Filters
7 months ago
Scanning
5 years ago
AbstractFileObject.php
1 year ago
AbstractFilesystemScanner.php
2 months ago
DebugLogReader.php
2 years ago
DirectoryListing.php
5 months ago
DiskWriteCheck.php
5 months ago
FileObject.php
1 year ago
Filesystem.php
7 months ago
FilesystemExceptions.php
5 years ago
FilesystemScanner.php
2 weeks ago
FilesystemScannerDto.php
2 weeks ago
FilterableDirectoryIterator.php
1 year ago
LegacyFileRulesTrait.php
2 weeks ago
LogCleanup.php
5 months ago
LogFiles.php
1 year ago
MissingFileException.php
3 years ago
OPcache.php
5 months ago
PartIdentifier.php
9 months ago
PathChecker.php
2 years ago
PathIdentifier.php
1 day ago
Permissions.php
1 week ago
WpUploadsFolderSymlinker.php
2 weeks ago
PathIdentifier.php
370 lines
| 1 | <?php |
| 2 | |
| 3 | namespace WPStaging\Framework\Filesystem; |
| 4 | |
| 5 | use WPStaging\Framework\Adapter\Directory; |
| 6 | use WPStaging\Framework\Adapter\DirectoryInterface; |
| 7 | |
| 8 | /** |
| 9 | * This class is used to shorten the full file path |
| 10 | * to reduce the overall file size of the backup file. |
| 11 | * |
| 12 | * A file like wp-content/uploads/wp-staging-pro/wp-staging-pro.zip turn into |
| 13 | * wpstg_p_/wp-staging-pro/wp-staging.zip |
| 14 | * |
| 15 | * @todo rename this class to PathShortener |
| 16 | */ |
| 17 | |
| 18 | class PathIdentifier |
| 19 | { |
| 20 | /** @var string */ |
| 21 | const IDENTIFIER_ABSPATH = 'wpstg_a_'; |
| 22 | |
| 23 | /** @var string */ |
| 24 | const IDENTIFIER_WP_CONTENT = 'wpstg_c_'; |
| 25 | |
| 26 | /** @var string */ |
| 27 | const IDENTIFIER_PLUGINS = 'wpstg_p_'; |
| 28 | |
| 29 | /** @var string */ |
| 30 | const IDENTIFIER_THEMES = 'wpstg_t_'; |
| 31 | |
| 32 | /** @var string */ |
| 33 | const IDENTIFIER_MUPLUGINS = 'wpstg_m_'; |
| 34 | |
| 35 | /** @var string */ |
| 36 | const IDENTIFIER_UPLOADS = 'wpstg_u_'; |
| 37 | |
| 38 | /** @var string */ |
| 39 | const IDENTIFIER_LANG = 'wpstg_l_'; |
| 40 | |
| 41 | /** |
| 42 | * @var string|null The identifier of the last match. |
| 43 | * We will try to match the path/identifier of the next item starting from this one. It's a form of cache, |
| 44 | * making it more efficient to transform long lists of similar paths. |
| 45 | */ |
| 46 | protected $lastIdentifier; |
| 47 | |
| 48 | /** @var DirectoryInterface */ |
| 49 | protected $directory; |
| 50 | |
| 51 | public function __construct(DirectoryInterface $directory) |
| 52 | { |
| 53 | $this->directory = $directory; |
| 54 | } |
| 55 | |
| 56 | /** @var string */ |
| 57 | public function getBackupDirectory() |
| 58 | { |
| 59 | return $this->directory->getBackupDirectory(); |
| 60 | } |
| 61 | |
| 62 | /** |
| 63 | * Convert an absolute file path of a file into an abbreviated path. |
| 64 | * |
| 65 | * E.g.: |
| 66 | * |
| 67 | * /var/www/single/wp-content/plugins/index.php => wpstg_p_index.php |
| 68 | * /var/www/single/wp-content/mu-plugins/index.php => wpstg_m_index.php |
| 69 | * /var/www/single/wp-content/uploads/2019/image.png => wpstg_c_uploads/2019/image.png |
| 70 | * /var/www/single/wp-content/themes/twentytwentyone/index.php => wpstg_t_twentytwentyone/index.php |
| 71 | * |
| 72 | * @param string $path /var/www/single/wp-content/plugins/index.php |
| 73 | * |
| 74 | * @return string wpstg_p_index.php |
| 75 | */ |
| 76 | public function transformPathToIdentifiable($path) |
| 77 | { |
| 78 | // Start looking from the same placeholder as the last item, unless it was wp-content, which would cause false-positives. |
| 79 | if (isset($this->lastIdentifier) && $this->lastIdentifier !== self::IDENTIFIER_WP_CONTENT) { |
| 80 | $basePath = $this->getIdentifierPath($this->lastIdentifier); |
| 81 | |
