Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
68.29% |
295 / 432 |
|
44.44% |
12 / 27 |
CRAP | |
0.00% |
0 / 1 |
| MimeAnalyzer | |
68.29% |
295 / 432 |
|
44.44% |
12 / 27 |
1068.18 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
1 | |||
| loadFiles | |
73.08% |
19 / 26 |
|
0.00% |
0 / 1 |
14.81 | |||
| parseMimeTypes | |
89.47% |
17 / 19 |
|
0.00% |
0 / 1 |
7.06 | |||
| parseMimeInfo | |
70.00% |
21 / 30 |
|
0.00% |
0 / 1 |
14.27 | |||
| setLogger | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addExtraTypes | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| addExtraInfo | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getExtensionsForType | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getExtensionsFromMimeType | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| getMimeTypesFromExtension | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getMimeTypeFromExtensionOrNull | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| guessTypesForExtension | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getTypesForExtension | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getExtensionFromMimeTypeOrNull | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isMatchingExtension | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| isPHPImageType | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
2 | |||
| isRecognizableExtension | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| improveTypeFromExtension | |
95.45% |
21 / 22 |
|
0.00% |
0 / 1 |
11 | |||
| guessMimeType | |
53.85% |
7 / 13 |
|
0.00% |
0 / 1 |
5.57 | |||
| doGuessMimeType | |
61.59% |
85 / 138 |
|
0.00% |
0 / 1 |
205.18 | |||
| detectZipTypeFromFile | |
92.00% |
23 / 25 |
|
0.00% |
0 / 1 |
8.03 | |||
| detectMicrosoftBinaryType | |
50.00% |
4 / 8 |
|
0.00% |
0 / 1 |
4.12 | |||
| detectMimeType | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
110 | |||
| getMediaType | |
81.40% |
35 / 43 |
|
0.00% |
0 / 1 |
25.12 | |||
| findMediaType | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
7 | |||
| getMediaTypes | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isValidMajorMimeType | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * This program is free software; you can redistribute it and/or modify |
| 4 | * it under the terms of the GNU General Public License as published by |
| 5 | * the Free Software Foundation; either version 2 of the License, or |
| 6 | * (at your option) any later version. |
| 7 | * |
| 8 | * This program is distributed in the hope that it will be useful, |
| 9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 | * GNU General Public License for more details. |
| 12 | * |
| 13 | * You should have received a copy of the GNU General Public License along |
| 14 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 15 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 16 | * http://www.gnu.org/copyleft/gpl.html |
| 17 | * |
| 18 | * @file |
| 19 | */ |
| 20 | use Psr\Log\LoggerAwareInterface; |
| 21 | use Psr\Log\LoggerInterface; |
| 22 | use Psr\Log\NullLogger; |
| 23 | use Wikimedia\AtEase\AtEase; |
| 24 | use Wikimedia\Mime\MimeMap; |
| 25 | use Wikimedia\Mime\MimeMapMinimal; |
| 26 | |
| 27 | /** |
| 28 | * @defgroup Mime Mime |
| 29 | * |
| 30 | * @ingroup Media |
| 31 | */ |
| 32 | |
| 33 | /** |
| 34 | * Detect MIME types of a file by mapping file extensions or parsing file contents. |
| 35 | * |
| 36 | * @since 1.28 |
| 37 | * @ingroup Mime |
| 38 | */ |
| 39 | class MimeAnalyzer implements LoggerAwareInterface { |
| 40 | /** @var string */ |
| 41 | protected $typeFile; |
| 42 | /** @var string */ |
| 43 | protected $infoFile; |
| 44 | /** @var string */ |
| 45 | protected $xmlTypes; |
| 46 | /** @var callable */ |
| 47 | protected $initCallback; |
| 48 | /** @var callable */ |
| 49 | protected $detectCallback; |
| 50 | /** @var callable */ |
| 51 | protected $guessCallback; |
| 52 | /** @var callable */ |
| 53 | protected $extCallback; |
| 54 | /** @var array Mapping of media types to arrays of MIME types */ |
| 55 | protected $mediaTypes = null; |
| 56 | /** @var array Map of MIME type aliases */ |
| 57 | protected $mimeTypeAliases = null; |
| 58 | /** @var array<string,string[]> Map of MIME types to file extensions */ |
| 59 | protected $mimeToExts = []; |
| 60 | /** @var array<string,string[]> Map of file extensions to MIME types */ |
| 61 | protected $extToMimes = []; |
| 62 | |
| 63 | /** @var array Map of file extensions types to MIME types (as a space separated list) */ |
| 64 | public $mExtToMime = []; // legacy name; field accessed by hooks |
| 65 | |
| 66 | /** @var string Extra MIME types, set for example by media handling extensions */ |
| 67 | private $extraTypes = ''; |
| 68 | /** @var string Extra MIME info, set for example by media handling extensions */ |
| 69 | private $extraInfo = ''; |
| 70 | |
| 71 | /** @var LoggerInterface */ |
| 72 | private $logger; |
| 73 | |
| 74 | /** @var string Use the full, built-in MIME mapping rather than load from a file */ |
| 75 | public const USE_INTERNAL = 'internal'; |
| 76 | |
| 77 | /** |
| 78 | * @param array $params Configuration map, includes: |
| 79 | * - typeFile: path to file with the list of known MIME types |
| 80 | * - infoFile: path to file with the MIME type info |
| 81 | * - xmlTypes: map of root element names to XML MIME types |
| 82 | * - initCallback: initialization callback that is passed this object [optional] |
| 83 | * - detectCallback: alternative to finfo that returns the mime type for a file. |
| 84 | * For example, the callback can return the output of "file -bi". [optional] |
| 85 | * - guessCallback: callback to improve the guessed MIME type using the file data. |
| 86 | * This is intended for fixing mistakes in fileinfo or "detectCallback". [optional] |
