Compression
1 year ago
Database
5 months ago
AbstractBackupsFinder.php
1 year ago
AbstractExtractor.php
5 months ago
AbstractServiceProvider.php
2 years ago
Archiver.php
5 months ago
BackupAssets.php
2 years ago
BackupContent.php
1 year ago
BackupMetadataEditor.php
1 year ago
BackupMetadataReader.php
5 months ago
BackupSigner.php
6 months ago
BackupsFinder.php
5 months ago
Extractor.php
5 months ago
FileBackupService.php
6 months ago
FileBackupServiceProvider.php
2 years ago
ServiceInterface.php
2 years ago
ZlibCompressor.php
11 months ago
Extractor.php
408 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\Queue\FinishedQueueException; |
| 32 | use WPStaging\Framework\Traits\ResourceTrait; |
| 33 | use WPStaging\Framework\Traits\RestoreFileExclusionTrait; |
| 34 | use WPStaging\Vendor\Psr\Log\LoggerInterface; |
| 35 | |
| 36 | class Extractor extends AbstractExtractor |
| 37 | { |
| 38 | use ResourceTrait; |
| 39 | use RestoreFileExclusionTrait; |
| 40 | |
| 41 | /** @var LoggerInterface */ |
| 42 | protected $logger; |
| 43 | |
| 44 | /** @var DiskWriteCheck */ |
| 45 | protected $diskWriteCheck; |
| 46 | |
| 47 | /** @var BackupValidator */ |
| 48 | protected $backupValidator; |
| 49 | |
| 50 | /** @var ZlibCompressor */ |
| 51 | protected $zlibCompressor; |
| 52 | |
| 53 | /** @var ExtractorTaskInterface */ |
| 54 | protected $extractorTask; |
| 55 | |
| 56 | /** @var bool */ |
| 57 | protected $isRepairMultipleHeadersIssue = false; |
| 58 | |
| 59 | /** @var bool */ |
| 60 | protected $isFastPerformanceMode = true; |
| 61 | |
| 62 | /** @var bool */ |
| 63 | protected $isLastRequestGracefulShutdown = true; |
| 64 | |
| 65 | public function __construct( |
| 66 | PathIdentifier $pathIdentifier, |
| 67 | Directory $directory, |
| 68 | DiskWriteCheck $diskWriteCheck, |
| 69 | ZlibCompressor $zlibCompressor, |
| 70 | BackupValidator $backupValidator, |
| 71 | BackupHeader $backupHeader, |
| 72 | Permissions $permissions |
| 73 | ) { |
| 74 | parent::__construct($pathIdentifier, $directory, $backupHeader, $permissions); |
| 75 | $this->zlibCompressor = $zlibCompressor; |
| 76 | $this->backupValidator = $backupValidator; |
| 77 | $this->diskWriteCheck = $diskWriteCheck; |
| 78 | } |
| 79 | |
| 80 | /** |
| 81 | * @param bool $isBackupFormatV1 |
| 82 | * @return void |
| 83 | */ |
| 84 | public function setIsBackupFormatV1(bool $isBackupFormatV1) |
| 85 | { |
| 86 | $this->isBackupFormatV1 = $isBackupFormatV1; |
| 87 | if ($isBackupFormatV1) { |
| 88 | $this->indexLineDto = new BackupFileIndex(); |
| 89 | } else { |
| 90 | $this->indexLineDto = WPStaging::make(FileHeader::class); |
| 91 | } |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * @param bool $isRepairMultipleHeadersIssue |
| 96 | * @return void |
| 97 | */ |
| 98 | public function setIsRepairMultipleHeadersIssue(bool $isRepairMultipleHeadersIssue) |
| 99 | { |
| 100 | $this->isRepairMultipleHeadersIssue = $isRepairMultipleHeadersIssue; |
| 101 | } |
| 102 | |
| 103 | public function setIsFastPerformanceMode(bool $isFastPerformanceMode) |
| 104 | { |
| 105 | $this->isFastPerformanceMode = $isFastPerformanceMode; |
| 106 | } |
| 107 | |
| 108 | public function setIsLastRequestGracefulShutdown(bool $isLastRequestGracefulShutdown) |
| 109 | { |
| 110 | $this->isLastRequestGracefulShutdown = $isLastRequestGracefulShutdown; |
| 111 | } |
| 112 | |
| 113 | /** |
| 114 | * @param ExtractorTaskInterface $extractorTask |
| 115 | * @param LoggerInterface $logger |
| 116 | * @return void |
| 117 | */ |
| 118 | public function inject(ExtractorTaskInterface $extractorTask, LoggerInterface $logger) |
| 119 | { |
| 120 | $this->extractorTask = $extractorTask; |
| 121 | $this->logger = $logger; |
| 122 | } |
| 123 | |
| 124 | /** |
| 125 | * @param bool $isValidateOnly |
| 126 | * @return void |
| 127 | */ |
