| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
require_once __DIR__ . '/../DatabaseCollationHelper.php'; |
| 8 |
|
| 9 |
class ABJ_404_Solution_DatabaseUpgradeOrphanAdoption extends ABJ_404_Solution_DatabaseUpgradeComponent { |
| 10 |
|
| 11 |
/** |
| 12 |
* Detect orphaned plugin tables under old prefixes and adopt their data |
| 13 |
* into the current-prefix tables. Uses slug verification against the logs |
| 14 |
* table to confirm ownership before adopting. |
| 15 |
* |
| 16 |
* @return void |
| 17 |
*/ |
| 18 |
public function adoptOrphanedTables(): void { |
| 19 |
global $wpdb; |
| 20 |
|
| 21 |
$dbNameRaw = $wpdb->dbname ?? ''; |
| 22 |
if ($dbNameRaw === '') { |
| 23 |
return; |
| 24 |
} |
| 25 |
// @utf8-audit: opt-out — $wpdb->dbname is set by WordPress at |
| 26 |
// bootstrap from wp-config.php; never user input. |
| 27 |
$dbNameEscaped = esc_sql($dbNameRaw); |
| 28 |
$dbName = is_array($dbNameEscaped) ? '' : $dbNameEscaped; |
| 29 |
|
| 30 |
// Find all abj404 tables in the database, grouped by prefix. |
| 31 |
$query = "SELECT table_name |
| 32 |
FROM information_schema.tables |
| 33 |
WHERE table_schema = '{$dbName}' |
| 34 |
AND LOWER(table_name) LIKE '%abj404\\_%'"; |
| 35 |
$results = $this->dbCore->queryAndGetResults($query); |
| 36 |
|
| 37 |
if (!is_array($results['rows']) || empty($results['rows'])) { |
| 38 |
return; |
| 39 |
} |
| 40 |
|
| 41 |
$currentPrefix = $this->dbCore->tableNameResolver()->getLowercasePrefix(); |
| 42 |
|
| 43 |
// Group tables by their prefix (everything before 'abj404_'). |
| 44 |
$tablesByPrefix = $this->groupAbj404TablesByPrefix($results['rows']); |
| 45 |
|
| 46 |
// Skip prefixes we've already adopted. |
| 47 |
$adoptedPrefixes = get_option('abj404_adopted_prefixes', array()); |
| 48 |
if (!is_array($adoptedPrefixes)) { |
| 49 |
$adoptedPrefixes = array(); |
| 50 |
} |
| 51 |
|
| 52 |
// Authoritative tenant-isolation boundary: any prefix that belongs to an |
| 53 |
// active multisite blog (a sibling subsite) is NEVER an orphan to adopt, |
| 54 |
// regardless of how well its content happens to match this site. The |
| 55 |
// slug-match heuristic below is only a tie-breaker among genuine |
| 56 |
// migration leftovers, never the authorization boundary. |
| 57 |
$activeBlogPrefixes = $this->getActiveBlogPrefixesLowercase(); |
| 58 |
|
| 59 |
// Process each OLD prefix (not the current one). |
| 60 |
foreach ($tablesByPrefix as $oldPrefix => $tables) { |
| 61 |
$oldPrefix = (string)$oldPrefix; |
| 62 |
if ($oldPrefix === $currentPrefix) { |
| 63 |
continue; |
| 64 |
} |
| 65 |
if (in_array($oldPrefix, $adoptedPrefixes, true)) { |
| 66 |
continue; |
| 67 |
} |
| 68 |
if (in_array($oldPrefix, $activeBlogPrefixes, true)) { |
| 69 |
$this->logger->infoMessage( |
| 70 |
"Orphaned tables under prefix '{$oldPrefix}' belong to an active " |
| 71 |
. "multisite blog (sibling subsite). Skipping adoption to preserve " |
| 72 |
. "tenant isolation." |
| 73 |
); |
| 74 |
continue; |
| 75 |
} |
| 76 |
|
| 77 |
$this->logger->infoMessage( |
| 78 |
"Found orphaned plugin tables under prefix '{$oldPrefix}' " |
| 79 |
. "(current prefix is '{$currentPrefix}'): " . implode(', ', $tables) |
