| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* The Urls Redirection functionality of the plugin. |
| 5 |
* |
| 6 |
* Defines the plugin name, version, and two examples hooks for how to |
| 7 |
* enqueue the admin-specific stylesheet and JavaScript. |
| 8 |
* |
| 9 |
* @link https://searchatlas.com |
| 10 |
* @since 1.0.0 |
| 11 |
* @package Metasync |
| 12 |
* @subpackage Metasync/redirections |
| 13 |
* @author Engineering Team <[email protected]> |
| 14 |
*/ |
| 15 |
|
| 16 |
// Abort if this file is accessed directly. |
| 17 |
if (!defined('ABSPATH')) { |
| 18 |
exit; |
| 19 |
} |
| 20 |
|
| 21 |
if (!class_exists('Metasync_Redirection_Validator')) { |
| 22 |
require_once dirname(__FILE__) . '/class-metasync-redirection-validator.php'; |
| 23 |
} |
| 24 |
|
| 25 |
class Metasync_Redirection |
| 26 |
{ |
| 27 |
|
| 28 |
private $db_redirection; |
| 29 |
private $common; |
| 30 |
private $importer; |
| 31 |
|
| 32 |
/** @var array|null Lazy-built exact-match lookup: normalized_path => row object */ |
| 33 |
private $exact_index = null; |
| 34 |
/** @var array|null Pattern-based rows (wildcard, regex, contain, start, end) */ |
| 35 |
private $pattern_index = null; |
| 36 |
|
| 37 |
public function __construct(&$db_redirection) |
| 38 |
{ |
| 39 |
$this->db_redirection = $db_redirection; |
| 40 |
$this->common = new Metasync_Common(); |
| 41 |
|
| 42 |
# Load importer class |
| 43 |
require_once dirname(__FILE__) . '/class-metasync-redirection-importer.php'; |
| 44 |
$this->importer = new Metasync_Redirection_Importer($db_redirection); |
| 45 |
} |
| 46 |
|
| 47 |
/** |
| 48 |
* Build the redirect lookup index (lazy, once per request). |
| 49 |
* Exact-match sources go into a hashmap for O(1) lookup. |
| 50 |
* Pattern-based sources (wildcard, regex, contain, start, end) stay in a list. |
| 51 |
*/ |
| 52 |
private function ensure_redirect_index() |
| 53 |
{ |
| 54 |
if ($this->exact_index !== null) { |
| 55 |
return; |
| 56 |
} |
| 57 |
|
| 58 |
$this->exact_index = array(); |
| 59 |
$this->pattern_index = array(); |
| 60 |
|
| 61 |
$redirections = $this->db_redirection->getAllActiveRecords(); |
| 62 |
if (empty($redirections)) { |
| 63 |
return; |
| 64 |
} |
| 65 |
|
| 66 |
foreach ($redirections as $row) { |
| 67 |
$sources_from = !empty($row->sources_from) |
| 68 |
? unserialize($row->sources_from, array('allowed_classes' => false)) |
| 69 |
: array(); |
| 70 |
$source_urls = is_array($sources_from) ? $sources_from : array(); |
| 71 |
$global_pattern_type = isset($row->pattern_type) ? $row->pattern_type : null; |
| 72 |
$is_pattern = false; |
| 73 |
|
| 74 |
foreach ($source_urls as $source_key => $source_value) { |
| 75 |
// Legacy list-format rows store the URL as the VALUE under a |
| 76 |
// numeric key ('0' => '/old'); modern rows key by URL with the |
| 77 |
// pattern type as the value. Remap so the legacy source lands |
| 78 |
// in the exact index instead of '/0'. |
| 79 |
if ((is_int($source_key) || ctype_digit((string) $source_key)) && is_string($source_value) && $source_value !== '') { |
| 80 |
$source_key = $source_value; |
| 81 |
} |
| 82 |
// Rows saved before the input-unslash fix may still hold |
| 83 |
// magic-quoted sources; heal at read time so they keep |
| 84 |
// matching the unslashed request URI. |
| 85 |
$source_key = wp_unslash((string) $source_key); |
| 86 |
$pattern_type = in_array($source_value, array('exact', 'contain', 'start', 'end', 'wildcard', 'regex')) |
| 87 |
? $source_value |
| 88 |
: ($global_pattern_type ? $global_pattern_type : 'exact'); |
| 89 |
|
| 90 |
if ($pattern_type === 'exact' && strpos((string) $source_key, '*') === false) { |
| 91 |
// Normalize: extract path from full URLs, ensure leading slash, strip trailing slash |
| 92 |
$norm = (string) $source_key; |
| 93 |
if (strpos($norm, 'http') === 0) { |
| 94 |
$parsed = wp_parse_url($norm); |
| 95 |
$norm = isset($parsed['path']) ? $parsed['path'] : '/'; |
| 96 |
} |
| 97 |
if ($norm === '' || $norm[0] !== '/') { |
| 98 |
$norm = '/' . $norm; |
| 99 |
} |
| 100 |
$norm = rtrim($norm, '/') ?: '/'; |
| 101 |
$this->exact_index[$norm] = $row; |
| 102 |
} else { |
| 103 |
$is_pattern = true; |
| 104 |
} |
| 105 |
} |
| 106 |
|
| 107 |
if ($is_pattern) { |
| 108 |
$this->pattern_index[] = $row; |
| 109 |
} |
| 110 |
} |
| 111 |
} |
| 112 |
|
| 113 |
function contains($haystack, $needle, $caseSensitive = false) |
| 114 |
{ |
| 115 |
return $caseSensitive ? |
| 116 |
(strpos($haystack, $needle) === FALSE ? FALSE : TRUE) : (stripos($haystack, $needle) === FALSE ? FALSE : TRUE); |
| 117 |
} |
| 118 |
|
| 119 |
public function create_admin_redirection_interface() |
| 120 |
{ |
| 121 |
# Check if we should show import interface |
| 122 |
$request_data = metasync_sanitize_input_array($_REQUEST); // phpcs:ignore WordPress.Security.NonceVerification.Recommended -- read-only view switch, the import verifies its own nonce |
| 123 |
if (isset($request_data['action']) && $request_data['action'] === 'import') { |
| 124 |
$this->show_import_interface(); |
| 125 |
return; |
| 126 |
} |
| 127 |
|
| 128 |
if (!class_exists('WP_List_Table')) { |
| 129 |
require_once ABSPATH . 'wp-admin/includes/class-wp-list-table.php'; |
| 130 |
} |
| 131 |
require dirname(__FILE__, 2) . '/redirections/class-metasync-redirection-list-table.php'; |
| 132 |
|
| 133 |
$MetasyncRedirection = new Metasync_Redirection_List_Table(); |
| 134 |
|
| 135 |
$MetasyncRedirection->setDatabaseResource($this->db_redirection); |
| 136 |
|
| 137 |
$MetasyncRedirection->prepare_items(); |
| 138 |
|
| 139 |
// Include the view markup. |
| 140 |
include dirname(__FILE__, 2) . '/views/metasync-redirection.php'; |
| 141 |
} |
| 142 |
|
| 143 |
/** |
| 144 |
* Show import interface |
| 145 |
*/ |
| 146 |
public function show_import_interface() |
| 147 |
{ |
| 148 |
$importer = $this->importer; |
| 149 |
include dirname(__FILE__, 2) . '/views/metasync-import-redirections.php'; |
| 150 |
} |
| 151 |
|
| 152 |
/** |
| 153 |
* Handle AJAX import request |
| 154 |
*/ |
| 155 |
public function handle_import_ajax() |
| 156 |
{ |
| 157 |
# Verify nonce |
| 158 |
check_ajax_referer('metasync_import_redirections', 'nonce'); |
| 159 |
|
| 160 |
# Check user capabilities |
| 161 |
if (!Metasync::current_user_has_plugin_access()) { |
| 162 |
wp_send_json_error(['message' => 'Insufficient permissions.']); |
| 163 |
return; |
| 164 |
} |
| 165 |
|
| 166 |
$plugin = isset($_POST['plugin']) ? sanitize_text_field(wp_unslash($_POST['plugin'])) : ''; |
| 167 |
|
| 168 |
if (empty($plugin)) { |
| 169 |
wp_send_json_error(['message' => 'No plugin specified.']); |
| 170 |
return; |
| 171 |
} |
| 172 |
|
| 173 |
# CSV is a file upload, not a plugin, so it takes its own path with |
| 174 |
# upload validation before the importer sees the temp file. |
| 175 |
if ($plugin === 'csv') { |
| 176 |
$this->handle_csv_import_ajax(); |
| 177 |
return; |
| 178 |
} |
| 179 |
|
| 180 |
try { |
| 181 |
# Perform import |
| 182 |
$result = $this->importer->import_from_plugin($plugin); |
| 183 |
|
| 184 |
Metasync_PostHog::import_batch($plugin, 'redirections', $result); |
| 185 |
|
| 186 |
if ($result['success']) { |
| 187 |
wp_send_json_success($result); |
| 188 |
} else { |
| 189 |
wp_send_json_error($result); |
| 190 |
} |
| 191 |
} catch (Exception $e) { |
| 192 |
wp_send_json_error([ |
| 193 |
'message' => 'Import failed. Please try again or contact support.', |
| 194 |
'imported' => 0, |
| 195 |
'skipped' => 0 |
| 196 |
]); |
| 197 |
} |
| 198 |
} |
| 199 |
|
| 200 |
/** |
| 201 |
* Handle the AJAX CSV import: validate the upload, then hand the temp |
| 202 |
* file to the importer. The readme has advertised CSV import since 2.5.x; |
| 203 |
* this is the first implementation of it. |
| 204 |
*/ |
| 205 |
