| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* URL normalization and multilingual redirect translation helpers. |
| 9 |
* Used by ABJ_404_Solution_PluginLogic via `use`. |
| 10 |
*/ |
| 11 |
trait ABJ_404_Solution_PluginLogicTrait_UrlNormalization { |
| 12 |
|
| 13 |
/** If a page's URL is /blogName/pageName then this returns /pageName. |
| 14 |
* @param string|null $urlRequest |
| 15 |
* @return string |
| 16 |
*/ |
| 17 |
function removeHomeDirectory($urlRequest): string { |
| 18 |
if ($urlRequest === null) { |
| 19 |
return ''; |
| 20 |
} |
| 21 |
$f = $this->f; |
| 22 |
$urlHomeDirectory = $this->urlHomeDirectory; |
| 23 |
$homeLen = $this->urlHomeDirectoryLength !== null ? $this->urlHomeDirectoryLength : 0; |
| 24 |
|
| 25 |
// Fix CRITICAL #1 (5th review): Skip processing for root installations |
| 26 |
// When WordPress is at domain root, urlHomeDirectoryLength is 0 |
| 27 |
// Without this check, substr($url, 0, 0) == '' is always TRUE, incorrectly stripping leading slash |
| 28 |
if ($homeLen === 0) { |
| 29 |
return $urlRequest; |
| 30 |
} |
| 31 |
|
| 32 |
// Fix CRITICAL #1 (2nd review): Check path boundary to prevent false positives |
| 33 |
// e.g., /blog should match /blog/page but NOT /blogpost or /blog-archive |
| 34 |
if ($this->f->substr($urlRequest, 0, $homeLen) == $urlHomeDirectory) { |
| 35 |
// Verify path boundary: next character must be '/', '?', '#', or end of string |
| 36 |
$nextChar = $this->f->substr($urlRequest, $homeLen, 1); |
| 37 |
if ($nextChar === '/' || $nextChar === '?' || $nextChar === '#' || $nextChar === '') { |
| 38 |
// Fix CRITICAL #2 (3rd review): Don't strip query/fragment markers |
| 39 |
if ($nextChar === '/' || $nextChar === '') { |
| 40 |
// Strip subdirectory + slash for paths: /blog/page → /page |
| 41 |
$urlRequest = $this->f->substr($urlRequest, ($homeLen + 1)); |
| 42 |
} else { |
| 43 |
// Fix HIGH #1 (4th review): Add leading slash for query/fragment |
| 44 |
// Strip subdirectory, add leading slash: /blog?q=1 → /?q=1 |
| 45 |
$urlRequest = '/' . $this->f->substr($urlRequest, $homeLen); |
| 46 |
} |
| 47 |
} |
| 48 |
// else: false positive (e.g., /blogpost when subdirectory is /blog) - don't strip |
| 49 |
} |
| 50 |
|
| 51 |
return $urlRequest; |
| 52 |
} |
| 53 |
|
| 54 |
/** |
| 55 |
* Normalize URL to relative path by removing WordPress subdirectory. |
| 56 |
* This ensures URLs are stored/matched independently of subdirectory changes. |
| 57 |
* Fixes Issue #24: Redirects now survive WordPress subdirectory changes. |
| 58 |
* |
| 59 |
* @param string|null $url Full URL or path |
| 60 |
* @return string Relative path without subdirectory |
| 61 |
*/ |
| 62 |
function normalizeToRelativePath($url): string { |
| 63 |
// Fix Issue #5: Handle empty URLs explicitly |
| 64 |
if ($url === '') { |
| 65 |
return '/'; |
| 66 |
} |
| 67 |
|
| 68 |
// Fix HIGH #2: Trim whitespace |
| 69 |
if ($url === null) { |
| 70 |
return '/'; |
| 71 |
} |
| 72 |
$url = trim($url); |
| 73 |
|
| 74 |
// Fix CRITICAL #2 (4th review): REMOVED rawurldecode() - URLs already decoded by UserRequest |
| 75 |
// Subdirectory decoding is now handled in constructor for consistency |
| 76 |
|
| 77 |
// Fix HIGH #2: If full URL, extract path only |
| 78 |
if (preg_match('#^https?://#i', $url)) { |
| 79 |
