Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
77.67% |
400 / 515 |
|
28.57% |
4 / 14 |
CRAP | |
0.00% |
0 / 1 |
| UploadVerification | |
77.67% |
400 / 515 |
|
28.57% |
4 / 14 |
329.96 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| verifyMimeType | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| verifyFile | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
156 | |||
| verifyPartialFile | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
90 | |||
| verifyExtension | |
72.50% |
29 / 40 |
|
0.00% |
0 / 1 |
9.33 | |||
| detectScript | |
65.96% |
31 / 47 |
|
0.00% |
0 / 1 |
26.10 | |||
| checkXMLEncodingMismatch | |
51.61% |
16 / 31 |
|
0.00% |
0 / 1 |
32.15 | |||
| detectScriptInSvg | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
4.01 | |||
| checkSvgPICallback | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| checkSvgExternalDTD | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
4 | |||
| checkSvgScriptCallback | |
90.84% |
238 / 262 |
|
0.00% |
0 / 1 |
45.49 | |||
| splitXmlNamespace | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| stripXmlNamespace | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| detectVirus | |
95.92% |
47 / 49 |
|
0.00% |
0 / 1 |
15 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Base class for the backend of file upload. |
| 4 | * |
| 5 | * @license GPL-2.0-or-later |
| 6 | * @file |
| 7 | * @ingroup Upload |
| 8 | */ |
| 9 | |
| 10 | namespace MediaWiki\Upload; |
| 11 | |
| 12 | use MediaWiki\Config\ConfigException; |
| 13 | use MediaWiki\Config\ServiceOptions; |
| 14 | use MediaWiki\MainConfigNames; |
| 15 | use MediaWiki\Media\MediaHandler; |
| 16 | use MediaWiki\Parser\Sanitizer; |
| 17 | use MediaWiki\Shell\Shell; |
| 18 | use Psr\Log\LoggerInterface; |
| 19 | use Wikimedia\Mime\MimeAnalyzer; |
| 20 | use Wikimedia\Mime\XmlTypeCheck; |
| 21 | |
| 22 | /** |
| 23 | * @ingroup Upload |
| 24 | * |
| 25 | * Service to verify file uploads are safe. |
| 26 | * |
| 27 | * This is responsible for checks on the file contents themselves. It |
| 28 | * is not responsible for on wiki checks like if the user has permission |
| 29 | * or if the upload target is protected. |
| 30 | * |
| 31 | * @author Brian Wolff |
| 32 | * @since 1.45 |
| 33 | */ |
| 34 | class UploadVerification { |
| 35 | |
| 36 | private const SAFE_XML_ENCODINGS = [ |
| 37 | 'UTF-8', |
| 38 | 'US-ASCII', |
| 39 | 'ISO-8859-1', |
| 40 | 'ISO-8859-2', |
| 41 | 'UTF-16', |
| 42 | 'UTF-32', |
| 43 | 'WINDOWS-1250', |
| 44 | 'WINDOWS-1251', |
| 45 | 'WINDOWS-1252', |
| 46 | 'WINDOWS-1253', |
| 47 | 'WINDOWS-1254', |
| 48 | 'WINDOWS-1255', |
| 49 | 'WINDOWS-1256', |
| 50 | 'WINDOWS-1257', |
| 51 | 'WINDOWS-1258', |
| 52 | ]; |
| 53 | |
| 54 | public const CONSTRUCTOR_OPTIONS = [ |
| 55 | MainConfigNames::VerifyMimeType, |
| 56 | MainConfigNames::MimeTypeExclusions, |
| 57 | MainConfigNames::DisableUploadScriptChecks, |
| 58 | MainConfigNames::Antivirus, |
| 59 | MainConfigNames::AntivirusSetup, |
| 60 | MainConfigNames::AntivirusRequired |
| 61 | ]; |
| 62 | |
| 63 | private SvgCssChecker $svgCssChecker; |
| 64 | |
| 65 | public function __construct( |
| 66 | private ServiceOptions $config, |
| 67 | private MimeAnalyzer $mimeAnalyzer, |
| 68 | private LoggerInterface $logger, |
| 69 | ) { |
| 70 | $config->assertRequiredOptions( self::CONSTRUCTOR_OPTIONS ); |
| 71 | $this->svgCssChecker = new SvgCssChecker; |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * Verify the MIME type. |
| 76 | * |
| 77 | * @note Only checks that it is not an evil MIME. |
| 78 | * The "does it have the correct file extension given its MIME type?" check is in verifyFile. |
| 79 | * @param string $mime Representing the MIME |
| 80 | * @return array|bool True if the file is verified, an array otherwise |
| 81 | */ |
| 82 | private function verifyMimeType( $mime ) { |
| 83 | $verifyMimeType = $this->config->get( MainConfigNames::VerifyMimeType ); |
| 84 | if ( $verifyMimeType ) { |
| 85 | $mimeTypeExclusions = $this->config |
| 86 | ->get( MainConfigNames::MimeTypeExclusions ); |
| 87 | if ( UploadBase::checkFileExtension( $mime, $mimeTypeExclusions ) ) { |
| 88 | return [ 'filetype-badmime', $mime ]; |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | return true; |
| 93 | } |
| 94 | |
| 95 | /** |
| 96 | * Verifies that the upload file is safe |
| 97 | * |
| 98 | * @note This verifies the contents of the file. It is not |
| 99 | * responsible for verifying if the file has a valid name, is |
| 100 | * too big, meets on wiki permission checks, etc. If you are |
| 101 | * implementing your own upload support, see |
| 102 | * UploadBase::verifyUpload for other necessary checks. |
| 103 | * |
| 104 | * @param string $path Path to the (temporary) file to check |
| 105 | * @param string $ext Final extension of file (UploadBase->mFinalExtension) |
| 106 | * @param array $fileProps Result of $mwProps->getPropsFromPath. |
| 107 | * FIXME final ext can sometimes be null, but should we require casting to string? |
| 108 | * @return array|true True of the file is verified, array otherwise. |
| 109 | */ |
| 110 | public function verifyFile( string $path, string $ext, array $fileProps ) { |
| 111 | $config = $this->config; |
| 112 | $verifyMimeType = $config->get( MainConfigNames::VerifyMimeType ); |
| 113 | $disableUploadScriptChecks = $config->get( MainConfigNames::DisableUploadScriptChecks ); |
| 114 | $status = $this->verifyPartialFile( $path, $ext, $fileProps ); |
| 115 | if ( $status !== true ) { |
