class-file-tree-producer.php
6 days ago
class-hmac-client.php
6 days ago
class-hmac-server.php
6 days ago
class-http-server.php
6 days ago
class-mysql-dump-producer.php
6 days ago
class-pdo-polyfill.php
6 days ago
class-sqlite-driver-pdo.php
6 days ago
class-staged-artifacts.php
6 days ago
class-staged-endpoints.php
6 days ago
class-staged-push-stream-protocol.php
6 days ago
class-wpdb-driver-pdo.php
6 days ago
export.php
6 days ago
utils.php
6 days ago
export.php
3823 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Unified export API for SQL and file operations. |
| 4 | */ |
| 5 | |
| 6 | use function WordPress\Reprint\Exporter\assert_valid_path; |
| 7 | use function WordPress\Reprint\Exporter\build_pdo_dsn; |
| 8 | use function WordPress\Reprint\Exporter\json_encode_or_throw; |
| 9 | use function WordPress\Reprint\Exporter\parse_size; |
| 10 | use function WordPress\Reprint\Exporter\path_is_within_root; |
| 11 | |
| 12 | // Capture any accidental output before headers are set so we can discard it |
| 13 | // when switching to streaming mode later. |
| 14 | if (!ob_get_level()) { |
| 15 | ob_start(); |
| 16 | } |
| 17 | |
| 18 | |
| 19 | /** |
| 20 | * The wire-protocol version this export plugin speaks. |
| 21 | * |
| 22 | * Both the export plugin (server) and the importer (client) are deployed |
| 23 | * independently. These two constants let them detect incompatibility at |
| 24 | * preflight time instead of producing silent corruption. |
| 25 | * |
| 26 | * EXPORT_PROTOCOL_VERSION is sent to the importer in the preflight JSON |
| 27 | * response as `protocol_version`. Bump it whenever a change to the wire |
| 28 | * protocol (cursor encoding, multipart structure, header names, endpoint |
| 29 | * parameters, response format) would break an older importer. |
| 30 | */ |
| 31 | define('EXPORT_PROTOCOL_VERSION', 1); |
| 32 | |
| 33 | /** |
| 34 | * The oldest *importer* protocol version this export plugin can talk to. |
| 35 | * |
| 36 | * Sent to the importer in the preflight response as `protocol_min_version`. |
| 37 | * The importer checks that its own IMPORT_PROTOCOL_VERSION is >= this value; |
| 38 | * if not, it tells the user to update the importer. |
| 39 | * |
| 40 | * Raise this when you drop backward-compatibility with old importers. |
| 41 | * Keep it equal to EXPORT_PROTOCOL_VERSION if no backward compat is needed. |
| 42 | */ |
| 43 | define('EXPORT_MIN_IMPORT_VERSION', 1); |
| 44 | |
| 45 | // File type mask + file type values (top bits of st_mode) |
| 46 | define('STAT_TYPE_MASK', 0170000); |
| 47 | define('STAT_TYPE_SOCKET', 0140000); |
| 48 | define('STAT_TYPE_LINK', 0120000); |
| 49 | define('STAT_TYPE_FILE', 0100000); |
| 50 | define('STAT_TYPE_BLOCK', 0060000); |
| 51 | define('STAT_TYPE_DIR', 0040000); |
| 52 | define('STAT_TYPE_CHAR', 0020000); |
| 53 | define('STAT_TYPE_FIFO', 0010000); |
| 54 | |
| 55 | /** |
| 56 | * Tracks time and memory limits for a single API request. |
| 57 | * |
| 58 | * Every export endpoint runs under resource constraints — a maximum |
| 59 | * execution time and a memory ceiling. Rather than threading four |
| 60 | * separate values through every function signature and every |
| 61 | * should_continue() call, this class bundles them into a single |
| 62 | * object with a simple has_remaining() check. |
| 63 | */ |
| 64 | class ResourceBudget |
| 65 | { |
| 66 | /** @var float */ |
| 67 | public $start_time; |
| 68 | /** @var int */ |
| 69 | public $max_time; |
| 70 | /** @var int */ |
| 71 | public $max_memory; |
| 72 | /** @var float */ |
| 73 | public $memory_threshold; |
| 74 | |
| 75 | public function __construct( |
| 76 | float $start_time, |
| 77 | int $max_time, |
| 78 | int $max_memory, |
| 79 | float $memory_threshold |
| 80 | ) { |
| 81 | $this->start_time = $start_time; |
| 82 | $this->max_time = $max_time; |
| 83 | $this->max_memory = $max_memory; |
| 84 | $this->memory_threshold = $memory_threshold; |
| 85 | } |
| 86 | |
| 87 | /** Returns false when the request should yield due to time or memory pressure. */ |
| 88 | public function has_remaining(): bool |
| 89 | { |
| 90 | if (microtime(true) - $this->start_time >= $this->max_time) { |
| 91 | return false; |
| 92 | } |
| 93 | |
| 94 | $memory_used = memory_get_usage(true); |
| 95 | if ($memory_used >= $this->max_memory * $this->memory_threshold) { |
| 96 | return false; |
| 97 | } |
| 98 | |
| 99 | return true; |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | /** |
| 104 | * Global streaming context. When set, the error handlers emit error chunks |
| 105 | * into the active gzip multipart stream instead of sending plain JSON |
| 106 | * (which would corrupt the compressed response). |
| 107 | * |
| 108 | * Set by each streaming endpoint right after creating $gz and $boundary. |
| 109 | * Keys: 'gz' => GzipOutputStream, 'boundary' => string |
| 110 | */ |
| 111 | $streaming_context = null; |
| 112 | |
| 113 | /** |
| 114 | * Initializes a multipart/mixed streaming response, optionally with gzip compression. |
| 115 | * |
| 116 | * Every streaming endpoint needs the same setup: a unique boundary, the |
| 117 | * Content-Type header, an output stream, and the global $streaming_context so |
| 118 | * error handlers can emit structured error chunks mid-stream. |
| 119 | * |
| 120 | * @param bool $require_headers If true, throws when headers were already sent |
| 121 | * (use for endpoints that can't degrade gracefully). |
| 122 | * @param bool $gzip If true, emit Content-Encoding: gzip and compress the body. |
| 123 | * @return array{gz: GzipOutputStream, boundary: string} |
| 124 | */ |
| 125 | function begin_multipart_stream(bool $require_headers = false, bool $gzip = true): array |
| 126 | { |
| 127 | global $streaming_context; |
| 128 | |
| 129 | /** |
| 130 | * We're choosing a random boundary without checking for its presence in the content. |
| 131 | * This may seem to contradict RFC 2046, where it says: |
| 132 | * |
| 133 | * > As stated previously, each body part is preceded by a boundary |
| 134 | * > delimiter line that contains the boundary delimiter. The boundary |
| 135 | * > delimiter MUST NOT appear inside any of the encapsulated parts, on a |
| 136 | * > line by itself or as the prefix of any line. This implies that it is |
| 137 | * > crucial that the composing agent be able to choose and specify a |
| 138 | * > unique boundary parameter value that does not contain the boundary |
| 139 | * > parameter value of an enclosing multipart as a prefix. |
| 140 | * > |
| 141 | * > https://www.rfc-editor.org/rfc/rfc2046.html |
| 142 | * |
| 143 | * But in practice, we're okay. We use 128 bits of randomness. The chance of |
| 144 | * it appearing in the data is about 1 in 2^128 — effectively zero. Curl does |
| 145 | * the same here: |
| 146 | * |
| 147 | * https://github.com/curl/curl/blob/462244447e8ba3a53b1ba9f0ba7baa52d8777daa/lib/mime.c#L1179-L1236 |
| 148 | * |
| 149 | * Also, most chunks declare their Content-Length, so the client may skip the |
| 150 | * boundary matching entirely and just consume that many bytes. |
| 151 | */ |
| 152 | $boundary = "boundary-" . bin2hex(random_bytes(16)); |
| 153 | $can_send_headers = !headers_sent(); |
| 154 | |
| 155 | if ($require_headers && !$can_send_headers) { |
| 156 | throw new RuntimeException( |
| 157 | "Cannot begin multipart stream: headers already sent" |
| 158 | ); |
| 159 | } |
| 160 | |
| 161 | if ($can_send_headers) { |
| 162 | @header("Content-Type: multipart/mixed; boundary=\"$boundary\""); |
| 163 | } |
| 164 | |
| 165 | $gz = new GzipOutputStream($can_send_headers && $gzip); |
| 166 | $streaming_context = ['gz' => $gz, 'boundary' => $boundary]; |
| 167 | |
| 168 | return $streaming_context; |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * Resolves database credentials from PHP constants and environment variables. |
| 173 | * |
| 174 | * Never reads from $config / HTTP parameters — credentials must come from |
| 175 | * the server environment (PHP constants or environment variables). |
| 176 | * |
| 177 | * @return array{db_host: string, db_name: string, db_user: string, db_password: string, |
| 178 | * wp_config_path: ?string, table_prefix: ?string} |
| 179 | * @throws InvalidArgumentException When required credentials are missing. |
| 180 | */ |
| 181 | function resolve_db_credentials(): array |
| 182 | { |
| 183 | $db_host = defined("DB_HOST") ? DB_HOST : getenv("DB_HOST"); |
| 184 | $db_name = defined("DB_NAME") ? DB_NAME : getenv("DB_NAME"); |
| 185 | $db_user = defined("DB_USER") ? DB_USER : getenv("DB_USER"); |
| 186 | $db_password = defined("DB_PASSWORD") ? DB_PASSWORD : getenv("DB_PASSWORD"); |
| 187 | |
| 188 | global $wpdb; |
| 189 | |
| 190 | $wp_config_path = null; |
| 191 | $table_prefix = null; |
| 192 | if (isset($GLOBALS['table_prefix']) && is_string($GLOBALS['table_prefix']) && $GLOBALS['table_prefix'] !== '') { |
| 193 | $table_prefix = $GLOBALS['table_prefix']; |
| 194 | } elseif (isset($wpdb) && is_object($wpdb) && isset($wpdb->prefix) && is_string($wpdb->prefix) && $wpdb->prefix !== '') { |
| 195 | $table_prefix = $wpdb->prefix; |
| 196 | } |
| 197 | |
| 198 | // On SQLite sites, the driver is already loaded by WordPress via the |
| 199 | // db.php drop-in. We just need to confirm it's available and skip the |
| 200 | // MySQL credential requirements. |
| 201 | if (is_sqlite_site()) { |
| 202 | return [ |
| 203 | "db_engine" => "sqlite", |
| 204 | "db_host" => "", |
| 205 | "db_name" => $db_name ?: "wordpress", |
| 206 | "db_user" => "", |
| 207 | "db_password" => "", |
| 208 | "wp_config_path" => $wp_config_path, |
| 209 | "table_prefix" => $table_prefix, |
| 210 | ]; |
| 211 | } |
| 212 | |
| 213 | $missing = []; |
| 214 | if (!$db_host) { $missing[] = "db_host"; } |
| 215 | if (!$db_name) { $missing[] = "db_name"; } |
| 216 | if (!$db_user) { $missing[] = "db_user"; } |
| 217 | if ($db_password === false || $db_password === null) { |
| 218 | $missing[] = "db_password"; |
| 219 | } |
| 220 | if (!empty($missing)) { |
| 221 | throw new InvalidArgumentException( |
| 222 | "Database credentials not found. Please provide via environment variables, " . |
| 223 | "PHP constants, or ensure wp-config.php exists with valid credentials. " . |
| 224 | "Missing: " . implode(", ", $missing) |
| 225 | ); |
| 226 | } |
| 227 | |
| 228 | return [ |
| 229 | "db_engine" => "mysql", |
| 230 | "db_host" => $db_host, |
| 231 | "db_name" => $db_name, |
| 232 | "db_user" => $db_user, |
| 233 | "db_password" => $db_password, |
| 234 | "wp_config_path" => $wp_config_path, |
| 235 | "table_prefix" => $table_prefix, |
| 236 | ]; |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * Returns true when the current WordPress site uses the SQLite backend. |
| 241 | * |
| 242 | * Detection is based on the WP_SQLite_Driver class being loaded and |
| 243 | * $wpdb->dbh being an instance of it. This is set up automatically by |
| 244 | * the sqlite-database-integration plugin's db.php drop-in when WordPress |
| 245 | * boots. |
| 246 | */ |
| 247 | function is_sqlite_site(): bool |
| 248 | { |
| 249 | global $wpdb; |
| 250 | // @TODO: Actually check for the WP_SQLite_Driver class being used here. |
| 251 | return defined('SQLITE_DB_DROPIN_VERSION') && isset($GLOBALS['@pdo']); |
| 252 | } |
| 253 | |
| 254 | /** |
| 255 | * Creates a database connection appropriate for the detected backend. |
| 256 | * |
| 257 | * For MySQL sites, returns a standard PDO connection. |
| 258 | * For SQLite sites, wraps the WP_SQLite_Driver that WordPress already |
| 259 | * loaded (via $wpdb->dbh) in a PDO-compatible adapter. The driver's |
| 260 | * AST-based translator converts every MySQL query to SQLite on the fly, |
| 261 | * so MySQLDumpProducer sees MySQL-shaped results and produces valid |
| 262 | * MySQL SQL output. |
| 263 | * |
| 264 | * @param array $creds Credentials from resolve_db_credentials(). |
| 265 | * @param array $options PDO options (only used for MySQL connections). |
| 266 | * @return PDO A real PDO for MySQL, or a PDO-compatible adapter for SQLite. |
| 267 | */ |
| 268 | function create_db_connection(array $creds, array $options = []) |
| 269 | { |
| 270 | if (($creds["db_engine"] ?? "mysql") === "sqlite") { |
| 271 | return create_sqlite_pdo_adapter(); |
| 272 | } |
| 273 | |
| 274 | // Gate on pdo_mysql, not pdo: ext-pdo core without the mysql driver |
| 275 | // can't drive MySQL exports. |
| 276 | if (!extension_loaded('pdo_mysql')) { |
| 277 | return create_wpdb_pdo_adapter(); |
| 278 | } |
| 279 | |
| 280 | // MySQL path (also works for HyperDB — wp-config.php credentials |
| 281 | // point to the write master). |
| 282 | $default_options = [ |
| 283 | PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION, |
| 284 | ]; |
| 285 | $merged_options = $options + $default_options; |
| 286 | |
| 287 | return new PDO( |
| 288 | "mysql:host={$creds['db_host']};dbname={$creds['db_name']};charset=utf8mb4", |
| 289 | $creds["db_user"], |
| 290 | $creds["db_password"], |
| 291 | $merged_options |
| 292 | ); |
| 293 | } |
| 294 | |
| 295 | /** |
| 296 | * Wraps the already-loaded WP_SQLite_Driver in a PDO-compatible adapter. |
| 297 | * |
| 298 | * Validates that the sqlite-database-integration plugin version is in the |
| 299 | * supported range, then extracts the driver and raw PDO from $wpdb->dbh. |
| 300 | * |
| 301 | * @return object PDO-compatible adapter (SqliteDriverPDO). |
| 302 | * @throws RuntimeException If the driver is not available or unsupported. |
| 303 | */ |
| 304 | function create_sqlite_pdo_adapter() |
| 305 | { |
| 306 | global $wpdb; |
| 307 | |
| 308 | /** |
| 309 | * Minimum sqlite-database-integration version that exposes the API we |
| 310 | * depend on: WP_SQLite_Driver::query(), get_query_results(), |
| 311 | * get_connection()->get_pdo(). |
| 312 | */ |
| 313 | $min_version = '2.1.0'; |
| 314 | |
| 315 | require_once __DIR__ . "/class-sqlite-driver-pdo.php"; |
| 316 | |
| 317 | if (!isset($wpdb) || !($wpdb->dbh instanceof WP_SQLite_Driver)) { |
| 318 | throw new RuntimeException( |
| 319 | "SQLite export requires WordPress loaded with the " . |
| 320 | "sqlite-database-integration plugin active." |
| 321 | ); |
| 322 | } |
| 323 | |
| 324 | // Verify the plugin version is in the supported range. |
| 325 | if (defined('SQLITE_DRIVER_VERSION')) { |
| 326 | if (version_compare(SQLITE_DRIVER_VERSION, $min_version, '<')) { |
| 327 | throw new RuntimeException( |
| 328 | "sqlite-database-integration plugin version " . SQLITE_DRIVER_VERSION . |
| 329 | " is too old. Minimum required: " . $min_version |
| 330 | ); |
| 331 | } |
| 332 | } |
| 333 | |
| 334 | $driver = $wpdb->dbh; |
| 335 | $raw_pdo = $driver->get_connection()->get_pdo(); |
| 336 | |
| 337 | return new SqliteDriverPDO($driver, $raw_pdo); |
| 338 | } |
| 339 | |
| 340 | /** |
| 341 | * Wraps the global $wpdb in a PDO-shaped adapter. |
| 342 | * |
| 343 | * Used on hosts without ext-pdo_mysql. Requires WordPress to be loaded |
| 344 | * (so $wpdb is available); throws otherwise. |
| 345 | */ |
| 346 | function create_wpdb_pdo_adapter() |
| 347 | { |
| 348 | global $wpdb; |
| 349 | |
| 350 | require_once __DIR__ . "/class-wpdb-driver-pdo.php"; |
| 351 | |
| 352 | // Guard against a clobbered/half-initialized $wpdb: isset() alone passes |
| 353 | // for non-object scalars, which would fatal inside the adapter constructor. |
| 354 | if (!isset($wpdb) || !is_object($wpdb)) { |
| 355 | throw new RuntimeException( |
| 356 | "MySQL export without PDO requires WordPress \$wpdb to be initialized." |
| 357 | ); |
| 358 | } |
| 359 | |
| 360 | return new WpdbDriverPDO($wpdb); |
| 361 | } |
| 362 | |
| 363 | // Guard with existence checks: when loaded via Composer autoloader, both |
| 364 | // files are already included from a path that may differ from __DIR__ |
| 365 | // (e.g. symlink vs realpath). With opcache.revalidate_path=0 (default), |
| 366 | // require_once does not resolve symlinks, so the same physical file can |
| 367 | // be loaded twice through different paths, causing "Cannot redeclare" |
| 368 | // fatal errors. |
| 369 | if (!function_exists('WordPress\\Reprint\\Exporter\\build_pdo_dsn')) { |
| 370 | require_once __DIR__ . "/utils.php"; |
| 371 | } |
| 372 | if (!class_exists('Site_Export_HTTP_Server', false)) { |
| 373 | require_once __DIR__ . "/class-http-server.php"; |
| 374 | } |
| 375 | |
| 376 | /** |
| 377 | * Emits an error chunk into a gzip multipart stream. |
| 378 | */ |
| 379 | function emit_error_chunk($gz, string $boundary, string $message): void |
| 380 | { |
| 381 | $json = json_encode([ |
| 382 | "error_type" => "php_error", |
| 383 | "path" => "", |
| 384 | "message" => $message, |
| 385 | ]); |
| 386 | if ($json === false) { |
| 387 | $json = '{"error_type":"php_error","path":"","message":"Error (json_encode failed)"}'; |
| 388 | } |
| 389 | $chunk = |
| 390 | "--{$boundary}\r\n" . |
| 391 | "Content-Type: application/json\r\n" . |
| 392 | "Content-Length: " . strlen($json) . "\r\n" . |
| 393 | "X-Chunk-Type: error\r\n" . |
| 394 | "\r\n" . |
| 395 | $json . "\r\n"; |
| 396 | try { |
| 397 | $gz->write($chunk); |
| 398 | $gz->sync(); |
| 399 | } catch (\Throwable $e) { |
| 400 | // Gzip stream is broken — fall back to raw output. |
| 401 | // The response is already partially gzipped so the client likely |
| 402 | // can't parse this, but it's better than silent failure. |
| 403 | echo $chunk; |
| 404 | flush(); |
| 405 | } |
| 406 | } |
| 407 | |
| 408 | // Streaming-aware error handler. Before streaming starts, errors produce |
| 409 | // a JSON response with HTTP 500. Mid-stream, errors become multipart |
| 410 | // error chunks so the client receives structured diagnostics. |
| 411 | // |
| 412 | // Respects the @ operator: suppressed errors are logged but never emitted |
| 413 | // into the stream or sent as responses, since the calling code already |
| 414 | // handles the failure (e.g. @readlink checks for false). |
| 415 | set_error_handler(function ($errno, $errstr, $errfile, $errline) { |
| 416 | global $streaming_context; |
| 417 | |
| 418 | $error = [ |
| 419 | "error" => "PHP Error: $errstr", |
| 420 | "file" => $errfile, |
| 421 | "line" => $errline, |
| 422 | "type" => $errno, |
| 423 | ]; |
| 424 | |
| 425 | if (!(error_reporting() & $errno)) { |
| 426 | error_log("Export error (suppressed): " . json_encode($error)); |
| 427 | return true; |
| 428 | } |
| 429 | |
| 430 | error_log("Export error: " . json_encode($error)); |
| 431 | |
| 432 | if ($streaming_context !== null) { |
| 433 | emit_error_chunk( |
| 434 | $streaming_context['gz'], |
| 435 | $streaming_context['boundary'], |
| 436 | "PHP Error ({$errno}): {$errstr} in {$errfile}:{$errline}" |
