Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
69.08% |
105 / 152 |
|
50.00% |
9 / 18 |
CRAP | |
0.00% |
0 / 1 |
| UploadFromUrl | |
69.54% |
105 / 151 |
|
50.00% |
9 / 18 |
140.52 | |
0.00% |
0 / 1 |
| isAllowed | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| isEnabled | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| isAllowedHost | |
94.74% |
18 / 19 |
|
0.00% |
0 / 1 |
9.01 | |||
| getCacheKey | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| getCacheKeyFromRequest | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| getAllowedHosts | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| isAllowedUrl | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| getUrl | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| initialize | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| initializeFromRequest | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| isValidRequest | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| getSourceType | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fetchFile | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| canFetchFile | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| canFetchFileFromUrl | |
71.43% |
5 / 7 |
|
0.00% |
0 / 1 |
4.37 | |||
| makeTemporaryFile | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| saveTempFileChunk | |
41.67% |
5 / 12 |
|
0.00% |
0 / 1 |
2.79 | |||
| reallyFetchFile | |
81.63% |
40 / 49 |
|
0.00% |
0 / 1 |
11.75 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Backend for uploading files from a HTTP resource. |
| 4 | * |
| 5 | * @license GPL-2.0-or-later |
| 6 | * @file |
| 7 | * @ingroup Upload |
| 8 | */ |
| 9 | |
| 10 | namespace MediaWiki\Upload; |
| 11 | |
| 12 | use MediaWiki\Context\RequestContext; |
| 13 | use MediaWiki\HookContainer\HookRunner; |
| 14 | use MediaWiki\Http\MWHttpRequest; |
| 15 | use MediaWiki\MainConfigNames; |
| 16 | use MediaWiki\MediaWikiServices; |
| 17 | use MediaWiki\Permissions\Authority; |
| 18 | use MediaWiki\Request\WebRequest; |
| 19 | use MediaWiki\Status\Status; |
| 20 | |
| 21 | /** |
| 22 | * Implements uploading from a HTTP resource. |
| 23 | * |
| 24 | * @ingroup Upload |
| 25 | * @author Bryan Tong Minh |
| 26 | * @author Michael Dale |
| 27 | */ |
| 28 | class UploadFromUrl extends UploadBase { |
| 29 | /** @var string */ |
| 30 | protected $mUrl; |
| 31 | |
| 32 | /** @var resource|null|false */ |
| 33 | protected $mTmpHandle; |
| 34 | |
| 35 | /** @var array<string,bool> */ |
| 36 | protected static $allowedUrls = []; |
| 37 | |
| 38 | /** |
| 39 | * Checks if the user is allowed to use the upload-by-URL feature. If the |
| 40 | * user is not allowed, return the name of the user right as a string. If |
| 41 | * the user is allowed, have the parent do further permissions checking. |
| 42 | * |
| 43 | * @param Authority $performer |
| 44 | * |
| 45 | * @return bool|string |
| 46 | */ |
| 47 | public static function isAllowed( Authority $performer ) { |
| 48 | if ( !$performer->isAllowed( 'upload_by_url' ) ) { |
| 49 | return 'upload_by_url'; |
| 50 | } |
| 51 | |
| 52 | return parent::isAllowed( $performer ); |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * Checks if the upload from URL feature is enabled |
| 57 | * @return bool |
| 58 | */ |
| 59 | public static function isEnabled() { |