| 80 |
); |
| 81 |
|
| 82 |
// Check if old tables have any data at all. |
| 83 |
$totalRows = $this->countOldPrefixRows($oldPrefix, $tables); |
| 84 |
if ($totalRows === 0) { |
| 85 |
$this->logger->infoMessage( |
| 86 |
"Orphaned tables under prefix '{$oldPrefix}' are all empty. Skipping adoption." |
| 87 |
); |
| 88 |
continue; |
| 89 |
} |
| 90 |
|
| 91 |
// Verify ownership via logs dest_url slug matching. |
| 92 |
$matchResult = $this->verifyOwnershipViaLogs($oldPrefix); |
| 93 |
|
| 94 |
if ($matchResult === null) { |
| 95 |
// Logs verification returned no data — fall back to redirects post-ID check. |
| 96 |
$matchResult = $this->verifyOwnershipViaRedirects($oldPrefix); |
| 97 |
} |
| 98 |
|
| 99 |
if ($matchResult !== true) { |
| 100 |
// false = data doesn't match this site; null = insufficient data to verify. |
| 101 |
// Either way, do not adopt — absence of veto is not permission. |
| 102 |
$reason = ($matchResult === false) |
| 103 |
? "Data does not appear to belong to this site." |
| 104 |
: "Insufficient data in logs and redirects to verify ownership."; |
| 105 |
$this->logger->infoMessage( |
| 106 |
"Orphaned tables under prefix '{$oldPrefix}' — skipping adoption. {$reason}" |
| 107 |
); |
| 108 |
continue; |
| 109 |
} |
| 110 |
|
| 111 |
// Ownership positively verified. Adopt the data. |
| 112 |
$this->adoptDataFromPrefix($oldPrefix, $currentPrefix, $tables); |
| 113 |
} |
| 114 |
} |
| 115 |
|
| 116 |
/** |
| 117 |
* Parse information_schema rows into a prefix => [lowercase table name, ...] |
| 118 |
* map. Each table name is lowercased and grouped by the substring before its |
| 119 |
* 'abj404_' segment; rows without a recognizable table_name or abj404_ |
| 120 |
* segment are skipped. Extracted from adoptOrphanedTables() so that method's |
| 121 |
* branching stays within the cyclomatic-complexity budget. |
| 122 |
* |
| 123 |
* @param array<array-key, mixed> $rows |
| 124 |
* @return array<string, array<int, string>> |
| 125 |
*/ |
| 126 |
private function groupAbj404TablesByPrefix(array $rows): array { |
| 127 |
$tablesByPrefix = []; |
| 128 |
foreach ($rows as $row) { |
| 129 |
if (!is_array($row)) { |
| 130 |
continue; |
| 131 |
} |
| 132 |
$tableName = null; |
| 133 |
foreach ($row as $key => $value) { |
| 134 |
if (strtolower((string)$key) === 'table_name' && is_scalar($value)) { |
| 135 |
$tableName = strtolower((string)$value); |
| 136 |
break; |
| 137 |
} |
| 138 |
} |
| 139 |
if ($tableName === null) { |
| 140 |
continue; |
| 141 |
} |
| 142 |
|
| 143 |
$abj404Pos = strpos($tableName, 'abj404_'); |
| 144 |
if ($abj404Pos === false) { |
| 145 |
continue; |
| 146 |
} |
| 147 |
|
| 148 |
$prefix = substr($tableName, 0, $abj404Pos); |
| 149 |
if (!is_string($prefix)) { |
| 150 |
continue; |
| 151 |
} |
| 152 |
$tablesByPrefix[$prefix][] = $tableName; |
| 153 |
} |
| 154 |
return $tablesByPrefix; |
| 155 |
} |
| 156 |
|
| 157 |
/** |
| 158 |
* Count total rows across all known plugin tables for a given prefix. |
| 159 |
* |
| 160 |
* @param string $oldPrefix |
| 161 |
* @param array<int, string> $knownTables Table names actually found in information_schema. |
| 162 |
* @return int |