| 437 | ); |
| 438 | return true; |
| 439 | } |
| 440 | |
| 441 | http_response_code(500); |
| 442 | @header("Content-Type: application/json"); |
| 443 | echo json_encode($error); |
| 444 | exit(1); |
| 445 | }); |
| 446 | |
| 447 | // Streaming-aware exception handler, mirrors the error handler above. |
| 448 | set_exception_handler(function ($e) { |
| 449 | global $streaming_context; |
| 450 | |
| 451 | $error = [ |
| 452 | "error" => get_class($e) . ": " . $e->getMessage(), |
| 453 | "file" => $e->getFile(), |
| 454 | "line" => $e->getLine(), |
| 455 | "trace" => $e->getTraceAsString(), |
| 456 | ]; |
| 457 | error_log("Export exception: " . json_encode($error)); |
| 458 | |
| 459 | if ($streaming_context !== null) { |
| 460 | emit_error_chunk( |
| 461 | $streaming_context['gz'], |
| 462 | $streaming_context['boundary'], |
| 463 | get_class($e) . ": " . $e->getMessage() |
| 464 | ); |
| 465 | return; |
| 466 | } |
| 467 | |
| 468 | http_response_code(500); |
| 469 | header("Content-Type: application/json"); |
| 470 | echo json_encode($error); |
| 471 | exit(1); |
| 472 | }); |
| 473 | |
| 474 | // Catches E_ERROR/E_PARSE fatals that set_error_handler cannot intercept. |
| 475 | register_shutdown_function(function () { |
| 476 | global $streaming_context; |
| 477 | |
| 478 | $error = error_get_last(); |
| 479 | if ($error === null) { |
| 480 | return; |
| 481 | } |
| 482 | $fatal_types = E_ERROR | E_PARSE | E_CORE_ERROR | E_COMPILE_ERROR; |
| 483 | if (!($error['type'] & $fatal_types)) { |
| 484 | return; |
| 485 | } |
| 486 | |
| 487 | $message = "Fatal: {$error['message']} in {$error['file']}:{$error['line']}"; |
| 488 | error_log("Export fatal: " . json_encode($error)); |
| 489 | |
| 490 | if ($streaming_context !== null) { |
| 491 | // Best-effort attempt to emit an error chunk into the stream. |
| 492 | // The stream may already be in a broken state, but this gives |
| 493 | // the client the best chance of receiving structured error info. |
| 494 | try { |
| 495 | emit_error_chunk( |
| 496 | $streaming_context['gz'], |
| 497 | $streaming_context['boundary'], |
| 498 | $message |
| 499 | ); |
| 500 | } catch (Throwable $ignored) { |
| 501 | // Stream is too broken to write to — nothing more we can do. |
| 502 | } |
| 503 | return; |
| 504 | } |
| 505 | |
| 506 | if (!headers_sent()) { |
| 507 | http_response_code(500); |
| 508 | @header("Content-Type: application/json"); |
| 509 | echo json_encode([ |
| 510 | "error" => $message, |
| 511 | "file" => $error['file'], |
| 512 | "line" => $error['line'], |
| 513 | "type" => $error['type'], |
| 514 | ]); |
| 515 | } |
| 516 | }); |
| 517 | |
| 518 | // ============================================================================ |
| 519 | // E2E Test Hook System (only active when SITE_EXPORT_TEST_MODE env var is set) |
| 520 | // We don't want anyone to interfere with the export process, which is why those |
| 521 | // hooks are not registered in production. |
| 522 | // ============================================================================ |
| 523 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 524 | /** |
| 525 | * Load test hooks from a well-known path relative to the site root. |
| 526 | * The hook file can define callback functions that are called at key |
| 527 | * points during export for testing error conditions and edge cases. |
| 528 | * |
| 529 | * Supported hook functions: |
| 530 | * test_hook_before_sql_batch(&$sql, $cursor) - Before SQL batch emitted |
| 531 | * test_hook_before_file_chunk($path, $offset, &$data) - Before file chunk |
| 532 | * test_hook_after_gzip_init($gz, $boundary) - After gzip stream init |
| 533 | * test_hook_before_completion($status, $gz, $boundary) - Before completion chunk |
| 534 | * test_hook_before_index_batch(&$batch_items, $stack) - Before index batch emitted |
| 535 | * test_hook_during_dir_scan($dir, &$entries) - During directory scanning |
| 536 | */ |
| 537 | $__test_hook_file_loaded = false; |
| 538 | function _e2e_load_test_hooks_if_needed(array $config): void { |
| 539 | global $__test_hook_file_loaded; |
| 540 | if ($__test_hook_file_loaded) { |
| 541 | return; |
| 542 | } |
| 543 | $candidates = []; |
| 544 | if (isset($config['directory'])) { |
| 545 | $dirs = is_array($config['directory']) ? $config['directory'] : [$config['directory']]; |
| 546 | foreach ($dirs as $d) { |
| 547 | $candidates[] = rtrim($d, '/') . '/wp-content/plugins/site-export/test-hooks.php'; |
| 548 | } |
| 549 | } |
| 550 | // Also check relative to this file's parent |
| 551 | $candidates[] = dirname(__DIR__) . '/test-hooks.php'; |
| 552 | foreach ($candidates as $candidate) { |
| 553 | if (file_exists($candidate)) { |
| 554 | if (function_exists('opcache_invalidate')) { |
| 555 | @opcache_invalidate($candidate, true); |
| 556 | } |
| 557 | require $candidate; |
| 558 | $__test_hook_file_loaded = true; |
| 559 | return; |
| 560 | } |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | function _e2e_call_hook(string $name, array &$args = []): void { |
| 565 | if (function_exists($name)) { |
| 566 | call_user_func_array($name, $args); |
| 567 | } |
| 568 | } |
| 569 | } |
| 570 | |
| 571 | require_once __DIR__ . "/class-mysql-dump-producer.php"; |
| 572 | require_once __DIR__ . "/class-file-tree-producer.php"; |
| 573 | |
| 574 | /** |
| 575 | * Prepares the PHP environment for streaming by disabling output buffering, |
| 576 | * compression layers, and proxy buffering. |
| 577 | */ |
| 578 | function prepare_streaming_response(): void |
| 579 | { |
| 580 | while (ob_get_level() > 0) { |
| 581 | @ob_end_clean(); |
| 582 | } |
| 583 | |
| 584 | if (!headers_sent()) { |
| 585 | @header("X-Accel-Buffering: no"); |
| 586 | @header("Cache-Control: no-store, no-cache, must-revalidate, max-age=0"); |
| 587 | @header("Pragma: no-cache"); |
| 588 | @header("Expires: 0"); |
| 589 | } |
| 590 | |
| 591 | /** |
| 592 | * zlib.output_compression buffers the entire response before compressing. The |
| 593 | * entire point of this plugin is to stream the response, therefore we use a custom |
| 594 | * GzipOutputStream. |
| 595 | */ |
| 596 | @ini_set("zlib.output_compression", "0"); |
| 597 | @ini_set("output_buffering", "0"); |
| 598 | @ini_set("implicit_flush", "1"); |
| 599 | |
| 600 | @ob_implicit_flush(true); |
| 601 | } |
| 602 | |
| 603 | /** |
| 604 | * Incremental gzip compressor that emits data as it arrives rather than |
| 605 | * buffering the entire response. |
| 606 | */ |
| 607 | class GzipOutputStream |
| 608 | { |
| 609 | private $deflate_ctx; |
| 610 | /** @var bool */ |
| 611 | private $enabled = true; |
| 612 | |
| 613 | public function __construct(bool $enabled = true) |
| 614 | { |
| 615 | $this->enabled = $enabled; |
| 616 | if ($this->enabled) { |
| 617 | $this->deflate_ctx = deflate_init(ZLIB_ENCODING_GZIP, ["level" => 6]); |
| 618 | if ($this->deflate_ctx === false) { |
| 619 | throw new \RuntimeException( |
| 620 | "deflate_init() failed — zlib may be misconfigured" |
| 621 | ); |
| 622 | } |
| 623 | if (!headers_sent()) { |
| 624 | @header("Content-Encoding: gzip"); |
| 625 | } |
| 626 | } |
| 627 | } |
| 628 | |
| 629 | /** |
| 630 | * Writes data without forcing a sync point. |
| 631 | * |
| 632 | * Uses ZLIB_NO_FLUSH so the compressor can build back-references across |
| 633 | * multiple write() calls, producing significantly better compression |
| 634 | * ratios than ZLIB_SYNC_FLUSH on every call. Data still flows out |
| 635 | * whenever zlib's internal buffer fills — the decompressor on the other |
| 636 | * end will decompress incrementally. |
| 637 | * |
| 638 | * Call sync() after each complete multipart part to guarantee the client |
| 639 | * can decompress everything emitted so far. |
| 640 | */ |
| 641 | public function write(string $data): void |
| 642 | { |
| 643 | if (!$this->enabled) { |
| 644 | echo $data; |
| 645 | return; |
| 646 | } |
| 647 | $compressed = deflate_add( |
| 648 | $this->deflate_ctx, |
| 649 | $data, |
| 650 | ZLIB_NO_FLUSH |
| 651 | ); |
| 652 | if ($compressed === false) { |
| 653 | throw new \RuntimeException("deflate_add() failed during gzip write"); |
| 654 | } |
| 655 | if ($compressed !== "") { |
| 656 | echo $compressed; |
| 657 | } |
| 658 | } |
| 659 | |
| 660 | /** |
| 661 | * Forces a sync flush so the client can decompress all data written so far. |
| 662 | */ |
| 663 | public function sync(): void |
| 664 | { |
| 665 | if (!$this->enabled) { |
| 666 | flush(); |
| 667 | return; |
| 668 | } |
| 669 | $compressed = deflate_add( |
| 670 | $this->deflate_ctx, |
| 671 | "", |
| 672 | ZLIB_SYNC_FLUSH |
| 673 | ); |
| 674 | if ($compressed === false) { |
| 675 | throw new \RuntimeException("deflate_add() failed during gzip sync"); |
| 676 | } |
| 677 | if ($compressed !== "") { |
| 678 | echo $compressed; |
| 679 | } |
| 680 | flush(); |
| 681 | } |
| 682 | |
| 683 | public function flush(): void |
| 684 | { |
| 685 | $this->sync(); |
| 686 | } |
| 687 | |
| 688 | /** |
| 689 | * Finalizes the gzip stream with ZLIB_FINISH. |
| 690 | */ |
| 691 | public function finish(): void |
| 692 | { |
| 693 | if (!$this->enabled) { |
| 694 | flush(); |
| 695 | return; |
| 696 | } |
| 697 | $final = deflate_add($this->deflate_ctx, "", ZLIB_FINISH); |
| 698 | if ($final === false) { |
| 699 | throw new \RuntimeException("deflate_add() failed during gzip finish"); |
| 700 | } |
| 701 | if ($final !== "") { |
| 702 | echo $final; |
| 703 | } |
| 704 | flush(); |
| 705 | } |
| 706 | } |
| 707 | |
| 708 | /** |
| 709 | * Deduplicates and resolves a list of paths, discarding empty entries. |
| 710 | */ |
| 711 | function normalize_path_list(array $paths): array |
| 712 | { |
| 713 | $normalized = []; |
| 714 | foreach ($paths as $path) { |
| 715 | if (!is_string($path)) { |
| 716 | continue; |
| 717 | } |
| 718 | $path = trim($path); |
| 719 | if ($path === "") { |
| 720 | continue; |
| 721 | } |
| 722 | $real = realpath($path); |
| 723 | $final = $real !== false ? $real : $path; |
| 724 | $final = rtrim($final, "/"); |
| 725 | if ($final === "") { |
| 726 | continue; |
| 727 | } |
| 728 | $normalized[$final] = true; |
| 729 | } |
| 730 | return array_keys($normalized); |
| 731 | } |
| 732 | |
| 733 | /** |
| 734 | * Walks parent directories upward from each start path to find WordPress installations. |
| 735 | */ |
| 736 | function detect_wp_roots(array $start_paths): array |
| 737 | { |
| 738 | $start_paths = normalize_path_list($start_paths); |
| 739 | $seen = []; |
| 740 | $roots = []; |
| 741 | |
| 742 | foreach ($start_paths as $start) { |
| 743 | $current = $start; |
| 744 | while ($current !== "" && !isset($seen[$current])) { |
| 745 | $seen[$current] = true; |
| 746 | $wp_load_path = $current . "/wp-load.php"; |
| 747 | $wp_config_path = $current . "/wp-config.php"; |
| 748 | $has_wp_load = file_exists($wp_load_path); |
| 749 | $has_wp_config = file_exists($wp_config_path); |
| 750 | $has_wp_content = is_dir($current . "/wp-content"); |
| 751 | if ($has_wp_load || $has_wp_config) { |
| 752 | $roots[$current] = [ |
| 753 | "path" => $current, |
| 754 | "wp_load" => $has_wp_load, |
| 755 | "wp_load_path" => $has_wp_load ? $wp_load_path : null, |
| 756 | "wp_config" => $has_wp_config, |
| 757 | "wp_config_path" => $has_wp_config ? $wp_config_path : null, |
| 758 | "wp_content" => $has_wp_content, |
| 759 | ]; |
| 760 | } |
| 761 | |
| 762 | $parent = dirname($current); |
| 763 | if ($parent === $current || $parent === "") { |
| 764 | break; |
| 765 | } |
| 766 | $current = $parent; |
| 767 | } |
| 768 | } |
| 769 | |
| 770 | return [ |
| 771 | "searched" => array_keys($seen), |
| 772 | "roots" => array_values($roots), |
| 773 | ]; |
| 774 | } |
| 775 | |
| 776 | /** |
| 777 | * Streams SQL dump fragments as gzipped multipart chunks. |
| 778 | */ |
| 779 | function endpoint_sql_chunk( |
| 780 | array $config, |
| 781 | ResourceBudget $budget |
| 782 | ): array { |
| 783 | prepare_streaming_response(); |
| 784 | $creds = resolve_db_credentials(); |
| 785 | |
| 786 | // -- Parse request parameters -- |
| 787 | $fragments_per_batch = $config["fragments_per_batch"] ?? 1000; |
| 788 | $fragments_per_batch = require_int_range( |
| 789 | "fragments_per_batch", |
| 790 | (int) $fragments_per_batch, |
| 791 | 1, |
| 792 | 10000 |
| 793 | ); |
| 794 | |
| 795 | $pdo_options = []; |
| 796 | if (!empty($config["db_unbuffered"])) { |
| 797 | $pdo_options[PDO::MYSQL_ATTR_USE_BUFFERED_QUERY] = false; |
| 798 | } |
| 799 | $mysql = create_db_connection($creds, $pdo_options); |
| 800 | |
| 801 | $producer_options = [ |
| 802 | "create_table_query" => $config["create_table_query"] ?? true, |
| 803 | ]; |
| 804 | |
| 805 | // -- Cap statement size to the smaller of client and server max_allowed_packet -- |
| 806 | // If the client sent its max_allowed_packet, cap the producer's |
| 807 | // max_statement_size so the dump stays importable on the client. |
| 808 | // We query the server's own max_allowed_packet too and use the |
| 809 | // smaller of the two (both scaled to 80% for protocol headroom). |
| 810 | if (!empty($config["max_allowed_packet"])) { |
| 811 | $client_max = (int) $config["max_allowed_packet"]; |
| 812 | if ($client_max >= 1048576 && $client_max <= 1073741824) { |
| 813 | $client_statement_size = (int) ($client_max * 0.8); |
| 814 | $server_statement_size = null; |
| 815 | try { |
| 816 | $row = $mysql |
| 817 | ->query("SELECT @@max_allowed_packet AS v") |
| 818 | ->fetch(PDO::FETCH_ASSOC); |
| 819 | if ($row && isset($row["v"])) { |
| 820 | $server_statement_size = (int) ((int) $row["v"] * 0.8); |
| 821 | } |
| 822 | } catch (Exception $e) { |
| 823 | // Ignore — producer will auto-detect |
| 824 | } |
| 825 | if ($server_statement_size !== null) { |
| 826 | $producer_options["max_statement_size"] = min( |
| 827 | $client_statement_size, |
| 828 | $server_statement_size |
| 829 | ); |
| 830 | } else { |
| 831 | $producer_options["max_statement_size"] = $client_statement_size; |
| 832 | } |
| 833 | } |
| 834 | } |
| 835 | |
| 836 | if (!empty($config["db_query_time_limit"])) { |
| 837 | $execution_budget_ms = (int) ($budget->max_time * 1000 * 0.8); |
| 838 | $query_time_limit = require_int_range( |
| 839 | "db_query_time_limit", |
| 840 | (int) $config["db_query_time_limit"], |
| 841 | 0, |
| 842 | 300000 |
| 843 | ); |
| 844 | $query_time_limit = min($query_time_limit, $execution_budget_ms); |
| 845 | if ($query_time_limit > 0) { |
| 846 | $producer_options["query_time_limit_ms"] = $query_time_limit; |
| 847 | } |
| 848 | } |
| 849 | |
| 850 | $exclude_rows = sql_exclude_rows_from_config($config, $creds["table_prefix"] ?? null); |
| 851 | if ($exclude_rows) { |
| 852 | $producer_options["exclude_rows"] = $exclude_rows; |
| 853 | } |
| 854 | |
| 855 | if (isset($config["cursor"])) { |
| 856 | $producer_options["cursor"] = $config["cursor"]; |
| 857 | } |
| 858 | |
| 859 | $reader = new WordPress\DataLiberation\MySQLDumpProducer( |
| 860 | $mysql, |
| 861 | $producer_options |
| 862 | ); |
| 863 | |
| 864 | if (ob_get_level()) { |
| 865 | ob_end_flush(); |
| 866 | } |
| 867 | |
| 868 | |
| 869 | ['gz' => $gz, 'boundary' => $boundary] = begin_multipart_stream(true); |
| 870 | |
| 871 | // E2E test hook: after gzip stream initialization |
| 872 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 873 | _e2e_load_test_hooks_if_needed($config); |
| 874 | $hook_args = [$gz, $boundary]; |
| 875 | _e2e_call_hook('test_hook_after_gzip_init', $hook_args); |
| 876 | } |
| 877 | |
| 878 | // -- Stream SQL fragments -- |
| 879 | // Pull SQL fragments from the producer in batches, writing each batch |
| 880 | // as a multipart chunk. Stop when the producer is exhausted or the |
| 881 | // resource budget (time/memory) runs out. |
| 882 | $batches_processed = 0; |
| 883 | $sql_bytes_processed = 0; |
| 884 | $aborted = false; |
| 885 | |
| 886 | try { |
| 887 | while ( |
| 888 | $budget->has_remaining() |
| 889 | ) { |
| 890 | $sql = []; |
| 891 | |
| 892 | $i = 0; |
| 893 | while ($reader->next_sql_fragment()) { |
| 894 | $sql[] = $reader->get_sql_fragment(); |
| 895 | $i++; |
| 896 | |
| 897 | if ($i >= $fragments_per_batch) { |
| 898 | break; |
| 899 | } |
| 900 | |
| 901 | if ( |
| 902 | !$budget->has_remaining() |
| 903 | ) { |
| 904 | break; |
| 905 | } |
| 906 | } |
| 907 | $sql = implode("", $sql); |
| 908 | $sql_bytes_processed += strlen($sql); |
| 909 | |
| 910 | // Does this chunk end on a complete statement boundary? |
| 911 | // The producer terminates complete statements with ";" and |
| 912 | // intermediate INSERT rows with ",", so checking the last |
| 913 | // character is sufficient. |
| 914 | $trimmed = rtrim($sql); |
| 915 | $query_complete = $trimmed !== "" && $trimmed[-1] === ";"; |
| 916 | |
| 917 | // E2E test hook: before SQL batch is emitted |
| 918 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 919 | $cursor_for_hook = $reader->get_reentrancy_cursor(); |
| 920 | $hook_args = [&$sql, $cursor_for_hook]; |
| 921 | _e2e_call_hook('test_hook_before_sql_batch', $hook_args); |
| 922 | } |
| 923 | |
| 924 | $cursor = $reader->get_reentrancy_cursor(); |
| 925 | $gz->write( |
| 926 | "--{$boundary}\r\n" . |
| 927 | "Content-Type: application/sql\r\n" . |
| 928 | "Content-Length: " . strlen($sql) . "\r\n" . |
| 929 | "X-Chunk-Type: sql\r\n" . |
| 930 | "X-Query-Complete: " . ($query_complete ? "1" : "0") . "\r\n" . |
| 931 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 932 | "\r\n" |
| 933 | ); |
| 934 | $gz->write($sql); |
| 935 | $gz->write("\r\n"); |
| 936 | $gz->sync(); |
| 937 | |
| 938 | $batches_processed++; |
| 939 | |
| 940 | if ($reader->is_finished()) { |
| 941 | break; |
| 942 | } |
| 943 | } |
| 944 | } catch (Throwable $e) { |
| 945 | $aborted = true; |
| 946 | error_log("SQL streaming error: " . $e->getMessage()); |
| 947 | emit_error_chunk($gz, $boundary, $e->getMessage()); |
| 948 | } |
| 949 | |
| 950 | // Best-effort completion chunk — the client already has the data chunks. |
| 951 | $status = $aborted ? "partial" : ($reader->is_finished() ? "complete" : "partial"); |
| 952 | |
| 953 | // E2E test hook: before completion chunk |
| 954 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 955 | $hook_args = [$status, $gz, $boundary]; |
| 956 | _e2e_call_hook('test_hook_before_completion', $hook_args); |
| 957 | } |
| 958 | |
| 959 | try { |
| 960 | $gz->write( |
| 961 | "--{$boundary}\r\n" . |
| 962 | "Content-Type: application/octet-stream\r\n" . |
| 963 | "Content-Length: 0\r\n" . |
| 964 | "X-Chunk-Type: completion\r\n" . |
| 965 | "X-Status: {$status}\r\n" . |
| 966 | "X-Batches-Processed: {$batches_processed}\r\n" . |
