| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
trait ABJ_404_Solution_DataAccess_RedirectsTrait { |
| 8 |
|
| 9 |
/** |
| 10 |
* @param int|string $id |
| 11 |
* @return void |
| 12 |
*/ |
| 13 |
function deleteRedirect($id) { |
| 14 |
global $wpdb; |
| 15 |
$cleanedID = absint(sanitize_text_field((string)$id)); |
| 16 |
|
| 17 |
// no nonce here because this action is not always user generated. |
| 18 |
|
| 19 |
if (is_numeric($id)) { |
| 20 |
$query = "delete from {wp_abj404_redirects} where id = %d"; |
| 21 |
$this->queryAndGetResults($query, array('query_params' => array($cleanedID))); |
| 22 |
|
| 23 |
// Invalidate caches |
| 24 |
$this->invalidateStatusCountsCache(); |
| 25 |
$this->clearRegexRedirectsCache(); |
| 26 |
} |
| 27 |
} |
| 28 |
|
| 29 |
/** |
| 30 |
* Remove auto-created redirects whose destination post no longer exists or is not published. |
| 31 |
* |
| 32 |
* @return int Number of orphaned redirects deleted. |
| 33 |
*/ |
| 34 |
public function cleanupOrphanedAutoRedirects(): int { |
| 35 |
// Guard: skip if redirects table doesn't exist (prevents recurring cron errors) |
| 36 |
$redirectsTable = $this->doTableNameReplacements('{wp_abj404_redirects}'); |
| 37 |
if (!$this->tableExists($redirectsTable)) { |
| 38 |
$this->logger->warn("Skipping orphaned redirect cleanup: table missing."); |
| 39 |
return 0; |
| 40 |
} |
| 41 |
|
| 42 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getOrphanedAutoRedirects.sql"); |
| 43 |
$query = $this->doTableNameReplacements($query); |
| 44 |
$query = $this->f->doNormalReplacements($query); |
| 45 |
|
| 46 |
$results = $this->queryAndGetResults($query); |
| 47 |
$rows = is_array($results['rows']) ? $results['rows'] : []; |
| 48 |
$deletedCount = 0; |
| 49 |
|
| 50 |
foreach ($rows as $row) { |
| 51 |
if (!is_array($row)) { |
| 52 |
continue; |
| 53 |
} |
| 54 |
$id = isset($row['id']) && is_scalar($row['id']) ? (string)$row['id'] : '0'; |
| 55 |
$url = isset($row['url']) && is_string($row['url']) ? $row['url'] : ''; |
| 56 |
$this->logger->debugMessage('Orphaned auto redirect deleted: "' . $url . '" (dest post ' . |
| 57 |
(isset($row['final_dest']) && is_scalar($row['final_dest']) ? (string)$row['final_dest'] : '?') . ' missing/unpublished).'); |
| 58 |
$this->deleteRedirect($id); |
| 59 |
$deletedCount++; |
| 60 |
} |
| 61 |
|
| 62 |
return $deletedCount; |
| 63 |
} |
| 64 |
|
| 65 |
/** Helper method to delete old redirects of a specific type. |
| 66 |
* Extracted common logic from deleteOldRedirectsCron() to eliminate duplication. |
| 67 |
* |
| 68 |
* @param array<string, mixed> $options Plugin options |
| 69 |
* @param int $now Current timestamp |
| 70 |
* @param string $optionKey Option key for deletion threshold ('capture_deletion', 'auto_deletion', 'manual_deletion') |
| 71 |
* @param string $statusList Comma-separated list of status codes to delete |
| 72 |
* @param string $debugMessageType Type description for debug logging ('Captured 404', 'Automatic redirect', 'Manual redirect') |
| 73 |
* @return int Count of deleted redirects |
| 74 |
*/ |
| 75 |
private function deleteOldRedirectsByType($options, $now, $optionKey, $statusList, $debugMessageType) { |
| 76 |
$abj404dao = abj_service('data_access'); |
| 77 |
$deletedCount = 0; |
| 78 |
|
| 79 |
// Calculate time threshold |
| 80 |
$rawDays = $options[$optionKey] ?? 0; |
| 81 |
$deletionDays = intval(is_scalar($rawDays) ? $rawDays : 0); |
| 82 |
if ($deletionDays <= 0) { |
| 83 |
return 0; |
| 84 |
} |
| 85 |
$deletionTime = $deletionDays * 86400; |
| 86 |
$then = $now - $deletionTime; |
| 87 |
|
| 88 |
// setSqlBigSelects() must run before any branch that touches the |
| 89 |
// deletion query path so a large logs_hits/redirects join cannot |
| 90 |
// trip MAX_JOIN_SIZE on legacy hosts that still default it small. |
| 91 |
// Kept above the rollup-existence guard so callers still see the |
| 92 |
// session pragma flip even on the skip path. |
| 93 |
$this->setSqlBigSelects(); |
| 94 |
|
| 95 |
// F6 audit: getMostUnusedRedirects.sql joins logs_hits. If the rollup |
| 96 |
// does not exist yet (fresh install, never built), schedule a rebuild |
| 97 |
// and skip this run. The next daily cron will have it. Without the |
| 98 |
// table the LEFT JOIN errors instead of degrading to "never used". |
| 99 |
if (!$this->logsHitsTableExists()) { |
| 100 |
$this->logger->debugMessage(__FUNCTION__ . " skipping: logs_hits table missing; scheduling rebuild."); |
| 101 |
$this->scheduleHitsTableRebuild(); |
| 102 |
return 0; |
| 103 |
} |
| 104 |
|
| 105 |
// Load and prepare SQL query |
| 106 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getMostUnusedRedirects.sql"); |
| 107 |
$query = $this->f->str_replace('{status_list}', $statusList, $query); |
| 108 |
$query = $this->f->str_replace('{timelimit}', (string)$then, $query); |
| 109 |
|
| 110 |
// Execute query and get results |
| 111 |
$results = $this->queryAndGetResults($query); |
| 112 |
$rows = is_array($results['rows']) ? $results['rows'] : array(); |
| 113 |
|
| 114 |
// Delete each redirect and log |
| 115 |
foreach ($rows as $rowRaw) { |
| 116 |
if (!is_array($rowRaw)) { |
| 117 |
continue; |
| 118 |
} |
| 119 |
$row = $rowRaw; |
| 120 |
// Build debug message based on redirect type |
| 121 |
if ($debugMessageType === 'Captured 404') { |
| 122 |
$this->logger->debugMessage("Captured 404 for \"" . (is_string($row['from_url'] ?? '') ? $row['from_url'] : '') . |
| 123 |
'" deleted (last used: ' . (is_string($row['last_used_formatted'] ?? '') ? $row['last_used_formatted'] : '') . ').'); |
| 124 |
} else { |
| 125 |
// Auto and Manual redirects show from/to URLs |
| 126 |
$this->logger->debugMessage($debugMessageType . " from: " . (is_string($row['from_url'] ?? '') ? $row['from_url'] : '') . ' to: ' . |
| 127 |
(is_string($row['best_guess_dest'] ?? '') ? $row['best_guess_dest'] : '') . ' deleted (last used: ' . (is_string($row['last_used_formatted'] ?? '') ? $row['last_used_formatted'] : '') . ').'); |
| 128 |
} |
| 129 |
|
| 130 |
$abj404dao->deleteRedirect(isset($row['id']) && is_scalar($row['id']) ? (string)$row['id'] : '0'); |
| 131 |
$deletedCount++; |
| 132 |
} |
| 133 |
|
| 134 |
return $deletedCount; |
| 135 |
} |
| 136 |
|
| 137 |
/** |
| 138 |
* Delete old rows from the logs/protocol table based on age. |
| 139 |
* |
| 140 |
* Uses batch deletes to avoid a long-running single table lock. |
| 141 |
* |
| 142 |
* @param int $daysToKeep |
| 143 |
* @param int $now |
| 144 |
* @return int |
| 145 |
*/ |
| 146 |
private function deleteOldLogsByAge(int $daysToKeep, int $now): int { |
| 147 |
if ($daysToKeep <= 0) { |
| 148 |
return 0; |
| 149 |
} |
| 150 |
|
| 151 |
$cutoffTimestamp = max(0, $now - ($daysToKeep * 86400)); |
| 152 |
$deletedTotal = 0; |
| 153 |
$batchSize = 2000; |
| 154 |
$maxBatches = 200; |
| 155 |
|
| 156 |
for ($i = 0; $i < $maxBatches; $i++) { |
| 157 |
$result = $this->queryAndGetResults( |
