Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
87.91% |
189 / 215 |
|
42.86% |
6 / 14 |
CRAP | |
0.00% |
0 / 1 |
| ApiDetailRetriever | |
87.91% |
189 / 215 |
|
42.86% |
6 / 14 |
63.95 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
2 | |||
| sendApiRequest | |
57.14% |
4 / 7 |
|
0.00% |
0 / 1 |
2.31 | |||
| getImportDetails | |
98.15% |
53 / 54 |
|
0.00% |
0 / 1 |
7 | |||
| reduceTitleList | |
60.00% |
3 / 5 |
|
0.00% |
0 / 1 |
3.58 | |||
| getMoreRevisions | |
79.31% |
23 / 29 |
|
0.00% |
0 / 1 |
9.72 | |||
| checkRevisionCount | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
5 | |||
| checkMaxRevisionAggregatedBytes | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| getFileRevisionsFromImageInfo | |
65.00% |
13 / 20 |
|
0.00% |
0 / 1 |
10.74 | |||
| getTextRevisionsFromRevisionsInfo | |
64.29% |
9 / 14 |
|
0.00% |
0 / 1 |
6.14 | |||
| getBaseParams | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| addTextRevisionsToParams | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
3 | |||
| addFileRevisionsToParams | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
3 | |||
| addTemplatesToParams | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
3.14 | |||
| addCategoriesToParams | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
3.14 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace FileImporter\Remote\MediaWiki; |
| 4 | |
| 5 | use FileImporter\Data\FileRevision; |
| 6 | use FileImporter\Data\FileRevisions; |
| 7 | use FileImporter\Data\ImportDetails; |
| 8 | use FileImporter\Data\SourceUrl; |
| 9 | use FileImporter\Data\TextRevision; |
| 10 | use FileImporter\Data\TextRevisions; |
| 11 | use FileImporter\Exceptions\HttpRequestException; |
| 12 | use FileImporter\Exceptions\ImportException; |
| 13 | use FileImporter\Exceptions\LocalizedImportException; |
| 14 | use FileImporter\Interfaces\DetailRetriever; |
| 15 | use FileImporter\Services\Http\HttpRequestExecutor; |
| 16 | use MediaWiki\Config\ConfigException; |
| 17 | use MediaWiki\MediaWikiServices; |
| 18 | use MediaWiki\Revision\SlotRecord; |
| 19 | use MediaWiki\Title\TitleValue; |
| 20 | use Psr\Log\LoggerInterface; |
| 21 | use Psr\Log\NullLogger; |
| 22 | |
| 23 | /** |
| 24 | * @license GPL-2.0-or-later |
| 25 | * @author Addshore |
| 26 | */ |
| 27 | class ApiDetailRetriever implements DetailRetriever { |
| 28 | use MediaWikiSourceUrlParser; |
| 29 | |
| 30 | /** |
| 31 | * @var string Placeholder name replacing usernames that have been suppressed as part of |
| 32 | * a steward action on the source site. |
| 33 | */ |
| 34 | private $suppressedUsername; |
| 35 | private int $maxRevisions; |
| 36 | private int $maxAggregatedBytes; |
| 37 | |
| 38 | private const API_RESULT_LIMIT = 500; |
| 39 | private const MAX_REVISIONS = 100; |
| 40 | private const MAX_AGGREGATED_BYTES = 250000000; |
| 41 | |
| 42 | /** |
| 43 | * @throws ConfigException when $wgFileImporterAccountForSuppressedUsername is invalid |
| 44 | */ |
| 45 | public function __construct( |
| 46 | private readonly HttpApiLookup $httpApiLookup, |
| 47 | private readonly HttpRequestExecutor $httpRequestExecutor, |
| 48 | private readonly int $maxBytes, |
| 49 | private readonly LoggerInterface $logger = new NullLogger(), |