| 116 | return $status; |
| 117 | } |
| 118 | |
| 119 | $mime = $fileProps['mime']; |
| 120 | |
| 121 | if ( $verifyMimeType ) { |
| 122 | # XXX: Missing extension will be caught by validateName() via getTitle() |
| 123 | if ( $ext !== '' && |
| 124 | !$this->verifyExtension( $mime, $ext ) |
| 125 | ) { |
| 126 | return [ 'filetype-mime-mismatch', $ext, $mime ]; |
| 127 | } |
| 128 | } |
| 129 | |
| 130 | # check for htmlish code and javascript |
| 131 | if ( !$disableUploadScriptChecks ) { |
| 132 | if ( $ext === 'svg' || $mime === 'image/svg+xml' ) { |
| 133 | $svgStatus = $this->detectScriptInSvg( $path, false ); |
| 134 | if ( $svgStatus !== false ) { |
| 135 | return $svgStatus; |
| 136 | } |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | $handler = $mime !== null ? MediaHandler::getHandler( $mime ) : null; |
| 141 | if ( $handler ) { |
| 142 | $handlerStatus = $handler->verifyUpload( $path ); |
| 143 | if ( !$handlerStatus->isOK() ) { |
| 144 | $errors = $handlerStatus->getErrorsArray(); |
| 145 | |
| 146 | return reset( $errors ); |
| 147 | } |
| 148 | } |
| 149 | |
| 150 | // TODO: Perhaps we should have a hook here akin to UploadVerifyFile |
| 151 | // except that it doesn't pass an UploadBase, and it would run |
| 152 | // even when someone is verifying a file not going through UploadBase. |
| 153 | |
| 154 | $this->logger->debug( 'verifyFile: all clear; passing.' ); |
| 155 | |
| 156 | return true; |
| 157 | } |
| 158 | |
| 159 | /** |
| 160 | * A verification routine suitable for partial files |
| 161 | * |
| 162 | * Runs the deny list checks, but not any checks that may |
| 163 | * assume the entire file is present. |
| 164 | * |
| 165 | * fileProps can be very expensive to calculate, so the calling class is |
| 166 | * responsible for caching it. |
| 167 | * |
| 168 | * @param string $path Path to the (temporary) file to check |
| 169 | * @param string $ext Final extension of file (UploadBase->mFinalExtension) |
| 170 | * @param array $fileProps Result of $mwProps->getPropsFromPath |
| 171 | * @return array|true True, if the file is valid, else an array with error message key. |
| 172 | * @phan-return non-empty-array|true |
| 173 | */ |
| 174 | public function verifyPartialFile( string $path, string $ext, array $fileProps ) { |
| 175 | $config = $this->config; |
| 176 | $disableUploadScriptChecks = $config->get( MainConfigNames::DisableUploadScriptChecks ); |
| 177 | |
| 178 | # check MIME type, if desired |
| 179 | $mime = $fileProps['file-mime']; |
| 180 | $status = $this->verifyMimeType( $mime ); |
| 181 | $this->logger->debug( 'verifyMimeType: {mime} {status}', |
| 182 | [ 'mime' => $mime, 'status' => $status === true ? 'OK' : 'bad' ] ); |
| 183 | if ( $status !== true ) { |
| 184 | return $status; |
| 185 | } |
| 186 | |
| 187 | # check for htmlish code and javascript |
| 188 | if ( !$disableUploadScriptChecks ) { |
| 189 | if ( $this->detectScript( $path, $mime, $ext ) ) { |
| 190 | return [ 'uploadscripted' ]; |
| 191 | } |
| 192 | if ( $ext === 'svg' || $mime === 'image/svg+xml' ) { |
| 193 | $svgStatus = $this->detectScriptInSvg( $path, true ); |
| 194 | if ( $svgStatus !== false ) { |
| 195 | return $svgStatus; |
| 196 | } |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | # Scan the uploaded file for viruses |
| 201 | $virus = $this->detectVirus( $path ); |
| 202 | if ( $virus ) { |
| 203 | return [ 'uploadvirus', $virus ]; |
| 204 | } |
| 205 | |
| 206 | return true; |
| 207 | } |
| 208 | |
| 209 | /** |
| 210 | * Checks if the MIME type of the uploaded file matches the file extension. |
| 211 | * |
| 212 | * @internal Will become private once UploadBase::verifyExtension is removed |
| 213 | * @param string $mime The MIME type of the uploaded file |
| 214 | * @param string $extension The filename extension that the file is to be served with |
| 215 | * @return bool |
| 216 | */ |
| 217 | public function verifyExtension( $mime, $extension ) { |
| 218 | $magic = $this->mimeAnalyzer; |
| 219 | $logContext = [ 'mime' => $mime, 'extension' => $extension ]; |
| 220 | |
| 221 | if ( !$mime || $mime === 'unknown' || $mime === 'unknown/unknown' ) { |
| 222 | if ( !$magic->isRecognizableExtension( $extension ) ) { |
| 223 | $this->logger->debug( |
| 224 | 'verifyExtension: passing file with unknown detected MIME type; ' . |
| 225 | "unrecognized extension '{extension}', can't verify", |
| 226 | $logContext |
| 227 | ); |
| 228 | |
| 229 | return true; |
| 230 | } |
| 231 | |
| 232 | $this->logger->debug( |
| 233 | 'verifyExtension: rejecting file with unknown detected MIME type; ' . |
| 234 | 'recognized extension "{extension}", so probably invalid file', |
| 235 | $logContext |
| 236 | ); |
| 237 | return false; |
| 238 | } |
| 239 | |
| 240 | $match = $magic->isMatchingExtension( $extension, $mime ); |
| 241 | |
| 242 | if ( $match === null ) { |
| 243 | if ( $magic->getMimeTypesFromExtension( $extension ) !== [] ) { |
| 244 | $this->logger->debug( |
| 245 | 'verifyExtension: No extension known for {mime}, but we know a MIME for {extension}', |
| 246 | $logContext |
| 247 | ); |
| 248 | return false; |
| 249 | } |
| 250 | |
| 251 | $this->logger->debug( |
| 252 | 'verifyExtension: no file extension known for MIME type {mime}, passing file', |
| 253 | $logContext |
| 254 | ); |
| 255 | return true; |
| 256 | } |
| 257 | |
| 258 | if ( $match ) { |
| 259 | $this->logger->debug( |
| 260 | 'verifyExtension: MIME type {mime} matches extension {extension}, passing file', |
