| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
require_once __DIR__ . '/RedirectsRetentionPolicy.php'; |
| 8 |
require_once __DIR__ . '/../view-build/ViewReadRuntimeState.php'; |
| 9 |
|
| 10 |
/** |
| 11 |
* Repository operations for scheduled redirect and log cleanup. |
| 12 |
*/ |
| 13 |
class ABJ_404_Solution_RedirectsCleanupRepository { |
| 14 |
|
| 15 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 16 |
private $dbCore; |
| 17 |
|
| 18 |
/** @var ABJ_404_Solution_RedirectsRepositoryInterface */ |
| 19 |
private $redirectsRepo; |
| 20 |
|
| 21 |
/** @var ABJ_404_Solution_Functions */ |
| 22 |
private $f; |
| 23 |
|
| 24 |
/** @var ABJ_404_Solution_Logging */ |
| 25 |
private $logger; |
| 26 |
|
| 27 |
/** @var ABJ_404_Solution_RedirectsRetentionPolicy */ |
| 28 |
private $retentionPolicy; |
| 29 |
|
| 30 |
/** |
| 31 |
* @param ABJ_404_Solution_DatabaseCore $dbCore |
| 32 |
* @param ABJ_404_Solution_RedirectsRepositoryInterface $redirectsRepo |
| 33 |
* @param ABJ_404_Solution_Functions $functions |
| 34 |
* @param ABJ_404_Solution_Logging $logger |
| 35 |
* @param ABJ_404_Solution_RedirectsRetentionPolicy|null $retentionPolicy |
| 36 |
*/ |
| 37 |
public function __construct( |
| 38 |
ABJ_404_Solution_DatabaseCore $dbCore, |
| 39 |
ABJ_404_Solution_RedirectsRepositoryInterface $redirectsRepo, |
| 40 |
ABJ_404_Solution_Functions $functions, |
| 41 |
ABJ_404_Solution_Logging $logger, |
| 42 |
?ABJ_404_Solution_RedirectsRetentionPolicy $retentionPolicy = null |
| 43 |
) { |
| 44 |
$this->dbCore = $dbCore; |
| 45 |
$this->redirectsRepo = $redirectsRepo; |
| 46 |
$this->f = $functions; |
| 47 |
$this->logger = $logger; |
| 48 |
$this->retentionPolicy = $retentionPolicy !== null ? $retentionPolicy : new ABJ_404_Solution_RedirectsRetentionPolicy(); |
| 49 |
} |
| 50 |
|
| 51 |
public function cleanupOrphanedAutoRedirects(): int { |
| 52 |
$redirectsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_redirects}'); |
| 53 |
if (!$this->dbCore->tableNameResolver()->tableExists($redirectsTable)) { |
| 54 |
$this->logger->warn("Skipping orphaned redirect cleanup: table missing."); |
| 55 |
return 0; |
| 56 |
} |
| 57 |
|
| 58 |
$query = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/getOrphanedAutoRedirects.sql"); |
| 59 |
$query = $this->dbCore->doTableNameReplacements($query); |
| 60 |
$query = $this->f->doNormalReplacements($query); |
| 61 |
|
| 62 |
$results = $this->dbCore->queryAndGetResults($query); |
| 63 |
$rows = is_array($results['rows']) ? $results['rows'] : []; |
| 64 |
$deletedCount = 0; |
| 65 |
|
| 66 |
foreach ($rows as $row) { |
| 67 |
if (!is_array($row)) { |
| 68 |
continue; |
| 69 |
} |
| 70 |
$id = isset($row['id']) && is_scalar($row['id']) ? (string)$row['id'] : '0'; |
| 71 |
$url = isset($row['url']) && is_string($row['url']) ? $row['url'] : ''; |
| 72 |
$this->logger->debugMessage('Orphaned auto redirect deleted: "' . $url . '" (dest post ' . |
