Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
95.61% |
196 / 205 |
|
33.33% |
2 / 6 |
CRAP | |
0.00% |
0 / 1 |
| MediaModerationImageContentsLookup | |
95.61% |
196 / 205 |
|
33.33% |
2 / 6 |
42 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| getImageContents | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
9 | |||
| getThumbnailMimeType | |
95.45% |
21 / 22 |
|
0.00% |
0 / 1 |
5 | |||
| getThumbnailForFile | |
92.31% |
60 / 65 |
|
0.00% |
0 / 1 |
9.04 | |||
| getThumbnailContents | |
97.22% |
35 / 36 |
|
0.00% |
0 / 1 |
8 | |||
| getFileContents | |
94.87% |
37 / 39 |
|
0.00% |
0 / 1 |
10.01 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\MediaModeration\Services; |
| 4 | |
| 5 | use ArchivedFile; |
| 6 | use File; |
| 7 | use FileBackend; |
| 8 | use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface; |
| 9 | use LocalRepo; |
| 10 | use MediaTransformError; |
| 11 | use MediaWiki\Config\ServiceOptions; |
| 12 | use MediaWiki\Extension\MediaModeration\Media\ThumborThumbnailImage; |
| 13 | use MediaWiki\Extension\MediaModeration\Status\ImageContentsLookupStatus; |
| 14 | use MediaWiki\Http\HttpRequestFactory; |
| 15 | use MediaWiki\Language\RawMessage; |
| 16 | use MimeAnalyzer; |
| 17 | use StatusValue; |
| 18 | use ThumbnailImage; |
| 19 | |
| 20 | /** |
| 21 | * This service looks up the contents of the given $file, either by getting a thumbnail |
| 22 | * with width as specified in the wgMediaModerationThumbnailWidth config or if this fails |
| 23 | * using the original source file. |
| 24 | */ |
| 25 | class MediaModerationImageContentsLookup { |
| 26 | |
| 27 | public const CONSTRUCTOR_OPTIONS = [ |
| 28 | 'MediaModerationThumbnailWidth', |
| 29 | ]; |
| 30 | |
| 31 | private FileBackend $fileBackend; |
| 32 | private StatsdDataFactoryInterface $perDbNameStatsdDataFactory; |
| 33 | private MimeAnalyzer $mimeAnalyzer; |
| 34 | private LocalRepo $localRepo; |
| 35 | |
| 36 | private HttpRequestFactory $httpRequestFactory; |
| 37 | |
| 38 | private int $thumbnailWidth; |
| 39 | |
| 40 | public function __construct( |
| 41 | ServiceOptions $options, |
| 42 | FileBackend $fileBackend, |
| 43 | StatsdDataFactoryInterface $perDbNameStatsdDataFactory, |
| 44 | MimeAnalyzer $mimeAnalyzer, |
| 45 | LocalRepo $localRepo, |
| 46 | HttpRequestFactory $httpRequestFactory |
| 47 | ) { |
| 48 | $options->assertRequiredOptions( self::CONSTRUCTOR_OPTIONS ); |
| 49 | $this->fileBackend = $fileBackend; |
| 50 | $this->perDbNameStatsdDataFactory = $perDbNameStatsdDataFactory; |
| 51 | $this->mimeAnalyzer = $mimeAnalyzer; |
| 52 | $this->localRepo = $localRepo; |
| 53 | $this->thumbnailWidth = $options->get( 'MediaModerationThumbnailWidth' ); |
| 54 | $this->httpRequestFactory = $httpRequestFactory; |
| 55 | } |
| 56 | |
| 57 | /** |
| 58 | * Gets the image contents as a string for the given $file to be sent to PhotoDNA. |
| 59 | * |
| 60 | * This function first tries to get a thumbnail for the $file and return the contents of the |
| 61 | * thumbnail. If this fails, then the function tries to return the contents of the $file |
| 62 | * if the $file is in a format supported by PhotoDNA. |
| 63 | * |
| 64 | * @param File|ArchivedFile $file |
| 65 | * @return ImageContentsLookupStatus |
| 66 | */ |
| 67 | public function getImageContents( $file ): ImageContentsLookupStatus { |
| 68 | // Create a status that will be returned, and if it is good will contain the |
| 69 | // thumbnail/original file contents and mime type. |
| 70 | $returnStatus = new ImageContentsLookupStatus(); |
| 71 | if ( $file instanceof File ) { |
| 72 | // Only try to use the thumbnail if the $file is an instance of the File class, |