| 50 | ) { |
| 51 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 52 | |
| 53 | $this->maxRevisions = (int)$config->get( 'FileImporterMaxRevisions' ); |
| 54 | $this->maxAggregatedBytes = (int)$config->get( 'FileImporterMaxAggregatedBytes' ); |
| 55 | $this->suppressedUsername = $config->get( 'FileImporterAccountForSuppressedUsername' ); |
| 56 | if ( !MediaWikiServices::getInstance()->getUserNameUtils()->isValid( $this->suppressedUsername ) ) { |
| 57 | throw new ConfigException( |
| 58 | 'Invalid username configured in wgFileImporterAccountForSuppressedUsername: "' . |
| 59 | $this->suppressedUsername . '"' |
| 60 | ); |
| 61 | } |
| 62 | } |
| 63 | |
| 64 | /** |
| 65 | * @return array[] |
| 66 | * @throws ImportException when the request failed |
| 67 | */ |
| 68 | private function sendApiRequest( SourceUrl $sourceUrl, array $apiParameters ) { |
| 69 | $apiUrl = $this->httpApiLookup->getApiUrl( $sourceUrl ); |
| 70 | |
| 71 | try { |
| 72 | $imageInfoRequest = $this->httpRequestExecutor->execute( $apiUrl, $apiParameters ); |
| 73 | } catch ( HttpRequestException $e ) { |
| 74 | throw new LocalizedImportException( [ 'fileimporter-api-failedtogetinfo', |
| 75 | $apiUrl ], $e ); |
| 76 | } |
| 77 | $requestData = json_decode( $imageInfoRequest->getContent(), true ); |
| 78 | return $requestData; |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * @throws ImportException e.g. when the file couldn't be found |
| 83 | */ |
| 84 | public function getImportDetails( SourceUrl $sourceUrl ): ImportDetails { |
| 85 | $params = $this->getBaseParams( $sourceUrl ); |
| 86 | $params = $this->addFileRevisionsToParams( $params ); |
| 87 | $params = $this->addTextRevisionsToParams( $params ); |
| 88 | $params = $this->addTemplatesToParams( $params ); |
| 89 | $params = $this->addCategoriesToParams( $params ); |
| 90 | |
| 91 | $requestData = $this->sendApiRequest( $sourceUrl, $params ); |
| 92 | |
| 93 | if ( count( $requestData['query']['pages'] ?? [] ) !== 1 ) { |
| 94 | $this->logger->warning( |
| 95 | 'No pages returned by the API', |
| 96 | [ |
| 97 | 'sourceUrl' => $sourceUrl->getUrl(), |
| 98 | 'apiParameters' => $params, |
| 99 | ] |
| 100 | ); |
| 101 | throw new LocalizedImportException( 'fileimporter-api-nopagesreturned' ); |
| 102 | } |
| 103 | |
| 104 | /** @var array $pageInfoData */ |
| 105 | $pageInfoData = array_last( $requestData['query']['pages'] ); |
| 106 | '@phan-var array $pageInfoData'; |
| 107 | |
| 108 | if ( ( $pageInfoData['missing'] ?? false ) !== false ) { |
| 109 | if ( ( $pageInfoData['imagerepository'] ?? null ) === 'shared' ) { |
| 110 | throw new LocalizedImportException( |
| 111 | [ 'fileimporter-cantimportfromsharedrepo', $sourceUrl->getHost() ] |
| 112 | ); |
| 113 | } |
| 114 | throw new LocalizedImportException( 'fileimporter-cantimportmissingfile' ); |
| 115 | } |
| 116 | |
| 117 | if ( empty( $pageInfoData['imageinfo'] ) || empty( $pageInfoData['revisions'] ) ) { |
| 118 | $this->logger->warning( |
| 119 | 'Bad image or revision info returned by the API', |
| 120 | [ |
| 121 | 'sourceUrl' => $sourceUrl->getUrl(), |
| 122 | 'apiParameters' => $params, |
| 123 | ] |
| 124 | ); |
