| 1447 |
* @param string $ip IP address to unblock. |
| 1448 |
* @return bool Whether the IP was found and removed. |
| 1449 |
*/ |
| 1450 |
public static function unblock_ip( $ip ) { |
| 1451 |
$blocks = get_option( 'vigilante_firewall_blocks', array() ); |
| 1452 |
|
| 1453 |
if ( ! isset( $blocks[ $ip ] ) ) { |
| 1454 |
return false; |
| 1455 |
} |
| 1456 |
|
| 1457 |
unset( $blocks[ $ip ] ); |
| 1458 |
update_option( 'vigilante_firewall_blocks', $blocks, false ); |
| 1459 |
|
| 1460 |
// Clean related transients |
| 1461 |
$hash = md5( $ip ); |
| 1462 |
delete_transient( 'vigilante_rate_block_' . $hash ); |
| 1463 |
delete_transient( 'vigilante_rate_' . $hash ); |
| 1464 |
delete_transient( 'vigilante_strikes_' . $hash ); |
| 1465 |
|
| 1466 |
return true; |
| 1467 |
} |
| 1468 |
} |
