Admin
1 week ago
OptimiseAssets
1 week ago
ThirdParty
1 year ago
AdminBar.php
1 week ago
AssetsManager.php
1 week ago
BulkChanges.php
1 week ago
CleanUp.php
2 months ago
Debug.php
1 week ago
DebugException.php
1 week ago
FileSystem.php
1 year ago
HardcodedAssets.php
1 week ago
LoadExceptions.php
1 week ago
Main.php
1 week ago
MainFront.php
1 week ago
Maintenance.php
1 week ago
Menu.php
1 week ago
MetaBoxes.php
1 week ago
Misc.php
1 week ago
MiscArray.php
1 week ago
ObjectCache.php
2 months ago
OwnAssets.php
1 week ago
PluginTracking.php
1 week ago
Preloads.php
1 week ago
Regex.php
1 week ago
Settings.php
1 week ago
Tips.php
1 year ago
Update.php
1 week ago
Regex.php
526 lines
| 1 | <?php |
| 2 | namespace WpAssetCleanUp; |
| 3 | |
| 4 | /** |
| 5 | * Handles rules that can be either: |
| 6 | * - plain strings, e.g. "contact-form-7" or "/wp-content/plugins/plugin/file.js" |
| 7 | * - explicit RegEx patterns, e.g. "#/wp-content/plugins/(plugin-a|plugin-b)/#i" |
| 8 | * - legacy loose RegEx patterns, e.g. "/wd-instagram-feed/(.*?).js" |
| 9 | * |
| 10 | * Important: |
| 11 | * * - "/" is not treated as a RegEx delimiter by default because asset URLs often start with "/wp-content/..." |
| 12 | * * - For request URI / Plugins Manager rules, "/" can be allowed via $allowSlashDelimiter. |
| 13 | * * - For new RegEx rules, prefer "#" as delimiter. |
| 14 | * |
| 15 | * PHP 5.6 compatible. |
| 16 | */ |
| 17 | class Regex |
| 18 | { |
| 19 | /** |
| 20 | * Preferred delimiter when wrapping legacy loose RegEx patterns. |
| 21 | * |
| 22 | * @var string |
| 23 | */ |
| 24 | const DEFAULT_DELIMITER = '#'; |
| 25 | |
| 26 | /** |
| 27 | * Match one rule against a subject. |
| 28 | * |
| 29 | * The rule can be: |
| 30 | * - plain string |
| 31 | * - explicitly delimited RegEx |
| 32 | * - legacy loose RegEx |
| 33 | * |
| 34 | * @param string $rule |
| 35 | * @param string $subject |
| 36 | * @param bool $allowSlashDelimiter |
| 37 | * @param bool $logInvalidPattern |
| 38 | * |
| 39 | * @return bool |
| 40 | */ |
| 41 | public static function matchesRule($rule, $subject, $allowSlashDelimiter = false, $logInvalidPattern = false) |
| 42 | { |
| 43 | $rule = trim((string)$rule); |
| 44 | $subject = (string)$subject; |
| 45 | |
| 46 | if ($rule === '' || $subject === '') { |
| 47 | return false; |
| 48 | } |
| 49 | |
| 50 | if (self::isDelimitedPattern($rule, $allowSlashDelimiter)) { |
| 51 | return self::matchesPattern($rule, $subject, $logInvalidPattern); |
| 52 | } |
| 53 | |
| 54 | if (self::startsLikeExplicitRegex($rule)) { |
| 55 | if ($logInvalidPattern && function_exists('error_log')) { |
| 56 | error_log('"Asset CleanUp Pro" / Invalid RegEx: ' . $rule . ' / Error: malformed explicit RegEx rule'); |
| 57 | } |
| 58 | |
| 59 | return false; |
| 60 | } |
| 61 | |
| 62 | if (self::looksLikeRegex($rule)) { |
| 63 | $wrappedPattern = self::wrapLoosePattern($rule); |
| 64 | |
| 65 | return self::matchesPattern($wrappedPattern, $subject, $logInvalidPattern); |
| 66 | } |
| 67 | |
| 68 | return strpos($subject, $rule) !== false; |
| 69 | } |
| 70 | |
| 71 | /** |
| 72 | * Match multiple rules against a subject. |
| 73 | * |
| 74 | * @param string|array $rules |
| 75 | * @param string $subject |
| 76 | * @param bool $logInvalidPattern |
| 77 | * |
| 78 | * @return bool |
| 79 | */ |
| 80 | public static function matchesAnyRule($rules, $subject, $allowSlashDelimiter = false, $logInvalidPattern = false) |