| 87 | * - extCallback: callback to improve the guessed MIME type using the extension. [optional] |
| 88 | * - logger: PSR-3 logger [optional] |
| 89 | * @note Constructing these instances is expensive due to file reads. |
| 90 | * A service or singleton pattern should be used to avoid creating instances again and again. |
| 91 | */ |
| 92 | public function __construct( array $params ) { |
| 93 | $this->typeFile = $params['typeFile']; |
| 94 | $this->infoFile = $params['infoFile']; |
| 95 | $this->xmlTypes = $params['xmlTypes']; |
| 96 | $this->initCallback = $params['initCallback'] ?? null; |
| 97 | $this->detectCallback = $params['detectCallback'] ?? null; |
| 98 | $this->guessCallback = $params['guessCallback'] ?? null; |
| 99 | $this->extCallback = $params['extCallback'] ?? null; |
| 100 | $this->logger = $params['logger'] ?? new NullLogger(); |
| 101 | |
| 102 | $this->loadFiles(); |
| 103 | } |
| 104 | |
| 105 | protected function loadFiles(): void { |
| 106 | # Allow media handling extensions adding MIME-types and MIME-info |
| 107 | if ( $this->initCallback ) { |
| 108 | call_user_func( $this->initCallback, $this ); |
| 109 | } |
| 110 | |
| 111 | $rawTypes = $this->extraTypes; |
| 112 | if ( $this->typeFile === self::USE_INTERNAL ) { |
| 113 | $this->mimeToExts = MimeMap::MIME_EXTENSIONS; |
| 114 | } else { |
| 115 | $this->mimeToExts = MimeMapMinimal::MIME_EXTENSIONS; |
| 116 | if ( $this->typeFile ) { |
| 117 | $rawTypes = file_get_contents( $this->typeFile ) . "\n" . $this->extraTypes; |
| 118 | } |
| 119 | } |
| 120 | if ( $rawTypes ) { |
| 121 | $this->parseMimeTypes( $rawTypes ); |
| 122 | } |
| 123 | |
| 124 | // Build the reverse mapping (extension => MIME type). |
| 125 | foreach ( $this->mimeToExts as $mime => $exts ) { |
| 126 | foreach ( $exts as $ext ) { |
| 127 | $this->extToMimes[$ext][] = $mime; |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | // Migrate items from the legacy $this->mExtToMime field. |
| 132 | // TODO: Remove this when mExtToMime is finally removed. |
| 133 | foreach ( $this->mExtToMime as $ext => $mimes ) { |
| 134 | foreach ( explode( ' ', $mimes ) as $mime ) { |
| 135 | $this->extToMimes[$ext][] = $mime; |
| 136 | } |
| 137 | } |
| 138 | |
| 139 | $rawInfo = $this->extraInfo; |
| 140 | if ( $this->infoFile === self::USE_INTERNAL ) { |
| 141 | $this->mimeTypeAliases = MimeMap::MIME_TYPE_ALIASES; |
| 142 | $this->mediaTypes = MimeMap::MEDIA_TYPES; |
| 143 | } else { |
| 144 | $this->mimeTypeAliases = MimeMapMinimal::MIME_TYPE_ALIASES; |
| 145 | $this->mediaTypes = MimeMapMinimal::MEDIA_TYPES; |
| 146 | if ( $this->infoFile ) { |
| 147 | $rawInfo = file_get_contents( $this->infoFile ) . "\n" . $this->extraInfo; |
| 148 | } |
| 149 | } |
| 150 | if ( $rawInfo ) { |
| 151 | $this->parseMimeInfo( $rawInfo ); |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | protected function parseMimeTypes( string $rawMimeTypes ): void { |
| 156 | $rawMimeTypes = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $rawMimeTypes ); |
| 157 | $rawMimeTypes = str_replace( "\t", " ", $rawMimeTypes ); |
| 158 | |
| 159 | $lines = explode( "\n", $rawMimeTypes ); |
| 160 | foreach ( $lines as $s ) { |
| 161 | $s = trim( $s ); |
| 162 | if ( $s === '' || str_starts_with( $s, '#' ) ) { |
| 163 | continue; |
| 164 | } |
| 165 | |
| 166 | $s = strtolower( $s ); |
| 167 | $i = strpos( $s, ' ' ); |
| 168 | |
| 169 | if ( $i === false ) { |
| 170 | continue; |
| 171 | } |
| 172 | |
| 173 | $ext = trim( substr( $s, $i + 1 ) ); |
| 174 | |
| 175 | if ( !$ext ) { |
| 176 | continue; |
| 177 | } |
| 178 | |
| 179 | $tokens = preg_split( '/\s+/', $s, -1, PREG_SPLIT_NO_EMPTY ); |
| 180 | if ( count( $tokens ) > 1 ) { |
| 181 | $mime = array_shift( $tokens ); |
| 182 | $this->mimeToExts[$mime] = array_values( array_unique( |
| 183 | array_merge( $this->mimeToExts[$mime] ?? [], $tokens ) ) ); |
| 184 | } |
| 185 | } |
| 186 | } |
| 187 | |
| 188 | protected function parseMimeInfo( string $rawMimeInfo ): void { |
| 189 | $rawMimeInfo = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $rawMimeInfo ); |
| 190 | $rawMimeInfo = str_replace( "\t", " ", $rawMimeInfo ); |
| 191 | |
| 192 | $lines = explode( "\n", $rawMimeInfo ); |
| 193 | foreach ( $lines as $s ) { |
| 194 | $s = trim( $s ); |
| 195 | if ( $s === '' || str_starts_with( $s, '#' ) ) { |
| 196 | continue; |
| 197 | } |
| 198 | |
| 199 | $s = strtolower( $s ); |
| 200 | $i = strpos( $s, ' ' ); |
| 201 | |
| 202 | if ( $i === false ) { |
| 203 | continue; |
| 204 | } |
| 205 | |
| 206 | # print "processing MIME INFO line $s<br>"; |
| 207 | |
| 208 | $match = []; |
| 209 | if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) { |
| 210 | $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s ); |
| 211 | $mtype = trim( strtoupper( $match[1] ) ); |
| 212 | } else { |
| 213 | $mtype = MEDIATYPE_UNKNOWN; |
| 214 | } |
| 215 | |
| 216 | $m = preg_split( '/\s+/', $s, -1, PREG_SPLIT_NO_EMPTY ); |
| 217 | |
| 218 | if ( !isset( $this->mediaTypes[$mtype] ) ) { |
| 219 | $this->mediaTypes[$mtype] = []; |
| 220 | } |
| 221 | |
| 222 | foreach ( $m as $mime ) { |
| 223 | $mime = trim( $mime ); |
| 224 | if ( !$mime ) { |
| 225 | continue; |
| 226 | } |
| 227 | |
| 228 | $this->mediaTypes[$mtype][] = $mime; |
| 229 | } |
| 230 | |
| 231 | if ( count( $m ) > 1 ) { |
| 232 | $main = $m[0]; |
| 233 | $mCount = count( $m ); |
| 234 | for ( $i = 1; $i < $mCount; $i += 1 ) { |
| 235 | $mime = $m[$i]; |
| 236 | $this->mimeTypeAliases[$mime] = $main; |
| 237 | } |
| 238 | } |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | public function setLogger( LoggerInterface $logger ) { |
| 243 | $this->logger = $logger; |
| 244 | } |
| 245 | |
| 246 | /** |