| 158 |
"DELETE FROM {wp_abj404_logsv2} WHERE timestamp <= %d LIMIT %d", |
| 159 |
array( |
| 160 |
'query_params' => array($cutoffTimestamp, $batchSize), |
| 161 |
'log_errors' => true, |
| 162 |
) |
| 163 |
); |
| 164 |
$rowsDeletedRaw = $result['rows_affected'] ?? 0; |
| 165 |
$rowsDeleted = (is_int($rowsDeletedRaw) || is_float($rowsDeletedRaw) || is_string($rowsDeletedRaw)) |
| 166 |
? (int)$rowsDeletedRaw |
| 167 |
: 0; |
| 168 |
if ($rowsDeleted <= 0) { |
| 169 |
break; |
| 170 |
} |
| 171 |
$deletedTotal += $rowsDeleted; |
| 172 |
if ($rowsDeleted < $batchSize) { |
| 173 |
break; |
| 174 |
} |
| 175 |
} |
| 176 |
|
| 177 |
return $deletedTotal; |
| 178 |
} |
| 179 |
|
| 180 |
/** Delete old redirects based on how old they are. This runs daily. |
| 181 |
* @return string |
| 182 |
*/ |
| 183 |
function deleteOldRedirectsCron() { |
| 184 |
global $wpdb; |
| 185 |
$abj404dao = abj_service('data_access'); |
| 186 |
$abj404logic = abj_service('plugin_logic'); |
| 187 |
|
| 188 |
$options = $abj404logic->getOptions(); |
| 189 |
$now = time(); |
| 190 |
$capturedURLsCount = 0; |
| 191 |
$autoRedirectsCount = 0; |
| 192 |
$manualRedirectsCount = 0; |
| 193 |
$oldLogRowsDeletedBySize = 0; |
| 194 |
$oldLogRowsDeletedByAge = 0; |
| 195 |
|
| 196 |
// If true then the user clicked the button to execute the mantenance. |
| 197 |
$manually_fired = $abj404dao->getPostOrGetSanitize('manually_fired', 'false'); |
| 198 |
if ($this->f->strtolower($manually_fired) == 'true') { |
| 199 |
$manually_fired = true; |
| 200 |
} else { |
| 201 |
$manually_fired = false; |
| 202 |
} |
| 203 |
|
| 204 |
$upgradesEtc = abj_service('database_upgrades'); |
| 205 |
$upgradesEtc->createDatabaseTables(false); |
| 206 |
|
| 207 |
// Ensure database connection is active for long-running maintenance operations |
| 208 |
// This prevents "MySQL server has gone away" errors |
| 209 |
$this->ensureConnection(); |
| 210 |
|
| 211 |
// delete the export file |
| 212 |
$tempFile = $abj404logic->getExportFilename(); |
| 213 |
if (file_exists($tempFile)) { |
| 214 |
ABJ_404_Solution_Functions::safeUnlink($tempFile); |
| 215 |
} |
| 216 |
|
| 217 |
$duplicateRowsDeleted = $abj404dao->removeDuplicatesCron(); |
| 218 |
|
| 219 |
// Remove Captured URLs |
| 220 |
if (array_key_exists('capture_deletion', $options) && $options['capture_deletion'] != '0') { |
| 221 |
$status_list = ABJ404_STATUS_CAPTURED . ", " . ABJ404_STATUS_IGNORED . ", " . ABJ404_STATUS_LATER; |
| 222 |
$capturedURLsCount = $this->deleteOldRedirectsByType($options, $now, 'capture_deletion', $status_list, 'Captured 404'); |
| 223 |
$captureDeletionDays = intval(is_scalar($options['capture_deletion']) ? $options['capture_deletion'] : 0); |
| 224 |
$oldLogRowsDeletedByAge = $this->deleteOldLogsByAge($captureDeletionDays, $now); |
| 225 |
} |
| 226 |
|
| 227 |
// Remove Automatic Redirects |
| 228 |
if (isset($options['auto_deletion']) && $options['auto_deletion'] != '0') { |
| 229 |
$status_list = (string)ABJ404_STATUS_AUTO; |
| 230 |
$autoRedirectsCount = $this->deleteOldRedirectsByType($options, $now, 'auto_deletion', $status_list, 'Automatic redirect'); |
| 231 |
} |
| 232 |
|
| 233 |
// Remove Manual Redirects |
| 234 |
if (isset($options['manual_deletion']) && $options['manual_deletion'] != '0') { |
| 235 |
$status_list = ABJ404_STATUS_MANUAL . ", " . ABJ404_STATUS_REGEX; |
| 236 |
$manualRedirectsCount = $this->deleteOldRedirectsByType($options, $now, 'manual_deletion', $status_list, 'Manual redirect'); |
| 237 |
} |
| 238 |
|
| 239 |
// Remove orphaned auto redirects (destination post deleted/unpublished) |
| 240 |
$orphanedCount = $this->cleanupOrphanedAutoRedirects(); |
| 241 |
|
| 242 |
// Auto-trash junk/bot captured URLs |
| 243 |
$junkTrashedCount = $this->autoTrashJunkCapturedUrls($options); |
| 244 |
|
| 245 |
//Clean up old logs. prepare the query. get the disk usage in bytes. compare to the max requested |
| 246 |
// disk usage (MB to bytes). delete 1k rows at a time until the size is acceptable. |
| 247 |
$logsSizeBytes = $abj404dao->getLogDiskUsage(); |
| 248 |
$maxLogSizeBytes = (array_key_exists('maximum_log_disk_usage', $options) ? $options['maximum_log_disk_usage'] : 100) * 1024 * 1000; |
| 249 |
|
| 250 |
// Disk-size gate first: skip the trim entirely when we're under budget. |
| 251 |
// This keeps the daily cron path fast in the common case (no over-quota, |
| 252 |
// no scan, no destructive query) without relying on a row-count |
| 253 |
// approximation for any decision that drives DELETE. |
| 254 |
// |
| 255 |
// When we ARE over budget, pay for an exact COUNT(id) before computing |
| 256 |
// logLinesToDelete. An information_schema.TABLE_ROWS approximation |
| 257 |
// would be cheap, but for InnoDB it can drift by orders of magnitude |
| 258 |
// — using it as the denominator of a destructive DELETE … LIMIT N |
| 259 |
// query risks over-deleting retained logs (approx too low → |
| 260 |
// averageSizePerLine inflated → logLinesToKeep too small) or |
| 261 |
// under-deleting enough to miss the disk cap (approx too high → |
| 262 |
// reverse). The exact COUNT cost is paid only on the rare ticks |
| 263 |
// where we actually need to trim. |
| 264 |
if ($logsSizeBytes > $maxLogSizeBytes) { |
| 265 |
$totalLogLines = $abj404dao->getLogsCount(0); |
| 266 |
$averageSizePerLine = max($logsSizeBytes, 1) / max($totalLogLines, 1); |
| 267 |
$logLinesToKeep = ceil($maxLogSizeBytes / $averageSizePerLine); |
| 268 |
$logLinesToDelete = max($totalLogLines - $logLinesToKeep, 0); |
| 269 |
if ($logLinesToDelete > 0) { |
| 270 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/deleteOldLogs.sql"); |
| 271 |
$query = $this->f->str_replace('{lines_to_delete}', (string)$logLinesToDelete, $query); |
| 272 |
$results = $this->queryAndGetResults($query); |
| 273 |
$oldLogRowsDeletedBySizeRaw = $results['rows_affected'] ?? 0; |
| 274 |
$oldLogRowsDeletedBySize = (is_int($oldLogRowsDeletedBySizeRaw) || is_float($oldLogRowsDeletedBySizeRaw) || is_string($oldLogRowsDeletedBySizeRaw)) |
| 275 |
? (int)$oldLogRowsDeletedBySizeRaw |
| 276 |
: 0; |
| 277 |
} |
| 278 |
} |
| 279 |
|
| 280 |
$logsSizeBytes = $abj404dao->getLogDiskUsage(); |
| 281 |
$logSizeMB = round($logsSizeBytes / (1024 * 1000), 2); |
| 282 |
|
| 283 |
$renamed = $abj404dao->limitDebugFileSize(); |
| 284 |
$renamed = $renamed ? "true" : "false"; |
| 285 |
|
| 286 |
$oldLogRowsDeleted = $oldLogRowsDeletedByAge + $oldLogRowsDeletedBySize; |
| 287 |
|
| 288 |
$message = "deleteOldRedirectsCron. Old captured URLs removed: " . |
| 289 |
$capturedURLsCount . ", Old automatic redirects removed: " . $autoRedirectsCount . |
| 290 |
", Old manual redirects removed: " . $manualRedirectsCount . |
| 291 |
", Orphaned auto redirects removed: " . $orphanedCount . |
| 292 |
", Junk URLs auto-trashed: " . $junkTrashedCount . |
| 293 |
", Old log lines removed: " . $oldLogRowsDeleted . |