| 967 | "X-SQL-Bytes: {$sql_bytes_processed}\r\n" . |
| 968 | "X-Memory-Used: " . memory_get_peak_usage(true) . "\r\n" . |
| 969 | "X-Memory-Limit: " . $budget->max_memory . "\r\n" . |
| 970 | "X-Time-Elapsed: " . (microtime(true) - $budget->start_time) . "\r\n" . |
| 971 | "\r\n" . |
| 972 | "\r\n" . |
| 973 | "--{$boundary}--\r\n" |
| 974 | ); |
| 975 | $gz->finish(); |
| 976 | } catch (\Throwable $e) { |
| 977 | error_log("Export: failed to write completion chunk: " . $e->getMessage()); |
| 978 | } |
| 979 | |
| 980 | return [ |
| 981 | "status" => $status, |
| 982 | "stats" => [ |
| 983 | "batches_processed" => $batches_processed, |
| 984 | "sql_bytes" => $sql_bytes_processed, |
| 985 | "memory_used" => memory_get_peak_usage(true), |
| 986 | "time_elapsed" => microtime(true) - $budget->start_time, |
| 987 | ], |
| 988 | ]; |
| 989 | } |
| 990 | |
| 991 | /** |
| 992 | * Streams table metadata (name, estimated rows, size) from INFORMATION_SCHEMA. |
| 993 | */ |
| 994 | function endpoint_db_index( |
| 995 | array $config, |
| 996 | ResourceBudget $budget |
| 997 | ): array { |
| 998 | prepare_streaming_response(); |
| 999 | |
| 1000 | $creds = resolve_db_credentials(); |
| 1001 | |
| 1002 | $tables_per_batch = $config["tables_per_batch"] ?? 1000; |
| 1003 | $tables_per_batch = require_int_range( |
| 1004 | "tables_per_batch", |
| 1005 | (int) $tables_per_batch, |
| 1006 | 10, |
| 1007 | 10000 |
| 1008 | ); |
| 1009 | |
| 1010 | $cursor = null; |
| 1011 | if (isset($config["cursor"])) { |
| 1012 | $cursor = json_decode($config["cursor"], true); |
| 1013 | if ($cursor === null && json_last_error() !== JSON_ERROR_NONE) { |
| 1014 | throw new InvalidArgumentException( |
| 1015 | "Invalid cursor format: " . json_last_error_msg() |
| 1016 | ); |
| 1017 | } |
| 1018 | } |
| 1019 | $last_table = $cursor["last_table"] ?? ""; |
| 1020 | |
| 1021 | $mysql = create_db_connection($creds); |
| 1022 | |
| 1023 | ['gz' => $gz, 'boundary' => $boundary] = begin_multipart_stream(); |
| 1024 | |
| 1025 | $tables_processed = 0; |
| 1026 | $rows_estimated = 0; |
| 1027 | $status = "partial"; |
| 1028 | $aborted = false; |
| 1029 | |
| 1030 | try { |
| 1031 | while ( |
| 1032 | $budget->has_remaining() |
| 1033 | ) { |
| 1034 | $sql = |
| 1035 | "SELECT TABLE_NAME, TABLE_ROWS, DATA_LENGTH, INDEX_LENGTH, ENGINE, " . |
| 1036 | "TABLE_COLLATION FROM INFORMATION_SCHEMA.TABLES " . |
| 1037 | "WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME > :last " . |
| 1038 | "ORDER BY TABLE_NAME ASC LIMIT {$tables_per_batch}"; |
| 1039 | $stmt = $mysql->prepare($sql); |
| 1040 | $stmt->bindValue(":last", $last_table, PDO::PARAM_STR); |
| 1041 | $stmt->execute(); |
| 1042 | $rows = $stmt->fetchAll(PDO::FETCH_ASSOC); |
| 1043 | |
| 1044 | if (!$rows) { |
| 1045 | $status = "complete"; |
| 1046 | break; |
| 1047 | } |
| 1048 | |
| 1049 | $tables = []; |
| 1050 | foreach ($rows as $row) { |
| 1051 | $name = (string) ($row["TABLE_NAME"] ?? ""); |
| 1052 | $tables[] = [ |
| 1053 | "name" => $name, |
| 1054 | "rows" => |
| 1055 | isset($row["TABLE_ROWS"]) && is_numeric($row["TABLE_ROWS"]) |
| 1056 | ? (int) $row["TABLE_ROWS"] |
| 1057 | : null, |
| 1058 | "data_bytes" => |
| 1059 | isset($row["DATA_LENGTH"]) && is_numeric($row["DATA_LENGTH"]) |
| 1060 | ? (int) $row["DATA_LENGTH"] |
| 1061 | : null, |
| 1062 | "index_bytes" => |
| 1063 | isset($row["INDEX_LENGTH"]) && is_numeric($row["INDEX_LENGTH"]) |
| 1064 | ? (int) $row["INDEX_LENGTH"] |
| 1065 | : null, |
| 1066 | "engine" => $row["ENGINE"] ?? null, |
| 1067 | "collation" => $row["TABLE_COLLATION"] ?? null, |
| 1068 | ]; |
| 1069 | $last_table = $name; |
| 1070 | $tables_processed++; |
| 1071 | if ( |
| 1072 | isset($row["TABLE_ROWS"]) && |
| 1073 | is_numeric($row["TABLE_ROWS"]) |
| 1074 | ) { |
| 1075 | $rows_estimated += (int) $row["TABLE_ROWS"]; |
| 1076 | } |
| 1077 | } |
| 1078 | |
| 1079 | $payload = json_encode_or_throw($tables); |
| 1080 | $cursor_json = json_encode_or_throw([ |
| 1081 | "phase" => "tables", |
| 1082 | "last_table" => $last_table, |
| 1083 | ]); |
| 1084 | |
| 1085 | $gz->write( |
| 1086 | "--{$boundary}\r\n" . |
| 1087 | "Content-Type: application/json\r\n" . |
| 1088 | "Content-Length: " . strlen($payload) . "\r\n" . |
| 1089 | "X-Chunk-Type: table_stats\r\n" . |
| 1090 | "X-Tables: " . count($tables) . "\r\n" . |
| 1091 | "X-Cursor: " . base64_encode($cursor_json) . "\r\n" . |
| 1092 | "\r\n" . |
| 1093 | $payload . "\r\n" |
| 1094 | ); |
| 1095 | $gz->sync(); |
| 1096 | |
| 1097 | if (count($rows) < $tables_per_batch) { |
| 1098 | $status = "complete"; |
| 1099 | break; |
| 1100 | } |
| 1101 | } |
| 1102 | } catch (\Throwable $e) { |
| 1103 | $aborted = true; |
| 1104 | emit_error_chunk($gz, $boundary, get_class($e) . ": " . $e->getMessage()); |
| 1105 | } |
| 1106 | |
| 1107 | try { |
| 1108 | $gz->write( |
| 1109 | "--{$boundary}\r\n" . |
| 1110 | "Content-Type: application/octet-stream\r\n" . |
| 1111 | "Content-Length: 0\r\n" . |
| 1112 | "X-Chunk-Type: completion\r\n" . |
| 1113 | "X-Status: " . ($aborted ? "partial" : $status) . "\r\n" . |
| 1114 | "X-Tables-Processed: {$tables_processed}\r\n" . |
| 1115 | "X-Rows-Estimated: {$rows_estimated}\r\n" . |
| 1116 | "X-Memory-Used: " . memory_get_peak_usage(true) . "\r\n" . |
| 1117 | "X-Memory-Limit: " . $budget->max_memory . "\r\n" . |
| 1118 | "X-Time-Elapsed: " . (microtime(true) - $budget->start_time) . "\r\n" . |
| 1119 | "\r\n" . |
| 1120 | "\r\n" . |
| 1121 | "--{$boundary}--\r\n" |
| 1122 | ); |
| 1123 | $gz->finish(); |
| 1124 | } catch (\Throwable $e) { |
| 1125 | error_log("Export: failed to write completion chunk: " . $e->getMessage()); |
| 1126 | } |
| 1127 | |
| 1128 | return [ |
| 1129 | "status" => $status, |
| 1130 | "stats" => [ |
| 1131 | "tables_processed" => $tables_processed, |
| 1132 | "rows_estimated" => $rows_estimated, |
| 1133 | "memory_used" => memory_get_peak_usage(true), |
| 1134 | "time_elapsed" => microtime(true) - $budget->start_time, |
| 1135 | ], |
| 1136 | ]; |
| 1137 | } |
| 1138 | |
| 1139 | /** |
| 1140 | * Resolves directory paths from config. |
| 1141 | */ |
| 1142 | function resolve_directories(array $config): array |
| 1143 | { |
| 1144 | $directories_input = $config["directory"] ?? null; |
| 1145 | if (!$directories_input) { |
| 1146 | throw new InvalidArgumentException( |
| 1147 | "directory is required for files operation" |
| 1148 | ); |
| 1149 | } |
| 1150 | |
| 1151 | $directories = []; |
| 1152 | $dir_list = is_array($directories_input) |
| 1153 | ? $directories_input |
| 1154 | : [$directories_input]; |
| 1155 | |
| 1156 | foreach ($dir_list as $directory) { |
| 1157 | if (!is_string($directory)) { |
| 1158 | throw new InvalidArgumentException( |
| 1159 | "directory entries must be non-empty strings" |
| 1160 | ); |
| 1161 | } |
| 1162 | $directory = trim($directory); |
| 1163 | assert_valid_path($directory, "directory entry"); |
| 1164 | |
| 1165 | $real_directory = realpath($directory); |
| 1166 | if ($real_directory === false) { |
| 1167 | throw new InvalidArgumentException( |
| 1168 | "directory does not exist or is not accessible: {$directory}\n" . |
| 1169 | "Current working directory: " . |
| 1170 | getcwd() . |
| 1171 | "\n" . |
| 1172 | "Script directory: " . |
| 1173 | __DIR__ |
| 1174 | ); |
| 1175 | } |
| 1176 | |
| 1177 | $directories[] = $real_directory; |
| 1178 | } |
| 1179 | |
| 1180 | if (empty($directories)) { |
| 1181 | throw new InvalidArgumentException("No valid directories specified"); |
| 1182 | } |
| 1183 | |
| 1184 | return $directories; |
| 1185 | } |
| 1186 | |
| 1187 | /** |
| 1188 | * Returns true when traversing $candidate would only duplicate or re-enter |
| 1189 | * one of the already-scheduled roots. |
| 1190 | * |
| 1191 | * Examples: |
| 1192 | * - candidate == root: duplicate root |
| 1193 | * - candidate is a parent of root: would expose outside-tree paths and then |
| 1194 | * re-enter the scheduled root again |
| 1195 | */ |
| 1196 | function should_skip_index_root(string $candidate, array $roots): bool |
| 1197 | { |
| 1198 | foreach ($roots as $root) { |
| 1199 | if ($candidate === $root) { |
| 1200 | return true; |
| 1201 | } |
| 1202 | if ($candidate === "/" || str_starts_with($root . "/", $candidate . "/")) { |
| 1203 | return true; |
| 1204 | } |
| 1205 | } |
| 1206 | |
| 1207 | return false; |
| 1208 | } |
| 1209 | |
| 1210 | /** |
| 1211 | * Returns lightweight preflight checks: filesystem accessibility, DB connectivity, |
| 1212 | * and environment details useful for diagnostics. |
| 1213 | */ |
| 1214 | function endpoint_preflight(array $config): array |
| 1215 | { |
| 1216 | // -- Resolve filesystem roots -- |
| 1217 | // Determine which directories to scan: either from the client-provided |
| 1218 | // "directory" config, or by auto-detecting from cwd/DOCUMENT_ROOT/__DIR__. |
| 1219 | $directories = []; |
| 1220 | $dir_error = null; |
| 1221 | $has_root_input = array_key_exists("directory", $config) && $config["directory"] !== null; |
| 1222 | if ($has_root_input) { |
| 1223 | try { |
| 1224 | $directories = resolve_directories($config); |
| 1225 | } catch (Exception $e) { |
| 1226 | $dir_error = $e->getMessage(); |
| 1227 | } |
| 1228 | } |
| 1229 | |
| 1230 | $search_roots = []; |
| 1231 | if (!empty($directories)) { |
| 1232 | $search_roots = $directories; |
| 1233 | } else { |
| 1234 | $filtered = array_filter( |
| 1235 | [ |
| 1236 | getcwd() ?: null, |
| 1237 | $_SERVER["DOCUMENT_ROOT"] ?? null, |
| 1238 | isset($_SERVER["SCRIPT_FILENAME"]) |
| 1239 | ? dirname($_SERVER["SCRIPT_FILENAME"]) |
| 1240 | : null, |
| 1241 | __DIR__, |
| 1242 | ], |
| 1243 | function ($value) { |
| 1244 | return $value !== null && $value !== ""; |
| 1245 | } |
| 1246 | ); |
| 1247 | $search_roots = normalize_path_list($filtered); |
| 1248 | } |
| 1249 | |
| 1250 | // -- Detect WordPress installations -- |
| 1251 | // Walk parent directories to find wp-load.php / wp-config.php. |
| 1252 | $wp_detect = detect_wp_roots($search_roots); |
| 1253 | $detected_root_paths = []; |
| 1254 | foreach ($wp_detect["roots"] as $root) { |
| 1255 | if (!empty($root["path"])) { |
| 1256 | $detected_root_paths[] = $root["path"]; |
| 1257 | } |
| 1258 | } |
| 1259 | $detected_root_paths = normalize_path_list($detected_root_paths); |
| 1260 | |
| 1261 | $wp_load_path = null; |
| 1262 | foreach ($wp_detect["roots"] as $root) { |
| 1263 | if (!empty($root["wp_load_path"]) && is_readable($root["wp_load_path"])) { |
| 1264 | $wp_load_path = $root["wp_load_path"]; |
| 1265 | break; |
| 1266 | } |
| 1267 | } |
| 1268 | $preflight_error = null; |
| 1269 | if (!$has_root_input && $wp_load_path === null) { |
| 1270 | $preflight_error = |
| 1271 | "wp-load.php not found and no root directories were provided"; |
| 1272 | } |
| 1273 | |
| 1274 | $scan_roots = !empty($directories) ? $directories : $detected_root_paths; |
| 1275 | if (empty($scan_roots)) { |
| 1276 | $scan_roots = $search_roots; |
| 1277 | } |
| 1278 | $scan_roots = normalize_path_list($scan_roots); |
| 1279 | |
| 1280 | $wp_scan_roots = normalize_path_list( |
| 1281 | array_merge($scan_roots, $detected_root_paths) |
| 1282 | ); |
| 1283 | |
| 1284 | // -- Probe each directory -- |
| 1285 | // Check accessibility, read .htaccess files, and collect disk space info. |
| 1286 | $dir_checks = []; |
| 1287 | $htaccess_files = []; |
| 1288 | $wp_paths = []; |
| 1289 | if (!empty($scan_roots)) { |
| 1290 | foreach ($scan_roots as $dir) { |
| 1291 | $exists = is_dir($dir); |
| 1292 | $readable = $exists && is_readable($dir); |
| 1293 | $openable = false; |
| 1294 | $disk_free = null; |
| 1295 | $disk_total = null; |
| 1296 | if ($readable) { |
| 1297 | $dh = @opendir($dir); |
| 1298 | if ($dh !== false) { |
| 1299 | $openable = true; |
| 1300 | @readdir($dh); |
| 1301 | closedir($dh); |
| 1302 | } |
| 1303 | } |
| 1304 | if ($openable) { |
| 1305 | $disk_free = @disk_free_space($dir); |
| 1306 | $disk_total = @disk_total_space($dir); |
| 1307 | } |
| 1308 | $dir_checks[] = [ |
| 1309 | "path" => $dir, |
| 1310 | "exists" => $exists, |
| 1311 | "readable" => $readable, |
| 1312 | "openable" => $openable, |
| 1313 | "disk_free_bytes" => $disk_free !== false ? $disk_free : null, |
| 1314 | "disk_total_bytes" => $disk_total !== false ? $disk_total : null, |
| 1315 | ]; |
| 1316 | |
| 1317 | $htaccess_path = rtrim($dir, "/") . "/.htaccess"; |
| 1318 | if (file_exists($htaccess_path)) { |
| 1319 | $htaccess_readable = is_readable($htaccess_path); |
| 1320 | $htaccess_size = @filesize($htaccess_path); |
| 1321 | $htaccess_mtime = @filemtime($htaccess_path); |
| 1322 | $htaccess_content = null; |
| 1323 | $htaccess_truncated = false; |
| 1324 | if ($htaccess_readable) { |
| 1325 | $limit = 8192; |
| 1326 | $fh = @fopen($htaccess_path, "r"); |
| 1327 | if ($fh) { |
| 1328 | $data = @fread($fh, $limit + 1); |
| 1329 | fclose($fh); |
| 1330 | if ($data !== false) { |
| 1331 | if (strlen($data) > $limit) { |
| 1332 | $htaccess_truncated = true; |
| 1333 | $data = substr($data, 0, $limit); |
| 1334 | } |
| 1335 | $htaccess_content = $data; |
| 1336 | } |
| 1337 | } |
| 1338 | } |
| 1339 | $htaccess_files[] = [ |
| 1340 | "path" => $htaccess_path, |
| 1341 | "readable" => $htaccess_readable, |
| 1342 | "size_bytes" => $htaccess_size !== false ? $htaccess_size : null, |
| 1343 | "mtime" => $htaccess_mtime !== false ? $htaccess_mtime : null, |
| 1344 | "content" => $htaccess_content, |
| 1345 | "truncated" => $htaccess_truncated, |
| 1346 | ]; |
| 1347 | } |
| 1348 | |
| 1349 | $plugins_dir = rtrim($dir, "/") . "/wp-content/plugins"; |
| 1350 | $mu_plugins_dir = rtrim($dir, "/") . "/wp-content/mu-plugins"; |
| 1351 | $themes_dir = rtrim($dir, "/") . "/wp-content/themes"; |
| 1352 | $wp_paths[] = [ |
| 1353 | "root" => $dir, |
| 1354 | "plugins_dir" => $plugins_dir, |
| 1355 | "mu_plugins_dir" => $mu_plugins_dir, |
| 1356 | "themes_dir" => $themes_dir, |
| 1357 | ]; |
| 1358 | } |
| 1359 | } |
| 1360 | |
| 1361 | if (!empty($wp_scan_roots)) { |
| 1362 | foreach ($wp_scan_roots as $dir) { |
| 1363 | $plugins_dir = rtrim($dir, "/") . "/wp-content/plugins"; |
| 1364 | $mu_plugins_dir = rtrim($dir, "/") . "/wp-content/mu-plugins"; |
| 1365 | $themes_dir = rtrim($dir, "/") . "/wp-content/themes"; |
| 1366 | $wp_paths[] = [ |
| 1367 | "root" => $dir, |
| 1368 | "plugins_dir" => $plugins_dir, |
| 1369 | "mu_plugins_dir" => $mu_plugins_dir, |
| 1370 | "themes_dir" => $themes_dir, |
| 1371 | ]; |
| 1372 | } |
| 1373 | } |
| 1374 | |
| 1375 | $wp_paths = normalize_path_list( |
| 1376 | array_map( |
| 1377 | function ($entry) { |
| 1378 | return $entry["root"] ?? null; |
| 1379 | }, |
| 1380 | $wp_paths |
| 1381 | ) |
| 1382 | ); |
| 1383 | $wp_paths = array_map(function ($root) { |
| 1384 | $root = rtrim($root, "/"); |
| 1385 | return [ |
| 1386 | "root" => $root, |
| 1387 | "plugins_dir" => $root . "/wp-content/plugins", |
| 1388 | "mu_plugins_dir" => $root . "/wp-content/mu-plugins", |
| 1389 | "themes_dir" => $root . "/wp-content/themes", |
| 1390 | ]; |
| 1391 | }, $wp_paths); |
| 1392 | |
| 1393 | $filesystem_ok = true; |
| 1394 | if ($dir_error !== null) { |
| 1395 | $filesystem_ok = false; |
| 1396 | } elseif (!empty($dir_checks)) { |
| 1397 | foreach ($dir_checks as $check) { |
| 1398 | if (empty($check["openable"])) { |
| 1399 | $filesystem_ok = false; |
| 1400 | break; |
| 1401 | } |
| 1402 | } |
| 1403 | } elseif ($wp_load_path === null) { |
| 1404 | $filesystem_ok = false; |
| 1405 | } |
| 1406 | |
| 1407 | // -- PHP resource limits -- |
| 1408 | // Gather memory, upload, and execution limits so the client can tune |
| 1409 | // its request sizes accordingly. |
| 1410 | $memory_limit_raw = ini_get("memory_limit"); |
| 1411 | $memory_limit_bytes = null; |
| 1412 | if ($memory_limit_raw !== false && $memory_limit_raw !== "") { |
| 1413 | if ($memory_limit_raw === "-1") { |
| 1414 | $memory_limit_bytes = PHP_INT_MAX; |
| 1415 | } else { |
| 1416 | $memory_limit_bytes = parse_size($memory_limit_raw); |
| 1417 | } |
| 1418 | } |
| 1419 | $memory_used = memory_get_usage(true); |
| 1420 | $memory_available = |
| 1421 | $memory_limit_bytes !== null && $memory_limit_bytes !== PHP_INT_MAX |
| 1422 | ? max(0, $memory_limit_bytes - $memory_used) |
| 1423 | : null; |
| 1424 | $post_max_size_raw = ini_get("post_max_size"); |
| 1425 | $upload_max_filesize_raw = ini_get("upload_max_filesize"); |
| 1426 | $post_max_bytes = |
| 1427 | $post_max_size_raw !== false && $post_max_size_raw !== "" |
| 1428 | ? parse_size($post_max_size_raw) |
| 1429 | : null; |
| 1430 | $upload_max_bytes = |
| 1431 | $upload_max_filesize_raw !== false && $upload_max_filesize_raw !== "" |
| 1432 | ? parse_size($upload_max_filesize_raw) |
| 1433 | : null; |
| 1434 | $max_request_bytes = null; |
| 1435 | if ($post_max_bytes !== null && $upload_max_bytes !== null) { |
| 1436 | $max_request_bytes = min($post_max_bytes, $upload_max_bytes); |
| 1437 | } elseif ($post_max_bytes !== null) { |
| 1438 | $max_request_bytes = $post_max_bytes; |
| 1439 | } elseif ($upload_max_bytes !== null) { |
| 1440 | $max_request_bytes = $upload_max_bytes; |
| 1441 | } |
| 1442 | |
| 1443 | // -- PHP extensions -- |
| 1444 | // Report loaded extensions and image processing capabilities. |
| 1445 | $extensions = get_loaded_extensions(); |
| 1446 | sort($extensions, SORT_STRING); |
| 1447 | $extension_versions = []; |
| 1448 | foreach ([ |
| 1449 | "curl", |
| 1450 | "gd", |
| 1451 | "imagick", |
| 1452 | "pdo_mysql", |
| 1453 | "mysqli", |
| 1454 | "mbstring", |
| 1455 | "zlib", |
| 1456 | "openssl", |
| 1457 | "fileinfo", |
| 1458 | "exif", |
| 1459 | ] as $ext) { |
| 1460 | if (extension_loaded($ext)) { |
| 1461 | $ver = phpversion($ext); |
| 1462 | $extension_versions[$ext] = $ver !== false ? $ver : true; |
| 1463 | } |
| 1464 | } |
| 1465 | |
| 1466 | $gd_info = function_exists("gd_info") ? gd_info() : null; |
| 1467 | $gd_formats = null; |
| 1468 | $gd_version = null; |
| 1469 | if (is_array($gd_info)) { |
| 1470 | $gd_version = $gd_info["GD Version"] ?? null; |
| 1471 | $gd_formats = [ |
| 1472 | "gif_create" => (bool) ($gd_info["GIF Create Support"] ?? false), |
| 1473 | "gif_read" => (bool) ($gd_info["GIF Read Support"] ?? false), |
| 1474 | "jpeg" => (bool) ($gd_info["JPEG Support"] ?? false), |
| 1475 | "png" => (bool) ($gd_info["PNG Support"] ?? false), |
| 1476 | "webp" => (bool) ($gd_info["WebP Support"] ?? false), |
| 1477 | "avif" => (bool) ($gd_info["AVIF Support"] ?? false), |
| 1478 | "bmp" => (bool) ($gd_info["BMP Support"] ?? false), |
| 1479 | "wbmp" => (bool) ($gd_info["WBMP Support"] ?? false), |
| 1480 | "xpm" => (bool) ($gd_info["XPM Support"] ?? false), |
| 1481 | ]; |
| 1482 | } |
| 1483 | $imagick_version = extension_loaded("imagick") |