| 261 | $logContext |
| 262 | ); |
| 263 | |
| 264 | /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */ |
| 265 | return true; |
| 266 | } |
| 267 | |
| 268 | $this->logger->debug( |
| 269 | 'verifyExtension: MIME type {mime} mismatches file extension {extension}, rejecting file', |
| 270 | $logContext |
| 271 | ); |
| 272 | |
| 273 | return false; |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Heuristic for detecting files that *could* contain JavaScript instructions or |
| 278 | * things that may look like HTML to a browser and are thus |
| 279 | * potentially harmful. The present implementation will produce false |
| 280 | * positives in some situations. |
| 281 | * |
| 282 | * @internal This is public for back-compat. Some extensions call this, however |
| 283 | * this is probably not the method they want. Instead they should call verifyFile(). |
| 284 | * Calling this outside this class should be considered deprecated and the method |
| 285 | * may become private in the future. |
| 286 | * @param string|null $file Pathname to the temporary upload file |
| 287 | * @param string $mime The MIME type of the file |
| 288 | * @param string|null $extension The extension of the file |
| 289 | * @return bool True if the file contains something looking like embedded scripts |
| 290 | */ |
| 291 | public function detectScript( $file, $mime, $extension ) { |
| 292 | # ugly hack: for text files, always look at the entire file. |
| 293 | # For binary field, just check the first K. |
| 294 | |
| 295 | if ( str_starts_with( $mime ?? '', 'text/' ) ) { |
| 296 | $chunk = file_get_contents( $file ); |
| 297 | } else { |
| 298 | $fp = fopen( $file, 'rb' ); |
| 299 | if ( !$fp ) { |
| 300 | return false; |
| 301 | } |
| 302 | $chunk = fread( $fp, 1024 ); |
| 303 | fclose( $fp ); |
| 304 | } |
| 305 | |
| 306 | $chunk = strtolower( $chunk ); |
| 307 | |
| 308 | if ( !$chunk ) { |
| 309 | return false; |
| 310 | } |
| 311 | |
| 312 | # decode from UTF-16 if needed (could be used for obfuscation). |
| 313 | if ( str_starts_with( $chunk, "\xfe\xff" ) ) { |
| 314 | $enc = 'UTF-16BE'; |
| 315 | } elseif ( str_starts_with( $chunk, "\xff\xfe" ) ) { |
| 316 | $enc = 'UTF-16LE'; |
| 317 | } else { |
| 318 | $enc = null; |
| 319 | } |
| 320 | |
| 321 | if ( $enc !== null ) { |
| 322 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 323 | $chunk = @iconv( $enc, "ASCII//IGNORE", $chunk ); |
| 324 | } |
| 325 | |
| 326 | $chunk = trim( $chunk ); |
| 327 | |
| 328 | /** @todo FIXME: Convert from UTF-16 if necessary! */ |
| 329 | |
| 330 | # check for HTML doctype |
| 331 | if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) { |
| 332 | $this->logger->debug( 'detectScript: found doctype html' ); |
| 333 | return true; |
| 334 | } |
| 335 | |
| 336 | // Some browsers will interpret obscure xml encodings as UTF-8, while |
| 337 | // PHP/expat will interpret the given encoding in the xml declaration (T49304) |
| 338 | if ( $extension === 'svg' || str_starts_with( $mime ?? '', 'image/svg' ) ) { |
| 339 | if ( $this->checkXMLEncodingMismatch( $file ) ) { |
| 340 | $this->logger->debug( 'detectScript: found SVG encoding mismatch' ); |
| 341 | return true; |
| 342 | } |
| 343 | } |
| 344 | |
| 345 | // Quick check for HTML heuristics in old IE and Safari. |
| 346 | // |
| 347 | // The exact heuristics IE uses are checked separately via verifyMimeType(), so we |
| 348 | // don't need them all here as it can cause many false positives. |
| 349 | // |
| 350 | // Check for `<script` and such still to forbid script tags and embedded HTML in SVG: |
| 351 | $tags = [ |
| 352 | '<body', |
| 353 | '<head', |
| 354 | '<html', # also in safari |
| 355 | '<script', # also in safari |
| 356 | ]; |
| 357 | |
| 358 | foreach ( $tags as $tag ) { |
| 359 | if ( str_contains( $chunk, $tag ) ) { |
| 360 | $this->logger->debug( 'detectScript: found HTML tag "{tag}"', [ 'tag' => $tag ] ); |
| 361 | return true; |
| 362 | } |
| 363 | } |
| 364 | |
| 365 | /* |
| 366 | * look for JavaScript |
| 367 | */ |
| 368 | |
| 369 | # resolve entity-refs to look at attributes. may be harsh on big files... cache result? |
| 370 | $chunk = Sanitizer::decodeCharReferences( $chunk ); |
| 371 | |
| 372 | # look for script-types |
| 373 | if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!im', $chunk ) ) { |
| 374 | $this->logger->debug( 'detectScript: found script types' ); |
| 375 | return true; |
| 376 | } |
| 377 | |
| 378 | # look for html-style script-urls |
| 379 | if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!im', $chunk ) ) { |
| 380 | $this->logger->debug( 'detectScript: found HTML-style script URLs' ); |
| 381 | return true; |
| 382 | } |
| 383 | |
| 384 | # look for css-style script-urls |
| 385 | if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!im', $chunk ) ) { |
| 386 | $this->logger->debug( 'detectScript: found CSS-style script URLs' ); |
| 387 | return true; |
| 388 | } |
| 389 | |
| 390 | $this->logger->debug( 'detectScript: no scripts found' ); |
| 391 | return false; |
| 392 | } |
| 393 | |
| 394 | /** |
| 395 | * Check an allowed list of xml encodings that are known not to be interpreted differently |
| 396 | * by the server's xml parser (expat) and some common browsers. |
| 397 | * |
| 398 | * @param string $file Pathname to the temporary upload file |
| 399 | * @return bool True if the file contains an encoding that could be misinterpreted |
| 400 | */ |
| 401 | private function checkXMLEncodingMismatch( $file ) { |