| 73 |
(isset($row['final_dest']) && is_scalar($row['final_dest']) ? (string)$row['final_dest'] : '?') . ' missing/unpublished).'); |
| 74 |
$this->redirectsRepo->deleteRedirect($id); |
| 75 |
$deletedCount++; |
| 76 |
} |
| 77 |
|
| 78 |
return $deletedCount; |
| 79 |
} |
| 80 |
|
| 81 |
/** |
| 82 |
* @param array<string, mixed> $options |
| 83 |
* @param int $now |
| 84 |
* @param string $optionKey |
| 85 |
* @param string $statusList |
| 86 |
* @param string $debugMessageType |
| 87 |
* @return int |
| 88 |
*/ |
| 89 |
public function deleteOldRedirectsByType($options, $now, $optionKey, $statusList, $debugMessageType) { |
| 90 |
$logsRepo = abj_service('logs_repository'); |
| 91 |
$deletedCount = 0; |
| 92 |
|
| 93 |
$deletionDays = $this->retentionPolicy->daysFromOptions(is_array($options) ? $options : array(), $optionKey); |
| 94 |
$then = $this->retentionPolicy->cutoffForDays($deletionDays, $now); |
| 95 |
if ($then === null) { |
| 96 |
return 0; |
| 97 |
} |
| 98 |
|
| 99 |
$this->dbCore->tableNameResolver()->setSqlBigSelects(); |
| 100 |
|
| 101 |
if (!$logsRepo->logsHitsTableExists()) { |
| 102 |
$this->logger->debugMessage(__FUNCTION__ . " skipping: logs_hits table missing; scheduling rebuild."); |
| 103 |
$logsRepo->scheduleHitsTableRebuild(); |
| 104 |
return 0; |
| 105 |
} |
| 106 |
|
| 107 |
$query = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/getMostUnusedRedirects.sql"); |
| 108 |
$query = $this->f->str_replace('{status_list}', $statusList, $query); |
| 109 |
$query = $this->f->str_replace('{timelimit}', (string)$then, $query); |
| 110 |
|
| 111 |
$results = $this->dbCore->queryAndGetResults($query); |
| 112 |
$rows = is_array($results['rows']) ? $results['rows'] : array(); |
| 113 |
|
| 114 |
foreach ($rows as $rowRaw) { |
| 115 |
if (!is_array($rowRaw)) { |
| 116 |
continue; |
| 117 |
} |
| 118 |
$row = $rowRaw; |
| 119 |
$fromUrl = isset($row['from_url']) && is_scalar($row['from_url']) ? (string)$row['from_url'] : ''; |
| 120 |
$lastUsed = isset($row['last_used_formatted']) && is_scalar($row['last_used_formatted']) |
| 121 |
? (string)$row['last_used_formatted'] |
| 122 |
: ''; |
| 123 |
$bestGuessDest = isset($row['best_guess_dest']) && is_scalar($row['best_guess_dest']) |
| 124 |
? (string)$row['best_guess_dest'] |
| 125 |
: ''; |
| 126 |
if ($debugMessageType === 'Captured 404') { |
| 127 |
$this->logger->debugMessage("Captured 404 for \"" . $fromUrl . |
| 128 |
'" deleted (last used: ' . $lastUsed . ').'); |
| 129 |
} else { |
| 130 |
$this->logger->debugMessage($debugMessageType . " from: " . $fromUrl . ' to: ' . |
| 131 |
$bestGuessDest . ' deleted (last used: ' . $lastUsed . ').'); |
| 132 |
} |
| 133 |
|
| 134 |
$this->redirectsRepo->deleteRedirect(isset($row['id']) && is_scalar($row['id']) ? (string)$row['id'] : '0'); |
| 135 |
$deletedCount++; |
| 136 |
} |
| 137 |
|
| 138 |
return $deletedCount; |
| 139 |
} |
| 140 |
|
| 141 |