| 247 | * Adds to the list mapping MIME to file extensions. |
| 248 | * |
| 249 | * As an extension author, you are encouraged to submit patches to |
| 250 | * MediaWiki's core to add new MIME types to MimeMap.php. |
| 251 | * |
| 252 | * @param string $types |
| 253 | */ |
| 254 | public function addExtraTypes( string $types ): void { |
| 255 | $this->extraTypes .= "\n" . $types; |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * Adds to the list mapping MIME to media type. |
| 260 | * |
| 261 | * As an extension author, you are encouraged to submit patches to |
| 262 | * MediaWiki's core to add new MIME info to MimeMap.php. |
| 263 | * |
| 264 | * @param string $info |
| 265 | */ |
| 266 | public function addExtraInfo( string $info ): void { |
| 267 | $this->extraInfo .= "\n" . $info; |
| 268 | } |
| 269 | |
| 270 | /** |
| 271 | * Returns a list of file extensions for a given MIME type as a space |
| 272 | * separated string or null if the MIME type was unrecognized. Resolves |
| 273 | * MIME type aliases. |
| 274 | * |
| 275 | * @deprecated since 1.35 Use getExtensionsFromMimeType() instead. |
| 276 | * @param string $mime |
| 277 | * @return string|null |
| 278 | */ |
| 279 | public function getExtensionsForType( $mime ) { |
| 280 | $exts = $this->getExtensionsFromMimeType( $mime ); |
| 281 | return $exts ? implode( ' ', $exts ) : null; |
| 282 | } |
| 283 | |
| 284 | /** |
| 285 | * Returns an array of file extensions associated with a given MIME type. |
| 286 | * The returned array is empty if the MIME type was unrecognized. Resolves |
| 287 | * MIME type aliases. |
| 288 | * |
| 289 | * @since 1.35 |
| 290 | * @param string $mime |
| 291 | * @return string[] |
| 292 | */ |
| 293 | public function getExtensionsFromMimeType( string $mime ): array { |
| 294 | $mime = strtolower( $mime ); |
| 295 | if ( !isset( $this->mimeToExts[$mime] ) && isset( $this->mimeTypeAliases[$mime] ) ) { |
| 296 | $mime = $this->mimeTypeAliases[$mime]; |
| 297 | } |
| 298 | return $this->mimeToExts[$mime] ?? []; |
| 299 | } |
| 300 | |
| 301 | /** |
| 302 | * Returns an array of MIME types associated with a given file extension. |
| 303 | * The returned array is empty if the file extension is not associated with |
| 304 | * any MIME types. |
| 305 | * |
| 306 | * @since 1.35 |
| 307 | * @param string $ext |
| 308 | * @return string[] |
| 309 | */ |
| 310 | public function getMimeTypesFromExtension( string $ext ): array { |
| 311 | $ext = strtolower( $ext ); |
| 312 | return $this->extToMimes[$ext] ?? []; |
| 313 | } |
| 314 | |
| 315 | /** |
| 316 | * Returns a single MIME type for a given file extension or null if unknown. |
| 317 | * This is always the first type from the list returned by getMimeTypesFromExtension($ext). |
| 318 | * |
| 319 | * @since 1.35 |
| 320 | * @param string $ext |
| 321 | * @return string|null |
| 322 | */ |
| 323 | public function getMimeTypeFromExtensionOrNull( string $ext ): ?string { |
| 324 | $types = $this->getMimeTypesFromExtension( $ext ); |
| 325 | return $types[0] ?? null; |
| 326 | } |
| 327 | |
| 328 | /** |
| 329 | * Returns a single file extension for a given MIME type or null if unknown. |
| 330 | * This is always the first type from the list returned by getExtensionsFromMimeType($mime). |
| 331 | * |
| 332 | * @deprecated since 1.35 Use getMimeTypeFromExtensionOrNull() instead. |
| 333 | * @param string $ext |
| 334 | * @return string|null |
| 335 | */ |
| 336 | public function guessTypesForExtension( $ext ) { |
| 337 | return $this->getMimeTypeFromExtensionOrNull( $ext ); |
| 338 | } |
| 339 | |
| 340 | /** |
| 341 | * Returns a list of MIME types for a given file extension as a space |
| 342 | * separated string or null if the extension was unrecognized. |
| 343 | * |
| 344 | * @deprecated since 1.35 Use getMimeTypesFromExtension() instead. |
| 345 | * @param string $ext |
| 346 | * @return string|null |
| 347 | */ |
| 348 | public function getTypesForExtension( $ext ) { |
| 349 | $types = $this->getMimeTypesFromExtension( $ext ); |
| 350 | return $types ? implode( ' ', $types ) : null; |
| 351 | } |
| 352 | |
| 353 | /** |
| 354 | * Returns a single file extension for a given MIME type or null if unknown. |
| 355 | * This is always the first type from the list returned by getExtensionsFromMimeType($mime). |
| 356 | * |
| 357 | * @since 1.35 |
| 358 | * @param string $mime |
| 359 | * @return string|null |
| 360 | */ |
| 361 | public function getExtensionFromMimeTypeOrNull( string $mime ): ?string { |
| 362 | $exts = $this->getExtensionsFromMimeType( $mime ); |
| 363 | return $exts[0] ?? null; |
| 364 | } |
| 365 | |
| 366 | /** |
| 367 | * Tests if the extension matches the given MIME type. Returns true if a |
| 368 | * match was found, null if the MIME type is unknown, and false if the |
| 369 | * MIME type is known but no matches where found. |
| 370 | * |
| 371 | * @param string $extension |
| 372 | * @param string $mime |
| 373 | * @return bool|null |
| 374 | */ |
| 375 | public function isMatchingExtension( string $extension, string $mime ): ?bool { |
| 376 | $exts = $this->getExtensionsFromMimeType( $mime ); |
| 377 | |
| 378 | if ( !$exts ) { |
| 379 | return null; // Unknown MIME type |
| 380 | } |
| 381 | |
| 382 | return in_array( strtolower( $extension ), $exts ); |
| 383 | } |
| 384 | |
| 385 | /** |
| 386 | * Returns true if the MIME type is known to represent an image format |
| 387 | * supported by the PHP GD library. |
| 388 | * |
| 389 | * @deprecated since 1.40 |
| 390 | * @param string $mime |
| 391 | * @return bool |
| 392 | */ |
| 393 | public function isPHPImageType( string $mime ): bool { |
| 394 | wfDeprecated( __METHOD__, '1.40' ); |
| 395 | // As defined by imagegetsize and image_type_to_mime |
| 396 | static $types = [ |