| 402 | // https://mimesniff.spec.whatwg.org/#resource-header says browsers |
| 403 | // should read the first 1445 bytes. Do 4096 bytes for good measure. |
| 404 | // XML Spec says XML declaration if present must be first thing in file |
| 405 | // other than BOM |
| 406 | $contents = file_get_contents( $file, false, null, 0, 4096 ); |
| 407 | $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si'; |
| 408 | |
| 409 | if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) { |
| 410 | if ( preg_match( $encodingRegex, $matches[1], $encMatch ) |
| 411 | && !in_array( strtoupper( $encMatch[1] ), self::SAFE_XML_ENCODINGS ) |
| 412 | ) { |
| 413 | $this->logger->debug( |
| 414 | 'checkXMLEncodingMismatch: Found unsafe XML encoding "{encoding}"', |
| 415 | [ 'encoding' => $encMatch[1] ] |
| 416 | ); |
| 417 | return true; |
| 418 | } |
| 419 | } elseif ( preg_match( "!<\?xml\b!i", $contents ) ) { |
| 420 | // Start of XML declaration without an end in the first 4096 bytes |
| 421 | // bytes. There shouldn't be a legitimate reason for this to happen. |
| 422 | $this->logger->debug( 'checkXMLEncodingMismatch: Unmatched XML declaration start' ); |
| 423 | return true; |
| 424 | } elseif ( str_starts_with( $contents, "\x4C\x6F\xA7\x94" ) ) { |
| 425 | // EBCDIC encoded XML |
| 426 | $this->logger->debug( 'checkXMLEncodingMismatch: EBCDIC Encoded XML' ); |
| 427 | return true; |
| 428 | } |
| 429 | |
| 430 | // It's possible the file is encoded with multibyte encoding, so re-encode attempt to |
| 431 | // detect the encoding in case it specifies an encoding not allowed in self::SAFE_XML_ENCODINGS |
| 432 | $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ]; |
| 433 | foreach ( $attemptEncodings as $encoding ) { |
| 434 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 435 | $str = @iconv( $encoding, 'UTF-8', $contents ); |
| 436 | if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) { |
| 437 | if ( preg_match( $encodingRegex, $matches[1], $encMatch ) |
| 438 | && !in_array( strtoupper( $encMatch[1] ), self::SAFE_XML_ENCODINGS ) |
| 439 | ) { |
| 440 | $this->logger->debug( |
| 441 | 'checkXMLEncodingMismatch: Found unsafe XML encoding "{encoding}"', |
| 442 | [ 'encoding' => $encMatch[1] ] |
| 443 | ); |
| 444 | return true; |
| 445 | } |
| 446 | } elseif ( $str != '' && preg_match( "!<\?xml\b!i", $str ) ) { |
| 447 | // Start of XML declaration without an end in the first 4096 bytes |
| 448 | // bytes. There shouldn't be a legitimate reason for this to happen. |
| 449 | $this->logger->debug( 'checkXMLEncodingMismatch: Unmatched XML declaration start' ); |
| 450 | |
| 451 | return true; |
| 452 | } |
| 453 | } |
| 454 | |
| 455 | return false; |
| 456 | } |
| 457 | |
| 458 | /** |
| 459 | * Looks for bad SVG files |
| 460 | * |
| 461 | * @warning This function is only safe for the XML serialization of SVGs. |
| 462 | * @param string $filename |
| 463 | * @param bool $partial |
| 464 | * @return bool|array |
| 465 | */ |
| 466 | private function detectScriptInSvg( $filename, $partial ) { |
| 467 | $check = new XmlTypeCheck( |
| 468 | $filename, |
| 469 | $this->checkSvgScriptCallback( ... ), |
| 470 | true, |
| 471 | [ |
| 472 | 'processing_instruction_handler' => $this->checkSvgPICallback( ... ), |
| 473 | 'external_dtd_handler' => $this->checkSvgExternalDTD( ... ), |
| 474 | ] |
| 475 | ); |
| 476 | if ( $check->wellFormed !== true ) { |
| 477 | // Invalid xml (T60553) |
| 478 | // But only when non-partial (T67724) |
| 479 | return $partial ? false : [ 'uploadinvalidxml' ]; |
| 480 | } |
| 481 | |
| 482 | if ( $check->filterMatch ) { |
| 483 | return $check->filterMatchType; |
| 484 | } |
| 485 | |
| 486 | return false; |
| 487 | } |
| 488 | |
| 489 | /** |
| 490 | * Callback to filter SVG Processing Instructions. |
| 491 | * |
| 492 | * @param string $target Processing instruction name |
| 493 | * @return bool|array |
| 494 | */ |
| 495 | private function checkSvgPICallback( $target ) { |
| 496 | // Don't allow external stylesheets (T59550) |
| 497 | if ( preg_match( '/xml-stylesheet/i', $target ) ) { |
| 498 | return [ 'upload-scripted-pi-callback' ]; |
| 499 | } |
| 500 | |
| 501 | return false; |
| 502 | } |
| 503 | |
| 504 | /** |
| 505 | * Verify that DTD URLs referenced are only the standard DTDs. |
| 506 | * |
| 507 | * Browsers seem to ignore external DTDs. |
| 508 | * |
| 509 | * However, just to be on the safe side, only allow DTDs from the SVG standard. |
| 510 | * |
| 511 | * @param string $type PUBLIC or SYSTEM |
| 512 | * @param string $publicId The well-known public identifier for the dtd |
| 513 | * @param string $systemId The url for the external dtd |
| 514 | * @return bool|array |
| 515 | */ |
| 516 | private function checkSvgExternalDTD( $type, $publicId, $systemId ) { |
| 517 | // This doesn't include the XHTML+MathML+SVG doctype since we don't |
| 518 | // allow XHTML anyway. |
| 519 | static $allowedDTDs = [ |
| 520 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd', |
| 521 | 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd', |
| 522 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd', |
| 523 | 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd', |
| 524 | // https://phabricator.wikimedia.org/T168856 |
| 525 | 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd', |
| 526 | ]; |
| 527 | if ( $type !== 'PUBLIC' |
| 528 | || !in_array( $systemId, $allowedDTDs ) |
| 529 | || !str_starts_with( $publicId, "-//W3C//" ) |
| 530 | ) { |