| 163 |
*/ |
| 164 |
private function countOldPrefixRows(string $oldPrefix, array $knownTables): int { |
| 165 |
$total = 0; |
| 166 |
foreach ($this->getPluginTableSuffixes() as $suffix) { |
| 167 |
$tableName = $oldPrefix . $suffix; |
| 168 |
if (!in_array($tableName, $knownTables, true)) { |
| 169 |
continue; |
| 170 |
} |
| 171 |
$result = $this->dbCore->queryAndGetResults( |
| 172 |
"SELECT COUNT(*) AS cnt FROM `{$tableName}`", |
| 173 |
['ignore_errors' => ["doesn't exist", "not found"]] |
| 174 |
); |
| 175 |
if (is_array($result['rows']) && !empty($result['rows'])) { |
| 176 |
$row = $result['rows'][0]; |
| 177 |
$countValue = is_array($row) ? ($row['cnt'] ?? $row['CNT'] ?? 0) : 0; |
| 178 |
$cnt = is_numeric($countValue) ? (int)$countValue : 0; |
| 179 |
$total += $cnt; |
| 180 |
} |
| 181 |
} |
| 182 |
return $total; |
| 183 |
} |
| 184 |
|
| 185 |
/** |
| 186 |
* Verify ownership of orphaned tables by matching logs dest_url against |
| 187 |
* current site's published post slugs. |
| 188 |
* |
| 189 |
* @param string $oldPrefix The old table prefix. |
| 190 |
* @return bool|null true = verified, false = failed, null = no data to verify. |
| 191 |
*/ |
| 192 |
private function verifyOwnershipViaLogs(string $oldPrefix): ?bool { |
| 193 |
global $wpdb; |
| 194 |
$logsTable = $oldPrefix . 'abj404_logsv2'; |
| 195 |
// WordPress core posts table uses the original $wpdb->prefix (possibly mixed-case), |
| 196 |
// NOT our lowercased prefix. Only plugin tables were renamed to lowercase. |
| 197 |
$postsTable = ($wpdb->prefix ?? 'wp_') . 'posts'; |
| 198 |
|
| 199 |
// p.post_name belongs to a WordPress CORE table and dest_url to a PLUGIN |
| 200 |
// table, so each carries whatever collation its own table was created |
| 201 |
// with. When those differ MySQL refuses the comparison outright |
| 202 |
// ("Illegal mix of collations ... for operation 'locate'", errno 1267), |
| 203 |
// the query returns nothing, ownership is never verified, and the |
| 204 |
// site's orphaned tables stay stranded under the old prefix. Converting |
| 205 |
// BOTH operands to one charset and one collation is the same treatment |
| 206 |
// updatePermalinkCache.sql and getPublishedPagesAndPostsIDs.sql already |
| 207 |
// give their wp_posts / wp_terms comparisons; this query is built |
| 208 |
// inline, which is how it missed out. The collation comes from the one |
| 209 |
// derivation that guarantees it is valid for the charset beside it. |
| 210 |
$collation = ABJ_404_Solution_DatabaseCollationHelper::utf8mb4CollationOrFallback( |
| 211 |
isset($wpdb->collate) && is_scalar($wpdb->collate) ? (string)$wpdb->collate : '' |
| 212 |
); |
| 213 |
$postNameOperand = "CONVERT(p.post_name USING utf8mb4) COLLATE " . $collation; |
| 214 |
$destUrlOperand = "CONVERT(dest_url USING utf8mb4) COLLATE " . $collation; |
| 215 |
|
| 216 |
// Check distinct internal dest_urls against published post slugs. |
| 217 |
$query = "SELECT COUNT(*) AS total, |
| 218 |
SUM(CASE WHEN matched = 1 THEN 1 ELSE 0 END) AS matches |
| 219 |
FROM ( |
| 220 |
SELECT DISTINCT dest_url, |
| 221 |
EXISTS(SELECT 1 FROM `{$postsTable}` p |
| 222 |
WHERE p.post_status = 'publish' |
