Compression
4 months ago
Database
1 week ago
AbstractBackupsFinder.php
1 year ago
AbstractExtractor.php
1 week ago
AbstractServiceProvider.php
2 years ago
Archiver.php
1 week ago
BackupAssets.php
1 month ago
BackupContent.php
1 year ago
BackupMetadataEditor.php
1 year ago
BackupMetadataReader.php
5 months ago
BackupSigner.php
6 months ago
BackupsDirectoryResolver.php
1 week ago
BackupsFinder.php
1 week ago
Extractor.php
1 week ago
FileBackupService.php
1 week ago
FileBackupServiceProvider.php
2 years ago
ServiceInterface.php
2 years ago
TmpBackupCleaner.php
1 week ago
ZlibCompressor.php
11 months ago
Extractor.php
607 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Extracts files from WP Staging backup archives to restore sites |
| 5 | * |
| 6 | * Handles the extraction process for both compressed and uncompressed backups, |
| 7 | * including validation, disk space checks, and file restoration with proper permissions. |
| 8 | */ |
| 9 | |
| 10 | namespace WPStaging\Backup\Service; |
| 11 | |
| 12 | use Exception; |
| 13 | use OutOfRangeException; |
| 14 | use RuntimeException; |
| 15 | use WPStaging\Backup\BackupFileIndex; |
| 16 | use WPStaging\Backup\BackupHeader; |
| 17 | use WPStaging\Backup\BackupValidator; |
| 18 | use WPStaging\Backup\Exceptions\EmptyChunkException; |
| 19 | use WPStaging\Backup\FileHeader; |
| 20 | use WPStaging\Backup\Interfaces\ExtractorTaskInterface; |
| 21 | use WPStaging\Core\WPStaging; |
| 22 | use WPStaging\Framework\Adapter\Directory; |
| 23 | use WPStaging\Framework\Job\Exception\DiskNotWritableException; |
| 24 | use WPStaging\Framework\Job\Exception\FileValidationException; |
| 25 | use WPStaging\Framework\Facades\Hooks; |
| 26 | use WPStaging\Framework\Filesystem\FileObject; |
| 27 | use WPStaging\Framework\Filesystem\DiskWriteCheck; |
| 28 | use WPStaging\Framework\Filesystem\MissingFileException; |
| 29 | use WPStaging\Framework\Filesystem\PathIdentifier; |
| 30 | use WPStaging\Framework\Filesystem\Permissions; |
| 31 | use WPStaging\Framework\Job\Dto\JobDataDto; |
| 32 | use WPStaging\Framework\Queue\FinishedQueueException; |
| 33 | use WPStaging\Framework\Traits\ResourceTrait; |
| 34 | use WPStaging\Framework\Traits\RestoreFileExclusionTrait; |
| 35 | use WPStaging\Vendor\Psr\Log\LoggerInterface; |
| 36 | |
| 37 | class Extractor extends AbstractExtractor |
| 38 | { |
| 39 | use ResourceTrait; |
| 40 | use RestoreFileExclusionTrait; |
| 41 | |
| 42 | /** @var LoggerInterface */ |
| 43 | protected $logger; |
| 44 | |
| 45 | /** @var DiskWriteCheck */ |
| 46 | protected $diskWriteCheck; |
| 47 | |
| 48 | /** @var BackupValidator */ |
| 49 | protected $backupValidator; |
| 50 | |
| 51 | /** @var ZlibCompressor */ |
| 52 | protected $zlibCompressor; |
| 53 | |
| 54 | /** @var ExtractorTaskInterface */ |
| 55 | protected $extractorTask; |
| 56 | |
| 57 | /** @var bool */ |
| 58 | protected $isRepairMultipleHeadersIssue = false; |
| 59 | |
| 60 | /** @var bool */ |
| 61 | protected $isFastPerformanceMode = true; |
| 62 | |
| 63 | /** @var bool */ |
| 64 | protected $isLastRequestGracefulShutdown = true; |
| 65 | |
| 66 | public function __construct( |
| 67 | PathIdentifier $pathIdentifier, |
| 68 | Directory $directory, |
| 69 | DiskWriteCheck $diskWriteCheck, |
| 70 | ZlibCompressor $zlibCompressor, |
| 71 | BackupValidator $backupValidator, |
| 72 | BackupHeader $backupHeader, |
| 73 | Permissions $permissions |