| 1484 | ? (phpversion("imagick") ?: null) |
| 1485 | : null; |
| 1486 | |
| 1487 | // -- Database connectivity -- |
| 1488 | // Find wp-config.php credentials, connect to MySQL, and probe server |
| 1489 | // variables (charset, collation, max_allowed_packet, sql_mode). |
| 1490 | // If WordPress is loadable, also read options like active_plugins, |
| 1491 | // theme, siteurl, multisite config, and WP constants. |
| 1492 | $db = [ |
| 1493 | "db_engine" => is_sqlite_site() ? "sqlite" : "mysql", |
| 1494 | "credentials_found" => false, |
| 1495 | "connected" => false, |
| 1496 | "can_query" => false, |
| 1497 | "version" => null, |
| 1498 | "db_charset" => null, |
| 1499 | "db_collation" => null, |
| 1500 | "server_charset" => null, |
| 1501 | "server_collation" => null, |
| 1502 | "table_listable" => null, |
| 1503 | "table_list_error" => null, |
| 1504 | "wp" => [ |
| 1505 | "wp_config_path" => null, |
| 1506 | "wp_load_path" => null, |
| 1507 | "wp_load_attempted" => false, |
| 1508 | "wp_load_loaded" => false, |
| 1509 | "wp_load_error" => null, |
| 1510 | "table_prefix" => null, |
| 1511 | "options_table" => null, |
| 1512 | "active_plugins" => null, |
| 1513 | "active_sitewide_plugins" => null, |
| 1514 | "theme_template" => null, |
| 1515 | "theme_stylesheet" => null, |
| 1516 | "siteurl" => null, |
| 1517 | "home" => null, |
| 1518 | "paths_urls" => null, |
| 1519 | "multisite" => null, |
| 1520 | "constants" => null, |
| 1521 | "constant_names" => null, |
| 1522 | "wpdb_charset" => null, |
| 1523 | "wpdb_collation" => null, |
| 1524 | "error" => null, |
| 1525 | ], |
| 1526 | "error" => null, |
| 1527 | ]; |
| 1528 | |
| 1529 | $credential_roots = []; |
| 1530 | if (!empty($directories)) { |
| 1531 | $credential_roots = $directories; |
| 1532 | } elseif (!empty($detected_root_paths)) { |
| 1533 | $credential_roots = $detected_root_paths; |
| 1534 | } elseif (!empty($search_roots)) { |
| 1535 | $credential_roots = $search_roots; |
| 1536 | } |
| 1537 | $credential_roots = normalize_path_list($credential_roots); |
| 1538 | |
| 1539 | $db["wp"]["wp_load_path"] = $wp_load_path; |
| 1540 | $db["wp"]["wp_load_loaded"] = function_exists("get_option"); |
| 1541 | |
| 1542 | $creds = null; |
| 1543 | try { |
| 1544 | $creds = resolve_db_credentials(); |
| 1545 | $db["wp"]["wp_config_path"] = $creds["wp_config_path"]; |
| 1546 | $db["wp"]["table_prefix"] = $creds["table_prefix"]; |
| 1547 | $db["db_engine"] = $creds["db_engine"] ?? $db["db_engine"]; |
| 1548 | $db["credentials_found"] = true; |
| 1549 | } catch (InvalidArgumentException $e) { |
| 1550 | $db["error"] = $e->getMessage(); |
| 1551 | } |
| 1552 | |
| 1553 | if ($creds !== null) { |
| 1554 | $required_ext = ($creds["db_engine"] ?? "mysql") === "sqlite" ? "pdo_sqlite" : "pdo_mysql"; |
| 1555 | if (!extension_loaded($required_ext)) { |
| 1556 | $db["error"] = "{$required_ext} extension not loaded"; |
| 1557 | } else { |
| 1558 | try { |
| 1559 | $mysql = create_db_connection($creds); |
| 1560 | $db["connected"] = true; |
| 1561 | |
| 1562 | $version = $mysql->query("SELECT VERSION()")->fetchColumn(); |
| 1563 | $db["version"] = $version !== false ? (string) $version : null; |
| 1564 | $db["can_query"] = true; |
| 1565 | |
| 1566 | $table_prefix = $db["wp"]["table_prefix"]; |
| 1567 | if ($table_prefix === null || $table_prefix === "") { |
| 1568 | try { |
| 1569 | $stmt = $mysql->query( |
| 1570 | "SELECT TABLE_NAME FROM INFORMATION_SCHEMA.TABLES " . |
| 1571 | "WHERE TABLE_SCHEMA = DATABASE() " . |
| 1572 | "AND TABLE_NAME LIKE '%\\_options' ESCAPE '\\\\' " . |
| 1573 | "LIMIT 5" |
| 1574 | ); |
| 1575 | if ($stmt !== false) { |
| 1576 | $names = $stmt->fetchAll(PDO::FETCH_COLUMN); |
| 1577 | foreach ($names as $name) { |
| 1578 | if (!is_string($name)) { |
| 1579 | continue; |
| 1580 | } |
| 1581 | $suffix = "options"; |
| 1582 | if ( |
| 1583 | strlen($name) > strlen($suffix) && |
| 1584 | substr($name, -strlen($suffix)) === $suffix |
| 1585 | ) { |
| 1586 | $table_prefix = substr( |
| 1587 | $name, |
| 1588 | 0, |
| 1589 | -strlen($suffix) |
| 1590 | ); |
| 1591 | break; |
| 1592 | } |
| 1593 | } |
| 1594 | } |
| 1595 | } catch (Exception $e) { |
| 1596 | if ($db["wp"]["error"] === null) { |
| 1597 | $db["wp"]["error"] = $e->getMessage(); |
| 1598 | } |
| 1599 | } |
| 1600 | } |
| 1601 | |
| 1602 | if ($table_prefix !== null && $table_prefix !== "") { |
| 1603 | $db["wp"]["table_prefix"] = $table_prefix; |
| 1604 | $db["wp"]["options_table"] = $table_prefix . "options"; |
| 1605 | } |
| 1606 | |
| 1607 | $wp_load_attempted = false; |
| 1608 | $wp_load_error = null; |
| 1609 | $wp_loaded = $db["wp"]["wp_load_loaded"]; |
| 1610 | if (!$wp_loaded && $wp_load_path !== null) { |
| 1611 | $wp_load_attempted = true; |
| 1612 | $errors = []; |
| 1613 | $handler = function ($errno, $errstr) use (&$errors) { |
| 1614 | $errors[] = $errstr; |
| 1615 | return true; |
| 1616 | }; |
| 1617 | set_error_handler($handler); |
| 1618 | $include_result = @include_once $wp_load_path; |
| 1619 | restore_error_handler(); |
| 1620 | if ($include_result === false) { |
| 1621 | $wp_load_error = !empty($errors) |
| 1622 | ? implode("; ", $errors) |
| 1623 | : "Failed to include wp-load.php"; |
| 1624 | } |
| 1625 | if (function_exists("get_option")) { |
| 1626 | $wp_loaded = true; |
| 1627 | } elseif ($wp_load_error === null) { |
| 1628 | $wp_load_error = "wp-load.php did not load WordPress functions"; |
| 1629 | } |
| 1630 | } |
| 1631 | |
| 1632 | $db["wp"]["wp_load_attempted"] = $wp_load_attempted; |
| 1633 | $db["wp"]["wp_load_loaded"] = $wp_loaded; |
| 1634 | if ($wp_load_error !== null) { |
| 1635 | $db["wp"]["wp_load_error"] = $wp_load_error; |
| 1636 | } |
| 1637 | |
| 1638 | if ($wp_loaded) { |
| 1639 | try { |
| 1640 | $wpdb_global = $GLOBALS["wpdb"] ?? null; |
| 1641 | if (is_object($wpdb_global)) { |
| 1642 | $wpdb_charset = (string) ($wpdb_global->charset ?? ""); |
| 1643 | $db["wp"]["wpdb_charset"] = $wpdb_charset !== "" ? $wpdb_charset : null; |
| 1644 | |
| 1645 | $wpdb_collation = (string) ($wpdb_global->collate ?? ""); |
| 1646 | $db["wp"]["wpdb_collation"] = $wpdb_collation !== "" ? $wpdb_collation : null; |
| 1647 | } |
| 1648 | |
| 1649 | $db["wp"]["active_plugins"] = get_option("active_plugins"); |
| 1650 | $db["wp"]["theme_stylesheet"] = get_option("stylesheet"); |
| 1651 | $db["wp"]["theme_template"] = get_option("template"); |
| 1652 | $db["wp"]["siteurl"] = get_option("siteurl"); |
| 1653 | $db["wp"]["home"] = get_option("home"); |
| 1654 | // Resolve wp-admin and wp-includes paths. |
| 1655 | // These are always ABSPATH/wp-admin and ABSPATH/WPINC |
| 1656 | // by WordPress convention, but on hosts like WP Cloud |
| 1657 | // they may be symlinks (e.g. __wp__/wp-admin -> /wordpress/wp-admin). |
| 1658 | // Use realpath() to resolve to the physical location so |
| 1659 | // the importer knows where the files actually live. |
| 1660 | $wp_admin_path = null; |
| 1661 | if (defined("ABSPATH")) { |
| 1662 | $wp_admin_candidate = ABSPATH . "wp-admin"; |
| 1663 | $wp_admin_real = realpath($wp_admin_candidate); |
| 1664 | if ($wp_admin_real !== false && is_dir($wp_admin_real)) { |
| 1665 | $wp_admin_path = $wp_admin_real; |
| 1666 | } |
| 1667 | } |
| 1668 | |
| 1669 | $wp_includes_path = null; |
| 1670 | if (defined("ABSPATH")) { |
| 1671 | $wpinc = defined("WPINC") ? WPINC : "wp-includes"; |
| 1672 | $wp_includes_candidate = ABSPATH . $wpinc; |
| 1673 | $wp_includes_real = realpath($wp_includes_candidate); |
| 1674 | if ($wp_includes_real !== false && is_dir($wp_includes_real)) { |
| 1675 | $wp_includes_path = $wp_includes_real; |
| 1676 | } |
| 1677 | } |
| 1678 | |
| 1679 | // Use realpath() to resolve any symlinks in |
| 1680 | // ABSPATH (e.g. /wordpress -> /srv/wpcloud/core/6.9.4 |
| 1681 | // on WP Cloud). This matches the convention used for |
| 1682 | // all other paths below and ensures the importer can |
| 1683 | // find the directory at the resolved location where |
| 1684 | // files are actually downloaded. |
| 1685 | $abspath_raw = defined("ABSPATH") |
| 1686 | ? rtrim(ABSPATH, "/") |
| 1687 | : null; |
| 1688 | $abspath_resolved = null; |
| 1689 | if ($abspath_raw !== null) { |
| 1690 | $abspath_real = realpath($abspath_raw); |
| 1691 | $abspath_resolved = $abspath_real !== false |
| 1692 | ? rtrim($abspath_real, "/") |
| 1693 | : $abspath_raw; |
| 1694 | } |
| 1695 | |
| 1696 | $paths_urls = [ |
| 1697 | "abspath" => $abspath_resolved, |
| 1698 | "wp_admin_path" => $wp_admin_path, |
| 1699 | "wp_includes_path" => $wp_includes_path, |
| 1700 | "content_dir" => defined("WP_CONTENT_DIR") |
| 1701 | ? realpath(rtrim(WP_CONTENT_DIR, "/")) |
| 1702 | : null, |
| 1703 | "content_url" => function_exists("content_url") |
| 1704 | ? content_url() |
| 1705 | : (defined("WP_CONTENT_URL") ? WP_CONTENT_URL : null), |
| 1706 | "plugins_dir" => defined("WP_PLUGIN_DIR") |
| 1707 | ? realpath(rtrim(WP_PLUGIN_DIR, "/")) |
| 1708 | : null, |
| 1709 | "plugins_url" => function_exists("plugins_url") |
| 1710 | ? plugins_url() |
| 1711 | : (defined("WP_PLUGIN_URL") ? WP_PLUGIN_URL : null), |
| 1712 | "mu_plugins_dir" => defined("WPMU_PLUGIN_DIR") |
| 1713 | ? realpath(rtrim(WPMU_PLUGIN_DIR, "/")) |
| 1714 | : null, |
| 1715 | "mu_plugins_url" => function_exists("content_url") |
| 1716 | ? content_url("/mu-plugins") |
| 1717 | : (defined("WPMU_PLUGIN_URL") ? WPMU_PLUGIN_URL : null), |
| 1718 | "uploads" => [ |
| 1719 | "basedir" => null, |
| 1720 | "baseurl" => null, |
| 1721 | "subdir" => null, |
| 1722 | ], |
| 1723 | "site_url" => function_exists("site_url") |
| 1724 | ? site_url() |
| 1725 | : null, |
| 1726 | "home_url" => function_exists("home_url") |
| 1727 | ? home_url() |
| 1728 | : null, |
| 1729 | "network_site_url" => function_exists("network_site_url") |
| 1730 | ? network_site_url() |
| 1731 | : null, |
| 1732 | "network_home_url" => function_exists("network_home_url") |
| 1733 | ? network_home_url() |
| 1734 | : null, |
| 1735 | ]; |
| 1736 | |
| 1737 | if (function_exists("wp_upload_dir")) { |
| 1738 | $uploads = wp_upload_dir(null, false); |
| 1739 | if (is_array($uploads)) { |
| 1740 | $raw_basedir = $uploads["basedir"] ?? null; |
| 1741 | $paths_urls["uploads"]["basedir"] = |
| 1742 | is_string($raw_basedir) ? realpath($raw_basedir) : null; |
| 1743 | $paths_urls["uploads"]["baseurl"] = |
| 1744 | $uploads["baseurl"] ?? null; |
| 1745 | $paths_urls["uploads"]["subdir"] = |
| 1746 | $uploads["subdir"] ?? null; |
| 1747 | } |
| 1748 | } |
| 1749 | $db["wp"]["paths_urls"] = $paths_urls; |
| 1750 | |
| 1751 | if ( |
| 1752 | function_exists("is_multisite") && |
| 1753 | is_multisite() && |
| 1754 | function_exists("get_site_option") |
| 1755 | ) { |
| 1756 | $db["wp"]["active_sitewide_plugins"] = get_site_option( |
| 1757 | "active_sitewide_plugins" |
| 1758 | ); |
| 1759 | } |
| 1760 | |
| 1761 | $multisite = [ |
| 1762 | "enabled" => false, |
| 1763 | "subdomain_install" => defined("SUBDOMAIN_INSTALL") |
| 1764 | ? (bool) SUBDOMAIN_INSTALL |
| 1765 | : null, |
| 1766 | "current_blog_id" => |
| 1767 | function_exists("get_current_blog_id") |
| 1768 | ? get_current_blog_id() |
| 1769 | : null, |
| 1770 | "current_network_id" => |
| 1771 | function_exists("get_current_network_id") |
| 1772 | ? get_current_network_id() |
| 1773 | : null, |
| 1774 | "domain_current_site" => defined("DOMAIN_CURRENT_SITE") |
| 1775 | ? DOMAIN_CURRENT_SITE |
| 1776 | : null, |
| 1777 | "path_current_site" => defined("PATH_CURRENT_SITE") |
| 1778 | ? PATH_CURRENT_SITE |
| 1779 | : null, |
| 1780 | "site_id_current_site" => |
| 1781 | defined("SITE_ID_CURRENT_SITE") |
| 1782 | ? SITE_ID_CURRENT_SITE |
| 1783 | : null, |
| 1784 | "blog_id_current_site" => |
| 1785 | defined("BLOG_ID_CURRENT_SITE") |
| 1786 | ? BLOG_ID_CURRENT_SITE |
| 1787 | : null, |
| 1788 | "network" => null, |
| 1789 | "site" => null, |
| 1790 | ]; |
| 1791 | |
| 1792 | if (function_exists("is_multisite") && is_multisite()) { |
| 1793 | $multisite["enabled"] = true; |
| 1794 | $network_id = $multisite["current_network_id"]; |
| 1795 | if ($network_id !== null && function_exists("get_network")) { |
| 1796 | $network = get_network($network_id); |
| 1797 | if (is_object($network)) { |
| 1798 | $multisite["network"] = [ |
| 1799 | "id" => $network->id ?? null, |
| 1800 | "domain" => $network->domain ?? null, |
| 1801 | "path" => $network->path ?? null, |
| 1802 | "site_id" => $network->site_id ?? null, |
| 1803 | "registered" => $network->registered ?? null, |
| 1804 | "last_updated" => $network->last_updated ?? null, |
| 1805 | ]; |
| 1806 | } |
| 1807 | } |
| 1808 | |
| 1809 | $blog_id = $multisite["current_blog_id"]; |
| 1810 | if ($blog_id !== null && function_exists("get_site")) { |
| 1811 | $site = get_site($blog_id); |
| 1812 | if (is_object($site)) { |
| 1813 | $multisite["site"] = [ |
| 1814 | "blog_id" => $site->blog_id ?? null, |
| 1815 | "domain" => $site->domain ?? null, |
| 1816 | "path" => $site->path ?? null, |
| 1817 | "site_id" => $site->site_id ?? null, |
| 1818 | "registered" => $site->registered ?? null, |
| 1819 | "last_updated" => $site->last_updated ?? null, |
| 1820 | "public" => $site->public ?? null, |
| 1821 | "archived" => $site->archived ?? null, |
| 1822 | "mature" => $site->mature ?? null, |
| 1823 | "spam" => $site->spam ?? null, |
| 1824 | "deleted" => $site->deleted ?? null, |
| 1825 | "lang_id" => $site->lang_id ?? null, |
| 1826 | ]; |
| 1827 | } |
| 1828 | } |
| 1829 | } |
| 1830 | $db["wp"]["multisite"] = $multisite; |
| 1831 | |
| 1832 | // Capture all WP_* constants plus a few other |
| 1833 | // WordPress-specific ones that don't follow the prefix. |
| 1834 | // We use the "user" category from get_defined_constants(true) |
| 1835 | // which only includes constants set via define(), excluding |
| 1836 | // the thousands of constants from PHP extensions. |
| 1837 | $user_constants = get_defined_constants(true)["user"] ?? []; |
| 1838 | // Include non-WP_* constants that are still |
| 1839 | // important for understanding a WordPress site. |
| 1840 | $extra_constants_names = [ |
| 1841 | "WPMU_PLUGIN_DIR", |
| 1842 | "WPMU_PLUGIN_URL", |
| 1843 | "UPLOADS", |
| 1844 | "ABSPATH", |
| 1845 | "DOMAIN_CURRENT_SITE", |
| 1846 | "PATH_CURRENT_SITE", |
| 1847 | "SITE_ID_CURRENT_SITE", |
| 1848 | "BLOG_ID_CURRENT_SITE", |
| 1849 | "SUBDOMAIN_INSTALL", |
| 1850 | "TEMPLATEPATH", |
| 1851 | "STYLESHEETPATH", |
| 1852 | "FORCE_SSL_LOGIN", |
| 1853 | "FORCE_SSL_ADMIN", |
| 1854 | "SAVEQUERIES", |
| 1855 | ]; |
| 1856 | $db["wp"]["constant_values"] = []; |
| 1857 | // Names of all runtime-defined constants (without values) |
| 1858 | // so the importer can use their presence as a detection |
| 1859 | // signal without leaking secret values. Only includes |
| 1860 | // constants set via define(), not PHP extension constants. |
| 1861 | $db["wp"]["constant_names"] = []; |
| 1862 | foreach ($user_constants as $name => $value) { |
| 1863 | if (strncmp($name, "WP_", 3) === 0 || in_array($name, $extra_constants_names)) { |
| 1864 | $db["wp"]["constant_values"][$name] = $value; |
| 1865 | } else { |
| 1866 | $db["wp"]["constant_names"][] = $name; |
| 1867 | } |
| 1868 | } |
| 1869 | |
| 1870 | global $wp_version; |
| 1871 | $db["wp"]["wp_version"] = isset($wp_version) && is_string($wp_version) |
| 1872 | ? $wp_version |
| 1873 | : null; |
| 1874 | } catch (Throwable $e) { |
| 1875 | if ($db["wp"]["error"] === null) { |
| 1876 | $db["wp"]["error"] = $e->getMessage(); |
| 1877 | } |
| 1878 | } |
| 1879 | } else { |
| 1880 | if ($db["wp"]["error"] === null) { |
| 1881 | if ($wp_load_error !== null) { |
| 1882 | $db["wp"]["error"] = $wp_load_error; |
| 1883 | } elseif ($wp_load_path === null) { |
| 1884 | $db["wp"]["error"] = "wp-load.php not found"; |
| 1885 | } else { |
| 1886 | $db["wp"]["error"] = "wp-load.php not loaded"; |
| 1887 | } |
| 1888 | } |
| 1889 | } |
| 1890 | |
| 1891 | // MySQL server variables — these don't apply to SQLite, |
| 1892 | // so wrap in a separate try/catch to avoid losing WP data |
| 1893 | // gathered earlier if the query fails. |
| 1894 | try { |
| 1895 | $vars = $mysql |
| 1896 | ->query( |
| 1897 | "SELECT @@character_set_database AS db_charset, " . |
| 1898 | "@@collation_database AS db_collation, " . |
| 1899 | "@@character_set_server AS server_charset, " . |
| 1900 | "@@collation_server AS server_collation, " . |
| 1901 | "@@character_set_connection AS connection_charset, " . |
| 1902 | "@@collation_connection AS connection_collation, " . |
| 1903 | "@@max_allowed_packet AS max_allowed_packet, " . |
| 1904 | "@@sql_mode AS sql_mode, " . |
| 1905 | "@@lower_case_table_names AS lower_case_table_names" |
| 1906 | ) |
| 1907 | ->fetch(PDO::FETCH_ASSOC); |
| 1908 | if (is_array($vars)) { |
| 1909 | $db["db_charset"] = $vars["db_charset"] ?? null; |
| 1910 | $db["db_collation"] = $vars["db_collation"] ?? null; |
| 1911 | $db["server_charset"] = $vars["server_charset"] ?? null; |
| 1912 | $db["server_collation"] = $vars["server_collation"] ?? null; |
| 1913 | $db["connection_charset"] = $vars["connection_charset"] ?? null; |
| 1914 | $db["connection_collation"] = $vars["connection_collation"] ?? null; |
| 1915 | $db["max_allowed_packet"] = isset($vars["max_allowed_packet"]) |
| 1916 | ? (int) $vars["max_allowed_packet"] |
| 1917 | : null; |
| 1918 | $db["sql_mode"] = $vars["sql_mode"] ?? null; |
| 1919 | $db["lower_case_table_names"] = isset( |
| 1920 | $vars["lower_case_table_names"] |
| 1921 | ) |
| 1922 | ? (int) $vars["lower_case_table_names"] |
| 1923 | : null; |
| 1924 | } |
| 1925 | } catch (Exception $e) { |
| 1926 | // Expected for SQLite — these MySQL system variables |
| 1927 | // don't exist. The null defaults are correct. |
| 1928 | } |
| 1929 | |
| 1930 | try { |
| 1931 | $stmt = $mysql->query( |
| 1932 | "SELECT TABLE_NAME FROM INFORMATION_SCHEMA.TABLES " . |
| 1933 | "WHERE TABLE_SCHEMA = DATABASE() LIMIT 1" |
| 1934 | ); |
| 1935 | if ($stmt !== false) { |
| 1936 | $stmt->fetchColumn(); |
| 1937 | $db["table_listable"] = true; |
| 1938 | $db["table_list_error"] = null; |
| 1939 | } else { |
| 1940 | $db["table_listable"] = false; |
| 1941 | $db["table_list_error"] = "SHOW TABLES failed"; |
| 1942 | } |
| 1943 | } catch (Exception $e) { |
| 1944 | $db["table_listable"] = false; |
| 1945 | $db["table_list_error"] = $e->getMessage(); |
| 1946 | } |
| 1947 | } catch (Exception $e) { |
| 1948 | $db["error"] = $e->getMessage(); |
| 1949 | } |
| 1950 | } |
| 1951 | } |
| 1952 | |
| 1953 | // -- WordPress content inventory -- |
| 1954 | // If WordPress was loaded, use its constants for the real plugin/theme/ |
| 1955 | // mu-plugin paths. Otherwise, fall back to conventional wp-content/ layout. |