| 125 | throw new LocalizedImportException( 'fileimporter-api-badinfo' ); |
| 126 | } |
| 127 | |
| 128 | // FIXME: Isn't this misplaced here, *before* more revisions are fetched? |
| 129 | $this->checkRevisionCount( $sourceUrl, $pageInfoData ); |
| 130 | $this->checkMaxRevisionAggregatedBytes( $pageInfoData ); |
| 131 | |
| 132 | while ( array_key_exists( 'continue', $requestData ) ) { |
| 133 | $this->getMoreRevisions( $sourceUrl, $requestData, $pageInfoData ); |
| 134 | } |
| 135 | |
| 136 | $pageTitle = $pageInfoData['title']; |
| 137 | $pageLanguage = $pageInfoData['pagelanguagehtmlcode'] ?? null; |
| 138 | |
| 139 | $imageInfoData = $pageInfoData['imageinfo']; |
| 140 | $revisionsData = $pageInfoData['revisions']; |
| 141 | $fileRevisions = $this->getFileRevisionsFromImageInfo( $imageInfoData, $pageTitle ); |
| 142 | $textRevisions = $this->getTextRevisionsFromRevisionsInfo( $revisionsData, $pageTitle ); |
| 143 | $templates = $this->reduceTitleList( $pageInfoData['templates'] ?? [], NS_TEMPLATE ); |
| 144 | $categories = $this->reduceTitleList( $pageInfoData['categories'] ?? [], NS_CATEGORY ); |
| 145 | |
| 146 | $titleAfterColon = array_last( explode( ':', $pageInfoData['title'], 2 ) ); |
| 147 | |
| 148 | $importDetails = new ImportDetails( |
| 149 | $sourceUrl, |
| 150 | new TitleValue( NS_FILE, $titleAfterColon ), |
| 151 | $textRevisions, |
| 152 | $fileRevisions |
| 153 | ); |
| 154 | // FIXME: Better use constructor parameters instead of setters? |
| 155 | $importDetails->setPageLanguage( $pageLanguage ); |
| 156 | $importDetails->setTemplates( $templates ); |
| 157 | $importDetails->setCategories( $categories ); |
| 158 | |
| 159 | return $importDetails; |
| 160 | } |
| 161 | |
| 162 | /** |
| 163 | * @param array[] $results Result set as returned by the API |
| 164 | * @param int $namespace |
| 165 | * |
| 166 | * @return string[] |
| 167 | */ |
| 168 | private function reduceTitleList( array $results, int $namespace ): array { |
| 169 | $titles = []; |
| 170 | foreach ( $results as $result ) { |
| 171 | if ( $result['ns'] === $namespace ) { |
| 172 | $titles[] = $result['title']; |
| 173 | } |
| 174 | } |
| 175 | return $titles; |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * Fetches the next set of revisions unless the number of revisions |
| 180 | * exceeds the max revisions limit |
| 181 | * |
| 182 | * @param SourceUrl $sourceUrl |
| 183 | * @param array[] &$requestData |
| 184 | * @param array[] &$pageInfoData |
| 185 | * |
| 186 | * @throws ImportException |
| 187 | */ |
| 188 | private function getMoreRevisions( |
| 189 | SourceUrl $sourceUrl, |
| 190 | array &$requestData, |
| 191 | array &$pageInfoData |
| 192 | ): void { |
| 193 | $rvContinue = $requestData['continue']['rvcontinue'] ?? null; |
| 194 | $iiStart = $requestData['continue']['iistart'] ?? null; |
| 195 | $tlContinue = $requestData['continue']['tlcontinue'] ?? null; |
| 196 | $clContinue = $requestData['continue']['clcontinue'] ?? null; |
| 197 | |
| 198 | $params = $this->getBaseParams( $sourceUrl ); |
| 199 | |
| 200 | if ( $iiStart ) { |
| 201 | $params = $this->addFileRevisionsToParams( $params, $iiStart ); |
| 202 | } |
| 203 | |
| 204 | if ( $rvContinue ) { |