| 74 | ) { |
| 75 | parent::__construct($pathIdentifier, $directory, $backupHeader, $permissions); |
| 76 | $this->zlibCompressor = $zlibCompressor; |
| 77 | $this->backupValidator = $backupValidator; |
| 78 | $this->diskWriteCheck = $diskWriteCheck; |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * @param bool $isBackupFormatV1 |
| 83 | * @return void |
| 84 | */ |
| 85 | public function setIsBackupFormatV1(bool $isBackupFormatV1) |
| 86 | { |
| 87 | $this->isBackupFormatV1 = $isBackupFormatV1; |
| 88 | if ($isBackupFormatV1) { |
| 89 | $this->indexLineDto = new BackupFileIndex(); |
| 90 | } else { |
| 91 | $this->indexLineDto = WPStaging::make(FileHeader::class); |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | /** |
| 96 | * @param bool $isRepairMultipleHeadersIssue |
| 97 | * @return void |
| 98 | */ |
| 99 | public function setIsRepairMultipleHeadersIssue(bool $isRepairMultipleHeadersIssue) |
| 100 | { |
| 101 | $this->isRepairMultipleHeadersIssue = $isRepairMultipleHeadersIssue; |
| 102 | } |
| 103 | |
| 104 | public function setIsFastPerformanceMode(bool $isFastPerformanceMode) |
| 105 | { |
| 106 | $this->isFastPerformanceMode = $isFastPerformanceMode; |
| 107 | } |
| 108 | |
| 109 | public function setIsLastRequestGracefulShutdown(bool $isLastRequestGracefulShutdown) |
| 110 | { |
| 111 | $this->isLastRequestGracefulShutdown = $isLastRequestGracefulShutdown; |
| 112 | } |
| 113 | |
| 114 | /** |
| 115 | * @param ExtractorTaskInterface $extractorTask |
| 116 | * @param LoggerInterface $logger |
| 117 | * @return void |
| 118 | */ |
| 119 | public function inject(ExtractorTaskInterface $extractorTask, LoggerInterface $logger) |
| 120 | { |
| 121 | $this->extractorTask = $extractorTask; |
| 122 | $this->logger = $logger; |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * @param bool $isValidateOnly |
| 127 | * @return void |
| 128 | */ |
| 129 | public function setIsValidateOnly(bool $isValidateOnly) |
| 130 | { |
| 131 | $this->isValidateOnly = $isValidateOnly; |
| 132 | if ($isValidateOnly) { |
| 133 | $this->throwExceptionOnValidationFailure = true; |
| 134 | } |
| 135 | } |
| 136 | |
| 137 | /** |
| 138 | * @return void |
| 139 | * @throws DiskNotWritableException |
| 140 | */ |
| 141 | public function execute() |
| 142 | { |
| 143 | while (!$this->isThreshold()) { |
| 144 | try { |
| 145 | $this->findFileToExtract(); |
| 146 | } catch (OutOfRangeException $e) { |
| 147 | // Done processing, or failed |
| 148 | $this->logger->warning('OutOfRangeException. Error: ' . $e->getMessage()); |
| 149 | return; |
| 150 | } catch (RuntimeException $e) { |
| 151 | $this->logger->warning($e->getMessage()); |
| 152 | continue; |
| 153 | } catch (MissingFileException $e) { |
| 154 | $this->logger->warning('MissingFileException. Error: ' . $e->getMessage()); |
| 155 | continue; |
| 156 | } catch (Exception $e) { |
| 157 | if ($e->getCode() === self::FILE_FILTERED_EXCEPTION_CODE) { |
| 158 | continue; |
| 159 | } |
| 160 | |
| 161 | if ($e->getCode() === self::FINISHED_QUEUE_EXCEPTION_CODE) { |
| 162 | throw new FinishedQueueException(); |
| 163 | } |
| 164 | |
| 165 | if ($e->getCode() === self::ITEM_SKIP_EXCEPTION_CODE) { |
| 166 | continue; |
| 167 | } |
| 168 | |
| 169 | throw $e; |
| 170 | } |
| 171 | |
| 172 | try { |
| 173 | $this->processCurrentFile(); |
| 174 | } catch (FileValidationException $e) { |
