Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
26.37% |
235 / 891 |
|
0.00% |
0 / 72 |
CRAP | |
0.00% |
0 / 1 |
| UploadBase | |
26.37% |
235 / 891 |
|
0.00% |
0 / 72 |
45123.65 | |
0.00% |
0 / 1 |
| getVerificationErrorCode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isEnabled | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isAllowed | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| isThrottled | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| createFromRequest | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
56 | |||
| isValidRequest | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDesiredDestName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getSourceType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| initializePathInfo | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| initializeFromRequest | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| setTempFile | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
30 | |||
| fetchFile | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| canFetchFile | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isEmptyFile | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFileSize | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getTempFileSha1Base36 | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| getRealPath | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| verifyUpload | |
50.00% |
9 / 18 |
|
0.00% |
0 / 1 |
8.12 | |||
| validateName | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
30 | |||
| verifyMimeType | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| verifyFile | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
210 | |||
| verifyPartialFile | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
90 | |||
| zipEntryCallback | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| verifyPermissions | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| verifyTitlePermissions | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
20 | |||
| checkWarnings | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
90 | |||
| makeWarningsSerializable | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| unserializeWarnings | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
30 | |||
| checkBadFileName | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| checkUnwantedFileExtensions | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| checkFileSize | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
20 | |||
| checkLocalFileExists | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
42 | |||
| checkLocalFileWasDeleted | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| checkAgainstExistingDupes | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
42 | |||
| checkAgainstArchiveDupes | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| performUpload | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
42 | |||
| postProcessUpload | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getTitle | |
67.65% |
46 / 68 |
|
0.00% |
0 / 1 |
35.93 | |||
| getLocalFile | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| getStashFile | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| tryStashFile | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| runUploadStashFileHook | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| doStashFile | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| cleanupTempFile | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| getTempPath | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| splitExtensions | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| checkFileExtension | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| checkFileExtensionList | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| verifyExtension | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
72 | |||
| detectScript | |
67.39% |
31 / 46 |
|
0.00% |
0 / 1 |
24.88 | |||
| checkXMLEncodingMissmatch | |
55.56% |
15 / 27 |
|
0.00% |
0 / 1 |
27.84 | |||
| detectScriptInSvg | |
88.24% |
15 / 17 |
|
0.00% |
0 / 1 |
5.04 | |||
| checkSvgPICallback | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| checkSvgExternalDTD | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
20 | |||
| checkSvgScriptCallback | |
90.84% |
119 / 131 |
|
0.00% |
0 / 1 |
45.49 | |||
| checkCssFragment | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
56 | |||
| splitXmlNamespace | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| stripXmlNamespace | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| detectVirus | |
0.00% |
0 / 47 |
|
0.00% |
0 / 1 |
240 | |||
| checkOverwrite | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
30 | |||
| userCanReUpload | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| getExistsWarning | |
0.00% |
0 / 53 |
|
0.00% |
0 / 1 |
132 | |||
| isThumbName | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
20 | |||
| getFilenamePrefixBlacklist | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
42 | |||
| getImageInfo | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| convertVerifyErrorToStatus | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| getMaxUploadSize | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
20 | |||
| getMaxPhpUploadSize | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
6 | |||
| getSessionStatus | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| setSessionStatus | |
0.00% |
0 / 35 |
|
0.00% |
0 / 1 |
30 | |||
| getUploadSessionKey | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| getUploadSessionStore | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Base class for the backend of file upload. |
| 4 | * |
| 5 | * This program is free software; you can redistribute it and/or modify |
| 6 | * it under the terms of the GNU General Public License as published by |
| 7 | * the Free Software Foundation; either version 2 of the License, or |
| 8 | * (at your option) any later version. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 | * GNU General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License along |
| 16 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 17 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 18 | * http://www.gnu.org/copyleft/gpl.html |
| 19 | * |
| 20 | * @file |
| 21 | * @ingroup Upload |
| 22 | */ |
| 23 | |
| 24 | use MediaWiki\Context\RequestContext; |
| 25 | use MediaWiki\HookContainer\HookRunner; |
| 26 | use MediaWiki\HookContainer\ProtectedHookAccessorTrait; |
| 27 | use MediaWiki\Logger\LoggerFactory; |
| 28 | use MediaWiki\MainConfigNames; |
| 29 | use MediaWiki\MediaWikiServices; |
| 30 | use MediaWiki\Parser\Sanitizer; |
| 31 | use MediaWiki\Permissions\Authority; |
| 32 | use MediaWiki\Permissions\PermissionStatus; |
| 33 | use MediaWiki\Request\WebRequest; |
| 34 | use MediaWiki\Shell\Shell; |
| 35 | use MediaWiki\Status\Status; |
| 36 | use MediaWiki\Title\Title; |
| 37 | use MediaWiki\User\User; |
| 38 | use MediaWiki\User\UserIdentity; |
| 39 | use Wikimedia\AtEase\AtEase; |
| 40 | |
| 41 | /** |
| 42 | * @defgroup Upload Upload related |
| 43 | */ |
| 44 | |
| 45 | /** |
| 46 | * @ingroup Upload |
| 47 | * |
| 48 | * UploadBase and subclasses are the backend of MediaWiki's file uploads. |
| 49 | * The frontends are formed by ApiUpload and SpecialUpload. |
| 50 | * |
| 51 | * @stable to extend |
| 52 | * |
| 53 | * @author Brooke Vibber |
| 54 | * @author Bryan Tong Minh |
| 55 | * @author Michael Dale |
| 56 | */ |
| 57 | abstract class UploadBase { |
| 58 | use ProtectedHookAccessorTrait; |
| 59 | |
| 60 | /** @var string|null Local file system path to the file to upload (or a local copy) */ |
| 61 | protected $mTempPath; |
| 62 | /** @var TempFSFile|null Wrapper to handle deleting the temp file */ |
| 63 | protected $tempFileObj; |
| 64 | /** @var string|null */ |
| 65 | protected $mDesiredDestName; |
| 66 | /** @var string|null */ |
| 67 | protected $mDestName; |
| 68 | /** @var bool|null */ |
| 69 | protected $mRemoveTempFile; |
| 70 | /** @var string|null */ |
| 71 | protected $mSourceType; |
| 72 | /** @var Title|false|null */ |
| 73 | protected $mTitle = false; |
| 74 | /** @var int */ |
| 75 | protected $mTitleError = 0; |
| 76 | /** @var string|null */ |
| 77 | protected $mFilteredName; |
| 78 | /** @var string|null */ |
| 79 | protected $mFinalExtension; |
| 80 | /** @var LocalFile|null */ |
| 81 | protected $mLocalFile; |
| 82 | /** @var UploadStashFile|null */ |
| 83 | protected $mStashFile; |
| 84 | /** @var int|null */ |
| 85 | protected $mFileSize; |
| 86 | /** @var array|null */ |
| 87 | protected $mFileProps; |
| 88 | /** @var string[] */ |
| 89 | protected $mBlackListedExtensions; |
| 90 | /** @var bool|null */ |
| 91 | protected $mJavaDetected; |
| 92 | /** @var string|false */ |
| 93 | protected $mSVGNSError; |
| 94 | |
| 95 | protected static $safeXmlEncodings = [ |
| 96 | 'UTF-8', |
| 97 | 'US-ASCII', |
| 98 | 'ISO-8859-1', |
| 99 | 'ISO-8859-2', |
| 100 | 'UTF-16', |
| 101 | 'UTF-32', |
| 102 | 'WINDOWS-1250', |
| 103 | 'WINDOWS-1251', |
| 104 | 'WINDOWS-1252', |
| 105 | 'WINDOWS-1253', |
| 106 | 'WINDOWS-1254', |
| 107 | 'WINDOWS-1255', |
| 108 | 'WINDOWS-1256', |
| 109 | 'WINDOWS-1257', |
| 110 | 'WINDOWS-1258', |
| 111 | ]; |
| 112 | |
| 113 | public const SUCCESS = 0; |
| 114 | public const OK = 0; |
| 115 | public const EMPTY_FILE = 3; |
| 116 | public const MIN_LENGTH_PARTNAME = 4; |
| 117 | public const ILLEGAL_FILENAME = 5; |
| 118 | public const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions() |
| 119 | public const FILETYPE_MISSING = 8; |
| 120 | public const FILETYPE_BADTYPE = 9; |
| 121 | public const VERIFICATION_ERROR = 10; |
| 122 | public const HOOK_ABORTED = 11; |
| 123 | public const FILE_TOO_LARGE = 12; |
| 124 | public const WINDOWS_NONASCII_FILENAME = 13; |
| 125 | public const FILENAME_TOO_LONG = 14; |
| 126 | |
| 127 | private const CODE_TO_STATUS = [ |
| 128 | self::EMPTY_FILE => 'empty-file', |
| 129 | self::FILE_TOO_LARGE => 'file-too-large', |
| 130 | self::FILETYPE_MISSING => 'filetype-missing', |
| 131 | self::FILETYPE_BADTYPE => 'filetype-banned', |
| 132 | self::MIN_LENGTH_PARTNAME => 'filename-tooshort', |
| 133 | self::ILLEGAL_FILENAME => 'illegal-filename', |
| 134 | self::OVERWRITE_EXISTING_FILE => 'overwrite', |
| 135 | self::VERIFICATION_ERROR => 'verification-error', |
| 136 | self::HOOK_ABORTED => 'hookaborted', |
| 137 | self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename', |
| 138 | self::FILENAME_TOO_LONG => 'filename-toolong', |
| 139 | ]; |
| 140 | |
| 141 | /** |
| 142 | * @param int $error |
| 143 | * @return string |
| 144 | */ |
| 145 | public function getVerificationErrorCode( $error ) { |
| 146 | return self::CODE_TO_STATUS[$error] ?? 'unknown-error'; |
| 147 | } |
| 148 | |
| 149 | /** |
| 150 | * Returns true if uploads are enabled. |
| 151 | * Can be override by subclasses. |
| 152 | * @stable to override |
| 153 | * @return bool |
| 154 | */ |
| 155 | public static function isEnabled() { |
| 156 | $enableUploads = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::EnableUploads ); |
| 157 | |
| 158 | return $enableUploads && wfIniGetBool( 'file_uploads' ); |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * Returns true if the user can use this upload module or else a string |
| 163 | * identifying the missing permission. |
| 164 | * Can be overridden by subclasses. |
| 165 | * |
| 166 | * @param Authority $performer |
| 167 | * @return bool|string |
| 168 | */ |
| 169 | public static function isAllowed( Authority $performer ) { |
| 170 | foreach ( [ 'upload', 'edit' ] as $permission ) { |
| 171 | if ( !$performer->isAllowed( $permission ) ) { |
| 172 | return $permission; |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | return true; |
| 177 | } |
| 178 | |
| 179 | /** |
| 180 | * Returns true if the user has surpassed the upload rate limit, false otherwise. |
| 181 | * |
| 182 | * @deprecated since 1.41, use verifyTitlePermissions() instead. |
| 183 | * Rate limit checks are now implicit in permission checks. |
| 184 | * |
| 185 | * @param User $user |