| 294 |
" (age: " . $oldLogRowsDeletedByAge . ", size: " . $oldLogRowsDeletedBySize . ")" . |
| 295 |
", New log size: " . $logSizeMB . "MB" . |
| 296 |
", Duplicate rows deleted: " . $duplicateRowsDeleted . ", Debug file size limited: " . |
| 297 |
$renamed; |
| 298 |
|
| 299 |
// only send a 404 notification email during daily maintenance. |
| 300 |
$adminEmailVal = array_key_exists('admin_notification_email', $options) ? $options['admin_notification_email'] : ''; |
| 301 |
if ($adminEmailVal !== null && |
| 302 |
$this->f->strlen(trim(is_string($adminEmailVal) ? $adminEmailVal : '')) > 5) { |
| 303 |
|
| 304 |
if ($manually_fired) { |
| 305 |
$message .= ', The admin email notification option is skipped for user ' |
| 306 |
. 'initiated maintenance runs.'; |
| 307 |
} else { |
| 308 |
$message .= ', ' . $abj404logic->emailCaptured404Notification(); |
| 309 |
} |
| 310 |
} else { |
| 311 |
$message .= ', Admin email notification option turned off.'; |
| 312 |
} |
| 313 |
|
| 314 |
if (isset($options['send_error_logs']) && |
| 315 |
$options['send_error_logs'] == '1') { |
| 316 |
if ($this->logger->emailErrorLogIfNecessary()) { |
| 317 |
$message .= ", Log file emailed to developer."; |
| 318 |
} else { |
| 319 |
// No error to report — roll the heartbeat dice. |
| 320 |
if ($this->logger->sendHeartbeatIfDueRandom(200)) { |
| 321 |
$message .= ", Heartbeat log emailed to developer."; |
| 322 |
} |
| 323 |
} |
| 324 |
} |
| 325 |
|
| 326 |
// Flag redirects whose destination URL is generating 404s (dead-destination detection). |
| 327 |
// This drives the "suspended redirect" warning in the admin table and allows |
| 328 |
// the frontend pipeline to skip known-bad destinations. |
| 329 |
$abj404dao->flagDeadDestinationRedirects(); |
| 330 |
|
| 331 |
// add some entries to the permalink cache if necessary |
| 332 |
$abj404permalinkCache = abj_service('permalink_cache'); |
| 333 |
$rowsUpdated = $abj404permalinkCache->updatePermalinkCache(15); |
| 334 |
$message .= ", Permlink cache rows updated: " . $rowsUpdated; |
| 335 |
|
| 336 |
$manually_fired_String = ($manually_fired) ? 'true' : 'false'; |
| 337 |
$message .= ", User initiated: " . $manually_fired_String; |
| 338 |
|
| 339 |
$this->logger->infoMessage($message); |
| 340 |
|
| 341 |
// fix any lingering errors |
| 342 |
$upgradesEtc = abj_service('database_upgrades'); |
| 343 |
$upgradesEtc->createDatabaseTables(); |
| 344 |
|
| 345 |
$this->queryAndGetResults("optimize table {wp_abj404_redirects}"); |
| 346 |
|
| 347 |
$upgradesEtc->updatePluginCheck(); |
| 348 |
|
| 349 |
return $message; |
| 350 |
} |
| 351 |
|
| 352 |
/** @return bool */ |
| 353 |
function limitDebugFileSize(): bool { |
| 354 |
$renamed = false; |
| 355 |
|
| 356 |
$mbFileSize = $this->logger->getDebugFileSize() / 1024 / 1000; |
| 357 |
if ($mbFileSize > 10) { |
| 358 |
$this->logger->limitDebugFileSize(); |
| 359 |
$renamed = true; |
| 360 |
} |
| 361 |
|
| 362 |
return $renamed; |
| 363 |
} |
| 364 |
|
| 365 |
/** Remove duplicates. |
| 366 |
* @return int |
| 367 |
*/ |
| 368 |
function removeDuplicatesCron(): int { |
| 369 |
$rowsDeleted = 0; |
| 370 |
$query = "SELECT COUNT(id) as repetitions, url FROM {wp_abj404_redirects} GROUP BY url HAVING repetitions > 1 "; |
| 371 |
$result = $this->queryAndGetResults($query); |
| 372 |
$outerRows = is_array($result['rows']) ? $result['rows'] : array(); |
| 373 |
foreach ($outerRows as $outerRow) { |
| 374 |
if (!is_array($outerRow)) { |
| 375 |
continue; |
| 376 |
} |
| 377 |
$row = $outerRow; |
| 378 |
$url = $row['url']; |
| 379 |
|
| 380 |
// Fix HIGH #2 (5th review): Use prepared statements instead of manual escaping |
| 381 |
$queryr1 = $this->prepare_query_wp( |
| 382 |
"select id from {wp_abj404_redirects} where url = {url} order by timestamp desc limit 0,1", |
| 383 |
array("url" => $url) |
| 384 |
); |
| 385 |
$result = $this->queryAndGetResults($queryr1); |
| 386 |
$innerRows = is_array($result['rows']) ? $result['rows'] : array(); |
| 387 |
if (count($innerRows) >= 1) { |
| 388 |
$row = is_array($innerRows[0]) ? $innerRows[0] : array(); |
| 389 |
$original = isset($row['id']) ? $row['id'] : 0; |
| 390 |
|
| 391 |
// Fix HIGH #2 (5th review): Use prepared statements instead of manual escaping |
| 392 |
$queryl = $this->prepare_query_wp( |
| 393 |
"delete from {wp_abj404_redirects} where url = {url} and id != {original}", |
| 394 |
array("url" => $url, "original" => $original) |
| 395 |
); |
| 396 |
$deleteResult = $this->queryAndGetResults($queryl); |
| 397 |
$affected = isset($deleteResult['rows_affected']) && is_numeric($deleteResult['rows_affected']) |
| 398 |
? (int)$deleteResult['rows_affected'] : 1; |
| 399 |
$rowsDeleted += max($affected, 1); |
| 400 |
} |
| 401 |
} |
| 402 |
|
| 403 |
// Invalidate status counts cache if any duplicates were removed |
| 404 |
if ($rowsDeleted > 0) { |
| 405 |
$this->invalidateStatusCountsCache(); |
| 406 |
} |
| 407 |
|
| 408 |
return $rowsDeleted; |
| 409 |
} |
| 410 |
|
| 411 |
/** |
| 412 |
* Canonical URL form stored in {wp_abj404_redirects}.canonical_url. |
| 413 |
* |
| 414 |
* Mirrors the SQL expression CONCAT('/', TRIM(BOTH '/' FROM url)) so the |
| 415 |
* captured-page JOIN against logs_hits.requested_url is a single indexed |
| 416 |
* equality lookup instead of evaluating CONCAT/TRIM per redirect row. |
| 417 |
* Both sides (the persisted column here and the rollup pre-aggregation |
| 418 |
* in getRedirectsForViewTempTable.sql) MUST produce byte-identical |
| 419 |
* output for the JOIN to match. |
| 420 |
* |
| 421 |
* Examples: 'foo' → '/foo', '/foo/' → '/foo', '' → '/', '/' → '/'. |
| 422 |
* |
| 423 |
* @param mixed $url |
| 424 |
* @return string |
| 425 |
*/ |
| 426 |
public static function computeRedirectsCanonicalUrl($url): string { |
| 427 |
if (!is_string($url)) { |
| 428 |
return '/'; |
| 429 |
} |
| 430 |
$trimmed = trim($url, '/'); |
| 431 |
if ($trimmed === '') { |
| 432 |
return '/'; |
| 433 |
} |
| 434 |
return '/' . $trimmed; |
| 435 |
} |
| 436 |
|
| 437 |
/** |
| 438 |
* Per-instance memoized cache of column-existence probes against the |
| 439 |
* redirects table. Keyed by lowercased column name. Empty array means |
| 440 |
* the cache has not been primed yet for any column. |
| 441 |
* |
| 442 |
* Per-instance (not static) so the DAO singleton resets on process end |
| 443 |
* without leaking state across test methods that build their own DAO. |
| 444 |
* |
| 445 |
* @var array<string, bool> |
| 446 |
*/ |
| 447 |
private $redirectsTableColumnsCache = array(); |
| 448 |
|
| 449 |
/** |
| 450 |
* Probe whether the redirects table has the named column right now. |
| 451 |
* |
| 452 |
* Used by setupRedirect() to drop columns from the INSERT payload when |
| 453 |
* they are missing on this site (e.g. canonical_url on installs where |
| 454 |
* dbDelta silently failed to ALTER ADD it). Result is cached for the |
| 455 |