| 1956 | // Scan each directory to list installed plugins, mu-plugins, and themes. |
| 1957 | $wp_runtime_paths = null; |
| 1958 | if ($db["wp"]["wp_load_loaded"]) { |
| 1959 | $runtime_root = defined("ABSPATH") ? rtrim(ABSPATH, "/") : null; |
| 1960 | $content_dir = defined("WP_CONTENT_DIR") |
| 1961 | ? rtrim(WP_CONTENT_DIR, "/") |
| 1962 | : null; |
| 1963 | $plugins_dir = defined("WP_PLUGIN_DIR") |
| 1964 | ? rtrim(WP_PLUGIN_DIR, "/") |
| 1965 | : null; |
| 1966 | $mu_plugins_dir = defined("WPMU_PLUGIN_DIR") |
| 1967 | ? rtrim(WPMU_PLUGIN_DIR, "/") |
| 1968 | : null; |
| 1969 | $themes_dir = null; |
| 1970 | if (function_exists("get_theme_root")) { |
| 1971 | $themes_dir = get_theme_root(); |
| 1972 | if (is_string($themes_dir)) { |
| 1973 | $themes_dir = rtrim($themes_dir, "/"); |
| 1974 | } else { |
| 1975 | $themes_dir = null; |
| 1976 | } |
| 1977 | } |
| 1978 | |
| 1979 | if ($content_dir !== null) { |
| 1980 | if ($plugins_dir === null) { |
| 1981 | $plugins_dir = $content_dir . "/plugins"; |
| 1982 | } |
| 1983 | if ($mu_plugins_dir === null) { |
| 1984 | $mu_plugins_dir = $content_dir . "/mu-plugins"; |
| 1985 | } |
| 1986 | if ($themes_dir === null) { |
| 1987 | $themes_dir = $content_dir . "/themes"; |
| 1988 | } |
| 1989 | } |
| 1990 | |
| 1991 | $wp_runtime_paths = [ |
| 1992 | "root" => $runtime_root ?? $content_dir, |
| 1993 | "content_dir" => $content_dir, |
| 1994 | "plugins_dir" => $plugins_dir, |
| 1995 | "mu_plugins_dir" => $mu_plugins_dir, |
| 1996 | "themes_dir" => $themes_dir, |
| 1997 | ]; |
| 1998 | } |
| 1999 | |
| 2000 | $wp_content = [ |
| 2001 | "roots" => [], |
| 2002 | ]; |
| 2003 | $wp_paths_to_scan = $wp_runtime_paths !== null ? [$wp_runtime_paths] : $wp_paths; |
| 2004 | foreach ($wp_paths_to_scan as $paths) { |
| 2005 | $root_entry = [ |
| 2006 | "root" => $paths["root"], |
| 2007 | "content_dir" => $paths["content_dir"] ?? null, |
| 2008 | "plugins" => [], |
| 2009 | "mu_plugins" => [], |
| 2010 | "themes" => [], |
| 2011 | ]; |
| 2012 | $plugins_dir = $paths["plugins_dir"] ?? null; |
| 2013 | if ($plugins_dir !== null && is_dir($plugins_dir) && is_readable($plugins_dir)) { |
| 2014 | $entries = @scandir($plugins_dir) ?: []; |
| 2015 | foreach ($entries as $entry) { |
| 2016 | if ($entry === "." || $entry === "..") { |
| 2017 | continue; |
| 2018 | } |
| 2019 | $path = $plugins_dir . "/" . $entry; |
| 2020 | $root_entry["plugins"][] = [ |
| 2021 | "name" => $entry, |
| 2022 | "type" => is_dir($path) ? "dir" : "file", |
| 2023 | ]; |
| 2024 | } |
| 2025 | usort( |
| 2026 | $root_entry["plugins"], |
| 2027 | function ($a, $b) { |
| 2028 | return strcmp($a["name"], $b["name"]); |
| 2029 | } |
| 2030 | ); |
| 2031 | } |
| 2032 | |
| 2033 | $mu_plugins_dir = $paths["mu_plugins_dir"] ?? null; |
| 2034 | if ($mu_plugins_dir !== null && is_dir($mu_plugins_dir) && is_readable($mu_plugins_dir)) { |
| 2035 | $entries = @scandir($mu_plugins_dir) ?: []; |
| 2036 | foreach ($entries as $entry) { |
| 2037 | if ($entry === "." || $entry === "..") { |
| 2038 | continue; |
| 2039 | } |
| 2040 | $path = $mu_plugins_dir . "/" . $entry; |
| 2041 | $root_entry["mu_plugins"][] = [ |
| 2042 | "name" => $entry, |
| 2043 | "type" => is_dir($path) ? "dir" : "file", |
| 2044 | ]; |
| 2045 | } |
| 2046 | usort( |
| 2047 | $root_entry["mu_plugins"], |
| 2048 | function ($a, $b) { |
| 2049 | return strcmp($a["name"], $b["name"]); |
| 2050 | } |
| 2051 | ); |
| 2052 | } |
| 2053 | |
| 2054 | $themes_dir = $paths["themes_dir"] ?? null; |
| 2055 | if ($themes_dir !== null && is_dir($themes_dir) && is_readable($themes_dir)) { |
| 2056 | $entries = @scandir($themes_dir) ?: []; |
| 2057 | foreach ($entries as $entry) { |
| 2058 | if ($entry === "." || $entry === "..") { |
| 2059 | continue; |
| 2060 | } |
| 2061 | $path = $themes_dir . "/" . $entry; |
| 2062 | if (is_dir($path)) { |
| 2063 | $root_entry["themes"][] = $entry; |
| 2064 | } |
| 2065 | } |
| 2066 | sort($root_entry["themes"]); |
| 2067 | } |
| 2068 | |
| 2069 | $wp_content["roots"][] = $root_entry; |
| 2070 | } |
| 2071 | |
| 2072 | // -- Assemble and return the preflight response -- |
| 2073 | $ok = |
| 2074 | $preflight_error === null && |
| 2075 | $filesystem_ok && |
| 2076 | (!empty($db["credentials_found"]) ? !empty($db["connected"]) : false); |
| 2077 | $response = [ |
| 2078 | "ok" => $ok, |
| 2079 | "error" => $preflight_error, |
| 2080 | "timestamp" => time(), |
| 2081 | "protocol_version" => EXPORT_PROTOCOL_VERSION, |
| 2082 | "protocol_min_version" => EXPORT_MIN_IMPORT_VERSION, |
| 2083 | "wp_detect" => [ |
| 2084 | "found" => !empty($wp_detect["roots"]), |
| 2085 | "searched" => $wp_detect["searched"], |
| 2086 | "roots" => $wp_detect["roots"], |
| 2087 | "error" => |
| 2088 | !empty($wp_detect["roots"]) |
| 2089 | ? null |
| 2090 | : "wp-load.php or wp-config.php not found in parent directories", |
| 2091 | ], |
| 2092 | "php" => [ |
| 2093 | "version" => PHP_VERSION, |
| 2094 | "sapi" => php_sapi_name(), |
| 2095 | "timezone" => date_default_timezone_get(), |
| 2096 | "extensions" => $extensions, |
| 2097 | "extension_versions" => $extension_versions, |
| 2098 | ], |
| 2099 | "limits" => [ |
| 2100 | "ini_max_execution_time" => (int) ini_get("max_execution_time"), |
| 2101 | "ini_max_input_time" => (int) ini_get("max_input_time"), |
| 2102 | "ini_default_socket_timeout" => (int) ini_get("default_socket_timeout"), |
| 2103 | "max_input_vars" => (int) ini_get("max_input_vars"), |
| 2104 | "max_file_uploads" => (int) ini_get("max_file_uploads"), |
| 2105 | "post_max_size" => $post_max_size_raw !== false ? $post_max_size_raw : null, |
| 2106 | "post_max_bytes" => $post_max_bytes, |
| 2107 | "upload_max_filesize" => |
| 2108 | $upload_max_filesize_raw !== false ? $upload_max_filesize_raw : null, |
| 2109 | "upload_max_bytes" => $upload_max_bytes, |
| 2110 | "max_request_bytes" => $max_request_bytes, |
| 2111 | "output_buffering" => ini_get("output_buffering") ?: null, |
| 2112 | "zlib_output_compression" => |
| 2113 | ini_get("zlib.output_compression") ?: null, |
| 2114 | "disable_functions" => ini_get("disable_functions") ?: null, |
| 2115 | "allow_url_fopen" => ini_get("allow_url_fopen") ?: null, |
| 2116 | "open_basedir" => ini_get("open_basedir") ?: null, |
| 2117 | ], |
| 2118 | "memory" => [ |
| 2119 | "limit_raw" => $memory_limit_raw !== false ? $memory_limit_raw : null, |
| 2120 | "limit_bytes" => $memory_limit_bytes, |
| 2121 | "used_bytes" => $memory_used, |
| 2122 | "available_bytes" => $memory_available, |
| 2123 | ], |
| 2124 | "images" => [ |
| 2125 | "gd" => [ |
| 2126 | "available" => is_array($gd_info), |
| 2127 | "version" => $gd_version, |
| 2128 | "formats" => $gd_formats, |
| 2129 | ], |
| 2130 | "imagick" => [ |
| 2131 | "available" => $imagick_version !== null, |
| 2132 | "version" => $imagick_version, |
| 2133 | ], |
| 2134 | ], |
| 2135 | "runtime" => [ |
| 2136 | "server_software" => $_SERVER["SERVER_SOFTWARE"] ?? null, |
| 2137 | // Every effective INI directive as computed by the PHP runtime |
| 2138 | // after merging php.ini, scanned .ini files, and htaccess |
| 2139 | // overrides. This captures the full configuration without |
| 2140 | // needing to read the .ini files themselves. |
| 2141 | "ini_get_all" => ini_get_all(null, false), |
| 2142 | "temp_dir" => sys_get_temp_dir(), |
| 2143 | "document_root" => $_SERVER["DOCUMENT_ROOT"] ?? null, |
| 2144 | "script_filename" => $_SERVER["SCRIPT_FILENAME"] ?? null, |
| 2145 | "cwd" => getcwd() ?: null, |
| 2146 | // Names of all defined environment variables (no values) so the |
| 2147 | // importer can use their presence as a webhost detection signal. |
| 2148 | "env_names" => array_values(array_unique(array_merge( |
| 2149 | array_keys($_ENV), |
| 2150 | array_keys(getenv()) |
| 2151 | ))), |
| 2152 | '$_SERVER_names' => array_keys($_SERVER), |
| 2153 | ], |
| 2154 | "filesystem" => [ |
| 2155 | "directories" => $dir_checks, |
| 2156 | "error" => $dir_error, |
| 2157 | "ok" => $filesystem_ok, |
| 2158 | ], |
| 2159 | "htaccess" => [ |
| 2160 | "files" => $htaccess_files, |
| 2161 | ], |
| 2162 | "wp_content" => $wp_content, |
| 2163 | "database" => $db, |
| 2164 | ]; |
| 2165 | |
| 2166 | header("Content-Type: application/json"); |
| 2167 | $json = json_encode($response); |
| 2168 | if ($json === false) { |
| 2169 | http_response_code(500); |
| 2170 | echo '{"error":"Failed to serialize preflight response: ' . json_last_error_msg() . '"}'; |
| 2171 | } else { |
| 2172 | echo $json; |
| 2173 | } |
| 2174 | |
| 2175 | return [ |
| 2176 | "status" => $response["ok"] ? "ok" : "error", |
| 2177 | "stats" => $response, |
| 2178 | ]; |
| 2179 | } |
| 2180 | |
| 2181 | /** |
| 2182 | * Streams file chunks from a producer as multipart/mixed. |
| 2183 | */ |
| 2184 | function stream_file_producer( |
| 2185 | $producer, |
| 2186 | ResourceBudget $budget, |
| 2187 | array $config = [], |
| 2188 | bool $gzip = false |
| 2189 | ): array { |
| 2190 | prepare_streaming_response(); |
| 2191 | |
| 2192 | ['gz' => $gz, 'boundary' => $boundary] = begin_multipart_stream(false, $gzip); |
| 2193 | |
| 2194 | // E2E test hook: after gzip stream initialization (file producer) |
| 2195 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 2196 | _e2e_load_test_hooks_if_needed($config); |
| 2197 | $hook_args = [$gz, $boundary]; |
| 2198 | _e2e_call_hook('test_hook_after_gzip_init', $hook_args); |
| 2199 | } |
| 2200 | |
| 2201 | $chunks_processed = 0; |
| 2202 | $files_completed = 0; |
| 2203 | $bytes_processed = 0; |
| 2204 | $last_progress_output = microtime(true); |
| 2205 | $metadata_sent = false; |
| 2206 | $iterations = 0; |
| 2207 | $aborted = false; |
| 2208 | $abort_payload = null; |
| 2209 | $last_cursor = ""; |
| 2210 | |
| 2211 | // -- Stream chunks from the producer -- |
| 2212 | // The producer yields file data, directories, symlinks, index entries, |
| 2213 | // and progress updates. Each chunk type is wrapped in a multipart part |
| 2214 | // with metadata headers (path, cursor, size, ctime). The loop runs |
| 2215 | // until the producer is exhausted or the resource budget runs out. |
| 2216 | try { |
| 2217 | $initial_progress = $producer->get_progress(); |
| 2218 | $initial_progress_json = json_encode_or_throw($initial_progress); |
| 2219 | $initial_cursor = $producer->get_reentrancy_cursor(); |
| 2220 | $last_cursor = $initial_cursor; |
| 2221 | $gz->write( |
| 2222 | "--{$boundary}\r\n" . |
| 2223 | "Content-Type: application/json\r\n" . |
| 2224 | "Content-Length: " . strlen($initial_progress_json) . "\r\n" . |
| 2225 | "X-Chunk-Type: progress\r\n" . |
| 2226 | "X-Cursor: " . base64_encode($initial_cursor) . "\r\n" . |
| 2227 | "\r\n" . |
| 2228 | $initial_progress_json . "\r\n" |
| 2229 | ); |
| 2230 | $gz->sync(); |
| 2231 | while (true) { |
| 2232 | if ( |
| 2233 | !$budget->has_remaining() |
| 2234 | ) { |
| 2235 | break; |
| 2236 | } |
| 2237 | |
| 2238 | if (!$producer->next_chunk()) { |
| 2239 | break; |
| 2240 | } |
| 2241 | |
| 2242 | $iterations++; |
| 2243 | $chunk = $producer->get_current_chunk(); |
| 2244 | $progress = $producer->get_progress(); |
| 2245 | |
| 2246 | if (!$metadata_sent && $progress["phase"] === "streaming") { |
| 2247 | $filesystem_root = $producer->get_filesystem_root(); |
| 2248 | $metadata = [ |
| 2249 | "filesystem_root" => base64_encode($filesystem_root ?? ""), |
| 2250 | ]; |
| 2251 | $metadata_json = json_encode_or_throw($metadata); |
| 2252 | |
| 2253 | $gz->write( |
| 2254 | "--{$boundary}\r\n" . |
| 2255 | "Content-Type: application/json\r\n" . |
| 2256 | "Content-Length: " . strlen($metadata_json) . "\r\n" . |
| 2257 | "X-Chunk-Type: metadata\r\n" . |
| 2258 | "X-Filesystem-Root: " . base64_encode($filesystem_root ?? "") . "\r\n" . |
| 2259 | "\r\n" . |
| 2260 | $metadata_json . "\r\n" |
| 2261 | ); |
| 2262 | $gz->sync(); |
| 2263 | |
| 2264 | $metadata_sent = true; |
| 2265 | } |
| 2266 | |
| 2267 | if ($chunk === null) { |
| 2268 | $now = microtime(true); |
| 2269 | if ($iterations === 1 || $now - $last_progress_output >= 3.0) { |
| 2270 | $progress_json = json_encode_or_throw($progress); |
| 2271 | $cursor = $producer->get_reentrancy_cursor(); |
| 2272 | $last_cursor = $cursor; |
| 2273 | |
| 2274 | $gz->write( |
| 2275 | "--{$boundary}\r\n" . |
| 2276 | "Content-Type: application/json\r\n" . |
| 2277 | "Content-Length: " . strlen($progress_json) . "\r\n" . |
| 2278 | "X-Chunk-Type: progress\r\n" . |
| 2279 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2280 | "\r\n" . |
| 2281 | $progress_json . "\r\n" |
| 2282 | ); |
| 2283 | $gz->sync(); |
| 2284 | |
| 2285 | $last_progress_output = $now; |
| 2286 | } |
| 2287 | |
| 2288 | continue; |
| 2289 | } |
| 2290 | |
| 2291 | $chunk_type = $chunk["type"] ?? "file"; |
| 2292 | $cursor = $producer->get_reentrancy_cursor(); |
| 2293 | $last_cursor = $cursor; |
| 2294 | |
| 2295 | if ($chunk_type === "directory") { |
| 2296 | $part = |
| 2297 | "--{$boundary}\r\n" . |
| 2298 | "Content-Type: application/octet-stream\r\n" . |
| 2299 | "Content-Length: 0\r\n" . |
| 2300 | "X-Chunk-Type: directory\r\n" . |
| 2301 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2302 | "X-Directory-Path: " . base64_encode($chunk["path"]) . "\r\n"; |
| 2303 | if (isset($chunk["ctime"])) { |
| 2304 | $part .= "X-Directory-Ctime: " . $chunk["ctime"] . "\r\n"; |
| 2305 | } |
| 2306 | $gz->write($part . "\r\n\r\n"); |
| 2307 | $gz->sync(); |
| 2308 | } elseif ($chunk_type === "symlink") { |
| 2309 | $gz->write( |
| 2310 | "--{$boundary}\r\n" . |
| 2311 | "Content-Type: application/octet-stream\r\n" . |
| 2312 | "Content-Length: 0\r\n" . |
| 2313 | "X-Chunk-Type: symlink\r\n" . |
| 2314 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2315 | "X-Symlink-Path: " . base64_encode($chunk["path"]) . "\r\n" . |
| 2316 | "X-Symlink-Target: " . base64_encode($chunk["target"]) . "\r\n" . |
| 2317 | "X-Symlink-Ctime: " . $chunk["ctime"] . "\r\n" . |
| 2318 | "\r\n\r\n" |
| 2319 | ); |
| 2320 | $gz->sync(); |
| 2321 | } elseif ($chunk_type === "index") { |
| 2322 | $gz->write( |
| 2323 | "--{$boundary}\r\n" . |
| 2324 | "Content-Type: application/octet-stream\r\n" . |
| 2325 | "Content-Length: 0\r\n" . |
| 2326 | "X-Chunk-Type: index\r\n" . |
| 2327 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2328 | "X-Index-Path: " . base64_encode($chunk["path"]) . "\r\n" . |
| 2329 | "X-File-Ctime: " . $chunk["ctime"] . "\r\n" . |
| 2330 | "X-File-Size: " . $chunk["size"] . "\r\n" . |
| 2331 | "\r\n\r\n" |
| 2332 | ); |
| 2333 | $gz->sync(); |
| 2334 | } elseif ($chunk_type === "missing") { |
| 2335 | $gz->write( |
| 2336 | "--{$boundary}\r\n" . |
| 2337 | "Content-Type: application/octet-stream\r\n" . |
| 2338 | "Content-Length: 0\r\n" . |
| 2339 | "X-Chunk-Type: missing\r\n" . |
| 2340 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2341 | "X-File-Path: " . base64_encode($chunk["path"]) . "\r\n" . |
| 2342 | "\r\n\r\n" |
| 2343 | ); |
| 2344 | $gz->sync(); |
| 2345 | } elseif ($chunk_type === "error") { |
| 2346 | $payload = [ |
| 2347 | "error_type" => $chunk["error_type"] ?? "unknown", |
| 2348 | "path" => base64_encode($chunk["path"] ?? ""), |
| 2349 | "message" => $chunk["message"] ?? "Error", |
| 2350 | ]; |
| 2351 | if (isset($chunk["expected_ctime"])) { |
| 2352 | $payload["expected_ctime"] = $chunk["expected_ctime"]; |
| 2353 | } |
| 2354 | if (isset($chunk["actual_ctime"])) { |
| 2355 | $payload["actual_ctime"] = $chunk["actual_ctime"]; |
| 2356 | } |
| 2357 | $json = json_encode_or_throw($payload); |
| 2358 | $gz->write( |
| 2359 | "--{$boundary}\r\n" . |
| 2360 | "Content-Type: application/json\r\n" . |
| 2361 | "Content-Length: " . strlen($json) . "\r\n" . |
| 2362 | "X-Chunk-Type: error\r\n" . |
| 2363 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2364 | "\r\n" . |
| 2365 | $json . "\r\n" |
| 2366 | ); |
| 2367 | $gz->sync(); |
| 2368 | } else { |
| 2369 | // E2E test hook: before file chunk is emitted |
| 2370 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 2371 | $hook_data = $chunk["data"]; |
| 2372 | $hook_args = [$chunk["path"], $chunk["offset"], &$hook_data]; |
| 2373 | _e2e_call_hook('test_hook_before_file_chunk', $hook_args); |
| 2374 | $chunk["data"] = $hook_data; |
| 2375 | } |
| 2376 | |
| 2377 | $chunks_processed++; |
| 2378 | $bytes_processed += strlen($chunk["data"]); |
| 2379 | if ($chunk["is_first_chunk"]) { |
| 2380 | $files_completed++; |
| 2381 | } |
| 2382 | |
| 2383 | $data = $chunk["data"]; |
| 2384 | |
| 2385 | $headers = |
| 2386 | "--{$boundary}\r\n" . |
| 2387 | "Content-Type: application/octet-stream\r\n" . |
| 2388 | "Content-Length: " . strlen($data) . "\r\n" . |
| 2389 | "X-Chunk-Type: file\r\n" . |
| 2390 | "X-Cursor: " . base64_encode($cursor) . "\r\n" . |
| 2391 | "X-File-Path: " . base64_encode($chunk["path"]) . "\r\n" . |
| 2392 | "X-File-Size: " . $chunk["size"] . "\r\n" . |
| 2393 | "X-File-Ctime: " . $chunk["ctime"] . "\r\n" . |
| 2394 | "X-Chunk-Offset: " . $chunk["offset"] . "\r\n" . |
| 2395 | "X-Chunk-Size: " . strlen($data) . "\r\n" . |
| 2396 | "X-First-Chunk: " . ($chunk["is_first_chunk"] ? "1" : "0") . "\r\n" . |
| 2397 | "X-Last-Chunk: " . ($chunk["is_last_chunk"] ? "1" : "0") . "\r\n"; |
| 2398 | if (!empty($chunk["file_changed"])) { |
| 2399 | $headers .= "X-File-Changed: 1\r\n"; |
| 2400 | if ($chunk["change_ctime"] !== null) { |
| 2401 | $headers .= "X-File-Change-Ctime: " . $chunk["change_ctime"] . "\r\n"; |
| 2402 | } |
| 2403 | if ($chunk["change_size"] !== null) { |
| 2404 | $headers .= "X-File-Change-Size: " . $chunk["change_size"] . "\r\n"; |
| 2405 | } |
| 2406 | } |
| 2407 | $gz->write($headers . "\r\n"); |
| 2408 | $gz->write($data); |
| 2409 | $gz->write("\r\n"); |
| 2410 | $gz->sync(); |
| 2411 | } |
| 2412 | } |
| 2413 | } catch (Throwable $e) { |
| 2414 | $aborted = true; |
| 2415 | $abort_payload = [ |
| 2416 | "error_type" => "exception", |
| 2417 | "path" => "", |
| 2418 | "message" => $e->getMessage(), |
| 2419 | ]; |
| 2420 | } |
| 2421 | |
| 2422 | // Best-effort error and completion chunks — the client already has the |
| 2423 | // data chunks. If the stream is broken at this point, log and move on. |
| 2424 | try { |
| 2425 | // @TODO: If an exception is thrown right after the previous chunk header, |
| 2426 | // it read the fixed Content-Length value and will consume this next |
| 2427 | // chunk as data. We should try and backfill the output up to the |
| 2428 | // previous content-length value if possible. |
| 2429 | if ($abort_payload !== null) { |
| 2430 | $json = json_encode_or_throw($abort_payload); |
| 2431 | $gz->write( |
| 2432 | "--{$boundary}\r\n" . |
| 2433 | "Content-Type: application/json\r\n" . |