| 186 | * @return bool |
| 187 | */ |
| 188 | public static function isThrottled( $user ) { |
| 189 | wfDeprecated( __METHOD__, '1.41' ); |
| 190 | return $user->pingLimiter( 'upload' ); |
| 191 | } |
| 192 | |
| 193 | /** @var string[] Upload handlers. Should probably just be a configuration variable. */ |
| 194 | private static $uploadHandlers = [ 'Stash', 'File', 'Url' ]; |
| 195 | |
| 196 | /** |
| 197 | * Create a form of UploadBase depending on wpSourceType and initializes it. |
| 198 | * |
| 199 | * @param WebRequest &$request |
| 200 | * @param string|null $type |
| 201 | * @return null|self |
| 202 | */ |
| 203 | public static function createFromRequest( &$request, $type = null ) { |
| 204 | $type = $type ?: $request->getVal( 'wpSourceType', 'File' ); |
| 205 | |
| 206 | if ( !$type ) { |
| 207 | return null; |
| 208 | } |
| 209 | |
| 210 | // Get the upload class |
| 211 | $type = ucfirst( $type ); |
| 212 | |
| 213 | // Give hooks the chance to handle this request |
| 214 | /** @var self|null $className */ |
| 215 | $className = null; |
| 216 | ( new HookRunner( MediaWikiServices::getInstance()->getHookContainer() ) ) |
| 217 | // @phan-suppress-next-line PhanTypeMismatchArgument Type mismatch on pass-by-ref args |
| 218 | ->onUploadCreateFromRequest( $type, $className ); |
| 219 | if ( $className === null ) { |
| 220 | $className = 'UploadFrom' . $type; |
| 221 | wfDebug( __METHOD__ . ": class name: $className" ); |
| 222 | if ( !in_array( $type, self::$uploadHandlers ) ) { |
| 223 | return null; |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | if ( !$className::isEnabled() || !$className::isValidRequest( $request ) ) { |
| 228 | return null; |
| 229 | } |
| 230 | |
| 231 | /** @var self $handler */ |
| 232 | $handler = new $className; |
| 233 | |
| 234 | $handler->initializeFromRequest( $request ); |
| 235 | |
| 236 | return $handler; |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * Check whether a request if valid for this handler. |
| 241 | * @param WebRequest $request |
| 242 | * @return bool |
| 243 | */ |
| 244 | public static function isValidRequest( $request ) { |
| 245 | return false; |
| 246 | } |
| 247 | |
| 248 | /** |
| 249 | * Get the desired destination name. |
| 250 | * @return string|null |
| 251 | */ |
| 252 | public function getDesiredDestName() { |
| 253 | return $this->mDesiredDestName; |
| 254 | } |
| 255 | |
| 256 | /** |
| 257 | * @stable to call |
| 258 | */ |
| 259 | public function __construct() { |
| 260 | } |
| 261 | |
| 262 | /** |
| 263 | * Returns the upload type. Should be overridden by child classes. |
| 264 | * |
| 265 | * @since 1.18 |
| 266 | * @stable to override |
| 267 | * @return string|null |
| 268 | */ |
| 269 | public function getSourceType() { |
| 270 | return null; |
| 271 | } |
| 272 | |
| 273 | /** |
| 274 | * @param string $name The desired destination name |
| 275 | * @param string|null $tempPath Callers should make sure this is not a storage path |
| 276 | * @param int|null $fileSize |
| 277 | * @param bool $removeTempFile (false) remove the temporary file? |
| 278 | */ |
| 279 | public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) { |
| 280 | $this->mDesiredDestName = $name; |
| 281 | if ( FileBackend::isStoragePath( $tempPath ) ) { |
| 282 | throw new InvalidArgumentException( __METHOD__ . " given storage path `$tempPath`." ); |
| 283 | } |
| 284 | |
| 285 | $this->setTempFile( $tempPath, $fileSize ); |
| 286 | $this->mRemoveTempFile = $removeTempFile; |
| 287 | } |
| 288 | |
| 289 | /** |
| 290 | * Initialize from a WebRequest. Override this in a subclass. |
| 291 | * |
| 292 | * @param WebRequest &$request |
| 293 | */ |
| 294 | abstract public function initializeFromRequest( &$request ); |
| 295 | |
| 296 | /** |
| 297 | * @param string|null $tempPath File system path to temporary file containing the upload |
| 298 | * @param int|null $fileSize |
| 299 | */ |
| 300 | protected function setTempFile( $tempPath, $fileSize = null ) { |
| 301 | $this->mTempPath = $tempPath ?? ''; |
| 302 | $this->mFileSize = $fileSize ?: null; |
| 303 | $this->mFileProps = null; |
| 304 | if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) { |
| 305 | $this->tempFileObj = new TempFSFile( $this->mTempPath ); |
| 306 | if ( !$fileSize ) { |
| 307 | $this->mFileSize = filesize( $this->mTempPath ); |
| 308 | } |
| 309 | } else { |
| 310 | $this->tempFileObj = null; |
| 311 | } |
| 312 | } |
| 313 | |
| 314 | /** |
| 315 | * Fetch the file. Usually a no-op. |
| 316 | * @stable to override |
| 317 | * @return Status |
| 318 | */ |
| 319 | public function fetchFile() { |
| 320 | return Status::newGood(); |
| 321 | } |
| 322 | |
| 323 | /** |
| 324 | * Perform checks to see if the file can be fetched. Usually a no-op. |
| 325 | * @stable to override |
| 326 | * @return Status |
| 327 | */ |
| 328 | public function canFetchFile() { |
| 329 | return Status::newGood(); |
| 330 | } |
| 331 | |
| 332 | /** |
| 333 | * Return true if the file is empty. |
| 334 | * @return bool |
| 335 | */ |
| 336 | public function isEmptyFile() { |
| 337 | return !$this->mFileSize; |
| 338 | } |
| 339 | |
| 340 | /** |
| 341 | * Return the file size. |
| 342 | * @return int |
| 343 | */ |
| 344 | public function getFileSize() { |
| 345 | return $this->mFileSize; |
| 346 | } |
| 347 | |
| 348 | /** |
| 349 | * Get the base 36 SHA1 of the file. |
| 350 | * @stable to override |
| 351 | * @return string|false |
| 352 | */ |
| 353 | public function getTempFileSha1Base36() { |
| 354 | // Use cached version if we already have it. |
| 355 | if ( $this->mFileProps && is_string( $this->mFileProps['sha1'] ) ) { |
| 356 | return $this->mFileProps['sha1']; |
| 357 | } |
| 358 | return FSFile::getSha1Base36FromPath( $this->mTempPath ); |
| 359 | } |
| 360 | |
| 361 | /** |
| 362 | * @param string $srcPath The source path |
| 363 | * @return string|false The real path if it was a virtual URL Returns false on failure |
| 364 | */ |
| 365 | public function getRealPath( $srcPath ) { |
| 366 | $repo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo(); |
| 367 | if ( FileRepo::isVirtualUrl( $srcPath ) ) { |
| 368 | /** @todo Just make uploads work with storage paths UploadFromStash |
| 369 | * loads files via virtual URLs. |
| 370 | */ |
| 371 | $tmpFile = $repo->getLocalCopy( $srcPath ); |
| 372 | if ( $tmpFile ) { |
| 373 | $tmpFile->bind( $this ); // keep alive with $this |
| 374 | } |
| 375 | $path = $tmpFile ? $tmpFile->getPath() : false; |
| 376 | } else { |
| 377 | $path = $srcPath; |
| 378 | } |
| 379 | |
| 380 | return $path; |
| 381 | } |
| 382 | |
| 383 | /** |
| 384 | * Verify whether the upload is sensible. |
| 385 | * |
| 386 | * Return a status array representing the outcome of the verification. |
| 387 | * Possible keys are: |
| 388 | * - 'status': set to self::OK in case of success, or to one of the error constants defined in |
| 389 | * this class in case of failure |
| 390 | * - 'max': set to the maximum allowed file size ($wgMaxUploadSize) if the upload is too large |
| 391 | * - 'details': set to error details if the file type is valid but contents are corrupt |
| 392 | * - 'filtered': set to the sanitized file name if the requested file name is invalid |
| 393 | * - 'finalExt': set to the file's file extension if it is not an allowed file extension |
| 394 | * - 'blacklistedExt': set to the list of disallowed file extensions if the current file extension |
| 395 | * is not allowed for uploads and the list is not empty |
| 396 | * |
| 397 | * @stable to override |
| 398 | * @return mixed[] array representing the result of the verification |
| 399 | */ |
| 400 | public function verifyUpload() { |
| 401 | /** |
| 402 | * If there was no filename or a zero size given, give up quick. |
| 403 | */ |
| 404 | if ( $this->isEmptyFile() ) { |
| 405 | return [ 'status' => self::EMPTY_FILE ]; |
| 406 | } |
| 407 | |
| 408 | /** |
| 409 | * Honor $wgMaxUploadSize |
| 410 | */ |
| 411 | $maxSize = self::getMaxUploadSize( $this->getSourceType() ); |
| 412 | if ( $this->mFileSize > $maxSize ) { |
| 413 | return [ |
| 414 | 'status' => self::FILE_TOO_LARGE, |
| 415 | 'max' => $maxSize, |
| 416 | ]; |
| 417 | } |
| 418 | |
| 419 | /** |
| 420 | * Look at the contents of the file; if we can recognize the |
| 421 | * type, but it's corrupt or data of the wrong type, we should |
| 422 | * probably not accept it. |
| 423 | */ |
| 424 | $verification = $this->verifyFile(); |
| 425 | if ( $verification !== true ) { |
| 426 | return [ |
| 427 | 'status' => self::VERIFICATION_ERROR, |
| 428 | 'details' => $verification |
| 429 | ]; |
| 430 | } |
| 431 | |
| 432 | /** |
| 433 | * Make sure this file can be created |
| 434 | */ |
| 435 | $result = $this->validateName(); |
| 436 | if ( $result !== true ) { |
| 437 | return $result; |
| 438 | } |
| 439 | |
| 440 | return [ 'status' => self::OK ]; |
| 441 | } |
| 442 | |
| 443 | /** |
| 444 | * Verify that the name is valid and, if necessary, that we can overwrite |
| 445 | * |
| 446 | * @return array|bool True if valid, otherwise an array with 'status' |
| 447 | * and other keys |
| 448 | */ |
| 449 | public function validateName() { |
| 450 | $nt = $this->getTitle(); |
| 451 | if ( $nt === null ) { |
| 452 | $result = [ 'status' => $this->mTitleError ]; |
| 453 | if ( $this->mTitleError === self::ILLEGAL_FILENAME ) { |
| 454 | $result['filtered'] = $this->mFilteredName; |
| 455 | } |
| 456 | if ( $this->mTitleError === self::FILETYPE_BADTYPE ) { |
| 457 | $result['finalExt'] = $this->mFinalExtension; |
| 458 | if ( count( $this->mBlackListedExtensions ) ) { |
| 459 | $result['blacklistedExt'] = $this->mBlackListedExtensions; |
| 460 | } |
| 461 | } |
| 462 | |
| 463 | return $result; |
| 464 | } |
| 465 | $this->mDestName = $this->getLocalFile()->getName(); |
| 466 | |
| 467 | return true; |
| 468 | } |
| 469 | |
| 470 | /** |
| 471 | * Verify the MIME type. |
| 472 | * |
| 473 | * @note Only checks that it is not an evil MIME. |
| 474 | * The "does it have the correct file extension given its MIME type?" check is in verifyFile. |
| 475 | * @param string $mime Representing the MIME |
| 476 | * @return array|bool True if the file is verified, an array otherwise |
| 477 | */ |
| 478 | protected function verifyMimeType( $mime ) { |
| 479 | $verifyMimeType = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::VerifyMimeType ); |
| 480 | if ( $verifyMimeType ) { |
| 481 | wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>" ); |
| 482 | $mimeTypeExclusions = MediaWikiServices::getInstance()->getMainConfig() |
| 483 | ->get( MainConfigNames::MimeTypeExclusions ); |
| 484 | if ( self::checkFileExtension( $mime, $mimeTypeExclusions ) ) { |
| 485 | return [ 'filetype-badmime', $mime ]; |
| 486 | } |
| 487 | } |
| 488 | |
| 489 | return true; |
| 490 | } |
| 491 | |
| 492 | /** |
| 493 | * Verifies that it's ok to include the uploaded file |
| 494 | * |
| 495 | * @return array|true True of the file is verified, array otherwise. |
| 496 | */ |
| 497 | protected function verifyFile() { |
| 498 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 499 | $verifyMimeType = $config->get( MainConfigNames::VerifyMimeType ); |
| 500 | $disableUploadScriptChecks = $config->get( MainConfigNames::DisableUploadScriptChecks ); |
| 501 | $status = $this->verifyPartialFile(); |
| 502 | if ( $status !== true ) { |
| 503 | return $status; |
| 504 | } |
| 505 | |
| 506 | // Calculating props calculates the sha1 which is expensive. |
| 507 | // reuse props if we already have them |
| 508 | if ( !is_array( $this->mFileProps ) ) { |
| 509 | $mwProps = new MWFileProps( MediaWikiServices::getInstance()->getMimeAnalyzer() ); |
| 510 | $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension ); |
| 511 | } |
| 512 | $mime = $this->mFileProps['mime']; |
| 513 | |
| 514 | if ( $verifyMimeType ) { |
| 515 | # XXX: Missing extension will be caught by validateName() via getTitle() |
| 516 | if ( (string)$this->mFinalExtension !== '' && |
| 517 | !self::verifyExtension( $mime, $this->mFinalExtension ) |
| 518 | ) { |
