Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
56.25% |
306 / 544 |
|
29.41% |
15 / 51 |
CRAP | |
0.00% |
0 / 1 |
| WikiImporter | |
56.35% |
306 / 543 |
|
29.41% |
15 / 51 |
3290.14 | |
0.00% |
0 / 1 |
| __construct | |
96.00% |
24 / 25 |
|
0.00% |
0 / 1 |
2 | |||
| getReader | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| throwXmlError | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| debug | |
50.00% |
1 / 2 |
|
0.00% |
0 / 1 |
2.50 | |||
| warn | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| notice | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| setDebug | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setNoUpdates | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setPageOffset | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setNoticeCallback | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setPageCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setPageOutCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setRevisionCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setUploadCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setLogItemCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setSiteInfoCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| setImportTitleFactory | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setTargetNamespace | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
20 | |||
| setTargetRootPage | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
56 | |||
| setImageBasePath | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setImportUploads | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setUsernamePrefix | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| disableStatisticsUpdate | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| beforeImportPage | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| importRevision | |
17.65% |
3 / 17 |
|
0.00% |
0 / 1 |
8.03 | |||
| importLogItem | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| importUpload | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| finishImportPage | |
72.22% |
13 / 18 |
|
0.00% |
0 / 1 |
5.54 | |||
| siteInfoCallback | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| pageCallback | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| pageOutCallback | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| revisionCallback | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| logItemCallback | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| nodeAttribute | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| nodeContents | |
80.00% |
8 / 10 |
|
0.00% |
0 / 1 |
7.39 | |||
| doImport | |
79.17% |
38 / 48 |
|
0.00% |
0 / 1 |
17.03 | |||
| handleSiteInfo | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
6 | |||
| handleLogItem | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
72 | |||
| processLogItem | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
72 | |||
| handlePage | |
90.70% |
39 / 43 |
|
0.00% |
0 / 1 |
17.23 | |||
| handleRevision | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
11 | |||
| handleContent | |
88.24% |
15 / 17 |
|
0.00% |
0 / 1 |
8.10 | |||
| makeContent | |
80.00% |
20 / 25 |
|
0.00% |
0 / 1 |
6.29 | |||
| processRevision | |
86.21% |
25 / 29 |
|
0.00% |
0 / 1 |
9.21 | |||
| handleUpload | |
0.00% |
0 / 31 |
|
0.00% |
0 / 1 |
240 | |||
| dumpTemp | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| processUpload | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
42 | |||
| handleContributor | |
92.31% |
12 / 13 |
|
0.00% |
0 / 1 |
6.02 | |||
| processTitle | |
65.22% |
15 / 23 |
|
0.00% |
0 / 1 |
7.51 | |||
| openReader | |
57.14% |
8 / 14 |
|
0.00% |
0 / 1 |
3.71 | |||
| syntaxCheckXML | |
92.31% |
12 / 13 |
|
0.00% |
0 / 1 |
4.01 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * MediaWiki page data importer. |
| 4 | * |
| 5 | * Copyright © 2003,2005 Brooke Vibber <bvibber@wikimedia.org> |
| 6 | * https://www.mediawiki.org/ |
| 7 | * |
| 8 | * @license GPL-2.0-or-later |
| 9 | * @file |
| 10 | * @ingroup SpecialPage |
| 11 | */ |
| 12 | |
| 13 | namespace MediaWiki\Import; |