| 81 | { |
| 82 | foreach (self::splitRules($rules) as $rule) { |
| 83 | if (self::matchesRule($rule, $subject, $allowSlashDelimiter, $logInvalidPattern)) { |
| 84 | return $rule; |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | return false; |
| 89 | } |
| 90 | |
| 91 | /** |
| 92 | * Alias for older code that expects RegEx matching terminology. |
| 93 | * |
| 94 | * @param string|array $rules |
| 95 | * @param string $subject |
| 96 | * |
| 97 | * @return bool |
| 98 | */ |
| 99 | public static function isRegExMatch($rules, $subject) |
| 100 | { |
| 101 | return self::matchesAnyRule($rules, $subject); |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * Check whether a rule is valid. |
| 106 | * |
| 107 | * Plain strings are always valid. |
| 108 | * Explicit RegEx and legacy loose RegEx must compile successfully. |
| 109 | * |
| 110 | * @param string $rule |
| 111 | * @param bool $allowSlashDelimiter |
| 112 | * |
| 113 | * @return bool |
| 114 | */ |
| 115 | public static function isValidRule($rule, $allowSlashDelimiter = false) |
| 116 | { |
| 117 | $rule = trim((string)$rule); |
| 118 | |
| 119 | if ($rule === '') { |
| 120 | return false; |
| 121 | } |
| 122 | |
| 123 | if (self::isDelimitedPattern($rule, $allowSlashDelimiter)) { |
| 124 | return self::isValidPattern($rule); |
| 125 | } |
| 126 | |
| 127 | // If it starts like an explicit RegEx but failed delimiter validation, |
| 128 | // do not keep it as a plain string. |
| 129 | // "/" is intentionally ignored here because it is too ambiguous. |
| 130 | if (self::startsLikeExplicitRegex($rule)) { |
| 131 | return false; |
| 132 | } |
| 133 | |
| 134 | if (self::looksLikeRegex($rule)) { |
| 135 | return self::isValidPattern(self::wrapLoosePattern($rule)); |
| 136 | } |
| 137 | |
| 138 | return true; |
| 139 | } |
| 140 | |
| 141 | /** |
| 142 | * Validate a PCRE pattern. |
| 143 | * |
| 144 | * @param string $pattern |
| 145 | * |
| 146 | * @return bool |
| 147 | */ |
| 148 | public static function isValidPattern($pattern) |
| 149 | { |
| 150 | $pattern = trim((string)$pattern); |
| 151 | |
| 152 | if ($pattern === '') { |
| 153 | return false; |
| 154 | } |
| 155 | |
| 156 | $hasWarning = false; |
| 157 | |
| 158 | set_error_handler(function () use (&$hasWarning) { |
| 159 | $hasWarning = true; |
| 160 | }); |
| 161 | |
| 162 | $result = preg_match($pattern, ''); |
| 163 | |
| 164 | restore_error_handler(); |
| 165 | |
| 166 | return ! $hasWarning && $result !== false && preg_last_error() === PREG_NO_ERROR; |
| 167 | } |
| 168 | |
| 169 | /** |
| 170 | * @param string $pattern |
| 171 | * @param string $subject |
| 172 | * @param bool $logInvalidPattern |
| 173 | * |
| 174 | * @return bool |
| 175 | */ |
| 176 | public static function matchesPattern($pattern, $subject, $logInvalidPattern = false) |
| 177 | { |
| 178 | $pattern = trim((string)$pattern); |
| 179 | |
| 180 | if ($pattern === '') { |
| 181 | return false; |
| 182 | } |
| 183 | |
| 184 | $hasWarning = false; |
| 185 | $errorMessage = ''; |
| 186 | $errorFile = ''; |
| 187 | $errorLine = ''; |
| 188 | |
| 189 | set_error_handler(function ($errno, $errstr, $errfile, $errline) use (&$hasWarning, &$errorMessage, &$errorFile, &$errorLine) { |
| 190 | $hasWarning = true; |
| 191 | $errorMessage = $errstr; |
| 192 | $errorFile = $errfile; |
| 193 | $errorLine = $errline; |
| 194 | }); |
| 195 | |