public function deleteOldLogsByAge(int $daysToKeep, int $now): int { |
| 142 |
$cutoffTimestamp = $this->retentionPolicy->cutoffForDays($daysToKeep, $now); |
| 143 |
if ($cutoffTimestamp === null) { |
| 144 |
return 0; |
| 145 |
} |
| 146 |
|
| 147 |
$deletedTotal = 0; |
| 148 |
$batchSize = 2000; |
| 149 |
$maxBatches = 200; |
| 150 |
|
| 151 |
for ($i = 0; $i < $maxBatches; $i++) { |
| 152 |
$result = $this->dbCore->queryAndGetResults( |
| 153 |
"DELETE FROM {wp_abj404_logsv2} WHERE timestamp <= %d LIMIT %d", |
| 154 |
array( |
| 155 |
'query_params' => array($cutoffTimestamp, $batchSize), |
| 156 |
'log_errors' => true, |
| 157 |
) |
| 158 |
); |
| 159 |
$rowsDeletedRaw = $result['rows_affected'] ?? 0; |
| 160 |
$rowsDeleted = (is_int($rowsDeletedRaw) || is_float($rowsDeletedRaw) || is_string($rowsDeletedRaw)) |
| 161 |
? (int)$rowsDeletedRaw |
| 162 |
: 0; |
| 163 |
if ($rowsDeleted <= 0) { |
| 164 |
break; |
| 165 |
} |
| 166 |
$deletedTotal += $rowsDeleted; |
| 167 |
if ($rowsDeleted < $batchSize) { |
| 168 |
break; |
| 169 |
} |
| 170 |
} |
| 171 |
|
| 172 |
return $deletedTotal; |
| 173 |
} |
| 174 |
|
| 175 |
public function removeDuplicatesCron(): int { |
| 176 |
$rowsDeleted = 0; |
| 177 |
$query = "SELECT COUNT(id) as repetitions, url FROM {wp_abj404_redirects} GROUP BY url HAVING repetitions > 1 "; |
| 178 |
$result = $this->dbCore->queryAndGetResults($query); |
| 179 |
$outerRows = is_array($result['rows']) ? $result['rows'] : array(); |
| 180 |
foreach ($outerRows as $outerRow) { |
| 181 |
if (!is_array($outerRow)) { |
| 182 |
continue; |
| 183 |
} |
| 184 |
$row = $outerRow; |
| 185 |
$url = $row['url']; |
| 186 |
|
| 187 |
$queryr1 = $this->prepareQueryWp( |
| 188 |
"select id from {wp_abj404_redirects} where url = {url} order by timestamp desc limit 0,1", |
| 189 |
array("url" => $url) |
| 190 |
); |
| 191 |
$result = $this->dbCore->queryAndGetResults($queryr1); |
| 192 |
$innerRows = is_array($result['rows']) ? $result['rows'] : array(); |
| 193 |
if (count($innerRows) >= 1) { |
| 194 |
$row = is_array($innerRows[0]) ? $innerRows[0] : array(); |
| 195 |
$original = isset($row['id']) ? $row['id'] : 0; |
| 196 |
|
| 197 |
$queryl = $this->prepareQueryWp( |
| 198 |
"delete from {wp_abj404_redirects} where url = {url} and id != {original}", |
| 199 |
array("url" => $url, "original" => $original) |
| 200 |
); |
| 201 |
$deleteResult = $this->dbCore->queryAndGetResults($queryl); |
| 202 |
$affected = isset($deleteResult['rows_affected']) && is_numeric($deleteResult['rows_affected']) |
| 203 |
? (int)$deleteResult['rows_affected'] : 1; |
| 204 |
$rowsDeleted += max($affected, 1); |
| 205 |
} |
| 206 |
} |
| 207 |
|
| 208 |
if ($rowsDeleted > 0) { |
| 209 |
abj_service('view_read_service')->invalidateStatusCountsCache(); |
| 210 |
} |
| 211 |
|
| 212 |
return $rowsDeleted; |
| 213 |
} |
| 214 |
|
| 215 |
/** |
| 216 |
* @param array<string, mixed> $options |
| 217 |
* @return int |
| 218 |