private function handle_csv_import_ajax() |
| 206 |
{ |
| 207 |
$file = isset($_FILES['csv_file']) && is_array($_FILES['csv_file']) ? $_FILES['csv_file'] : []; // phpcs:ignore WordPress.Security.NonceVerification.Missing -- private, caller verifies a nonce and capability first |
| 208 |
|
| 209 |
// wp_send_json_error() ends the request, so no return is needed (or |
| 210 |
// reachable) after any of these guards. |
| 211 |
if (empty($file['tmp_name']) || (isset($file['error']) && (int) $file['error'] !== UPLOAD_ERR_OK)) { |
| 212 |
wp_send_json_error(['message' => 'Upload failed. Please choose a .csv file and try again.']); |
| 213 |
} |
| 214 |
|
| 215 |
if (!is_uploaded_file($file['tmp_name'])) { |
| 216 |
wp_send_json_error(['message' => 'Invalid upload.']); |
| 217 |
} |
| 218 |
|
| 219 |
if (!preg_match('/\.csv$/i', (string) $file['name'])) { |
| 220 |
wp_send_json_error(['message' => 'Only .csv files are supported.']); |
| 221 |
} |
| 222 |
|
| 223 |
if ((int) $file['size'] > 2 * MB_IN_BYTES) { |
| 224 |
wp_send_json_error(['message' => 'The CSV file is too large. Maximum size is 2 MB.']); |
| 225 |
} |
| 226 |
|
| 227 |
$result = $this->importer->import_csv_file($file['tmp_name']); |
| 228 |
|
| 229 |
Metasync_PostHog::import_batch('csv', 'redirections', $result); |
| 230 |
|
| 231 |
if (!empty($result['success'])) { |
| 232 |
wp_send_json_success($result); |
| 233 |
} else { |
| 234 |
wp_send_json_error($result); |
| 235 |
} |
| 236 |
} |
| 237 |
|
| 238 |
public function get_current_page_url() |
| 239 |
{ |
| 240 |
$server_data = metasync_sanitize_input_array($_SERVER); |
| 241 |
$link = '://' . $server_data['HTTP_HOST'] . $server_data['REQUEST_URI']; |
| 242 |
$link = (is_ssl() ? 'https' : 'http') . $link; |
| 243 |
return sanitize_url($link); |
| 244 |
} |
| 245 |
|
| 246 |
public function source_url_redirection(object $row, string $uri, string $incoming_query = '') |
| 247 |
{ |
| 248 |
// Redirection feature switched off: rows owned by the per-post Redirection |
| 249 |
// meta box must not fire. Returning false rather than aborting means the |
| 250 |
// caller carries on evaluating the remaining rules, so rules added by hand |
| 251 |
// on the Redirections screen keep working exactly as before. |
| 252 |
if (Metasync_Feature_Flags::is_disabled(Metasync_Feature_Flags::REDIRECTION) |
| 253 |
&& class_exists('Metasync_Post_Meta_Settings') |
| 254 |
&& Metasync_Post_Meta_Settings::owns_redirect_row($row)) { |
| 255 |
return false; |
| 256 |
} |
| 257 |
|
| 258 |
// Optimize: unserialize only once |
| 259 |
$sources_from = unserialize($row->sources_from, ['allowed_classes' => false]); // phpcs:ignore WordPress.PHP.DiscouragedPHPFunctions.serialize_unserialize |
| 260 |
$source_urls = is_array($sources_from) ? $sources_from : []; |
| 261 |
$global_pattern_type = isset($row->pattern_type) ? $row->pattern_type : null; |
| 262 |
$regex_pattern = isset($row->regex_pattern) ? $row->regex_pattern : null; |
| 263 |
|
| 264 |
foreach ($source_urls as $source_key => $source_value) { |
| 265 |
$match_found = false; |
| 266 |
$captured_path = ''; // Store captured path for wildcard replacement |
| 267 |
|
| 268 |
// Ensure source_key is always a string (handles numeric array indexes) |
| 269 |
$source_key = (string) $source_key; |
| 270 |
|
| 271 |
// Legacy list-format rows store the URL as the VALUE under a |
| 272 |
// numeric key ('0' => '/old'); remap so matching uses the URL. |
| 273 |
if (ctype_digit($source_key) && is_string($source_value) && $source_value !== '') { |
| 274 |
$source_key = $source_value; |
| 275 |
} |
| 276 |
|
| 277 |
// Heal magic-quoted sources saved before the input-unslash fix so |
| 278 |
// pattern matching sees the literal value the user typed. |
| 279 |
$source_key = wp_unslash((string) $source_key); |
| 280 |
|
| 281 |
// Determine pattern type: use source value if it's a valid pattern, otherwise use global pattern_type |
| 282 |
$pattern_type = in_array($source_value, ['exact', 'contain', 'start', 'end', 'wildcard', 'regex']) |
| 283 |
? $source_value |
| 284 |
: ($global_pattern_type ? $global_pattern_type : 'exact'); |
| 285 |
|
| 286 |
// Normalize both URI and source for comparison |
| 287 |
$normalized_uri = $uri; |
| 288 |
$normalized_source = $source_key; |
| 289 |
|
| 290 |
// If source is a full URL, extract just the path part |
| 291 |
if (strpos($source_key, 'http') === 0) { |
| 292 |
$parsed_url = wp_parse_url($source_key); |
| 293 |
$normalized_source = $parsed_url['path'] ?? ''; |
| 294 |
} |
| 295 |
|
| 296 |
# Ensure both have leading slashes for proper comparison |
| 297 |
# This handles cases where database stores URLs with or without leading slash |
| 298 |
# Convert to string to handle cases where source_key might be an integer |
| 299 |
$normalized_source = (string) $normalized_source; |
| 300 |
if (!empty($normalized_source) && $normalized_source[0] !== '/') { |
| 301 |
$normalized_source = '/' . $normalized_source; |
| 302 |
} |
| 303 |
if (!empty($normalized_uri) && $normalized_uri[0] !== '/') { |
| 304 |
$normalized_uri = '/' . $normalized_uri; |
| 305 |
} |
| 306 |
|
| 307 |
// Normalize trailing slashes so /path and /path/ match equivalently |
| 308 |
$normalized_source = rtrim($normalized_source, '/') ?: '/'; |
| 309 |
$normalized_uri = rtrim($normalized_uri, '/') ?: '/'; |
| 310 |
|
| 311 |
// Keep full URI path for matching (don't strip leading slash) |
| 312 |
// This allows matching against full URLs like /wordpress/index.php/category/article/* |
| 313 |
|
| 314 |
// Handle regex patterns |
| 315 |
if ($pattern_type === 'regex' && $regex_pattern) { |
| 316 |
// Validate regex pattern before using it |
| 317 |
if (!$this->validate_regex_pattern($regex_pattern)) { |
| 318 |
// Skip invalid regex patterns to prevent errors |
| 319 |
continue; |
| 320 |
} |
| 321 |
|
| 322 |
# Normalize pattern (add delimiters if missing) |
| 323 |
$normalized_pattern = $this->normalize_regex_pattern($regex_pattern); |
| 324 |
|
| 325 |
|
| 326 |
$matches = []; |
| 327 |
// Suppress warnings for invalid regex and check result |
| 328 |
# $result = @preg_match($regex_pattern, $normalized_uri, $matches); |
| 329 |
$result = @preg_match($normalized_pattern, $normalized_uri, $matches); |
| 330 |
|
| 331 |
if ($result === 1) { |
| 332 |
$match_found = true; |
| 333 |
// Store captured groups for replacement |
| 334 |
if (isset($matches[1])) { |
| 335 |
$captured_path = $matches[1]; |
| 336 |
} |
| 337 |
} |
| 338 |
// If $result === false, regex is invalid - skip silently |
| 339 |
} else { |
| 340 |
// Check if source has wildcard |
| 341 |
$has_wildcard = strpos($normalized_source, '*') !== false; |
| 342 |
|
| 343 |
if ($has_wildcard) { |
| 344 |
// Handle wildcard pattern |
| 345 |
$match_result = $this->match_wildcard($normalized_source, $normalized_uri); |
| 346 |
if ($match_result !== false) { |
| 347 |
$match_found = true; |
| 348 |
$captured_path = $match_result; |
| 349 |
} |
| 350 |
} else { |
| 351 |
// Handle legacy pattern matching (non-wildcard) |
| 352 |
switch ($source_value) { |
| 353 |
case 'exact': |
| 354 |
if ($normalized_source === $normalized_uri) { |
| 355 |
$match_found = true; |
| 356 |
} |
| 357 |
break; |
| 358 |
case 'contain': |
| 359 |
if ($this->contains($normalized_uri, $normalized_source)) { |
| 360 |
$match_found = true; |
| 361 |
} |
| 362 |
break; |
| 363 |
case 'start': |
| 364 |
if (str_starts_with($normalized_uri, $normalized_source)) { |
| 365 |
$match_found = true; |
| 366 |
} |
| 367 |
break; |
| 368 |
case 'end': |
| 369 |
if (str_ends_with($normalized_uri, $normalized_source)) { |
| 370 |
$match_found = true; |
| 371 |
} |
| 372 |
break; |
| 373 |
default: |
| 374 |
// Handle new pattern_type field |
| 375 |
switch ($pattern_type) { |
| 376 |
case 'exact': |
| 377 |