$parsed = parse_url($url); |
| 80 |
if ($parsed === false || !isset($parsed['path'])) { |
| 81 |
return '/'; |
| 82 |
} |
| 83 |
$url = $parsed['path']; |
| 84 |
// Preserve query and fragment |
| 85 |
if (!empty($parsed['query'])) { |
| 86 |
$url .= '?' . $parsed['query']; |
| 87 |
} |
| 88 |
if (!empty($parsed['fragment'])) { |
| 89 |
$url .= '#' . $parsed['fragment']; |
| 90 |
} |
| 91 |
} |
| 92 |
|
| 93 |
// Fix HIGH #2: Handle protocol-relative URLs (//example.com/path) |
| 94 |
if (strpos($url, '//') === 0) { |
| 95 |
$parsed = parse_url('http:' . $url); |
| 96 |
if ($parsed !== false && isset($parsed['path'])) { |
| 97 |
$url = $parsed['path']; |
| 98 |
if (!empty($parsed['query'])) { |
| 99 |
$url .= '?' . $parsed['query']; |
| 100 |
} |
| 101 |
if (!empty($parsed['fragment'])) { |
| 102 |
$url .= '#' . $parsed['fragment']; |
| 103 |
} |
| 104 |
} else { |
| 105 |
return '/'; |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
// Remove home directory if present |
| 110 |
$relativePath = $this->removeHomeDirectory($url); |
| 111 |
|
| 112 |
// Fix Issue #5: Check if removeHomeDirectory() returned empty unexpectedly |
| 113 |
if ($relativePath === '') { |
| 114 |
// Return root path for empty results |
| 115 |
return '/'; |
| 116 |
} |
| 117 |
|
| 118 |
// Fix HIGH #2: Normalize multiple slashes to single slash |
| 119 |
$relativePathCleaned = preg_replace('#/+#', '/', $relativePath); |
| 120 |
$relativePath = is_string($relativePathCleaned) ? $relativePathCleaned : $relativePath; |
| 121 |
|
| 122 |
// Ensure consistent leading slash (but not multiple) |
| 123 |
$relativePath = '/' . ltrim($relativePath, '/'); |
| 124 |
|
| 125 |
return $relativePath; |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Normalize a user-provided path for storage/matching. |
| 130 |
* Decodes percent-encoded octets and strips invalid UTF-8/control bytes. |
| 131 |
* |
| 132 |
* @param string|null $url |
| 133 |
* @return string |
| 134 |
*/ |
| 135 |
private function normalizeUserProvidedPath($url) { |
| 136 |
$url = $this->f->normalizeUrlString($url); |
| 137 |
if ($url === '') { |
| 138 |
return ''; |
| 139 |
} |
| 140 |
|
| 141 |
return $this->normalizeToRelativePath($url); |
| 142 |
} |
| 143 |
|
| 144 |
/** |
| 145 |
* Normalize an external destination URL for storage. |
| 146 |
* Decodes percent-encoded octets and strips invalid UTF-8/control bytes. |
| 147 |
* |
| 148 |
* @param string|null $url |
| 149 |
* @return string |
| 150 |
*/ |
| 151 |
private function normalizeExternalDestinationUrl($url) { |
| 152 |
return $this->f->normalizeUrlString($url); |
| 153 |
} |
| 154 |
|
| 155 |
/** |
| 156 |
* Generate normalized lookup variants for URL matching. |
| 157 |
* Includes decoded form and a legacy encoded fallback. |
| 158 |
* |
| 159 |
* @param string|null $url |
| 160 |
* @return array<int, string> |
| 161 |
*/ |
| 162 |
function getNormalizedUrlCandidates($url) { |
| 163 |
$decoded = $this->normalizeUserProvidedPath($url); |
| 164 |
if ($decoded === '') { |
| 165 |
return array(); |
| 166 |
} |
| 167 |
|
| 168 |
$candidates = array($decoded); |
| 169 |
|
| 170 |
// Case-insensitive fallback: URLs are case-insensitive in practice, |
| 171 |
// but the DB uses BINARY comparison for performance. Try the lowercase |
| 172 |