*/ |
| 219 |
public function autoTrashJunkCapturedUrls(array $options): int { |
| 220 |
$enabled = $options['auto_trash_junk_urls'] ?? '0'; |
| 221 |
if ($enabled !== '1') { |
| 222 |
return 0; |
| 223 |
} |
| 224 |
|
| 225 |
$transientKey = 'abj404_last_auto_trash'; |
| 226 |
if (get_transient($transientKey) !== false) { |
| 227 |
return 0; |
| 228 |
} |
| 229 |
// allow-cache-empty: timestamp marker rate-limits automatic trash cleanup; not a cached query payload. |
| 230 |
set_transient($transientKey, abj_clock()->now(), HOUR_IN_SECONDS); |
| 231 |
|
| 232 |
$patternsRaw = $options['auto_trash_junk_patterns'] ?? ''; |
| 233 |
$patternsStr = is_string($patternsRaw) ? $patternsRaw : ''; |
| 234 |
$lines = array_filter(array_map('trim', explode("\n", $patternsStr))); |
| 235 |
|
| 236 |
if (empty($lines)) { |
| 237 |
return 0; |
| 238 |
} |
| 239 |
|
| 240 |
global $wpdb; |
| 241 |
$totalTrashed = 0; |
| 242 |
|
| 243 |
$likeClauses = array(); |
| 244 |
foreach ($lines as $pattern) { |
| 245 |
$escaped = $wpdb->esc_like($pattern); |
| 246 |
// DAO-bypass-approved: $wpdb->prepare is read-only string formatting; result goes through queryAndGetResults |
| 247 |
$likeClauses[] = $wpdb->prepare("url LIKE %s", '%' . $escaped . '%'); |
| 248 |
} |
| 249 |
|
| 250 |
$wherePatterns = implode(' OR ', $likeClauses); |
| 251 |
$query = "UPDATE {wp_abj404_redirects} |
| 252 |
SET disabled = 1 |
| 253 |
WHERE status = " . ABJ404_STATUS_CAPTURED . " |
| 254 |
AND disabled = 0 |
| 255 |
AND (" . $wherePatterns . ")"; |
| 256 |
$query = $this->dbCore->doTableNameReplacements($query); |
| 257 |
|
| 258 |
$result = $this->dbCore->queryAndGetResults($query); |
| 259 |
$affected = $result['rows_affected'] ?? 0; |
| 260 |
$totalTrashed += is_numeric($affected) ? (int)$affected : 0; |
| 261 |
|
| 262 |
$cutoff = abj_clock()->now() - (14 * DAY_IN_SECONDS); |
| 263 |
// DAO-bypass-approved: $wpdb->prepare is read-only string formatting; result goes through queryAndGetResults |
| 264 |
$query = $wpdb->prepare("UPDATE {wp_abj404_redirects} r |
| 265 |
SET r.disabled = 1 |
| 266 |
WHERE r.status = " . ABJ404_STATUS_CAPTURED . " |
| 267 |
AND r.disabled = 0 |
| 268 |
AND r.timestamp < %d |
| 269 |
AND NOT EXISTS ( |
| 270 |
SELECT 1 FROM {wp_abj404_logsv2} l |
| 271 |
WHERE l.requested_url = r.url |
| 272 |
LIMIT 1 |
| 273 |
)", |
| 274 |
$cutoff |
| 275 |
); |
| 276 |
$query = $this->dbCore->doTableNameReplacements($query); |
| 277 |
|
| 278 |
$result = $this->dbCore->queryAndGetResults($query); |
| 279 |
$affected = $result['rows_affected'] ?? 0; |
| 280 |
$totalTrashed += is_numeric($affected) ? (int)$affected : 0; |
| 281 |
|
| 282 |
if ($totalTrashed > 0) { |
| 283 |
$this->logger->infoMessage("Auto-trashed " . $totalTrashed . " junk/stale captured URLs during maintenance."); |
| 284 |
delete_transient(ABJ_404_Solution_ViewReadRuntimeState::CACHE_KEY_CAPTURED_STATUS); |
| 285 |
} |
| 286 |
|
| 287 |