| 128 | public function setIsValidateOnly(bool $isValidateOnly) |
| 129 | { |
| 130 | $this->isValidateOnly = $isValidateOnly; |
| 131 | if ($isValidateOnly) { |
| 132 | $this->throwExceptionOnValidationFailure = true; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | /** |
| 137 | * @return void |
| 138 | * @throws DiskNotWritableException |
| 139 | */ |
| 140 | public function execute() |
| 141 | { |
| 142 | while (!$this->isThreshold()) { |
| 143 | try { |
| 144 | $this->findFileToExtract(); |
| 145 | } catch (OutOfRangeException $e) { |
| 146 | // Done processing, or failed |
| 147 | $this->logger->warning('OutOfRangeException. Error: ' . $e->getMessage()); |
| 148 | return; |
| 149 | } catch (RuntimeException $e) { |
| 150 | $this->logger->warning($e->getMessage()); |
| 151 | continue; |
| 152 | } catch (MissingFileException $e) { |
| 153 | $this->logger->warning('MissingFileException. Error: ' . $e->getMessage()); |
| 154 | continue; |
| 155 | } catch (Exception $e) { |
| 156 | if ($e->getCode() === self::FILE_FILTERED_EXCEPTION_CODE) { |
| 157 | continue; |
| 158 | } |
| 159 | |
| 160 | if ($e->getCode() === self::FINISHED_QUEUE_EXCEPTION_CODE) { |
| 161 | throw new FinishedQueueException(); |
| 162 | } |
| 163 | |
| 164 | if ($e->getCode() === self::ITEM_SKIP_EXCEPTION_CODE) { |
| 165 | continue; |
| 166 | } |
| 167 | |
| 168 | throw $e; |
| 169 | } |
| 170 | |
| 171 | try { |
| 172 | $this->processCurrentFile(); |
| 173 | } catch (FileValidationException $e) { |
| 174 | if ($this->isValidateOnly || $this->throwExceptionOnValidationFailure) { |
| 175 | throw $e; |
| 176 | } |
| 177 | |
| 178 | $this->logger->warning('Unable to validate file. Error: ' . $e->getMessage()); |
| 179 | } |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | /** |
| 184 | * @param Exception $ex |
| 185 | * @param string $filePath |
| 186 | * @return void |
| 187 | */ |
| 188 | protected function throwMissingFileException(Exception $ex, string $filePath) |
| 189 | { |
| 190 | throw new MissingFileException(sprintf("Following backup part missing: %s", $filePath), 0, $ex); |
| 191 | } |
| 192 | |
| 193 | protected function isBigFile(): bool |
| 194 | { |
| 195 | $sizeToConsiderAsBigFile = Hooks::applyFilters('wpstg.tests.restore.bigFileSize', 10 * MB_IN_BYTES); |
| 196 | |
| 197 | return $this->extractingFile->getTotalBytes() > $sizeToConsiderAsBigFile; |
| 198 | } |
| 199 | |
| 200 | protected function cleanExistingFile(string $identifier) |
| 201 | { |
| 202 | if ($this->isValidateOnly) { |
| 203 | return; |
| 204 | } |
| 205 | |
| 206 | if ($identifier !== PathIdentifier::IDENTIFIER_UPLOADS || $this->extractingFile->getWrittenBytes() > 0) { |
| 207 | return; |
| 208 | } |
| 209 | |
| 210 | if (file_exists($this->extractingFile->getBackupPath())) { |
| 211 | // Delete the original upload file |
| 212 | if (!unlink($this->extractingFile->getBackupPath())) { |
| 213 | throw new \RuntimeException(sprintf(__('Could not delete original media library file %s. Skipping restore of it...', 'wp-staging'), $this->extractingFile->getRelativePath())); |
| 214 | } |
| 215 | } |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * Fixes issue https://github.com/wp-staging/wp-staging-pro/issues/2861 |
| 220 | * @return void |
| 221 | */ |
| 222 | protected function maybeRemoveLastAccidentalCharFromLastExtractedFile() |
| 223 | { |
| 224 | if ($this->backupMetadata->getTotalFiles() !== $this->extractorDto->getTotalFilesExtracted()) { |
| 225 | return; |
| 226 | } |
| 227 | |
| 228 | if ($this->backupValidator->validateFileIndexFirstLine($this->wpstgFile, $this->backupMetadata)) { |
| 229 | return; |
| 230 | } |
| 231 | |
| 232 | $this->removeLastCharInExtractedFile(); |
| 233 | } |
| 234 | |
| 235 | protected function getExtractFolder(string $identifier): string |
| 236 | { |