| 205 | $params = $this->addTextRevisionsToParams( $params, $rvContinue ); |
| 206 | } |
| 207 | |
| 208 | if ( $tlContinue ) { |
| 209 | $params = $this->addTemplatesToParams( $params, $tlContinue ); |
| 210 | } |
| 211 | |
| 212 | if ( $clContinue ) { |
| 213 | $params = $this->addCategoriesToParams( $params, $clContinue ); |
| 214 | } |
| 215 | |
| 216 | $requestData = $this->sendApiRequest( $sourceUrl, $params ); |
| 217 | |
| 218 | $newPageInfoData = array_last( $requestData['query']['pages'] ); |
| 219 | |
| 220 | if ( array_key_exists( 'revisions', $newPageInfoData ) ) { |
| 221 | $pageInfoData['revisions'] = |
| 222 | array_merge( $pageInfoData['revisions'], $newPageInfoData['revisions'] ); |
| 223 | } |
| 224 | |
| 225 | if ( array_key_exists( 'imageinfo', $newPageInfoData ) ) { |
| 226 | $pageInfoData['imageinfo'] = |
| 227 | array_merge( $pageInfoData['imageinfo'], $newPageInfoData['imageinfo'] ); |
| 228 | } |
| 229 | |
| 230 | if ( array_key_exists( 'templates', $newPageInfoData ) ) { |
| 231 | $pageInfoData['templates'] = |
| 232 | array_merge( $pageInfoData['templates'], $newPageInfoData['templates'] ); |
| 233 | } |
| 234 | |
| 235 | if ( array_key_exists( 'categories', $newPageInfoData ) ) { |
| 236 | $pageInfoData['categories'] = |
| 237 | array_merge( $pageInfoData['categories'], $newPageInfoData['categories'] ); |
| 238 | } |
| 239 | |
| 240 | $this->checkRevisionCount( $sourceUrl, $pageInfoData ); |
| 241 | $this->checkMaxRevisionAggregatedBytes( $pageInfoData ); |
| 242 | } |
| 243 | |
| 244 | /** |
| 245 | * Throws an exception if the number of revisions to be imported exceeds |
| 246 | * the maximum revision limit |
| 247 | * |
| 248 | * @param SourceUrl $sourceUrl |
| 249 | * @param array[] $pageInfoData |
| 250 | * |
| 251 | * @throws ImportException when exceeding the acceptable maximum |
| 252 | */ |
| 253 | private function checkRevisionCount( SourceUrl $sourceUrl, array $pageInfoData ): void { |
| 254 | if ( count( $pageInfoData['revisions'] ) > $this->maxRevisions || |
| 255 | count( $pageInfoData['imageinfo'] ) > $this->maxRevisions || |
| 256 | count( $pageInfoData['revisions'] ) > static::MAX_REVISIONS || |
| 257 | count( $pageInfoData['imageinfo'] ) > static::MAX_REVISIONS ) { |
| 258 | $this->logger->warning( |
| 259 | 'Too many revisions were being fetched', |
| 260 | [ |
| 261 | 'sourceUrl' => $sourceUrl->getUrl(), |
| 262 | ] |
| 263 | ); |
| 264 | |
| 265 | throw new LocalizedImportException( 'fileimporter-api-toomanyrevisions' ); |
| 266 | } |
| 267 | } |
| 268 | |
| 269 | /** |
| 270 | * @param array[] $pageInfoData |
| 271 | * @phan-param array{imageinfo:array{size:int}[]} $pageInfoData |
| 272 | * |
| 273 | * @throws ImportException when exceeding the maximum file size |
| 274 | */ |
| 275 | private function checkMaxRevisionAggregatedBytes( array $pageInfoData ): void { |
| 276 | $aggregatedFileBytes = 0; |
| 277 | foreach ( $pageInfoData['imageinfo'] as $fileVersion ) { |
| 278 | $aggregatedFileBytes += $fileVersion['size'] ?? 0; |
| 279 | if ( $aggregatedFileBytes > $this->maxAggregatedBytes || |
| 280 | $aggregatedFileBytes > static::MAX_AGGREGATED_BYTES ) { |
| 281 | $versions = count( $pageInfoData['imageinfo'] ); |