| 2434 | "Content-Length: " . strlen($json) . "\r\n" . |
| 2435 | "X-Chunk-Type: error\r\n" . |
| 2436 | "X-Cursor: " . base64_encode($last_cursor) . "\r\n" . |
| 2437 | "\r\n" . |
| 2438 | $json . "\r\n" |
| 2439 | ); |
| 2440 | $gz->sync(); |
| 2441 | } |
| 2442 | |
| 2443 | $progress = $producer->get_progress(); |
| 2444 | $is_complete = $progress["phase"] === "finished" && !$aborted; |
| 2445 | $status = $is_complete ? "complete" : "partial"; |
| 2446 | |
| 2447 | // E2E test hook: before completion chunk (file producer) |
| 2448 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 2449 | $hook_args = [$status, $gz, $boundary]; |
| 2450 | _e2e_call_hook('test_hook_before_completion', $hook_args); |
| 2451 | } |
| 2452 | |
| 2453 | error_log( |
| 2454 | "Export completion: status={$status}, phase={$progress["phase"]}, " . |
| 2455 | "chunks={$chunks_processed}, files={$files_completed}, bytes={$bytes_processed}" |
| 2456 | ); |
| 2457 | |
| 2458 | $gz->write( |
| 2459 | "--{$boundary}\r\n" . |
| 2460 | "Content-Type: application/octet-stream\r\n" . |
| 2461 | "Content-Length: 0\r\n" . |
| 2462 | "X-Chunk-Type: completion\r\n" . |
| 2463 | "X-Status: {$status}\r\n" . |
| 2464 | "X-Chunks-Processed: {$chunks_processed}\r\n" . |
| 2465 | "X-Files-Completed: {$files_completed}\r\n" . |
| 2466 | "X-Bytes-Processed: {$bytes_processed}\r\n" . |
| 2467 | "X-Memory-Used: " . memory_get_peak_usage(true) . "\r\n" . |
| 2468 | "X-Memory-Limit: " . $budget->max_memory . "\r\n" . |
| 2469 | "X-Time-Elapsed: " . (microtime(true) - $budget->start_time) . "\r\n" . |
| 2470 | "\r\n" . |
| 2471 | "\r\n" . |
| 2472 | "--{$boundary}--\r\n" |
| 2473 | ); |
| 2474 | $gz->finish(); |
| 2475 | } catch (\Throwable $e) { |
| 2476 | error_log("Export: failed to write completion chunk: " . $e->getMessage()); |
| 2477 | } |
| 2478 | |
| 2479 | $status = $aborted ? "partial" : ($status ?? "partial"); |
| 2480 | |
| 2481 | return [ |
| 2482 | "status" => $status, |
| 2483 | "stats" => [ |
| 2484 | "chunks_processed" => $chunks_processed, |
| 2485 | "files_completed" => $files_completed, |
| 2486 | "bytes_processed" => $bytes_processed, |
| 2487 | "memory_used" => memory_get_peak_usage(true), |
| 2488 | "time_elapsed" => microtime(true) - $budget->start_time, |
| 2489 | ], |
| 2490 | ]; |
| 2491 | } |
| 2492 | |
| 2493 | /** |
| 2494 | * Encodes a file_index stack for JSON serialization. |
| 2495 | * |
| 2496 | * Paths may contain non-UTF8 bytes, so dir and after are base64-encoded. |
| 2497 | */ |
| 2498 | function encode_index_stack(array $stack): array |
| 2499 | { |
| 2500 | $encoded = []; |
| 2501 | foreach ($stack as $frame) { |
| 2502 | $encoded[] = [ |
| 2503 | "dir" => base64_encode($frame["dir"]), |
| 2504 | "after" => $frame["after"] !== null ? base64_encode($frame["after"]) : null, |
| 2505 | ]; |
| 2506 | } |
| 2507 | return $encoded; |
| 2508 | } |
| 2509 | |
| 2510 | /** |
| 2511 | * Resolve "." and ".." segments in a path without resolving symlinks. |
| 2512 | * |
| 2513 | * Unlike realpath(), this only performs textual normalization — it collapses |
| 2514 | * "." and ".." but leaves symlink components intact. This is useful when |
| 2515 | * you need a clean absolute path to inspect which components are symlinks. |
| 2516 | * |
| 2517 | * @param string $path An absolute path that may contain "." or ".." segments. |
| 2518 | * @return string The normalized absolute path. |
| 2519 | */ |
| 2520 | function normalize_dot_segments(string $path): string |
| 2521 | { |
| 2522 | $parts = explode("/", $path); |
| 2523 | $normalized = []; |
| 2524 | foreach ($parts as $p) { |
| 2525 | if ($p === "" || $p === ".") { |
| 2526 | if (empty($normalized)) { |
| 2527 | $normalized[] = ""; |
| 2528 | } |
| 2529 | continue; |
| 2530 | } |
| 2531 | if ($p === "..") { |
| 2532 | if (count($normalized) > 1) { |
| 2533 | array_pop($normalized); |
| 2534 | } |
| 2535 | continue; |
| 2536 | } |
| 2537 | $normalized[] = $p; |
| 2538 | } |
| 2539 | return implode("/", $normalized); |
| 2540 | } |
| 2541 | |
| 2542 | /** |
| 2543 | * Given a path, such as `/srv/wordpress/wp-content/plugins/akismet/assets`, returns |
| 2544 | * a list of all the parent paths that are symlinks. It will check `/srv`, |
| 2545 | * `/srv/wordpress`, `/srv/wordpress/wp-content`, etc. |
| 2546 | * |
| 2547 | * For example, given the following filesystem layout: |
| 2548 | * |
| 2549 | * /srv/wordpress/wp-content -> /htdocs/wp-content |
| 2550 | * /srv/wordpress/wp-content/plugins/akismet -> /wordpress/plugins/akismet/latest |
| 2551 | * /wordpress/plugins/akismet/latest -> /wordpress/plugins/akismet/5.0.5 |
| 2552 | * |
| 2553 | * Calling |
| 2554 | * |
| 2555 | * find_parents_symlinks("/srv/wordpress/wp-content/plugins/akismet/assets") |
| 2556 | * |
| 2557 | * will return the following symlinks: |
| 2558 | * |
| 2559 | * ['path' => '/srv/wordpress/wp-content', 'target' => '/htdocs/wp-content'] |
| 2560 | * ['path' => '/htdocs/wp-content/plugins/akismet', 'target' => '/wordpress/plugins/akismet/latest'] |
| 2561 | * |
| 2562 | * Note: |
| 2563 | * |
| 2564 | * * Every found `path` is a resolved realpath(), which means that all the parents are |
| 2565 | * regular directories, not symlinks. |
| 2566 | * * It is intentionally not recursive. That last `akismet/latest` -> `akismet/5.0.5` |
| 2567 | * symlink was not returned. The client is free to recursively request the files from |
| 2568 | * any additional directories outside of the initial content root based on the parent |
| 2569 | * symlinks resolved by this function. |
| 2570 | * |
| 2571 | * @param string $absolute_path An absolute path to a file or directory. |
| 2572 | * @return array An array of symlinks found in the path. |
| 2573 | * Each array element is an associative array with the following keys: |
| 2574 | * - "path": The path to the symlink. |
| 2575 | * - "ctime": The creation time of the symlink. |
| 2576 | * - "size": The size of the symlink. |
| 2577 | * - "type": The type of the symlink. |
| 2578 | * - "target": The target of the symlink. |
| 2579 | * - "intermediate": Whether the symlink is an intermediate symlink. |
| 2580 | */ |
| 2581 | function find_parents_symlinks(string $absolute_path): array |
| 2582 | { |
| 2583 | $entries = []; |
| 2584 | $parts = explode('/', $absolute_path); |
| 2585 | $current = ""; |
| 2586 | // Walk through /srv, /srv/wordpress, /srv/wordpress/wp-content, etc. |
| 2587 | foreach ($parts as $part) { |
| 2588 | if ($part === "") { |
| 2589 | $current = "/"; |
| 2590 | continue; |
| 2591 | } |
| 2592 | $current = rtrim($current, "/") . "/" . $part; |
| 2593 | // If the path up to this point is not a symlink, we can just |
| 2594 | // expand to the next path segment. |
| 2595 | if (!@is_link($current)) { |
| 2596 | continue; |
| 2597 | } |
| 2598 | |
| 2599 | // If we're looking at a valid symlink, record it. |
| 2600 | $target = @readlink($current); |
| 2601 | if ($target !== false && $target !== "") { |
| 2602 | $stat = @lstat($current); |
| 2603 | $entries[] = [ |
| 2604 | "path" => $current, |
| 2605 | "ctime" => (int) ($stat["ctime"] ?? 0), |
| 2606 | "size" => 0, |
| 2607 | "type" => "link", |
| 2608 | "target" => $target, |
| 2609 | "intermediate" => true, |
| 2610 | ]; |
| 2611 | } |
| 2612 | // Swap the current path for the resolved realpath(). |
| 2613 | // e.g. if $current is a symlink at /srv/wordpress/wp-content pointing |
| 2614 | // to /htdocs/wp-content, then from now on we'll use /htdocs/wp-content |
| 2615 | // as our $current and append the next path segments to it. |
| 2616 | $real = @realpath($current); |
| 2617 | if ($real !== false) { |
| 2618 | $current = $real; |
| 2619 | } |
| 2620 | } |
| 2621 | return $entries; |
| 2622 | } |
| 2623 | |
| 2624 | /** |
| 2625 | * Resolves a symlink's target to a canonical path for the file index. |
| 2626 | * |
| 2627 | * On many WordPress hosts (wp.com, SiteGround, etc.), the filesystem |
| 2628 | * contains chains of symlinks. For example, /srv might point to /, |
| 2629 | * /srv/wordpress might point to /wordpress, and readlink() returns |
| 2630 | * relative paths like "../wordpress/core/latest" that still contain |
| 2631 | * intermediate symlinks. realpath() cuts through all of this and |
| 2632 | * returns the final canonical path — e.g. /htdocs instead of /srv/htdocs. |
| 2633 | * |
| 2634 | * The client uses symlink targets to discover additional directories to |
| 2635 | * index, so only directory symlinks get a resolved target. File symlink |
| 2636 | * targets are ignored because the client doesn't need to recurse into them. |
| 2637 | * |
| 2638 | * Also walks the raw readlink() path to find intermediate symlinks that |
| 2639 | * realpath() skips. For example, if readlink() returns a relative path |
| 2640 | * like "../../../wordpress/plugins/akismet/latest", the absolute form |
| 2641 | * might be /srv/wordpress/plugins/akismet/latest — and /srv/wordpress is |
| 2642 | * itself a symlink to /wordpress. realpath() jumps straight to |
| 2643 | * /wordpress/..., so we'd never record the /srv/wordpress intermediate. |
| 2644 | * find_parents_symlinks() catches those. |
| 2645 | * |
| 2646 | * @param string $path Absolute path to the symlink. |
| 2647 | * @return array{target: string|null, intermediates: array} The resolved |
| 2648 | * canonical target (null for file symlinks or unresolvable |
| 2649 | * paths), and any intermediate symlink entries found. |
| 2650 | */ |
| 2651 | function resolve_symlink_target(string $path): array |
| 2652 | { |
| 2653 | clearstatcache(true, $path); |
| 2654 | $resolved_target = @realpath($path); |
| 2655 | |
| 2656 | // Only directory symlinks matter — the client uses targets to discover |
| 2657 | // additional directories to index. Also skip unresolvable symlinks |
| 2658 | // and self-referencing paths. |
| 2659 | if ( |
| 2660 | $resolved_target === false || |
| 2661 | $resolved_target === $path || |
| 2662 | !is_dir($resolved_target) |
| 2663 | ) { |
| 2664 | return ['target' => null, 'intermediates' => []]; |
| 2665 | } |
| 2666 | |
| 2667 | $intermediates = []; |
| 2668 | $raw_target = @readlink($path); |
| 2669 | if ($raw_target !== false && $raw_target !== "") { |
| 2670 | if ($raw_target[0] !== "/") { |
| 2671 | $raw_target = dirname($path) . "/" . $raw_target; |
| 2672 | } |
| 2673 | $abs_raw = normalize_dot_segments($raw_target); |
| 2674 | if ($abs_raw !== "" && $abs_raw[0] === "/" && $abs_raw !== $resolved_target) { |
| 2675 | $intermediates = find_parents_symlinks($abs_raw); |
| 2676 | } |
| 2677 | } |
| 2678 | |
| 2679 | return ['target' => $resolved_target, 'intermediates' => $intermediates]; |
| 2680 | } |
| 2681 | |
| 2682 | /** |
| 2683 | * Encodes batch items for JSON serialization, base64-encoding paths |
| 2684 | * to handle non-UTF8 filesystem bytes. |
| 2685 | */ |
| 2686 | function encode_index_batch(array $batch_items): array |
| 2687 | { |
| 2688 | $encoded = []; |
| 2689 | foreach ($batch_items as $item) { |
| 2690 | $entry = [ |
| 2691 | "path" => base64_encode($item["path"]), |
| 2692 | "ctime" => $item["ctime"], |
| 2693 | "size" => $item["size"], |
| 2694 | "type" => $item["type"], |
| 2695 | ]; |
| 2696 | if (isset($item["target"])) { |
| 2697 | $entry["target"] = base64_encode($item["target"]); |
| 2698 | } |
| 2699 | if (!empty($item["intermediate"])) { |
| 2700 | $entry["intermediate"] = true; |
| 2701 | } |
| 2702 | $encoded[] = $entry; |
| 2703 | } |
| 2704 | return $encoded; |
| 2705 | } |
| 2706 | |
| 2707 | /** |
| 2708 | * Streams a directory index as gzipped JSON batches of {path, ctime, size, type}. |
| 2709 | * |
| 2710 | * The client supplies list_dir and drives traversal depth-first by |
| 2711 | * enqueuing directories as they are discovered. Resumption is supported |
| 2712 | * via cursor containing the directory stack and last-seen entry. |
| 2713 | */ |
| 2714 | function endpoint_file_index( |
| 2715 | array $config, |
| 2716 | ResourceBudget $budget |
| 2717 | ): array { |
| 2718 | // This endpoint may run repeatedly in the same PHP process (e.g. PHP built-in |
| 2719 | // server, long-lived workers). Clear stale stat/realpath cache from previous |
| 2720 | // requests so path type transitions (symlink/file/dir) are seen correctly. |
| 2721 | clearstatcache(true); |
| 2722 | |
| 2723 | $directories = resolve_directories($config); |
| 2724 | $batch_size = $config["batch_size"] ?? 5000; |
| 2725 | $batch_size = require_int_range( |
| 2726 | "batch_size", |
| 2727 | (int) $batch_size, |
| 2728 | 100, |
| 2729 | 100000 |
| 2730 | ); |
| 2731 | |
| 2732 | $list_dir = $config["list_dir"] ?? null; |
| 2733 | $list_dir_real = null; |
| 2734 | $stack = []; |
| 2735 | $ordered = []; |
| 2736 | $follow_symlinks = !empty($config["follow_symlinks"]); |
| 2737 | $cursor_provided = isset($config["cursor"]); |
| 2738 | // Default-skip generated caches, VCS metadata, OS junk, and editor |
| 2739 | // scratch files unless the client explicitly opts in. See |
| 2740 | // path_is_default_skipped() for the full deny-list and rationale. |
| 2741 | $include_caches = !empty($config["include_caches"]); |
| 2742 | // Reprint's own storage (the push staging area and apply bookkeeping) |
| 2743 | // must never appear in an index: it can sit inside the document root on |
| 2744 | // hosts that allow writing nowhere else, and indexing it would sync or |
| 2745 | // delete reprint's own records mid-transfer. storage_path is the |
| 2746 | // server's own setting, so it is known here for every request — a |
| 2747 | // pulling peer's request does not need to mention it. |
| 2748 | $storage_path = isset($config["storage_path"]) && is_string($config["storage_path"]) |
| 2749 | ? rtrim($config["storage_path"], "/") |
| 2750 | : ""; |
| 2751 | if ($storage_path !== "") { |
| 2752 | // The traversal canonicalizes every path with realpath() (see the |
| 2753 | // wp.com note further down), so the setting must be compared in the |
| 2754 | // same form. path_is_within_root() compares plain strings, and a |
| 2755 | // trailing slash on the setting would make it miss — hence the |
| 2756 | // rtrim above. When the directory does not exist yet there is |
| 2757 | // nothing to exclude and the trimmed value stays as a harmless |
| 2758 | // fallback. |
| 2759 | $storage_real = realpath($storage_path); |
| 2760 | if ($storage_real !== false) { |
| 2761 | $storage_path = $storage_real; |
| 2762 | } |
| 2763 | } |
| 2764 | |
| 2765 | // Find the starting point – either by parsing the cursor, or by |
| 2766 | // sourcing it from the filesystem. |
| 2767 | |
| 2768 | // -- Restore or initialize the directory traversal stack -- |
| 2769 | // On resumption, the cursor encodes the stack of directories and the |
| 2770 | // last-processed entry in each. On first request, build the stack from |
| 2771 | // the list_dir and any extra allowed roots. |
| 2772 | if ($cursor_provided) { |
| 2773 | $cursor_data = json_decode($config["cursor"], true); |
| 2774 | if (!is_array($cursor_data)) { |
| 2775 | throw new InvalidArgumentException("Invalid index cursor format"); |
| 2776 | } |
| 2777 | if (!isset($cursor_data["stack"]) || !is_array($cursor_data["stack"])) { |
| 2778 | throw new InvalidArgumentException("Index cursor missing stack"); |
| 2779 | } |
| 2780 | foreach ($cursor_data["stack"] as $frame) { |
| 2781 | if (!is_array($frame)) { |
| 2782 | throw new InvalidArgumentException("Invalid index cursor frame"); |
| 2783 | } |
| 2784 | $dir_encoded = $frame["dir"] ?? null; |
| 2785 | if (!is_string($dir_encoded) || $dir_encoded === "") { |
| 2786 | throw new InvalidArgumentException("Index cursor frame missing dir"); |
| 2787 | } |
| 2788 | $dir = base64_decode($dir_encoded, true); |
| 2789 | if ($dir === false || $dir === "") { |
| 2790 | throw new InvalidArgumentException("Index cursor frame has invalid dir encoding"); |
| 2791 | } |
| 2792 | $after_encoded = $frame["after"] ?? null; |
| 2793 | if ($after_encoded !== null && !is_string($after_encoded)) { |
| 2794 | throw new InvalidArgumentException("Index cursor frame invalid after"); |
| 2795 | } |
| 2796 | $after = null; |
| 2797 | if ($after_encoded !== null) { |
| 2798 | $after = base64_decode($after_encoded, true); |
| 2799 | if ($after === false) { |
| 2800 | throw new InvalidArgumentException("Index cursor frame has invalid after encoding"); |
| 2801 | } |
| 2802 | } |
| 2803 | $stack[] = [ |
| 2804 | "dir" => $dir, |
| 2805 | "after" => $after, |
| 2806 | ]; |
| 2807 | } |
| 2808 | } else { |
| 2809 | if (!$list_dir) { |
| 2810 | throw new InvalidArgumentException("list_dir is required for file_index"); |
| 2811 | } |
| 2812 | |
| 2813 | clearstatcache(true, $list_dir); |
| 2814 | $list_dir_real = realpath($list_dir); |
| 2815 | if ($list_dir_real === false || !is_dir($list_dir_real)) { |
| 2816 | throw new InvalidArgumentException( |
| 2817 | "list_dir does not exist or is not accessible: {$list_dir}" |
| 2818 | ); |
| 2819 | } |
| 2820 | |
| 2821 | $allowed = false; |
| 2822 | foreach ($directories as $root) { |
| 2823 | if ( |
| 2824 | $list_dir_real === $root || |
| 2825 | str_starts_with($list_dir_real, $root . "/") |
| 2826 | ) { |
| 2827 | $allowed = true; |
| 2828 | break; |
| 2829 | } |
| 2830 | } |
| 2831 | // When follow_symlinks is enabled, allow any directory that the |
| 2832 | // authenticated client requests. The client is already authenticated |
| 2833 | // via HMAC, so there is no untrusted-input risk. |
| 2834 | if (!$allowed && !$follow_symlinks) { |
| 2835 | throw new InvalidArgumentException( |
| 2836 | "list_dir is outside of allowed roots: {$list_dir_real}" |
| 2837 | ); |
| 2838 | } |
| 2839 | |
| 2840 | $ordered = [$list_dir_real]; |
| 2841 | $extra_roots = []; |
| 2842 | foreach ($directories as $root) { |
| 2843 | if ($root === $list_dir_real) { |
| 2844 | continue; |
| 2845 | } |
| 2846 | $extra_roots[] = $root; |
| 2847 | } |
| 2848 | if (!empty($extra_roots)) { |
| 2849 | sort($extra_roots, SORT_STRING); |
| 2850 | foreach ($extra_roots as $root) { |
| 2851 | // Skip exact duplicates of already-ordered roots. |
| 2852 | // Do NOT skip parent roots — on hosts like wp.com Atomic |
| 2853 | // the document root (/srv/htdocs) is a parent of the |
| 2854 | // primary root (/srv/htdocs/__wp__) but contains a separate |
| 2855 | // wp-content with the site's actual plugins and themes. |
| 2856 | // The during-traversal dedup in the main loop already |
| 2857 | // prevents re-entering child roots (i.e. when traversing |
| 2858 | // /srv/htdocs we won't descend back into __wp__/). |
| 2859 | if (in_array($root, $ordered, true)) { |
| 2860 | continue; |
| 2861 | } |
| 2862 | $ordered[] = $root; |
| 2863 | } |