* life of the DAO instance: the first call runs SHOW COLUMNS FROM |
| 456 |
* {wp_abj404_redirects}, every subsequent call hits the cache. |
| 457 |
* |
| 458 |
* Defensive against host failure: if SHOW COLUMNS errors out (table |
| 459 |
* missing, permission denied, etc.), returns true so the INSERT path |
| 460 |
* proceeds with the full payload. Better to surface the eventual |
| 461 |
* INSERT failure (which is already routed through queryAndGetResults |
| 462 |
* with auto-repair) than to silently strip a column we cannot probe. |
| 463 |
* |
| 464 |
* @param string $columnName |
| 465 |
* @return bool |
| 466 |
*/ |
| 467 |
private function redirectsTableHasColumn(string $columnName): bool { |
| 468 |
$key = strtolower($columnName); |
| 469 |
if ($this->redirectsTableColumnsCache !== array()) { |
| 470 |
// Cache primed: definitive yes/no for any column we saw on the |
| 471 |
// table. Absence in the cache means the column does not exist. |
| 472 |
return isset($this->redirectsTableColumnsCache[$key]); |
| 473 |
} |
| 474 |
global $wpdb; |
| 475 |
if (!isset($wpdb)) { |
| 476 |
// No DB handle (early-bootstrap path or test harness without |
| 477 |
// wpdb global). Default permissive so the caller's normal |
| 478 |
// payload path runs; do not cache so a later call retries. |
| 479 |
return true; |
| 480 |
} |
| 481 |
$redirectsTable = $this->doTableNameReplacements("{wp_abj404_redirects}"); |
| 482 |
// Probe via queryAndGetResults so the DAO's centralized error |
| 483 |
// classification, retry, and timeout handling apply. log_errors=false |
| 484 |
// because a missing-table or permission-denied response is a probe |
| 485 |
// signal here, not a bug to surface to the admin. |
| 486 |
// @utf8-audit: opt-out - $redirectsTable is built by doTableNameReplacements |
| 487 |
// from $wpdb->prefix plus the literal token {wp_abj404_redirects}; no user |
| 488 |
// input flows in, so invalid-UTF-8 to SQL is structurally impossible. |
| 489 |
$result = $this->queryAndGetResults( |
| 490 |
"SHOW COLUMNS FROM `" . esc_sql($redirectsTable) . "`", |
| 491 |
array('log_errors' => false, 'log_too_slow' => false) |
| 492 |
); |
| 493 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 494 |
if ($rows === array()) { |
| 495 |
// Probe failed (table missing, permission denied, host-side |
| 496 |
// restriction). Default permissive so the centralized DAO error |
| 497 |
// handler classifies any resulting INSERT failure with full |
| 498 |
// context. Do not cache: the next call may succeed once the |
| 499 |
// missing-table auto-repair runs. |
| 500 |
return true; |
| 501 |
} |
| 502 |
$primed = array(); |
| 503 |
foreach ($rows as $row) { |
| 504 |
if (!is_array($row)) { continue; } |
| 505 |
foreach ($row as $field => $value) { |
| 506 |
if (strtolower((string)$field) !== 'field') { continue; } |
| 507 |
$primed[strtolower((string)$value)] = true; |
| 508 |
} |
| 509 |
} |
| 510 |
if ($primed === array()) { |
| 511 |
// SHOW COLUMNS returned rows but none had a parseable Field |
| 512 |
// entry. Treat as a probe failure (do not cache). |
| 513 |
return true; |
| 514 |
} |
| 515 |
$this->redirectsTableColumnsCache = $primed; |
| 516 |
return isset($this->redirectsTableColumnsCache[$key]); |
| 517 |
} |
| 518 |
|
| 519 |
/** |
| 520 |
* SQL expression that emits the canonical form of an arbitrary URL column. |
| 521 |
* |
| 522 |
* Used by the (rare) callsites that still need to canonicalize at JOIN |
| 523 |
* time — e.g. logs_hits rebuild, which canonicalizes logsv2.requested_url |
| 524 |
* before grouping. Persistent canonical_url on redirects rows is the fast |
| 525 |
* path; this expression is only for columns we cannot pre-compute. |
| 526 |
* |
| 527 |
* @param string $columnExpr A SQL column reference, e.g. "r.url". |
| 528 |
* @return string SQL fragment. |
| 529 |
*/ |
| 530 |
public static function hitsCanonicalUrlSqlExpression(string $columnExpr): string { |
| 531 |
return "CONCAT('/', TRIM(BOTH '/' FROM " . $columnExpr . "))"; |
| 532 |
} |
| 533 |
|
| 534 |
/** |
| 535 |
* Store a redirect for future use. |
| 536 |
* @global type $wpdb |
| 537 |
* @param string $fromURL |
| 538 |
* @param string $status ABJ404_STATUS_MANUAL etc |
| 539 |
* @param string $type ABJ404_TYPE_POST, ABJ404_TYPE_CAT, ABJ404_TYPE_TAG, etc. |
| 540 |
* @param string $final_dest |
| 541 |
* @param string $code |
| 542 |
* @param int $disabled |
| 543 |
* @param string|null $engine The matching engine that created this redirect (null for manual/unknown) |
| 544 |
* @param float|null $score Match confidence score (0-100), NULL for manual redirects |
| 545 |
* @return int |
| 546 |
*/ |
| 547 |
function setupRedirect($fromURL, $status, $type, $final_dest, $code, $disabled = 0, $engine = null, $score = null) { |
| 548 |
global $wpdb; |
| 549 |
|
| 550 |
// nonce is verified outside of this method. We can't verify here because |
| 551 |
// automatic redirects are sometimes created without user interaction. |
| 552 |
|
| 553 |
if (!is_numeric($type)) { |
| 554 |
$this->logger->errorMessage("Wrong data type for redirect. TYPE is non-numeric. From: " . |
| 555 |
esc_url($fromURL) . " to: " . esc_url($final_dest) . ", Type: " .esc_html($type) . ", Status: " . $status); |
| 556 |
} else if (!is_numeric($status)) { |
| 557 |
$this->logger->errorMessage("Wrong data type for redirect. STATUS is non-numeric. From: " . |
| 558 |
esc_url($fromURL) . " to: " . esc_url($final_dest) . ", Type: " .esc_html($type) . ", Status: " . $status); |
| 559 |
} |
| 560 |
|
| 561 |
$statusAsInt = is_numeric($status) ? absint($status) : -1; |
| 562 |
$typeAsInt = is_numeric($type) ? absint($type) : -1; |
| 563 |
|
| 564 |
// Guard: automatic redirects must point to a currently valid destination. |
| 565 |
// This prevents persisting "auto" rows with missing/unpublished targets. |
| 566 |
if ($statusAsInt === ABJ404_STATUS_AUTO && |
| 567 |
!$this->isValidAutomaticRedirectDestination($typeAsInt, $final_dest)) { |
| 568 |
$this->logger->debugMessage("Skipping automatic redirect with invalid destination. " . |
| 569 |
"From: " . esc_url($fromURL) . ", Dest: " . esc_html((string)$final_dest) . |
| 570 |
", Type: " . esc_html((string)$type) . ", Status: " . esc_html((string)$status)); |
| 571 |
return 0; |
| 572 |
} |
| 573 |
|
| 574 |
$insertId = 0; |
| 575 |
|
| 576 |
// if we should not capture a 404 then don't. |
| 577 |
if (!abj_service('request_context')->ignore_doprocess) { |
| 578 |
$now = time(); |
| 579 |
$redirectsTable = $this->doTableNameReplacements("{wp_abj404_redirects}"); |
| 580 |
|
| 581 |
// Normalize to relative path before storing (Issue #24) |
| 582 |
// Fix HIGH #1 (5th review): Abort operation if normalization fails |
| 583 |
// Storing un-normalized URLs causes permanent lookup failures |
| 584 |
$abj404logic = abj_service('plugin_logic'); |
| 585 |
$fromURL = $abj404logic->normalizeToRelativePath($fromURL); |