if ($normalized_source === $normalized_uri) { |
| 378 |
$match_found = true; |
| 379 |
} |
| 380 |
break; |
| 381 |
case 'contain': |
| 382 |
if ($this->contains($normalized_uri, $normalized_source)) { |
| 383 |
$match_found = true; |
| 384 |
} |
| 385 |
break; |
| 386 |
case 'start': |
| 387 |
if (str_starts_with($normalized_uri, $normalized_source)) { |
| 388 |
$match_found = true; |
| 389 |
} |
| 390 |
break; |
| 391 |
case 'end': |
| 392 |
if (str_ends_with($normalized_uri, $normalized_source)) { |
| 393 |
$match_found = true; |
| 394 |
} |
| 395 |
break; |
| 396 |
case 'wildcard': |
| 397 |
$match_result = $this->match_wildcard($normalized_source, $normalized_uri); |
| 398 |
if ($match_result !== false) { |
| 399 |
$match_found = true; |
| 400 |
$captured_path = $match_result; |
| 401 |
} |
| 402 |
break; |
| 403 |
} |
| 404 |
break; |
| 405 |
} |
| 406 |
} |
| 407 |
} |
| 408 |
|
| 409 |
if ($match_found) { |
| 410 |
$this->db_redirection->update_counter($row); |
| 411 |
|
| 412 |
if ($row->http_code === '410') { |
| 413 |
status_header(410); |
| 414 |
die; |
| 415 |
} |
| 416 |
if ($row->http_code === '451') { |
| 417 |
status_header(451, 'Unavailable For Legal Reasons'); |
| 418 |
die; |
| 419 |
} |
| 420 |
if ($row->url_redirect_to) { |
| 421 |
// Replace wildcards or $1 placeholders in destination URL |
| 422 |
$destination = $this->process_destination_url($row->url_redirect_to, $captured_path); |
| 423 |
$destination = $this->append_query_string($destination, $incoming_query); |
| 424 |
$is_exact = isset($row->pattern_type) && $row->pattern_type === 'exact'; |
| 425 |
if (get_option('metasync_allow_external_redirects', 0) && $is_exact) { |
| 426 |
$destination = esc_url_raw($destination); |
| 427 |
if (empty($destination)) { |
| 428 |
$destination = home_url(); |
| 429 |
} |
| 430 |
// Same syntax hygiene as the default branch: the opt-in |
| 431 |
// permits external hosts, never syntactic evasion, so a |
| 432 |
// backslash trick like '/\evil.com' still bounces home. |
| 433 |
$checked = Metasync_Redirection_Validator::normalize_destination($destination); |
| 434 |
if (!Metasync_Redirection_Validator::is_safe_destination_syntax($checked)) { |
| 435 |
$checked = home_url(); |
| 436 |
} |
| 437 |
$this->safe_redirect_to_destination($checked, $row->http_code); |
| 438 |
} else { |
| 439 |
// Stored destinations can carry backslashes (imports, |
| 440 |
// older rows). Browsers read them as path separators, so |
| 441 |
// '/\evil.com' would slip past wp_validate_redirect and |
| 442 |
// still navigate off-site. Normalize first, and bounce |
| 443 |
// anything still syntactically evasive to the home page. |
| 444 |
$checked = Metasync_Redirection_Validator::normalize_destination($destination); |
| 445 |
if (!Metasync_Redirection_Validator::is_safe_destination_syntax($checked)) { |
| 446 |
$checked = home_url(); |
| 447 |
} |
| 448 |
$this->safe_redirect_to_destination(wp_validate_redirect($checked, home_url()), $row->http_code); |
| 449 |
} |
| 450 |
// Both branches terminate inside safe_redirect_to_destination(). |
| 451 |
} |
| 452 |
// Match found and processed, return true to stop checking other rules |
| 453 |
return true; |
| 454 |
} |
| 455 |
} |
| 456 |
|
| 457 |
// No match found |
| 458 |
return false; |
| 459 |
} |
| 460 |
|
| 461 |
/** |
| 462 |
* Emit a redirect through wp_safe_redirect() and end execution. |
| 463 |
* |
| 464 |
* wp_safe_redirect() confines the Location header to hosts WordPress |
| 465 |
* trusts (this site plus anything on the allowed_redirect_hosts |
| 466 |
* allowlist). The one intentional external destination is an exact-match |
| 467 |
* rule saved by an admin while the metasync_allow_external_redirects |
| 468 |
* opt-in is enabled, so the parsed host of the already syntax-validated |
| 469 |
* destination is allowlisted for this request only. Everything else — |
| 470 |
* same-site destinations, relative paths — passes through untouched. |
| 471 |
* |
| 472 |
* @param string $destination Validated destination URL or path. |
| 473 |
* @param string|int $http_code Redirect status code from the rule row. |
| 474 |
* @return void Never returns; wp_safe_redirect() is followed by exit. |
| 475 |
*/ |
| 476 |
private function safe_redirect_to_destination($destination, $http_code) |
| 477 |
{ |
| 478 |
$host = wp_parse_url($destination, PHP_URL_HOST); |
| 479 |
$grant_host = null; |
| 480 |
if (is_string($host) && $host !== '') { |
| 481 |
// Scope the temporary host grant to this one redirect. The filter |
| 482 |
// runs while wp_safe_redirect() validates the Location header and |
| 483 |
// is removed before exit: shutdown callbacks registered elsewhere |
| 484 |
// still fire after exit, so the grant must not outlive the call. |
| 485 |
$grant_host = static function ($allowed_hosts, $checked_host = '') use ($host) { |
| 486 |
if (is_array($allowed_hosts)) { |
| 487 |
$allowed_hosts[] = $host; |
| 488 |
} |
| 489 |
return $allowed_hosts; |
| 490 |
}; |
| 491 |
add_filter('allowed_redirect_hosts', $grant_host, 10, 2); |
| 492 |
} |
| 493 |
wp_safe_redirect($destination, $http_code); |
| 494 |
if (null !== $grant_host) { |
| 495 |
// Removal matches on priority + callback identity; core's |
| 496 |
// remove_filter() takes no accepted-args parameter. |
| 497 |
remove_filter('allowed_redirect_hosts', $grant_host, 10); |
| 498 |
} |
| 499 |
exit; |
| 500 |
} |
| 501 |
|
| 502 |
/** |
| 503 |
* Resolve a URL through the redirect table to its final destination (follows redirect chains). |
| 504 |
* Used by OTTO and other backend processing so the final canonical URL is used before 404 checks. |
| 505 |
* |
| 506 |
* @param string $url Full URL (e.g. https://example.com/old-page) |
| 507 |
* @param int $max_hops Maximum redirect hops to follow (default 10, prevents infinite loops) |
| 508 |
* @return string Final destination URL, or original $url if no redirect matches |
| 509 |
*/ |
| 510 |
public function resolve_url_to_final_destination($url, $max_hops = 10) |
| 511 |
{ |
| 512 |
if (empty($url) || !is_string($url)) { |
| 513 |
return $url; |
| 514 |
} |
| 515 |
$parsed = wp_parse_url($url); |
| 516 |
$scheme = isset($parsed['scheme']) ? $parsed['scheme'] : 'https'; |
| 517 |
$host = isset($parsed['host']) ? $parsed['host'] : ''; |
| 518 |
$base = $scheme . '://' . $host; |
| 519 |
// Track the query separately from the path: sources match on the path |
| 520 |
// only, and each destination may carry its own query. Folding the |
| 521 |
// query into $uri (as this used to) made '/old?x=1' miss an '/old' |
| 522 |
// rule and swallowed destination queries along the chain. |
| 523 |
$uri = isset($parsed['path']) ? $parsed['path'] : '/'; |
| 524 |
$query = isset($parsed['query']) ? $parsed['query'] : ''; |
| 525 |
$seen = array(); |
| 526 |
for ($i = 0; $i < $max_hops; $i++) { |
| 527 |
$uri_key = $query === '' ? $uri : $uri . '?' . $query; |
| 528 |
if (isset($seen[$uri_key])) { |
| 529 |
break; // cycle detected |
| 530 |
} |
| 531 |
$seen[$uri_key] = true; |
| 532 |
$dest = $this->get_redirect_destination_for_uri($uri); |
| 533 |
if ($dest === null) { |
| 534 |
break; |
| 535 |
} |
| 536 |
$dest = $this->append_query_string($dest, $query); |
| 537 |
if (strpos($dest, 'http') === 0) { |
| 538 |
$url = $dest; |
| 539 |
$parsed = wp_parse_url($url); |
| 540 |
$scheme = isset($parsed['scheme']) ? $parsed['scheme'] : 'https'; |
| 541 |
$host = isset($parsed['host']) ? $parsed['host'] : ''; |
| 542 |
$base = $scheme . '://' . $host; |
| 543 |
$uri = isset($parsed['path']) ? $parsed['path'] : '/'; |
| 544 |
$query = isset($parsed['query']) ? $parsed['query'] : ''; |
| 545 |
} else { |
| 546 |
$dest_parts = explode('?', $dest, 2); |
| 547 |
$uri = (isset($dest_parts[0][0]) && $dest_parts[0][0] === '/') ? $dest_parts[0] : '/' . $dest_parts[0]; |