| 175 | if ($this->isValidateOnly || $this->throwExceptionOnValidationFailure) { |
| 176 | throw $e; |
| 177 | } |
| 178 | |
| 179 | $this->logger->warning('Unable to validate file. Error: ' . $e->getMessage()); |
| 180 | } |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * @param Exception $ex |
| 186 | * @param string $filePath |
| 187 | * @return void |
| 188 | */ |
| 189 | protected function throwMissingFileException(Exception $ex, string $filePath) |
| 190 | { |
| 191 | throw new MissingFileException(sprintf("Following backup part missing: %s", $filePath), 0, $ex); |
| 192 | } |
| 193 | |
| 194 | protected function isBigFile(): bool |
| 195 | { |
| 196 | $sizeToConsiderAsBigFile = Hooks::applyFilters('wpstg.tests.restore.bigFileSize', 10 * MB_IN_BYTES); |
| 197 | |
| 198 | return $this->extractingFile->getTotalBytes() > $sizeToConsiderAsBigFile; |
| 199 | } |
| 200 | |
| 201 | protected function cleanExistingFile(string $identifier) |
| 202 | { |
| 203 | if ($this->isValidateOnly) { |
| 204 | return; |
| 205 | } |
| 206 | |
| 207 | if ($identifier !== PathIdentifier::IDENTIFIER_UPLOADS || $this->extractingFile->getWrittenBytes() > 0) { |
| 208 | return; |
| 209 | } |
| 210 | |
| 211 | // Continuation segment of a multipart-split upload file — the existing bytes on disk |
| 212 | // are the previous parts' segments and must be preserved so the file can be stitched. |
| 213 | if ($this->indexLineDto instanceof FileHeader && $this->indexLineDto->getIsPreviousPartRequired()) { |
| 214 | return; |
| 215 | } |
| 216 | |
| 217 | if (file_exists($this->extractingFile->getBackupPath())) { |
| 218 | // Delete the original upload file |
| 219 | if (!unlink($this->extractingFile->getBackupPath())) { |
| 220 | throw new \RuntimeException(sprintf(__('Could not delete original media library file %s. Skipping restore of it...', 'wp-staging'), $this->extractingFile->getRelativePath())); |
| 221 | } |
| 222 | } |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * Fixes issue https://github.com/wp-staging/wp-staging-pro/issues/2861 |
| 227 | * @return void |
| 228 | */ |
| 229 | protected function maybeRemoveLastAccidentalCharFromLastExtractedFile() |
| 230 | { |
| 231 | if ($this->backupMetadata->getTotalFiles() !== $this->extractorDto->getTotalFilesExtracted()) { |
| 232 | return; |
| 233 | } |
| 234 | |
| 235 | if ($this->backupValidator->validateFileIndexFirstLine($this->wpstgFile, $this->backupMetadata)) { |
| 236 | return; |
| 237 | } |
| 238 | |
| 239 | $this->removeLastCharInExtractedFile(); |
| 240 | } |
| 241 | |
| 242 | protected function getExtractFolder(string $identifier): string |
| 243 | { |
| 244 | if ($this->isValidateOnly) { |
| 245 | return trailingslashit($this->dirRestore . self::VALIDATE_DIRECTORY); |
| 246 | } |
| 247 | |
| 248 | if ($identifier === PathIdentifier::IDENTIFIER_UPLOADS) { |
| 249 | return $this->directory->getUploadsDirectory(); |
| 250 | } |
| 251 | |
| 252 | return $this->dirRestore . $identifier; |
| 253 | } |
| 254 | |
| 255 | /** |
| 256 | * @return void |
| 257 | * @throws DiskNotWritableException |
| 258 | */ |
| 259 | private function processCurrentFile() |
| 260 | { |
| 261 | $destinationFilePath = $this->extractingFile->getBackupPath(); |
| 262 | if ($this->currentIdentifier === PathIdentifier::IDENTIFIER_UPLOADS && $this->isExcludedFile($destinationFilePath)) { |
| 263 | $this->extractorDto->incrementTotalFilesSkipped(); |