// variant so /E2E-Case matches a redirect stored as /e2e-case. |
| 173 |
$lower = function_exists('mb_strtolower') ? mb_strtolower($decoded, 'UTF-8') : strtolower($decoded); |
| 174 |
if ($lower !== $decoded) { |
| 175 |
$candidates[] = $lower; |
| 176 |
} |
| 177 |
|
| 178 |
// Legacy fallback for stored percent-encoded slugs. |
| 179 |
$encoded = $this->normalizeToRelativePath($this->f->encodeUrlForLegacyMatch($decoded)); |
| 180 |
if ($encoded !== $decoded) { |
| 181 |
$candidates[] = $encoded; |
| 182 |
} |
| 183 |
|
| 184 |
return array_values(array_unique($candidates)); |
| 185 |
} |
| 186 |
|
| 187 |
/** |
| 188 |
* Translate a redirect destination URL to the current language when possible. |
| 189 |
* |
| 190 |
* @param string $location Full URL or path to redirect to. |
| 191 |
* @param string $requestedURL Original requested path/URL that triggered the 404. |
| 192 |
* @return string URL to use for redirect. |
| 193 |
*/ |
| 194 |
function maybeTranslateRedirectUrl($location, $requestedURL = '') { |
| 195 |
if (!is_string($location) || $location === '') { |
| 196 |
return $location; |
| 197 |
} |
| 198 |
|
| 199 |
$translated = $this->translatePressRedirectUrl($location, $requestedURL); |
| 200 |
if ($translated !== null && $translated !== '') { |
| 201 |
$location = $translated; |
| 202 |
} |
| 203 |
|
| 204 |
if ($translated === null || $translated === '') { |
| 205 |
$translated = $this->wpmlRedirectUrl($location, $requestedURL); |
| 206 |
if ($translated !== null && $translated !== '') { |
| 207 |
$location = $translated; |
| 208 |
} |
| 209 |
} |
| 210 |
|
| 211 |
if ($translated === null || $translated === '') { |
| 212 |
$translated = $this->polylangRedirectUrl($location, $requestedURL); |
| 213 |
if ($translated !== null && $translated !== '') { |
| 214 |
$location = $translated; |
| 215 |
} |
| 216 |
} |
| 217 |
|
| 218 |
// Allow other multilingual plugins/themes to override redirect destinations. |
| 219 |
return apply_filters('abj404_translate_redirect_url', $location, $requestedURL); |
| 220 |
} |
| 221 |
|
| 222 |
/** @return string|null */ |
| 223 |
private function translatePressRedirectUrl(string $location, string $requestedURL) { |
| 224 |
if (!$this->translatePressIntegrationAvailable()) { |
| 225 |
return null; |
| 226 |
} |
| 227 |
|
| 228 |
if (!$this->isLocalUrl($location)) { |
| 229 |
return null; |
| 230 |
} |
| 231 |
|
| 232 |
$language = $this->getTranslatePressLanguageFromRequest($requestedURL); |
| 233 |
if ($language === '') { |
| 234 |
return null; |
| 235 |
} |
| 236 |
|
| 237 |
$translated = $this->translatePressTranslateUrl($location, $language); |
| 238 |
if (!is_string($translated) || $translated === '' || $translated === $location) { |
| 239 |
return null; |
| 240 |
} |
| 241 |
|
| 242 |
if (!$this->isLocalUrl($translated)) { |
| 243 |
return null; |
| 244 |
} |
| 245 |
|
| 246 |
return $translated; |
| 247 |
} |
| 248 |
|
| 249 |
/** @return bool */ |
| 250 |
private function translatePressIntegrationAvailable(): bool { |
| 251 |
return function_exists('trp_get_language_from_url') || |
| 252 |
function_exists('trp_get_current_language') || |
| 253 |
function_exists('trp_get_url_for_language') || |
| 254 |
function_exists('trp_translate_url') || |
| 255 |
has_filter('trp_translate_url'); |
| 256 |