| 73 | // as support for generating a thumbnail for deleted files is not implemented. |
| 74 | $thumbnailStatus = $this->getThumbnailForFile( $file ); |
| 75 | if ( $thumbnailStatus->isGood() ) { |
| 76 | // If we could get the ThumbnailImage object for the $file, then |
| 77 | // try to get the contents of the thumbnail along with the mime type. |
| 78 | $thumbnail = $thumbnailStatus->getValue(); |
| 79 | $thumbnailContentsStatus = $this->getThumbnailContents( $thumbnail ); |
| 80 | $thumbnailMimeTypeStatus = $this->getThumbnailMimeType( $thumbnail ); |
| 81 | if ( $thumbnailContentsStatus->isGood() && $thumbnailMimeTypeStatus->isGood() ) { |
| 82 | // If we were able to get the thumbnail contents and mime type, then return with them. |
| 83 | return $returnStatus |
| 84 | ->setImageContents( $thumbnailContentsStatus->getValue() ) |
| 85 | ->setMimeType( $file->getMimeType() ); |
| 86 | } |
| 87 | // Add the failures to the return status for the caller to see. |
| 88 | $returnStatus |
| 89 | ->merge( $thumbnailMimeTypeStatus ) |
| 90 | ->merge( $thumbnailContentsStatus ); |
| 91 | } |
| 92 | // Add the failures to the return status for the caller to see. |
| 93 | $returnStatus->merge( $thumbnailStatus ); |
| 94 | } |
| 95 | // If no thumbnail could be generated for the $file or the $file is an ArchivedFile instance, then we will |
| 96 | // reach here. Now try to get the contents of the $file if the mime type type is supported by PhotoDNA. |
| 97 | if ( in_array( $file->getMimeType(), MediaModerationFileProcessor::ALLOWED_MIME_TYPES, true ) ) { |
| 98 | $fileContentsStatus = $this->getFileContents( $file ); |
| 99 | if ( $fileContentsStatus->isGood() ) { |
| 100 | // We were able to get the contents of the $file |
| 101 | if ( $file instanceof File ) { |
| 102 | // Add to the RuntimeException count if $file was a File, as we should |
| 103 | // have been able to generate a thumbnail for it. |
| 104 | $this->perDbNameStatsdDataFactory->increment( |
| 105 | 'MediaModeration.PhotoDNAServiceProvider.Execute.SourceFileUsedForFileObject' |
| 106 | ); |
| 107 | } |
| 108 | // Set the result as OK as we got the original file, but still include the |
| 109 | // errors from the thumbnail generation for tracking. |
| 110 | return $returnStatus |
| 111 | ->setOK( true ) |
| 112 | ->setImageContents( $fileContentsStatus->getValue() ) |
| 113 | ->setMimeType( $file->getMimeType() ); |
| 114 | } |
| 115 | // If we were unable to get the contents of the $file, then add the errors from |
| 116 | // this to the return status. |
| 117 | $returnStatus->merge( $fileContentsStatus ); |
| 118 | } |
| 119 | // If we get here, then we have failed to get any image contents and so should return a fatal status. |
| 120 | if ( $returnStatus->isOK() ) { |
| 121 | // The $returnStatus can be good and have no message if the image was deleted and the source image is |
| 122 | // not supported by PhotoDNA (such as a deleted SVG). |
| 123 | $returnStatus->fatal( new RawMessage( |
| 124 | 'Failed to get image contents for $1', |
| 125 | [ $file->getName() ] |
| 126 | ) ); |
| 127 | } |
| 128 | // Increment the RuntimeException statsd counter, as we have reached a point where |
| 129 | // we could not generate a thumbnail where we should have been able to. |
| 130 | $this->perDbNameStatsdDataFactory->increment( |
| 131 | 'MediaModeration.PhotoDNAServiceProvider.Execute.RuntimeException' |
| 132 | ); |
| 133 | return $returnStatus; |
| 134 | } |
| 135 | |
| 136 | /** |
| 137 | * Gets the mime type (or best guess for it) of the given $thumbnail. |
| 138 | * |
| 139 | * @param ThumbnailImage $thumbnail |
| 140 | * @return StatusValue |
| 141 | */ |
| 142 | protected function getThumbnailMimeType( ThumbnailImage $thumbnail ): StatusValue { |