| 519 | return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ]; |
| 520 | } |
| 521 | } |
| 522 | |
| 523 | # check for htmlish code and javascript |
| 524 | if ( !$disableUploadScriptChecks ) { |
| 525 | if ( $this->mFinalExtension === 'svg' || $mime === 'image/svg+xml' ) { |
| 526 | $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false ); |
| 527 | if ( $svgStatus !== false ) { |
| 528 | return $svgStatus; |
| 529 | } |
| 530 | } |
| 531 | } |
| 532 | |
| 533 | $handler = MediaHandler::getHandler( $mime ); |
| 534 | if ( $handler ) { |
| 535 | $handlerStatus = $handler->verifyUpload( $this->mTempPath ); |
| 536 | if ( !$handlerStatus->isOK() ) { |
| 537 | $errors = $handlerStatus->getErrorsArray(); |
| 538 | |
| 539 | return reset( $errors ); |
| 540 | } |
| 541 | } |
| 542 | |
| 543 | $error = true; |
| 544 | $this->getHookRunner()->onUploadVerifyFile( $this, $mime, $error ); |
| 545 | if ( $error !== true ) { |
| 546 | if ( !is_array( $error ) ) { |
| 547 | $error = [ $error ]; |
| 548 | } |
| 549 | return $error; |
| 550 | } |
| 551 | |
| 552 | wfDebug( __METHOD__ . ": all clear; passing." ); |
| 553 | |
| 554 | return true; |
| 555 | } |
| 556 | |
| 557 | /** |
| 558 | * A verification routine suitable for partial files |
| 559 | * |
| 560 | * Runs the deny list checks, but not any checks that may |
| 561 | * assume the entire file is present. |
| 562 | * |
| 563 | * @return array|true True, if the file is valid, else an array with error message key. |
| 564 | * @phan-return non-empty-array|true |
| 565 | */ |
| 566 | protected function verifyPartialFile() { |
| 567 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 568 | $disableUploadScriptChecks = $config->get( MainConfigNames::DisableUploadScriptChecks ); |
| 569 | # getTitle() sets some internal parameters like $this->mFinalExtension |
| 570 | $this->getTitle(); |
| 571 | |
| 572 | // Calculating props calculates the sha1 which is expensive. |
| 573 | // reuse props if we already have them (e.g. During stashed upload) |
| 574 | if ( !is_array( $this->mFileProps ) ) { |
| 575 | $mwProps = new MWFileProps( MediaWikiServices::getInstance()->getMimeAnalyzer() ); |
| 576 | $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension ); |
| 577 | } |
| 578 | |
| 579 | # check MIME type, if desired |
| 580 | $mime = $this->mFileProps['file-mime']; |
| 581 | $status = $this->verifyMimeType( $mime ); |
| 582 | if ( $status !== true ) { |
| 583 | return $status; |
| 584 | } |
| 585 | |
| 586 | # check for htmlish code and javascript |
| 587 | if ( !$disableUploadScriptChecks ) { |
| 588 | if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) { |
| 589 | return [ 'uploadscripted' ]; |
| 590 | } |
| 591 | if ( $this->mFinalExtension === 'svg' || $mime === 'image/svg+xml' ) { |
| 592 | $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true ); |
| 593 | if ( $svgStatus !== false ) { |
| 594 | return $svgStatus; |
| 595 | } |
| 596 | } |
| 597 | } |
| 598 | |
| 599 | # Scan the uploaded file for viruses |
| 600 | $virus = self::detectVirus( $this->mTempPath ); |
| 601 | if ( $virus ) { |
| 602 | return [ 'uploadvirus', $virus ]; |
| 603 | } |
| 604 | |
| 605 | return true; |
| 606 | } |
| 607 | |
| 608 | /** |
| 609 | * Callback for ZipDirectoryReader to detect Java class files. |
| 610 | * |
| 611 | * @param array $entry |
| 612 | */ |
| 613 | public function zipEntryCallback( $entry ) { |
| 614 | $names = [ $entry['name'] ]; |
| 615 | |
| 616 | // If there is a null character, cut off the name at it, because JDK's |
| 617 | // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name |
| 618 | // were constructed which had ".class\0" followed by a string chosen to |
| 619 | // make the hash collide with the truncated name, that file could be |
| 620 | // returned in response to a request for the .class file. |
| 621 | $nullPos = strpos( $entry['name'], "\000" ); |
| 622 | if ( $nullPos !== false ) { |
| 623 | $names[] = substr( $entry['name'], 0, $nullPos ); |
| 624 | } |
| 625 | |
| 626 | // If there is a trailing slash in the file name, we have to strip it, |
| 627 | // because that's what ZIP_GetEntry() does. |
| 628 | if ( preg_grep( '!\.class/?$!', $names ) ) { |
| 629 | $this->mJavaDetected = true; |
| 630 | } |
| 631 | } |
| 632 | |
| 633 | /** |
| 634 | * Alias for verifyTitlePermissions. The function was originally |
| 635 | * 'verifyPermissions', but that suggests it's checking the user, when it's |
| 636 | * really checking the title + user combination. |
| 637 | * |
| 638 | * @param Authority $performer to verify the permissions against |
| 639 | * @return array|bool An array as returned by getPermissionErrors or true |
| 640 | * in case the user has proper permissions. |
| 641 | */ |
| 642 | public function verifyPermissions( Authority $performer ) { |
| 643 | return $this->verifyTitlePermissions( $performer ); |
| 644 | } |
| 645 | |
| 646 | /** |
| 647 | * Check whether the user can edit, upload and create the image. This |
| 648 | * checks only against the current title; if it returns errors, it may |
| 649 | * very well be that another title will not give errors. Therefore |
| 650 | * isAllowed() should be called as well for generic is-user-blocked or |
| 651 | * can-user-upload checking. |
| 652 | * |
| 653 | * @param Authority $performer to verify the permissions against |
| 654 | * @return array|bool An array as returned by getPermissionErrors or true |
| 655 | * in case the user has proper permissions. |
| 656 | */ |
| 657 | public function verifyTitlePermissions( Authority $performer ) { |
| 658 | /** |
| 659 | * If the image is protected, non-sysop users won't be able |
| 660 | * to modify it by uploading a new revision. |
| 661 | */ |
| 662 | $nt = $this->getTitle(); |
| 663 | if ( $nt === null ) { |
| 664 | return true; |
| 665 | } |
| 666 | |
| 667 | $status = PermissionStatus::newEmpty(); |
| 668 | $performer->authorizeWrite( 'edit', $nt, $status ); |
| 669 | $performer->authorizeWrite( 'upload', $nt, $status ); |
| 670 | if ( !$status->isGood() ) { |
| 671 | return $status->toLegacyErrorArray(); |
| 672 | } |
| 673 | |
| 674 | $overwriteError = $this->checkOverwrite( $performer ); |
| 675 | if ( $overwriteError !== true ) { |
| 676 | return [ $overwriteError ]; |
| 677 | } |
| 678 | |
| 679 | return true; |
| 680 | } |
| 681 | |
| 682 | /** |
| 683 | * Check for non fatal problems with the file. |
| 684 | * |
| 685 | * This should not assume that mTempPath is set. |
| 686 | * |
| 687 | * @param User|null $user Accepted since 1.35 |
| 688 | * |
| 689 | * @return mixed[] Array of warnings |
| 690 | */ |
| 691 | public function checkWarnings( $user = null ) { |
| 692 | if ( $user === null ) { |
| 693 | // TODO check uses and hard deprecate |
| 694 | $user = RequestContext::getMain()->getUser(); |
| 695 | } |
| 696 | |
| 697 | $warnings = []; |
| 698 | |
| 699 | $localFile = $this->getLocalFile(); |
| 700 | $localFile->load( IDBAccessObject::READ_LATEST ); |
| 701 | $filename = $localFile->getName(); |
| 702 | $hash = $this->getTempFileSha1Base36(); |
| 703 | |
| 704 | $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName ); |
| 705 | if ( $badFileName !== null ) { |
| 706 | $warnings['badfilename'] = $badFileName; |
| 707 | } |
| 708 | |
| 709 | $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( (string)$this->mFinalExtension ); |
| 710 | if ( $unwantedFileExtensionDetails !== null ) { |
| 711 | $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails; |
| 712 | } |
| 713 | |
| 714 | $fileSizeWarnings = $this->checkFileSize( $this->mFileSize ); |
| 715 | if ( $fileSizeWarnings ) { |
| 716 | $warnings = array_merge( $warnings, $fileSizeWarnings ); |
| 717 | } |
| 718 | |
| 719 | $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash ); |
| 720 | if ( $localFileExistsWarnings ) { |
| 721 | $warnings = array_merge( $warnings, $localFileExistsWarnings ); |
| 722 | } |
| 723 | |
| 724 | if ( $this->checkLocalFileWasDeleted( $localFile ) ) { |
| 725 | $warnings['was-deleted'] = $filename; |
| 726 | } |
| 727 | |
| 728 | // If a file with the same name exists locally then the local file has already been tested |
| 729 | // for duplication of content |
| 730 | $ignoreLocalDupes = isset( $warnings['exists'] ); |
| 731 | $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes ); |
| 732 | if ( $dupes ) { |
| 733 | $warnings['duplicate'] = $dupes; |
| 734 | } |
| 735 | |
| 736 | $archivedDupes = $this->checkAgainstArchiveDupes( $hash, $user ); |
| 737 | if ( $archivedDupes !== null ) { |
| 738 | $warnings['duplicate-archive'] = $archivedDupes; |
| 739 | } |
| 740 | |
| 741 | return $warnings; |
| 742 | } |
| 743 | |
| 744 | /** |
| 745 | * Convert the warnings array returned by checkWarnings() to something that |
| 746 | * can be serialized. File objects will be converted to an associative array |
| 747 | * with the following keys: |
| 748 | * |
| 749 | * - fileName: The name of the file |
| 750 | * - timestamp: The upload timestamp |
| 751 | * |
| 752 | * @param mixed[] $warnings |
| 753 | * @return mixed[] |
| 754 | */ |
| 755 | public static function makeWarningsSerializable( $warnings ) { |
| 756 | array_walk_recursive( $warnings, static function ( &$param, $key ) { |
| 757 | if ( $param instanceof File ) { |
| 758 | $param = [ |
| 759 | 'fileName' => $param->getName(), |
| 760 | 'timestamp' => $param->getTimestamp() |
| 761 | ]; |
| 762 | } elseif ( is_object( $param ) ) { |
| 763 | throw new InvalidArgumentException( |
| 764 | 'UploadBase::makeWarningsSerializable: ' . |
| 765 | 'Unexpected object of class ' . get_class( $param ) ); |
| 766 | } |
| 767 | } ); |
| 768 | return $warnings; |
| 769 | } |
| 770 | |
| 771 | /** |
| 772 | * Convert the serialized warnings array created by makeWarningsSerializable() |
| 773 | * back to the output of checkWarnings(). |
| 774 | * |
| 775 | * @param mixed[] $warnings |
| 776 | * @return mixed[] |
| 777 | */ |
| 778 | public static function unserializeWarnings( $warnings ) { |
| 779 | foreach ( $warnings as $key => $value ) { |
| 780 | if ( is_array( $value ) ) { |
| 781 | if ( isset( $value['fileName'] ) && isset( $value['timestamp'] ) ) { |
| 782 | $warnings[$key] = MediaWikiServices::getInstance()->getRepoGroup()->findFile( |
| 783 | $value['fileName'], |
| 784 | [ 'time' => $value['timestamp'] ] |
| 785 | ); |
| 786 | } else { |
| 787 | $warnings[$key] = self::unserializeWarnings( $value ); |
| 788 | } |
| 789 | } |
| 790 | } |
| 791 | return $warnings; |
| 792 | } |
| 793 | |
| 794 | /** |
| 795 | * Check whether the resulting filename is different from the desired one, |
| 796 | * but ignore things like ucfirst() and spaces/underscore things |
| 797 | * |
| 798 | * @param string $filename |
| 799 | * @param string $desiredFileName |
| 800 | * |
| 801 | * @return string|null String that was determined to be bad or null if the filename is okay |
| 802 | */ |
| 803 | private function checkBadFileName( $filename, $desiredFileName ) { |
| 804 | $comparableName = str_replace( ' ', '_', $desiredFileName ); |
| 805 | $comparableName = Title::capitalize( $comparableName, NS_FILE ); |
| 806 | |
| 807 | if ( $desiredFileName != $filename && $comparableName != $filename ) { |
| 808 | return $filename; |
| 809 | } |
| 810 | |
| 811 | return null; |
| 812 | } |
| 813 | |
| 814 | /** |
| 815 | * @param string $fileExtension The file extension to check |
| 816 | * |
| 817 | * @return array|null array with the following keys: |
| 818 | * 0 => string The final extension being used |
| 819 | * 1 => string[] The extensions that are allowed |
| 820 | * 2 => int The number of extensions that are allowed. |
| 821 | */ |
| 822 | private function checkUnwantedFileExtensions( $fileExtension ) { |
| 823 | $checkFileExtensions = MediaWikiServices::getInstance()->getMainConfig() |
| 824 | ->get( MainConfigNames::CheckFileExtensions ); |
| 825 | $fileExtensions = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::FileExtensions ); |
| 826 | if ( $checkFileExtensions ) { |
| 827 | $extensions = array_unique( $fileExtensions ); |
| 828 | if ( !self::checkFileExtension( $fileExtension, $extensions ) ) { |