| 60 | $allowCopyUploads = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::AllowCopyUploads ); |
| 61 | |
| 62 | return $allowCopyUploads && parent::isEnabled(); |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * Checks whether the URL is for an allowed host |
| 67 | * The domains in the allowlist can include wildcard characters (*) in place |
| 68 | * of any of the domain levels, e.g. '*.flickr.com' or 'upload.*.gov.uk'. |
| 69 | * |
| 70 | * @param string $url |
| 71 | * @return bool |
| 72 | */ |
| 73 | public static function isAllowedHost( $url ) { |
| 74 | $urlUtils = MediaWikiServices::getInstance()->getURLUtils(); |
| 75 | $domains = self::getAllowedHosts(); |
| 76 | if ( !count( $domains ) ) { |
| 77 | return true; |
| 78 | } |
| 79 | $parsedUrl = $urlUtils->parse( $url ); |
| 80 | if ( !$parsedUrl ) { |
| 81 | return false; |
| 82 | } |
| 83 | $valid = false; |
| 84 | foreach ( $domains as $domain ) { |
| 85 | // See if the domain for the upload matches this allowed domain |
| 86 | $domainPieces = explode( '.', $domain ); |
| 87 | $uploadDomainPieces = explode( '.', $parsedUrl['host'] ); |
| 88 | if ( count( $domainPieces ) === count( $uploadDomainPieces ) ) { |
| 89 | $valid = true; |
| 90 | // See if all the pieces match or not (excluding wildcards) |
| 91 | foreach ( $domainPieces as $index => $piece ) { |
| 92 | if ( $piece !== '*' && $piece !== $uploadDomainPieces[$index] ) { |
| 93 | $valid = false; |
| 94 | } |
| 95 | } |
| 96 | if ( $valid ) { |
| 97 | // We found a match, so quit comparing against the list |
| 98 | break; |
| 99 | } |
| 100 | } |
| 101 | /* Non-wildcard test |
| 102 | if ( $parsedUrl['host'] === $domain ) { |
| 103 | $valid = true; |
| 104 | break; |
| 105 | } |
| 106 | */ |
| 107 | } |
| 108 | |
| 109 | return $valid; |
| 110 | } |
| 111 | |
| 112 | /** |
| 113 | * Provides a caching key for an upload from url set of parameters |
| 114 | * Used to set the status of an async job in UploadFromUrlJob |
| 115 | * and retrieve it in frontend clients like ApiUpload. Will return the |
| 116 | * empty string if not all parameters are present. |
| 117 | * |
| 118 | * @param array $params |
| 119 | * @return string |
| 120 | */ |
| 121 | public static function getCacheKey( $params ) { |
| 122 | if ( !isset( $params['filename'] ) || !isset( $params['url'] ) ) { |
| 123 | return ""; |
| 124 | } else { |
| 125 | // We use sha1 here to ensure we have a fixed-length string of printable |
| 126 | // characters. There is no cryptography involved, so we just need a |
| 127 | // relatively fast function. |
| 128 | return sha1( sprintf( "%s|||%s", $params['filename'], $params['url'] ) ); |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | /** |
| 133 | * Get the caching key from a web request |
| 134 | * @param WebRequest &$request |
| 135 | * |
| 136 | * @return string |
| 137 | */ |
| 138 | public static function getCacheKeyFromRequest( &$request ) { |
| 139 | $uploadCacheKey = $request->getText( 'wpCacheKey', $request->getText( 'key', '' ) ); |
| 140 | if ( $uploadCacheKey !== '' ) { |
| 141 | return $uploadCacheKey; |
| 142 | } |
| 143 | $desiredDestName = $request->getText( 'wpDestFile' ); |
| 144 | if ( !$desiredDestName ) { |
| 145 | $desiredDestName = $request->getText( 'wpUploadFileURL' ); |
| 146 | } |
| 147 | return self::getCacheKey( |
| 148 | [ |