| 143 | // Attempt to work out what the mime type of the file is based on the extension, and if that |
| 144 | // fails then try based on the contents of the thumbnail. |
| 145 | $thumbnailMimeType = $thumbnail instanceof ThumborThumbnailImage ? |
| 146 | $thumbnail->getContentType() : |
| 147 | $this->mimeAnalyzer->getMimeTypeFromExtensionOrNull( $thumbnail->getExtension() ); |
| 148 | if ( $thumbnailMimeType === null ) { |
| 149 | $thumbnailMimeType = $this->mimeAnalyzer->guessMimeType( $thumbnail->getLocalCopyPath() ); |
| 150 | } |
| 151 | if ( !$thumbnailMimeType ) { |
| 152 | // We cannot send a request to PhotoDNA without knowing what the mime type is. |
| 153 | $this->perDbNameStatsdDataFactory->increment( |
| 154 | 'MediaModeration.ImageContentsLookup.Thumbnail.MimeType.LookupFailed' |
| 155 | ); |
| 156 | return StatusValue::newFatal( new RawMessage( |
| 157 | 'Could not get mime type of thumbnail for $1', |
| 158 | [ $thumbnail->getFile()->getName() ] |
| 159 | ) ); |
| 160 | } |
| 161 | if ( !in_array( $thumbnailMimeType, MediaModerationFileProcessor::ALLOWED_MIME_TYPES, true ) ) { |
| 162 | // We cannot send a request to PhotoDNA with a thumbnail type that is unsupported by the API. |
| 163 | $this->perDbNameStatsdDataFactory->increment( |
| 164 | 'MediaModeration.ImageContentsLookup.Thumbnail.MimeType.Unsupported' |
| 165 | ); |
| 166 | return StatusValue::newFatal( new RawMessage( |
| 167 | 'Mime type of thumbnail for $1 is not supported by PhotoDNA.', |
| 168 | [ $thumbnail->getFile()->getName() ] |
| 169 | ) ); |
| 170 | } |
| 171 | return StatusValue::newGood( $thumbnailMimeType ); |
| 172 | } |
| 173 | |
| 174 | /** |
| 175 | * @param File $file |
| 176 | * @return StatusValue<ThumbnailImage|ThumborThumbnailImage> A StatusValue with a ThumbnailImage object as the value |
| 177 | * if it is a good status. |
| 178 | */ |
| 179 | protected function getThumbnailForFile( File $file ): StatusValue { |
| 180 | $genericErrorMessage = 'Could not transform file ' . $file->getName(); |
| 181 | $start = microtime( true ); |
| 182 | $thumbName = $file->thumbName( [ 'width' => $this->thumbnailWidth ] ); |
| 183 | $thumbProxyUrl = $file->getRepo()->getThumbProxyUrl(); |
| 184 | $secret = $file->getRepo()->getThumbProxySecret(); |
| 185 | if ( $thumbProxyUrl && $secret ) { |
| 186 | // Specific to Wikimedia setup only: proxy the request to Thumbor, |
| 187 | // which should result in the thumbnail being generated on disk |
| 188 | // @see wfProxyThumbnailRequest() |
| 189 | $req = $this->httpRequestFactory->create( |
| 190 | $thumbProxyUrl . $file->getThumbRel( $thumbName ) |
| 191 | ); |
| 192 | $req->setHeader( 'X-Swift-Secret', $secret ); |
| 193 | $result = $req->execute(); |
| 194 | if ( $result->isGood() ) { |
| 195 | $imageContent = $req->getContent(); |
| 196 | // getimagesizefromstring() can return a PHP Notice if |
| 197 | // the contents are invalid. Suppress the notice, and check |
| 198 | // instead of the result is truthy. |
| 199 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 200 | $imageMetadata = @getimagesizefromstring( $imageContent ); |
| 201 | if ( $imageMetadata ) { |
| 202 | $thumbnail = new ThumborThumbnailImage( |
| 203 | $file, |
| 204 | $file->getThumbUrl( $thumbName ), |
| 205 | [ |
| 206 | 'width' => $imageMetadata[0], |
| 207 | 'height' => $imageMetadata[1] |
| 208 | ], |
| 209 | $imageContent, |
| 210 | $req->getResponseHeader( 'content-type' ) |
| 211 | ); |
| 212 | $this->perDbNameStatsdDataFactory->timing( |
| 213 | 'MediaModeration.PhotoDNAServiceProviderThumbnailTransformThumborRequest', |
| 214 | 1000 * ( microtime( true ) - $start ) |
| 215 | ); |