| 586 |
|
| 587 |
$insertData = array( |
| 588 |
'url' => $fromURL, |
| 589 |
'status' => $status, |
| 590 |
'type' => $type, |
| 591 |
'final_dest' => $final_dest, |
| 592 |
'code' => $code, |
| 593 |
'disabled' => $disabled, |
| 594 |
'timestamp' => $now, |
| 595 |
); |
| 596 |
$insertFormats = array('%s', '%d', '%d', '%s', '%d', '%d', '%d'); |
| 597 |
// Schema-drift tolerance: canonical_url shipped in 4.1.11. On |
| 598 |
// installs where dbDelta silently failed to add it, every |
| 599 |
// captured-404 INSERT errors with "Unknown column 'canonical_url' |
| 600 |
// in 'field list'" (1671 such errors observed on a single 4.1.12 |
| 601 |
// site over 10 days in the May 10 debug zip). Probe before |
| 602 |
// referencing the column. The probe result is cached per-request |
| 603 |
// by self::redirectsTableHasColumn() so the SHOW COLUMNS cost |
| 604 |
// amortizes across multiple captured-404 hits in the same request. |
| 605 |
// Pre-compute the canonical form so the captured-page JOIN to |
| 606 |
// logs_hits.requested_url is a single indexed equality lookup |
| 607 |
// instead of CONCAT('/', TRIM(...)) per row at query time. The |
| 608 |
// formula must stay in lockstep with hitsCanonicalUrlSqlExpression() |
| 609 |
// (read side) and the buildRedirectsCanonicalUrlChunk() backfill. |
| 610 |
if ($this->redirectsTableHasColumn('canonical_url')) { |
| 611 |
$insertData['canonical_url'] = self::computeRedirectsCanonicalUrl($fromURL); |
| 612 |
$insertFormats[] = '%s'; |
| 613 |
} |
| 614 |
if ($engine !== null) { |
| 615 |
$insertData['engine'] = substr((string)$engine, 0, 64); |
| 616 |
$insertFormats[] = '%s'; |
| 617 |
} |
| 618 |
if ($score !== null) { |
| 619 |
$insertData['score'] = round((float)$score, 2); |
| 620 |
$insertFormats[] = '%f'; |
| 621 |
} |
| 622 |
|
| 623 |
$insertSql = "INSERT INTO `" . $redirectsTable . "` (`" . |
| 624 |
implode('`, `', array_keys($insertData)) . "`) VALUES (" . |
| 625 |
implode(', ', $insertFormats) . ")"; |
| 626 |
$insertResult = $this->queryAndGetResults($insertSql, array( |
| 627 |
'query_params' => array_values($insertData), |
| 628 |
)); |
| 629 |
$insertIdRaw = $insertResult['insert_id'] ?? 0; |
| 630 |
$insertId = is_scalar($insertIdRaw) ? (int)$insertIdRaw : 0; |
| 631 |
|
| 632 |
// Invalidate caches |
| 633 |
$this->invalidateStatusCountsCache(); |
| 634 |
// Clear regex cache in case a regex redirect was added |
| 635 |
if ($status == ABJ404_STATUS_REGEX) { |
| 636 |
$this->clearRegexRedirectsCache(); |
| 637 |
} |
| 638 |
} |
| 639 |
|
| 640 |
return $insertId; |
| 641 |
} |
| 642 |
|
| 643 |
/** |
| 644 |
* Automatic redirects are only valid for published posts or existing terms. |
| 645 |
* If a destination is missing or unpublished, skip creating the auto redirect. |
| 646 |
* |
| 647 |
* @param int $type |
| 648 |
* @param mixed $finalDest |
| 649 |
* @return bool |
| 650 |
*/ |
| 651 |
private function isValidAutomaticRedirectDestination($type, $finalDest) { |
| 652 |
$destId = absint(is_scalar($finalDest) ? $finalDest : 0); |
| 653 |
|
| 654 |
if ($type === ABJ404_TYPE_POST) { |
| 655 |
if ($destId <= 0) { |
| 656 |
return false; |
| 657 |
} |
| 658 |
if (!function_exists('get_post')) { |
| 659 |
return true; |
| 660 |
} |
| 661 |
|
| 662 |
// Boundary normalizer: WP_Post shape-probing lives in the VO. |
| 663 |
$ref = ABJ_404_Solution_PostRef::fromWpPost(get_post($destId)); |
| 664 |
if ($ref === null) { |
| 665 |
return false; |
| 666 |
} |
| 667 |
return $ref->isPublished(); |
| 668 |
} |
| 669 |
|
| 670 |
if ($type === ABJ404_TYPE_CAT || $type === ABJ404_TYPE_TAG) { |
| 671 |
if ($destId <= 0) { |
| 672 |
return false; |
| 673 |
} |
| 674 |
if (!function_exists('get_term')) { |
| 675 |
return true; |
| 676 |
} |
| 677 |
|
| 678 |
$taxonomy = ($type === ABJ404_TYPE_CAT) ? 'category' : 'post_tag'; |
| 679 |
$term = get_term($destId, $taxonomy); |
| 680 |
if ($term === null || is_wp_error($term)) { |
| 681 |
return false; |
| 682 |
} |
| 683 |
return is_object($term); |
| 684 |
} |
| 685 |
|
| 686 |
// Homepage is always a valid auto redirect destination. |
| 687 |
if ($type === ABJ404_TYPE_HOME) { |
| 688 |
return true; |
| 689 |
} |
| 690 |
|
| 691 |
// Auto redirects should not target other types. |
| 692 |
return false; |
| 693 |
} |
| 694 |
|
| 695 |
/** Get the redirect for the URL. |
| 696 |
* |
| 697 |
* @param string $url |
| 698 |
* @param bool $degradedMode When true, the lookup tolerates a partially- |
| 699 |
* migrated schema by stripping predicates that reference columns |
| 700 |
* not yet present (e.g. r.start_ts / r.end_ts on installs that |
| 701 |
* have not run the 4.1.x scheduled-redirect migration). Skipping |
| 702 |
* scheduled-redirect filtering is far better than 100% of redirects |
| 703 |
* failing silently. The pipeline only enables this mode when |
| 704 |
* DB_VERSION lags ABJ404_VERSION and recovery has not closed the |
| 705 |
* gap, so the happy path pays no extra cost. |
| 706 |
* @return array<string, mixed> |
| 707 |
*/ |
| 708 |
function getActiveRedirectForURL($url, $degradedMode = false) { |
| 709 |
// Strip invalid UTF-8/control bytes but keep valid unicode for multilingual slugs. |
| 710 |
$url = $this->f->sanitizeInvalidUTF8($url); |
| 711 |
|
| 712 |
// Reject URLs still invalid after sanitization (bot garbage like %c0, null bytes) |
| 713 |
if (function_exists('mb_check_encoding') && !mb_check_encoding($url, 'UTF-8')) { |
| 714 |
return array('id' => 0); |
| 715 |
} |
| 716 |
|
| 717 |
// Normalize to relative path before querying (Issue #24) |
| 718 |
// Fix HIGH #1 (5th review): Abort operation if normalization fails |
| 719 |
// Querying with un-normalized URLs causes lookup failures |
| 720 |
$abj404logic = abj_service('plugin_logic'); |
| 721 |
$candidates = $abj404logic->getNormalizedUrlCandidates($url); |
| 722 |
foreach ($candidates as $candidate) { |
| 723 |
$redirect = $this->getActiveRedirectForNormalizedUrl($candidate, $degradedMode); |
| 724 |
if ($redirect['id'] !== 0) { |
| 725 |
return $redirect; |
| 726 |
} |
| 727 |
} |
| 728 |
|
| 729 |
return array('id' => 0); |
| 730 |
} |
| 731 |
|
| 732 |
/** Get the redirect for the URL. |
| 733 |
* @param string $url |
| 734 |
* @return array<string, mixed> |
| 735 |
*/ |
| 736 |
function getExistingRedirectForURL($url) { |
| 737 |
// Strip invalid UTF-8/control bytes but keep valid unicode for multilingual slugs. |
| 738 |
$url = $this->f->sanitizeInvalidUTF8($url); |
| 739 |
|
| 740 |
// Reject URLs still invalid after sanitization (bot garbage like %c0, null bytes) |
| 741 |
if (function_exists('mb_check_encoding') && !mb_check_encoding($url, 'UTF-8')) { |
| 742 |
return array('id' => 0); |
| 743 |
} |
| 744 |
|
| 745 |
// Normalize to relative path before querying (Issue #24) |
| 746 |
// Fix HIGH #1 (5th review): Abort operation if normalization fails |
| 747 |
// Querying with un-normalized URLs causes lookup failures |