return $totalTrashed; |
| 288 |
} |
| 289 |
|
| 290 |
public function expireOldAutoRedirects(int $days, int $now): int { |
| 291 |
$cutoff = $this->retentionPolicy->cutoffForDays($days, $now); |
| 292 |
if ($cutoff === null) { |
| 293 |
return 0; |
| 294 |
} |
| 295 |
|
| 296 |
$redirectsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_redirects}'); |
| 297 |
if (!$this->dbCore->tableNameResolver()->tableExists($redirectsTable)) { |
| 298 |
$this->logger->warn("expireOldAutoRedirects: redirects table missing, skipping."); |
| 299 |
return 0; |
| 300 |
} |
| 301 |
|
| 302 |
$sql = "SELECT id FROM `{$redirectsTable}` |
| 303 |
WHERE status = %d |
| 304 |
AND disabled = 0 |
| 305 |
AND `timestamp` > 0 |
| 306 |
AND `timestamp` < %d"; |
| 307 |
|
| 308 |
$result = $this->dbCore->queryAndGetResults($sql, array( |
| 309 |
'query_params' => array(ABJ404_STATUS_AUTO, $cutoff), |
| 310 |
)); |
| 311 |
|
| 312 |
if (!empty($result['timed_out']) || (isset($result['last_error']) && $result['last_error'] != '')) { |
| 313 |
return 0; |
| 314 |
} |
| 315 |
|
| 316 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 317 |
$ids = array(); |
| 318 |
foreach ($rows as $row) { |
| 319 |
if (is_array($row)) { |
| 320 |
$value = $row['id'] ?? reset($row); |
| 321 |
} elseif (is_object($row)) { |
| 322 |
$value = $row->id ?? null; |
| 323 |
} else { |
| 324 |
$value = $row; |
| 325 |
} |
| 326 |
if ($value !== null && $value !== '') { |
| 327 |
$ids[] = absint($value); |
| 328 |
} |
| 329 |
} |
| 330 |
|
| 331 |
if (empty($ids)) { |
| 332 |
return 0; |
| 333 |
} |
| 334 |
|
| 335 |
$moved = 0; |
| 336 |
foreach ($ids as $id) { |
| 337 |
$this->redirectsRepo->moveRedirectsToTrash($id, 1); |
| 338 |
$moved++; |
| 339 |
} |
| 340 |
|
| 341 |
$this->logger->infoMessage("expireOldAutoRedirects: moved {$moved} expired auto-redirect(s) to trash (threshold: {$days} days)."); |
| 342 |
return $moved; |
| 343 |
} |
| 344 |
|
| 345 |
/** |
| 346 |
* Token-style wpdb prepare helper used by removeDuplicatesCron. |
| 347 |
* |
| 348 |
* @param string $query |
| 349 |
* @param array<string, mixed> $data |
| 350 |
* @return string |
| 351 |
*/ |
| 352 |
private function prepareQueryWp($query, $data) { |
| 353 |
global $wpdb; |
| 354 |
$orderedValues = []; |
| 355 |
$preparedQuery = preg_replace_callback('/\{(\w+)\}/', function($matches) use ($data, &$orderedValues) { |
| 356 |
$key = $matches[1]; |
| 357 |
if (!isset($data[$key])) { |
| 358 |
return $matches[0]; |
| 359 |
} |
| 360 |
$value = $data[$key]; |
| 361 |
$orderedValues[] = $value; |
| 362 |
return is_int($value) ? '%d' : '%s'; |
| 363 |
}, $query); |
| 364 |
$preparedQuery = $preparedQuery !== null ? $preparedQuery : $query; |
| 365 |
// DAO-bypass-approved: $wpdb->prepare is read-only string formatting; callers execute the result through queryAndGetResults |
| 366 |
return $wpdb->prepare($preparedQuery, $orderedValues); |
| 367 |
} |
| 368 |
} |
| 369 |
|