| 216 | return StatusValue::newGood( $thumbnail ); |
| 217 | } |
| 218 | } |
| 219 | // The request failed, so increment the failure counter and use regular ::transform |
| 220 | // for checks done farther on. |
| 221 | $this->perDbNameStatsdDataFactory->increment( |
| 222 | 'MediaModeration.ImageContentsLookup.Thumbnail.ThumborTransform.Failed' |
| 223 | ); |
| 224 | $thumbnail = $file->transform( [ 'width' => $this->thumbnailWidth ] ); |
| 225 | } else { |
| 226 | // For non Wikimedia setups, use RENDER_NOW to ensure we have |
| 227 | // a file to work with. |
| 228 | $thumbnail = $file->transform( [ 'width' => $this->thumbnailWidth ], File::RENDER_NOW ); |
| 229 | } |
| 230 | $delay = microtime( true ) - $start; |
| 231 | $this->perDbNameStatsdDataFactory->timing( |
| 232 | 'MediaModeration.PhotoDNAServiceProviderThumbnailTransform', |
| 233 | 1000 * $delay |
| 234 | ); |
| 235 | if ( !$thumbnail ) { |
| 236 | $this->perDbNameStatsdDataFactory->increment( |
| 237 | 'MediaModeration.ImageContentsLookup.Thumbnail.Transform.Failed' |
| 238 | ); |
| 239 | return StatusValue::newFatal( new RawMessage( $genericErrorMessage ) ); |
| 240 | } |
| 241 | if ( $thumbnail instanceof MediaTransformError ) { |
| 242 | $this->perDbNameStatsdDataFactory->increment( |
| 243 | 'MediaModeration.ImageContentsLookup.Thumbnail.Transform.Failed' |
| 244 | ); |
| 245 | return StatusValue::newFatal( new RawMessage( $genericErrorMessage . ': ' . $thumbnail->toText() ) ); |
| 246 | } |
| 247 | if ( !( $thumbnail instanceof ThumbnailImage ) ) { |
| 248 | $this->perDbNameStatsdDataFactory->increment( |
| 249 | 'MediaModeration.ImageContentsLookup.Thumbnail.Transform.Failed' |
| 250 | ); |
| 251 | return StatusValue::newFatal( new RawMessage( |
| 252 | $genericErrorMessage . ': not an instance of ThumbnailImage, got ' . get_class( $thumbnail ) |
| 253 | ) ); |
| 254 | } |
| 255 | if ( !$thumbnail->hasFile() ) { |
| 256 | $this->perDbNameStatsdDataFactory->increment( |
| 257 | 'MediaModeration.ImageContentsLookup.Thumbnail.Transform.Failed' |
| 258 | ); |
| 259 | return StatusValue::newFatal( new RawMessage( |
| 260 | $genericErrorMessage . ', got a ' . get_class( $thumbnail ) . ' but ::hasFile() returns false.' |
| 261 | ) ); |
| 262 | } |
| 263 | return StatusValue::newGood( $thumbnail ); |
| 264 | } |
| 265 | |
| 266 | protected function getThumbnailContents( ThumbnailImage $thumbnail ): StatusValue { |
| 267 | if ( $thumbnail->getHeight() < 160 || $thumbnail->getWidth() < 160 ) { |
| 268 | $this->perDbNameStatsdDataFactory->increment( |
| 269 | 'MediaModeration.ImageContentsLookup.Thumbnail.Contents.TooSmall' |
| 270 | ); |
| 271 | // PhotoDNA requires that images be at least 160px by 160px, so don't use the |
| 272 | // thumbnail if either dimension is too small. |
| 273 | return StatusValue::newFatal( new RawMessage( |
| 274 | 'Thumbnail does not meet dimension requirements for $1', |
| 275 | [ $thumbnail->getFile()->getName() ] |
| 276 | ) ); |
| 277 | } |
| 278 | if ( !( $thumbnail instanceof ThumborThumbnailImage ) && !$thumbnail->getStoragePath() ) { |
| 279 | $this->perDbNameStatsdDataFactory->increment( |
| 280 | 'MediaModeration.ImageContentsLookup.Thumbnail.Contents.LookupFailed' |
| 281 | ); |
| 282 | return StatusValue::newFatal( new RawMessage( |
| 283 | 'Could not get storage path of thumbnail for $1', |
| 284 | [ $thumbnail->getFile()->getName() ] |
| 285 | ) ); |
| 286 | } |
| 287 | $fileContents = $thumbnail instanceof ThumborThumbnailImage ? |
| 288 | $thumbnail->getContent() : |
| 289 | $this->fileBackend->getFileContents( [ 'src' => $thumbnail->getStoragePath() ] ); |
| 290 | if ( !$fileContents ) { |
| 291 | $this->perDbNameStatsdDataFactory->increment( |