| 829 | return [ |
| 830 | $fileExtension, |
| 831 | Message::listParam( $extensions, 'comma' ), |
| 832 | count( $extensions ) |
| 833 | ]; |
| 834 | } |
| 835 | } |
| 836 | |
| 837 | return null; |
| 838 | } |
| 839 | |
| 840 | /** |
| 841 | * @param int $fileSize |
| 842 | * |
| 843 | * @return array warnings |
| 844 | */ |
| 845 | private function checkFileSize( $fileSize ) { |
| 846 | $uploadSizeWarning = MediaWikiServices::getInstance()->getMainConfig() |
| 847 | ->get( MainConfigNames::UploadSizeWarning ); |
| 848 | |
| 849 | $warnings = []; |
| 850 | |
| 851 | if ( $uploadSizeWarning && ( $fileSize > $uploadSizeWarning ) ) { |
| 852 | $warnings['large-file'] = [ |
| 853 | Message::sizeParam( $uploadSizeWarning ), |
| 854 | Message::sizeParam( $fileSize ), |
| 855 | ]; |
| 856 | } |
| 857 | |
| 858 | if ( $fileSize == 0 ) { |
| 859 | $warnings['empty-file'] = true; |
| 860 | } |
| 861 | |
| 862 | return $warnings; |
| 863 | } |
| 864 | |
| 865 | /** |
| 866 | * @param LocalFile $localFile |
| 867 | * @param string|false $hash sha1 hash of the file to check |
| 868 | * |
| 869 | * @return array warnings |
| 870 | */ |
| 871 | private function checkLocalFileExists( LocalFile $localFile, $hash ) { |
| 872 | $warnings = []; |
| 873 | |
| 874 | $exists = self::getExistsWarning( $localFile ); |
| 875 | if ( $exists !== false ) { |
| 876 | $warnings['exists'] = $exists; |
| 877 | |
| 878 | // check if file is an exact duplicate of current file version |
| 879 | if ( $hash !== false && $hash === $localFile->getSha1() ) { |
| 880 | $warnings['no-change'] = $localFile; |
| 881 | } |
| 882 | |
| 883 | // check if file is an exact duplicate of older versions of this file |
| 884 | $history = $localFile->getHistory(); |
| 885 | foreach ( $history as $oldFile ) { |
| 886 | if ( $hash === $oldFile->getSha1() ) { |
| 887 | $warnings['duplicate-version'][] = $oldFile; |
| 888 | } |
| 889 | } |
| 890 | } |
| 891 | |
| 892 | return $warnings; |
| 893 | } |
| 894 | |
| 895 | private function checkLocalFileWasDeleted( LocalFile $localFile ) { |
| 896 | return $localFile->wasDeleted() && !$localFile->exists(); |
| 897 | } |
| 898 | |
| 899 | /** |
| 900 | * @param string|false $hash sha1 hash of the file to check |
| 901 | * @param bool $ignoreLocalDupes True to ignore local duplicates |
| 902 | * |
| 903 | * @return File[] Duplicate files, if found. |
| 904 | */ |
| 905 | private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) { |
| 906 | if ( $hash === false ) { |
| 907 | return []; |
| 908 | } |
| 909 | $dupes = MediaWikiServices::getInstance()->getRepoGroup()->findBySha1( $hash ); |
| 910 | $title = $this->getTitle(); |
| 911 | foreach ( $dupes as $key => $dupe ) { |
| 912 | if ( |
| 913 | ( $dupe instanceof LocalFile ) && |
| 914 | $ignoreLocalDupes && |
| 915 | $title->equals( $dupe->getTitle() ) |
| 916 | ) { |
| 917 | unset( $dupes[$key] ); |
| 918 | } |
| 919 | } |
| 920 | |
| 921 | return $dupes; |
| 922 | } |
| 923 | |
| 924 | /** |
| 925 | * @param string|false $hash sha1 hash of the file to check |
| 926 | * @param Authority $performer |
| 927 | * |
| 928 | * @return string|null Name of the dupe or empty string if discovered (depending on visibility) |
| 929 | * null if the check discovered no dupes. |
| 930 | */ |
| 931 | private function checkAgainstArchiveDupes( $hash, Authority $performer ) { |
| 932 | if ( $hash === false ) { |
| 933 | return null; |
| 934 | } |
| 935 | $archivedFile = new ArchivedFile( null, 0, '', $hash ); |
| 936 | if ( $archivedFile->getID() > 0 ) { |
| 937 | if ( $archivedFile->userCan( File::DELETED_FILE, $performer ) ) { |
| 938 | return $archivedFile->getName(); |
| 939 | } |
| 940 | return ''; |
| 941 | } |
| 942 | |
| 943 | return null; |
| 944 | } |
| 945 | |
| 946 | /** |
| 947 | * Really perform the upload. Stores the file in the local repo, watches |
| 948 | * if necessary and runs the UploadComplete hook. |
| 949 | * |
| 950 | * @param string $comment |
| 951 | * @param string|false $pageText |
| 952 | * @param bool $watch Whether the file page should be added to user's watchlist. |
| 953 | * (This doesn't check $user's permissions.) |
| 954 | * @param User $user |
| 955 | * @param string[] $tags Change tags to add to the log entry and page revision. |
| 956 | * (This doesn't check $user's permissions.) |
| 957 | * @param string|null $watchlistExpiry Optional watchlist expiry timestamp in any format |
| 958 | * acceptable to wfTimestamp(). |
| 959 | * @return Status Indicating the whether the upload succeeded. |
| 960 | * |
| 961 | * @since 1.35 Accepts $watchlistExpiry parameter. |
| 962 | */ |
| 963 | public function performUpload( |
| 964 | $comment, $pageText, $watch, $user, $tags = [], ?string $watchlistExpiry = null |
| 965 | ) { |
| 966 | $this->getLocalFile()->load( IDBAccessObject::READ_LATEST ); |
| 967 | $props = $this->mFileProps; |
| 968 | |
| 969 | $error = null; |
| 970 | $this->getHookRunner()->onUploadVerifyUpload( $this, $user, $props, $comment, $pageText, $error ); |
| 971 | if ( $error ) { |
| 972 | if ( !is_array( $error ) ) { |
| 973 | $error = [ $error ]; |
| 974 | } |
| 975 | return Status::newFatal( ...$error ); |
| 976 | } |
| 977 | |
| 978 | $status = $this->getLocalFile()->upload( |
| 979 | $this->mTempPath, |
| 980 | $comment, |
| 981 | $pageText !== false ? $pageText : '', |
| 982 | File::DELETE_SOURCE, |
| 983 | $props, |
| 984 | false, |
| 985 | $user, |
| 986 | $tags |
| 987 | ); |
| 988 | |
| 989 | if ( $status->isGood() ) { |
| 990 | if ( $watch ) { |
| 991 | MediaWikiServices::getInstance()->getWatchlistManager()->addWatchIgnoringRights( |
| 992 | $user, |
| 993 | $this->getLocalFile()->getTitle(), |
| 994 | $watchlistExpiry |
| 995 | ); |
| 996 | } |
| 997 | $this->getHookRunner()->onUploadComplete( $this ); |
| 998 | |
| 999 | $this->postProcessUpload(); |
| 1000 | } |
| 1001 | |
| 1002 | return $status; |
| 1003 | } |
| 1004 | |
| 1005 | /** |
| 1006 | * Perform extra steps after a successful upload. |
| 1007 | * |
| 1008 | * @stable to override |
| 1009 | * @since 1.25 |
| 1010 | */ |
| 1011 | public function postProcessUpload() { |
| 1012 | } |
| 1013 | |
| 1014 | /** |
| 1015 | * Returns the title of the file to be uploaded. Sets mTitleError in case |
| 1016 | * the name was illegal. |
| 1017 | * |
| 1018 | * @return Title|null The title of the file or null in case the name was illegal |
| 1019 | */ |
| 1020 | public function getTitle() { |
| 1021 | if ( $this->mTitle !== false ) { |
| 1022 | return $this->mTitle; |
| 1023 | } |
| 1024 | if ( !is_string( $this->mDesiredDestName ) ) { |
| 1025 | $this->mTitleError = self::ILLEGAL_FILENAME; |
| 1026 | $this->mTitle = null; |
| 1027 | |
| 1028 | return $this->mTitle; |
| 1029 | } |
| 1030 | /* Assume that if a user specified File:Something.jpg, this is an error |
| 1031 | * and that the namespace prefix needs to be stripped of. |
| 1032 | */ |
| 1033 | $title = Title::newFromText( $this->mDesiredDestName ); |
| 1034 | if ( $title && $title->getNamespace() === NS_FILE ) { |
| 1035 | $this->mFilteredName = $title->getDBkey(); |
| 1036 | } else { |
| 1037 | $this->mFilteredName = $this->mDesiredDestName; |
| 1038 | } |
| 1039 | |
| 1040 | # oi_archive_name is max 255 bytes, which include a timestamp and an |
| 1041 | # exclamation mark, so restrict file name to 240 bytes. |
| 1042 | if ( strlen( $this->mFilteredName ) > 240 ) { |
| 1043 | $this->mTitleError = self::FILENAME_TOO_LONG; |
| 1044 | $this->mTitle = null; |
| 1045 | |
| 1046 | return $this->mTitle; |
| 1047 | } |
| 1048 | |
| 1049 | /** |
| 1050 | * Chop off any directories in the given filename. Then |
| 1051 | * filter out illegal characters, and try to make a legible name |
| 1052 | * out of it. We'll strip some silently that Title would die on. |
| 1053 | */ |
| 1054 | $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName ); |
| 1055 | /* Normalize to title form before we do any further processing */ |
| 1056 | $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName ); |
| 1057 | if ( $nt === null ) { |
| 1058 | $this->mTitleError = self::ILLEGAL_FILENAME; |
| 1059 | $this->mTitle = null; |
| 1060 | |
| 1061 | return $this->mTitle; |
| 1062 | } |
| 1063 | $this->mFilteredName = $nt->getDBkey(); |
| 1064 | |
| 1065 | /** |
| 1066 | * We'll want to prevent against *any* 'extension', and use |
| 1067 | * only the final one for the allow list. |
| 1068 | */ |
| 1069 | [ $partname, $ext ] = self::splitExtensions( $this->mFilteredName ); |
| 1070 | |
| 1071 | if ( $ext !== [] ) { |
| 1072 | $this->mFinalExtension = trim( end( $ext ) ); |
| 1073 | } else { |
| 1074 | $this->mFinalExtension = ''; |
| 1075 | |
| 1076 | // No extension, try guessing one from the temporary file |
| 1077 | // FIXME: Sometimes we mTempPath isn't set yet here, possibly due to an unrealistic |
| 1078 | // or incomplete test case in UploadBaseTest (T272328) |
| 1079 | if ( $this->mTempPath !== null ) { |
| 1080 | $magic = MediaWikiServices::getInstance()->getMimeAnalyzer(); |
| 1081 | $mime = $magic->guessMimeType( $this->mTempPath ); |
| 1082 | if ( $mime !== 'unknown/unknown' ) { |
| 1083 | # Get a space separated list of extensions |
| 1084 | $mimeExt = $magic->getExtensionFromMimeTypeOrNull( $mime ); |
| 1085 | if ( $mimeExt !== null ) { |
| 1086 | # Set the extension to the canonical extension |
| 1087 | $this->mFinalExtension = $mimeExt; |
| 1088 | |
| 1089 | # Fix up the other variables |
| 1090 | $this->mFilteredName .= ".{$this->mFinalExtension}"; |
| 1091 | $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName ); |
| 1092 | $ext = [ $this->mFinalExtension ]; |
| 1093 | } |
| 1094 | } |
| 1095 | } |
| 1096 | } |
| 1097 | |
| 1098 | // Don't allow users to override the list of prohibited file extensions (check file extension) |
| 1099 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 1100 | $checkFileExtensions = $config->get( MainConfigNames::CheckFileExtensions ); |
| 1101 | $strictFileExtensions = $config->get( MainConfigNames::StrictFileExtensions ); |
| 1102 | $fileExtensions = $config->get( MainConfigNames::FileExtensions ); |
| 1103 | $prohibitedFileExtensions = $config->get( MainConfigNames::ProhibitedFileExtensions ); |
| 1104 | |
| 1105 | $badList = self::checkFileExtensionList( $ext, $prohibitedFileExtensions ); |
| 1106 | |
| 1107 | if ( $this->mFinalExtension == '' ) { |
| 1108 | $this->mTitleError = self::FILETYPE_MISSING; |
| 1109 | $this->mTitle = null; |
| 1110 | |
| 1111 | return $this->mTitle; |
| 1112 | } |
| 1113 | |
| 1114 | if ( $badList || |
| 1115 | ( $checkFileExtensions && $strictFileExtensions && |
| 1116 | !self::checkFileExtension( $this->mFinalExtension, $fileExtensions ) ) |
| 1117 | ) { |
| 1118 | $this->mBlackListedExtensions = $badList; |
| 1119 | $this->mTitleError = self::FILETYPE_BADTYPE; |
| 1120 | $this->mTitle = null; |
| 1121 | |
| 1122 | return $this->mTitle; |
| 1123 | } |
| 1124 | |
| 1125 | // Windows may be broken with special characters, see T3780 |
| 1126 | if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) |
| 1127 | && !MediaWikiServices::getInstance()->getRepoGroup() |
| 1128 | ->getLocalRepo()->backendSupportsUnicodePaths() |
| 1129 | ) { |
| 1130 | $this->mTitleError = self::WINDOWS_NONASCII_FILENAME; |
| 1131 | $this->mTitle = null; |
| 1132 | |
| 1133 | return $this->mTitle; |
| 1134 | } |
| 1135 | |
| 1136 | # If there was more than one file "extension", reassemble the base |
| 1137 | # filename to prevent bogus complaints about length |
| 1138 | if ( count( $ext ) > 1 ) { |
| 1139 | $iterations = count( $ext ) - 1; |
| 1140 | for ( $i = 0; $i < $iterations; $i++ ) { |
| 1141 | $partname .= '.' . $ext[$i]; |
| 1142 | } |
| 1143 | } |
| 1144 | |
| 1145 | if ( strlen( $partname ) < 1 ) { |
| 1146 | $this->mTitleError = self::MIN_LENGTH_PARTNAME; |
| 1147 | $this->mTitle = null; |
| 1148 | |
| 1149 | return $this->mTitle; |
| 1150 | } |
| 1151 | |
| 1152 | $this->mTitle = $nt; |
| 1153 | |
| 1154 | return $this->mTitle; |
| 1155 | } |
| 1156 | |
| 1157 | /** |
| 1158 | * Return the local file and initializes if necessary. |