} |
| 257 |
|
| 258 |
/** @return mixed */ |
| 259 |
private function translatePressTranslateUrl(string $url, string $language) { |
| 260 |
if (function_exists('trp_get_url_for_language')) { |
| 261 |
return trp_get_url_for_language($language, $url); |
| 262 |
} |
| 263 |
|
| 264 |
if (function_exists('trp_translate_url')) { |
| 265 |
return trp_translate_url($url, $language); |
| 266 |
} |
| 267 |
|
| 268 |
return apply_filters('trp_translate_url', $url, $language); |
| 269 |
} |
| 270 |
|
| 271 |
private function getTranslatePressLanguageFromRequest(string $requestedURL): string { |
| 272 |
$fullRequestedUrl = $this->buildFullUrlFromRequest($requestedURL); |
| 273 |
|
| 274 |
if (function_exists('trp_get_language_from_url')) { |
| 275 |
$language = trp_get_language_from_url($fullRequestedUrl); |
| 276 |
if (is_string($language) && $language !== '') { |
| 277 |
return $language; |
| 278 |
} |
| 279 |
} |
| 280 |
|
| 281 |
if (function_exists('trp_get_current_language')) { |
| 282 |
$language = trp_get_current_language(); |
| 283 |
if (is_string($language) && $language !== '') { |
| 284 |
return $language; |
| 285 |
} |
| 286 |
} |
| 287 |
|
| 288 |
return ''; |
| 289 |
} |
| 290 |
|
| 291 |
/** @return string|null */ |
| 292 |
private function wpmlRedirectUrl(string $location, string $requestedURL) { |
| 293 |
if (!$this->wpmlIntegrationAvailable()) { |
| 294 |
return null; |
| 295 |
} |
| 296 |
|
| 297 |
if (!$this->isLocalUrl($location)) { |
| 298 |
return null; |
| 299 |
} |
| 300 |
|
| 301 |
$language = $this->getWpmlLanguageFromRequest($requestedURL); |
| 302 |
if ($language === '') { |
| 303 |
return null; |
| 304 |
} |
| 305 |
|
| 306 |
$translated = $this->wpmlTranslateUrl($location, $language); |
| 307 |
if (!is_string($translated) || $translated === '' || $translated === $location) { |
| 308 |
return null; |
| 309 |
} |
| 310 |
|
| 311 |
if (!$this->isLocalUrl($translated)) { |
| 312 |
return null; |
| 313 |
} |
| 314 |
|
| 315 |
return $translated; |
| 316 |
} |
| 317 |
|
| 318 |
private function wpmlIntegrationAvailable(): bool { |
| 319 |
return function_exists('wpml_current_language') || |
| 320 |
has_filter('wpml_current_language') || |
| 321 |
has_filter('wpml_language_from_url') || |
| 322 |
has_filter('wpml_permalink'); |
| 323 |
} |
| 324 |
|
| 325 |
/** @return mixed */ |
| 326 |
private function wpmlTranslateUrl(string $url, string $language) { |
| 327 |
if (has_filter('wpml_permalink')) { |
| 328 |
return apply_filters('wpml_permalink', $url, $language); |
| 329 |
} |
| 330 |
|
| 331 |
return null; |
| 332 |
} |
| 333 |
|
| 334 |
private function getWpmlLanguageFromRequest(string $requestedURL): string { |
| 335 |
$fullRequestedUrl = $this->buildFullUrlFromRequest($requestedURL); |
| 336 |
|
| 337 |
if (has_filter('wpml_language_from_url')) { |
| 338 |
$language = apply_filters('wpml_language_from_url', '', $fullRequestedUrl); |
| 339 |
if (is_string($language) && $language !== '') { |
| 340 |
return $language; |
| 341 |
} |
| 342 |
} |
| 343 |
|
| 344 |
if (function_exists('wpml_current_language')) { |
| 345 |
$language = wpml_current_language(); |
| 346 |
if (is_string($language) && $language !== '') { |
| 347 |
return $language; |
| 348 |
} |
| 349 |
} |
| 350 |
|
| 351 |
if (has_filter('wpml_current_language')) { |