| 292 | 'MediaModeration.ImageContentsLookup.Thumbnail.Contents.LookupFailed' |
| 293 | ); |
| 294 | return StatusValue::newFatal( new RawMessage( |
| 295 | 'Could not get thumbnail contents for $1', |
| 296 | [ $thumbnail->getFile()->getName() ] |
| 297 | ) ); |
| 298 | } |
| 299 | if ( strlen( $fileContents ) > 4000000 ) { |
| 300 | $this->perDbNameStatsdDataFactory->increment( |
| 301 | 'MediaModeration.ImageContentsLookup.Thumbnail.Contents.TooLarge' |
| 302 | ); |
| 303 | // Check that the size of the file does not exceed 4MB, as PhotoDNA returns an |
| 304 | // error for files that are any larger. |
| 305 | // strlen returns the size of the string in bytes and 4MB is 4,000,000 bytes. |
| 306 | return StatusValue::newFatal( new RawMessage( |
| 307 | 'Original file contents exceeds 4MB for $1', |
| 308 | [ $thumbnail->getFile()->getName() ] |
| 309 | ) ); |
| 310 | } |
| 311 | return StatusValue::newGood( $fileContents ); |
| 312 | } |
| 313 | |
| 314 | /** |
| 315 | * Gets the contents of the originally uploaded file referenced by $file. |
| 316 | * |
| 317 | * @param File|ArchivedFile $file |
| 318 | * @return StatusValue |
| 319 | */ |
| 320 | protected function getFileContents( $file ): StatusValue { |
| 321 | if ( $file->getSize() && $file->getSize() > 4000000 ) { |
| 322 | $this->perDbNameStatsdDataFactory->increment( |
| 323 | 'MediaModeration.ImageContentsLookup.File.Contents.TooLarge' |
| 324 | ); |
| 325 | // Check that the size of the file does not exceed 4MB, as PhotoDNA returns an |
| 326 | // error for files that are any larger. |
| 327 | return StatusValue::newFatal( new RawMessage( |
| 328 | 'Original file contents exceeds 4MB for $1', |
| 329 | [ $file->getName() ] |
| 330 | ) ); |
| 331 | } |
| 332 | if ( |
| 333 | ( $file->getHeight() && $file->getHeight() < 160 ) || |
| 334 | ( $file->getWidth() && $file->getWidth() < 160 ) |
| 335 | ) { |
| 336 | $this->perDbNameStatsdDataFactory->increment( |
| 337 | 'MediaModeration.ImageContentsLookup.File.Contents.TooSmall' |
| 338 | ); |
| 339 | // Check that the height and width is at least 160px by 160px |
| 340 | // as PhotoDNA requires that the file be at least that size. |
| 341 | // If the height or width is false, then just ignore this check |
| 342 | // as PhotoDNA will verify this for us. |
| 343 | return StatusValue::newFatal( new RawMessage( |
| 344 | 'Original file does not meet dimension requirements for $1', |
| 345 | [ $file->getName() ] |
| 346 | ) ); |
| 347 | } |
| 348 | if ( $file instanceof ArchivedFile ) { |
| 349 | // Format for the URL is copied from SpecialUndelete::showFile |
| 350 | $filePath = $this->localRepo->getZonePath( 'deleted' ) . '/' . |
| 351 | $this->localRepo->getDeletedHashPath( $file->getStorageKey() ) . $file->getStorageKey(); |
| 352 | } else { |
| 353 | $filePath = $file->getPath(); |
| 354 | } |
| 355 | if ( !$filePath ) { |
| 356 | $this->perDbNameStatsdDataFactory->increment( |
| 357 | 'MediaModeration.ImageContentsLookup.File.Contents.LookupFailed' |
| 358 | ); |
| 359 | return StatusValue::newFatal( new RawMessage( |
| 360 | 'Could not get storage path of original file for $1', |
| 361 | [ $file->getName() ] |
| 362 | ) ); |
| 363 | } |
| 364 | $fileContents = $this->fileBackend->getFileContents( [ 'src' => $filePath ] ); |
| 365 | if ( !$fileContents ) { |
| 366 | $this->perDbNameStatsdDataFactory->increment( |
| 367 | 'MediaModeration.ImageContentsLookup.File.Contents.LookupFailed' |
| 368 | ); |
| 369 | return StatusValue::newFatal( new RawMessage( |
| 370 | 'Could not get original file contents for $1', |
| 371 | [ $file->getName() ] |
| 372 | ) ); |
| 373 | } |
| 374 | return StatusValue::newGood( $fileContents ); |
| 375 | } |
| 376 | } |