| 548 |
$query = isset($dest_parts[1]) ? $dest_parts[1] : ''; |
| 549 |
$url = $base . $uri . ($query === '' ? '' : '?' . $query); |
| 550 |
} |
| 551 |
} |
| 552 |
return $url; |
| 553 |
} |
| 554 |
|
| 555 |
/** |
| 556 |
* Get redirect destination for a URI without redirecting (no wp_redirect, no counter update). |
| 557 |
* Returns the destination URL/path if this URI matches a redirect source, else null. |
| 558 |
* Used by resolve_url_to_final_destination. 410/451 are treated as "no destination". |
| 559 |
* |
| 560 |
* @param string $uri URI path (and optional query), e.g. /old-page or /old?x=1 |
| 561 |
* @return string|null Destination URL or path, or null if no match |
| 562 |
*/ |
| 563 |
private function get_redirect_destination_for_uri($uri) |
| 564 |
{ |
| 565 |
// Defensive: callers may pass a path?query form; sources match on the |
| 566 |
// path only, exactly like the live template-redirect path. |
| 567 |
$qpos = strpos((string) $uri, '?'); |
| 568 |
if ($qpos !== false) { |
| 569 |
$uri = substr($uri, 0, $qpos); |
| 570 |
} |
| 571 |
|
| 572 |
$redirections = $this->db_redirection->getAllActiveRecords(); |
| 573 |
if (empty($redirections)) { |
| 574 |
return null; |
| 575 |
} |
| 576 |
foreach ($redirections as $row) { |
| 577 |
if (isset($row->http_code) && in_array((int) $row->http_code, array(410, 451), true)) { |
| 578 |
continue; // Gone / Unavailable – no destination to follow |
| 579 |
} |
| 580 |
if (empty($row->url_redirect_to)) { |
| 581 |
continue; |
| 582 |
} |
| 583 |
$dest = $this->get_destination_for_row_and_uri($row, $uri); |
| 584 |
if ($dest !== null) { |
| 585 |
return $dest; |
| 586 |
} |
| 587 |
} |
| 588 |
return null; |
| 589 |
} |
| 590 |
|
| 591 |
/** |
| 592 |
* Determine whether creating a redirect from $source to $destination would |
| 593 |
* produce a redirect loop or chain longer than the configured hop budget. |
| 594 |
* |
| 595 |
* Read-only: no DB writes, no side effects. Reuses |
| 596 |
* get_redirect_destination_for_uri() so chain traversal matches the read path. |
| 597 |
* |
| 598 |
* @param string $source Source URL or path being created |
| 599 |
* @param string $destination Destination URL or path being created |
| 600 |
* @param array|null $chain Out param populated with the visited URI chain |
| 601 |
* @param int|null $max_hops Optional hop budget; defaults to filter `metasync_redirect_loop_max_hops` (5) |
| 602 |
* @return bool True if a loop or budget overrun is detected |
| 603 |
*/ |
| 604 |
public function would_create_loop($source, $destination, &$chain = null, $max_hops = null) |
| 605 |
{ |
| 606 |
if ($max_hops === null) { |
| 607 |
$max_hops = (int) apply_filters('metasync_redirect_loop_max_hops', 5); |
| 608 |
} |
| 609 |
if ($max_hops < 1) { |
| 610 |
$max_hops = 1; |
| 611 |
} |
| 612 |
|
| 613 |
$source_uri = $this->normalize_uri_path($source); |
| 614 |
$current = $this->normalize_uri_path($destination); |
| 615 |
$chain = array($source_uri, $current); |
| 616 |
|
| 617 |
// Trivial direct loop: redirect points back at itself |
| 618 |
if ($current === $source_uri) { |
| 619 |
return true; |
| 620 |
} |
| 621 |
|
| 622 |
$seen = array(); |
| 623 |
for ($i = 0; $i < $max_hops; $i++) { |
| 624 |
if (isset($seen[$current])) { |
| 625 |
// Pre-existing cycle that does not involve the new source — not our concern |
| 626 |
return false; |
| 627 |
} |
| 628 |
$seen[$current] = true; |
| 629 |
|
| 630 |
// Try matching with and without trailing slash to handle inconsistent storage |
| 631 |
$dest = $this->get_redirect_destination_for_uri($current); |
| 632 |
if ($dest === null) { |
| 633 |
$alt = (substr($current, -1) === '/') ? rtrim($current, '/') : $current . '/'; |
| 634 |
if ($alt !== '' && $alt !== $current) { |
| 635 |
$dest = $this->get_redirect_destination_for_uri($alt); |
| 636 |
} |
| 637 |
} |
| 638 |
if ($dest === null) { |
| 639 |
return false; |
| 640 |
} |
| 641 |
|
| 642 |
$current = $this->normalize_uri_path($dest); |
| 643 |
|
| 644 |
$chain[] = $current; |
| 645 |
|
| 646 |
if ($current === $source_uri) { |
| 647 |
return true; |
| 648 |
} |
| 649 |
} |
| 650 |
|
| 651 |
// Hop budget exhausted without resolving — treat as a loop-equivalent warning |
| 652 |
return true; |
| 653 |
} |
| 654 |
|
| 655 |
/** |
| 656 |
* Validate that a redirect would not create a loop. |
| 657 |
* |
| 658 |
* @param string $source Source URL or path |
| 659 |
* @param string $destination Destination URL or path |
| 660 |
* @return string|null Null if no loop, error message string if loop detected |
| 661 |
*/ |
| 662 |
public function validate_no_loop($source, $destination) { |
| 663 |
$chain = []; |
| 664 |
if ($this->would_create_loop($source, $destination, $chain)) { |
| 665 |
return 'Redirect would create a loop: ' . implode(' → ', $chain); |
| 666 |
} |
| 667 |
return null; |
| 668 |
} |
| 669 |
|
| 670 |
/** |
| 671 |
* Lightweight, non-blocking check that a redirect destination appears to resolve. |
| 672 |
* |
| 673 |
* Reuses the terminal dead-end logic from check_redirect_health(): internal |
| 674 |
* destinations are matched against url_to_postid(); external destinations are |
| 675 |
* only probed (wp_remote_head) when off-site redirects are explicitly allowed. |
| 676 |
* Returns a human-readable warning string when the destination looks |
| 677 |
* unreachable, or null when it resolves (or cannot be meaningfully checked). |
| 678 |
* |
| 679 |
* @param string $destination Destination URL or path. |
| 680 |
* @param int $http_code Redirect HTTP code (410/451 are skipped). |
| 681 |
* @param string $pattern_type Source pattern type ('regex' is skipped). |
| 682 |
* @return string|null Warning message, or null if the destination resolves / is unchecked. |
| 683 |
*/ |
| 684 |
public function destination_resolves_warning($destination, $http_code = 301, $pattern_type = 'exact') |
| 685 |
{ |
| 686 |
// 410/451 redirects intentionally have no live destination. |
| 687 |
if (in_array((int) $http_code, array(410, 451), true)) { |
| 688 |
return null; |
| 689 |
} |
| 690 |
|
| 691 |
// Regex sources have no single concrete destination path to resolve. |
| 692 |
if ($pattern_type === 'regex') { |
| 693 |
return null; |
| 694 |
} |
| 695 |
|
| 696 |
if (!is_string($destination) || trim($destination) === '') { |
| 697 |
return null; |
| 698 |
} |
| 699 |
|
| 700 |
// Determine if the destination is external by checking for a host component |
| 701 |
// that differs from the local site. Relative paths (e.g. /about) have no host |
| 702 |
// and are always internal. |
| 703 |
$is_external = false; |
| 704 |
$parsed_dest = wp_parse_url($destination); |
| 705 |
if (!empty($parsed_dest['host'])) { |
| 706 |
$site_host = wp_parse_url(site_url(), PHP_URL_HOST); |
| 707 |
$is_external = false === $site_host || strcasecmp($parsed_dest['host'], (string) $site_host) !== 0; |
| 708 |
} |
| 709 |
|
| 710 |
if ($is_external) { |
| 711 |
// Only probe external targets when off-site redirects are explicitly enabled, |
| 712 |
// and only with a short, single-hop HEAD request to keep the save lightweight. |
| 713 |
// 3xx responses are intentionally not flagged — the destination itself is |
| 714 |
// reachable even if it redirects further. |
| 715 |
if (get_option('metasync_allow_external_redirects')) { |
| 716 |
$response = wp_remote_head($destination, array('timeout' => 5, 'redirection' => 0)); |
| 717 |
if (is_wp_error($response)) { |
| 718 |
return 'Destination URL appears unreachable — the redirect may lead to an error.'; |
| 719 |
} |
| 720 |
$code = (int) wp_remote_retrieve_response_code($response); |
| 721 |
if ($code >= 400) { |
| 722 |
return 'Destination URL returned an error status (' . $code . ') — the redirect may lead to a broken page.'; |