| 2864 | } |
| 2865 | |
| 2866 | for ($i = count($ordered) - 1; $i >= 0; $i--) { |
| 2867 | $stack[] = [ |
| 2868 | "dir" => $ordered[$i], |
| 2869 | "after" => null, |
| 2870 | ]; |
| 2871 | } |
| 2872 | } |
| 2873 | |
| 2874 | if ($list_dir_real === null) { |
| 2875 | if (!empty($stack)) { |
| 2876 | $list_dir_real = $stack[count($stack) - 1]["dir"]; |
| 2877 | } else { |
| 2878 | $list_dir_real = $directories[0] ?? "/"; |
| 2879 | } |
| 2880 | } |
| 2881 | |
| 2882 | prepare_streaming_response(); |
| 2883 | |
| 2884 | ['gz' => $gz, 'boundary' => $boundary] = begin_multipart_stream(); |
| 2885 | |
| 2886 | $filesystem_root = $directories[0] ?? "/"; |
| 2887 | $batches_emitted = 0; |
| 2888 | $total_entries = 0; |
| 2889 | $batch_items = []; |
| 2890 | $status = "partial"; |
| 2891 | $aborted = false; |
| 2892 | $abort_payload = null; |
| 2893 | |
| 2894 | // -- Pre-scan: discover intermediate symlinks -- |
| 2895 | // When following symlinks, discover intermediate symlinks along each |
| 2896 | // directory path being traversed. For example, if list_dir is |
| 2897 | // /srv/wordpress/plugins/akismet/latest and /srv/wordpress is itself |
| 2898 | // a symlink to /wordpress, emit that intermediate symlink so the |
| 2899 | // client can recreate the full chain locally. |
| 2900 | if (!$cursor_provided && $follow_symlinks) { |
| 2901 | foreach ($ordered as $dir) { |
| 2902 | $path_symlinks = find_parents_symlinks($dir); |
| 2903 | foreach ($path_symlinks as $entry) { |
| 2904 | $batch_items[] = $entry; |
| 2905 | } |
| 2906 | } |
| 2907 | } |
| 2908 | |
| 2909 | // -- Depth-first directory traversal -- |
| 2910 | // Walk the directory tree using the stack. Each directory's entries are |
| 2911 | // read with scandir (sorted ascending), yielding files, symlinks, and |
| 2912 | // subdirectories. Subdirectories push new frames onto the stack. Entries |
| 2913 | // are batched into JSON index_batch chunks and streamed to the client. |
| 2914 | |
| 2915 | $current_dir = $list_dir_real; |
| 2916 | |
| 2917 | try { |
| 2918 | $metadata = [ |
| 2919 | "filesystem_root" => base64_encode($filesystem_root), |
| 2920 | "list_dir" => base64_encode($list_dir_real), |
| 2921 | ]; |
| 2922 | $metadata_json = json_encode_or_throw($metadata); |
| 2923 | |
| 2924 | $gz->write( |
| 2925 | "--{$boundary}\r\n" . |
| 2926 | "Content-Type: application/json\r\n" . |
| 2927 | "Content-Length: " . strlen($metadata_json) . "\r\n" . |
| 2928 | "X-Chunk-Type: metadata\r\n" . |
| 2929 | "X-Filesystem-Root: " . base64_encode($filesystem_root ?? "") . "\r\n" . |
| 2930 | "X-Index-Dir: " . base64_encode($list_dir_real ?? "") . "\r\n" . |
| 2931 | "\r\n" . |
| 2932 | $metadata_json . "\r\n" |
| 2933 | ); |
| 2934 | $gz->sync(); |
| 2935 | $stop = false; |
| 2936 | |
| 2937 | while (!$stop) { |
| 2938 | if (empty($stack)) { |
| 2939 | $status = "complete"; |
| 2940 | break; |
| 2941 | } |
| 2942 | |
| 2943 | $frame_index = count($stack) - 1; |
| 2944 | $frame = $stack[$frame_index]; |
| 2945 | $current_dir = $frame["dir"]; |
| 2946 | $current_after = $frame["after"] ?? null; |
| 2947 | |
| 2948 | clearstatcache(true, $current_dir); |
| 2949 | $current_real = realpath($current_dir); |
| 2950 | if ($current_real === false || !is_dir($current_real)) { |
| 2951 | $abort_payload = [ |
| 2952 | "error_type" => "dir_open", |
| 2953 | "path" => base64_encode($current_dir), |
| 2954 | "message" => "Directory does not exist or is not accessible", |
| 2955 | ]; |
| 2956 | array_pop($stack); |
| 2957 | $json = json_encode_or_throw($abort_payload); |
| 2958 | $cursor_json = json_encode_or_throw( |
| 2959 | ["stack" => encode_index_stack($stack)], |
| 2960 | JSON_UNESCAPED_SLASHES |
| 2961 | ); |
| 2962 | $cursor_b64 = base64_encode($cursor_json); |
| 2963 | $gz->write( |
| 2964 | "--{$boundary}\r\n" . |
| 2965 | "Content-Type: application/json\r\n" . |
| 2966 | "Content-Length: " . strlen($json) . "\r\n" . |
| 2967 | "X-Chunk-Type: error\r\n" . |
| 2968 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 2969 | "\r\n" . |
| 2970 | $json . "\r\n" |
| 2971 | ); |
| 2972 | $gz->sync(); |
| 2973 | $abort_payload = null; |
| 2974 | continue; |
| 2975 | } |
| 2976 | |
| 2977 | $allowed = $follow_symlinks; |
| 2978 | if (!$allowed) { |
| 2979 | foreach ($directories as $root) { |
| 2980 | if ( |
| 2981 | $current_real === $root || |
| 2982 | str_starts_with($current_real, $root . "/") |
| 2983 | ) { |
| 2984 | $allowed = true; |
| 2985 | break; |
| 2986 | } |
| 2987 | } |
| 2988 | } |
| 2989 | if (!$allowed) { |
| 2990 | $abort_payload = [ |
| 2991 | "error_type" => "dir_outside_root", |
| 2992 | "path" => base64_encode($current_real), |
| 2993 | "message" => "Directory is outside allowed roots", |
| 2994 | ]; |
| 2995 | array_pop($stack); |
| 2996 | $json = json_encode_or_throw($abort_payload); |
| 2997 | $cursor_json = json_encode_or_throw( |
| 2998 | ["stack" => encode_index_stack($stack)], |
| 2999 | JSON_UNESCAPED_SLASHES |
| 3000 | ); |
| 3001 | $cursor_b64 = base64_encode($cursor_json); |
| 3002 | $gz->write( |
| 3003 | "--{$boundary}\r\n" . |
| 3004 | "Content-Type: application/json\r\n" . |
| 3005 | "Content-Length: " . strlen($json) . "\r\n" . |
| 3006 | "X-Chunk-Type: error\r\n" . |
| 3007 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3008 | "\r\n" . |
| 3009 | $json . "\r\n" |
| 3010 | ); |
| 3011 | $gz->sync(); |
| 3012 | $abort_payload = null; |
| 3013 | continue; |
| 3014 | } |
| 3015 | |
| 3016 | // Use realpath() consistently for all paths. On hosts like wp.com, |
| 3017 | // /srv is a symlink to / and /srv/wordpress is a symlink to |
| 3018 | // /wordpress, so realpath() canonicalizes everything into one |
| 3019 | // namespace: /srv/htdocs → /htdocs, /srv/wordpress/... → /wordpress/... |
| 3020 | // This keeps root dirs and symlink-followed dirs consistent. |
| 3021 | $stack[$frame_index]["dir"] = $current_real; |
| 3022 | $current_dir = $current_real; |
| 3023 | |
| 3024 | |
| 3025 | |
| 3026 | clearstatcache(true, $current_real); |
| 3027 | $entries = @scandir($current_real, SCANDIR_SORT_ASCENDING); |
| 3028 | if ($entries === false) { |
| 3029 | $abort_payload = [ |
| 3030 | "error_type" => "dir_open", |
| 3031 | "path" => base64_encode($current_real), |
| 3032 | "message" => "Failed to open directory", |
| 3033 | ]; |
| 3034 | $json = json_encode_or_throw($abort_payload); |
| 3035 | $cursor_json = json_encode_or_throw( |
| 3036 | ["stack" => encode_index_stack($stack)], |
| 3037 | JSON_UNESCAPED_SLASHES |
| 3038 | ); |
| 3039 | $cursor_b64 = base64_encode($cursor_json); |
| 3040 | $gz->write( |
| 3041 | "--{$boundary}\r\n" . |
| 3042 | "Content-Type: application/json\r\n" . |
| 3043 | "Content-Length: " . strlen($json) . "\r\n" . |
| 3044 | "X-Chunk-Type: error\r\n" . |
| 3045 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3046 | "\r\n" . |
| 3047 | $json . "\r\n" |
| 3048 | ); |
| 3049 | $gz->sync(); |
| 3050 | $abort_payload = null; |
| 3051 | array_pop($stack); |
| 3052 | continue; |
| 3053 | } |
| 3054 | |
| 3055 | // E2E test hook: during directory scanning |
| 3056 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 3057 | _e2e_load_test_hooks_if_needed($config); |
| 3058 | $hook_args = [$current_real, &$entries]; |
| 3059 | _e2e_call_hook('test_hook_during_dir_scan', $hook_args); |
| 3060 | } |
| 3061 | |
| 3062 | $filtered = []; |
| 3063 | foreach ($entries as $entry) { |
| 3064 | if ($entry === "." || $entry === "..") { |
| 3065 | continue; |
| 3066 | } |
| 3067 | $filtered[] = $entry; |
| 3068 | } |
| 3069 | |
| 3070 | $position = 0; |
| 3071 | if ($current_after !== null && $current_after !== "") { |
| 3072 | $position = position_after_entry($filtered, $current_after); |
| 3073 | } |
| 3074 | |
| 3075 | while (true) { |
| 3076 | if ($position >= count($filtered)) { |
| 3077 | array_pop($stack); |
| 3078 | break; |
| 3079 | } |
| 3080 | $entry = $filtered[$position]; |
| 3081 | $position++; |
| 3082 | |
| 3083 | $stack[$frame_index]["after"] = $entry; |
| 3084 | $path = $current_dir . "/" . $entry; |
| 3085 | // Default deny-list. Applied before stat() to save a syscall |
| 3086 | // per skipped entry, and before the traversal push so we |
| 3087 | // don't recurse into skipped directories. The "after" cursor |
| 3088 | // is updated above this check, so resume correctly skips |
| 3089 | // past the filtered entry on the next request. |
| 3090 | if (!$include_caches && path_is_default_skipped($path)) { |
| 3091 | continue; |
| 3092 | } |
| 3093 | // The "" guard matters: path_is_within_root() with an empty |
| 3094 | // root would match every absolute path. |
| 3095 | if ($storage_path !== "" && path_is_within_root($path, $storage_path)) { |
| 3096 | continue; |
| 3097 | } |
| 3098 | clearstatcache(true, $path); |
| 3099 | $stat = @lstat($path); |
| 3100 | if ($stat === false) { |
| 3101 | if ( |
| 3102 | !$budget->has_remaining() |
| 3103 | ) { |
| 3104 | $status = "partial"; |
| 3105 | $stop = true; |
| 3106 | break; |
| 3107 | } |
| 3108 | continue; |
| 3109 | } |
| 3110 | |
| 3111 | $mode = $stat["mode"] & STAT_TYPE_MASK; |
| 3112 | $type = "file"; |
| 3113 | $link_target = null; |
| 3114 | if ($mode === STAT_TYPE_LINK) { |
| 3115 | $type = "link"; |
| 3116 | $resolved = resolve_symlink_target($path); |
| 3117 | $link_target = $resolved['target']; |
| 3118 | if ($follow_symlinks && !empty($resolved['intermediates'])) { |
| 3119 | $batch_items = array_merge($batch_items, $resolved['intermediates']); |
| 3120 | } |
| 3121 | } elseif ($mode === STAT_TYPE_DIR) { |
| 3122 | $type = "dir"; |
| 3123 | } elseif ($mode !== STAT_TYPE_FILE) { |
| 3124 | $type = "other"; |
| 3125 | } |
| 3126 | |
| 3127 | $ctime = (int) ($stat["ctime"] ?? 0); |
| 3128 | $size = $type === "file" ? (int) ($stat["size"] ?? 0) : 0; |
| 3129 | |
| 3130 | $item = [ |
| 3131 | "path" => $path, |
| 3132 | "ctime" => $ctime, |
| 3133 | "size" => $size, |
| 3134 | "type" => $type, |
| 3135 | ]; |
| 3136 | if ($link_target !== null) { |
| 3137 | $item["target"] = $link_target; |
| 3138 | } |
| 3139 | $batch_items[] = $item; |
| 3140 | |
| 3141 | if (count($batch_items) >= $batch_size) { |
| 3142 | // E2E test hook: before index batch is emitted |
| 3143 | if (getenv('SITE_EXPORT_TEST_MODE')) { |
| 3144 | _e2e_load_test_hooks_if_needed($config); |
| 3145 | $hook_args = [&$batch_items, $stack]; |
| 3146 | _e2e_call_hook('test_hook_before_index_batch', $hook_args); |
| 3147 | } |
| 3148 | |
| 3149 | $cursor_json = json_encode_or_throw( |
| 3150 | ["stack" => encode_index_stack($stack)], |
| 3151 | JSON_UNESCAPED_SLASHES |
| 3152 | ); |
| 3153 | $cursor_b64 = base64_encode($cursor_json); |
| 3154 | $json = json_encode_or_throw( |
| 3155 | encode_index_batch($batch_items), |
| 3156 | JSON_UNESCAPED_SLASHES |
| 3157 | ); |
| 3158 | |
| 3159 | $gz->write( |
| 3160 | "--{$boundary}\r\n" . |
| 3161 | "Content-Type: application/json\r\n" . |
| 3162 | "Content-Length: " . strlen($json) . "\r\n" . |
| 3163 | "X-Chunk-Type: index_batch\r\n" . |
| 3164 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3165 | "X-Batch-Size: " . count($batch_items) . "\r\n" . |
| 3166 | "\r\n" |
| 3167 | ); |
| 3168 | $gz->write($json); |
| 3169 | $gz->write("\r\n"); |
| 3170 | $gz->sync(); |
| 3171 | |
| 3172 | $batches_emitted++; |
| 3173 | $total_entries += count($batch_items); |
| 3174 | $batch_items = []; |
| 3175 | } |
| 3176 | |
| 3177 | if ($type === "dir") { |
| 3178 | // Skip traversing directories whose realpath is already |
| 3179 | // covered by the configured roots (duplicate root), or is a |
| 3180 | // parent of one of them (would expose outside-tree files and |
| 3181 | // re-enter a scheduled root). O(k) where k = number of roots. |
| 3182 | $dir_real = realpath($path); |
| 3183 | if ($dir_real !== false && should_skip_index_root($dir_real, $directories)) { |
| 3184 | // Don't push — emit the entry but skip traversal |
| 3185 | continue; |
| 3186 | } |
| 3187 | $stack[] = [ |
| 3188 | "dir" => $path, |
| 3189 | "after" => null, |
| 3190 | ]; |
| 3191 | break; |
| 3192 | } |
| 3193 | |
| 3194 | if ( |
| 3195 | !$budget->has_remaining() |
| 3196 | ) { |
| 3197 | $status = "partial"; |
| 3198 | $stop = true; |
| 3199 | break; |
| 3200 | } |
| 3201 | } |
| 3202 | |
| 3203 | if ($stop) { |
| 3204 | break; |
| 3205 | } |
| 3206 | |
| 3207 | if ( |
| 3208 | !$budget->has_remaining() |
| 3209 | ) { |
| 3210 | $status = "partial"; |
| 3211 | break; |
| 3212 | } |
| 3213 | } |
| 3214 | } catch (Throwable $e) { |
| 3215 | $aborted = true; |
| 3216 | $abort_payload = [ |
| 3217 | "error_type" => "exception", |
| 3218 | "path" => base64_encode($current_dir), |
| 3219 | "message" => $e->getMessage(), |
| 3220 | ]; |
| 3221 | } |
| 3222 | |
| 3223 | // -- Flush remaining items and write completion chunk -- |
| 3224 | if (!empty($batch_items)) { |
| 3225 | $cursor_json = json_encode_or_throw( |
| 3226 | ["stack" => encode_index_stack($stack)], |
| 3227 | JSON_UNESCAPED_SLASHES |
| 3228 | ); |
| 3229 | $cursor_b64 = base64_encode($cursor_json); |
| 3230 | $json = json_encode_or_throw( |
| 3231 | encode_index_batch($batch_items), |
| 3232 | JSON_UNESCAPED_SLASHES |
| 3233 | ); |
| 3234 | |
| 3235 | $gz->write( |
| 3236 | "--{$boundary}\r\n" . |
| 3237 | "Content-Type: application/json\r\n" . |
| 3238 | "Content-Length: " . strlen($json) . "\r\n" . |
| 3239 | "X-Chunk-Type: index_batch\r\n" . |
| 3240 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3241 | "X-Batch-Size: " . count($batch_items) . "\r\n" . |
| 3242 | "\r\n" |
| 3243 | ); |
| 3244 | $gz->write($json); |
| 3245 | $gz->write("\r\n"); |
| 3246 | $gz->sync(); |
| 3247 | |
| 3248 | $batches_emitted++; |
| 3249 | $total_entries += count($batch_items); |
| 3250 | } |
| 3251 | |
| 3252 | try { |
| 3253 | if ($abort_payload !== null) { |
| 3254 | $json = json_encode_or_throw($abort_payload); |
| 3255 | $cursor_json = json_encode_or_throw( |
| 3256 | ["stack" => encode_index_stack($stack)], |
| 3257 | JSON_UNESCAPED_SLASHES |
| 3258 | ); |
| 3259 | $cursor_b64 = base64_encode($cursor_json); |
| 3260 | $gz->write( |
| 3261 | "--{$boundary}\r\n" . |
| 3262 | "Content-Type: application/json\r\n" . |
| 3263 | "Content-Length: " . strlen($json) . "\r\n" . |
| 3264 | "X-Chunk-Type: error\r\n" . |
| 3265 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3266 | "\r\n" . |
| 3267 | $json . "\r\n" |
| 3268 | ); |
| 3269 | $gz->sync(); |
| 3270 | $status = "partial"; |
| 3271 | } |
| 3272 | |
| 3273 | $cursor_json = json_encode_or_throw( |
| 3274 | ["stack" => encode_index_stack($stack)], |
| 3275 | JSON_UNESCAPED_SLASHES |
| 3276 | ); |
| 3277 | $cursor_b64 = base64_encode($cursor_json); |
| 3278 | |
| 3279 | $gz->write( |
| 3280 | "--{$boundary}\r\n" . |
| 3281 | "Content-Type: application/octet-stream\r\n" . |
| 3282 | "Content-Length: 0\r\n" . |
| 3283 | "X-Chunk-Type: completion\r\n" . |
| 3284 | "X-Status: " . ($aborted ? "partial" : $status) . "\r\n" . |
| 3285 | "X-Cursor: " . $cursor_b64 . "\r\n" . |
| 3286 | "X-Index-Dir: " . base64_encode($list_dir_real) . "\r\n" . |
| 3287 | "X-Batches-Emitted: {$batches_emitted}\r\n" . |
| 3288 | "X-Total-Entries: {$total_entries}\r\n" . |
| 3289 | "X-Memory-Used: " . memory_get_peak_usage(true) . "\r\n" . |
| 3290 | "X-Memory-Limit: " . $budget->max_memory . "\r\n" . |
| 3291 | "X-Time-Elapsed: " . (microtime(true) - $budget->start_time) . "\r\n" . |
| 3292 | "\r\n" . |
| 3293 | "\r\n" . |
| 3294 | "--{$boundary}--\r\n" |
| 3295 | ); |
| 3296 | $gz->finish(); |
| 3297 | } catch (\Throwable $e) { |
| 3298 | error_log("Export: failed to write completion chunk: " . $e->getMessage()); |
| 3299 | } |
| 3300 | |
| 3301 | return [ |
| 3302 | "status" => $aborted ? "partial" : $status, |
| 3303 | "stats" => [ |
| 3304 | "batches_emitted" => $batches_emitted, |
| 3305 | "total_entries" => $total_entries, |
| 3306 | "memory_used" => memory_get_peak_usage(true), |
| 3307 | "time_elapsed" => microtime(true) - $budget->start_time, |
| 3308 | ], |
| 3309 | ]; |
| 3310 | } |
| 3311 | |
| 3312 | /** |
| 3313 | * Streams files from a client-provided path list (uploaded as JSON). |
| 3314 | */ |
| 3315 | function endpoint_file_fetch( |
| 3316 | array $config, |
| 3317 | ResourceBudget $budget |
| 3318 | ): array { |
| 3319 | // Same rationale as endpoint_file_index(): avoid stale path metadata across |
| 3320 | // requests in long-lived PHP processes. |
| 3321 | clearstatcache(true); |
| 3322 | |
| 3323 | $directories = resolve_directories($config); |
| 3324 | |
| 3325 | $list_path = $config["file_list_path"] ?? null; |
| 3326 | if ($list_path === null && isset($_FILES["file_list"])) { |
| 3327 | $tmp_name = $_FILES["file_list"]["tmp_name"] ?? ""; |
| 3328 | if ($tmp_name === "" || !is_uploaded_file($tmp_name)) { |
| 3329 | throw new InvalidArgumentException( |
| 3330 | "file_list upload missing or invalid" |
| 3331 | ); |
| 3332 | } |
| 3333 | $list_path = $tmp_name; |
| 3334 | } |
| 3335 | |
| 3336 | if ($list_path === null) { |
| 3337 | throw new InvalidArgumentException( |
| 3338 | "file_list is required for file_fetch endpoint" |
| 3339 | ); |
| 3340 | } |
| 3341 | |
| 3342 | $raw = file_get_contents($list_path); |
| 3343 | if ($raw === false) { |
| 3344 | throw new InvalidArgumentException("Failed to read file_list"); |
| 3345 | } |
| 3346 | $decoded = json_decode($raw, true); |
| 3347 | if (!is_array($decoded)) { |
| 3348 | throw new InvalidArgumentException( |
| 3349 | "file_list must be a JSON array of paths" |
| 3350 | ); |
| 3351 | } |
| 3352 | $paths = []; |
| 3353 | foreach ($decoded as $path) { |
| 3354 | if (!is_string($path) || $path === "") { |
| 3355 | continue; |
| 3356 | } |
| 3357 | $paths[] = $path; |
| 3358 | } |
| 3359 | |
| 3360 | $chunk_size = $config["chunk_size"] ?? FileTreeProducer::DEFAULT_CHUNK_SIZE; |
| 3361 | $chunk_size = require_int_range( |
| 3362 | "chunk_size", |
| 3363 | (int) $chunk_size, |
| 3364 | 16 * 1024, |
| 3365 | 32 * 1024 * 1024 |
| 3366 | ); |
| 3367 | |
| 3368 | $sync_options = [ |
| 3369 | "chunk_size" => $chunk_size, |
| 3370 | "paths" => $paths, |
| 3371 | ]; |
| 3372 | if (isset($config["cursor"])) { |
| 3373 | $sync_options["cursor"] = $config["cursor"]; |
| 3374 | } |
| 3375 | |
| 3376 | $producer = new FileTreeProducer($directories, $sync_options); |
| 3377 | return stream_file_producer( |
| 3378 | $producer, |
| 3379 | $budget, |
| 3380 | $config, |
| 3381 | file_fetch_paths_should_gzip($paths) |
| 3382 | ); |
| 3383 | } |
| 3384 | |
| 3385 | /** |
| 3386 | * Decides whether to gzip a file_fetch multipart response based on the path |
| 3387 | * list it will carry. |
| 3388 | * |
| 3389 | * Encoding is set per response (Content-Encoding is a response-level header), |