| 223 |
AND LENGTH(p.post_name) >= 3 |
| 224 |
AND LOCATE({$postNameOperand}, {$destUrlOperand}) > 0) AS matched |
| 225 |
FROM `{$logsTable}` l |
| 226 |
WHERE dest_url IS NOT NULL |
| 227 |
AND dest_url != '' |
| 228 |
AND dest_url != '404' |
| 229 |
AND dest_url NOT LIKE 'http://%' |
| 230 |
AND dest_url NOT LIKE 'https://%' |
| 231 |
LIMIT 500 |
| 232 |
) sub"; |
| 233 |
|
| 234 |
$result = $this->dbCore->queryAndGetResults($query, |
| 235 |
['ignore_errors' => ["doesn't exist", "not found"]]); |
| 236 |
|
| 237 |
if (!is_array($result['rows']) || empty($result['rows'])) { |
| 238 |
return null; |
| 239 |
} |
| 240 |
|
| 241 |
$row = is_array($result['rows'][0] ?? null) ? $result['rows'][0] : []; |
| 242 |
$total = 0; |
| 243 |
$matches = 0; |
| 244 |
foreach ($row as $key => $value) { |
| 245 |
$lk = strtolower((string)$key); |
| 246 |
if ($lk === 'total' && is_numeric($value)) { $total = (int)$value; } |
| 247 |
if ($lk === 'matches' && is_numeric($value)) { $matches = (int)$value; } |
| 248 |
} |
| 249 |
|
| 250 |
if ($total === 0) { |
| 251 |
return null; // No internal dest_urls to verify. |
| 252 |
} |
| 253 |
|
| 254 |
$matchPct = ($matches / max(1, $total)) * 100; |
| 255 |
$this->logger->infoMessage( |
| 256 |
"Logs ownership verification for prefix '{$oldPrefix}': " |
| 257 |
. "{$matches}/{$total} distinct internal dest_urls match published post slugs " |
| 258 |
. "({$matchPct}%)" |
| 259 |
); |
| 260 |
|
| 261 |
return $matchPct >= 80; |
| 262 |
} |
| 263 |
|
| 264 |
/** |
| 265 |
* Fallback ownership verification using redirects table post-ID existence. |
| 266 |
* Weaker than slug matching but useful when logs have no internal dest_urls. |
| 267 |
* |
| 268 |
* @param string $oldPrefix |
| 269 |
* @return bool|null true = verified, false = failed, null = no data. |
| 270 |
*/ |
| 271 |
private function verifyOwnershipViaRedirects(string $oldPrefix): ?bool { |
| 272 |
global $wpdb; |
| 273 |
$redirectsTable = $oldPrefix . 'abj404_redirects'; |
| 274 |
$postsTable = ($wpdb->prefix ?? 'wp_') . 'posts'; |
| 275 |
|
| 276 |
$query = "SELECT COUNT(*) AS total, |
| 277 |
SUM(CASE WHEN p.ID IS NOT NULL THEN 1 ELSE 0 END) AS matches |
| 278 |
FROM `{$redirectsTable}` r |
| 279 |
LEFT JOIN `{$postsTable}` p |
| 280 |
ON p.ID = CAST(r.final_dest AS UNSIGNED) |
| 281 |
AND p.post_status IN ('publish', 'draft', 'private') |
| 282 |
WHERE r.type IN (1, 2, 3)"; |
| 283 |
|
| 284 |
$result = $this->dbCore->queryAndGetResults($query, |
| 285 |
['ignore_errors' => ["doesn't exist", "not found"]]); |
| 286 |
|
| 287 |
if (!is_array($result['rows']) || empty($result['rows'])) { |
| 288 |
return null; |
| 289 |
} |
| 290 |
|
| 291 |
$row = is_array($result['rows'][0] ?? null) ? $result['rows'][0] : []; |
| 292 |
$total = 0; |
| 293 |
$matches = 0; |
| 294 |
foreach ($row as $key => $value) { |
| 295 |
$lk = strtolower((string)$key); |
| 296 |
if ($lk === 'total' && is_numeric($value)) { $total = (int)$value; } |
| 297 |
if ($lk === 'matches' && is_numeric($value)) { $matches = (int)$value; } |
| 298 |
} |
| 299 |
|
| 300 |
if ($total === 0) { |
| 301 |
return null; |
| 302 |
} |
| 303 |
|
| 304 |
$matchPct = ($matches / max(1, $total)) * 100; |