| 397 | 'image/gif', 'image/jpeg', 'image/png', |
| 398 | 'image/x-bmp', 'image/xbm', 'image/tiff', |
| 399 | 'image/jp2', 'image/jpeg2000', 'image/iff', |
| 400 | 'image/xbm', 'image/x-xbitmap', |
| 401 | 'image/vnd.wap.wbmp', 'image/vnd.xiff', |
| 402 | 'image/x-photoshop', |
| 403 | 'application/x-shockwave-flash', |
| 404 | ]; |
| 405 | |
| 406 | return in_array( $mime, $types ); |
| 407 | } |
| 408 | |
| 409 | /** |
| 410 | * Returns true if the extension represents a type which can |
| 411 | * be reliably detected from its content. Use this to determine |
| 412 | * whether strict content checks should be applied to reject |
| 413 | * invalid uploads; if we can't identify the type we won't |
| 414 | * be able to say if it's invalid. |
| 415 | * |
| 416 | * @todo Be more accurate when using fancy MIME detector plugins; |
| 417 | * right now this is the bare minimum getimagesize() list. |
| 418 | * @param string $extension |
| 419 | * @return bool |
| 420 | */ |
| 421 | public function isRecognizableExtension( string $extension ): bool { |
| 422 | static $types = [ |
| 423 | // Types recognized by getimagesize() |
| 424 | 'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd', |
| 425 | 'bmp', 'tiff', 'tif', 'jpc', 'jp2', |
| 426 | 'jpx', 'jb2', 'swc', 'iff', 'wbmp', |
| 427 | 'xbm', |
| 428 | |
| 429 | // Formats we recognize magic numbers for |
| 430 | 'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx', 'opus', |
| 431 | 'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka', |
| 432 | 'webp', 'mp3', |
| 433 | |
| 434 | // XML formats we sure hope we recognize reliably |
| 435 | 'svg', |
| 436 | |
| 437 | // 3D formats |
| 438 | 'stl', |
| 439 | ]; |
| 440 | return in_array( strtolower( $extension ), $types ); |
| 441 | } |
| 442 | |
| 443 | /** |
| 444 | * Improves a MIME type using the file extension. Some file formats are very generic, |
| 445 | * so their MIME type is not very meaningful. A more useful MIME type can be derived |
| 446 | * by looking at the file extension. Typically, this method would be called on the |
| 447 | * result of guessMimeType(). |
| 448 | * |
| 449 | * XXX: Null-returning behavior is probably an accident and definitely confusing (T253483). |
| 450 | * |
| 451 | * @param string $mime The MIME type, typically guessed from a file's content. |
| 452 | * @param string $ext The file extension, as taken from the file name |
| 453 | * @return string|null The improved MIME type, or null if the MIME type is |
| 454 | * unknown/unknown and the extension is not recognized. |
| 455 | */ |
| 456 | public function improveTypeFromExtension( string $mime, string $ext ): ?string { |
| 457 | if ( $mime === 'unknown/unknown' ) { |
| 458 | if ( $this->isRecognizableExtension( $ext ) ) { |
| 459 | $this->logger->info( __METHOD__ . ': refusing to guess mime type for .' . |
| 460 | "$ext file, we should have recognized it" ); |
| 461 | } else { |
| 462 | // Not something we can detect, so simply |
| 463 | // trust the file extension |
| 464 | $mime = $this->getMimeTypeFromExtensionOrNull( $ext ); |
| 465 | } |
| 466 | } elseif ( $mime === 'application/x-opc+zip' |
| 467 | || $mime === 'application/vnd.oasis.opendocument' |
| 468 | ) { |
| 469 | if ( $this->isMatchingExtension( $ext, $mime ) ) { |
| 470 | // A known file extension for an OPC/ODF file, |
| 471 | // find the proper MIME type for that file extension |
| 472 | $mime = $this->getMimeTypeFromExtensionOrNull( $ext ); |
| 473 | } else { |
| 474 | $this->logger->info( __METHOD__ . |
| 475 | ": refusing to guess better type for $mime file, " . |
| 476 | ".$ext is not a known OPC/ODF extension." ); |
| 477 | $mime = 'application/zip'; |
| 478 | } |
| 479 | } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) { |
| 480 | // Textual types are sometimes not recognized properly. |
| 481 | // If detected as text/plain, and has an extension which is textual |
| 482 | // improve to the extension's type. For example, csv and json are often |
| 483 | // misdetected as text/plain. |
| 484 | $mime = $this->getMimeTypeFromExtensionOrNull( $ext ); |
| 485 | } |
| 486 | |
| 487 | # Media handling extensions can improve the MIME detected |
| 488 | $callback = $this->extCallback; |
| 489 | if ( $callback ) { |
| 490 | $callback( $this, $ext, $mime /* by reference */ ); |
| 491 | } |
| 492 | |
| 493 | if ( $mime !== null && isset( $this->mimeTypeAliases[$mime] ) ) { |
| 494 | $mime = $this->mimeTypeAliases[$mime]; |
| 495 | } |
| 496 | |
| 497 | $this->logger->info( __METHOD__ . ": improved mime type for .$ext: $mime" ); |
| 498 | return $mime; |
| 499 | } |
| 500 | |
| 501 | /** |
| 502 | * MIME type detection. This uses detectMimeType to detect the MIME type |
| 503 | * of the file, but applies additional checks to determine some well known |
| 504 | * file formats that may be missed or misinterpreted by the default MIME |
| 505 | * detection (namely XML based formats like XHTML or SVG, as well as ZIP |
| 506 | * based formats like OPC/ODF files). |
| 507 | * |
| 508 | * @param string $file The file to check |
| 509 | * @param string|bool $ext The file extension, or true (default) to extract |
| 510 | * it from the filename. Set it to false to ignore the extension. DEPRECATED! |
| 511 | * Set to false, use improveTypeFromExtension($mime, $ext) later to improve MIME type. |
| 512 | * @return string The MIME type of $file |
| 513 | */ |
| 514 | public function guessMimeType( string $file, $ext = true ): string { |
| 515 | if ( $ext ) { // TODO: make $ext default to false. Or better, remove it. |
| 516 | $this->logger->info( __METHOD__ . |
| 517 | ": WARNING: use of the \$ext parameter is deprecated. " . |
| 518 | "Use improveTypeFromExtension(\$mime, \$ext) instead." ); |
| 519 | } |
| 520 | |