| 264 | $this->extractorDto->setCurrentIndexOffset($this->wpstgIndexOffsetForNextFile); |
| 265 | $this->debugLog('Skipping excluded upload file: ' . rtrim($destinationFilePath, "\n")); |
| 266 | return; |
| 267 | } |
| 268 | |
| 269 | if ($this->extractingFile->getWrittenBytes() > 0) { |
| 270 | $this->logger->debug(sprintf('Resuming extraction of file %s from byte %d. Total size: %d...', $this->extractingFile->getRelativePath(), $this->extractingFile->getWrittenBytes(), $this->extractingFile->getTotalBytes())); |
| 271 | } |
| 272 | |
| 273 | $uncompressedSize = $this->indexLineDto->getUncompressedSize(); |
| 274 | $shouldExtractToMemory = $this->isValidateOnly |
| 275 | && !$this->isBackupFormatV1 |
| 276 | && !$this->isCurrentSegmentedFileHeader() |
| 277 | && $this->extractingFile->getWrittenBytes() === 0 |
| 278 | && $this->extractingFile->getReadBytes() === 0 |
| 279 | && $this->isWithinMemoryExtractionLimit($uncompressedSize) |
| 280 | && Hooks::applyFilters(JobDataDto::FILTER_BACKUP_USE_INMEMORY_EXTRACTION, true); |
| 281 | try { |
| 282 | if ($this->isThreshold()) { |
| 283 | // Prevent considering a file as big just because we start extracting at the threshold |
| 284 | return; |
| 285 | } |
| 286 | |
| 287 | if ($shouldExtractToMemory) { |
| 288 | $this->extractAndValidateInMemory(); |
| 289 | return; |
| 290 | } |
| 291 | |
| 292 | $this->extractFileToDisk(); |
| 293 | } catch (DiskNotWritableException $e) { |
| 294 | // Re-throw |
| 295 | throw $e; |
| 296 | } catch (OutOfRangeException $e) { |
| 297 | // Backup header, should be ignored silently |
| 298 | $this->extractingFile->setWrittenBytes($this->extractingFile->getTotalBytes()); |
| 299 | } catch (Exception $e) { |
| 300 | // Set this file as "written", so that we can skip to the next file. |
| 301 | $this->extractingFile->setWrittenBytes($this->extractingFile->getTotalBytes()); |
| 302 | |
| 303 | if (defined('WPSTG_DEBUG') && WPSTG_DEBUG) { |
| 304 | $this->logger->warning(sprintf('Skipped file %s. Reason: %s', $this->extractingFile->getRelativePath(), $e->getMessage())); |
| 305 | } |
| 306 | } |
| 307 | |
| 308 | if ($this->isFastPerformanceMode) { |
| 309 | return; |
| 310 | } |
| 311 | |
| 312 | $this->extractorTask->persistDto($this->extractorDto); |
| 313 | } |
| 314 | |
| 315 | /** |
| 316 | * @return void |
| 317 | * @throws DiskNotWritableException |
| 318 | * @throws \WPStaging\Framework\Filesystem\FilesystemExceptions |
| 319 | */ |
| 320 | private function fileBatchWrite() |
| 321 | { |
| 322 | $destinationFilePath = $this->extractingFile->getBackupPath(); |
| 323 | |
| 324 | if (strpos($destinationFilePath, '.sql') !== false) { |
| 325 | $this->logger->debug(sprintf('DEBUG: Extracting SQL file %s', $destinationFilePath)); |
| 326 | } |
| 327 | |
| 328 | $this->maybeResetFilePointerAfterInMemoryFallback(); |
| 329 | wp_mkdir_p(dirname($destinationFilePath)); |
| 330 | |
| 331 | /** |
| 332 | * On some servers, it is required to create empty file first, so we will create empty files. |
| 333 | * On some servers, touch doesn't work consistently, so we will use fwrite, see the reason below. |
| 334 | * On sites hosted on SiteGround, creating files using file_puts_contents uses a lot of memory, |
| 335 | * so by default we will use fwrite to create the empty file. |
| 336 | * If creating the empty file using fwrite fails, let try creating it using file_put_contents |