| 723 |
} |
| 724 |
} |
| 725 |
return null; |
| 726 |
} |
| 727 |
|
| 728 |
// Internal destination: confirm it maps to a known page on this site. |
| 729 |
$path = $this->normalize_uri_path($destination); |
| 730 |
$post_id = url_to_postid(site_url($path)); |
| 731 |
if ($post_id === 0) { |
| 732 |
$post_id = url_to_postid(site_url($path . '/')); |
| 733 |
} |
| 734 |
|
| 735 |
if ($post_id === 0) { |
| 736 |
return 'Destination URL does not match a known page on this site — the redirect may lead to a 404.'; |
| 737 |
} |
| 738 |
|
| 739 |
return null; |
| 740 |
} |
| 741 |
|
| 742 |
/** |
| 743 |
* Normalise a URL or path to a leading-slash URI path used for chain comparison. |
| 744 |
* |
| 745 |
* @param string $url |
| 746 |
* @return string |
| 747 |
*/ |
| 748 |
private function normalize_uri_path($url) |
| 749 |
{ |
| 750 |
if (!is_string($url) || $url === '') { |
| 751 |
return '/'; |
| 752 |
} |
| 753 |
if (strpos($url, 'http') === 0) { |
| 754 |
$parsed = wp_parse_url($url); |
| 755 |
$path = isset($parsed['path']) ? $parsed['path'] : '/'; |
| 756 |
} else { |
| 757 |
$path = $url; |
| 758 |
} |
| 759 |
if ($path === '' || $path[0] !== '/') { |
| 760 |
$path = '/' . $path; |
| 761 |
} |
| 762 |
$path = rtrim($path, '/') ?: '/'; |
| 763 |
return $path; |
| 764 |
} |
| 765 |
|
| 766 |
/** |
| 767 |
* Check health of one or all active redirects. |
| 768 |
* |
| 769 |
* Returns per-redirect diagnostics: loop, chain_too_long, dead_end, or ok. |
| 770 |
* Uses a prebuilt lookup index for O(1) exact-match chain walking instead |
| 771 |
* of re-scanning all records per hop. |
| 772 |
* |
| 773 |
* @param int|null $redirect_id Optional single redirect ID to check. |
| 774 |
* @param int $max_hops Hops beyond which a chain is flagged (default 3). |
| 775 |
* @return array Array of health result objects. |
| 776 |
*/ |
| 777 |
public function check_redirect_health($redirect_id = null, $max_hops = 3) |
| 778 |
{ |
| 779 |
// Preload all active records once and build a lookup index |
| 780 |
$all_records = $this->db_redirection->getAllActiveRecords(); |
| 781 |
$exact_map = array(); // normalized_path => destination (O(1) lookup) |
| 782 |
$pattern_rows = array(); // non-exact rows requiring linear scan |
| 783 |
|
| 784 |
foreach ($all_records as $row) { |
| 785 |
if (isset($row->http_code) && in_array((int) $row->http_code, array(410, 451), true)) { |
| 786 |
continue; |
| 787 |
} |
| 788 |
if (empty($row->url_redirect_to)) { |
| 789 |
continue; |
| 790 |
} |
| 791 |
$sources_from = !empty($row->sources_from) |
| 792 |
? unserialize($row->sources_from, array('allowed_classes' => false)) |
| 793 |
: array(); |
| 794 |
$source_urls = is_array($sources_from) ? $sources_from : array(); |
| 795 |
$global_pattern_type = isset($row->pattern_type) ? $row->pattern_type : null; |
| 796 |
|
| 797 |
foreach ($source_urls as $source_key => $source_value) { |
| 798 |
// Legacy list-format rows store the URL as the VALUE under a |
| 799 |
// numeric key ('0' => '/old'); remap so health checks match |
| 800 |
// the URL rather than a bogus '/0'. |
| 801 |
if ((is_int($source_key) || ctype_digit((string) $source_key)) && is_string($source_value) && $source_value !== '') { |
| 802 |
$source_key = $source_value; |
| 803 |
} |
| 804 |
$pattern_type = in_array($source_value, array('exact', 'contain', 'start', 'end', 'wildcard', 'regex')) |
| 805 |
? $source_value |
| 806 |
: ($global_pattern_type ? $global_pattern_type : 'exact'); |
| 807 |
|
| 808 |
if ($pattern_type === 'exact' && strpos((string) $source_key, '*') === false) { |
| 809 |
$norm = $this->normalize_uri_path($source_key); |
| 810 |
$exact_map[$norm] = $row->url_redirect_to; |
| 811 |
} else { |
| 812 |
$pattern_rows[] = $row; |
| 813 |
break; // row already added, no need to check other sources |
| 814 |
} |
| 815 |
} |
| 816 |
} |
| 817 |
|
| 818 |
// Determine which records to check |
| 819 |
if ($redirect_id !== null) { |
| 820 |
$record = $this->db_redirection->find((int) $redirect_id); |
| 821 |
$records = $record ? array($record) : array(); |
| 822 |
} else { |
| 823 |
$records = $all_records; |
| 824 |
} |
| 825 |
|
| 826 |
$results = array(); |
| 827 |
|
| 828 |
foreach ($records as $row) { |
| 829 |
$id = isset($row->id) ? (int) $row->id : 0; |
| 830 |
$destination = isset($row->url_redirect_to) ? $row->url_redirect_to : ''; |
| 831 |
$http_code = isset($row->http_code) ? (int) $row->http_code : 301; |
| 832 |
|
| 833 |
// Extract first source path for display |
| 834 |
$sources_from = !empty($row->sources_from) |
| 835 |
? unserialize($row->sources_from, array('allowed_classes' => false)) |
| 836 |
: array(); |
| 837 |
$source_keys = is_array($sources_from) ? array_keys($sources_from) : array(); |
| 838 |
$source = !empty($source_keys) ? $source_keys[0] : ''; |
| 839 |
$source_path = $this->normalize_uri_path($source); |
| 840 |
|
| 841 |
// 410/451 have no destination — always ok |
| 842 |
if (in_array($http_code, array(410, 451), true)) { |
| 843 |
$results[] = array( |
| 844 |
'id' => $id, |
| 845 |
'source' => $source_path, |
| 846 |
'destination' => $destination, |
| 847 |
'final_destination' => null, |
| 848 |
'chain_length' => 0, |
| 849 |
'chain' => array($source_path), |
| 850 |
'status' => 'ok', |
| 851 |
); |
| 852 |
continue; |
| 853 |
} |
| 854 |
|
| 855 |
// Walk chain using the prebuilt index |
| 856 |
$current = $this->normalize_uri_path($destination); |
| 857 |
$chain = array($source_path, $current); |
| 858 |
$seen = array(); |
| 859 |
$is_loop = false; |
| 860 |
$hard_limit = 20; |
| 861 |
|
| 862 |
for ($i = 0; $i < $hard_limit; $i++) { |
| 863 |
if (isset($seen[$current])) { |
| 864 |
$is_loop = true; |
| 865 |
break; |
| 866 |
} |
| 867 |
$seen[$current] = true; |
| 868 |
|
| 869 |
// O(1) exact-match lookup first |
| 870 |
$dest = isset($exact_map[$current]) ? $exact_map[$current] : null; |
| 871 |
if ($dest === null) { |
| 872 |
// Try trailing slash variant |
| 873 |
$alt = (substr($current, -1) === '/') ? rtrim($current, '/') : $current . '/'; |
| 874 |
if ($alt !== '' && $alt !== $current) { |
| 875 |
$dest = isset($exact_map[$alt]) ? $exact_map[$alt] : null; |
| 876 |
} |
| 877 |
} |
| 878 |
// Fallback: scan pattern-based rows only (small set) |
| 879 |
if ($dest === null && !empty($pattern_rows)) { |
| 880 |
foreach ($pattern_rows as $prow) { |
| 881 |
$dest = $this->get_destination_for_row_and_uri($prow, $current); |
| 882 |
if ($dest !== null) { |
| 883 |
break; |
| 884 |
} |
| 885 |
// Try trailing slash alt for patterns too |
| 886 |
if (isset($alt)) { |
| 887 |
$dest = $this->get_destination_for_row_and_uri($prow, $alt); |
| 888 |
if ($dest !== null) { |
| 889 |
break; |
| 890 |
} |
| 891 |
} |
| 892 |
} |
| 893 |
} |
| 894 |
|
| 895 |
if ($dest === null) { |
| 896 |
break; |
| 897 |
} |
| 898 |
|
| 899 |
$current = $this->normalize_uri_path($dest); |
| 900 |
$chain[] = $current; |
| 901 |
} |
| 902 |
|
| 903 |
// Classify: loop > chain_too_long > dead_end > ok |
| 904 |
$chain_hops = count($chain) - 1; |
| 905 |
if ($is_loop) { |
| 906 |
$status = 'loop'; |
| 907 |
} elseif ($chain_hops > $max_hops) { |
| 908 |
$status = 'chain_too_long'; |
| 909 |
} else { |
| 910 |
// Check if terminal destination resolves to a real page |
| 911 |
$is_dead = false; |
| 912 |
if (function_exists('url_to_postid')) { |
| 913 |
$post_id = url_to_postid(site_url($current)); |
| 914 |
if ($post_id === 0) { |
| 915 |
$post_id = url_to_postid(site_url($current . '/')); |
| 916 |
} |
| 917 |
$is_dead = ($post_id === 0); |
| 918 |
// url_to_postid() knows only posts and pages — a taxonomy |
| 919 |
// archive (category/tag/custom term URL) returns 0 and a |
| 920 |
// healthy destination was flagged dead. Resolve those too. |
| 921 |
