| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Resolves and reconciles MySQL collation for plugin tables and columns. |
| 9 |
* |
| 10 |
* Extracted from DatabaseCore as part of the (3/6) DatabaseCore decomposition. |
| 11 |
* Owns four cohesive responsibilities that all turn on knowing the effective |
| 12 |
* collation of plugin tables/columns so cross-collation comparisons do not |
| 13 |
* blow up plugin queries: |
| 14 |
* |
| 15 |
* 1. Sanitizing raw collation identifiers (strip non-word characters so they |
| 16 |
* are safe to interpolate in SQL). |
| 17 |
* 2. Discovering the effective table-level and column-level collation via |
| 18 |
* SHOW CREATE TABLE and information_schema, with safe fallbacks. |
| 19 |
* 3. Resolving the preferred utf8mb4 collation from the wpdb connection. |
| 20 |
* 4. Responding to a query-time collation mismatch by scheduling a |
| 21 |
* schema-wide correction outside the foreground request. |
| 22 |
* |
| 23 |
* This class holds no DatabaseCore back-reference. It receives: |
| 24 |
* - a query-runner callable bound over DatabaseCore::queryAndGetResults |
| 25 |
* (signature: function(string, array<string,mixed>): array<string,mixed>); |
| 26 |
* - a DDL reader callable bound over the table-name resolver |
| 27 |
* (signature: function(string): string); |
| 28 |
* - getter/setter callables for runtime flags |
| 29 |
* (signatures: function(string): mixed, function(string, mixed, int): void); |
| 30 |
* - the plugin logger and an optional clock. |
| 31 |
* |
| 32 |
* The recursion guard is a static property on this class (it must survive across |
| 33 |
* helper invocations within a single request) and is functionally identical to |
| 34 |
* the former DatabaseCore::$collationRecoveryInProgress. |
| 35 |
*/ |
| 36 |
class ABJ_404_Solution_DatabaseCollationHelper { |
| 37 |
|
| 38 |
/** @var int Cooldown after a collation-recovery attempt (seconds). */ |
| 39 |
const COLLATION_RECOVERY_COOLDOWN_SECONDS = 3600; |
| 40 |
|
| 41 |
/** @var bool Prevent recursive collation-repair scheduling within one request. */ |
| 42 |
private static $collationSchedulingInProgress = false; |
| 43 |
|
| 44 |
/** @var callable(string, array<string,mixed>): array<string,mixed> */ |
| 45 |
private $queryRunner; |
| 46 |
|
| 47 |
/** @var callable(string): string */ |
| 48 |
private $ddlReader; |
| 49 |
|
| 50 |
/** @var callable(string): mixed */ |
| 51 |
private $runtimeFlagGetter; |
| 52 |
|
| 53 |
/** @var callable(string, mixed, int): void */ |
| 54 |
private $runtimeFlagSetter; |
| 55 |
|
| 56 |
/** @var ABJ_404_Solution_Logging */ |
| 57 |
private $logger; |
| 58 |
|
| 59 |
/** @var ABJ_404_Solution_Clock|null */ |
| 60 |
private $clock; |
| 61 |
|
| 62 |
/** |
| 63 |
* @param callable(string, array<string,mixed>): array<string,mixed> $queryRunner |
| 64 |
* Runs a SQL query through the centralized error-handling pipeline and |
| 65 |
* returns its result array. Bound by DatabaseCore over queryAndGetResults(). |
| 66 |
* @param callable(string): string $ddlReader |
| 67 |
* Returns the SHOW CREATE TABLE output for the given table name (or ''). |
| 68 |
* @param callable(string): mixed $runtimeFlagGetter |
| 69 |
* Returns the current value of a runtime flag (transient with option fallback). |
| 70 |
* @param callable(string, mixed, int): void $runtimeFlagSetter |
| 71 |
* Persists a runtime-flag value with a TTL. |
| 72 |
* @param ABJ_404_Solution_Logging $logger |
| 73 |
* @param ABJ_404_Solution_Clock|null $clock Optional; lazily resolved when null. |