| 282 | throw new LocalizedImportException( [ 'fileimporter-filetoolarge', $versions ] ); |
| 283 | } |
| 284 | } |
| 285 | } |
| 286 | |
| 287 | /** |
| 288 | * @param array[] $imageInfo |
| 289 | * @param string $pageTitle |
| 290 | * |
| 291 | * @throws ImportException when the file is not acceptable, e.g. hidden or to big |
| 292 | */ |
| 293 | private function getFileRevisionsFromImageInfo( array $imageInfo, string $pageTitle ): FileRevisions { |
| 294 | $revisions = []; |
| 295 | foreach ( $imageInfo as $revisionInfo ) { |
| 296 | if ( ( $revisionInfo['filehidden'] ?? false ) !== false ) { |
| 297 | throw new LocalizedImportException( 'fileimporter-cantimportfilehidden' ); |
| 298 | } |
| 299 | |
| 300 | if ( ( $revisionInfo['filemissing'] ?? false ) !== false ) { |
| 301 | throw new LocalizedImportException( 'fileimporter-filemissinginrevision' ); |
| 302 | } |
| 303 | |
| 304 | if ( ( $revisionInfo['userhidden'] ?? false ) !== false ) { |
| 305 | $revisionInfo['user'] ??= $this->suppressedUsername; |
| 306 | } |
| 307 | |
| 308 | if ( ( $revisionInfo['size'] ?? 0 ) > $this->maxBytes ) { |
| 309 | $versions = count( $imageInfo ); |
| 310 | throw new LocalizedImportException( [ 'fileimporter-filetoolarge', $versions ] ); |
| 311 | } |
| 312 | |
| 313 | if ( isset( $revisionInfo['sha1'] ) ) { |
| 314 | // Convert from API sha1 format to DB sha1 format. The conversion can be se inside |
| 315 | // ApiQueryImageInfo. |
| 316 | // * API sha1 format is base 16 padded to 40 chars |
| 317 | // * DB sha1 format is base 36 padded to 31 chars |
| 318 | $revisionInfo['sha1'] = \Wikimedia\base_convert( $revisionInfo['sha1'], 16, 36, 31 ); |
| 319 | } |
| 320 | |
| 321 | if ( ( $revisionInfo['commenthidden'] ?? false ) !== false ) { |
| 322 | $revisionInfo['comment'] ??= |
| 323 | wfMessage( 'fileimporter-revision-removed-comment' )->plain(); |
| 324 | } |
| 325 | |
| 326 | $revisionInfo['name'] = $pageTitle; |
| 327 | $revisionInfo['description'] = $revisionInfo['comment'] ?? null; |
| 328 | |
| 329 | $revisions[] = new FileRevision( $revisionInfo ); |
| 330 | } |
| 331 | return new FileRevisions( $revisions ); |
| 332 | } |
| 333 | |
| 334 | /** |
| 335 | * @param array[] $revisionsInfo |
| 336 | * @param string $pageTitle |
| 337 | */ |
| 338 | private function getTextRevisionsFromRevisionsInfo( array $revisionsInfo, string $pageTitle ): TextRevisions { |
| 339 | $revisions = []; |
| 340 | foreach ( $revisionsInfo as $revisionInfo ) { |
| 341 | if ( ( $revisionInfo['userhidden'] ?? false ) !== false ) { |
| 342 | $revisionInfo['user'] ??= $this->suppressedUsername; |
| 343 | } |
| 344 | |
| 345 | if ( ( $revisionInfo['texthidden'] ?? false ) !== false ) { |
| 346 | $revisionInfo['slots'][SlotRecord::MAIN]['content'] ??= |
| 347 | wfMessage( 'fileimporter-revision-removed-text' )->plain(); |
| 348 | } |
| 349 | |
| 350 | if ( ( $revisionInfo['commenthidden'] ?? false ) !== false ) { |
| 351 | $revisionInfo['comment'] ??= |
| 352 | wfMessage( 'fileimporter-revision-removed-comment' )->plain(); |
| 353 | } |
| 354 | |
| 355 | // Backwards-compatible with formatversion=1 where "" means true |
| 356 | $revisionInfo['minor'] = ( $revisionInfo['minor'] ?? false ) !== false; |