| 531 | return [ 'upload-scripted-dtd' ]; |
| 532 | } |
| 533 | return false; |
| 534 | } |
| 535 | |
| 536 | /** |
| 537 | * Callback to XML parsing checking individual tags for evilness |
| 538 | * @warning This assumes that the SVG is using the XML serialization. |
| 539 | * It is not safe if the SVG is directly embedded in HTML. |
| 540 | * @todo Replace this with a allow list filter! |
| 541 | * @param string $element |
| 542 | * @param array $attribs |
| 543 | * @param string|null $data |
| 544 | * @return bool|array |
| 545 | */ |
| 546 | private function checkSvgScriptCallback( $element, $attribs, $data = null ) { |
| 547 | [ $namespace, $strippedElement ] = self::splitXmlNamespace( $element ); |
| 548 | |
| 549 | $logContext = [ |
| 550 | 'element' => $element, |
| 551 | 'namespace' => $namespace, |
| 552 | 'localPart' => $strippedElement |
| 553 | ]; |
| 554 | |
| 555 | // We specifically don't include: |
| 556 | // http://www.w3.org/1999/xhtml (T62771) |
| 557 | static $validNamespaces = [ |
| 558 | '', |
| 559 | 'adobe:ns:meta/', |
| 560 | 'http://cipa.jp/exif/1.0/', |
| 561 | 'http://creativecommons.org/ns#', |
| 562 | 'http://developer.sonyericsson.com/cell/1.0/', |
| 563 | 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd', |
| 564 | 'http://iptc.org/std/iptc4xmpcore/1.0/xmlns/', |
| 565 | 'http://iptc.org/std/iptc4xmpext/2008-02-29/', |
| 566 | 'http://leica-camera.com/digital-shift-assistant/1.0/', |
| 567 | 'http://ns.acdsee.com/iptc/1.0/', |
| 568 | 'http://ns.acdsee.com/regions/', |
| 569 | 'http://ns.adobe.com/adobeillustrator/10.0/', |
| 570 | 'http://ns.adobe.com/adobesvgviewerextensions/3.0/', |
| 571 | 'http://ns.adobe.com/album/1.0/', |
| 572 | 'http://ns.adobe.com/camera-raw-defaults/1.0/', |
| 573 | 'http://ns.adobe.com/camera-raw-embedded-lens-profile/1.0/', |
| 574 | 'http://ns.adobe.com/camera-raw-saved-settings/1.0/', |
| 575 | 'http://ns.adobe.com/camera-raw-settings/1.0/', |
| 576 | 'http://ns.adobe.com/creatoratom/1.0/', |
| 577 | 'http://ns.adobe.com/dicom/', |
| 578 | 'http://ns.adobe.com/exif/1.0/', |
| 579 | 'http://ns.adobe.com/exif/1.0/aux/', |
| 580 | 'http://ns.adobe.com/extensibility/1.0/', |
| 581 | 'http://ns.adobe.com/flows/1.0/', |
| 582 | 'http://ns.adobe.com/hdr-gain-map/1.0/', |
| 583 | 'http://ns.adobe.com/hdr-metadata/1.0/', |
| 584 | 'http://ns.adobe.com/ix/1.0/', |
| 585 | 'http://ns.adobe.com/lightroom/1.0/', |
| 586 | 'http://ns.adobe.com/illustrator/1.0/', |
| 587 | 'http://ns.adobe.com/imagereplacement/1.0/', |
| 588 | 'http://ns.adobe.com/pdf/1.3/', |
| 589 | 'http://ns.adobe.com/pdfx/1.3/', |
| 590 | 'http://ns.adobe.com/photoshop/1.0/', |
| 591 | 'http://ns.adobe.com/photoshop/1.0/camera-profile', |
| 592 | 'http://ns.adobe.com/photoshop/1.0/panorama-profile', |
| 593 | 'http://ns.adobe.com/raw/1.0/', |
| 594 | 'http://ns.adobe.com/swf/1.0/', |
| 595 | 'http://ns.adobe.com/saveforweb/1.0/', |
| 596 | 'http://ns.adobe.com/tiff/1.0/', |
| 597 | 'http://ns.adobe.com/variables/1.0/', |
| 598 | 'http://ns.adobe.com/xap/1.0/', |
| 599 | 'http://ns.adobe.com/xap/1.0/bj/', |
| 600 | 'http://ns.adobe.com/xap/1.0/g/', |
| 601 | 'http://ns.adobe.com/xap/1.0/g/img/', |
| 602 | 'http://ns.adobe.com/xap/1.0/mm/', |
| 603 | 'http://ns.adobe.com/xap/1.0/plus/', |
| 604 | 'http://ns.adobe.com/xap/1.0/rights/', |
| 605 | 'http://ns.adobe.com/xap/1.0/stype/dimensions#', |
| 606 | 'http://ns.adobe.com/xap/1.0/stype/font#', |
| 607 | 'http://ns.adobe.com/xap/1.0/stype/manifestitem#', |
| 608 | 'http://ns.adobe.com/xap/1.0/stype/resourceevent#', |
| 609 | 'http://ns.adobe.com/xap/1.0/stype/resourceref#', |
| 610 | 'http://ns.adobe.com/xap/1.0/stype/version#', |
| 611 | 'http://ns.adobe.com/xap/1.0/t/pg/', |
| 612 | 'http://ns.adobe.com/xmp/1.0/dynamicmedia/', |
| 613 | 'http://ns.adobe.com/xmp/identifier/qual/1.0/', |
| 614 | 'http://ns.adobe.com/xmp/note/', |
| 615 | 'http://ns.adobe.com/xmp/stype/area#', |
| 616 | 'http://ns.apple.com/adjustment-settings/1.0/', |
| 617 | 'http://ns.apple.com/faceinfo/1.0/', |
| 618 | 'http://ns.apple.com/hdrgainmap/1.0/', |
| 619 | 'http://ns.apple.com/pixeldatainfo/1.0/', |
| 620 | 'http://ns.exiftool.org/1.0/', |
| 621 | 'http://ns.extensis.com/extensis/1.0/', |
| 622 | 'http://ns.fastpictureviewer.com/fpv/1.0/', |
| 623 | 'http://ns.google.com/photos/1.0/audio/', |
| 624 | 'http://ns.google.com/photos/1.0/camera/', |
| 625 | 'http://ns.google.com/photos/1.0/container/', |
| 626 | 'http://ns.google.com/photos/1.0/creations/', |
| 627 | 'http://ns.google.com/photos/1.0/depthmap/', |
| 628 | 'http://ns.google.com/photos/1.0/focus/', |
| 629 | 'http://ns.google.com/photos/1.0/image/', |
| 630 | 'http://ns.google.com/photos/1.0/panorama/', |
| 631 | 'http://ns.google.com/photos/dd/1.0/profile/', |
| 632 | 'http://ns.google.com/videos/1.0/spherical/', |
| 633 | 'http://ns.idimager.com/ics/1.0/', |
| 634 | 'http://ns.iview-multimedia.com/mediapro/1.0/', |
| 635 | 'http://ns.leiainc.com/photos/1.0/image/', |
| 636 | 'http://ns.microsoft.com/expressionmedia/1.0/', |
| 637 | 'http://ns.microsoft.com/photo/1.0', |
| 638 | 'http://ns.microsoft.com/photo/1.1', |
| 639 | 'http://ns.microsoft.com/photo/1.2/', |
| 640 | 'http://ns.microsoft.com/photo/1.2/t/region#', |
| 641 | 'http://ns.microsoft.com/photo/1.2/t/regioninfo#', |
| 642 | 'http://ns.nikon.com/asteroid/1.0/', |
| 643 | 'http://ns.nikon.com/nine/1.0/', |
| 644 | 'http://ns.nikon.com/sdc/1.0/', |
| 645 | 'http://ns.optimasc.com/dex/1.0/', |
| 646 | 'http://ns.seal/2024/1.0/', |