| 352 |
$language = apply_filters('wpml_current_language', null); |
| 353 |
if (is_string($language) && $language !== '') { |
| 354 |
return $language; |
| 355 |
} |
| 356 |
} |
| 357 |
|
| 358 |
return ''; |
| 359 |
} |
| 360 |
|
| 361 |
/** @return string|null */ |
| 362 |
private function polylangRedirectUrl(string $location, string $requestedURL) { |
| 363 |
if (!$this->polylangIntegrationAvailable()) { |
| 364 |
return null; |
| 365 |
} |
| 366 |
|
| 367 |
if (!$this->isLocalUrl($location)) { |
| 368 |
return null; |
| 369 |
} |
| 370 |
|
| 371 |
$language = $this->getPolylangLanguageFromRequest($requestedURL); |
| 372 |
if ($language === '') { |
| 373 |
return null; |
| 374 |
} |
| 375 |
|
| 376 |
$translated = $this->polylangTranslateUrl($location, $language); |
| 377 |
if (!is_string($translated) || $translated === '' || $translated === $location) { |
| 378 |
return null; |
| 379 |
} |
| 380 |
|
| 381 |
if (!$this->isLocalUrl($translated)) { |
| 382 |
return null; |
| 383 |
} |
| 384 |
|
| 385 |
return $translated; |
| 386 |
} |
| 387 |
|
| 388 |
private function polylangIntegrationAvailable(): bool { |
| 389 |
return function_exists('pll_current_language') || |
| 390 |
function_exists('pll_translate_url'); |
| 391 |
} |
| 392 |
|
| 393 |
/** @return mixed */ |
| 394 |
private function polylangTranslateUrl(string $url, string $language) { |
| 395 |
if (function_exists('pll_translate_url')) { |
| 396 |
return pll_translate_url($url, $language); |
| 397 |
} |
| 398 |
|
| 399 |
return null; |
| 400 |
} |
| 401 |
|
| 402 |
private function getPolylangLanguageFromRequest(string $requestedURL): string { |
| 403 |
if (function_exists('pll_current_language')) { |
| 404 |
$language = pll_current_language(); |
| 405 |
if (is_string($language) && $language !== '') { |
| 406 |
return $language; |
| 407 |
} |
| 408 |
} |
| 409 |
|
| 410 |
return ''; |
| 411 |
} |
| 412 |
|
| 413 |
private function buildFullUrlFromRequest(string $requestedURL): string { |
| 414 |
$path = $requestedURL; |
| 415 |
if ($path === '') { |
| 416 |
$userRequest = ABJ_404_Solution_UserRequest::getInstance(); |
| 417 |
if ($userRequest !== null) { |
| 418 |
$path = $userRequest->getPathWithSortedQueryString(); |
| 419 |
} |
| 420 |
} |
| 421 |
|
| 422 |
if ($path === '') { |
| 423 |
return home_url('/'); |
| 424 |
} |
| 425 |
|
| 426 |
if (preg_match('#^https?://#i', $path)) { |
| 427 |
return $path; |
| 428 |
} |
| 429 |
|
| 430 |
return home_url($path); |
| 431 |
} |
| 432 |
|
| 433 |
private function isLocalUrl(string $url): bool { |
| 434 |
if (!is_string($url) || $url === '') { |
| 435 |
return false; |
| 436 |
} |
| 437 |
|
| 438 |
$parsedUrl = function_exists('wp_parse_url') ? wp_parse_url($url) : parse_url($url); |
| 439 |
if (!is_array($parsedUrl) || !isset($parsedUrl['host'])) { |
| 440 |
// Relative URLs are treated as local. |
| 441 |
return true; |
| 442 |
} |
| 443 |
|
| 444 |
$siteUrl = home_url(); |
| 445 |
$parsedSite = function_exists('wp_parse_url') ? wp_parse_url($siteUrl) : parse_url($siteUrl); |
| 446 |
$siteHost = is_array($parsedSite) && isset($parsedSite['host']) ? strtolower($parsedSite['host']) : ''; |
| 447 |
|
| 448 |
return $siteHost !== '' && strtolower($parsedUrl['host']) === $siteHost; |
| 449 |
} |
| 450 |
|
| 451 |
} |
| 452 |
|