| 748 |
$abj404logic = abj_service('plugin_logic'); |
| 749 |
$candidates = $abj404logic->getNormalizedUrlCandidates($url); |
| 750 |
foreach ($candidates as $candidate) { |
| 751 |
$redirect = $this->getExistingRedirectForNormalizedUrl($candidate); |
| 752 |
if ($redirect['id'] !== 0) { |
| 753 |
return $redirect; |
| 754 |
} |
| 755 |
} |
| 756 |
|
| 757 |
return array('id' => 0); |
| 758 |
} |
| 759 |
|
| 760 |
/** |
| 761 |
* @param string $url |
| 762 |
* @param bool $degradedMode See getActiveRedirectForURL(). |
| 763 |
* @return array<string, mixed> |
| 764 |
*/ |
| 765 |
private function getActiveRedirectForNormalizedUrl($url, $degradedMode = false) { |
| 766 |
$redirect = array(); |
| 767 |
|
| 768 |
// we look for two URLs that might match. one with a trailing slash and one without. |
| 769 |
// the one the user entered takes priority in case the admin added separate redirects for |
| 770 |
// cases with and without the slash (and for backward compatibility). |
| 771 |
$url1 = $url; |
| 772 |
$url2 = $url; |
| 773 |
if (substr($url, -1) === '/') { |
| 774 |
$url2 = rtrim($url, '/'); |
| 775 |
} else { |
| 776 |
$url2 = $url2 . '/'; |
| 777 |
} |
| 778 |
|
| 779 |
// join to the wp_posts table to make sure the post exists. |
| 780 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPermalinkFromURL.sql"); |
| 781 |
|
| 782 |
// Degraded mode: when the migration that adds r.start_ts/r.end_ts has |
| 783 |
// not run (or is failing repeatedly), the columns may be missing from |
| 784 |
// wp_abj404_redirects on this install. Strip the scheduled-redirect |
| 785 |
// predicates so the lookup still serves manual + automatic redirects. |
| 786 |
// Trade-off: scheduled redirects whose start/end window is currently |
| 787 |
// active will still match (good); scheduled redirects whose window has |
| 788 |
// not yet opened or has already closed will *also* match during this |
| 789 |
// window (acceptable — far better than 100% of redirects failing). |
| 790 |
if ($degradedMode && $this->redirectsTableMissingScheduledColumns()) { |
| 791 |
$query = $this->stripScheduledRedirectPredicates($query); |
| 792 |
} |
| 793 |
|
| 794 |
// Fix HIGH #2 (5th review): Use prepared statements instead of manual escaping |
| 795 |
$query = $this->prepare_query_wp($query, array("url1" => $url1, "url2" => $url2)); |
| 796 |
$query = $this->doTableNameReplacements($query); |
| 797 |
$query = $this->f->doNormalReplacements($query); |
| 798 |
$results = $this->queryAndGetResults($query); |
| 799 |
$rows = $results['rows']; |
| 800 |
|
| 801 |
if (is_array($rows)) { |
| 802 |
if (empty($rows)) { |
| 803 |
$redirect['id'] = 0; |
| 804 |
} else { |
| 805 |
foreach ($rows[0] as $key => $value) { |
| 806 |
$redirect[$key] = $value; |
| 807 |
} |
| 808 |
} |
| 809 |
} |
| 810 |
|
| 811 |
if (!isset($redirect['id'])) { |
| 812 |
$redirect['id'] = 0; |
| 813 |
} |
| 814 |
|
| 815 |
return $redirect; |
| 816 |
} |
| 817 |
|
| 818 |
/** |
| 819 |
* Determine whether wp_abj404_redirects is missing the scheduled-redirect |
| 820 |
* columns (start_ts / end_ts) added in the 4.1.x migration. Result is |
| 821 |
* cached in a transient so subsequent 404s do not re-run SHOW COLUMNS. |
| 822 |
* |
| 823 |
* The cache is short-lived on the "missing" branch (5 min) so that once |
| 824 |
* the migration finally runs we pick up the new columns quickly. On the |
| 825 |
* "present" branch we cache for 24 h — columns don't disappear once added, |
| 826 |
* so a long TTL keeps the happy path fast. |
| 827 |
*/ |
| 828 |
private function redirectsTableMissingScheduledColumns(): bool { |
| 829 |
$cacheKey = 'abj404_redirects_scheduled_cols_status'; |
| 830 |
if (function_exists('get_transient')) { |
| 831 |
$cached = get_transient($cacheKey); |
| 832 |
if ($cached === 'missing') { return true; } |
| 833 |
if ($cached === 'present') { return false; } |
| 834 |
} |
| 835 |
|
| 836 |
$tableName = $this->doTableNameReplacements('{wp_abj404_redirects}'); |
| 837 |
$columns = $this->getRedirectsTableColumns($tableName); |
| 838 |
|
| 839 |
// If we couldn't read the schema at all, do NOT strip predicates — |
| 840 |
// returning false keeps the standard query, which fails loudly rather |
| 841 |
// than masking a deeper problem. |
| 842 |
if (empty($columns)) { |
| 843 |
return false; |
| 844 |
} |
| 845 |
|
| 846 |
$colsLower = array_map('strtolower', $columns); |
| 847 |
$missing = !in_array('start_ts', $colsLower, true) |
| 848 |
|| !in_array('end_ts', $colsLower, true); |
| 849 |
|
| 850 |
if (function_exists('set_transient')) { |
| 851 |
$hour = defined('HOUR_IN_SECONDS') ? (int) HOUR_IN_SECONDS : 3600; |
| 852 |
set_transient( |
| 853 |
$cacheKey, |
| 854 |
$missing ? 'missing' : 'present', |
| 855 |
$missing ? 5 * 60 : 24 * $hour |
| 856 |
); |
| 857 |
} |
| 858 |
|
| 859 |
return $missing; |
| 860 |
} |
| 861 |
|
| 862 |
/** |
| 863 |
* Read column names for a table via SHOW COLUMNS. Returns [] on failure |
| 864 |
* so callers can decide to fall back to the standard query. |
| 865 |
* |
| 866 |
* @return array<int, string> |
| 867 |
*/ |
| 868 |
private function getRedirectsTableColumns(string $tableName): array { |
| 869 |
global $wpdb; |
| 870 |
if (!isset($wpdb)) { |
| 871 |
return []; |
| 872 |
} |
| 873 |
// @utf8-audit: opt-out — $tableName is always a system value |
| 874 |
// (doTableNameReplacements / $wpdb->prefix); never user input. |
| 875 |
$result = $this->queryAndGetResults( |
| 876 |
"SHOW COLUMNS FROM `" . esc_sql($tableName) . "`", |
| 877 |
array('log_errors' => false) |
| 878 |
); |
| 879 |
$rows = isset($result['rows']) && is_array($result['rows']) ? $result['rows'] : []; |
| 880 |
$lastError = isset($result['last_error']) && is_string($result['last_error']) ? $result['last_error'] : ''; |
| 881 |
if ($lastError !== '') { |
| 882 |
return []; |
| 883 |
} |
| 884 |
$columns = []; |
| 885 |
foreach ($rows as $row) { |
| 886 |
if (is_array($row) && isset($row['Field']) && is_string($row['Field'])) { |
| 887 |
$columns[] = $row['Field']; |
| 888 |
} |
| 889 |
} |
| 890 |
return $columns; |
| 891 |
} |
| 892 |
|
| 893 |
/** |
| 894 |
* Strip lines from getPermalinkFromURL.sql that reference r.start_ts or |
| 895 |
* r.end_ts. Used in degraded mode when those columns are missing on this |
| 896 |
* install. |
| 897 |
*/ |
| 898 |
private function stripScheduledRedirectPredicates(string $sql): string { |
| 899 |
$stripped = preg_replace( |
| 900 |
'/^[^\n]*\br\.(?:start_ts|end_ts)\b[^\n]*\R?/m', |
| 901 |
'', |
| 902 |
$sql |
| 903 |
); |
| 904 |
return is_string($stripped) ? $stripped : $sql; |
| 905 |
} |
| 906 |
|
| 907 |
/** |
| 908 |
* @param string $url |
| 909 |
* @return array<string, mixed> |
| 910 |
*/ |
| 911 |
private function getExistingRedirectForNormalizedUrl($url) { |
| 912 |
$redirect = array(); |
| 913 |
|
| 914 |
// a disabled value of '1' means in the trash. |
| 915 |