| 647 | 'http://ns.useplus.org/ldf/xmp/1.0/', |
| 648 | 'http://prismstandard.org/namespaces/basic/2.0/', |
| 649 | 'http://prismstandard.org/namespaces/pmi/2.2/', |
| 650 | 'http://prismstandard.org/namespaces/prismusagerights/2.1/', |
| 651 | 'http://prismstandard.org/namespaces/prl/2.1/', |
| 652 | 'http://prismstandard.org/namespaces/prm/3.0/', |
| 653 | 'http://purl.org/dc/elements/1.1/', |
| 654 | 'http://purl.org/dc/elements/1.1', |
| 655 | 'http://rs.tdwg.org/dwc/index.htm', |
| 656 | 'http://schemas.microsoft.com/visio/2003/svgextensions/', |
| 657 | 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd', |
| 658 | 'http://taptrix.com/inkpad/svg_extensions', |
| 659 | 'http://www.digikam.org/ns/1.0/', |
| 660 | 'http://www.dji.com/drone-dji/1.0/', |
| 661 | 'http://www.metadataworkinggroup.com/schemas/collections/', |
| 662 | 'http://www.metadataworkinggroup.com/schemas/keywords/', |
| 663 | 'http://www.metadataworkinggroup.com/schemas/regions/', |
| 664 | 'http://web.resource.org/cc/', |
| 665 | 'http://www.freesoftware.fsf.org/bkchem/cdml', |
| 666 | 'http://www.inkscape.org/namespaces/inkscape', |
| 667 | 'http://www.opengis.net/gml', |
| 668 | 'http://www.w3.org/1999/02/22-rdf-syntax-ns#', |
| 669 | 'http://www.w3.org/2000/01/rdf-schema#', |
| 670 | 'http://www.w3.org/2000/svg', |
| 671 | 'http://www.w3.org/2000/02/svg/testsuite/description/', // T278044 |
| 672 | 'http://www.w3.org/tr/rec-rdf-syntax/', |
| 673 | 'http://xmp.gettyimages.com/gift/1.0/', |
| 674 | ]; |
| 675 | |
| 676 | // Inkscape mangles namespace definitions created by Adobe Illustrator. |
| 677 | // This is nasty but harmless. (T144827) |
| 678 | $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace ); |
| 679 | |
| 680 | if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) { |
| 681 | $this->logger->debug( |
| 682 | 'detectScriptInSvg: Non-SVG namespace "{namespace}" in uploaded file.', |
| 683 | $logContext |
| 684 | ); |
| 685 | return [ 'uploadscriptednamespace', $namespace ]; |
| 686 | } |
| 687 | |
| 688 | // check for elements that can contain javascript |
| 689 | if ( $strippedElement === 'script' ) { |
| 690 | $this->logger->debug( |
| 691 | 'detectScriptInSvg: Found script element "{element}" in uploaded file.', |
| 692 | $logContext |
| 693 | ); |
| 694 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 695 | } |
| 696 | |
| 697 | // e.g., <svg xmlns="http://www.w3.org/2000/svg"> |
| 698 | // <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg> |
| 699 | if ( $strippedElement === 'handler' ) { |
| 700 | $this->logger->debug( |
| 701 | 'detectScriptInSvg: Found scriptable element "{element}" in uploaded file.', |
| 702 | $logContext |
| 703 | ); |
| 704 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 705 | } |
| 706 | |
| 707 | // SVG reported in Feb '12 that used xml:stylesheet to generate javascript block |
| 708 | if ( $strippedElement === 'stylesheet' ) { |
| 709 | $this->logger->debug( |
| 710 | 'detectScriptInSvg: Found scriptable element "{element}" in uploaded file.', |
| 711 | $logContext |
| 712 | ); |
| 713 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 714 | } |
| 715 | |
| 716 | // Block iframes, in case they pass the namespace check |
| 717 | if ( $strippedElement === 'iframe' ) { |
| 718 | $this->logger->debug( "detectScriptInSvg: iframe in uploaded file.", $logContext ); |
| 719 | return [ 'uploaded-script-svg', $strippedElement ]; |
| 720 | } |
| 721 | |
| 722 | // Check <style> css |
| 723 | if ( $strippedElement === 'style' ) { |
| 724 | $cssCheck = $this->svgCssChecker->checkStyleTag( $data ); |
| 725 | if ( $cssCheck !== true ) { |
| 726 | $this->logger->debug( |
| 727 | "detectScriptInSvg: hostile CSS in style element. {tag}", |
| 728 | [ 'tag' => $cssCheck[0] ] |
| 729 | ); |
| 730 | return [ 'uploaded-hostile-svg', $cssCheck[0], $cssCheck[1], $cssCheck[2] ]; |
| 731 | } |
| 732 | } |
| 733 | |
| 734 | static $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker', |
| 735 | 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke', 'cursor' ]; |
| 736 | |
| 737 | foreach ( $attribs as $attrib => $value ) { |
| 738 | // If attributeNamespace is '', it is relative to its element's namespace |
| 739 | [ $attributeNamespace, $stripped ] = self::splitXmlNamespace( $attrib ); |
| 740 | $value = strtolower( $value ); |
| 741 | $attribLogContext = [ 'attrib' => $attrib, 'value' => $value ] + $logContext; |
| 742 | |
| 743 | if ( !( |
| 744 | // Inkscape element's have valid attribs that start with on and are safe, fail all others |
| 745 | // We are assuming here that the SVG will be interpreted |
| 746 | // under XML serialization. This is not safe for SVGs |
| 747 | // embedded directly in HTML. |
| 748 | $namespace === 'http://www.inkscape.org/namespaces/inkscape' && |
| 749 | $attributeNamespace === '' |
| 750 | ) && str_starts_with( $stripped, 'on' ) |
| 751 | ) { |
| 752 | $this->logger->debug( |
| 753 | 'detectScriptInSvg: Found event-handler attribute ' . |
| 754 | '{attrib}="{value}" in uploaded file.', |
| 755 | $attribLogContext |
| 756 | ); |
| 757 | return [ 'uploaded-event-handler-on-svg', $attrib, $value ]; |
| 758 | } |
| 759 | |
| 760 | // Do not allow relative links, or unsafe url schemas. |
| 761 | // For <a> tags, only data:, http: and https: and same-document |
| 762 | // fragment links are allowed. |
| 763 | // For all other tags, only 'data:' and fragments (#) are allowed. |
| 764 | if ( |
| 765 | $stripped === 'href' |
| 766 | && $value !== '' |