| 1159 | * |
| 1160 | * @stable to override |
| 1161 | * @return LocalFile|null |
| 1162 | */ |
| 1163 | public function getLocalFile() { |
| 1164 | if ( $this->mLocalFile === null ) { |
| 1165 | $nt = $this->getTitle(); |
| 1166 | $this->mLocalFile = $nt === null |
| 1167 | ? null |
| 1168 | : MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo()->newFile( $nt ); |
| 1169 | } |
| 1170 | |
| 1171 | return $this->mLocalFile; |
| 1172 | } |
| 1173 | |
| 1174 | /** |
| 1175 | * @return UploadStashFile|null |
| 1176 | */ |
| 1177 | public function getStashFile() { |
| 1178 | return $this->mStashFile; |
| 1179 | } |
| 1180 | |
| 1181 | /** |
| 1182 | * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must |
| 1183 | * be called before calling this method (unless $isPartial is true). |
| 1184 | * |
| 1185 | * Upload stash exceptions are also caught and converted to an error status. |
| 1186 | * |
| 1187 | * @since 1.28 |
| 1188 | * @stable to override |
| 1189 | * @param User $user |
| 1190 | * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file). |
| 1191 | * @return Status If successful, value is an UploadStashFile instance |
| 1192 | */ |
| 1193 | public function tryStashFile( User $user, $isPartial = false ) { |
| 1194 | if ( !$isPartial ) { |
| 1195 | $error = $this->runUploadStashFileHook( $user ); |
| 1196 | if ( $error ) { |
| 1197 | return Status::newFatal( ...$error ); |
| 1198 | } |
| 1199 | } |
| 1200 | try { |
| 1201 | $file = $this->doStashFile( $user ); |
| 1202 | return Status::newGood( $file ); |
| 1203 | } catch ( UploadStashException $e ) { |
| 1204 | return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() ); |
| 1205 | } |
| 1206 | } |
| 1207 | |
| 1208 | /** |
| 1209 | * @param User $user |
| 1210 | * @return array|null Error message and parameters, null if there's no error |
| 1211 | */ |
| 1212 | protected function runUploadStashFileHook( User $user ) { |
| 1213 | $props = $this->mFileProps; |
| 1214 | $error = null; |
| 1215 | $this->getHookRunner()->onUploadStashFile( $this, $user, $props, $error ); |
| 1216 | if ( $error && !is_array( $error ) ) { |
| 1217 | $error = [ $error ]; |
| 1218 | } |
| 1219 | return $error; |
| 1220 | } |
| 1221 | |
| 1222 | /** |
| 1223 | * Implementation for stashFile() and tryStashFile(). |
| 1224 | * |
| 1225 | * @stable to override |
| 1226 | * @param User|null $user |
| 1227 | * @return UploadStashFile Stashed file |
| 1228 | */ |
| 1229 | protected function doStashFile( User $user = null ) { |
| 1230 | $stash = MediaWikiServices::getInstance()->getRepoGroup() |
| 1231 | ->getLocalRepo()->getUploadStash( $user ); |
| 1232 | $file = $stash->stashFile( $this->mTempPath, $this->getSourceType(), $this->mFileProps ); |
| 1233 | $this->mStashFile = $file; |
| 1234 | |
| 1235 | return $file; |
| 1236 | } |
| 1237 | |
| 1238 | /** |
| 1239 | * If we've modified the upload file, then we need to manually remove it |
| 1240 | * on exit to clean up. |
| 1241 | */ |
| 1242 | public function cleanupTempFile() { |
| 1243 | if ( $this->mRemoveTempFile && $this->tempFileObj ) { |
| 1244 | // Delete when all relevant TempFSFile handles go out of scope |
| 1245 | wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal" ); |
| 1246 | $this->tempFileObj->autocollect(); |
| 1247 | } |
| 1248 | } |
| 1249 | |
| 1250 | /** |
| 1251 | * @return string|null |
| 1252 | */ |
| 1253 | public function getTempPath() { |
| 1254 | return $this->mTempPath; |
| 1255 | } |
| 1256 | |
| 1257 | /** |
| 1258 | * Split a file into a base name and all dot-delimited 'extensions' |
| 1259 | * on the end. Some web server configurations will fall back to |
| 1260 | * earlier pseudo-'extensions' to determine type and execute |
| 1261 | * scripts, so we need to check them all. |
| 1262 | * |
| 1263 | * @param string $filename |
| 1264 | * @return array [ string, string[] ] |
| 1265 | */ |
| 1266 | public static function splitExtensions( $filename ) { |
| 1267 | $bits = explode( '.', $filename ); |
| 1268 | $basename = array_shift( $bits ); |
| 1269 | |
| 1270 | return [ $basename, $bits ]; |
| 1271 | } |
| 1272 | |
| 1273 | /** |
| 1274 | * Perform case-insensitive match against a list of file extensions. |
| 1275 | * |
| 1276 | * @param string $ext File extension |
| 1277 | * @param array $list |
| 1278 | * @return bool Returns true if the extension is in the list. |
| 1279 | */ |
| 1280 | public static function checkFileExtension( $ext, $list ) { |
| 1281 | return in_array( strtolower( $ext ?? '' ), $list, true ); |
| 1282 | } |
| 1283 | |
| 1284 | /** |
| 1285 | * Perform case-insensitive match against a list of file extensions. |
| 1286 | * Returns an array of matching extensions. |
| 1287 | * |
| 1288 | * @param string[] $ext File extensions |
| 1289 | * @param string[] $list |
| 1290 | * @return string[] |
| 1291 | */ |
| 1292 | public static function checkFileExtensionList( $ext, $list ) { |
| 1293 | return array_intersect( array_map( 'strtolower', $ext ), $list ); |
| 1294 | } |
| 1295 | |
| 1296 | /** |
| 1297 | * Checks if the MIME type of the uploaded file matches the file extension. |
| 1298 | * |
| 1299 | * @param string $mime The MIME type of the uploaded file |
| 1300 | * @param string $extension The filename extension that the file is to be served with |
| 1301 | * @return bool |
| 1302 | */ |
| 1303 | public static function verifyExtension( $mime, $extension ) { |
| 1304 | $magic = MediaWikiServices::getInstance()->getMimeAnalyzer(); |
| 1305 | |
| 1306 | if ( !$mime || $mime === 'unknown' || $mime === 'unknown/unknown' ) { |
| 1307 | if ( !$magic->isRecognizableExtension( $extension ) ) { |
| 1308 | wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " . |
| 1309 | "unrecognized extension '$extension', can't verify" ); |
| 1310 | |
| 1311 | return true; |
| 1312 | } |
| 1313 | |
| 1314 | wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " . |
| 1315 | "recognized extension '$extension', so probably invalid file" ); |
| 1316 | return false; |
| 1317 | } |
| 1318 | |
| 1319 | $match = $magic->isMatchingExtension( $extension, $mime ); |
| 1320 | |
| 1321 | if ( $match === null ) { |
| 1322 | if ( $magic->getMimeTypesFromExtension( $extension ) !== [] ) { |
| 1323 | wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension" ); |
| 1324 | |
| 1325 | return false; |
| 1326 | } |
| 1327 | |
| 1328 | wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file" ); |
| 1329 | return true; |
| 1330 | } |
| 1331 | |
| 1332 | if ( $match ) { |
| 1333 | wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file" ); |
| 1334 | |
| 1335 | /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */ |
| 1336 | return true; |
| 1337 | } |
| 1338 | |
| 1339 | wfDebug( __METHOD__ |
| 1340 | . ": mime type $mime mismatches file extension $extension, rejecting file" ); |
| 1341 | |
| 1342 | return false; |
| 1343 | } |
| 1344 | |
| 1345 | /** |
| 1346 | * Heuristic for detecting files that *could* contain JavaScript instructions or |
| 1347 | * things that may look like HTML to a browser and are thus |
| 1348 | * potentially harmful. The present implementation will produce false |
| 1349 | * positives in some situations. |
| 1350 | * |
| 1351 | * @param string|null $file Pathname to the temporary upload file |
| 1352 | * @param string $mime The MIME type of the file |
| 1353 | * @param string|null $extension The extension of the file |
| 1354 | * @return bool True if the file contains something looking like embedded scripts |
| 1355 | */ |
| 1356 | public static function detectScript( $file, $mime, $extension ) { |
| 1357 | # ugly hack: for text files, always look at the entire file. |
| 1358 | # For binary field, just check the first K. |
| 1359 | |
| 1360 | if ( str_starts_with( $mime ?? '', 'text/' ) ) { |
| 1361 | $chunk = file_get_contents( $file ); |
| 1362 | } else { |
| 1363 | $fp = fopen( $file, 'rb' ); |
| 1364 | if ( !$fp ) { |
| 1365 | return false; |
| 1366 | } |
| 1367 | $chunk = fread( $fp, 1024 ); |
| 1368 | fclose( $fp ); |
| 1369 | } |
| 1370 | |
| 1371 | $chunk = strtolower( $chunk ); |
| 1372 | |
| 1373 | if ( !$chunk ) { |
| 1374 | return false; |
| 1375 | } |
| 1376 | |
| 1377 | # decode from UTF-16 if needed (could be used for obfuscation). |
| 1378 | if ( str_starts_with( $chunk, "\xfe\xff" ) ) { |
| 1379 | $enc = 'UTF-16BE'; |
| 1380 | } elseif ( str_starts_with( $chunk, "\xff\xfe" ) ) { |
| 1381 | $enc = 'UTF-16LE'; |
| 1382 | } else { |
| 1383 | $enc = null; |
| 1384 | } |
| 1385 | |
| 1386 | if ( $enc !== null ) { |
| 1387 | $chunk = iconv( $enc, "ASCII//IGNORE", $chunk ); |
| 1388 | } |
| 1389 | |
| 1390 | $chunk = trim( $chunk ); |
| 1391 | |
| 1392 | /** @todo FIXME: Convert from UTF-16 if necessary! */ |
| 1393 | wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff" ); |
| 1394 | |
| 1395 | # check for HTML doctype |
| 1396 | if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) { |
| 1397 | return true; |
| 1398 | } |
| 1399 | |
| 1400 | // Some browsers will interpret obscure xml encodings as UTF-8, while |
| 1401 | // PHP/expat will interpret the given encoding in the xml declaration (T49304) |
| 1402 | if ( $extension === 'svg' || str_starts_with( $mime ?? '', 'image/svg' ) ) { |
| 1403 | if ( self::checkXMLEncodingMissmatch( $file ) ) { |
| 1404 | return true; |
| 1405 | } |
| 1406 | } |
| 1407 | |
| 1408 | // Quick check for HTML heuristics in old IE and Safari. |
| 1409 | // |
| 1410 | // The exact heuristics IE uses are checked separately via verifyMimeType(), so we |
| 1411 | // don't need them all here as it can cause many false positives. |
| 1412 | // |
| 1413 | // Check for `<script` and such still to forbid script tags and embedded HTML in SVG: |
| 1414 | $tags = [ |
| 1415 | '<body', |
| 1416 | '<head', |
| 1417 | '<html', # also in safari |
| 1418 | '<script', # also in safari |
| 1419 | ]; |
| 1420 | |
| 1421 | foreach ( $tags as $tag ) { |
| 1422 | if ( strpos( $chunk, $tag ) !== false ) { |
| 1423 | wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag" ); |
| 1424 | |
| 1425 | return true; |
| 1426 | } |
| 1427 | } |
| 1428 | |
| 1429 | /* |
| 1430 | * look for JavaScript |
| 1431 | */ |
| 1432 | |
| 1433 | # resolve entity-refs to look at attributes. may be harsh on big files... cache result? |
| 1434 | $chunk = Sanitizer::decodeCharReferences( $chunk ); |
| 1435 | |
| 1436 | # look for script-types |
| 1437 | if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!im', $chunk ) ) { |
| 1438 | wfDebug( __METHOD__ . ": found script types" ); |
| 1439 | |
| 1440 | return true; |
| 1441 | } |
| 1442 | |
| 1443 | # look for html-style script-urls |
| 1444 | if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!im', $chunk ) ) { |
| 1445 | wfDebug( __METHOD__ . ": found html-style script urls" ); |
| 1446 | |
| 1447 | return true; |
| 1448 | } |
| 1449 | |
| 1450 | # look for css-style script-urls |
| 1451 | if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!im', $chunk ) ) { |
| 1452 | wfDebug( __METHOD__ . ": found css-style script urls" ); |
| 1453 | |
| 1454 | return true; |
| 1455 | } |
| 1456 | |
| 1457 | wfDebug( __METHOD__ . ": no scripts found" ); |
| 1458 | |
| 1459 | return false; |
| 1460 | } |
| 1461 | |
| 1462 | /** |
| 1463 | * Check an allowed list of xml encodings that are known not to be interpreted differently |
| 1464 | * by the server's xml parser (expat) and some common browsers. |
| 1465 | * |
| 1466 | * @param string $file Pathname to the temporary upload file |
| 1467 | * @return bool True if the file contains an encoding that could be misinterpreted |
| 1468 | */ |
| 1469 | public static function checkXMLEncodingMissmatch( $file ) { |
| 1470 | // https://mimesniff.spec.whatwg.org/#resource-header says browsers |
| 1471 | // should read the first 1445 bytes. Do 4096 bytes for good measure. |
| 1472 | // XML Spec says XML declaration if present must be first thing in file |
| 1473 | // other than BOM |
| 1474 | $contents = file_get_contents( $file, false, null, 0, 4096 ); |
| 1475 | $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si'; |
| 1476 | |