| 337 | * @see https://github.com/wp-staging/wp-staging-pro/issues/3272 why it was needed. |
| 338 | */ |
| 339 | if (!$this->createEmptyFile($destinationFilePath)) { |
| 340 | file_put_contents($destinationFilePath, ''); |
| 341 | } |
| 342 | |
| 343 | $destinationFileResource = @fopen($destinationFilePath, FileObject::MODE_APPEND); |
| 344 | if (!$destinationFileResource) { |
| 345 | $this->diskWriteCheck->testDiskIsWriteable(); |
| 346 | throw new Exception("Can not extract file $destinationFilePath"); |
| 347 | } |
| 348 | |
| 349 | /** |
| 350 | * When last request is not graceful shutdown and it is not fast performance mode (i.e. safe performance mode), |
| 351 | * we need to set the file pointer to the correct position in the backup file to continue extraction from where it left off. |
| 352 | * But this solution only works for non-compressed backups |
| 353 | */ |
| 354 | if (!$this->isLastRequestGracefulShutdown && !$this->isFastPerformanceMode && !$this->extractingFile->getIsCompressed()) { |
| 355 | $fileSize = $this->getRecoverableCurrentSegmentBytes($destinationFilePath); |
| 356 | $this->wpstgFile->fseek($this->extractingFile->getStart() + $fileSize); |
| 357 | $this->extractingFile->setReadBytes($fileSize); |
| 358 | $this->extractingFile->setWrittenBytes($fileSize); |
| 359 | $this->logger->debug(sprintf('DEBUG: Seeking to byte %d in backup file to continue extraction of %s...', $this->extractingFile->getStart() + $fileSize, $this->extractingFile->getRelativePath())); |
| 360 | } |
| 361 | |
| 362 | $lastDebugMessage = ''; |
| 363 | $processedChunks = 0; |
| 364 | while (!$this->extractingFile->isFinished() && !$this->isThreshold()) { |
| 365 | $readBytesBefore = $this->wpstgFile->ftell(); |
| 366 | try { |
| 367 | $chunk = $this->readAndPrepareChunk(); |
| 368 | } catch (DiskNotWritableException $ex) { |
| 369 | $this->diskWriteCheck->testDiskIsWriteable(); |
| 370 | throw new Exception("Unable to extract file to $destinationFilePath. Please check if there is enough disk space available."); |
| 371 | } |
| 372 | |
| 373 | if ($chunk === null) { |
| 374 | continue; |
| 375 | } |
| 376 | |
| 377 | $processedChunks++; |
| 378 | $this->updateProgressTracking($processedChunks, $lastDebugMessage); |
| 379 | $writtenBytes = $this->writeChunkToFile($destinationFileResource, $chunk); |
| 380 | |
| 381 | $this->trackChunkProgress($readBytesBefore, $writtenBytes); |
| 382 | $this->persistDto(); |
| 383 | } |
| 384 | |
| 385 | if (!empty($lastDebugMessage)) { |
| 386 | $this->logger->debug($lastDebugMessage); |
| 387 | } |
| 388 | |
| 389 | fclose($destinationFileResource); |
| 390 | $destinationFileResource = null; |
| 391 | } |
| 392 | |
| 393 | protected function persistDto() |
| 394 | { |
| 395 | if ($this->isFastPerformanceMode) { |
| 396 | return; |
| 397 | } |
| 398 | |
| 399 | $this->updateExtractorDto(); |
| 400 | $this->extractorTask->persistDto($this->extractorDto); |
| 401 | } |
| 402 | |
| 403 | /** |
| 404 | * @return void |
| 405 | * @throws Exception |
| 406 | */ |
| 407 | private function extractFileToDisk() |
| 408 | { |
| 409 | $this->fileBatchWrite(); |
| 410 | $isFileExtracted = $this->isExtractingFileExtracted(function ($message) { |
| 411 | $this->logger->info($message); |
| 412 | }); |
| 413 | |
| 414 | if (!$isFileExtracted) { |
| 415 | return; |
| 416 | } |
| 417 | |