| 149 | 'filename' => $desiredDestName, |
| 150 | 'url' => trim( $request->getVal( 'wpUploadFileURL' ) ) |
| 151 | ] |
| 152 | ); |
| 153 | } |
| 154 | |
| 155 | /** |
| 156 | * @since 1.45 public |
| 157 | * @return string[] |
| 158 | */ |
| 159 | public static function getAllowedHosts(): array { |
| 160 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 161 | $domains = $config->get( MainConfigNames::CopyUploadsDomains ); |
| 162 | |
| 163 | if ( $config->get( MainConfigNames::CopyUploadAllowOnWikiDomainConfig ) ) { |
| 164 | $page = wfMessage( 'copyupload-allowed-domains' )->inContentLanguage()->plain(); |
| 165 | |
| 166 | foreach ( explode( "\n", $page ) as $line ) { |
| 167 | // Strip comments |
| 168 | $line = preg_replace( "/^\\s*([^#]*)\\s*((.*)?)$/", "\\1", $line ); |
| 169 | // Trim whitespace |
| 170 | $line = trim( $line ); |
| 171 | |
| 172 | if ( $line !== '' ) { |
| 173 | $domains[] = $line; |
| 174 | } |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | return $domains; |
| 179 | } |
| 180 | |
| 181 | /** |
| 182 | * Checks whether the URL is not allowed. |
| 183 | * |
| 184 | * @param string $url |
| 185 | * @return bool |
| 186 | */ |
| 187 | public static function isAllowedUrl( $url ) { |
| 188 | if ( !isset( self::$allowedUrls[$url] ) ) { |
| 189 | $allowed = true; |
| 190 | ( new HookRunner( MediaWikiServices::getInstance()->getHookContainer() ) ) |
| 191 | ->onIsUploadAllowedFromUrl( $url, $allowed ); |
| 192 | self::$allowedUrls[$url] = $allowed; |
| 193 | } |
| 194 | |
| 195 | return self::$allowedUrls[$url]; |
| 196 | } |
| 197 | |
| 198 | /** |
| 199 | * Get the URL of the file to be uploaded |
| 200 | * @return string |
| 201 | */ |
| 202 | public function getUrl() { |
| 203 | return $this->mUrl; |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * Entry point for API upload |
| 208 | * |
| 209 | * @param string $name |
| 210 | * @param string $url |
| 211 | * @param bool $initTempFile |
| 212 | */ |
| 213 | public function initialize( $name, $url, $initTempFile = true ) { |
| 214 | $this->mUrl = $url; |
| 215 | |
| 216 | $tempPath = $initTempFile ? $this->makeTemporaryFile() : null; |
| 217 | $fileSize = $initTempFile ? 0 : null; |
| 218 | # File size and removeTempFile will be filled in later |
| 219 | $this->initializePathInfo( $name, $tempPath, $fileSize, false ); |
| 220 | } |
| 221 | |
| 222 | /** |
| 223 | * Entry point for SpecialUpload |
| 224 | * @param WebRequest &$request |
| 225 | */ |
| 226 | public function initializeFromRequest( &$request ) { |
| 227 | $desiredDestName = $request->getText( 'wpDestFile' ); |
| 228 | if ( !$desiredDestName ) { |
| 229 | $desiredDestName = $request->getText( 'wpUploadFileURL' ); |
| 230 | } |
| 231 | $this->initialize( |
| 232 | $desiredDestName, |
| 233 | trim( $request->getVal( 'wpUploadFileURL' ) ) |
| 234 | ); |
| 235 | } |
| 236 | |
| 237 | /** |
| 238 | * @param WebRequest $request |
| 239 | * @return bool |
| 240 | */ |
| 241 | public static function isValidRequest( $request ) { |
| 242 | $user = RequestContext::getMain()->getUser(); |
| 243 | |
| 244 | $url = $request->getVal( 'wpUploadFileURL' ); |
| 245 | |
| 246 | return $url |
| 247 | && MediaWikiServices::getInstance() |
| 248 | ->getPermissionManager() |
| 249 | ->userHasRight( $user, 'upload_by_url' ); |
| 250 | } |
| 251 | |
| 252 | /** |
| 253 | * @return string |