| 82 | // Early bail: This item has the same type as the previous one. |
| 83 | if (strpos($path, $basePath) === 0) { |
| 84 | return $this->lastIdentifier . substr($path, strlen($basePath)); |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | // Uploads are usually the largest folders, so let's start with them. |
| 89 | if (strpos($path, $this->directory->getUploadsDirectory()) === 0) { |
| 90 | $this->lastIdentifier = self::IDENTIFIER_UPLOADS; |
| 91 | |
| 92 | return $this->lastIdentifier . substr($path, strlen($this->directory->getUploadsDirectory())); |
| 93 | } |
| 94 | |
| 95 | if ($this->directory->getPluginUploadsDirectory() !== $this->directory->getUploadsDirectory()) { |
| 96 | if (strpos($path, $this->directory->getPluginUploadsDirectory()) === 0) { |
| 97 | $this->lastIdentifier = self::IDENTIFIER_UPLOADS; |
| 98 | |
| 99 | return $this->lastIdentifier . substr($path, strlen($this->directory->getPluginUploadsDirectory())); |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | if (strpos($path, $this->directory->getPluginsDirectory()) === 0) { |
| 104 | $this->lastIdentifier = self::IDENTIFIER_PLUGINS; |
| 105 | |
| 106 | return $this->lastIdentifier . substr($path, strlen($this->directory->getPluginsDirectory())); |
| 107 | } |
| 108 | |
| 109 | foreach ($this->directory->getAllThemesDirectories() as $themesDirectory) { |
| 110 | if (strpos($path, $themesDirectory) === 0) { |
| 111 | $this->lastIdentifier = self::IDENTIFIER_THEMES; |
| 112 | |
| 113 | return $this->lastIdentifier . substr($path, strlen($themesDirectory)); |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | if (strpos($path, $this->directory->getMuPluginsDirectory()) === 0) { |
| 118 | $this->lastIdentifier = self::IDENTIFIER_MUPLUGINS; |
| 119 | |
| 120 | return $this->lastIdentifier . substr($path, strlen($this->directory->getMuPluginsDirectory())); |
| 121 | } |
| 122 | |
| 123 | if (strpos($path, $this->directory->getLangsDirectory()) === 0) { |
| 124 | $this->lastIdentifier = self::IDENTIFIER_LANG; |
| 125 | |
| 126 | return $this->lastIdentifier . substr($path, strlen($this->directory->getLangsDirectory())); |
| 127 | } |
| 128 | |
| 129 | if (strpos($path, $this->directory->getWpContentDirectory()) === 0) { |
| 130 | $this->lastIdentifier = self::IDENTIFIER_WP_CONTENT; |
| 131 | |
| 132 | return $this->lastIdentifier . substr($path, strlen($this->directory->getWpContentDirectory())); |
| 133 | } |
| 134 | |
| 135 | if (strpos($path, $this->directory->getAbspath()) === 0) { |
| 136 | $this->lastIdentifier = self::IDENTIFIER_ABSPATH; |
| 137 | |
| 138 | return $this->lastIdentifier . substr($path, strlen($this->directory->getAbspath())); |
| 139 | } |
| 140 | |
| 141 | // This should never happen on Backups, as we only scan the folders above explicitly and don't follow links. |
| 142 | throw new \RuntimeException(sprintf( |
| 143 | 'Could not classify %s for backup: it is not inside any known WordPress content directory (plugins, themes, mu-plugins, uploads, languages, wp-content, or the WordPress root).', |
| 144 | $path === '' ? 'an empty path' : "the path \"$path\"" |
| 145 | )); |
| 146 | } |
| 147 | |
| 148 | /** |
| 149 | * @param string $path wpstg_p_index.php |
| 150 | * |
| 151 | * @return string /var/www/single/wp-content/plugins/index.php |
| 152 | */ |
| 153 | public function transformIdentifiableToPath($path) |
| 154 | { |
| 155 | $identifier = $this->getIdentifierFromPath($path); |
| 156 | $pathWithoutIdentifier = $this->getPathWithoutIdentifier($path); |
| 157 | |
| 158 | return $this->getIdentifierPath($identifier) . $pathWithoutIdentifier; |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * @param string $path wpstg_p_index.php |
| 163 | * |