| 521 | $mime = $this->doGuessMimeType( $file ); |
| 522 | |
| 523 | if ( !$mime ) { |
| 524 | $this->logger->info( __METHOD__ . |
| 525 | ": internal type detection failed for $file (.$ext)..." ); |
| 526 | $mime = $this->detectMimeType( $file, $ext ); |
| 527 | } |
| 528 | |
| 529 | if ( isset( $this->mimeTypeAliases[$mime] ) ) { |
| 530 | $mime = $this->mimeTypeAliases[$mime]; |
| 531 | } |
| 532 | |
| 533 | $this->logger->info( __METHOD__ . ": guessed mime type of $file: $mime" ); |
| 534 | return $mime; |
| 535 | } |
| 536 | |
| 537 | /** |
| 538 | * Guess the MIME type from the file contents. |
| 539 | * |
| 540 | * @param string $file |
| 541 | * @return bool|string |
| 542 | * @throws UnexpectedValueException |
| 543 | */ |
| 544 | private function doGuessMimeType( string $file ) { |
| 545 | // Read a chunk of the file |
| 546 | AtEase::suppressWarnings(); |
| 547 | $f = fopen( $file, 'rb' ); |
| 548 | AtEase::restoreWarnings(); |
| 549 | |
| 550 | if ( !$f ) { |
| 551 | return 'unknown/unknown'; |
| 552 | } |
| 553 | |
| 554 | $fsize = filesize( $file ); |
| 555 | if ( $fsize === false ) { |
| 556 | return 'unknown/unknown'; |
| 557 | } |
| 558 | |
| 559 | $head = fread( $f, 1024 ); |
| 560 | $head16k = $head . fread( $f, 16384 - 1024 ); // some WebM files have big headers |
| 561 | $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR |
| 562 | if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) { |
| 563 | throw new UnexpectedValueException( |
| 564 | "Seeking $tailLength bytes from EOF failed in " . __METHOD__ ); |
| 565 | } |
| 566 | $tail = $tailLength ? fread( $f, $tailLength ) : ''; |
| 567 | |
| 568 | $this->logger->info( __METHOD__ . |
| 569 | ": analyzing head and tail of $file for magic numbers." ); |
| 570 | |
| 571 | // Hardcode a few magic number checks... |
| 572 | $headers = [ |
| 573 | // Multimedia... |
| 574 | 'MThd' => 'audio/midi', |
| 575 | 'OggS' => 'application/ogg', |
| 576 | 'ID3' => 'audio/mpeg', |
| 577 | "\xff\xfb" => 'audio/mpeg', // MPEG-1 layer 3 |
| 578 | "\xff\xf3" => 'audio/mpeg', // MPEG-2 layer 3 (lower sample rates) |
| 579 | "\xff\xe3" => 'audio/mpeg', // MPEG-2.5 layer 3 (very low sample rates) |
| 580 | |
| 581 | // Image formats... |
| 582 | // Note that WMF may have a bare header, no magic number. |
| 583 | "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives? |
| 584 | "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile', |
| 585 | '%PDF' => 'application/pdf', |
| 586 | 'gimp xcf' => 'image/x-xcf', |
| 587 | |
| 588 | // Some forbidden fruit... |
| 589 | 'MZ' => 'application/octet-stream', // DOS/Windows executable |
| 590 | "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary |
| 591 | "\x7fELF" => 'application/octet-stream', // ELF binary |
| 592 | ]; |
| 593 | |
| 594 | foreach ( $headers as $magic => $candidate ) { |
| 595 | if ( str_starts_with( $head, $magic ) ) { |
| 596 | $this->logger->info( __METHOD__ . |
| 597 | ": magic header in $file recognized as $candidate" ); |
| 598 | return $candidate; |
| 599 | } |
| 600 | } |
| 601 | |
| 602 | /* Look for WebM and Matroska files */ |
| 603 | if ( str_starts_with( $head16k, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ) ) ) { |
| 604 | $doctype = strpos( $head16k, "\x42\x82" ); |
| 605 | if ( $doctype ) { |
| 606 | // Next byte is datasize, then data (sizes larger than 1 byte are stupid muxers) |
| 607 | $data = substr( $head16k, $doctype + 3, 8 ); |
| 608 | if ( str_starts_with( $data, "matroska" ) ) { |
| 609 | $this->logger->info( __METHOD__ . ": recognized file as video/x-matroska" ); |
| 610 | return "video/x-matroska"; |
| 611 | } |
| 612 | |
| 613 | if ( str_starts_with( $data, "webm" ) ) { |
| 614 | // XXX HACK look for a video track, if we don't find it, this is an audio file |
| 615 | // This detection is very naive and doesn't parse the actual fields |
| 616 | // 0x86 byte indicates start of codecname field |
| 617 | // next byte is a variable length integer (vint) for the size of the value following it |
| 618 | // 8 (first bit is 1) indicates the smallest size vint, a single byte |
| 619 | // (unlikely we see larger vints here) |
| 620 | // 5 indicates a length of 5 ( V_VP8 or V_VP9 or V_AV1 ) |
| 621 | // Sometimes we see 0x86 instead of 0x85 because a |
| 622 | // non-conforming muxer wrote a null terminated string |
| 623 | $videotrack = str_contains( $head16k, "\x86\x85V_VP8" ) || |
| 624 | str_contains( $head16k, "\x86\x85V_VP9" ) || |
| 625 | str_contains( $head16k, "\x86\x85V_AV1" ) || |
| 626 | str_contains( $head16k, "\x86\x86V_VP8\x0" ) || |
| 627 | str_contains( $head16k, "\x86\x86V_VP9\x0" ) || |
| 628 | str_contains( $head16k, "\x86\x86V_AV1\x0" ); |
| 629 | |
| 630 | if ( $videotrack ) { |
| 631 | // There is a video track, so this is a video file. |
| 632 | $this->logger->info( __METHOD__ . ": recognized file as video/webm" ); |
| 633 | return "video/webm"; |
| 634 | } |
| 635 | |
| 636 | $this->logger->info( __METHOD__ . ": recognized file as audio/webm" ); |
| 637 | return "audio/webm"; |
| 638 | } |
| 639 | } |
| 640 | $this->logger->info( __METHOD__ . ": unknown EBML file" ); |
| 641 | return "unknown/unknown"; |
| 642 | } |
| 643 | |
| 644 | /* Look for WebP */ |
| 645 | if ( str_starts_with( $head, "RIFF" ) && substr( $head, 8, 7 ) === "WEBPVP8" ) { |
| 646 | $this->logger->info( __METHOD__ . ": recognized file as image/webp" ); |
| 647 | return "image/webp"; |
| 648 | } |
| 649 | |
| 650 | /* Look for JPEG2000 */ |
| 651 | if ( str_starts_with( $head, "\x00\x00\x00\x0cjP\x20\x20\x0d\x0a\x87\x0a" ) ) { |
| 652 | $this->logger->info( __METHOD__ . ": recognized as JPEG2000" ); |
| 653 | // we skip 4 bytes |
| 654 | if ( substr( $head, 16, 8 ) === "ftypjp2 " ) { |
| 655 | $this->logger->info( __METHOD__ . ": recognized file as image/jp2" ); |