| 254 | */ |
| 255 | public function getSourceType() { |
| 256 | return 'url'; |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * Download the file |
| 261 | * |
| 262 | * @param array $httpOptions Array of options for MWHttpRequest. |
| 263 | * This could be used to override the timeout on the http request. |
| 264 | * @return Status |
| 265 | */ |
| 266 | public function fetchFile( $httpOptions = [] ) { |
| 267 | $status = $this->canFetchFile(); |
| 268 | if ( !$status->isGood() ) { |
| 269 | return $status; |
| 270 | } |
| 271 | return $this->reallyFetchFile( $httpOptions ); |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * verify we can actually download the file |
| 276 | * |
| 277 | * @return Status |
| 278 | */ |
| 279 | public function canFetchFile() { |
| 280 | return $this->canFetchFileFromUrl( $this->mUrl ); |
| 281 | } |
| 282 | |
| 283 | private function canFetchFileFromUrl( string $url ): Status { |
| 284 | if ( !MWHttpRequest::isValidURI( $url ) ) { |
| 285 | return Status::newFatal( 'http-invalid-url', $url ); |
| 286 | } |
| 287 | if ( !self::isAllowedHost( $url ) ) { |
| 288 | return Status::newFatal( 'upload-copy-upload-invalid-domain' ); |
| 289 | } |
| 290 | if ( !self::isAllowedUrl( $url ) ) { |
| 291 | return Status::newFatal( 'upload-copy-upload-invalid-url' ); |
| 292 | } |
| 293 | return Status::newGood(); |
| 294 | } |
| 295 | |
| 296 | /** |
| 297 | * Create a new temporary file in the URL subdirectory of wfTempDir(). |
| 298 | * |
| 299 | * @return string Path to the file |
| 300 | */ |
| 301 | protected function makeTemporaryFile() { |
| 302 | $tmpFile = MediaWikiServices::getInstance()->getTempFSFileFactory() |
| 303 | ->newTempFSFile( 'URL', 'urlupload_' ); |
| 304 | $tmpFile->bind( $this ); |
| 305 | |
| 306 | return $tmpFile->getPath(); |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Callback: save a chunk of the result of a HTTP request to the temporary file |
| 311 | * |
| 312 | * @param mixed $req |
| 313 | * @param string $buffer |
| 314 | * @return int Number of bytes handled |
| 315 | */ |
| 316 | public function saveTempFileChunk( $req, $buffer ) { |
| 317 | wfDebugLog( 'fileupload', 'Received chunk of ' . strlen( $buffer ) . ' bytes' ); |
| 318 | $nbytes = fwrite( $this->mTmpHandle, $buffer ); |
| 319 | |
| 320 | if ( $nbytes == strlen( $buffer ) ) { |
| 321 | $this->mFileSize += $nbytes; |
| 322 | } else { |
| 323 | // Well... that's not good! |
| 324 | wfDebugLog( |
| 325 | 'fileupload', |
| 326 | 'Short write ' . $nbytes . '/' . strlen( $buffer ) . |
| 327 | ' bytes, aborting with ' . $this->mFileSize . ' uploaded so far' |
| 328 | ); |
| 329 | fclose( $this->mTmpHandle ); |
| 330 | $this->mTmpHandle = false; |
| 331 | } |
| 332 | |
| 333 | return $nbytes; |
| 334 | } |
| 335 | |
| 336 | /** |
| 337 | * Download the file, save it to the temporary file and update the file |
| 338 | * size and set $mRemoveTempFile to true. |
| 339 | * |
| 340 | * @param array $httpOptions Array of options for MWHttpRequest |
| 341 | * @return Status |
| 342 | */ |
| 343 | protected function reallyFetchFile( $httpOptions = [] ) { |
| 344 | $copyUploadProxy = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::CopyUploadProxy ); |
| 345 | $copyUploadTimeout = MediaWikiServices::getInstance()->getMainConfig() |