$query = $this->prepare_query_wp('select * from {wp_abj404_redirects} where BINARY url = BINARY {url} ' . |
| 916 |
" and disabled = 0 ", array("url" => $url)); |
| 917 |
$results = $this->queryAndGetResults($query); |
| 918 |
$rows = $results['rows']; |
| 919 |
|
| 920 |
if (is_array($rows)) { |
| 921 |
if (empty($rows)) { |
| 922 |
$redirect['id'] = 0; |
| 923 |
} else { |
| 924 |
foreach ($rows[0] as $key => $value) { |
| 925 |
$redirect[$key] = $value; |
| 926 |
} |
| 927 |
} |
| 928 |
} |
| 929 |
|
| 930 |
if (!isset($redirect['id'])) { |
| 931 |
$redirect['id'] = 0; |
| 932 |
} |
| 933 |
|
| 934 |
return $redirect; |
| 935 |
} |
| 936 |
|
| 937 |
// Published-content lookup queries (getPublishedPagesAndPostsIDs, |
| 938 |
// getPublishedImagesIDs, getPublishedTags, addURLToTermsRows, |
| 939 |
// getPublishedCategories) live in DataAccessTrait_PublishedContent. |
| 940 |
|
| 941 |
/** Delete stored redirects based on passed in POST data. |
| 942 |
* @return string |
| 943 |
*/ |
| 944 |
function deleteSpecifiedRedirects() { |
| 945 |
global $wpdb; |
| 946 |
$message = ""; |
| 947 |
|
| 948 |
// nonce already verified. |
| 949 |
|
| 950 |
if (!array_key_exists('sanity_purge', $_POST) || $_POST['sanity_purge'] != "1") { |
| 951 |
$message = __('Error: You didn\'t check the I understand checkbox. No purging of records for you!', '404-solution'); |
| 952 |
return $message; |
| 953 |
} |
| 954 |
|
| 955 |
if (!isset($_POST['types']) || $_POST['types'] == '') { |
| 956 |
$message = __('Error: No redirect types were selected. No purges will be done.', '404-solution'); |
| 957 |
return $message; |
| 958 |
} |
| 959 |
|
| 960 |
if (is_array($_POST['types'])) { |
| 961 |
$type = array_map('sanitize_text_field', $_POST['types']); |
| 962 |
} else { |
| 963 |
$type = sanitize_text_field($_POST['types']); |
| 964 |
} |
| 965 |
|
| 966 |
if (!is_array($type)) { |
| 967 |
$message = __('An unknown error has occurred.', '404-solution'); |
| 968 |
return $message; |
| 969 |
} |
| 970 |
|
| 971 |
$redirectTypes = array(); |
| 972 |
foreach ($type as $aType) { |
| 973 |
if (('' . $aType != ABJ404_TYPE_HOME) && ('' . $aType != ABJ404_TYPE_404_DISPLAYED)) { |
| 974 |
array_push($redirectTypes, absint($aType)); |
| 975 |
} |
| 976 |
} |
| 977 |
|
| 978 |
if (empty($redirectTypes)) { |
| 979 |
$message = __('Error: No valid redirect types were selected. Exiting.', '404-solution'); |
| 980 |
$this->logger->debugMessage("Error: No valid redirect types were selected. Types: " . |
| 981 |
wp_kses_post((string)json_encode($redirectTypes))); |
| 982 |
return $message; |
| 983 |
} |
| 984 |
$purge = isset($_POST['purgetype']) ? sanitize_text_field($_POST['purgetype']) : ''; |
| 985 |
|
| 986 |
if ($purge != 'abj404_logs' && $purge != 'abj404_redirects') { |
| 987 |
$message = __('Error: An invalid purge type was selected. Exiting.', '404-solution'); |
| 988 |
$this->logger->debugMessage("Error: An invalid purge type was selected. Type: " . |
| 989 |
wp_kses_post((string)json_encode($purge))); |
| 990 |
return $message; |
| 991 |
} |
| 992 |
|
| 993 |
// always add the type "0" because it's an invalid type that may exist in the databse. |
| 994 |
// Adding it here does some cleanup if any is necessary. |
| 995 |
array_push($redirectTypes, 0); |
| 996 |
|
| 997 |
// Ensure all values are integers to prevent SQL injection |
| 998 |
$redirectTypes = array_map('absint', $redirectTypes); |
| 999 |
$typesForSQL = implode(',', $redirectTypes); |
| 1000 |
|
| 1001 |
if ($purge == 'abj404_redirects') { |
| 1002 |
$query = "update {wp_abj404_redirects} set disabled = 1 where status in (" . $typesForSQL . ")"; |
| 1003 |
$purgeResult = $this->queryAndGetResults($query); |
| 1004 |
$rowsAffectedRaw = $purgeResult['rows_affected'] ?? 0; |
| 1005 |
$redirectCount = is_scalar($rowsAffectedRaw) ? (int)$rowsAffectedRaw : 0; |
| 1006 |
|
| 1007 |
// Invalidate caches so the admin table reflects the purge immediately |
| 1008 |
$this->invalidateStatusCountsCache(); |
| 1009 |
$this->clearRegexRedirectsCache(); |
| 1010 |
|
| 1011 |
$message .= sprintf( _n( '%s redirect entry was moved to the trash.', |
| 1012 |
'%s redirect entries were moved to the trash.', $redirectCount, '404-solution'), $redirectCount); |
| 1013 |
} |
| 1014 |
|
| 1015 |
return $message; |
| 1016 |
} |
| 1017 |
|
| 1018 |
/** |
| 1019 |
* This returns only the first column of the first row of the result. |
| 1020 |
* @global type $wpdb |
| 1021 |
* @param string $query a query that starts with "select count(id) from ..." |
| 1022 |
* @param array<int, mixed> $valueParams values to use to prepare the query. |
| 1023 |
* @return int the count (result) of the query. |
| 1024 |
*/ |
| 1025 |
|
| 1026 |
/** |
| 1027 |
* Get all conditions for a redirect, ordered by sort_order. |
| 1028 |
* |
| 1029 |
* Returns an empty array when the redirect has no conditions or when the |
| 1030 |
* conditions table does not yet exist (graceful degradation). |
| 1031 |
* |
| 1032 |
* @param int $redirectId |
| 1033 |
* @return array<int, array<string, mixed>> |
| 1034 |
*/ |
| 1035 |
public function getRedirectConditions(int $redirectId): array { |
| 1036 |
$table = $this->doTableNameReplacements('{wp_abj404_redirect_conditions}'); |
| 1037 |
|
| 1038 |
// Guard: conditions table may not exist on older installs before upgrade runs. |
| 1039 |
if (!$this->tableExists($table)) { |
| 1040 |
return []; |
| 1041 |
} |
| 1042 |
|
| 1043 |
$result = $this->queryAndGetResults( |
| 1044 |
"SELECT id, redirect_id, logic, condition_type, operator, value, sort_order |
| 1045 |
FROM `{$table}` |
| 1046 |
WHERE redirect_id = %d |
| 1047 |
ORDER BY sort_order ASC, id ASC", |
| 1048 |
array('query_params' => array($redirectId), 'log_errors' => false) |
| 1049 |
); |
| 1050 |
|
| 1051 |
$lastError = isset($result['last_error']) && is_string($result['last_error']) ? $result['last_error'] : ''; |
| 1052 |
if ($lastError !== '') { |
| 1053 |
$this->logger->warn("getRedirectConditions: DB error for redirect_id={$redirectId}: " . $lastError); |
| 1054 |
return []; |
| 1055 |
} |
| 1056 |
|
| 1057 |
$rows = isset($result['rows']) && is_array($result['rows']) ? $result['rows'] : []; |
| 1058 |
return $rows; |
| 1059 |
} |
| 1060 |
|
| 1061 |
/** |
| 1062 |
* Save conditions for a redirect (replaces all existing conditions). |
| 1063 |
* |
| 1064 |
* Passes through empty arrays safely — all existing conditions are deleted |
| 1065 |
* and nothing is inserted, which is the correct behaviour for "no conditions". |
| 1066 |
* |
| 1067 |
* @param int $redirectId |
| 1068 |
* @param array<int, array<string, mixed>> $conditions Array of condition arrays. |
| 1069 |
* Each must contain: logic, condition_type, operator, value, sort_order. |
| 1070 |
* @return void |
| 1071 |
*/ |
| 1072 |
public function saveRedirectConditions(int $redirectId, array $conditions): void { |
| 1073 |