| 656 | return 'image/jp2'; |
| 657 | } elseif ( substr( $head, 16, 8 ) === "ftypjpx " ) { |
| 658 | $this->logger->info( __METHOD__ . ": recognized file as image/jpx" ); |
| 659 | return 'image/jpx'; |
| 660 | } |
| 661 | } |
| 662 | |
| 663 | /* Look for MS Compound Binary (OLE) files */ |
| 664 | if ( str_starts_with( $head, "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" ) ) { |
| 665 | $this->logger->info( __METHOD__ . ': recognized MS CFB (OLE) file' ); |
| 666 | return $this->detectMicrosoftBinaryType( $f ); |
| 667 | } |
| 668 | |
| 669 | /** |
| 670 | * Look for PHP. Check for this before HTML/XML... Warning: this is a |
| 671 | * heuristic, and won't match a file with a lot of non-PHP before. It |
| 672 | * will also match text files which could be PHP. :) |
| 673 | * |
| 674 | * @todo FIXME: For this reason, the check is probably useless -- an attacker |
| 675 | * could almost certainly just pad the file with a lot of nonsense to |
| 676 | * circumvent the check in any case where it would be a security |
| 677 | * problem. On the other hand, it causes harmful false positives (bug |
| 678 | * 16583). The heuristic has been cut down to exclude three-character |
| 679 | * strings like "<? ", but should it be axed completely? |
| 680 | */ |
| 681 | if ( ( strpos( $head, '<?php' ) !== false ) || |
| 682 | ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) || |
| 683 | ( strpos( $head, "<\x00?\x00 " ) !== false ) || |
| 684 | ( strpos( $head, "<\x00?\x00\n" ) !== false ) || |
| 685 | ( strpos( $head, "<\x00?\x00\t" ) !== false ) || |
| 686 | ( strpos( $head, "<\x00?\x00=" ) !== false ) |
| 687 | ) { |
| 688 | $this->logger->info( __METHOD__ . ": recognized $file as application/x-php" ); |
| 689 | return 'application/x-php'; |
| 690 | } |
| 691 | |
| 692 | /** |
| 693 | * look for XML formats (XHTML and SVG) |
| 694 | */ |
| 695 | AtEase::suppressWarnings(); |
| 696 | $xml = new XmlTypeCheck( $file ); |
| 697 | AtEase::restoreWarnings(); |
| 698 | if ( $xml->wellFormed ) { |
| 699 | $xmlTypes = $this->xmlTypes; |
| 700 | // @phan-suppress-next-line PhanTypeMismatchDimFetch False positive |
| 701 | return $xmlTypes[$xml->getRootElement()] ?? 'application/xml'; |
| 702 | } |
| 703 | |
| 704 | /** |
| 705 | * look for shell scripts |
| 706 | */ |
| 707 | $script_type = null; |
| 708 | |
| 709 | # detect by shebang |
| 710 | if ( str_starts_with( $head, "#!" ) ) { |
| 711 | $script_type = "ASCII"; |
| 712 | } elseif ( str_starts_with( $head, "\xef\xbb\xbf#!" ) ) { |
| 713 | $script_type = "UTF-8"; |
| 714 | } elseif ( str_starts_with( $head, "\xfe\xff\x00#\x00!" ) ) { |
| 715 | $script_type = "UTF-16BE"; |
| 716 | } elseif ( str_starts_with( $head, "\xff\xfe#\x00!" ) ) { |
| 717 | $script_type = "UTF-16LE"; |
| 718 | } |
| 719 | |
| 720 | if ( $script_type ) { |
| 721 | if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) { |
| 722 | // Quick and dirty fold down to ASCII! |
| 723 | $pack = [ 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' ]; |
| 724 | $chars = unpack( $pack[$script_type], substr( $head, 2 ) ); |
| 725 | $head = ''; |
| 726 | foreach ( $chars as $codepoint ) { |
| 727 | if ( $codepoint < 128 ) { |
| 728 | $head .= chr( $codepoint ); |
| 729 | } else { |
| 730 | $head .= '?'; |
| 731 | } |
| 732 | } |
| 733 | } |
| 734 | |
| 735 | $match = []; |
| 736 | |
| 737 | if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) { |
| 738 | $mime = "application/x-{$match[2]}"; |
| 739 | $this->logger->info( __METHOD__ . ": shell script recognized as $mime" ); |
| 740 | return $mime; |
| 741 | } |
| 742 | } |
| 743 | |
| 744 | // Check for ZIP variants (before getimagesize) |
| 745 | $eocdrPos = strpos( $tail, "PK\x05\x06" ); |
| 746 | if ( $eocdrPos !== false && $eocdrPos <= strlen( $tail ) - 22 ) { |
| 747 | $this->logger->info( __METHOD__ . ": ZIP signature present in $file" ); |
| 748 | // Check if it really is a ZIP file, make sure the EOCDR is at the end (T40432) |
| 749 | $commentLength = unpack( "n", substr( $tail, $eocdrPos + 20 ) )[1]; |
| 750 | if ( $eocdrPos + 22 + $commentLength !== strlen( $tail ) ) { |
| 751 | $this->logger->info( __METHOD__ . ": ZIP EOCDR not at end. Not a ZIP file." ); |
| 752 | } else { |
| 753 | return $this->detectZipTypeFromFile( $f ); |
| 754 | } |
| 755 | } |
| 756 | |
| 757 | // Check for STL (3D) files |
| 758 | // @see https://en.wikipedia.org/wiki/STL_(file_format) |
| 759 | if ( $fsize >= 15 && |
| 760 | stripos( $head, 'SOLID ' ) === 0 && |
| 761 | preg_match( '/\RENDSOLID .*$/i', $tail ) ) { |
| 762 | // ASCII STL file |
| 763 | return 'application/sla'; |
| 764 | } elseif ( $fsize > 84 ) { |
| 765 | // binary STL file |
| 766 | $triangles = substr( $head, 80, 4 ); |
| 767 | $triangles = unpack( 'V', $triangles ); |
| 768 | $triangles = reset( $triangles ); |
| 769 | if ( $triangles !== false && $fsize === 84 + ( $triangles * 50 ) ) { |
| 770 | return 'application/sla'; |
| 771 | } |
| 772 | } |
| 773 | |
| 774 | AtEase::suppressWarnings(); |
| 775 | $gis = getimagesize( $file ); |
| 776 | AtEase::restoreWarnings(); |
| 777 | |
| 778 | if ( $gis && isset( $gis['mime'] ) ) { |
| 779 | $mime = $gis['mime']; |
| 780 | $this->logger->info( __METHOD__ . ": getimagesize detected $file as $mime" ); |
| 781 | return $mime; |
| 782 | } |
| 783 | |
| 784 | # Media handling extensions can guess the MIME by content |
| 785 | # It's intentionally here so that if core is wrong about a type (false positive), |
| 786 | # people will hopefully nag and submit patches :) |
| 787 | $mime = false; |
| 788 | # Some strings by reference for performance - assuming well-behaved hooks |
| 789 | $callback = $this->guessCallback; |
| 790 | if ( $callback ) { |
| 791 | $callback( $this, $head, $tail, $file, $mime /* by reference */ ); |