| 767 | && !str_starts_with( $value, 'data:' ) |
| 768 | && !str_starts_with( $value, '#' ) |
| 769 | && !( $strippedElement === 'a' && preg_match( '!^https?://!i', $value ) ) |
| 770 | ) { |
| 771 | $this->logger->debug( |
| 772 | 'detectScriptInSvg: Found href attribute <{localPart} {attrib}="{value}"> ' . |
| 773 | 'in uploaded file.', |
| 774 | $attribLogContext |
| 775 | ); |
| 776 | return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ]; |
| 777 | } |
| 778 | |
| 779 | // Only allow 'data:\' targets that should be safe. |
| 780 | // This prevents vectors like image/svg, text/xml, application/xml, and text/html, which can contain scripts |
| 781 | if ( $stripped === 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) { |
| 782 | // RFC2397 parameters. |
| 783 | // This is only slightly slower than (;[\w;]+)*. |
| 784 | // phpcs:ignore Generic.Files.LineLength |
| 785 | $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?'; |
| 786 | |
| 787 | if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|a?png|webp|avif)$parameters,!i", $value ) ) { |
| 788 | $this->logger->debug( |
| 789 | 'detectScriptInSvg: Found href with data URI with MIME type that is not ' . |
| 790 | 'allowed: <{localPart} {attrib}="{value}"> in uploaded file.', |
| 791 | $attribLogContext |
| 792 | ); |
| 793 | return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ]; |
| 794 | } |
| 795 | } |
| 796 | |
| 797 | // Change href with animate from (http://html5sec.org/#137). |
| 798 | if ( $stripped === 'attributename' |
| 799 | && $strippedElement === 'animate' |
| 800 | && $this->stripXmlNamespace( $value ) === 'href' |
| 801 | ) { |
| 802 | $this->logger->debug( |
| 803 | 'detectScriptInSvg: Found animate that might be changing href using "from": ' . |
| 804 | '<{localPart} {attrib}="{value}"> in uploaded file.', |
| 805 | $attribLogContext |
| 806 | ); |
| 807 | return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ]; |
| 808 | } |
| 809 | |
| 810 | // Use set/animate to add event-handler attribute to parent. |
| 811 | if ( ( $strippedElement === 'set' || $strippedElement === 'animate' ) |
| 812 | && $stripped === 'attributename' |
| 813 | && str_starts_with( $value, 'on' ) |
| 814 | ) { |
| 815 | $this->logger->debug( |
| 816 | 'detectScriptInSvg: Found SVG setting event-handler attribute with ' . |
| 817 | '<{localPart} {attrib}="{value}"> in uploaded file.', |
| 818 | $attribLogContext |
| 819 | ); |
| 820 | return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ]; |
| 821 | } |
| 822 | |
| 823 | // use set to add href attribute to parent element. |
| 824 | if ( $strippedElement === 'set' |
| 825 | && $stripped === 'attributename' |
| 826 | && str_contains( $value, 'href' ) |
| 827 | ) { |
| 828 | $this->logger->debug( |
| 829 | 'detectScriptInSvg: Found SVG setting href attribute "{value}" in uploaded file.', |
| 830 | $attribLogContext |
| 831 | ); |
| 832 | return [ 'uploaded-setting-href-svg' ]; |
| 833 | } |
| 834 | |
| 835 | // use set to add a remote / data / script target to an element. |
| 836 | if ( $strippedElement === 'set' |
| 837 | && $stripped === 'to' |
| 838 | && preg_match( '!(http|https|data|script):!im', $value ) |
| 839 | ) { |
| 840 | $this->logger->debug( |
| 841 | 'detectScriptInSvg: Found SVG setting attribute to "{value}" in uploaded file.', |
| 842 | $attribLogContext |
| 843 | ); |
| 844 | return [ 'uploaded-wrong-setting-svg', $value ]; |
| 845 | } |
| 846 | |
| 847 | // use handler attribute with remote / data / script. |
| 848 | if ( $stripped === 'handler' && preg_match( '!(http|https|data|script):!im', $value ) ) { |
| 849 | $this->logger->debug( |
| 850 | 'detectScriptInSvg: Found SVG setting handler with remote/data/script ' . |
| 851 | '{attrib}="value" in uploaded file.', |
| 852 | $attribLogContext |
| 853 | ); |
| 854 | return [ 'uploaded-setting-handler-svg', $attrib, $value ]; |
| 855 | } |
| 856 | |
| 857 | // use CSS styles to bring in remote code. |
| 858 | if ( $stripped === 'style' |
| 859 | && $this->svgCssChecker->checkStyleAttribute( $value ) !== true |
| 860 | ) { |
| 861 | $this->logger->debug( |
| 862 | 'detectScriptInSvg: Found SVG setting a style with remote url ' . |
| 863 | '{attrib}="{value}" in uploaded file.', |
| 864 | $attribLogContext |
| 865 | ); |
| 866 | return [ 'uploaded-remote-url-svg', $attrib, $value ]; |
| 867 | } |
| 868 | |
| 869 | // Several attributes can include css, css character escaping isn't allowed. |
| 870 | if ( in_array( $stripped, $cssAttrs, true ) |
| 871 | && $this->svgCssChecker->checkPresentationalAttribute( $value ) !== true |
| 872 | ) { |
| 873 | $this->logger->debug( |
| 874 | 'detectScriptInSvg: Found SVG setting a style with ' . |
| 875 | '{attrib}="{value}" in uploaded file.', |
| 876 | $attribLogContext |
| 877 | ); |
| 878 | return [ 'uploaded-remote-url-svg', $attrib, $value ]; |
| 879 | } |
| 880 | |
| 881 | // image filters can pull in url, which could be svg that executes scripts. |
| 882 | // Only allow url( "#foo" ). |
| 883 | // Do not allow url( http://example.com ) |
| 884 | // TODO: It seems like the line above already does this check. |
| 885 | if ( $strippedElement === 'image' |
| 886 | && $stripped === 'filter' |
| 887 | && preg_match( '!url\s*\(\s*["\']?[^#]!im', $value ) |
| 888 | ) { |
| 889 | $this->logger->debug( |
| 890 | 'detectScriptInSvg: Found image filter with URL: ' . |
| 891 | '<{localPart} {attrib}="{value}"> in uploaded file.', |
| 892 | $attribLogContext |
| 893 | ); |