| 1477 | if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) { |
| 1478 | if ( preg_match( $encodingRegex, $matches[1], $encMatch ) |
| 1479 | && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings ) |
| 1480 | ) { |
| 1481 | wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'" ); |
| 1482 | |
| 1483 | return true; |
| 1484 | } |
| 1485 | } elseif ( preg_match( "!<\?xml\b!i", $contents ) ) { |
| 1486 | // Start of XML declaration without an end in the first 4096 bytes |
| 1487 | // bytes. There shouldn't be a legitimate reason for this to happen. |
| 1488 | wfDebug( __METHOD__ . ": Unmatched XML declaration start" ); |
| 1489 | |
| 1490 | return true; |
| 1491 | } elseif ( str_starts_with( $contents, "\x4C\x6F\xA7\x94" ) ) { |
| 1492 | // EBCDIC encoded XML |
| 1493 | wfDebug( __METHOD__ . ": EBCDIC Encoded XML" ); |
| 1494 | |
| 1495 | return true; |
| 1496 | } |
| 1497 | |
| 1498 | // It's possible the file is encoded with multibyte encoding, so re-encode attempt to |
| 1499 | // detect the encoding in case it specifies an encoding not allowed in self::$safeXmlEncodings |
| 1500 | $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ]; |
| 1501 | foreach ( $attemptEncodings as $encoding ) { |
| 1502 | AtEase::suppressWarnings(); |
| 1503 | $str = iconv( $encoding, 'UTF-8', $contents ); |
| 1504 | AtEase::restoreWarnings(); |
| 1505 | if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) { |
| 1506 | if ( preg_match( $encodingRegex, $matches[1], $encMatch ) |
| 1507 | && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings ) |
| 1508 | ) { |
| 1509 | wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'" ); |
| 1510 | |
| 1511 | return true; |
| 1512 | } |
| 1513 | } elseif ( $str != '' && preg_match( "!<\?xml\b!i", $str ) ) { |
| 1514 | // Start of XML declaration without an end in the first 4096 bytes |
| 1515 | // bytes. There shouldn't be a legitimate reason for this to happen. |
| 1516 | wfDebug( __METHOD__ . ": Unmatched XML declaration start" ); |
| 1517 | |
| 1518 | return true; |
| 1519 | } |
| 1520 | } |
| 1521 | |
| 1522 | return false; |
| 1523 | } |
| 1524 | |
| 1525 | /** |
| 1526 | * @param string $filename |
| 1527 | * @param bool $partial |
| 1528 | * @return bool|array |
| 1529 | */ |
| 1530 | protected function detectScriptInSvg( $filename, $partial ) { |
| 1531 | $this->mSVGNSError = false; |
| 1532 | $check = new XmlTypeCheck( |
| 1533 | $filename, |
| 1534 | [ $this, 'checkSvgScriptCallback' ], |
| 1535 | true, |
| 1536 | [ |
| 1537 | 'processing_instruction_handler' => [ __CLASS__, 'checkSvgPICallback' ], |
| 1538 | 'external_dtd_handler' => [ __CLASS__, 'checkSvgExternalDTD' ], |
| 1539 | ] |
| 1540 | ); |
| 1541 | if ( $check->wellFormed !== true ) { |
| 1542 | // Invalid xml (T60553) |
| 1543 | // But only when non-partial (T67724) |
| 1544 | return $partial ? false : [ 'uploadinvalidxml' ]; |
| 1545 | } |
| 1546 | |
| 1547 | if ( $check->filterMatch ) { |
| 1548 | if ( $this->mSVGNSError ) { |
| 1549 | return [ 'uploadscriptednamespace', $this->mSVGNSError ]; |
| 1550 | } |
| 1551 | return $check->filterMatchType; |
| 1552 | } |
| 1553 | |
| 1554 | return false; |
| 1555 | } |
| 1556 | |
| 1557 | /** |
| 1558 | * Callback to filter SVG Processing Instructions. |
| 1559 | * |
| 1560 | * @param string $target Processing instruction name |
| 1561 | * @param string $data Processing instruction attribute and value |
| 1562 | * @return bool|array |
| 1563 | */ |
| 1564 | public static function checkSvgPICallback( $target, $data ) { |
| 1565 | // Don't allow external stylesheets (T59550) |
| 1566 | if ( preg_match( '/xml-stylesheet/i', $target ) ) { |
| 1567 | return [ 'upload-scripted-pi-callback' ]; |
| 1568 | } |
| 1569 | |
| 1570 | return false; |
| 1571 | } |
| 1572 | |
| 1573 | /** |
| 1574 | * Verify that DTD URLs referenced are only the standard DTDs. |
| 1575 | * |
| 1576 | * Browsers seem to ignore external DTDs. |
| 1577 | * |
| 1578 | * However, just to be on the safe side, only allow DTDs from the SVG standard. |
| 1579 | * |
| 1580 | * @param string $type PUBLIC or SYSTEM |
| 1581 | * @param string $publicId The well-known public identifier for the dtd |
| 1582 | * @param string $systemId The url for the external dtd |
| 1583 | * @return bool|array |
| 1584 | */ |
| 1585 | public static function checkSvgExternalDTD( $type, $publicId, $systemId ) { |
| 1586 | // This doesn't include the XHTML+MathML+SVG doctype since we don't |
| 1587 | // allow XHTML anyway. |
| 1588 | static $allowedDTDs = [ |
| 1589 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd', |
| 1590 | 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd', |
| 1591 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd', |
| 1592 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd', |
| 1593 | // https://phabricator.wikimedia.org/T168856 |
| 1594 | 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd', |
| 1595 | ]; |
| 1596 | if ( $type !== 'PUBLIC' |
| 1597 | || !in_array( $systemId, $allowedDTDs ) |
| 1598 | || !str_starts_with( $publicId, "-//W3C//" ) |
| 1599 | ) { |
| 1600 | return [ 'upload-scripted-dtd' ]; |
| 1601 | } |
| 1602 | return false; |
| 1603 | } |
| 1604 | |
| 1605 | /** |
| 1606 | * @todo Replace this with a allow list filter! |
| 1607 | * @param string $element |
| 1608 | * @param array $attribs |
| 1609 | * @param string|null $data |
| 1610 | * @return bool|array |
| 1611 | */ |
| 1612 | public function checkSvgScriptCallback( $element, $attribs, $data = null ) { |
| 1613 | [ $namespace, $strippedElement ] = self::splitXmlNamespace( $element ); |
| 1614 | |
| 1615 | // We specifically don't include: |
| 1616 | // http://www.w3.org/1999/xhtml (T62771) |
| 1617 | static $validNamespaces = [ |
| 1618 | '', |
| 1619 | 'adobe:ns:meta/', |
| 1620 | 'http://creativecommons.org/ns#', |
| 1621 | 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd', |
| 1622 | 'http://ns.adobe.com/adobeillustrator/10.0/', |
| 1623 | 'http://ns.adobe.com/adobesvgviewerextensions/3.0/', |
| 1624 | 'http://ns.adobe.com/extensibility/1.0/', |
| 1625 | 'http://ns.adobe.com/flows/1.0/', |
| 1626 | 'http://ns.adobe.com/illustrator/1.0/', |
| 1627 | 'http://ns.adobe.com/imagereplacement/1.0/', |
| 1628 | 'http://ns.adobe.com/pdf/1.3/', |
| 1629 | 'http://ns.adobe.com/photoshop/1.0/', |
| 1630 | 'http://ns.adobe.com/saveforweb/1.0/', |
| 1631 | 'http://ns.adobe.com/variables/1.0/', |
| 1632 | 'http://ns.adobe.com/xap/1.0/', |
| 1633 | 'http://ns.adobe.com/xap/1.0/g/', |
| 1634 | 'http://ns.adobe.com/xap/1.0/g/img/', |
| 1635 | 'http://ns.adobe.com/xap/1.0/mm/', |
| 1636 | 'http://ns.adobe.com/xap/1.0/rights/', |
| 1637 | 'http://ns.adobe.com/xap/1.0/stype/dimensions#', |
| 1638 | 'http://ns.adobe.com/xap/1.0/stype/font#', |
| 1639 | 'http://ns.adobe.com/xap/1.0/stype/manifestitem#', |
| 1640 | 'http://ns.adobe.com/xap/1.0/stype/resourceevent#', |
| 1641 | 'http://ns.adobe.com/xap/1.0/stype/resourceref#', |
| 1642 | 'http://ns.adobe.com/xap/1.0/t/pg/', |
| 1643 | 'http://purl.org/dc/elements/1.1/', |
| 1644 | 'http://purl.org/dc/elements/1.1', |
| 1645 | 'http://schemas.microsoft.com/visio/2003/svgextensions/', |
| 1646 | 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd', |
| 1647 | 'http://taptrix.com/inkpad/svg_extensions', |
| 1648 | 'http://web.resource.org/cc/', |
| 1649 | 'http://www.freesoftware.fsf.org/bkchem/cdml', |
| 1650 | 'http://www.inkscape.org/namespaces/inkscape', |
| 1651 | 'http://www.opengis.net/gml', |
| 1652 | 'http://www.w3.org/1999/02/22-rdf-syntax-ns#', |
| 1653 | 'http://www.w3.org/2000/svg', |
| 1654 | 'http://www.w3.org/tr/rec-rdf-syntax/', |
| 1655 | 'http://www.w3.org/2000/01/rdf-schema#', |
| 1656 | 'http://www.w3.org/2000/02/svg/testsuite/description/', // https://phabricator.wikimedia.org/T278044 |
| 1657 | ]; |
| 1658 | |
| 1659 | // Inkscape mangles namespace definitions created by Adobe Illustrator. |
| 1660 | // This is nasty but harmless. (T144827) |
| 1661 | $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace ); |
| 1662 | |
| 1663 | if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) { |
| 1664 | wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file." ); |
| 1665 | /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */ |
| 1666 | $this->mSVGNSError = $namespace; |
| 1667 | |
| 1668 | return true; |
| 1669 | } |
| 1670 | |
| 1671 | // check for elements that can contain javascript |
| 1672 | if ( $strippedElement === 'script' ) { |
| 1673 | wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file." ); |
| 1674 | |
| 1675 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 1676 | } |
| 1677 | |
| 1678 | // e.g., <svg xmlns="http://www.w3.org/2000/svg"> |
| 1679 | // <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg> |
| 1680 | if ( $strippedElement === 'handler' ) { |
| 1681 | wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file." ); |
| 1682 | |
| 1683 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 1684 | } |
| 1685 | |
| 1686 | // SVG reported in Feb '12 that used xml:stylesheet to generate javascript block |
| 1687 | if ( $strippedElement === 'stylesheet' ) { |
| 1688 | wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file." ); |
| 1689 | |
| 1690 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 1691 | } |
| 1692 | |
| 1693 | // Block iframes, in case they pass the namespace check |
| 1694 | if ( $strippedElement === 'iframe' ) { |
| 1695 | wfDebug( __METHOD__ . ": iframe in uploaded file." ); |
| 1696 | |
| 1697 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 1698 | } |
| 1699 | |
| 1700 | // Check <style> css |
| 1701 | if ( $strippedElement === 'style' |
| 1702 | && self::checkCssFragment( Sanitizer::normalizeCss( $data ) ) |
| 1703 | ) { |
| 1704 | wfDebug( __METHOD__ . ": hostile css in style element." ); |
| 1705 | |
| 1706 | return [ 'uploaded-hostile-svg' ]; |
| 1707 | } |
| 1708 | |
| 1709 | static $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker', |
| 1710 | 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ]; |
| 1711 | |
| 1712 | foreach ( $attribs as $attrib => $value ) { |
| 1713 | // If attributeNamespace is '', it is relative to its element's namespace |
| 1714 | [ $attributeNamespace, $stripped ] = self::splitXmlNamespace( $attrib ); |
| 1715 | $value = strtolower( $value ); |
| 1716 | |
| 1717 | if ( !( |
| 1718 | // Inkscape element's have valid attribs that start with on and are safe, fail all others |
| 1719 | $namespace === 'http://www.inkscape.org/namespaces/inkscape' && |
| 1720 | $attributeNamespace === '' |
| 1721 | ) && str_starts_with( $stripped, 'on' ) |
| 1722 | ) { |
| 1723 | wfDebug( __METHOD__ |
| 1724 | . ": Found event-handler attribute '$attrib'='$value' in uploaded file." ); |
| 1725 | |
| 1726 | return [ 'uploaded-event-handler-on-svg', $attrib, $value ]; |
| 1727 | } |
| 1728 | |
| 1729 | // Do not allow relative links, or unsafe url schemas. |
| 1730 | // For <a> tags, only data:, http: and https: and same-document |
| 1731 | // fragment links are allowed. |
| 1732 | // For all other tags, only 'data:' and fragments (#) are allowed. |
| 1733 | if ( |
| 1734 | $stripped === 'href' |
| 1735 | && $value !== '' |
| 1736 | && !str_starts_with( $value, 'data:' ) |
| 1737 | && !str_starts_with( $value, '#' ) |
| 1738 | && !( $strippedElement === 'a' && preg_match( '!^https?://!i', $value ) ) |
| 1739 | ) { |
| 1740 | wfDebug( __METHOD__ . ": Found href attribute <$strippedElement " |
| 1741 | . "'$attrib'='$value' in uploaded file." ); |
| 1742 | |
| 1743 | return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ]; |
| 1744 | } |
| 1745 | |
| 1746 | // Only allow 'data:\' targets that should be safe. |
| 1747 | // This prevents vectors like image/svg, text/xml, application/xml, and text/html, which can contain scripts |
| 1748 | if ( $stripped === 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) { |
| 1749 | // RFC2397 parameters. |
| 1750 | // This is only slightly slower than (;[\w;]+)*. |
| 1751 | // phpcs:ignore Generic.Files.LineLength |
| 1752 | $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?'; |
| 1753 | |