| 237 | if ($this->isValidateOnly) { |
| 238 | return trailingslashit($this->dirRestore . self::VALIDATE_DIRECTORY); |
| 239 | } |
| 240 | |
| 241 | if ($identifier === PathIdentifier::IDENTIFIER_UPLOADS) { |
| 242 | return $this->directory->getUploadsDirectory(); |
| 243 | } |
| 244 | |
| 245 | return $this->dirRestore . $identifier; |
| 246 | } |
| 247 | |
| 248 | /** |
| 249 | * @return void |
| 250 | * @throws DiskNotWritableException |
| 251 | */ |
| 252 | private function processCurrentFile() |
| 253 | { |
| 254 | $destinationFilePath = $this->extractingFile->getBackupPath(); |
| 255 | if ($this->currentIdentifier === PathIdentifier::IDENTIFIER_UPLOADS && $this->isExcludedFile($destinationFilePath)) { |
| 256 | $this->extractorDto->incrementTotalFilesSkipped(); |
| 257 | $this->extractorDto->setCurrentIndexOffset($this->wpstgIndexOffsetForNextFile); |
| 258 | return; |
| 259 | } |
| 260 | |
| 261 | if ($this->extractingFile->getWrittenBytes() > 0) { |
| 262 | $this->logger->debug(sprintf('Resuming extraction of file %s from byte %d. Total size: %d...', $this->extractingFile->getRelativePath(), $this->extractingFile->getWrittenBytes(), $this->extractingFile->getTotalBytes())); |
| 263 | } |
| 264 | |
| 265 | try { |
| 266 | if ($this->isThreshold()) { |
| 267 | // Prevent considering a file as big just because we start extracting at the threshold |
| 268 | return; |
| 269 | } |
| 270 | |
| 271 | $this->fileBatchWrite(); |
| 272 | |
| 273 | $isFileExtracted = $this->isExtractingFileExtracted(function ($message) { |
| 274 | $this->logger->info($message); |
| 275 | }); |
| 276 | |
| 277 | if (!$isFileExtracted) { |
| 278 | return; |
| 279 | } |
| 280 | } catch (DiskNotWritableException $e) { |
| 281 | // Re-throw |
| 282 | throw $e; |
| 283 | } catch (OutOfRangeException $e) { |
| 284 | // Backup header, should be ignored silently |
| 285 | $this->extractingFile->setWrittenBytes($this->extractingFile->getTotalBytes()); |
| 286 | } catch (Exception $e) { |
| 287 | // Set this file as "written", so that we can skip to the next file. |
| 288 | $this->extractingFile->setWrittenBytes($this->extractingFile->getTotalBytes()); |
| 289 | |
| 290 | if (defined('WPSTG_DEBUG') && WPSTG_DEBUG) { |
| 291 | $this->logger->warning(sprintf('Skipped file %s. Reason: %s', $this->extractingFile->getRelativePath(), $e->getMessage())); |
| 292 | } |
| 293 | } |
| 294 | |
| 295 | $this->validateExtractedFileAndMoveNext(); |
| 296 | if ($this->isFastPerformanceMode) { |
| 297 | return; |
| 298 | } |
| 299 | |
| 300 | $this->extractorTask->persistDto($this->extractorDto); |
| 301 | } |
| 302 | |
| 303 | /** |
| 304 | * @return void |
| 305 | * @throws DiskNotWritableException |
| 306 | * @throws \WPStaging\Framework\Filesystem\FilesystemExceptions |
| 307 | */ |
| 308 | private function fileBatchWrite() |
| 309 | { |
| 310 | $destinationFilePath = $this->extractingFile->getBackupPath(); |
| 311 | |
| 312 | if (strpos($destinationFilePath, '.sql') !== false) { |
| 313 | $this->logger->debug(sprintf('DEBUG: Extracting SQL file %s', $destinationFilePath)); |
| 314 | } |
| 315 | |
| 316 | wp_mkdir_p(dirname($destinationFilePath)); |
| 317 | |
| 318 | /** |
| 319 | * On some servers, it is required to create empty file first, so we will create empty files. |
| 320 | * On some servers, touch doesn't work consistently, so we will use fwrite, see the reason below. |
| 321 | * On sites hosted on SiteGround, creating files using file_puts_contents uses a lot of memory, |
| 322 | * so by default we will use fwrite to create the empty file. |
| 323 | * If creating the empty file using fwrite fails, let try creating it using file_put_contents |
| 324 | * @see https://github.com/wp-staging/wp-staging-pro/issues/3272 why it was needed. |
| 325 | */ |
| 326 | if (!$this->createEmptyFile($destinationFilePath)) { |