| 74 |
*/ |
| 75 |
public function __construct( |
| 76 |
callable $queryRunner, |
| 77 |
callable $ddlReader, |
| 78 |
callable $runtimeFlagGetter, |
| 79 |
callable $runtimeFlagSetter, |
| 80 |
$logger, |
| 81 |
$clock = null |
| 82 |
) { |
| 83 |
$this->queryRunner = $queryRunner; |
| 84 |
$this->ddlReader = $ddlReader; |
| 85 |
$this->runtimeFlagGetter = $runtimeFlagGetter; |
| 86 |
$this->runtimeFlagSetter = $runtimeFlagSetter; |
| 87 |
$this->logger = $logger; |
| 88 |
$this->clock = $clock; |
| 89 |
} |
| 90 |
|
| 91 |
/** |
| 92 |
* Reset the static recursion guard. Intended for test setUp/tearDown only. |
| 93 |
* |
| 94 |
* @return void |
| 95 |
*/ |
| 96 |
public static function resetRecursionGuardForTests(): void { |
| 97 |
self::$collationSchedulingInProgress = false; |
| 98 |
} |
| 99 |
|
| 100 |
/** |
| 101 |
* Sanitize a raw collation identifier so it is safe to interpolate in SQL. |
| 102 |
* |
| 103 |
* Strips every character that is not [A-Za-z0-9_]. |
| 104 |
* |
| 105 |
* @param string $collation |
| 106 |
* @return string |
| 107 |
*/ |
| 108 |
public function sanitizeCollationIdentifier($collation): string { |
| 109 |
if (!is_string($collation) || $collation === '') { |
| 110 |
return ''; |
| 111 |
} |
| 112 |
$sanitized = preg_replace('/[^A-Za-z0-9_]/', '', $collation); |
| 113 |
return $sanitized !== null ? $sanitized : ''; |
| 114 |
} |
| 115 |
|
| 116 |
/** |
| 117 |
* Get the table-level default collation for a given table. |
| 118 |
* |
| 119 |
* Queries SHOW CREATE TABLE for the COLLATE clause; falls back to |
| 120 |
* information_schema.TABLES.TABLE_COLLATION; then to utf8mb4_unicode_ci. |
| 121 |
* Result is validated through sanitizeCollationIdentifier(). |
| 122 |
* |
| 123 |
* @param string $tableName Fully-qualified table name (including prefix). |
| 124 |
* @return string |
| 125 |
*/ |
| 126 |
public function getTableCollationString(string $tableName): string { |
| 127 |
$fallback = 'utf8mb4_unicode_ci'; |
| 128 |
$ddl = ($this->ddlReader)($tableName); |
| 129 |
if (preg_match('/COLLATE[= ]([A-Za-z0-9_]+)/i', $ddl, $m)) { |
| 130 |
$sanitized = $this->sanitizeCollationIdentifier($m[1]); |
| 131 |
return $sanitized !== '' ? $sanitized : $fallback; |
| 132 |
} |
| 133 |
global $wpdb; |
| 134 |
if (isset($wpdb) && method_exists($wpdb, 'prepare')) { |
| 135 |
/** @var wpdb $wpdb */ |
| 136 |
$sql = $wpdb->prepare( |
| 137 |
"SELECT TABLE_COLLATION FROM information_schema.TABLES " |
| 138 |
. "WHERE TABLE_SCHEMA = DATABASE() " |
| 139 |
. "AND TABLE_NAME = %s " |
| 140 |
. "LIMIT 1", |
| 141 |
$tableName |
| 142 |
); |
| 143 |
if (is_string($sql) && $sql !== '') { |
| 144 |
$result = ($this->queryRunner)($sql, array('log_errors' => false)); |
| 145 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 146 |
if (!empty($rows) && is_array($rows[0])) { |
| 147 |
$row = array_change_key_case($rows[0]); |
| 148 |
$collation = $row['table_collation'] ?? ''; |
| 149 |
if (is_string($collation) && $collation !== '') { |
| 150 |
$sanitized = $this->sanitizeCollationIdentifier($collation); |
| 151 |
return $sanitized !== '' ? $sanitized : $fallback; |
| 152 |
} |
| 153 |
} |
| 154 |
} |
| 155 |
} |
| 156 |
return $fallback; |
| 157 |
} |
| 158 |
|
| 159 |
/** |
| 160 |
* Get the column-level collation for a specific column in a table. |
| 161 |
* |
| 162 |
* Queries information_schema.COLUMNS for the COLLATION_NAME. Falls back |
| 163 |
* to getTableCollationString() if the column query fails, then ultimately |
| 164 |