| 1754 | if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) { |
| 1755 | wfDebug( __METHOD__ . ": Found href to allow listed data: uri " |
| 1756 | . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file." ); |
| 1757 | return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ]; |
| 1758 | } |
| 1759 | } |
| 1760 | |
| 1761 | // Change href with animate from (http://html5sec.org/#137). |
| 1762 | if ( $stripped === 'attributename' |
| 1763 | && $strippedElement === 'animate' |
| 1764 | && $this->stripXmlNamespace( $value ) === 'href' |
| 1765 | ) { |
| 1766 | wfDebug( __METHOD__ . ": Found animate that might be changing href using from " |
| 1767 | . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file." ); |
| 1768 | |
| 1769 | return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ]; |
| 1770 | } |
| 1771 | |
| 1772 | // Use set/animate to add event-handler attribute to parent. |
| 1773 | if ( ( $strippedElement === 'set' || $strippedElement === 'animate' ) |
| 1774 | && $stripped === 'attributename' |
| 1775 | && str_starts_with( $value, 'on' ) |
| 1776 | ) { |
| 1777 | wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with " |
| 1778 | . "\"<$strippedElement $stripped='$value'...\" in uploaded file." ); |
| 1779 | |
| 1780 | return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ]; |
| 1781 | } |
| 1782 | |
| 1783 | // use set to add href attribute to parent element. |
| 1784 | if ( $strippedElement === 'set' |
| 1785 | && $stripped === 'attributename' |
| 1786 | && str_contains( $value, 'href' ) |
| 1787 | ) { |
| 1788 | wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file." ); |
| 1789 | |
| 1790 | return [ 'uploaded-setting-href-svg' ]; |
| 1791 | } |
| 1792 | |
| 1793 | // use set to add a remote / data / script target to an element. |
| 1794 | if ( $strippedElement === 'set' |
| 1795 | && $stripped === 'to' |
| 1796 | && preg_match( '!(http|https|data|script):!im', $value ) |
| 1797 | ) { |
| 1798 | wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file." ); |
| 1799 | |
| 1800 | return [ 'uploaded-wrong-setting-svg', $value ]; |
| 1801 | } |
| 1802 | |
| 1803 | // use handler attribute with remote / data / script. |
| 1804 | if ( $stripped === 'handler' && preg_match( '!(http|https|data|script):!im', $value ) ) { |
| 1805 | wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script " |
| 1806 | . "'$attrib'='$value' in uploaded file." ); |
| 1807 | |
| 1808 | return [ 'uploaded-setting-handler-svg', $attrib, $value ]; |
| 1809 | } |
| 1810 | |
| 1811 | // use CSS styles to bring in remote code. |
| 1812 | if ( $stripped === 'style' |
| 1813 | && self::checkCssFragment( Sanitizer::normalizeCss( $value ) ) |
| 1814 | ) { |
| 1815 | wfDebug( __METHOD__ . ": Found svg setting a style with " |
| 1816 | . "remote url '$attrib'='$value' in uploaded file." ); |
| 1817 | return [ 'uploaded-remote-url-svg', $attrib, $value ]; |
| 1818 | } |
| 1819 | |
| 1820 | // Several attributes can include css, css character escaping isn't allowed. |
| 1821 | if ( in_array( $stripped, $cssAttrs, true ) |
| 1822 | && self::checkCssFragment( $value ) |
| 1823 | ) { |
| 1824 | wfDebug( __METHOD__ . ": Found svg setting a style with " |
| 1825 | . "remote url '$attrib'='$value' in uploaded file." ); |
| 1826 | return [ 'uploaded-remote-url-svg', $attrib, $value ]; |
| 1827 | } |
| 1828 | |
| 1829 | // image filters can pull in url, which could be svg that executes scripts. |
| 1830 | // Only allow url( "#foo" ). |
| 1831 | // Do not allow url( http://example.com ) |
| 1832 | if ( $strippedElement === 'image' |
| 1833 | && $stripped === 'filter' |
| 1834 | && preg_match( '!url\s*\(\s*["\']?[^#]!im', $value ) |
| 1835 | ) { |
| 1836 | wfDebug( __METHOD__ . ": Found image filter with url: " |
| 1837 | . "\"<$strippedElement $stripped='$value'...\" in uploaded file." ); |
| 1838 | |
| 1839 | return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ]; |
| 1840 | } |
| 1841 | } |
| 1842 | |
| 1843 | return false; // No scripts detected |
| 1844 | } |
| 1845 | |
| 1846 | /** |
| 1847 | * Check a block of CSS or CSS fragment for anything that looks like |
| 1848 | * it is bringing in remote code. |
| 1849 | * @param string $value a string of CSS |
| 1850 | * @return bool true if the CSS contains an illegal string, false if otherwise |
| 1851 | */ |
| 1852 | private static function checkCssFragment( $value ) { |
| 1853 | # Forbid external stylesheets, for both reliability and to protect viewer's privacy |
| 1854 | if ( stripos( $value, '@import' ) !== false ) { |
| 1855 | return true; |
| 1856 | } |
| 1857 | |
| 1858 | # We allow @font-face to embed fonts with data: urls, so we snip the string |
| 1859 | # 'url' out so that this case won't match when we check for urls below |
| 1860 | $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im'; |
| 1861 | $value = preg_replace( $pattern, '$1$2', $value ); |
| 1862 | |
| 1863 | # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS |
| 1864 | # properties filter and accelerator don't seem to be useful for xss in SVG files. |
| 1865 | # Expression and -o-link don't seem to work either, but filtering them here in case. |
| 1866 | # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:..., |
| 1867 | # but not local ones such as url("#..., url('#..., url(#.... |
| 1868 | if ( preg_match( '!expression |
| 1869 | | -o-link\s*: |
| 1870 | | -o-link-source\s*: |
| 1871 | | -o-replace\s*:!imx', $value ) ) { |
| 1872 | return true; |
| 1873 | } |
| 1874 | |
| 1875 | if ( preg_match_all( |
| 1876 | "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim", |
| 1877 | $value, |
| 1878 | $matches |
| 1879 | ) !== 0 |
| 1880 | ) { |
| 1881 | # TODO: redo this in one regex. Until then, url("#whatever") matches the first |
| 1882 | foreach ( $matches[1] as $match ) { |
| 1883 | if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) { |
| 1884 | return true; |
| 1885 | } |
| 1886 | } |
| 1887 | } |
| 1888 | |
| 1889 | if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) { |
| 1890 | return true; |
| 1891 | } |
| 1892 | |
| 1893 | return false; |
| 1894 | } |
| 1895 | |
| 1896 | /** |
| 1897 | * Divide the element name passed by the XML parser to the callback into URI and prefix. |
| 1898 | * @param string $element |
| 1899 | * @return array Containing the namespace URI and prefix |
| 1900 | */ |
| 1901 | private static function splitXmlNamespace( $element ) { |
| 1902 | // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ] |
| 1903 | $parts = explode( ':', strtolower( $element ) ); |
| 1904 | $name = array_pop( $parts ); |
| 1905 | $ns = implode( ':', $parts ); |
| 1906 | |
| 1907 | return [ $ns, $name ]; |
| 1908 | } |
| 1909 | |
| 1910 | /** |
| 1911 | * @param string $element |
| 1912 | * @return string |
| 1913 | */ |
| 1914 | private function stripXmlNamespace( $element ) { |
| 1915 | // 'http://www.w3.org/2000/svg:script' -> 'script' |
| 1916 | return self::splitXmlNamespace( $element )[1]; |
| 1917 | } |
| 1918 | |
| 1919 | /** |
| 1920 | * Generic wrapper function for a virus scanner program. |
| 1921 | * This relies on the $wgAntivirus and $wgAntivirusSetup variables. |
| 1922 | * $wgAntivirusRequired may be used to deny upload if the scan fails. |
| 1923 | * |
| 1924 | * @param string $file Pathname to the temporary upload file |
| 1925 | * @return bool|null|string False if not virus is found, null if the scan fails or is disabled, |
| 1926 | * or a string containing feedback from the virus scanner if a virus was found. |
| 1927 | * If textual feedback is missing but a virus was found, this function returns true. |
| 1928 | */ |
| 1929 | public static function detectVirus( $file ) { |
| 1930 | global $wgOut; |
| 1931 | $mainConfig = MediaWikiServices::getInstance()->getMainConfig(); |
| 1932 | $antivirus = $mainConfig->get( MainConfigNames::Antivirus ); |
| 1933 | $antivirusSetup = $mainConfig->get( MainConfigNames::AntivirusSetup ); |
| 1934 | $antivirusRequired = $mainConfig->get( MainConfigNames::AntivirusRequired ); |
| 1935 | if ( !$antivirus ) { |
| 1936 | wfDebug( __METHOD__ . ": virus scanner disabled" ); |
| 1937 | |
| 1938 | return null; |
| 1939 | } |
| 1940 | |
| 1941 | if ( !$antivirusSetup[$antivirus] ) { |
| 1942 | wfDebug( __METHOD__ . ": unknown virus scanner: {$antivirus}" ); |
| 1943 | $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>", |
| 1944 | [ 'virus-badscanner', $antivirus ] ); |
| 1945 | |
| 1946 | return wfMessage( 'virus-unknownscanner' )->text() . " {$antivirus}"; |
| 1947 | } |
| 1948 | |
| 1949 | # look up scanner configuration |
| 1950 | $command = $antivirusSetup[$antivirus]['command']; |
| 1951 | $exitCodeMap = $antivirusSetup[$antivirus]['codemap']; |
| 1952 | $msgPattern = $antivirusSetup[$antivirus]['messagepattern'] ?? null; |
| 1953 | |
| 1954 | if ( !str_contains( $command, "%f" ) ) { |
| 1955 | # simple pattern: append file to scan |
| 1956 | $command .= " " . Shell::escape( $file ); |
| 1957 | } else { |
| 1958 | # complex pattern: replace "%f" with file to scan |
| 1959 | $command = str_replace( "%f", Shell::escape( $file ), $command ); |
| 1960 | } |
| 1961 | |
| 1962 | wfDebug( __METHOD__ . ": running virus scan: $command " ); |
| 1963 | |
| 1964 | # execute virus scanner |
| 1965 | $exitCode = false; |
| 1966 | |
| 1967 | # NOTE: there's a 50-line workaround to make stderr redirection work on windows, too. |
| 1968 | # that does not seem to be worth the pain. |
| 1969 | # Ask me (Duesentrieb) about it if it's ever needed. |
| 1970 | $output = wfShellExecWithStderr( $command, $exitCode ); |
| 1971 | |
| 1972 | # map exit code to AV_xxx constants. |
| 1973 | $mappedCode = $exitCode; |
| 1974 | if ( $exitCodeMap ) { |
| 1975 | if ( isset( $exitCodeMap[$exitCode] ) ) { |
| 1976 | $mappedCode = $exitCodeMap[$exitCode]; |
| 1977 | } elseif ( isset( $exitCodeMap["*"] ) ) { |
| 1978 | $mappedCode = $exitCodeMap["*"]; |
| 1979 | } |
| 1980 | } |
| 1981 | |
| 1982 | # NB: AV_NO_VIRUS is 0, but AV_SCAN_FAILED is false, |
| 1983 | # so we need the strict equalities === and thus can't use a switch here |
| 1984 | if ( $mappedCode === AV_SCAN_FAILED ) { |
| 1985 | # scan failed (code was mapped to false by $exitCodeMap) |
| 1986 | wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode)." ); |
| 1987 | |
| 1988 | $output = $antivirusRequired |
| 1989 | ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text() |
| 1990 | : null; |
| 1991 | } elseif ( $mappedCode === AV_SCAN_ABORTED ) { |
| 1992 | # scan failed because filetype is unknown (probably immune) |
| 1993 | wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode)." ); |
| 1994 | $output = null; |
| 1995 | } elseif ( $mappedCode === AV_NO_VIRUS ) { |
| 1996 | # no virus found |
| 1997 | wfDebug( __METHOD__ . ": file passed virus scan." ); |
| 1998 | $output = false; |
| 1999 | } else { |
| 2000 | $output = trim( $output ); |
| 2001 | |
| 2002 | if ( !$output ) { |
| 2003 | $output = true; # if there's no output, return true |
| 2004 | } elseif ( $msgPattern ) { |
| 2005 | $groups = []; |
| 2006 | if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) { |
| 2007 | $output = $groups[1]; |
| 2008 | } |
| 2009 | } |
| 2010 | |
| 2011 | wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output" ); |
| 2012 | } |
| 2013 | |
| 2014 | return $output; |
| 2015 | } |
| 2016 | |
| 2017 | /** |
| 2018 | * Check if there's a file overwrite conflict and, if so, if restrictions |
| 2019 | * forbid this user from performing the upload. |
| 2020 | * |
| 2021 | * @param Authority $performer |
| 2022 | * |
| 2023 | * @return bool|array |
| 2024 | */ |
| 2025 | private function checkOverwrite( Authority $performer ) { |
| 2026 | // First check whether the local file can be overwritten |
| 2027 | $file = $this->getLocalFile(); |
| 2028 | $file->load( IDBAccessObject::READ_LATEST ); |
| 2029 | if ( $file->exists() ) { |
| 2030 | if ( !self::userCanReUpload( $performer, $file ) ) { |
| 2031 | return [ 'fileexists-forbidden', $file->getName() ]; |
| 2032 | } |
| 2033 | |
| 2034 | return true; |
| 2035 | } |
| 2036 | |