| 14 | |
| 15 | use InvalidArgumentException; |
| 16 | use MediaWiki\Config\Config; |
| 17 | use MediaWiki\Content\Content; |
| 18 | use MediaWiki\Content\ContentSerializationException; |
| 19 | use MediaWiki\Content\IContentHandlerFactory; |
| 20 | use MediaWiki\Deferred\DeferredUpdates; |
| 21 | use MediaWiki\Deferred\SiteStatsUpdate; |
| 22 | use MediaWiki\HookContainer\HookContainer; |
| 23 | use MediaWiki\HookContainer\HookRunner; |
| 24 | use MediaWiki\Language\Language; |
| 25 | use MediaWiki\MainConfigNames; |
| 26 | use MediaWiki\Page\CacheKeyHelper; |
| 27 | use MediaWiki\Page\PageIdentity; |
| 28 | use MediaWiki\Page\WikiPageFactory; |
| 29 | use MediaWiki\Permissions\Authority; |
| 30 | use MediaWiki\Revision\SlotRecord; |
| 31 | use MediaWiki\Revision\SlotRoleRegistry; |
| 32 | use MediaWiki\Status\Status; |
| 33 | use MediaWiki\Title\ForeignTitle; |
| 34 | use MediaWiki\Title\ImportTitleFactory; |
| 35 | use MediaWiki\Title\NaiveForeignTitleFactory; |
| 36 | use MediaWiki\Title\NaiveImportTitleFactory; |
| 37 | use MediaWiki\Title\NamespaceAwareForeignTitleFactory; |
| 38 | use MediaWiki\Title\NamespaceImportTitleFactory; |
| 39 | use MediaWiki\Title\NamespaceInfo; |
| 40 | use MediaWiki\Title\SubpageImportTitleFactory; |
| 41 | use MediaWiki\Title\Title; |
| 42 | use MediaWiki\Title\TitleFactory; |
| 43 | use MediaWiki\User\ExternalUserNames; |
| 44 | use RuntimeException; |
| 45 | use UnexpectedValueException; |
| 46 | use Wikimedia\Message\MessageParam; |
| 47 | use Wikimedia\Message\MessageSpecifier; |
| 48 | use Wikimedia\NormalizedException\NormalizedException; |
| 49 | use Wikimedia\Rdbms\IDBAccessObject; |
| 50 | use XMLReader; |
| 51 | |
| 52 | /** |
| 53 | * XML file reader for the page data importer. |
| 54 | * |
| 55 | * implements Special:Import |
| 56 | * @ingroup SpecialPage |
| 57 | */ |
| 58 | class WikiImporter { |
| 59 | /** @var XMLReader|null */ |
| 60 | private $reader; |
| 61 | |
| 62 | /** @var string */ |
| 63 | private $sourceAdapterId; |
| 64 | |
| 65 | /** @var array|null */ |
| 66 | private $foreignNamespaces = null; |
| 67 | |
| 68 | /** @var callable|null */ |
| 69 | private $mLogItemCallback; |
| 70 | |
| 71 | /** @var callable */ |
| 72 | private $mUploadCallback; |
| 73 | |
| 74 | /** @var callable|null */ |
| 75 | private $mRevisionCallback; |
| 76 | |
| 77 | /** @var callable|null */ |
| 78 | private $mPageCallback; |
| 79 | |
| 80 | /** @var callable|null */ |
| 81 | private $mSiteInfoCallback; |
| 82 | |
| 83 | /** @var callable|null */ |
| 84 | private $mPageOutCallback; |
| 85 | |
| 86 | /** @var callable|null */ |
| 87 | private $mNoticeCallback; |
| 88 | |
| 89 | /** @var bool|null */ |
| 90 | private $mDebug; |
| 91 | |
| 92 | /** @var bool|null */ |
| 93 | private $mImportUploads; |
| 94 | |
| 95 | /** @var string|null */ |
| 96 | private $mImageBasePath; |
| 97 | |
| 98 | /** @var bool */ |
| 99 | private $mNoUpdates = false; |
| 100 | |
| 101 | /** @var int */ |
| 102 | private $pageOffset = 0; |
| 103 | |
| 104 | private ImportTitleFactory $importTitleFactory; |
| 105 | private ExternalUserNames $externalUserNames; |
| 106 | |
| 107 | /** @var array */ |
| 108 | private $countableCache = []; |
| 109 | |
| 110 | /** @var bool */ |
| 111 | private $disableStatisticsUpdate = false; |
| 112 | |
| 113 | /** |
| 114 | * Authority used for permission checks only (to ensure that the user performing the import is |
| 115 | * allowed to edit the pages they're importing). To skip the checks, use UltimateAuthority. |
| 116 | * |
| 117 | * If you want to also log the import actions, see ImportReporter. |
| 118 | */ |
| 119 | private Authority $performer; |
| 120 | |
| 121 | private Config $config; |
| 122 | private HookRunner $hookRunner; |
| 123 | private Language $contentLanguage; |
| 124 | private NamespaceInfo $namespaceInfo; |
| 125 | private TitleFactory $titleFactory; |
| 126 | private WikiPageFactory $wikiPageFactory; |
| 127 | private UploadRevisionImporter $uploadRevisionImporter; |
| 128 | private IContentHandlerFactory $contentHandlerFactory; |