| 196 | $result = preg_match($pattern, (string)$subject); |
| 197 | |
| 198 | restore_error_handler(); |
| 199 | |
| 200 | if ($hasWarning || $result === false || preg_last_error() !== PREG_NO_ERROR) { |
| 201 | if ($logInvalidPattern && function_exists('error_log')) { |
| 202 | $pregErrorMessage = function_exists('preg_last_error_msg') ? preg_last_error_msg() : 'PREG error code: ' . preg_last_error(); |
| 203 | |
| 204 | error_log( |
| 205 | '"Asset CleanUp Pro" / Invalid RegEx: ' . $pattern . |
| 206 | ' / Error: ' . ($errorMessage !== '' ? $errorMessage : $pregErrorMessage) . |
| 207 | ($errorFile !== '' ? ' / File: ' . $errorFile : '') . |
| 208 | ($errorLine !== '' ? ' / Line: ' . $errorLine : '') |
| 209 | ); |
| 210 | } |
| 211 | |
| 212 | return false; |
| 213 | } |
| 214 | |
| 215 | return $result === 1; |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * Get the delimiters accepted for explicit RegEx rules. |
| 220 | * |
| 221 | * "/" is intentionally excluded because asset URLs often start with "/". |
| 222 | * Paired delimiters such as "()", "[]", "{}" are intentionally excluded |
| 223 | * because rules such as "(plugin-a|plugin-b)" should be treated as legacy loose RegEx. |
| 224 | * |
| 225 | * @param bool $allowSlashDelimiter |
| 226 | * |
| 227 | * @return array |
| 228 | */ |
| 229 | protected static function getAllowedExplicitDelimiters($allowSlashDelimiter = false) |
| 230 | { |
| 231 | $delimiters = array('#', '~', '@', '!', '%'); |
| 232 | |
| 233 | if ($allowSlashDelimiter) { |
| 234 | $delimiters[] = '/'; |
| 235 | } |
| 236 | |
| 237 | return $delimiters; |
| 238 | } |
| 239 | |
| 240 | /** |
| 241 | * Detect rules that appear to start as explicit RegEx patterns, |
| 242 | * but might be malformed. |
| 243 | * |
| 244 | * Example: |
| 245 | * - #valid# |
| 246 | * - #invalid#123 |
| 247 | * - ~something~ |
| 248 | * |
| 249 | * "/" is intentionally ignored because asset URLs often start with "/". |
| 250 | * |
| 251 | * @param string $value |
| 252 | * |
| 253 | * @return bool |
| 254 | */ |
| 255 | public static function startsLikeExplicitRegex($value) |
| 256 | { |
| 257 | $value = trim((string)$value); |
| 258 | |
| 259 | if ($value === '') { |
| 260 | return false; |
| 261 | } |
| 262 | |
| 263 | $delimiter = substr($value, 0, 1); |
| 264 | |
| 265 | // Never treat "/" as a malformed explicit RegEx starter. |
| 266 | // It is too ambiguous because request URIs and asset URLs commonly start with "/". |
| 267 | if ($delimiter === '/') { |
| 268 | return false; |
| 269 | } |
| 270 | |
| 271 | return in_array($delimiter, self::getAllowedExplicitDelimiters(), true); |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * Detect explicit PCRE delimiters. |
| 276 | * |
| 277 | * Important: |
| 278 | * * "/" is intentionally not accepted as a delimiter by default because normal asset URLs |
| 279 | * * often start with "/wp-content/...". It can be allowed via $allowSlashDelimiter |
| 280 | * * for request URI / Plugins Manager rules. |
| 281 | * |
| 282 | * Examples considered RegEx: |
| 283 | * - #something#i |
| 284 | * - ~something~i |
| 285 | * - @something@ |
| 286 | * - !something! |
| 287 | * |
| 288 | * Examples considered plain strings: |
| 289 | * - /wp-content/plugins/plugin/file.js |
| 290 | * - /contact-form-7/ |
| 291 | * |
| 292 | * @param string $value |
| 293 | * @param bool $allowSlashDelimiter |
| 294 | * |
| 295 | * @return bool |
| 296 | */ |