| 2037 | $services = MediaWikiServices::getInstance(); |
| 2038 | |
| 2039 | /* Check shared conflicts: if the local file does not exist, but |
| 2040 | * RepoGroup::findFile finds a file, it exists in a shared repository. |
| 2041 | */ |
| 2042 | $file = $services->getRepoGroup()->findFile( $this->getTitle(), [ 'latest' => true ] ); |
| 2043 | if ( $file && !$performer->isAllowed( 'reupload-shared' ) ) { |
| 2044 | return [ 'fileexists-shared-forbidden', $file->getName() ]; |
| 2045 | } |
| 2046 | |
| 2047 | return true; |
| 2048 | } |
| 2049 | |
| 2050 | /** |
| 2051 | * Check if a user is the last uploader |
| 2052 | * |
| 2053 | * @param Authority $performer |
| 2054 | * @param File $img |
| 2055 | * @return bool |
| 2056 | */ |
| 2057 | public static function userCanReUpload( Authority $performer, File $img ) { |
| 2058 | if ( $performer->isAllowed( 'reupload' ) ) { |
| 2059 | return true; // non-conditional |
| 2060 | } |
| 2061 | |
| 2062 | if ( !$performer->isAllowed( 'reupload-own' ) ) { |
| 2063 | return false; |
| 2064 | } |
| 2065 | |
| 2066 | if ( !( $img instanceof LocalFile ) ) { |
| 2067 | return false; |
| 2068 | } |
| 2069 | |
| 2070 | return $performer->getUser()->equals( $img->getUploader( File::RAW ) ); |
| 2071 | } |
| 2072 | |
| 2073 | /** |
| 2074 | * Helper function that does various existence checks for a file. |
| 2075 | * The following checks are performed: |
| 2076 | * - If the file exists |
| 2077 | * - If an article with the same name as the file exists |
| 2078 | * - If a file exists with normalized extension |
| 2079 | * - If the file looks like a thumbnail and the original exists |
| 2080 | * |
| 2081 | * @param File $file The File object to check |
| 2082 | * @return array|false False if the file does not exist, else an array |
| 2083 | */ |
| 2084 | public static function getExistsWarning( $file ) { |
| 2085 | if ( $file->exists() ) { |
| 2086 | return [ 'warning' => 'exists', 'file' => $file ]; |
| 2087 | } |
| 2088 | |
| 2089 | if ( $file->getTitle()->getArticleID() ) { |
| 2090 | return [ 'warning' => 'page-exists', 'file' => $file ]; |
| 2091 | } |
| 2092 | |
| 2093 | $n = strrpos( $file->getName(), '.' ); |
| 2094 | if ( $n > 0 ) { |
| 2095 | $partname = substr( $file->getName(), 0, $n ); |
| 2096 | $extension = substr( $file->getName(), $n + 1 ); |
| 2097 | } else { |
| 2098 | $partname = $file->getName(); |
| 2099 | $extension = ''; |
| 2100 | } |
| 2101 | $normalizedExtension = File::normalizeExtension( $extension ); |
| 2102 | $localRepo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo(); |
| 2103 | |
| 2104 | if ( $normalizedExtension != $extension ) { |
| 2105 | // We're not using the normalized form of the extension. |
| 2106 | // Normal form is lowercase, using most common of alternate |
| 2107 | // extensions (e.g. 'jpg' rather than 'JPEG'). |
| 2108 | |
| 2109 | // Check for another file using the normalized form... |
| 2110 | $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" ); |
| 2111 | $file_lc = $localRepo->newFile( $nt_lc ); |
| 2112 | |
| 2113 | if ( $file_lc->exists() ) { |
| 2114 | return [ |
| 2115 | 'warning' => 'exists-normalized', |
| 2116 | 'file' => $file, |
| 2117 | 'normalizedFile' => $file_lc |
| 2118 | ]; |
| 2119 | } |
| 2120 | } |
| 2121 | |
| 2122 | // Check for files with the same name but a different extension |
| 2123 | $similarFiles = $localRepo->findFilesByPrefix( "{$partname}.", 1 ); |
| 2124 | if ( count( $similarFiles ) ) { |
| 2125 | return [ |
| 2126 | 'warning' => 'exists-normalized', |
| 2127 | 'file' => $file, |
| 2128 | 'normalizedFile' => $similarFiles[0], |
| 2129 | ]; |
| 2130 | } |
| 2131 | |
| 2132 | if ( self::isThumbName( $file->getName() ) ) { |
| 2133 | // Check for filenames like 50px- or 180px-, these are mostly thumbnails |
| 2134 | $nt_thb = Title::newFromText( |
| 2135 | substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension, |
| 2136 | NS_FILE |
| 2137 | ); |
| 2138 | $file_thb = $localRepo->newFile( $nt_thb ); |
| 2139 | if ( $file_thb->exists() ) { |
| 2140 | return [ |
| 2141 | 'warning' => 'thumb', |
| 2142 | 'file' => $file, |
| 2143 | 'thumbFile' => $file_thb |
| 2144 | ]; |
| 2145 | } |
| 2146 | |
| 2147 | // The file does not exist, but we just don't like the name |
| 2148 | return [ |
| 2149 | 'warning' => 'thumb-name', |
| 2150 | 'file' => $file, |
| 2151 | 'thumbFile' => $file_thb |
| 2152 | ]; |
| 2153 | } |
| 2154 | |
| 2155 | foreach ( self::getFilenamePrefixBlacklist() as $prefix ) { |
| 2156 | if ( str_starts_with( $partname, $prefix ) ) { |
| 2157 | return [ |
| 2158 | 'warning' => 'bad-prefix', |
| 2159 | 'file' => $file, |
| 2160 | 'prefix' => $prefix |
| 2161 | ]; |
| 2162 | } |
| 2163 | } |
| 2164 | |
| 2165 | return false; |
| 2166 | } |
| 2167 | |
| 2168 | /** |
| 2169 | * Helper function that checks whether the filename looks like a thumbnail |
| 2170 | * @param string $filename |
| 2171 | * @return bool |
| 2172 | */ |
| 2173 | public static function isThumbName( $filename ) { |
| 2174 | $n = strrpos( $filename, '.' ); |
| 2175 | $partname = $n ? substr( $filename, 0, $n ) : $filename; |
| 2176 | |
| 2177 | return ( |
| 2178 | substr( $partname, 3, 3 ) === 'px-' || |
| 2179 | substr( $partname, 2, 3 ) === 'px-' |
| 2180 | ) && preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) ); |
| 2181 | } |
| 2182 | |
| 2183 | /** |
| 2184 | * Get a list of disallowed filename prefixes from [[MediaWiki:Filename-prefix-blacklist]] |
| 2185 | * |
| 2186 | * @return string[] List of prefixes |
| 2187 | */ |
| 2188 | public static function getFilenamePrefixBlacklist() { |
| 2189 | $list = []; |
| 2190 | $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage(); |
| 2191 | if ( !$message->isDisabled() ) { |
| 2192 | $lines = explode( "\n", $message->plain() ); |
| 2193 | foreach ( $lines as $line ) { |
| 2194 | // Remove comment lines |
| 2195 | $comment = substr( trim( $line ), 0, 1 ); |
| 2196 | if ( $comment === '#' || $comment == '' ) { |
| 2197 | continue; |
| 2198 | } |
| 2199 | // Remove additional comments after a prefix |
| 2200 | $comment = strpos( $line, '#' ); |
| 2201 | if ( $comment > 0 ) { |
| 2202 | $line = substr( $line, 0, $comment - 1 ); |
| 2203 | } |
| 2204 | $list[] = trim( $line ); |
| 2205 | } |
| 2206 | } |
| 2207 | |
| 2208 | return $list; |
| 2209 | } |
| 2210 | |
| 2211 | /** |
| 2212 | * Gets image info about the file just uploaded. |
| 2213 | * |
| 2214 | * @deprecated since 1.42, subclasses of ApiUpload can use |
| 2215 | * ApiUpload::getUploadImageInfo() instead. |
| 2216 | * |
| 2217 | * @param ?ApiResult $result unused since 1.42 |
| 2218 | * @return array Image info |
| 2219 | */ |
| 2220 | public function getImageInfo( $result = null ) { |
| 2221 | $apiUpload = ApiUpload::getDummyInstance(); |
| 2222 | return $apiUpload->getUploadImageInfo( $this ); |
| 2223 | } |
| 2224 | |
| 2225 | /** |
| 2226 | * @param array $error |
| 2227 | * @return Status |
| 2228 | */ |
| 2229 | public function convertVerifyErrorToStatus( $error ) { |
| 2230 | $code = $error['status']; |
| 2231 | unset( $code['status'] ); |
| 2232 | |
| 2233 | return Status::newFatal( $this->getVerificationErrorCode( $code ), $error ); |
| 2234 | } |
| 2235 | |
| 2236 | /** |
| 2237 | * Get MediaWiki's maximum uploaded file size for a given type of upload, based on |
| 2238 | * $wgMaxUploadSize. |
| 2239 | * |
| 2240 | * @param null|string $forType |
| 2241 | * @return int |
| 2242 | */ |
| 2243 | public static function getMaxUploadSize( $forType = null ) { |
| 2244 | $maxUploadSize = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::MaxUploadSize ); |
| 2245 | |
| 2246 | if ( is_array( $maxUploadSize ) ) { |
| 2247 | if ( $forType !== null && isset( $maxUploadSize[$forType] ) ) { |
| 2248 | return $maxUploadSize[$forType]; |
| 2249 | } |
| 2250 | return $maxUploadSize['*']; |
| 2251 | } |
| 2252 | return intval( $maxUploadSize ); |
| 2253 | } |
| 2254 | |
| 2255 | /** |
| 2256 | * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the |
| 2257 | * limit can't be guessed, return a very large number (PHP_INT_MAX) instead. |
| 2258 | * |
| 2259 | * @since 1.27 |
| 2260 | * @return int |
| 2261 | */ |
| 2262 | public static function getMaxPhpUploadSize() { |
| 2263 | $phpMaxFileSize = wfShorthandToInteger( |
| 2264 | ini_get( 'upload_max_filesize' ), |
| 2265 | PHP_INT_MAX |
| 2266 | ); |
| 2267 | $phpMaxPostSize = wfShorthandToInteger( |
| 2268 | ini_get( 'post_max_size' ), |
| 2269 | PHP_INT_MAX |
| 2270 | ) ?: PHP_INT_MAX; |
| 2271 | return min( $phpMaxFileSize, $phpMaxPostSize ); |
| 2272 | } |
| 2273 | |
| 2274 | /** |
| 2275 | * Get the current status of a chunked upload (used for polling). |
| 2276 | * |
| 2277 | * This should only be called during POST requests since we |
| 2278 | * fetch from dc-local MainStash, and from a GET request we can't |
| 2279 | * know that the value is available or up-to-date. |
| 2280 | * |
| 2281 | * @param UserIdentity $user |
| 2282 | * @param string $statusKey |
| 2283 | * @return mixed[]|false |
| 2284 | */ |
| 2285 | public static function getSessionStatus( UserIdentity $user, $statusKey ) { |
| 2286 | $store = self::getUploadSessionStore(); |
| 2287 | $key = self::getUploadSessionKey( $store, $user, $statusKey ); |
| 2288 | |
| 2289 | return $store->get( $key ); |
| 2290 | } |
| 2291 | |
| 2292 | /** |
| 2293 | * Set the current status of a chunked upload (used for polling). |
| 2294 | * |
| 2295 | * The value will be set in cache for 1 day. |
| 2296 | * |
| 2297 | * This should only be called during POST requests. |
| 2298 | * |
| 2299 | * @param UserIdentity $user |
| 2300 | * @param string $statusKey |
| 2301 | * @param array|false $value |
| 2302 | * @return void |
| 2303 | */ |
| 2304 | public static function setSessionStatus( UserIdentity $user, $statusKey, $value ) { |
| 2305 | $store = self::getUploadSessionStore(); |
| 2306 | $key = self::getUploadSessionKey( $store, $user, $statusKey ); |
| 2307 | $logger = LoggerFactory::getInstance( 'upload' ); |
| 2308 | |
| 2309 | if ( is_array( $value ) && ( $value['result'] ?? '' ) === 'Failure' ) { |
| 2310 | $logger->info( 'Upload session {key} for {user} set to failure {status} at {stage}', |
| 2311 | [ |
| 2312 | 'result' => $value['result'] ?? '', |
| 2313 | 'stage' => $value['stage'] ?? 'unknown', |
| 2314 | 'user' => $user->getName(), |
| 2315 | 'status' => (string)( $value['status'] ?? '-' ), |
| 2316 | 'filekey' => $value['filekey'] ?? '', |
| 2317 | 'key' => $statusKey |
| 2318 | ] |
| 2319 | ); |
| 2320 | } elseif ( is_array( $value ) ) { |
| 2321 | $logger->debug( 'Upload session {key} for {user} changed {status} at {stage}', |
| 2322 | [ |
| 2323 | 'result' => $value['result'] ?? '', |
| 2324 | 'stage' => $value['stage'] ?? 'unknown', |
| 2325 | 'user' => $user->getName(), |
| 2326 | 'status' => (string)( $value['status'] ?? '-' ), |
| 2327 | 'filekey' => $value['filekey'] ?? '', |
| 2328 | 'key' => $statusKey |
| 2329 | ] |
| 2330 | ); |
| 2331 | } else { |
| 2332 | $logger->debug( "Upload session {key} deleted for {user}", |
| 2333 | [ |
| 2334 | 'value' => $value, |
| 2335 | 'key' => $statusKey, |
| 2336 | 'user' => $user->getName() |
| 2337 | ] |
| 2338 | ); |
| 2339 | } |
| 2340 | |
| 2341 | if ( $value === false ) { |
| 2342 | $store->delete( $key ); |
| 2343 | } else { |
| 2344 | $store->set( $key, $value, $store::TTL_DAY ); |
| 2345 | } |
| 2346 | } |
| 2347 | |
| 2348 | /** |
| 2349 | * @param BagOStuff $store |
| 2350 | * @param UserIdentity $user |
| 2351 | * @param string $statusKey |
| 2352 | * @return string |
| 2353 | */ |
| 2354 | private static function getUploadSessionKey( BagOStuff $store, UserIdentity $user, $statusKey ) { |
| 2355 | return $store->makeKey( |
| 2356 | 'uploadstatus', |
| 2357 | $user->isRegistered() ? $user->getId() : md5( $user->getName() ), |
| 2358 | $statusKey |
| 2359 | ); |
| 2360 | } |
| 2361 | |
| 2362 | /** |
| 2363 | * @return BagOStuff |
| 2364 | */ |
| 2365 | private static function getUploadSessionStore() { |
| 2366 | return MediaWikiServices::getInstance()->getMainObjectStash(); |
| 2367 | } |
| 2368 | } |