| 164 | * @return string index.php |
| 165 | */ |
| 166 | public function getPathWithoutIdentifier($path) |
| 167 | { |
| 168 | return substr($path, 8); |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * @param string $identifiablePath e.g. wpstg_u_2019/image.png |
| 173 | * |
| 174 | * @return bool True when the remainder is unsafe and the entry must be refused. |
| 175 | */ |
| 176 | public function hasPathTraversal(string $identifiablePath): bool |
| 177 | { |
| 178 | $relativePath = $this->getPathWithoutIdentifier($identifiablePath); |
| 179 | if ($relativePath === '') { |
| 180 | return true; |
| 181 | } |
| 182 | |
| 183 | if (strpos($relativePath, "\0") !== false) { |
| 184 | return true; |
| 185 | } |
| 186 | |
| 187 | $normalizedPath = str_replace('\\', '/', $relativePath); |
| 188 | if ($normalizedPath[0] === '/' || preg_match('#^[a-zA-Z]:#', $normalizedPath) === 1) { |
| 189 | return true; |
| 190 | } |
| 191 | |
| 192 | return in_array('..', explode('/', $normalizedPath), true); |
| 193 | } |
| 194 | |
| 195 | public function isPathWithinRoot(string $targetPath, string $root): bool |
| 196 | { |
| 197 | $realRoot = realpath($root); |
| 198 | if ($realRoot === false) { |
| 199 | return false; |
| 200 | } |
| 201 | |
| 202 | if (is_link($targetPath)) { |
| 203 | return false; |
| 204 | } |
| 205 | |
| 206 | $deepestExisting = $targetPath; |
| 207 | while (!file_exists($deepestExisting)) { |
| 208 | if (is_link($deepestExisting)) { |
| 209 | return false; |
| 210 | } |
| 211 | |
| 212 | $parent = dirname($deepestExisting); |
| 213 | if ($parent === $deepestExisting) { |
| 214 | return false; |
| 215 | } |
| 216 | |
| 217 | $deepestExisting = $parent; |
| 218 | } |
| 219 | |
| 220 | $realExisting = realpath($deepestExisting); |
| 221 | if ($realExisting === false) { |
| 222 | return false; |
| 223 | } |
| 224 | |
| 225 | $realExisting = rtrim($realExisting, '/\\') . DIRECTORY_SEPARATOR; |
| 226 | $realRoot = rtrim($realRoot, '/\\') . DIRECTORY_SEPARATOR; |
| 227 | |
| 228 | return strpos($realExisting, $realRoot) === 0; |
| 229 | } |
| 230 | |
| 231 | /** |
| 232 | * @param string $path wpstg_p_index.php |
| 233 | * |
| 234 | * @return string wpstg_p_ |
| 235 | */ |
| 236 | public function getIdentifierFromPath($path) |
| 237 | { |
| 238 | return substr($path, 0, 8); |
| 239 | } |
| 240 | |
| 241 | /** |
| 242 | * @return string |
| 243 | */ |
| 244 | public function transformIdentifiableToRelativePath(string $string): string |
| 245 | { |
| 246 | $string = trim($string); |
| 247 | if (empty($string)) { |
| 248 | return $string; |
| 249 | } |
| 250 | |
| 251 | $key = substr($string, 0, 8); |
| 252 | $path = $this->getRelativePath($key); |
| 253 | if ($path !== $key && is_string($path)) { |
| 254 | return substr_replace($string, $path, 0, 8); |
| 255 | } |
| 256 | |
| 257 | return $string; |
| 258 | } |
| 259 | |
| 260 | /** |
| 261 | * @return string |
| 262 | */ |
| 263 | public function getRelativePath(string $identifier): string |
| 264 | { |
| 265 | static $cache = []; |
| 266 | |
| 267 | if (!empty($cache) && !empty($identifier) && isset($cache[$identifier])) { |
| 268 | return $cache[$identifier]; |
| 269 | } |
| 270 | |
| 271 | $path = [ |
| 272 | self::IDENTIFIER_ABSPATH => '', |
| 273 | self::IDENTIFIER_WP_CONTENT => 'wp-content/', |
| 274 | self::IDENTIFIER_PLUGINS => 'wp-content/plugins/', |
| 275 | self::IDENTIFIER_THEMES => 'wp-content/themes/', |
| 276 | self::IDENTIFIER_MUPLUGINS => 'wp-content/mu-plugins/', |
| 277 | self::IDENTIFIER_UPLOADS => 'wp-content/uploads/', |
| 278 | self::IDENTIFIER_LANG => 'wp-content/languages/', |
| 279 | ]; |
| 280 | |
| 281 | if (!empty($identifier) && isset($path[$identifier])) { |