* to utf8mb4_unicode_ci. Result is validated through |
| 165 |
* sanitizeCollationIdentifier(). |
| 166 |
* |
| 167 |
* @param string $tableName Fully-qualified table name (including prefix). |
| 168 |
* @param string $columnName Column name to look up. |
| 169 |
* @return string |
| 170 |
*/ |
| 171 |
public function getColumnCollationString(string $tableName, string $columnName): string { |
| 172 |
$fallback = 'utf8mb4_unicode_ci'; |
| 173 |
global $wpdb; |
| 174 |
if (!isset($wpdb) || !method_exists($wpdb, 'prepare')) { |
| 175 |
return $this->getTableCollationString($tableName); |
| 176 |
} |
| 177 |
/** @var wpdb $wpdb */ |
| 178 |
$sql = $wpdb->prepare( |
| 179 |
"SELECT COLLATION_NAME FROM information_schema.COLUMNS " |
| 180 |
. "WHERE TABLE_SCHEMA = DATABASE() " |
| 181 |
. "AND TABLE_NAME = %s " |
| 182 |
. "AND COLUMN_NAME = %s " |
| 183 |
. "LIMIT 1", |
| 184 |
$tableName, |
| 185 |
$columnName |
| 186 |
); |
| 187 |
if (!is_string($sql) || $sql === '') { |
| 188 |
return $this->getTableCollationString($tableName); |
| 189 |
} |
| 190 |
$result = ($this->queryRunner)($sql, array('log_errors' => false)); |
| 191 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 192 |
if (empty($rows) || !is_array($rows[0])) { |
| 193 |
return $this->getTableCollationString($tableName); |
| 194 |
} |
| 195 |
$row = array_change_key_case($rows[0]); |
| 196 |
$collation = $row['collation_name'] ?? ''; |
| 197 |
if (!is_string($collation) || $collation === '') { |
| 198 |
return $this->getTableCollationString($tableName); |
| 199 |
} |
| 200 |
$sanitized = $this->sanitizeCollationIdentifier($collation); |
| 201 |
return $sanitized !== '' ? $sanitized : $fallback; |
| 202 |
} |
| 203 |
|
| 204 |
/** |
| 205 |
* Coerce a SQL expression to the charset and collation of an indexed |
| 206 |
* comparison column without wrapping that indexed column. |
| 207 |
* |
| 208 |
* Mixed-collation sites are common during upgrades and partial restores. |
| 209 |
* Applying CONVERT/COLLATE only to the non-indexed operand makes the |
| 210 |
* equality deterministic while leaving the target column sargable. |
| 211 |
* Callers must supply an internally constructed SQL expression; this |
| 212 |
* method sanitizes metadata identifiers, not arbitrary SQL text. |
| 213 |
* |
| 214 |
* @param string $expression SQL expression used opposite the target column. |
| 215 |
* @param array{table: string, column: string} $targetColumn Indexed target column metadata. |
| 216 |
* @return string |
| 217 |
*/ |
| 218 |
public function coerceExpressionToColumnCollation(string $expression, array $targetColumn): string { |
| 219 |
$tableName = $targetColumn['table'] ?? ''; |
| 220 |
$columnName = $targetColumn['column'] ?? ''; |
| 221 |
if ($tableName === '' || $columnName === '') { |
| 222 |
throw new InvalidArgumentException('Target table and column are required for a collation-safe SQL comparison.'); |
| 223 |
} |
| 224 |
|
| 225 |
$collation = $this->getColumnCollationString($tableName, $columnName); |
| 226 |
$charsetParts = explode('_', $collation, 2); |
| 227 |
$charset = $this->sanitizeCollationIdentifier($charsetParts[0] ?? ''); |
| 228 |
if ($charset === '' || $collation === '') { |
| 229 |
throw new InvalidArgumentException('Target column collation could not be converted to a safe SQL identifier.'); |
| 230 |
} |
| 231 |
|
| 232 |
return 'CONVERT(' . $expression . ' USING ' . $charset . ') COLLATE ' . $collation; |
| 233 |
} |
| 234 |
|
| 235 |
/** |
| 236 |
* Return the preferred utf8mb4 collation for this wpdb connection. |