| 894 | return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ]; |
| 895 | } |
| 896 | } |
| 897 | |
| 898 | return false; // No scripts detected |
| 899 | } |
| 900 | |
| 901 | /** |
| 902 | * Divide the element name passed by the XML parser to the callback into URI and prefix. |
| 903 | * |
| 904 | * @param string $element |
| 905 | * @return array Containing the namespace URI and prefix |
| 906 | */ |
| 907 | private function splitXmlNamespace( $element ) { |
| 908 | // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ] |
| 909 | $parts = explode( ':', strtolower( $element ) ); |
| 910 | $name = array_pop( $parts ); |
| 911 | $ns = implode( ':', $parts ); |
| 912 | |
| 913 | return [ $ns, $name ]; |
| 914 | } |
| 915 | |
| 916 | /** |
| 917 | * @param string $element |
| 918 | * @return string |
| 919 | */ |
| 920 | private function stripXmlNamespace( $element ) { |
| 921 | // 'http://www.w3.org/2000/svg:script' -> 'script' |
| 922 | return self::splitXmlNamespace( $element )[1]; |
| 923 | } |
| 924 | |
| 925 | /** |
| 926 | * Generic wrapper function for a virus scanner program. |
| 927 | * This relies on the $wgAntivirus and $wgAntivirusSetup variables. |
| 928 | * $wgAntivirusRequired may be used to deny upload if the scan fails. |
| 929 | * |
| 930 | * @note In most cases, external callers would call verifyFile() to run |
| 931 | * all tests, instead of just doing a virus scan. |
| 932 | * @param string $file Pathname to the temporary upload file |
| 933 | * @return bool|null|string False if not virus is found, null if the scan fails or is disabled, |
| 934 | * or a string containing feedback from the virus scanner if a virus was found. |
| 935 | * If textual feedback is missing but a virus was found, this function returns true. |
| 936 | */ |
| 937 | public function detectVirus( $file ) { |
| 938 | $mainConfig = $this->config; |
| 939 | $antivirus = $mainConfig->get( MainConfigNames::Antivirus ); |
| 940 | $antivirusSetup = $mainConfig->get( MainConfigNames::AntivirusSetup ); |
| 941 | $antivirusRequired = $mainConfig->get( MainConfigNames::AntivirusRequired ); |
| 942 | if ( !$antivirus ) { |
| 943 | $this->logger->debug( 'detectVirus: virus scanner disabled' ); |
| 944 | return null; |
| 945 | } |
| 946 | |
| 947 | if ( !( $antivirusSetup[$antivirus] ?? false ) ) { |
| 948 | throw new ConfigException( "Unknown virus scanner: $antivirus" ); |
| 949 | } |
| 950 | |
| 951 | # look up scanner configuration |
| 952 | $command = $antivirusSetup[$antivirus]['command']; |
| 953 | $exitCodeMap = $antivirusSetup[$antivirus]['codemap']; |
| 954 | $msgPattern = $antivirusSetup[$antivirus]['messagepattern'] ?? null; |
| 955 | |
| 956 | if ( !str_contains( $command, "%f" ) ) { |
| 957 | # simple pattern: append file to scan |
| 958 | $command .= " " . Shell::escape( $file ); |
| 959 | } else { |
| 960 | # complex pattern: replace "%f" with file to scan |
| 961 | $command = str_replace( "%f", Shell::escape( $file ), $command ); |
| 962 | } |
| 963 | |
| 964 | $this->logger->debug( 'detectVirus: running virus scan: {command}', |
| 965 | [ 'command' => $command ] ); |
| 966 | |
| 967 | # execute virus scanner |
| 968 | $exitCode = false; |
| 969 | |
| 970 | # NOTE: there's a 50-line workaround to make stderr redirection work on windows, too. |
| 971 | # that does not seem to be worth the pain. |
| 972 | # Ask me (Duesentrieb) about it if it's ever needed. |
| 973 | $output = Shell::command()->unsafeCommand( $command )->includeStderr()->execute(); |
| 974 | $exitCode = $output->getExitCode(); |
| 975 | |
| 976 | # map exit code to AV_xxx constants. |
| 977 | $mappedCode = $exitCode; |
| 978 | if ( $exitCodeMap ) { |
| 979 | if ( isset( $exitCodeMap[$exitCode] ) ) { |
| 980 | $mappedCode = $exitCodeMap[$exitCode]; |
| 981 | } elseif ( isset( $exitCodeMap["*"] ) ) { |
| 982 | $mappedCode = $exitCodeMap["*"]; |
| 983 | } |
| 984 | } |
| 985 | |
| 986 | # NB: AV_NO_VIRUS is 0, but AV_SCAN_FAILED is false, |
| 987 | # so we need the strict equalities === and thus can't use a switch here |
| 988 | if ( $mappedCode === AV_SCAN_FAILED ) { |
| 989 | # scan failed (code was mapped to false by $exitCodeMap) |
| 990 | $this->logger->debug( 'detectVirus: failed to scan {file} (code {exitCode}).', |
| 991 | [ 'file' => $file, 'exitCode' => $exitCode ] ); |
| 992 | |
| 993 | $output = $antivirusRequired |
| 994 | ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text() |
| 995 | : null; |
| 996 | } elseif ( $mappedCode === AV_SCAN_ABORTED ) { |
| 997 | # scan failed because filetype is unknown (probably immune) |
| 998 | $this->logger->debug( 'detectVirus: unsupported file type {file} (code {exitCode}).', |
| 999 | [ 'file' => $file, 'exitCode' => $exitCode ] ); |
| 1000 | $output = null; |
| 1001 | } elseif ( $mappedCode === AV_NO_VIRUS ) { |
| 1002 | # no virus found |
| 1003 | $this->logger->debug( 'detectVirus: file passed virus scan.' ); |
| 1004 | $output = false; |
| 1005 | } else { |
| 1006 | $output = trim( $output->getStdout() ); |
| 1007 | |
| 1008 | if ( !$output ) { |
| 1009 | $output = true; # if there's no output, return true |
| 1010 | } elseif ( $msgPattern ) { |
| 1011 | $groups = []; |
| 1012 | if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) { |
| 1013 | $output = $groups[1]; |
| 1014 | } |
| 1015 | } |
| 1016 | |
| 1017 | $this->logger->debug( 'detectVirus: FOUND VIRUS! scanner feedback: {output}', |
| 1018 | [ 'output' => $output ] ); |
| 1019 | } |
| 1020 | |
| 1021 | return $output; |
| 1022 | } |
| 1023 | } |