| 282 | $cache[$identifier] = $path[$identifier]; |
| 283 | return $cache[$identifier]; |
| 284 | } |
| 285 | |
| 286 | // Add __METHOD__ for debugging in wpstg-restore |
| 287 | trigger_error(sprintf('[%s] Could not find a path for the placeholder: %s', __METHOD__, filter_var($identifier, FILTER_SANITIZE_SPECIAL_CHARS))); |
| 288 | return $identifier; |
| 289 | } |
| 290 | |
| 291 | public function getAbsolutePath(string $identifier): string |
| 292 | { |
| 293 | return $this->getIdentifierPath($identifier); |
| 294 | } |
| 295 | |
| 296 | /** |
| 297 | * @return string |
| 298 | */ |
| 299 | public function getIdentifierByPartName(string $key): string |
| 300 | { |
| 301 | static $cache = []; |
| 302 | |
| 303 | if (!empty($cache) && !empty($key) && !empty($cache[$key])) { |
| 304 | return $cache[$key]; |
| 305 | } |
| 306 | |
| 307 | $list = [ |
| 308 | PartIdentifier::WP_CONTENT_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_WP_CONTENT, |
| 309 | PartIdentifier::PLUGIN_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_PLUGINS, |
| 310 | PartIdentifier::THEME_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_THEMES, |
| 311 | PartIdentifier::MU_PLUGIN_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_MUPLUGINS, |
| 312 | PartIdentifier::UPLOAD_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_UPLOADS, |
| 313 | PartIdentifier::LANGUAGE_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_LANG, |
| 314 | PartIdentifier::DATABASE_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_UPLOADS, |
| 315 | PartIdentifier::WP_ROOT_PART_IDENTIFIER => PathIdentifier::IDENTIFIER_ABSPATH, |
| 316 | ]; |
| 317 | |
| 318 | if (!empty($key) && !empty($list[$key])) { |
| 319 | $cache[$key] = $list[$key]; |
| 320 | return $cache[$key]; |
| 321 | } |
| 322 | |
| 323 | return ''; |
| 324 | } |
| 325 | |
| 326 | /** |
| 327 | * @param string $identifier wpstg_p_ |
| 328 | * |
| 329 | * @return string /var/www/single/wp-content/plugins/ |
| 330 | */ |
| 331 | protected function getIdentifierPath($identifier) |
| 332 | { |
| 333 | // It is crucial that generic paths are placed last in this list. Eg: wp-content directory must be last. |
| 334 | switch ($identifier) { |
| 335 | case self::IDENTIFIER_ABSPATH: |
| 336 | return $this->directory->getAbspath(); |
| 337 | case self::IDENTIFIER_UPLOADS: |
| 338 | return $this->directory->getUploadsDirectory(); |
| 339 | case self::IDENTIFIER_PLUGINS: |
| 340 | return $this->directory->getPluginsDirectory(); |
| 341 | case self::IDENTIFIER_THEMES: |
| 342 | return $this->directory->getActiveThemeParentDirectory(); |
| 343 | case self::IDENTIFIER_MUPLUGINS: |
| 344 | return $this->directory->getMuPluginsDirectory(); |
| 345 | case self::IDENTIFIER_LANG: |
| 346 | return $this->directory->getLangsDirectory(); |
| 347 | case self::IDENTIFIER_WP_CONTENT: |
| 348 | return $this->directory->getWpContentDirectory(); |
| 349 | default: |
| 350 | throw new \UnexpectedValueException(sprintf("[%s] Could not find a path for the placeholder: %s", __METHOD__, filter_var($identifier, FILTER_SANITIZE_SPECIAL_CHARS))); |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | /** |
| 355 | * @param string $identifiablePath wpstg_p_db.php |
| 356 | * |
| 357 | * @return bool |
| 358 | */ |
| 359 | public function hasDropinsFile(string $identifiablePath): bool |
| 360 | { |
| 361 | if (!(strpos($identifiablePath, self::IDENTIFIER_WP_CONTENT) === 0)) { |
| 362 | return false; |
| 363 | } |
| 364 | |
| 365 | $dropinsFile = implode('|', PartIdentifier::DROP_IN_FILES); |
| 366 | |
| 367 | return preg_match('@^' . self::IDENTIFIER_WP_CONTENT . '(' . $dropinsFile . ')@', $identifiablePath) ? true : false; |
| 368 | } |
| 369 | } |
| 370 |