| 792 | } |
| 793 | |
| 794 | return $mime; |
| 795 | } |
| 796 | |
| 797 | /** |
| 798 | * Detect application-specific file type of a given ZIP file. |
| 799 | * If it can't tell, return 'application/zip'. |
| 800 | * |
| 801 | * @internal |
| 802 | * @param resource $handle |
| 803 | * @return string |
| 804 | */ |
| 805 | public function detectZipTypeFromFile( $handle ) { |
| 806 | $types = []; |
| 807 | $status = ZipDirectoryReader::readHandle( |
| 808 | $handle, |
| 809 | static function ( $entry ) use ( &$types ) { |
| 810 | $name = $entry['name']; |
| 811 | $names = [ $name ]; |
| 812 | |
| 813 | // If there is a null character, cut off the name at it, because JDK's |
| 814 | // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name |
| 815 | // were constructed which had ".class\0" followed by a string chosen to |
| 816 | // make the hash collide with the truncated name, that file could be |
| 817 | // returned in response to a request for the .class file. |
| 818 | $nullPos = strpos( $entry['name'], "\000" ); |
| 819 | if ( $nullPos !== false ) { |
| 820 | $names[] = substr( $entry['name'], 0, $nullPos ); |
| 821 | } |
| 822 | |
| 823 | // If there is a trailing slash in the file name, we have to strip it, |
| 824 | // because that's what ZIP_GetEntry() does. |
| 825 | if ( preg_grep( '!\.class/?$!', $names ) ) { |
| 826 | $types[] = 'application/java'; |
| 827 | } |
| 828 | |
| 829 | if ( $name === '[Content_Types].xml' ) { |
| 830 | $types[] = 'application/x-opc+zip'; |
| 831 | } elseif ( $name === 'mimetype' ) { |
| 832 | $types[] = 'application/vnd.oasis.opendocument'; |
| 833 | } |
| 834 | } |
| 835 | ); |
| 836 | if ( !$status->isOK() ) { |
| 837 | $this->logger->info( "Error reading zip file: " . (string)$status ); |
| 838 | // This could be unknown/unknown but we have some weird phpunit test cases |
| 839 | return 'application/zip'; |
| 840 | } |
| 841 | if ( in_array( 'application/java', $types ) ) { |
| 842 | // For security, java detection takes precedence |
| 843 | return 'application/java'; |
| 844 | } elseif ( count( $types ) ) { |
| 845 | return $types[0]; |
| 846 | } else { |
| 847 | return 'application/zip'; |
| 848 | } |
| 849 | } |
| 850 | |
| 851 | /** |
| 852 | * Detect the type of a Microsoft Compound Binary a.k.a. OLE file. |
| 853 | * These are old style pre-ODF files such as .doc and .xls |
| 854 | * |
| 855 | * @param resource $handle An opened seekable file handle |
| 856 | * @return string The detected MIME type |
| 857 | */ |
| 858 | private function detectMicrosoftBinaryType( $handle ): string { |
| 859 | $info = MSCompoundFileReader::readHandle( $handle ); |
| 860 | if ( !$info['valid'] ) { |
| 861 | $this->logger->info( __METHOD__ . ': invalid file format' ); |
| 862 | return 'unknown/unknown'; |
| 863 | } |
| 864 | if ( !$info['mime'] ) { |
| 865 | $this->logger->info( __METHOD__ . ": unrecognised document subtype" ); |
| 866 | return 'unknown/unknown'; |
| 867 | } |
| 868 | return $info['mime']; |
| 869 | } |
| 870 | |
| 871 | /** |
| 872 | * Internal MIME type detection. Detection is done using the fileinfo |
| 873 | * extension if it is available. It can be overridden by callback, which could |
| 874 | * use an external program, for example. If detection fails and $ext is not false, |
| 875 | * the MIME type is guessed from the file extension, using getMimeTypeFromExtensionOrNull. |
| 876 | * |
| 877 | * If the MIME type is still unknown, getimagesize is used to detect the |
| 878 | * MIME type if the file is an image. If no MIME type can be determined, |
| 879 | * this function returns 'unknown/unknown'. |
| 880 | * |
| 881 | * @param string $file The file to check |
| 882 | * @param string|bool $ext The file extension, or true (default) to extract it from the filename. |
| 883 | * Set it to false to ignore the extension. DEPRECATED! Set to false, use |
| 884 | * improveTypeFromExtension($mime, $ext) later to improve MIME type. |
| 885 | * @return string The MIME type of $file |
| 886 | */ |
| 887 | private function detectMimeType( string $file, $ext = true ): string { |
| 888 | /** @todo Make $ext default to false. Or better, remove it. */ |
| 889 | if ( $ext ) { |
| 890 | $this->logger->info( __METHOD__ . |
| 891 | ": WARNING: use of the \$ext parameter is deprecated. " |
| 892 | . "Use improveTypeFromExtension(\$mime, \$ext) instead." ); |
| 893 | } |
| 894 | |
| 895 | $callback = $this->detectCallback; |
| 896 | if ( $callback ) { |
| 897 | $m = $callback( $file ); |
| 898 | } else { |
| 899 | $m = mime_content_type( $file ); |
| 900 | } |
| 901 | |
| 902 | if ( $m ) { |
| 903 | # normalize |
| 904 | $m = preg_replace( '![;, ].*$!', '', $m ); # strip charset, etc |
| 905 | $m = trim( $m ); |
| 906 | $m = strtolower( $m ); |
| 907 | |
| 908 | if ( !str_contains( $m, 'unknown' ) ) { |
| 909 | $this->logger->info( __METHOD__ . ": magic mime type of $file: $m" ); |
| 910 | return $m; |
| 911 | } |
| 912 | } |
| 913 | |
| 914 | // If desired, look at extension as a fallback. |
| 915 | if ( $ext === true ) { |
| 916 | $i = strrpos( $file, '.' ); |
| 917 | $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' ); |
| 918 | } |
| 919 | if ( $ext ) { |
| 920 | if ( $this->isRecognizableExtension( $ext ) ) { |
| 921 | $this->logger->info( __METHOD__ . ": refusing to guess mime type for .$ext file, " |
| 922 | . "we should have recognized it" ); |
| 923 | } else { |
| 924 | $m = $this->getMimeTypeFromExtensionOrNull( $ext ); |
| 925 | if ( $m ) { |
| 926 | $this->logger->info( __METHOD__ . ": extension mime type of $file: $m" ); |
| 927 | return $m; |
| 928 | } |
| 929 | } |
| 930 | } |
| 931 | |
| 932 | // Unknown type |
| 933 | $this->logger->info( __METHOD__ . ": failed to guess mime type for $file!" ); |