| 129 | private SlotRoleRegistry $slotRoleRegistry; |
| 130 | |
| 131 | /** |
| 132 | * Creates an ImportXMLReader drawing from the source provided |
| 133 | */ |
| 134 | public function __construct( |
| 135 | ImportSource $source, |
| 136 | Authority $performer, |
| 137 | Config $config, |
| 138 | HookContainer $hookContainer, |
| 139 | Language $contentLanguage, |
| 140 | NamespaceInfo $namespaceInfo, |
| 141 | TitleFactory $titleFactory, |
| 142 | WikiPageFactory $wikiPageFactory, |
| 143 | UploadRevisionImporter $uploadRevisionImporter, |
| 144 | IContentHandlerFactory $contentHandlerFactory, |
| 145 | SlotRoleRegistry $slotRoleRegistry |
| 146 | ) { |
| 147 | $this->performer = $performer; |
| 148 | $this->config = $config; |
| 149 | $this->hookRunner = new HookRunner( $hookContainer ); |
| 150 | $this->contentLanguage = $contentLanguage; |
| 151 | $this->namespaceInfo = $namespaceInfo; |
| 152 | $this->titleFactory = $titleFactory; |
| 153 | $this->wikiPageFactory = $wikiPageFactory; |
| 154 | $this->uploadRevisionImporter = $uploadRevisionImporter; |
| 155 | $this->contentHandlerFactory = $contentHandlerFactory; |
| 156 | $this->slotRoleRegistry = $slotRoleRegistry; |
| 157 | |
| 158 | if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) { |
| 159 | stream_wrapper_register( 'uploadsource', UploadSourceAdapter::class ); |
| 160 | } |
| 161 | $this->sourceAdapterId = UploadSourceAdapter::registerSource( $source ); |
| 162 | |
| 163 | $this->openReader(); |
| 164 | |
| 165 | // Default callbacks |
| 166 | $this->setPageCallback( $this->beforeImportPage( ... ) ); |
| 167 | $this->setRevisionCallback( $this->importRevision( ... ) ); |
| 168 | $this->setUploadCallback( $this->importUpload( ... ) ); |
| 169 | $this->setLogItemCallback( $this->importLogItem( ... ) ); |
| 170 | $this->setPageOutCallback( $this->finishImportPage( ... ) ); |
| 171 | |
| 172 | $this->importTitleFactory = new NaiveImportTitleFactory( |
| 173 | $this->contentLanguage, |
| 174 | $this->namespaceInfo, |
| 175 | $this->titleFactory |
| 176 | ); |
| 177 | $this->externalUserNames = new ExternalUserNames( 'imported', false ); |
| 178 | } |
| 179 | |
| 180 | /** |
| 181 | * @return null|XMLReader |
| 182 | */ |
| 183 | public function getReader() { |
| 184 | return $this->reader; |
| 185 | } |
| 186 | |
| 187 | /** |
| 188 | * @param string $err |
| 189 | */ |
| 190 | public function throwXmlError( $err ) { |
| 191 | $this->debug( "FAILURE: $err" ); |
| 192 | wfDebug( "WikiImporter XML error: $err" ); |
| 193 | } |
| 194 | |
| 195 | /** |
| 196 | * @param string $data |
| 197 | */ |
| 198 | public function debug( $data ) { |
| 199 | if ( $this->mDebug ) { |
| 200 | wfDebug( "IMPORT: $data" ); |
| 201 | } |
| 202 | } |
| 203 | |
| 204 | /** |
| 205 | * @param string $data |
| 206 | */ |
| 207 | public function warn( $data ) { |
| 208 | wfDebug( "IMPORT: $data" ); |
| 209 | } |
| 210 | |
| 211 | /** |
| 212 | * @param string $msg |
| 213 | * @phpcs:ignore Generic.Files.LineLength |
| 214 | * @param MessageParam|MessageSpecifier|string|int|float|list<MessageParam|MessageSpecifier|string|int|float> ...$params |
| 215 | * See Message::params() |
| 216 | */ |
| 217 | public function notice( $msg, ...$params ) { |
| 218 | if ( is_callable( $this->mNoticeCallback ) ) { |
| 219 | ( $this->mNoticeCallback )( $msg, $params ); |
| 220 | } else { # No ImportReporter -> CLI |
| 221 | // T177997: the command line importers should call setNoticeCallback() |
| 222 | // for their own custom callback to echo the notice |
| 223 | wfDebug( wfMessage( $msg, $params )->text() ); |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | /** |
| 228 | * Set debug mode... |
| 229 | * @param bool $debug |
| 230 | */ |
| 231 | public function setDebug( $debug ) { |
| 232 | $this->mDebug = $debug; |
| 233 | } |
| 234 | |
| 235 | /** |
| 236 | * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer |
| 237 | * @param bool $noupdates |
| 238 | */ |
| 239 | public function setNoUpdates( $noupdates ) { |
| 240 | $this->mNoUpdates = $noupdates; |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * Sets 'pageOffset' value. So it will skip the first n-1 pages |