| 297 | public static function isDelimitedPattern($value, $allowSlashDelimiter = false) |
| 298 | { |
| 299 | $value = trim((string)$value); |
| 300 | |
| 301 | if (strlen($value) < 3) { |
| 302 | return false; |
| 303 | } |
| 304 | |
| 305 | $delimiter = substr($value, 0, 1); |
| 306 | |
| 307 | if (! in_array($delimiter, self::getAllowedExplicitDelimiters($allowSlashDelimiter), true)) { |
| 308 | return false; |
| 309 | } |
| 310 | |
| 311 | $closingDelimiterPosition = self::findClosingDelimiterPosition($value, $delimiter); |
| 312 | |
| 313 | if ($closingDelimiterPosition === false || $closingDelimiterPosition < 2) { |
| 314 | return false; |
| 315 | } |
| 316 | |
| 317 | $modifiers = substr($value, $closingDelimiterPosition + 1); |
| 318 | |
| 319 | if ($modifiers !== '' && preg_match('/^[a-zA-Z]*$/', $modifiers) !== 1) { |
| 320 | return false; |
| 321 | } |
| 322 | |
| 323 | return self::isValidPattern($value); |
| 324 | } |
| 325 | |
| 326 | /** |
| 327 | * Detect legacy loose RegEx rules. |
| 328 | * |
| 329 | * Do NOT treat dots alone as RegEx because many asset URLs end in ".js" or ".css". |
| 330 | * |
| 331 | * Examples returning true: |
| 332 | * - /wd-instagram-feed/(.*?).js |
| 333 | * - wp-content/plugins/(plugin-a|plugin-b) |
| 334 | * - ^/wp-content/ |
| 335 | * - \.min\.js$ |
| 336 | * |
| 337 | * Examples returning false: |
| 338 | * - /wp-content/plugins/plugin/file.js |
| 339 | * - contact-form-7 |
| 340 | * |
| 341 | * @param string $value |
| 342 | * |
| 343 | * @return bool |
| 344 | */ |
| 345 | public static function looksLikeRegex($value) |
| 346 | { |
| 347 | $value = trim((string)$value); |
| 348 | |
| 349 | if ($value === '') { |
| 350 | return false; |
| 351 | } |
| 352 | |
| 353 | // Strong RegEx indicators. |
| 354 | // Do not include "?", "*" or "+" here because they can appear in URLs/query strings |
| 355 | // or filenames and should not automatically convert a plain string into RegEx. |
| 356 | if (preg_match('/[\[\]\(\)\{\}\|\^\$]/', $value) === 1) { |
| 357 | return true; |
| 358 | } |
| 359 | |
| 360 | // Common wildcard-style RegEx sequences. |
| 361 | if (strpos($value, '.*') !== false || strpos($value, '.+') !== false || strpos($value, '.?') !== false) { |
| 362 | return true; |
| 363 | } |
| 364 | |
| 365 | // Escaped RegEx tokens such as \.js, \d, \w, \s, etc. |
| 366 | if (preg_match('/\\\\[.dDsSwWbBAZzGQE]/', $value) === 1) { |
| 367 | return true; |
| 368 | } |
| 369 | |
| 370 | return false; |
| 371 | } |
| 372 | |
| 373 | /** |
| 374 | * Wrap a legacy loose RegEx pattern using the default delimiter. |
| 375 | * |
| 376 | * @param string $pattern |
| 377 | * |
| 378 | * @return string |
| 379 | */ |
| 380 | public static function wrapLoosePattern($pattern) |
| 381 | { |
| 382 | $pattern = trim((string)$pattern); |
| 383 | |
| 384 | if ($pattern === '') { |
| 385 | return ''; |
| 386 | } |
| 387 | |
| 388 | return self::DEFAULT_DELIMITER . self::escapeDelimiter($pattern, self::DEFAULT_DELIMITER) . self::DEFAULT_DELIMITER; |
| 389 | } |
| 390 | |
| 391 | /** |
| 392 | * Split textarea/string/array rules into clean non-empty rows. |
| 393 | * |
| 394 | * @param string|array $rules |
| 395 | * |
| 396 | * @return array |
| 397 | */ |
| 398 | public static function splitRules($rules) |
| 399 | { |
| 400 | if (is_array($rules)) { |
| 401 | $rows = $rules; |
| 402 | } else { |
| 403 | $rules = str_replace(array("\r\n", "\r"), "\n", (string)$rules); |