| 237 |
* |
| 238 |
* If wpdb->collate already names a utf8mb4_* collation, use it; otherwise |
| 239 |
* fall back to utf8mb4_unicode_ci. |
| 240 |
* |
| 241 |
* @return string |
| 242 |
*/ |
| 243 |
public function getPreferredUtf8mb4Collation(): string { |
| 244 |
global $wpdb; |
| 245 |
if (isset($wpdb) && isset($wpdb->collate) && !empty($wpdb->collate)) { |
| 246 |
$wpdbCollation = $this->sanitizeCollationIdentifier((string)$wpdb->collate); |
| 247 |
if ($wpdbCollation !== '' && stripos($wpdbCollation, 'utf8mb4') !== false) { |
| 248 |
return $wpdbCollation; |
| 249 |
} |
| 250 |
} |
| 251 |
return 'utf8mb4_unicode_ci'; |
| 252 |
} |
| 253 |
|
| 254 |
/** |
| 255 |
* Schedule broad collation correction outside the foreground request. |
| 256 |
* |
| 257 |
* correctCollations() discovers every plugin table and may issue ALTER |
| 258 |
* TABLE ... CONVERT for each drifted table. Running that work inline can |
| 259 |
* exhaust an admin AJAX request on large sites, so the query that detected |
| 260 |
* the mismatch keeps its normal degraded error result while a dedicated, |
| 261 |
* deduplicated WP-Cron event performs the repair. Scheduler failures remain |
| 262 |
* retryable and are logged with the adapter's underlying failure detail. |
| 263 |
* |
| 264 |
* @return void |
| 265 |
*/ |
| 266 |
public function scheduleCollationRecovery(): void { |
| 267 |
if (self::$collationSchedulingInProgress) { |
| 268 |
return; |
| 269 |
} |
| 270 |
|
| 271 |
$cooldownKey = 'abj404_collation_recovery_cooldown'; |
| 272 |
$cooldownUntil = ($this->runtimeFlagGetter)($cooldownKey); |
| 273 |
$onCooldown = is_scalar($cooldownUntil) && (int)$cooldownUntil > $this->clock()->now(); |
| 274 |
|
| 275 |
if ($onCooldown) { |
| 276 |
return; |
| 277 |
} |
| 278 |
|
| 279 |
self::$collationSchedulingInProgress = true; |
| 280 |
try { |
| 281 |
$scheduler = abj_cron_scheduler(); |
| 282 |
$scheduled = $scheduler->scheduleSingleIfMissing( |
| 283 |
ABJ_404_Solution_CronScheduler::HOOK_REPAIR_COLLATIONS, |
| 284 |
1 |
| 285 |
); |
| 286 |
if ($scheduled) { |
| 287 |
($this->runtimeFlagSetter)( |
| 288 |
$cooldownKey, |
| 289 |
$this->clock()->now() + self::COLLATION_RECOVERY_COOLDOWN_SECONDS, |
| 290 |
self::COLLATION_RECOVERY_COOLDOWN_SECONDS |
| 291 |
); |
| 292 |
$this->logger->infoMessage( |
| 293 |
'Collation mismatch detected: schema correction scheduled for background repair.' |
| 294 |
); |
| 295 |
return; |
| 296 |
} |
| 297 |
$this->logger->warn( |
| 298 |
'Could not schedule background collation repair: ' . $scheduler->lastFailureDetail() |
| 299 |
); |
| 300 |
} catch (Throwable $e) { |
| 301 |
$this->logger->warn( |
| 302 |
'Could not schedule background collation repair: ' . $e->getMessage() |
| 303 |
); |
| 304 |
} finally { |
| 305 |
self::$collationSchedulingInProgress = false; |
| 306 |
} |
| 307 |
} |
| 308 |
|
| 309 |
/** |
| 310 |
* Lazily resolve the clock instance. |
| 311 |
* |
| 312 |
* @return ABJ_404_Solution_Clock |
| 313 |
*/ |
| 314 |
private function clock() { |
| 315 |
if ($this->clock === null) { |
| 316 |
if (class_exists('ABJ_404_Solution_ServiceContainer')) { |
| 317 |
$resolved = ABJ_404_Solution_ServiceContainer::safeGet('clock'); |
| 318 |
if ($resolved instanceof ABJ_404_Solution_Clock) { |
| 319 |
$this->clock = $resolved; |
| 320 |
return $this->clock; |
| 321 |
} |
| 322 |
} |
| 323 |
$this->clock = new ABJ_404_Solution_SystemClock(); |
| 324 |
} |
| 325 |
return $this->clock; |
| 326 |
} |
| 327 |
} |
| 328 |
|