| 3390 | * so we have to commit before any byte is sent. The trade-off: |
| 3391 | * - Text-y bodies (PHP/JS/CSS/JSON/SQL/HTML/etc.) compress 5–60×. Gzip is |
| 3392 | * a clear win on wire size and total wall time. |
| 3393 | * - Image/video/audio/font/archive bodies are already compressed; passing |
| 3394 | * them through gzip costs ~4 ms per 200 KB and produces ~0% size |
| 3395 | * reduction (deflate falls back to literal stored blocks for incompressible |
| 3396 | * input). Negligible per individual file, but unbounded if the batch is |
| 3397 | * all-binary multiplied by request volume. |
| 3398 | * |
| 3399 | * Rule: gzip the response if **any** file in the batch is compressible. |
| 3400 | * |
| 3401 | * The previous all-or-nothing rule ("gzip only if every file is compressible") |
| 3402 | * was over-conservative — a single PNG in a 200-CSS batch flipped the whole |
| 3403 | * response to identity, losing ~50 % of wire size that would have compressed. |
| 3404 | * The wasted CPU on the small binary portion of mixed batches is bounded by |
| 3405 | * request size (capped server-side), so this trade-off favors smaller wire |
| 3406 | * bytes on the common WordPress mixed batch (theme dirs, wp-content/uploads |
| 3407 | * mixed with plugin assets) without harming the all-binary uploads case |
| 3408 | * (which has zero compressible files and stays identity). |
| 3409 | */ |
| 3410 | function file_fetch_paths_should_gzip(array $paths): bool |
| 3411 | { |
| 3412 | if ($paths === []) { |
| 3413 | return false; |
| 3414 | } |
| 3415 | $any_compressible = false; |
| 3416 | foreach ($paths as $path) { |
| 3417 | if (!is_string($path)) { |
| 3418 | // Defensive: an unexpected non-string entry is a bad input we |
| 3419 | // shouldn't compress around. Treat as a hard reject. |
| 3420 | return false; |
| 3421 | } |
| 3422 | // Once true, we can skip checking the subsequent files. |
| 3423 | if ($any_compressible) { |
| 3424 | continue; |
| 3425 | } |
| 3426 | $ext = path_extension_compressibility($path); |
| 3427 | if ($ext === 'yes') { |
| 3428 | $any_compressible = true; |
| 3429 | continue; |
| 3430 | } |
| 3431 | if ($ext === 'unknown') { |
| 3432 | // Extension didn't match a known-text or known-binary list. Peek |
| 3433 | // at the first 64 bytes and let the bytes decide. Cheap (one |
| 3434 | // open/read/close per file) and means we don't have to grow the |
| 3435 | // whitelist every time a plugin invents a new template suffix. |
| 3436 | if (path_head_looks_like_text($path)) { |
| 3437 | $any_compressible = true; |
| 3438 | } |
| 3439 | continue; |
| 3440 | } |
| 3441 | // 'no' — known binary. Skip; doesn't disqualify the batch. |
| 3442 | } |
| 3443 | return $any_compressible; |
| 3444 | } |
| 3445 | |
| 3446 | /** |
| 3447 | * Returns true if a path's basename suggests text content gzip will shrink. |
| 3448 | * |
| 3449 | * Files with no extension (`.htaccess`, `LICENSE`, `README`, dotfiles) are |
| 3450 | * treated as text by convention — that's almost always how they're stored |
| 3451 | * in WordPress installs. |
| 3452 | */ |
| 3453 | function path_extension_is_compressible(string $path): bool |
| 3454 | { |
| 3455 | return path_extension_compressibility($path) === 'yes'; |
| 3456 | } |
| 3457 | |
| 3458 | /** |
| 3459 | * Three-state classifier for a path's extension. |
| 3460 | * |
| 3461 | * - 'yes' known text-y extension (or dotfile / extensionless name). |
| 3462 | * - 'no' known binary/already-compressed extension. |
| 3463 | * - 'unknown' neither list matches; caller may probe the file bytes. |
| 3464 | */ |
| 3465 | function path_extension_compressibility(string $path): string |
| 3466 | { |
| 3467 | $basename = basename($path); |
| 3468 | if ($basename === '') { |
| 3469 | return 'no'; |
| 3470 | } |
| 3471 | // Dotfiles like .htaccess / .env / .gitignore have no "real" extension — |
| 3472 | // pathinfo() reports the part after the leading dot as the extension, |
| 3473 | // but they're text by convention. Treat the whole class as compressible. |
| 3474 | if ($basename[0] === '.' && strpos($basename, '.', 1) === false) { |
| 3475 | return 'yes'; |
| 3476 | } |
| 3477 | $ext = strtolower((string) pathinfo($basename, PATHINFO_EXTENSION)); |
| 3478 | // Files with truly no extension (LICENSE, README, Makefile) — treat as text. |
| 3479 | if ($ext === '') { |
| 3480 | return 'yes'; |
| 3481 | } |
| 3482 | static $compressible = [ |
| 3483 | // Source / markup |
| 3484 | 'php', 'phtml', 'js', 'jsx', 'ts', 'tsx', 'mjs', 'cjs', |
| 3485 | 'css', 'scss', 'sass', 'less', |
| 3486 | 'html', 'htm', 'xml', 'xsl', 'xslt', 'svg', |
| 3487 | 'vue', 'astro', 'twig', 'mustache', 'hbs', 'liquid', |
| 3488 | // Data / config |
| 3489 | 'json', 'jsonl', 'yaml', 'yml', 'toml', 'csv', 'tsv', |
| 3490 | 'sql', 'ini', 'conf', 'cfg', 'env', 'properties', |
| 3491 | // Docs / plain text |
| 3492 | 'md', 'markdown', 'txt', 'log', 'rst', 'adoc', |
| 3493 | // Translations / feeds / captions |
| 3494 | 'pot', 'po', 'rss', 'atom', 'srt', 'vtt', 'webvtt', |
| 3495 | // Misc text-y |
| 3496 | 'sh', 'bash', 'patch', 'diff', |
| 3497 | ]; |
| 3498 | if (in_array($ext, $compressible, true)) { |
| 3499 | return 'yes'; |
| 3500 | } |
| 3501 | static $incompressible = [ |
| 3502 | // Already-compressed / encrypted archives |
| 3503 | 'zip', 'gz', 'tgz', 'bz2', 'xz', '7z', 'rar', 'tar', |
| 3504 | // Images |
| 3505 | 'jpg', 'jpeg', 'png', 'gif', 'webp', 'heic', 'heif', 'avif', |
| 3506 | 'tiff', 'tif', 'bmp', 'ico', |
| 3507 | // Audio |
| 3508 | 'mp3', 'm4a', 'aac', 'ogg', 'opus', 'flac', 'wav', |
| 3509 | // Video |
| 3510 | 'mp4', 'm4v', 'mov', 'webm', 'mkv', 'avi', |
| 3511 | // Fonts (already deflate-compressed in woff/woff2) |
| 3512 | 'woff', 'woff2', 'ttf', 'otf', 'eot', |
| 3513 | // Misc binary blobs |
| 3514 | 'pdf', 'psd', 'sketch', 'fig', 'iso', 'dmg', 'mo', 'phar', |
| 3515 | ]; |
| 3516 | if (in_array($ext, $incompressible, true)) { |
| 3517 | return 'no'; |
| 3518 | } |
| 3519 | return 'unknown'; |
| 3520 | } |
| 3521 | |
| 3522 | /** |
| 3523 | * Probes the first bytes of a file to decide if it looks like text. |
| 3524 | * |
| 3525 | * Used as a fallback when the extension didn't match either the text or the |
| 3526 | * binary list. The cost is one open + read + close per file in the |
| 3527 | * file_fetch batch, which is negligible relative to streaming the file |
| 3528 | * itself; the upside is we don't need to grow the extension lists every |
| 3529 | * time a plugin invents a new template suffix. |
| 3530 | * |
| 3531 | * The check is deliberately strict: any NUL or other ASCII control byte |
| 3532 | * (outside tab/newline/CR/form-feed) means binary, and the head must also |
| 3533 | * decode as valid UTF-8. UTF-8 happens to reject most random binary |
| 3534 | * sequences naturally because high-bit bytes only validate in well-formed |
| 3535 | * multi-byte runs — so PNG, JPEG, ZIP, etc. fail this within a handful of |
| 3536 | * bytes even when their headers look ASCII. |
| 3537 | */ |
| 3538 | function path_head_looks_like_text(string $path): bool |
| 3539 | { |
| 3540 | if (!is_file($path)) { |
| 3541 | return false; |
| 3542 | } |
| 3543 | $fp = @fopen($path, 'rb'); |
| 3544 | if ($fp === false) { |
| 3545 | // Producer will surface a clearer error later; don't compress on |
| 3546 | // unreadable paths. |
| 3547 | return false; |
| 3548 | } |
| 3549 | $head = (string) fread($fp, 64); |
| 3550 | fclose($fp); |
| 3551 | if ($head === '') { |
| 3552 | // Empty file: nothing to compress, default to identity. |
| 3553 | return false; |
| 3554 | } |
| 3555 | // Any NUL byte → binary. Cheapest signal, catches PNG/ZIP/woff/etc. |
| 3556 | if (strpos($head, "\x00") !== false) { |
| 3557 | return false; |
| 3558 | } |
| 3559 | // Other ASCII control bytes (excluding TAB \x09, LF \x0A, FF \x0C, CR \x0D) |
| 3560 | // shouldn't appear in source/data files. Also reject DEL \x7F. |
| 3561 | if (preg_match('/[\x01-\x08\x0B\x0E-\x1F\x7F]/', $head)) { |
| 3562 | return false; |
| 3563 | } |
| 3564 | // Must decode cleanly as UTF-8. mb_check_encoding handles the case where |
| 3565 | // a multi-byte sequence is sliced by our 64-byte window: it returns false, |
| 3566 | // which we treat as "not obviously text" — biased toward identity, which |
| 3567 | // is the safe direction. |
| 3568 | if (function_exists('mb_check_encoding') && !mb_check_encoding($head, 'UTF-8')) { |
| 3569 | return false; |
| 3570 | } |
| 3571 | return true; |
| 3572 | } |
| 3573 | |
| 3574 | /** |
| 3575 | * Returns true if $path is a generated cache file, version-control or |
| 3576 | * dev-tooling artifact, or OS-level junk that is not worth shipping in |
| 3577 | * a typical site migration. |
| 3578 | * |
| 3579 | * Matching rules: |
| 3580 | * |
| 3581 | * - Path-component-aware: a segment that *contains* a skipped name as a |
| 3582 | * substring (e.g. "cache-control" or "node_modules-backup") does NOT |
| 3583 | * trigger a skip. Only whole-segment matches do. This is done by |
| 3584 | * wrapping `/` around both the haystack and needle and doing a |
| 3585 | * substring check. |
| 3586 | * |
| 3587 | * - Cache/upgrade dirs are matched only under `wp-content/` so a user |
| 3588 | * directory literally called `cache` in some other tree doesn't |
| 3589 | * silently disappear. |
| 3590 | * |
| 3591 | * - Dotfiles that ship in real WordPress sites — `.htaccess`, |
| 3592 | * `.user.ini`, `.well-known/` — are preserved. Editor/VCS dotfiles |
| 3593 | * and macOS metadata are not. |
| 3594 | * |
| 3595 | * The default deny-list is conservative: false-negatives (something we |
| 3596 | * could have skipped but didn't) are mere wire-byte waste; false-positives |
| 3597 | * (something the user actually wanted) are silent data loss. Callers |
| 3598 | * opting in to a more aggressive filter can pass extra patterns; callers |
| 3599 | * who want everything can set include_caches=1 on the request. |
| 3600 | */ |
| 3601 | function path_is_default_skipped(string $path): bool |
| 3602 | { |
| 3603 | // Sentinel slashes on each side make "starts-with" / "ends-with" / |
| 3604 | // "anywhere-in-middle" the same str_contains() check. |
| 3605 | $needle_haystack = '/' . trim($path, '/') . '/'; |
| 3606 | |
| 3607 | // Generated content under wp-content/. WordPress regenerates these |
| 3608 | // on demand (cache via the page lifecycle, upgrade via wp-admin |
| 3609 | // updates), so transferring them is pure waste. |
| 3610 | // |
| 3611 | // Notable specific entries: |
| 3612 | // - wp-content/wpcomsh-cache: wp.com Atomic's Memcached-backed |
| 3613 | // filesystem cache shadow. |
| 3614 | // - wp-content/wflogs: Wordfence's per-request scan logs; can |
| 3615 | // reach gigabytes on long-running sites. |
| 3616 | static $cache_dirs = [ |
| 3617 | '/wp-content/cache/', |
| 3618 | '/wp-content/upgrade/', |
| 3619 | '/wp-content/wpcomsh-cache/', |
| 3620 | '/wp-content/wflogs/', |
| 3621 | ]; |
| 3622 | foreach ($cache_dirs as $needle) { |
| 3623 | if (strpos($needle_haystack, $needle) !== false) { |
| 3624 | return true; |
| 3625 | } |
| 3626 | } |
| 3627 | |
| 3628 | // VCS metadata + local dev tooling. Match any path component exactly. |
| 3629 | static $junk_components = [ |
| 3630 | '.git', '.svn', '.hg', '.bzr', |
| 3631 | 'node_modules', |
| 3632 | '.idea', '.vscode', |
| 3633 | '.cache', '.npm', '.yarn', '.pnpm-store', |
| 3634 | ]; |
| 3635 | foreach ($junk_components as $needle) { |
| 3636 | if (strpos($needle_haystack, '/' . $needle . '/') !== false) { |
| 3637 | return true; |
| 3638 | } |
| 3639 | } |
| 3640 | |
| 3641 | // OS junk + filesystem metadata files (basename match). |
| 3642 | $basename = basename($path); |
| 3643 | static $junk_basenames = [ |
| 3644 | '.DS_Store', '._.DS_Store', |
| 3645 | 'Thumbs.db', 'desktop.ini', 'ehthumbs.db', |
| 3646 | ]; |
| 3647 | if (in_array($basename, $junk_basenames, true)) { |
| 3648 | return true; |
| 3649 | } |
| 3650 | |
| 3651 | // Editor / merge scratch files (basename pattern): |
| 3652 | // `.#name` Emacs lock |
| 3653 | // `#name#` Emacs autosave |
| 3654 | // `name~` Editor backup |
| 3655 | // `name.swp` Vim swap (also .swo, .swn) |
| 3656 | // `name.bak` generic backup |
| 3657 | // `name.orig` merge conflict leftover |
| 3658 | // `name.rej` merge conflict leftover |
| 3659 | if ($basename !== '' && $basename[0] === '.' && isset($basename[1]) && $basename[1] === '#') { |
| 3660 | return true; |
| 3661 | } |
| 3662 | if (strlen($basename) >= 3 && $basename[0] === '#' && substr($basename, -1) === '#') { |
| 3663 | return true; |
| 3664 | } |
| 3665 | if (preg_match('/(?:~|\.(?:swp|swo|swn|bak|orig|rej))$/', $basename) === 1) { |
| 3666 | return true; |
| 3667 | } |
| 3668 | |
| 3669 | return false; |
| 3670 | } |
| 3671 | |
| 3672 | /** |
| 3673 | * Maps importer-requested SQL row filters to producer row-exclusion rules. |
| 3674 | * |
| 3675 | * Rules are data, not exporter-known tokens. A client may provide: |
| 3676 | * |
| 3677 | * skip_rows[0][table_name_without_prefix]=postmeta |
| 3678 | * skip_rows[0][column]=meta_key |
| 3679 | * skip_rows[0][value_base64]=X2VkaXRfbG9jaw== |
| 3680 | * |
| 3681 | * `table_name_without_prefix` is appended to the server-side WordPress table prefix. The |
| 3682 | * prefix must come from WordPress; if it cannot be resolved, clients must use |
| 3683 | * explicit `table` instead. Values are base64-encoded so raw bytes never travel |
| 3684 | * as SQL text. |
| 3685 | * |
| 3686 | * @return list<array{table: string, column: string, value: string}> |
| 3687 | */ |
| 3688 | function sql_exclude_rows_from_config(array $config, ?string $table_prefix): array |
| 3689 | { |
| 3690 | if (!isset($config["skip_rows"])) { |
| 3691 | return []; |
| 3692 | } |
| 3693 | |
| 3694 | $requested = $config["skip_rows"]; |
| 3695 | if (is_string($requested)) { |
| 3696 | $decoded = json_decode($requested, true); |
| 3697 | if (!is_array($decoded)) { |
| 3698 | throw new InvalidArgumentException("skip_rows string must be a JSON array"); |
| 3699 | } |
| 3700 | $requested = $decoded; |
| 3701 | } |
| 3702 | if (!is_array($requested)) { |
| 3703 | throw new InvalidArgumentException("skip_rows must be an array"); |
| 3704 | } |
| 3705 | |
| 3706 | $rules = []; |
| 3707 | foreach ($requested as $index => $rule) { |
| 3708 | if (!is_array($rule)) { |
| 3709 | throw new InvalidArgumentException("skip_rows[{$index}] must be an object"); |
| 3710 | } |
| 3711 | |
| 3712 | $has_table = isset($rule["table"]); |
| 3713 | $has_table_name_without_prefix = isset($rule["table_name_without_prefix"]); |
| 3714 | if ($has_table === $has_table_name_without_prefix) { |
| 3715 | throw new InvalidArgumentException("skip_rows[{$index}] must include exactly one of table or table_name_without_prefix"); |
| 3716 | } |
| 3717 | if (!isset($rule["column"], $rule["value_base64"])) { |
| 3718 | throw new InvalidArgumentException("skip_rows[{$index}] must include column and value_base64"); |
| 3719 | } |
| 3720 | if (!is_string($rule["column"]) || $rule["column"] === "") { |
| 3721 | throw new InvalidArgumentException("skip_rows[{$index}].column must be a non-empty string"); |
| 3722 | } |
| 3723 | if (!is_string($rule["value_base64"])) { |
| 3724 | throw new InvalidArgumentException("skip_rows[{$index}].value_base64 must be a string"); |
| 3725 | } |
| 3726 | |
| 3727 | if ($has_table_name_without_prefix) { |
| 3728 | if (!is_string($rule["table_name_without_prefix"]) || $rule["table_name_without_prefix"] === "") { |
| 3729 | throw new InvalidArgumentException("skip_rows[{$index}].table_name_without_prefix must be a non-empty string"); |
| 3730 | } |
| 3731 | if ($table_prefix === null || $table_prefix === "") { |
| 3732 | throw new InvalidArgumentException("skip_rows[{$index}].table_name_without_prefix requires a table_prefix"); |
| 3733 | } |
| 3734 | $table = $table_prefix . $rule["table_name_without_prefix"]; |
| 3735 | } else { |
| 3736 | if (!is_string($rule["table"]) || $rule["table"] === "") { |
| 3737 | throw new InvalidArgumentException("skip_rows[{$index}].table must be a non-empty string"); |
| 3738 | } |
| 3739 | $table = $rule["table"]; |
| 3740 | } |
| 3741 | |
| 3742 | $value = base64_decode($rule["value_base64"], true); |
| 3743 | if ($value === false) { |
| 3744 | throw new InvalidArgumentException("skip_rows[{$index}].value_base64 must be valid base64"); |
| 3745 | } |
| 3746 | |
| 3747 | $rules[] = [ |
| 3748 | "table" => $table, |
| 3749 | "column" => $rule["column"], |
| 3750 | "value" => $value, |
| 3751 | ]; |
| 3752 | } |
| 3753 | |
| 3754 | return $rules; |
| 3755 | } |
| 3756 | |
| 3757 | /** |
| 3758 | * Validates that an integer falls within the given range, or throws. |
| 3759 | */ |
| 3760 | function require_int_range( |
| 3761 | string $name, |
| 3762 | int $value, |
| 3763 | int $min, |
| 3764 | int $max |
| 3765 | ): int { |
| 3766 | if ($value < $min || $value > $max) { |
| 3767 | throw new InvalidArgumentException( |
| 3768 | "{$name} out of range. Expected {$min}-{$max}, got {$value}" |
| 3769 | ); |
| 3770 | } |
| 3771 | return $value; |
| 3772 | } |
| 3773 | |
| 3774 | /** |
| 3775 | * Validates that a float falls within the given range, or throws. |
| 3776 | */ |
| 3777 | function require_float_range( |
| 3778 | string $name, |
| 3779 | float $value, |
| 3780 | float $min, |
| 3781 | float $max |
| 3782 | ): float { |
| 3783 | if ($value < $min || $value > $max) { |
| 3784 | throw new InvalidArgumentException( |
| 3785 | "{$name} out of range. Expected {$min}-{$max}, got {$value}" |
| 3786 | ); |
| 3787 | } |
| 3788 | return $value; |
| 3789 | } |
| 3790 | |
| 3791 | /** |
| 3792 | * Returns the index of the first entry lexicographically after $after (binary search). |
| 3793 | */ |
| 3794 | function position_after_entry(array $entries, string $after): int |
| 3795 | { |
| 3796 | $low = 0; |
| 3797 | $high = count($entries); |
| 3798 | while ($low < $high) { |
| 3799 | $mid = (int) (($low + $high) / 2); |
| 3800 | $entry = $entries[$mid]; |
| 3801 | if (strcmp($entry, $after) <= 0) { |
| 3802 | $low = $mid + 1; |
| 3803 | } else { |
| 3804 | $high = $mid; |
| 3805 | } |
| 3806 | } |
| 3807 | return $low; |
| 3808 | } |
| 3809 | |
| 3810 | /** |
| 3811 | * Builds the config array from HTTP GET/POST parameters and optional JSON body. |
| 3812 | */ |
| 3813 | function parse_http_config(): array |
| 3814 | { |
| 3815 | $body = file_get_contents('php://input'); |
| 3816 | if ($body === false) { |
| 3817 | $body = ''; |
| 3818 | } |
| 3819 | |
| 3820 | $server = new Site_Export_HTTP_Server(); |
| 3821 | return $server->parse_http_config($_GET, $_POST, $_SERVER, $body); |
| 3822 | } |
| 3823 |