| 305 |
$this->logger->infoMessage( |
| 306 |
"Redirects fallback ownership verification for prefix '{$oldPrefix}': " |
| 307 |
. "{$matches}/{$total} type 1/2/3 redirects point to existing posts ({$matchPct}%)" |
| 308 |
); |
| 309 |
|
| 310 |
return $matchPct >= 80; |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* Adopt data from orphaned tables under an old prefix into current-prefix tables. |
| 315 |
* Uses INSERT IGNORE to avoid duplicate key conflicts. |
| 316 |
* |
| 317 |
* @param string $oldPrefix |
| 318 |
* @param string $currentPrefix |
| 319 |
* @param array<string> $knownTables |
| 320 |
* @return void |
| 321 |
*/ |
| 322 |
private function adoptDataFromPrefix(string $oldPrefix, string $currentPrefix, array $knownTables): void { |
| 323 |
$this->logger->infoMessage( |
| 324 |
"Beginning adoption of data from prefix '{$oldPrefix}' to '{$currentPrefix}'" |
| 325 |
); |
| 326 |
|
| 327 |
$totalAdopted = 0; |
| 328 |
|
| 329 |
foreach ($this->getPluginTableSuffixes() as $suffix) { |
| 330 |
$oldTable = $oldPrefix . $suffix; |
| 331 |
if (!in_array($oldTable, $knownTables, true)) { |
| 332 |
continue; |
| 333 |
} |
| 334 |
$newTable = $currentPrefix . $suffix; |
| 335 |
|
| 336 |
// Check if old table exists and has rows. |
| 337 |
$countResult = $this->dbCore->queryAndGetResults( |
| 338 |
"SELECT COUNT(*) AS cnt FROM `{$oldTable}`", |
| 339 |
['ignore_errors' => ["doesn't exist", "not found"]] |
| 340 |
); |
| 341 |
if (!is_array($countResult['rows']) || empty($countResult['rows'])) { |
| 342 |
continue; |
| 343 |
} |
| 344 |
$row = $countResult['rows'][0]; |
| 345 |
$countValue = is_array($row) ? ($row['cnt'] ?? $row['CNT'] ?? 0) : 0; |
| 346 |
$oldCount = is_numeric($countValue) ? (int)$countValue : 0; |
| 347 |
if ($oldCount === 0) { |
| 348 |
continue; |
| 349 |
} |
| 350 |
|
| 351 |
// Check if new table exists (it should — auto-repair creates them). |
| 352 |
$newExists = $this->dbCore->queryAndGetResults( |
| 353 |
"SELECT 1 FROM `{$newTable}` LIMIT 1", |
| 354 |
['ignore_errors' => ["doesn't exist", "not found"]] |
| 355 |
); |
| 356 |
if (!empty($newExists['last_error'])) { |
| 357 |
$this->logger->infoMessage( |
| 358 |
"Target table '{$newTable}' does not exist yet. Skipping adoption for '{$suffix}'." |
| 359 |
); |
| 360 |
continue; |
| 361 |
} |
| 362 |
|
| 363 |
// Build a column-matched INSERT to handle schema drift between old and new tables. |
| 364 |
// Old tables from older plugin versions may have fewer or different columns. |
| 365 |
$commonColumns = $this->getCommonColumns($oldTable, $newTable); |
| 366 |
if (empty($commonColumns)) { |
| 367 |
$this->logger->infoMessage( |
| 368 |
"No common columns found between '{$oldTable}' and '{$newTable}'. Skipping." |
| 369 |
); |
| 370 |
continue; |
| 371 |
} |
| 372 |
|
| 373 |
$columnList = implode('`, `', $commonColumns); |
| 374 |
$insertQuery = "INSERT IGNORE INTO `{$newTable}` (`{$columnList}`) " |
| 375 |
. "SELECT `{$columnList}` FROM `{$oldTable}`"; |
| 376 |
$insertResult = $this->dbCore->queryAndGetResults($insertQuery, |
| 377 |
['ignore_errors' => ["doesn't exist", "not found", "Duplicate"]]); |
| 378 |
|