| 346 | ->get( MainConfigNames::CopyUploadTimeout ); |
| 347 | |
| 348 | // Note the temporary file should already be created by makeTemporaryFile() |
| 349 | $this->mTmpHandle = fopen( $this->mTempPath, 'wb' ); |
| 350 | if ( !$this->mTmpHandle ) { |
| 351 | return Status::newFatal( 'tmp-create-error' ); |
| 352 | } |
| 353 | wfDebugLog( 'fileupload', 'Temporary file created "' . $this->mTempPath . '"' ); |
| 354 | |
| 355 | $this->mRemoveTempFile = true; |
| 356 | $this->mFileSize = 0; |
| 357 | |
| 358 | $options = $httpOptions + [ 'followRedirects' => false ]; |
| 359 | |
| 360 | if ( $copyUploadProxy !== false ) { |
| 361 | $options['proxy'] = $copyUploadProxy; |
| 362 | } |
| 363 | |
| 364 | if ( $copyUploadTimeout && !isset( $options['timeout'] ) ) { |
| 365 | $options['timeout'] = $copyUploadTimeout; |
| 366 | } |
| 367 | wfDebugLog( |
| 368 | 'fileupload', |
| 369 | 'Starting download from "' . $this->mUrl . '" ' . |
| 370 | '<' . implode( ',', array_keys( array_filter( $options ) ) ) . '>' |
| 371 | ); |
| 372 | |
| 373 | // Manually follow any redirects up to the limit and reset the output file before each new request to prevent |
| 374 | // capturing the redirect response as part of the file. |
| 375 | $attemptsLeft = $options['maxRedirects'] ?? 5; |
| 376 | $targetUrl = $this->mUrl; |
| 377 | $requestFactory = MediaWikiServices::getInstance()->getHttpRequestFactory(); |
| 378 | while ( $attemptsLeft > 0 ) { |
| 379 | $req = $requestFactory->create( $targetUrl, $options, __METHOD__ ); |
| 380 | $req->setCallback( $this->saveTempFileChunk( ... ) ); |
| 381 | $status = $req->execute(); |
| 382 | if ( !$req->isRedirect() ) { |
| 383 | break; |
| 384 | } |
| 385 | $targetUrl = $req->getFinalUrl(); |
| 386 | $redirectedUrlFetchable = $this->canFetchFileFromUrl( $targetUrl ); |
| 387 | if ( !$redirectedUrlFetchable->isGood() ) { |
| 388 | return $redirectedUrlFetchable; |
| 389 | } |
| 390 | // Remove redirect response content from file. |
| 391 | ftruncate( $this->mTmpHandle, 0 ); |
| 392 | rewind( $this->mTmpHandle ); |
| 393 | $attemptsLeft--; |
| 394 | } |
| 395 | |
| 396 | if ( $attemptsLeft == 0 ) { |
| 397 | return Status::newFatal( 'upload-too-many-redirects' ); |
| 398 | } |
| 399 | |
| 400 | if ( $this->mTmpHandle ) { |
| 401 | // File got written ok... |
| 402 | fclose( $this->mTmpHandle ); |
| 403 | $this->mTmpHandle = null; |
| 404 | } else { |
| 405 | // We encountered a write error during the download... |
| 406 | return Status::newFatal( 'tmp-write-error' ); |
| 407 | } |
| 408 | |
| 409 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop |
| 410 | if ( $status->isOK() ) { |
| 411 | wfDebugLog( 'fileupload', 'Download by URL completed successfully.' ); |
| 412 | } else { |
| 413 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop |
| 414 | wfDebugLog( 'fileupload', $status->getWikiText( false, false, 'en' ) ); |
| 415 | wfDebugLog( |
| 416 | 'fileupload', |
| 417 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop |
| 418 | 'Download by URL completed with HTTP status ' . $req->getStatus() |
| 419 | ); |
| 420 | } |
| 421 | |
| 422 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Always set after loop |
| 423 | return $status; |
| 424 | } |
| 425 | } |
| 426 | |
| 427 | /** @deprecated class alias since 1.46 */ |
| 428 | class_alias( UploadFromUrl::class, 'UploadFromUrl' ); |