| 357 | $revisionInfo['title'] = $pageTitle; |
| 358 | $revisions[] = new TextRevision( $revisionInfo ); |
| 359 | } |
| 360 | return new TextRevisions( $revisions ); |
| 361 | } |
| 362 | |
| 363 | private function getBaseParams( SourceUrl $sourceUrl ): array { |
| 364 | return [ |
| 365 | 'action' => 'query', |
| 366 | 'errorformat' => 'plaintext', |
| 367 | 'format' => 'json', |
| 368 | 'formatversion' => '2', |
| 369 | 'titles' => $this->parseTitleFromSourceUrl( $sourceUrl ), |
| 370 | 'prop' => 'info' |
| 371 | ]; |
| 372 | } |
| 373 | |
| 374 | /** |
| 375 | * Adds to params base the properties for getting Text Revisions |
| 376 | */ |
| 377 | private function addTextRevisionsToParams( array $params, ?string $rvContinue = null ): array { |
| 378 | $params['prop'] .= ( $params['prop'] ) ? '|revisions' : 'revisions'; |
| 379 | |
| 380 | if ( $rvContinue ) { |
| 381 | $params['rvcontinue'] = $rvContinue; |
| 382 | } |
| 383 | |
| 384 | return $params + [ |
| 385 | 'rvlimit' => static::API_RESULT_LIMIT, |
| 386 | 'rvdir' => 'newer', |
| 387 | 'rvslots' => SlotRecord::MAIN, |
| 388 | 'rvprop' => implode( |
| 389 | '|', |
| 390 | [ |
| 391 | 'flags', |
| 392 | 'timestamp', |
| 393 | 'user', |
| 394 | 'sha1', |
| 395 | 'contentmodel', |
| 396 | 'comment', |
| 397 | 'content', |
| 398 | 'tags', |
| 399 | ] |
| 400 | ) |
| 401 | ]; |
| 402 | } |
| 403 | |
| 404 | /** |
| 405 | * Adds to params base the properties for getting File Revisions |
| 406 | */ |
| 407 | private function addFileRevisionsToParams( array $params, ?string $iiStart = null ): array { |
| 408 | $params['prop'] .= ( $params['prop'] ) ? '|imageinfo' : 'imageinfo'; |
| 409 | |
| 410 | if ( $iiStart ) { |
| 411 | $params['iistart'] = $iiStart; |
| 412 | } |
| 413 | |
| 414 | return $params + [ |
| 415 | 'iilimit' => static::API_RESULT_LIMIT, |
| 416 | 'iiurlwidth' => 800, |
| 417 | 'iiurlheight' => 400, |
| 418 | 'iiprop' => implode( |
| 419 | '|', |
| 420 | [ |
| 421 | 'timestamp', |
| 422 | 'user', |
| 423 | 'userid', |
| 424 | 'comment', |
| 425 | 'canonicaltitle', |
| 426 | 'url', |
| 427 | 'size', |
| 428 | 'sha1', |
| 429 | 'archivename', |
| 430 | ] |
| 431 | ) |
| 432 | ]; |
| 433 | } |
| 434 | |
| 435 | /** |
| 436 | * Adds to params base the properties for getting Templates |
| 437 | */ |
| 438 | private function addTemplatesToParams( array $params, ?string $tlContinue = null ): array { |
| 439 | $params['prop'] .= ( $params['prop'] ) ? '|templates' : 'templates'; |
| 440 | |
| 441 | if ( $tlContinue ) { |
| 442 | $params['tlcontinue'] = $tlContinue; |
| 443 | } |
| 444 | |
| 445 | return $params + [ 'tlnamespace' => NS_TEMPLATE, 'tllimit' => static::API_RESULT_LIMIT ]; |
| 446 | } |
| 447 | |
| 448 | /** |
| 449 | * Adds to params base the properties for getting Categories |
| 450 | */ |
| 451 | private function addCategoriesToParams( array $params, ?string $clContinue = null ): array { |
| 452 | $params['prop'] .= ( $params['prop'] ) ? '|categories' : 'categories'; |
| 453 | |
| 454 | if ( $clContinue ) { |
| 455 | $params['clcontinue'] = $clContinue; |
| 456 | } |
| 457 | |
| 458 | return $params + [ 'cllimit' => static::API_RESULT_LIMIT ]; |
| 459 | } |
| 460 | |
| 461 | } |