| 418 | $this->validateExtractedFileAndMoveNext(); |
| 419 | } |
| 420 | |
| 421 | /** |
| 422 | * @return string|null |
| 423 | * @throws DiskNotWritableException |
| 424 | * @throws Exception |
| 425 | */ |
| 426 | private function readAndPrepareChunk() |
| 427 | { |
| 428 | try { |
| 429 | $chunk = $this->zlibCompressor->getService()->readChunk($this->wpstgFile, $this->extractingFile); |
| 430 | } catch (EmptyChunkException $ex) { |
| 431 | return null; |
| 432 | } |
| 433 | |
| 434 | if ($this->isRepairMultipleHeadersIssue) { |
| 435 | $chunk = $this->maybeRepairMultipleHeadersIssue($chunk); |
| 436 | } |
| 437 | |
| 438 | return $chunk; |
| 439 | } |
| 440 | |
| 441 | /** |
| 442 | * @return void |
| 443 | */ |
| 444 | private function updateProgressTracking(int $processedChunks, string &$lastDebugMessage) |
| 445 | { |
| 446 | if ($processedChunks % 200 === 0 || $processedChunks === $this->extractorDto->getTotalChunks()) { |
| 447 | $lastDebugMessage = sprintf('DEBUG: Extracting chunk %d/%d', $processedChunks, $this->extractorDto->getTotalChunks()); |
| 448 | } |
| 449 | } |
| 450 | |
| 451 | /** |
| 452 | * @param resource $fileResource |
| 453 | * @param string $chunk |
| 454 | * @return int |
| 455 | * @throws DiskNotWritableException |
| 456 | */ |
| 457 | private function writeChunkToFile($fileResource, string $chunk): int |
| 458 | { |
| 459 | $writtenBytes = fwrite($fileResource, $chunk, (int)$this->getScriptMemoryLimit()); |
| 460 | |
| 461 | if ($writtenBytes === false || $writtenBytes <= 0) { |
| 462 | fclose($fileResource); |
| 463 | throw DiskNotWritableException::diskNotWritable(); |
| 464 | } |
| 465 | |
| 466 | return $writtenBytes; |
| 467 | } |
| 468 | |
| 469 | private function getRecoverableCurrentSegmentBytes(string $destinationFilePath): int |
| 470 | { |
| 471 | clearstatcache(); |
| 472 | $diskSize = filesize($destinationFilePath); |
| 473 | if ($diskSize === false) { |
| 474 | return 0; |
| 475 | } |
| 476 | |
| 477 | $writtenBytes = (int) $diskSize; |
| 478 | if ($this->indexLineDto instanceof FileHeader && $this->indexLineDto->getIsPreviousPartRequired()) { |
| 479 | $writtenBytes -= $this->extractorDto->getExtractorFileBaseBytes(); |
| 480 | } |
| 481 | |
| 482 | return max(0, min($this->extractingFile->getTotalBytes(), $writtenBytes)); |
| 483 | } |
| 484 | |
| 485 | /** |
| 486 | * @return void |
| 487 | */ |
| 488 | private function trackChunkProgress(int $readBytesBefore, int $chunkSize) |
| 489 | { |
| 490 | $readBytesAfter = $this->wpstgFile->ftell() - $readBytesBefore; |
| 491 | $this->extractingFile->addReadBytes($readBytesAfter); |
| 492 | $this->extractingFile->addWrittenBytes($chunkSize); |
| 493 | } |
| 494 | |
| 495 | /** |
| 496 | * @return void |
| 497 | * @throws FileValidationException |
| 498 | */ |
| 499 | private function validateFileContent(string $fileContent, string $pathForErrorLogging) |
| 500 | { |
| 501 | $actualSize = strlen($fileContent); |
| 502 | $expectedSize = $this->indexLineDto->getUncompressedSize(); |
| 503 | if ($expectedSize !== $actualSize) { |
| 504 | throw new FileValidationException( |
| 505 | sprintf( |
| 506 | 'Filesize validation failed for file %s. Expected: %s. Actual: %s', |
| 507 | $pathForErrorLogging, |
| 508 | $this->formatSize($expectedSize, 2), |
| 509 | $this->formatSize($actualSize, 2) |
| 510 | ) |
| 511 | ); |
| 512 | } |
| 513 | |
| 514 | if (!$this->extractingFile->areHeaderBytesRemoved()) { |