| 327 | file_put_contents($destinationFilePath, ''); |
| 328 | } |
| 329 | |
| 330 | $destinationFileResource = @fopen($destinationFilePath, FileObject::MODE_APPEND); |
| 331 | if (!$destinationFileResource) { |
| 332 | $this->diskWriteCheck->testDiskIsWriteable(); |
| 333 | throw new Exception("Can not extract file $destinationFilePath"); |
| 334 | } |
| 335 | |
| 336 | /** |
| 337 | * When last request is not graceful shutdown and it is not fast performance mode (i.e. safe performance mode), |
| 338 | * we need to set the file pointer to the correct position in the backup file to continue extraction from where it left off. |
| 339 | * But this solution only works for non-compressed backups |
| 340 | */ |
| 341 | if (!$this->isLastRequestGracefulShutdown && !$this->isFastPerformanceMode && !$this->extractingFile->getIsCompressed()) { |
| 342 | $fileSize = filesize($destinationFilePath); |
| 343 | $this->wpstgFile->fseek($this->extractingFile->getStart() + $fileSize); |
| 344 | $this->extractingFile->setReadBytes($fileSize); |
| 345 | $this->extractingFile->setWrittenBytes($fileSize); |
| 346 | $this->logger->debug(sprintf('DEBUG: Seeking to byte %d in backup file to continue extraction of %s...', $this->extractingFile->getStart() + $fileSize, $this->extractingFile->getRelativePath())); |
| 347 | } |
| 348 | |
| 349 | $lastDebugMessage = ''; |
| 350 | while (!$this->extractingFile->isFinished() && !$this->isThreshold()) { |
| 351 | $readBytesBefore = $this->wpstgFile->ftell(); |
| 352 | |
| 353 | $chunk = null; |
| 354 | try { |
| 355 | $chunk = $this->zlibCompressor->getService()->readChunk($this->wpstgFile, $this->extractingFile, function ($currentChunkNumber) use (&$lastDebugMessage) { |
| 356 | // Log every 200 chunks to provide progress updates without overwhelming the logs. |
| 357 | if ($currentChunkNumber % 200 === 0 || $currentChunkNumber === $this->extractorDto->getTotalChunks()) { |
| 358 | $lastDebugMessage = sprintf('DEBUG: Extracting chunk %d/%d', $currentChunkNumber, $this->extractorDto->getTotalChunks()); |
| 359 | } |
| 360 | }); |
| 361 | } catch (DiskNotWritableException $ex) { |
| 362 | $this->diskWriteCheck->testDiskIsWriteable(); |
| 363 | // If empty chunk, it is an empty file, so we can skip it |
| 364 | throw new Exception("Unable to extract file to $destinationFilePath. Please check if there is enough disk space available."); |
| 365 | } catch (EmptyChunkException $ex) { |
| 366 | // If empty chunk, it is an empty file, so we can skip it |
| 367 | continue; |
| 368 | } |
| 369 | |
| 370 | if ($this->isRepairMultipleHeadersIssue) { |
| 371 | $chunk = $this->maybeRepairMultipleHeadersIssue($chunk); |
| 372 | } |
| 373 | |
| 374 | $writtenBytes = fwrite($destinationFileResource, $chunk, (int)$this->getScriptMemoryLimit()); |
| 375 | |
| 376 | if ($writtenBytes === false || $writtenBytes <= 0) { |
| 377 | fclose($destinationFileResource); |
| 378 | $destinationFileResource = null; |
| 379 | throw DiskNotWritableException::diskNotWritable(); |
| 380 | } |
| 381 | |
| 382 | $readBytesAfter = $this->wpstgFile->ftell() - $readBytesBefore; |
| 383 | |
| 384 | $this->extractingFile->addReadBytes($readBytesAfter); |
| 385 | $this->extractingFile->addWrittenBytes($writtenBytes); |
| 386 | |
| 387 | $this->persistDto(); |
| 388 | } |
| 389 | |
| 390 | if (!empty($lastDebugMessage)) { |
| 391 | $this->logger->debug($lastDebugMessage); |
| 392 | } |
| 393 | |
| 394 | fclose($destinationFileResource); |
| 395 | $destinationFileResource = null; |
| 396 | } |
| 397 | |
| 398 | protected function persistDto() |
| 399 | { |
| 400 | if ($this->isFastPerformanceMode) { |
| 401 | return; |
| 402 | } |
| 403 | |
| 404 | $this->updateExtractorDto(); |
| 405 | $this->extractorTask->persistDto($this->extractorDto); |
| 406 | } |
| 407 | } |
| 408 |