if ($is_dead && $this->uri_resolves_to_term_archive($current)) { |
| 922 |
$is_dead = false; |
| 923 |
} |
| 924 |
} |
| 925 |
$status = $is_dead ? 'dead_end' : 'ok'; |
| 926 |
} |
| 927 |
|
| 928 |
$results[] = array( |
| 929 |
'id' => $id, |
| 930 |
'source' => $source_path, |
| 931 |
'destination' => $this->normalize_uri_path($destination), |
| 932 |
'final_destination' => $current, |
| 933 |
'chain_length' => $chain_hops, |
| 934 |
'chain' => $chain, |
| 935 |
'status' => $status, |
| 936 |
); |
| 937 |
} |
| 938 |
|
| 939 |
return $results; |
| 940 |
} |
| 941 |
|
| 942 |
/** |
| 943 |
* Whether a URI path resolves to a taxonomy archive (category, tag, or a |
| 944 |
* custom taxonomy term). url_to_postid() returns 0 for these, so without |
| 945 |
* this check the health check reports them as dead ends. |
| 946 |
* |
| 947 |
* @param string $path URI path, e.g. /category/news/. |
| 948 |
* @return bool True when the path maps to an existing term archive. |
| 949 |
*/ |
| 950 |
private function uri_resolves_to_term_archive($path) |
| 951 |
{ |
| 952 |
if (!function_exists('get_term_by') || !function_exists('get_taxonomies')) { |
| 953 |
return false; |
| 954 |
} |
| 955 |
|
| 956 |
$path = trim((string) $path, '/'); |
| 957 |
if ($path === '') { |
| 958 |
return false; |
| 959 |
} |
| 960 |
$segments = explode('/', $path); |
| 961 |
$first = array_shift($segments); |
| 962 |
if ($first === '' || empty($segments)) { |
| 963 |
return false; |
| 964 |
} |
| 965 |
|
| 966 |
// Built-in taxonomies honor the category_base/tag_base options; the |
| 967 |
// defaults are 'category' and 'tag'. |
| 968 |
$category_base = trim((string) get_option('category_base'), '/'); |
| 969 |
$tag_base = trim((string) get_option('tag_base'), '/'); |
| 970 |
$bases = array( |
| 971 |
$category_base !== '' ? $category_base : 'category' => 'category', |
| 972 |
$tag_base !== '' ? $tag_base : 'tag' => 'post_tag', |
| 973 |
); |
| 974 |
$taxonomy = isset($bases[$first]) ? $bases[$first] : null; |
| 975 |
|
| 976 |
// Custom taxonomies carry their own rewrite base (defaults to the |
| 977 |
// taxonomy name). |
| 978 |
if ($taxonomy === null) { |
| 979 |
foreach (get_taxonomies(array('publicly_queryable' => true, '_builtin' => false), 'objects') as $tax_object) { |
| 980 |
$base = isset($tax_object->rewrite['slug']) ? trim((string) $tax_object->rewrite['slug'], '/') : $tax_object->name; |
| 981 |
if ($base === $first) { |
| 982 |
$taxonomy = $tax_object->name; |
| 983 |
break; |
| 984 |
} |
| 985 |
} |
| 986 |
} |
| 987 |
if ($taxonomy === null) { |
| 988 |
return false; |
| 989 |
} |
| 990 |
|
| 991 |
// get_term_by() returns WP_Term|false for a registered taxonomy — never |
| 992 |
// a WP_Error — so a false check is the whole story here. |
| 993 |
$term = get_term_by('slug', implode('/', $segments), $taxonomy); |
| 994 |
if ($term === false) { |
| 995 |
// Hierarchical terms live at parent/child paths, and the child's |
| 996 |
// own slug is only the last segment. |
| 997 |
$term = get_term_by('slug', end($segments), $taxonomy); |
| 998 |
} |
| 999 |
return ($term !== false); |
| 1000 |
} |
| 1001 |
|
| 1002 |
/** |
| 1003 |
* Handle AJAX health check request from admin UI. |
| 1004 |
*/ |
| 1005 |
public function handle_health_check_ajax() |
| 1006 |
{ |
| 1007 |
check_ajax_referer('metasync_redirect_health_check', 'nonce'); |
| 1008 |
|
| 1009 |
if (!Metasync::current_user_has_plugin_access()) { |
| 1010 |
wp_send_json_error(array('message' => 'Insufficient permissions.')); |
| 1011 |
return; |
| 1012 |
} |
| 1013 |
|
| 1014 |
$redirect_id = isset($_POST['redirect_id']) ? intval($_POST['redirect_id']) : null; |
| 1015 |
if ($redirect_id === 0) { |
| 1016 |
$redirect_id = null; |
| 1017 |
} |
| 1018 |
|
| 1019 |
$results = $this->check_redirect_health($redirect_id); |
| 1020 |
wp_send_json_success(array('results' => $results)); |
| 1021 |
} |
| 1022 |
|
| 1023 |
/** |
| 1024 |
* Get destination for a single row and URI if it matches. Same matching logic as source_url_redirection. |
| 1025 |
* |
| 1026 |
* @param object $row Redirect row |
| 1027 |
* @param string $uri URI to match |
| 1028 |
* @return string|null Destination URL/path or null |
| 1029 |
*/ |
| 1030 |
private function get_destination_for_row_and_uri($row, $uri) |
| 1031 |
{ |
| 1032 |
// Parse stored redirect sources (serialized: source path/URL => pattern type per source, or single list) |
| 1033 |
$sources_from = unserialize($row->sources_from, ['allowed_classes' => false]); // phpcs:ignore WordPress.PHP.DiscouragedPHPFunctions.serialize_unserialize |
| 1034 |
$source_urls = is_array($sources_from) ? $sources_from : array(); |
| 1035 |
$global_pattern_type = isset($row->pattern_type) ? $row->pattern_type : null; |
| 1036 |
$regex_pattern = isset($row->regex_pattern) ? $row->regex_pattern : null; |
| 1037 |
|
| 1038 |
foreach ($source_urls as $source_key => $source_value) { |
| 1039 |
$match_found = false; |
| 1040 |
$captured_path = ''; // Used for wildcard/regex replacement in destination (e.g. * or $1) |
| 1041 |
|
| 1042 |
// Legacy list-format rows store the URL as the VALUE under a |
| 1043 |
// numeric key ('0' => '/old'); remap so matching uses the URL. |
| 1044 |
if ((is_int($source_key) || ctype_digit((string) $source_key)) && is_string($source_value) && $source_value !== '') { |
| 1045 |
$source_key = $source_value; |
| 1046 |
} |
| 1047 |
|
| 1048 |
// Heal magic-quoted sources saved before the input-unslash fix so |
| 1049 |
// this health-check view matches with the same semantics as the |
| 1050 |
// live front-end matcher. |
| 1051 |
$source_key = wp_unslash((string) $source_key); |
| 1052 |
|
| 1053 |
// Resolve pattern type: per-source value (exact, contain, start, end, wildcard, regex) or row-level default |
| 1054 |
$pattern_type = in_array($source_value, array('exact', 'contain', 'start', 'end', 'wildcard', 'regex')) |
| 1055 |
? $source_value |
| 1056 |
: ($global_pattern_type ? $global_pattern_type : 'exact'); |
| 1057 |
|
| 1058 |
$normalized_uri = $uri; |
| 1059 |
$normalized_source = $source_key; |
| 1060 |
|
| 1061 |
// If source is a full URL, use only the path for matching (consistent with front-end redirect behavior) |
| 1062 |
if (strpos($source_key, 'http') === 0) { |
| 1063 |
$parsed_src = wp_parse_url($source_key); |
| 1064 |
$normalized_source = isset($parsed_src['path']) ? $parsed_src['path'] : ''; |
| 1065 |
} |
| 1066 |
|
| 1067 |
// Ensure leading slash for reliable path comparison |
| 1068 |
if (!empty($normalized_source) && $normalized_source[0] !== '/') { |
| 1069 |
$normalized_source = '/' . $normalized_source; |
| 1070 |
} |
| 1071 |
if (!empty($normalized_uri) && $normalized_uri[0] !== '/') { |
| 1072 |
$normalized_uri = '/' . $normalized_uri; |
| 1073 |
} |
| 1074 |
|
| 1075 |
// Normalize trailing slashes so /path and /path/ match equivalently |
| 1076 |
$normalized_source = rtrim($normalized_source, '/') ?: '/'; |
| 1077 |
$normalized_uri = rtrim($normalized_uri, '/') ?: '/'; |
| 1078 |
|
| 1079 |
// --- Matching: regex, wildcard, or legacy pattern types --- |
| 1080 |
|
| 1081 |
if ($pattern_type === 'regex' && $regex_pattern) { |
| 1082 |
// Regex: validate and normalize pattern, then match; capture group 1 for $1 in destination |
| 1083 |
if (!$this->validate_regex_pattern($regex_pattern)) { |
| 1084 |
continue; |
| 1085 |
} |
| 1086 |
$normalized_pattern = $this->normalize_regex_pattern($regex_pattern); |
| 1087 |
$matches = array(); |
| 1088 |
if (@preg_match($normalized_pattern, $normalized_uri, $matches) === 1) { |
| 1089 |
$match_found = true; |
| 1090 |
$captured_path = isset($matches[1]) ? $matches[1] : ''; |
| 1091 |
} |
| 1092 |
} else { |
| 1093 |
// Non-regex: check for * in source (wildcard) or use exact/contain/start/end |
| 1094 |