$table = $this->doTableNameReplacements('{wp_abj404_redirect_conditions}'); |
| 1074 |
|
| 1075 |
// Guard: conditions table may not exist yet. |
| 1076 |
if (!$this->tableExists($table)) { |
| 1077 |
$this->logger->warn("saveRedirectConditions: conditions table missing — skipping save for redirect_id={$redirectId}."); |
| 1078 |
return; |
| 1079 |
} |
| 1080 |
|
| 1081 |
// Delete existing conditions for this redirect. |
| 1082 |
$deleteResult = $this->queryAndGetResults( |
| 1083 |
"DELETE FROM `{$table}` WHERE redirect_id = %d", |
| 1084 |
array('query_params' => array($redirectId), 'log_errors' => false) |
| 1085 |
); |
| 1086 |
$deleteError = isset($deleteResult['last_error']) && is_string($deleteResult['last_error']) ? $deleteResult['last_error'] : ''; |
| 1087 |
if ($deleteError !== '') { |
| 1088 |
$this->logger->warn("saveRedirectConditions: error deleting old conditions for redirect_id={$redirectId}: " . $deleteError); |
| 1089 |
} |
| 1090 |
|
| 1091 |
if (empty($conditions)) { |
| 1092 |
return; |
| 1093 |
} |
| 1094 |
|
| 1095 |
$allowedTypes = [ |
| 1096 |
'login_status', 'user_role', 'referrer', |
| 1097 |
'user_agent', 'ip_range', 'http_header', |
| 1098 |
]; |
| 1099 |
$allowedOperators = [ |
| 1100 |
'equals', 'contains', 'regex', |
| 1101 |
'not_equals', 'not_contains', 'cidr', |
| 1102 |
]; |
| 1103 |
$allowedLogic = ['AND', 'OR']; |
| 1104 |
|
| 1105 |
foreach ($conditions as $index => $cond) { |
| 1106 |
if (!is_array($cond)) { |
| 1107 |
continue; |
| 1108 |
} |
| 1109 |
|
| 1110 |
$logic = isset($cond['logic']) && is_string($cond['logic']) |
| 1111 |
? strtoupper(trim($cond['logic'])) : 'AND'; |
| 1112 |
$type = isset($cond['condition_type']) && is_string($cond['condition_type']) |
| 1113 |
? trim($cond['condition_type']) : ''; |
| 1114 |
$operator = isset($cond['operator']) && is_string($cond['operator']) |
| 1115 |
? trim($cond['operator']) : 'equals'; |
| 1116 |
$value = isset($cond['value']) && is_string($cond['value']) |
| 1117 |
? trim($cond['value']) : ''; |
| 1118 |
$sortOrder = isset($cond['sort_order']) ? absint(is_scalar($cond['sort_order']) ? $cond['sort_order'] : 0) : $index; |
| 1119 |
|
| 1120 |
// Validate required fields. |
| 1121 |
if (!in_array($logic, $allowedLogic, true)) { |
| 1122 |
$logic = 'AND'; |
| 1123 |
} |
| 1124 |
if (!in_array($type, $allowedTypes, true)) { |
| 1125 |
$this->logger->warn("saveRedirectConditions: unknown condition_type '{$type}' — skipping."); |
| 1126 |
continue; |
| 1127 |
} |
| 1128 |
if (!in_array($operator, $allowedOperators, true)) { |
| 1129 |
$operator = 'equals'; |
| 1130 |
} |
| 1131 |
// Truncate value to column max (1024 chars). |
| 1132 |
if (strlen($value) > 1024) { |
| 1133 |
$value = substr($value, 0, 1024); |
| 1134 |
} |
| 1135 |
|
| 1136 |
$insertResult = $this->queryAndGetResults( |
| 1137 |
"INSERT INTO `{$table}` (`redirect_id`, `logic`, `condition_type`, `operator`, `value`, `sort_order`) |
| 1138 |
VALUES (%d, %s, %s, %s, %s, %d)", |
| 1139 |
array( |
| 1140 |
'query_params' => array($redirectId, $logic, $type, $operator, $value, $sortOrder), |
| 1141 |
'log_errors' => false, |
| 1142 |
) |
| 1143 |
); |
| 1144 |
$insertError = isset($insertResult['last_error']) && is_string($insertResult['last_error']) ? $insertResult['last_error'] : ''; |
| 1145 |
if ($insertError !== '') { |
| 1146 |
$this->logger->warn("saveRedirectConditions: error inserting condition #{$index} for redirect_id={$redirectId}: " . $insertError); |
| 1147 |
} |
| 1148 |
} |
| 1149 |
} |
| 1150 |
|
| 1151 |
/** |
| 1152 |
* Auto-trash captured URLs that match known junk/bot patterns, and |
| 1153 |
* captured URLs with 0 hits older than 14 days. |
| 1154 |
* |
| 1155 |
* Rate-limited to once per hour via transient. |
| 1156 |
* |
| 1157 |
* @param array<string, mixed> $options Plugin options. |
| 1158 |
* @return int Number of URLs trashed. |
| 1159 |
*/ |
| 1160 |
function autoTrashJunkCapturedUrls(array $options): int { |
| 1161 |
// Feature must be enabled |
| 1162 |
$enabled = $options['auto_trash_junk_urls'] ?? '0'; |
| 1163 |
if ($enabled !== '1') { |
| 1164 |
return 0; |
| 1165 |
} |
| 1166 |
|
| 1167 |
// Rate limit: once per hour |
| 1168 |
$transientKey = 'abj404_last_auto_trash'; |
| 1169 |
if (get_transient($transientKey) !== false) { |
| 1170 |
return 0; |
| 1171 |
} |
| 1172 |
set_transient($transientKey, time(), HOUR_IN_SECONDS); |
| 1173 |
|
| 1174 |
$patternsRaw = $options['auto_trash_junk_patterns'] ?? ''; |
| 1175 |
$patternsStr = is_string($patternsRaw) ? $patternsRaw : ''; |
| 1176 |
$lines = array_filter(array_map('trim', explode("\n", $patternsStr))); |
| 1177 |
|
| 1178 |
if (empty($lines)) { |
| 1179 |
return 0; |
| 1180 |
} |
| 1181 |
|
| 1182 |
global $wpdb; |
| 1183 |
$totalTrashed = 0; |
| 1184 |
|
| 1185 |
// Build LIKE conditions for each pattern |
| 1186 |
$likeClauses = array(); |
| 1187 |
foreach ($lines as $pattern) { |
| 1188 |
$escaped = $wpdb->esc_like($pattern); |
| 1189 |
$likeClauses[] = $wpdb->prepare("url LIKE %s", '%' . $escaped . '%'); |
| 1190 |
} |
| 1191 |
|
| 1192 |
// Trash captured URLs matching junk patterns (case-insensitive via LIKE) |
| 1193 |
$wherePatterns = implode(' OR ', $likeClauses); |
| 1194 |
$query = "UPDATE {wp_abj404_redirects} |
| 1195 |
SET disabled = 1 |
| 1196 |
WHERE status = " . ABJ404_STATUS_CAPTURED . " |
| 1197 |
AND disabled = 0 |
| 1198 |
AND (" . $wherePatterns . ")"; |
| 1199 |
$query = $this->doTableNameReplacements($query); |
| 1200 |
|
| 1201 |
$result = $this->queryAndGetResults($query); |
| 1202 |
$affected = $result['rows_affected'] ?? 0; |
| 1203 |
$totalTrashed += is_numeric($affected) ? (int)$affected : 0; |
| 1204 |
|
| 1205 |
// Trash captured URLs with 0 log hits older than 14 days. |
| 1206 |
// logshits is not a column — it's computed from the logs table. |
| 1207 |
$cutoff = time() - (14 * DAY_IN_SECONDS); |
| 1208 |
$query = $wpdb->prepare( |
| 1209 |
"UPDATE {wp_abj404_redirects} r |
| 1210 |
SET r.disabled = 1 |
| 1211 |
WHERE r.status = " . ABJ404_STATUS_CAPTURED . " |
| 1212 |
AND r.disabled = 0 |
| 1213 |
AND r.timestamp < %d |
| 1214 |
AND NOT EXISTS ( |
| 1215 |
SELECT 1 FROM {wp_abj404_logsv2} l |
| 1216 |
WHERE l.requested_url = r.url |
| 1217 |
LIMIT 1 |
| 1218 |
)", |
| 1219 |
$cutoff |
| 1220 |
); |
| 1221 |
$query = $this->doTableNameReplacements($query); |
| 1222 |
|
| 1223 |
$result = $this->queryAndGetResults($query); |
| 1224 |
$affected = $result['rows_affected'] ?? 0; |
| 1225 |
$totalTrashed += is_numeric($affected) ? (int)$affected : 0; |
| 1226 |
|
| 1227 |
if ($totalTrashed > 0) { |
| 1228 |
$this->logger->infoMessage("Auto-trashed " . $totalTrashed . " junk/stale captured URLs during maintenance."); |
| 1229 |
// Invalidate the cached status counts so the UI reflects the change |
| 1230 |
delete_transient(self::CACHE_KEY_CAPTURED_STATUS); |
| 1231 |
} |
| 1232 |
|
| 1233 |
return $totalTrashed; |
| 1234 |
} |
| 1235 |
} |
| 1236 |
|