| 379 |
$affectedRows = 0; |
| 380 |
if (is_array($insertResult) && isset($insertResult['rows_affected'])) { |
| 381 |
$rawAffected = $insertResult['rows_affected']; |
| 382 |
$affectedRows = is_numeric($rawAffected) ? (int)$rawAffected : 0; |
| 383 |
} |
| 384 |
|
| 385 |
if ($affectedRows > 0) { |
| 386 |
$totalAdopted += $affectedRows; |
| 387 |
$this->logger->infoMessage( |
| 388 |
"Adopted {$affectedRows} rows from '{$oldTable}' into '{$newTable}'" |
| 389 |
); |
| 390 |
} |
| 391 |
} |
| 392 |
|
| 393 |
$this->logger->infoMessage( |
| 394 |
"Adoption complete: {$totalAdopted} total rows adopted from prefix '{$oldPrefix}' to '{$currentPrefix}'" |
| 395 |
); |
| 396 |
|
| 397 |
// Record this prefix as adopted so we don't re-detect it on every page load. |
| 398 |
$adoptedPrefixes = get_option('abj404_adopted_prefixes', array()); |
| 399 |
if (!is_array($adoptedPrefixes)) { |
| 400 |
$adoptedPrefixes = array(); |
| 401 |
} |
| 402 |
if (!in_array($oldPrefix, $adoptedPrefixes, true)) { |
| 403 |
$adoptedPrefixes[] = $oldPrefix; |
| 404 |
update_option('abj404_adopted_prefixes', $adoptedPrefixes, false); |
| 405 |
} |
| 406 |
} |
| 407 |
|
| 408 |
/** |
| 409 |
* Get the list of column names that exist in both tables. |
| 410 |
* Used by adoptDataFromPrefix() to build column-matched INSERTs |
| 411 |
* that survive schema drift between plugin versions. |
| 412 |
* |
| 413 |
* @param string $tableA |
| 414 |
* @param string $tableB |
| 415 |
* @return array<int, string> Column names present in both tables (lowercase). |
| 416 |
*/ |
| 417 |
private function getCommonColumns(string $tableA, string $tableB): array { |
| 418 |
$colsA = $this->getTableColumns($tableA); |
| 419 |
$colsB = $this->getTableColumns($tableB); |
| 420 |
|
| 421 |
if (empty($colsA) || empty($colsB)) { |
| 422 |
return []; |
| 423 |
} |
| 424 |
|
| 425 |
return array_values(array_intersect($colsA, $colsB)); |
| 426 |
} |
| 427 |
|
| 428 |
/** |
| 429 |
* Get column names for a table via SHOW COLUMNS. |
| 430 |
* |
| 431 |
* @param string $tableName |
| 432 |
* @return array<int, string> Column names (lowercase). |
| 433 |
*/ |
| 434 |
private function getTableColumns(string $tableName): array { |
| 435 |
$result = $this->dbCore->queryAndGetResults( |
| 436 |
"SHOW COLUMNS FROM `{$tableName}`", |
| 437 |
['ignore_errors' => ["doesn't exist", "not found"]] |
| 438 |
); |
| 439 |
|
| 440 |
if (!is_array($result['rows']) || empty($result['rows'])) { |
| 441 |
return []; |
| 442 |
} |
| 443 |
|
| 444 |
$columns = []; |
| 445 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : []; |
| 446 |
foreach ($rows as $row) { |
| 447 |
if (!is_array($row)) { |
| 448 |
continue; |
| 449 |
} |
| 450 |
// SHOW COLUMNS returns 'Field' key — case-insensitive lookup. |
| 451 |
$colName = null; |
| 452 |
foreach ($row as $key => $value) { |
| 453 |
if (strtolower((string)$key) === 'field' && is_scalar($value)) { |
| 454 |
$colName = strtolower((string)$value); |
| 455 |
break; |
| 456 |
} |
| 457 |
} |
| 458 |
if ($colName !== null) { |
| 459 |
$columns[] = $colName; |
| 460 |
} |
| 461 |
} |
| 462 |
|
| 463 |
return $columns; |
| 464 |
} |
| 465 |
|
| 466 |
} |
| 467 |
|