$has_wildcard = strpos($normalized_source, '*') !== false; |
| 1095 |
if ($has_wildcard) { |
| 1096 |
// Wildcard: e.g. /old/* matches /old/page and captures "page" for destination |
| 1097 |
$match_result = $this->match_wildcard($normalized_source, $normalized_uri); |
| 1098 |
if ($match_result !== false) { |
| 1099 |
$match_found = true; |
| 1100 |
$captured_path = $match_result; |
| 1101 |
} |
| 1102 |
} else { |
| 1103 |
// Legacy pattern: source_value can be the pattern type when key is the path |
| 1104 |
switch ($source_value) { |
| 1105 |
case 'exact': |
| 1106 |
if ($normalized_source === $normalized_uri) { |
| 1107 |
$match_found = true; |
| 1108 |
} |
| 1109 |
break; |
| 1110 |
case 'contain': |
| 1111 |
if ($this->contains($normalized_uri, $normalized_source)) { |
| 1112 |
$match_found = true; |
| 1113 |
} |
| 1114 |
break; |
| 1115 |
case 'start': |
| 1116 |
if (str_starts_with($normalized_uri, $normalized_source)) { |
| 1117 |
$match_found = true; |
| 1118 |
} |
| 1119 |
break; |
| 1120 |
case 'end': |
| 1121 |
if (str_ends_with($normalized_uri, $normalized_source)) { |
| 1122 |
$match_found = true; |
| 1123 |
} |
| 1124 |
break; |
| 1125 |
default: |
| 1126 |
// Fallback to row-level pattern_type when source_value is not a known type |
| 1127 |
switch ($pattern_type) { |
| 1128 |
case 'exact': |
| 1129 |
if ($normalized_source === $normalized_uri) { |
| 1130 |
$match_found = true; |
| 1131 |
} |
| 1132 |
break; |
| 1133 |
case 'contain': |
| 1134 |
if ($this->contains($normalized_uri, $normalized_source)) { |
| 1135 |
$match_found = true; |
| 1136 |
} |
| 1137 |
break; |
| 1138 |
case 'start': |
| 1139 |
if (str_starts_with($normalized_uri, $normalized_source)) { |
| 1140 |
$match_found = true; |
| 1141 |
} |
| 1142 |
break; |
| 1143 |
case 'end': |
| 1144 |
if (str_ends_with($normalized_uri, $normalized_source)) { |
| 1145 |
$match_found = true; |
| 1146 |
} |
| 1147 |
break; |
| 1148 |
case 'wildcard': |
| 1149 |
$match_result = $this->match_wildcard($normalized_source, $normalized_uri); |
| 1150 |
if ($match_result !== false) { |
| 1151 |
$match_found = true; |
| 1152 |
$captured_path = $match_result; |
| 1153 |
} |
| 1154 |
break; |
| 1155 |
} |
| 1156 |
break; |
| 1157 |
} |
| 1158 |
} |
| 1159 |
} |
| 1160 |
|
| 1161 |
// First match wins: return destination with * / $1 replaced by captured_path |
| 1162 |
if ($match_found && !empty($row->url_redirect_to)) { |
| 1163 |
return $this->process_destination_url($row->url_redirect_to, $captured_path); |
| 1164 |
} |
| 1165 |
} |
| 1166 |
|
| 1167 |
return null; |
| 1168 |
} |
| 1169 |
|
| 1170 |
/** |
| 1171 |
* Match wildcard pattern against URI |
| 1172 |
* |
| 1173 |
* @param string $pattern Pattern with * wildcard |
| 1174 |
* @param string $uri URI to match against |
| 1175 |
* @return string|false Returns captured path on match, false otherwise |
| 1176 |
*/ |
| 1177 |
private function match_wildcard($pattern, $uri) |
| 1178 |
{ |
| 1179 |
// Sanitize: escape the entire pattern for safe regex use, then restore wildcards |
| 1180 |
$escaped = preg_quote($pattern, '/'); |
| 1181 |
// preg_quote escapes *, so replace the escaped \* back with a capturing group |
| 1182 |
$regex = '/^' . str_replace('\\*', '(.*)', $escaped) . '$/'; |
| 1183 |
|
| 1184 |
$matches = []; |
| 1185 |
$result = @preg_match($regex, $uri, $matches); |
| 1186 |
|
| 1187 |
if ($result && $result !== false) { |
| 1188 |
// Return the captured path (first capturing group) |
| 1189 |
return isset($matches[1]) ? $matches[1] : ''; |
| 1190 |
} |
| 1191 |
|
| 1192 |
return false; |
| 1193 |
} |
| 1194 |
|
| 1195 |
/** |
| 1196 |
* Process destination URL with captured path |
| 1197 |
* |
| 1198 |
* @param string $destination Destination URL (may contain * or $1) |
| 1199 |
* @param string $captured_path Captured path from source |
| 1200 |
* @return string Processed destination URL |
| 1201 |
*/ |
| 1202 |
private function process_destination_url($destination, $captured_path) |
| 1203 |
{ |
| 1204 |
// Replace * wildcard with captured path |
| 1205 |
if (strpos($destination, '*') !== false) { |
| 1206 |
$destination = str_replace('*', $captured_path ?? '', $destination); |
| 1207 |
} |
| 1208 |
|
| 1209 |
// Replace $1 placeholder with captured path (for regex compatibility) |
| 1210 |
if (strpos($destination, '$1') !== false) { |
| 1211 |
$destination = str_replace('$1', $captured_path ?? '', $destination); |
| 1212 |
} |
| 1213 |
|
| 1214 |
return $destination; |
| 1215 |
} |
| 1216 |
|
| 1217 |
/** |
| 1218 |
* Preserve query parameters from the incoming request when redirecting. |
| 1219 |
* |
| 1220 |
* Redirect matching intentionally ignores the query string, but the |
| 1221 |
* browser-visible redirect must not silently discard it. Keep the stored |
| 1222 |
* destination query intact and append the incoming query verbatim so |
| 1223 |
* encoded values and repeated parameters survive unchanged. |
| 1224 |
* |
| 1225 |
* @param string $destination Destination URL or path. |
| 1226 |
* @param string $query Incoming query without the leading '?'. |
| 1227 |
* @return string Destination with the incoming query appended. |
| 1228 |
*/ |
| 1229 |
private function append_query_string($destination, $query) |
| 1230 |
{ |
| 1231 |
$destination = (string) $destination; |
| 1232 |
$query = ltrim((string) $query, '?'); |
| 1233 |
if ($query === '') { |
| 1234 |
return $destination; |
| 1235 |
} |
| 1236 |
|
| 1237 |
$fragment = ''; |
| 1238 |
$fragment_pos = strpos($destination, '#'); |
| 1239 |
if ($fragment_pos !== false) { |
| 1240 |
$fragment = substr($destination, $fragment_pos); |
| 1241 |
$destination = substr($destination, 0, $fragment_pos); |
| 1242 |
} |
| 1243 |
|
| 1244 |
$separator = strpos($destination, '?') === false ? '?' : '&'; |
| 1245 |
return $destination . $separator . $query . $fragment; |
| 1246 |
} |
| 1247 |
|
| 1248 |
/** |
| 1249 |
* Handle template redirect for frontend redirections |
| 1250 |
*/ |
| 1251 |
public function handle_template_redirect() |
| 1252 |
{ |
| 1253 |
// Only process on frontend |
| 1254 |
if (is_admin()) { |
| 1255 |
return; |
| 1256 |
} |
| 1257 |
|
| 1258 |
// Get current URI. wp_magic_quotes() slashes $_SERVER too, so unslash |
| 1259 |
// here to compare against the literal (unslashed) stored sources. |
| 1260 |
$request_uri = isset($_SERVER['REQUEST_URI']) ? wp_unslash((string) $_SERVER['REQUEST_URI']) : '/'; |
| 1261 |
|
| 1262 |
// Keep the raw query for the redirect destination. It is deliberately |
| 1263 |
// excluded from matching below, but must survive the browser redirect. |
| 1264 |
$query_pos = strpos($request_uri, '?'); |
| 1265 |
$incoming_query = $query_pos === false ? '' : substr($request_uri, $query_pos + 1); |
| 1266 |
|
| 1267 |
// Remove query string for matching |
| 1268 |
$uri = $query_pos === false ? $request_uri : substr($request_uri, 0, $query_pos); |
| 1269 |
|
| 1270 |
// REQUEST_URI arrives percent-encoded while stored sources are raw |
| 1271 |
// UTF-8, so a stored '/日本' could never match '/%E6%97%A5%E6%9C%AC'. |
| 1272 |
// Decode before matching; the stored destination still drives output. |
| 1273 |
$decoded_uri = rawurldecode((string) $uri); |
| 1274 |
if ($decoded_uri !== '') { |
| 1275 |
$uri = $decoded_uri; |
| 1276 |
} |
| 1277 |
|
| 1278 |
// Build the lookup index (lazy, once per request) |
| 1279 |
$this->ensure_redirect_index(); |
| 1280 |
|
| 1281 |
// O(1) exact-match lookup first |
| 1282 |
$normalized_uri = $uri; |
| 1283 |
if (!empty($normalized_uri) && $normalized_uri[0] !== '/') { |
| 1284 |
$normalized_uri = '/' . $normalized_uri; |
| 1285 |
} |
| 1286 |
$normalized_uri = rtrim($normalized_uri, '/') ?: '/'; |
| 1287 |
|
| 1288 |
if (isset($this->exact_index[$normalized_uri])) { |
| 1289 |