| 515 | $crc32Checksum = hash(FileHeader::CRC32_CHECKSUM_ALGO, $fileContent); |
| 516 | /** @var FileHeader $fileHeader */ |
| 517 | $fileHeader = $this->indexLineDto; |
| 518 | $expectedChecksum = $fileHeader->getCrc32Checksum(); |
| 519 | if ($expectedChecksum !== $crc32Checksum) { |
| 520 | throw new FileValidationException( |
| 521 | sprintf( |
| 522 | 'CRC32 Checksum validation failed for file %s. Expected: %s. Actual: %s', |
| 523 | $pathForErrorLogging, |
| 524 | $expectedChecksum, |
| 525 | $crc32Checksum |
| 526 | ) |
| 527 | ); |
| 528 | } |
| 529 | } else { |
| 530 | $this->debugLog('Skipping validation for file because duplicate file headers were removed: ' . $pathForErrorLogging); |
| 531 | } |
| 532 | } |
| 533 | |
| 534 | /** |
| 535 | * @return void |
| 536 | */ |
| 537 | private function switchFromInMemoryToDiskExtraction(string $pathForErrorLogging) |
| 538 | { |
| 539 | $this->logger->debug(sprintf( |
| 540 | 'Threshold reached during in-memory extraction of %s. Switching to disk-based extraction on next request.', |
| 541 | $pathForErrorLogging |
| 542 | )); |
| 543 | |
| 544 | $this->extractingFile->setWrittenBytes(0); |
| 545 | } |
| 546 | |
| 547 | /** |
| 548 | * @return void |
| 549 | * @throws FileValidationException |
| 550 | * @throws Exception |
| 551 | */ |
| 552 | private function extractAndValidateInMemory() |
| 553 | { |
| 554 | $pathForErrorLogging = $this->pathIdentifier->transformIdentifiableToPath($this->indexLineDto->getIdentifiablePath()); |
| 555 | $chunks = []; |
| 556 | while (!$this->extractingFile->isFinished() && !$this->isThreshold()) { |
| 557 | $readBytesBefore = $this->wpstgFile->ftell(); |
| 558 | $chunk = $this->readAndPrepareChunk(); |
| 559 | if ($chunk === null) { |
| 560 | continue; |
| 561 | } |
| 562 | |
| 563 | $chunks[] = $chunk; |
| 564 | $this->trackChunkProgress($readBytesBefore, strlen($chunk)); |
| 565 | } |
| 566 | |
| 567 | if (!$this->extractingFile->isFinished()) { |
| 568 | $this->switchFromInMemoryToDiskExtraction($pathForErrorLogging); |
| 569 | $this->persistDto(); |
| 570 | return; |
| 571 | } |
| 572 | |
| 573 | $fileContent = implode('', $chunks); |
| 574 | $this->validateFileContent($fileContent, $pathForErrorLogging); |
| 575 | $this->moveToNextFile(); |
| 576 | } |
| 577 | |
| 578 | /** |
| 579 | * @return void |
| 580 | * @throws RuntimeException |
| 581 | */ |
| 582 | private function maybeResetFilePointerAfterInMemoryFallback() |
| 583 | { |
| 584 | if ($this->extractingFile->getWrittenBytes() !== 0 || $this->extractingFile->getReadBytes() === 0) { |
| 585 | return; |
| 586 | } |
| 587 | |
| 588 | $this->logger->debug(sprintf( |
| 589 | 'Starting disk extraction for %s after in-memory fallback (resetting state)', |
| 590 | $this->extractingFile->getRelativePath() |
| 591 | )); |
| 592 | |
| 593 | $this->extractingFile->setReadBytes(0); |
| 594 | $seekResult = $this->wpstgFile->fseek($this->extractingFile->getStart()); |
| 595 | if ($seekResult !== 0) { |
| 596 | $message = sprintf( |
| 597 | 'Failed to seek backup file to start offset %d for %s during disk extraction fallback.', |
| 598 | $this->extractingFile->getStart(), |
| 599 | $this->extractingFile->getRelativePath() |
| 600 | ); |
| 601 | |
| 602 | $this->logger->warning($message); |
| 603 | throw new RuntimeException($message); |
| 604 | } |
| 605 | } |
| 606 | } |
| 607 |