| 404 | $rows = explode("\n", $rules); |
| 405 | } |
| 406 | |
| 407 | $cleanRules = array(); |
| 408 | |
| 409 | foreach ($rows as $row) { |
| 410 | $row = trim((string)$row); |
| 411 | |
| 412 | if ($row !== '') { |
| 413 | $cleanRules[] = $row; |
| 414 | } |
| 415 | } |
| 416 | |
| 417 | return $cleanRules; |
| 418 | } |
| 419 | |
| 420 | /** |
| 421 | * Purify textarea rules. |
| 422 | * |
| 423 | * Keeps: |
| 424 | * - plain strings |
| 425 | * - valid explicit RegEx rules |
| 426 | * - valid legacy loose RegEx rules |
| 427 | * |
| 428 | * Removes: |
| 429 | * - invalid explicit RegEx rules |
| 430 | * - invalid legacy loose RegEx rules |
| 431 | * |
| 432 | * It does not rewrite valid loose legacy rules for storage. |
| 433 | * |
| 434 | * @param string $textareaValue |
| 435 | * @param bool $allowSlashDelimiter |
| 436 | * |
| 437 | * @return string |
| 438 | */ |
| 439 | public static function purifyTextareaRules($textareaValue, $allowSlashDelimiter = false) |
| 440 | { |
| 441 | $validRules = array(); |
| 442 | |
| 443 | foreach (self::splitRules($textareaValue) as $rule) { |
| 444 | if (self::isValidRule($rule, $allowSlashDelimiter)) { |
| 445 | $validRules[] = trim($rule); |
| 446 | } |
| 447 | } |
| 448 | |
| 449 | return implode("\n", $validRules); |
| 450 | } |
| 451 | |
| 452 | /** |
| 453 | * Backward-compatible alias for old naming. |
| 454 | * |
| 455 | * @param string $textareaValue |
| 456 | * @param bool $allowSlashDelimiter |
| 457 | * |
| 458 | * @return string |
| 459 | */ |
| 460 | public static function purifyTextareaRegexValue($textareaValue, $allowSlashDelimiter = false) |
| 461 | { |
| 462 | return self::purifyTextareaRules($textareaValue, $allowSlashDelimiter); |
| 463 | } |
| 464 | |
| 465 | /** |
| 466 | * Find the real closing delimiter, ignoring escaped delimiters. |
| 467 | * |
| 468 | * @param string $pattern |
| 469 | * @param string $delimiter |
| 470 | * |
| 471 | * @return int|false |
| 472 | */ |
| 473 | protected static function findClosingDelimiterPosition($pattern, $delimiter) |
| 474 | { |
| 475 | $length = strlen($pattern); |
| 476 | |
| 477 | for ($i = $length - 1; $i > 0; $i--) { |
| 478 | if ($pattern[$i] !== $delimiter) { |
| 479 | continue; |
| 480 | } |
| 481 | |
| 482 | if (! self::isEscapedPosition($pattern, $i)) { |
| 483 | return $i; |
| 484 | } |
| 485 | } |
| 486 | |
| 487 | return false; |
| 488 | } |
| 489 | |
| 490 | /** |
| 491 | * Check whether a character at a given offset is escaped. |
| 492 | * |
| 493 | * @param string $string |
| 494 | * @param int $position |
| 495 | * |
| 496 | * @return bool |
| 497 | */ |
| 498 | protected static function isEscapedPosition($string, $position) |
| 499 | { |
| 500 | $backslashes = 0; |
| 501 | |
| 502 | for ($i = $position - 1; $i >= 0; $i--) { |
| 503 | if ($string[$i] !== '\\') { |
| 504 | break; |
| 505 | } |
| 506 | |
| 507 | $backslashes++; |
| 508 | } |
| 509 | |
| 510 | return ($backslashes % 2) === 1; |
| 511 | } |
| 512 | |
| 513 | /** |
| 514 | * Escape the delimiter when wrapping a loose pattern. |
| 515 | * |
| 516 | * @param string $pattern |
| 517 | * @param string $delimiter |
| 518 | * |
| 519 | * @return string |
| 520 | */ |
| 521 | protected static function escapeDelimiter($pattern, $delimiter) |
| 522 | { |
| 523 | return str_replace($delimiter, '\\' . $delimiter, $pattern); |
| 524 | } |
| 525 | } |
| 526 |