if ($this->source_url_redirection($this->exact_index[$normalized_uri], $uri, $incoming_query)) { |
| 1290 |
return; |
| 1291 |
} |
| 1292 |
} |
| 1293 |
|
| 1294 |
// Fallback: scan only pattern-based rows (wildcard, regex, contain, start, end) |
| 1295 |
foreach ($this->pattern_index as $redirection) { |
| 1296 |
if ($this->source_url_redirection($redirection, $uri, $incoming_query)) { |
| 1297 |
return; |
| 1298 |
} |
| 1299 |
} |
| 1300 |
} |
| 1301 |
|
| 1302 |
/** |
| 1303 |
* Prevent WordPress from redirecting to draft posts |
| 1304 |
* This stops WordPress from auto-redirecting URLs to ?p=POST_ID for draft posts |
| 1305 |
* |
| 1306 |
* @param string $redirect_url The redirect URL |
| 1307 |
* @param string $requested_url The requested URL |
| 1308 |
* @return string|false The redirect URL or false to cancel redirect |
| 1309 |
*/ |
| 1310 |
public function prevent_draft_post_redirects($redirect_url, $requested_url) |
| 1311 |
{ |
| 1312 |
# If no redirect is happening, return as-is |
| 1313 |
if (empty($redirect_url)) { |
| 1314 |
return $redirect_url; |
| 1315 |
} |
| 1316 |
|
| 1317 |
# Check if WordPress is trying to redirect to a ?p= or ?page_id= URL |
| 1318 |
if (strpos($redirect_url, '?p=') !== false || strpos($redirect_url, '?page_id=') !== false) { |
| 1319 |
# Extract the post ID |
| 1320 |
$post_id = null; |
| 1321 |
if (preg_match('/[?&]p=(\d+)/', $redirect_url, $matches)) { |
| 1322 |
$post_id = intval($matches[1]); |
| 1323 |
} elseif (preg_match('/[?&]page_id=(\d+)/', $redirect_url, $matches)) { |
| 1324 |
$post_id = intval($matches[1]); |
| 1325 |
} |
| 1326 |
|
| 1327 |
# If we found a post ID, check if it's a draft |
| 1328 |
if ($post_id) { |
| 1329 |
$post = get_post($post_id); |
| 1330 |
|
| 1331 |
# If post is draft, auto-draft, pending, or private, prevent the redirect |
| 1332 |
if ($post && in_array($post->post_status, ['draft', 'auto-draft', 'pending', 'private'])) { |
| 1333 |
# Return false to cancel the redirect and show 404 instead |
| 1334 |
return false; |
| 1335 |
} |
| 1336 |
} |
| 1337 |
} |
| 1338 |
|
| 1339 |
# Allow the redirect for published posts |
| 1340 |
return $redirect_url; |
| 1341 |
} |
| 1342 |
|
| 1343 |
/** |
| 1344 |
* Prevent wp_old_slug_redirect from redirecting to draft posts |
| 1345 |
* |
| 1346 |
* This uses WordPress's built-in 'old_slug_redirect_post_id' filter to selectively |
| 1347 |
* block redirects ONLY to unpublished posts, while allowing redirects to published posts. |
| 1348 |
* - It doesn't break existing WordPress functionality |
| 1349 |
* - Published posts can still use old slug redirects (good for SEO) |
| 1350 |
* - Only protects unpublished content from exposure |
| 1351 |
* - Non-invasive and backwards compatible |
| 1352 |
* |
| 1353 |
* @param int $post_id The post ID that WordPress wants to redirect to |
| 1354 |
* @return int|false The post ID to redirect to, or false to prevent redirect |
| 1355 |
*/ |
| 1356 |
public function prevent_old_slug_redirect_to_drafts($post_id) |
| 1357 |
{ |
| 1358 |
# If no post ID provided, don't redirect |
| 1359 |
if (empty($post_id)) { |
| 1360 |
return false; |
| 1361 |
} |
| 1362 |
|
| 1363 |
# Get the post |
| 1364 |
$post = get_post($post_id); |
| 1365 |
|
| 1366 |
# If post doesn't exist, don't redirect |
| 1367 |
if (!$post) { |
| 1368 |
return false; |
| 1369 |
} |
| 1370 |
|
| 1371 |
// Check if this post is unpublished (draft, pending, private, auto-draft) |
| 1372 |
if (in_array($post->post_status, ['draft', 'auto-draft', 'pending', 'private'])) { |
| 1373 |
# Return false to prevent the redirect to unpublished content |
| 1374 |
# This will cause WordPress to show 404 instead, protecting draft content |
| 1375 |
return false; |
| 1376 |
} |
| 1377 |
|
| 1378 |
# Allow redirect for published posts (preserves normal WordPress functionality) |
| 1379 |
return $post_id; |
| 1380 |
} |
| 1381 |
|
| 1382 |
/** |
| 1383 |
* Normalize regex pattern by adding delimiters if missing |
| 1384 |
* |
| 1385 |
* @param string $pattern The regex pattern |
| 1386 |
* @return string Pattern with delimiters |
| 1387 |
*/ |
| 1388 |
public static function normalize_regex_pattern($pattern) |
| 1389 |
{ |
| 1390 |
if (empty($pattern)) { |
| 1391 |
return $pattern; |
| 1392 |
} |
| 1393 |
|
| 1394 |
// Check if pattern starts with a common delimiter |
| 1395 |
$common_delimiters = ['/', '#', '~', '%', '@']; |
| 1396 |
$starts_with_delimiter = in_array($pattern[0], $common_delimiters); |
| 1397 |
|
| 1398 |
if ($starts_with_delimiter) { |
| 1399 |
// Pattern starts with delimiter, check if it has proper structure |
| 1400 |
$first_char = $pattern[0]; |
| 1401 |
$last_delimiter_pos = strrpos($pattern, $first_char); |
| 1402 |
|
| 1403 |
// If there's a closing delimiter at a different position, pattern likely has delimiters |
| 1404 |
if ($last_delimiter_pos !== false && $last_delimiter_pos > 0) { |
| 1405 |
// Check if what comes after the last delimiter are valid modifiers |
| 1406 |
$after_last_delimiter = substr($pattern, $last_delimiter_pos + 1); |
| 1407 |
// Valid modifiers: i, m, s, x, A, D, S, U, X, J, u |
| 1408 |
if (empty($after_last_delimiter) || preg_match('/^[imsxADSUXJu]*$/', $after_last_delimiter)) { |
| 1409 |
// Pattern appears to have proper delimiters, return as-is |
| 1410 |
return $pattern; |
| 1411 |
} |
| 1412 |
} |
| 1413 |
} |
| 1414 |
|
| 1415 |
// Pattern doesn't have delimiters or is malformed, add them |
| 1416 |
// Choose delimiter that's not in the pattern |
| 1417 |
$delimiters = ['/', '#', '~', '%', '@']; |
| 1418 |
$delimiter = '/'; |
| 1419 |
|
| 1420 |
foreach ($delimiters as $test_delimiter) { |
| 1421 |
if (strpos($pattern, $test_delimiter) === false) { |
| 1422 |
$delimiter = $test_delimiter; |
| 1423 |
break; |
| 1424 |
} |
| 1425 |
} |
| 1426 |
|
| 1427 |
return $delimiter . $pattern . $delimiter; |
| 1428 |
} |
| 1429 |
|
| 1430 |
/** |
| 1431 |
* Validate regex pattern |
| 1432 |
*/ |
| 1433 |
public function validate_regex_pattern($pattern) |
| 1434 |
{ |
| 1435 |
if (empty($pattern)) { |
| 1436 |
return true; // Empty pattern is valid (not required) |
| 1437 |
} |
| 1438 |
|
| 1439 |
// Normalize pattern (add delimiters if missing, fix malformed patterns) |
| 1440 |
$normalized_pattern = $this->normalize_regex_pattern($pattern); |
| 1441 |
|
| 1442 |
// Test if the regex pattern is valid |
| 1443 |
# $test_result = @preg_match($pattern, ''); |
| 1444 |
# return $test_result !== false; |
| 1445 |
|
| 1446 |
// Use error handler to catch warnings from malformed patterns |
| 1447 |
$error_occurred = false; |
| 1448 |
set_error_handler(function() use (&$error_occurred) { // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_set_error_handler -- genuine failure path, bounded, no secrets |
| 1449 |
$error_occurred = true; |
| 1450 |
return true; // Suppress the error |
| 1451 |
}, E_WARNING); |
| 1452 |
|
| 1453 |
$test_result = preg_match($normalized_pattern, ''); |
| 1454 |
|
| 1455 |
restore_error_handler(); |
| 1456 |
|
| 1457 |
// Return false if preg_match failed or if an error occurred |
| 1458 |
return $test_result !== false && !$error_occurred; |
| 1459 |
} |
| 1460 |
|
| 1461 |
/** |
| 1462 |
* Sanitize URL for redirection |
| 1463 |
*/ |
| 1464 |
public function sanitize_redirect_url($url) |
| 1465 |
{ |
| 1466 |
// Remove any dangerous protocols |
| 1467 |
$url = str_replace(['javascript:', 'data:', 'vbscript:'], '', $url); |
| 1468 |
|
| 1469 |
// Ensure it's a valid URL |
| 1470 |
if (!filter_var($url, FILTER_VALIDATE_URL) && !str_starts_with($url, '/')) { |
| 1471 |
// If it's not a full URL and doesn't start with /, assume it's a relative path |
| 1472 |
$url = '/' . ltrim($url, '/'); |
| 1473 |
} |
| 1474 |
|
| 1475 |
return esc_url_raw($url); |
| 1476 |
} |
| 1477 |
} |
| 1478 |
|