| 934 | return 'unknown/unknown'; |
| 935 | } |
| 936 | |
| 937 | /** |
| 938 | * Determine the media type code for a file, using its MIME type, name and |
| 939 | * possibly its contents. |
| 940 | * |
| 941 | * This function relies on the findMediaType(), mapping extensions and MIME |
| 942 | * types to media types. |
| 943 | * |
| 944 | * @todo analyse file if need be |
| 945 | * @todo look at multiple extension, separately and together. |
| 946 | * |
| 947 | * @param string|null $path Full path to the image file, in case we have to look at the contents |
| 948 | * (if null, only the MIME type is used to determine the media type code). |
| 949 | * @param string|null $mime MIME type. If null it will be guessed using guessMimeType. |
| 950 | * @return string A value to be used with the MEDIATYPE_xxx constants. |
| 951 | */ |
| 952 | public function getMediaType( string $path = null, string $mime = null ): string { |
| 953 | if ( !$mime && !$path ) { |
| 954 | return MEDIATYPE_UNKNOWN; |
| 955 | } |
| 956 | |
| 957 | // If MIME type is unknown, guess it |
| 958 | if ( !$mime ) { |
| 959 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullable False positive |
| 960 | $mime = $this->guessMimeType( $path, false ); |
| 961 | } |
| 962 | |
| 963 | // Special code for ogg - detect if it's video (theora), |
| 964 | // else label it as sound. |
| 965 | if ( $mime == 'application/ogg' && is_string( $path ) && file_exists( $path ) ) { |
| 966 | // Read a chunk of the file |
| 967 | $f = fopen( $path, "rt" ); |
| 968 | if ( !$f ) { |
| 969 | return MEDIATYPE_UNKNOWN; |
| 970 | } |
| 971 | $head = fread( $f, 256 ); |
| 972 | fclose( $f ); |
| 973 | |
| 974 | $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) ); |
| 975 | |
| 976 | // This is an UGLY HACK, file should be parsed correctly |
| 977 | if ( strpos( $head, 'theora' ) !== false ) { |
| 978 | return MEDIATYPE_VIDEO; |
| 979 | } elseif ( strpos( $head, 'vorbis' ) !== false ) { |
| 980 | return MEDIATYPE_AUDIO; |
| 981 | } elseif ( strpos( $head, 'flac' ) !== false ) { |
| 982 | return MEDIATYPE_AUDIO; |
| 983 | } elseif ( strpos( $head, 'speex' ) !== false ) { |
| 984 | return MEDIATYPE_AUDIO; |
| 985 | } elseif ( strpos( $head, 'opus' ) !== false ) { |
| 986 | return MEDIATYPE_AUDIO; |
| 987 | } else { |
| 988 | return MEDIATYPE_MULTIMEDIA; |
| 989 | } |
| 990 | } |
| 991 | |
| 992 | $type = null; |
| 993 | // Check for entry for full MIME type |
| 994 | if ( $mime ) { |
| 995 | $type = $this->findMediaType( $mime ); |
| 996 | if ( $type !== MEDIATYPE_UNKNOWN ) { |
| 997 | return $type; |
| 998 | } |
| 999 | } |
| 1000 | |
| 1001 | // Check for entry for file extension |
| 1002 | if ( $path ) { |
| 1003 | $i = strrpos( $path, '.' ); |
| 1004 | $e = strtolower( $i ? substr( $path, $i + 1 ) : '' ); |
| 1005 | |
| 1006 | // TODO: look at multi-extension if this fails, parse from full path |
| 1007 | $type = $this->findMediaType( '.' . $e ); |
| 1008 | if ( $type !== MEDIATYPE_UNKNOWN ) { |
| 1009 | return $type; |
| 1010 | } |
| 1011 | } |
| 1012 | |
| 1013 | // Check major MIME type |
| 1014 | if ( $mime ) { |
| 1015 | $i = strpos( $mime, '/' ); |
| 1016 | if ( $i !== false ) { |
| 1017 | $major = substr( $mime, 0, $i ); |
| 1018 | $type = $this->findMediaType( $major ); |
| 1019 | if ( $type !== MEDIATYPE_UNKNOWN ) { |
| 1020 | return $type; |
| 1021 | } |
| 1022 | } |
| 1023 | } |
| 1024 | |
| 1025 | if ( !$type ) { |
| 1026 | $type = MEDIATYPE_UNKNOWN; |
| 1027 | } |
| 1028 | |
| 1029 | return $type; |
| 1030 | } |
| 1031 | |
| 1032 | /** |
| 1033 | * Returns a media code matching the given MIME type or file extension. |
| 1034 | * |
| 1035 | * File extensions are represented by a string starting with a dot (.) to |
| 1036 | * distinguish them from MIME types. |
| 1037 | * |
| 1038 | * @param string $extMime |
| 1039 | * @return int|string |
| 1040 | */ |
| 1041 | private function findMediaType( string $extMime ) { |
| 1042 | if ( strpos( $extMime, '.' ) === 0 ) { |
| 1043 | // If it's an extension, look up the MIME types |
| 1044 | $m = $this->getMimeTypesFromExtension( substr( $extMime, 1 ) ); |
| 1045 | if ( !$m ) { |
| 1046 | return MEDIATYPE_UNKNOWN; |
| 1047 | } |
| 1048 | } else { |
| 1049 | // Normalize MIME type |
| 1050 | if ( isset( $this->mimeTypeAliases[$extMime] ) ) { |
| 1051 | $extMime = $this->mimeTypeAliases[$extMime]; |
| 1052 | } |
| 1053 | |
| 1054 | $m = [ $extMime ]; |
| 1055 | } |
| 1056 | |
| 1057 | foreach ( $m as $mime ) { |
| 1058 | foreach ( $this->mediaTypes as $type => $codes ) { |
| 1059 | if ( in_array( $mime, $codes, true ) ) { |
| 1060 | return $type; |
| 1061 | } |
| 1062 | } |
| 1063 | } |
| 1064 | |
| 1065 | return MEDIATYPE_UNKNOWN; |
| 1066 | } |
| 1067 | |
| 1068 | /** |
| 1069 | * Returns an array of media types (MEDIATYPE_xxx constants) |
| 1070 | * |
| 1071 | * @return string[] |
| 1072 | */ |
| 1073 | public function getMediaTypes(): array { |
| 1074 | return array_keys( $this->mediaTypes ); |
| 1075 | } |
| 1076 | |
| 1077 | /** |
| 1078 | * Check if major_mime has a value accepted by enum in a database schema. |
| 1079 | * |
| 1080 | * @since 1.42.0 (also backported to 1.39.7, 1.40.3 and 1.41.1) |
| 1081 | * |
| 1082 | * @param string $type |
| 1083 | * @return bool |
| 1084 | */ |
| 1085 | public function isValidMajorMimeType( string $type ): bool { |
| 1086 | // From maintenance/tables-generated.sql => img_major_mime |
| 1087 | $types = [ |
| 1088 | 'unknown', |
| 1089 | 'application', |
| 1090 | 'audio', |
| 1091 | 'image', |
| 1092 | 'text', |
| 1093 | 'video', |
| 1094 | 'message', |
| 1095 | 'model', |
| 1096 | 'multipart', |
| 1097 | 'chemical', |
| 1098 | ]; |
| 1099 | |
| 1100 | return in_array( $type, $types ); |
| 1101 | } |
| 1102 | } |