| 245 | * and start from the nth page. It's 1-based indexing. |
| 246 | * @param int $nthPage |
| 247 | * @since 1.29 |
| 248 | */ |
| 249 | public function setPageOffset( $nthPage ) { |
| 250 | $this->pageOffset = $nthPage; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * Set a callback that displays notice messages |
| 255 | * |
| 256 | * @param callable $callback |
| 257 | * @return callable |
| 258 | */ |
| 259 | public function setNoticeCallback( $callback ) { |
| 260 | return wfSetVar( $this->mNoticeCallback, $callback ); |
| 261 | } |
| 262 | |
| 263 | /** |
| 264 | * Sets the action to perform as each new page in the stream is reached. |
| 265 | * @param callable|null $callback |
| 266 | * @return callable|null |
| 267 | */ |
| 268 | public function setPageCallback( $callback ) { |
| 269 | $previous = $this->mPageCallback; |
| 270 | $this->mPageCallback = $callback; |
| 271 | return $previous; |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * Sets the action to perform as each page in the stream is completed. |
| 276 | * Callback accepts the page title (as a Title object), a second object |
| 277 | * with the original title form (in case it's been overridden into a |
| 278 | * local namespace), and a count of revisions. |
| 279 | * |
| 280 | * @param callable|null $callback |
| 281 | * @return callable|null |
| 282 | */ |
| 283 | public function setPageOutCallback( $callback ) { |
| 284 | $previous = $this->mPageOutCallback; |
| 285 | $this->mPageOutCallback = $callback; |
| 286 | return $previous; |
| 287 | } |
| 288 | |
| 289 | /** |
| 290 | * Sets the action to perform as each page revision is reached. |
| 291 | * @param callable|null $callback |
| 292 | * @return callable|null |
| 293 | */ |
| 294 | public function setRevisionCallback( $callback ) { |
| 295 | $previous = $this->mRevisionCallback; |
| 296 | $this->mRevisionCallback = $callback; |
| 297 | return $previous; |
| 298 | } |
| 299 | |
| 300 | /** |
| 301 | * Sets the action to perform as each file upload version is reached. |
| 302 | * @param callable $callback |
| 303 | * @return callable |
| 304 | */ |
| 305 | public function setUploadCallback( $callback ) { |
| 306 | $previous = $this->mUploadCallback; |
| 307 | $this->mUploadCallback = $callback; |
| 308 | return $previous; |
| 309 | } |
| 310 | |
| 311 | /** |
| 312 | * Sets the action to perform as each log item reached. |
| 313 | * @param callable $callback |
| 314 | * @return callable |
| 315 | */ |
| 316 | public function setLogItemCallback( $callback ) { |
| 317 | $previous = $this->mLogItemCallback; |
| 318 | $this->mLogItemCallback = $callback; |
| 319 | return $previous; |
| 320 | } |
| 321 | |
| 322 | /** |
| 323 | * Sets the action to perform when site info is encountered |
| 324 | * @param callable $callback |
| 325 | * @return callable |
| 326 | */ |
| 327 | public function setSiteInfoCallback( $callback ) { |
| 328 | $previous = $this->mSiteInfoCallback; |
| 329 | $this->mSiteInfoCallback = $callback; |
| 330 | return $previous; |
| 331 | } |
| 332 | |
| 333 | /** |
| 334 | * Sets the factory object to use to convert ForeignTitle objects into local |
| 335 | * Title objects |
| 336 | * @param ImportTitleFactory $factory |
| 337 | */ |
| 338 | public function setImportTitleFactory( $factory ) { |
| 339 | $this->importTitleFactory = $factory; |
| 340 | } |
| 341 | |
| 342 | /** |
| 343 | * Set a target namespace to override the defaults |
| 344 | * @param null|int $namespace |
| 345 | * @return bool |
| 346 | */ |
| 347 | public function setTargetNamespace( $namespace ) { |
| 348 | if ( $namespace === null ) { |
| 349 | // Don't override namespaces |
| 350 | $this->setImportTitleFactory( |
| 351 | new NaiveImportTitleFactory( |
| 352 | $this->contentLanguage, |
| 353 | $this->namespaceInfo, |
| 354 | $this->titleFactory |
| 355 | ) |
| 356 | ); |
| 357 | return true; |
| 358 | } elseif ( |
| 359 | $namespace >= 0 && |
| 360 | $this->namespaceInfo->exists( intval( $namespace ) ) |
| 361 | ) { |
| 362 | $namespace = intval( $namespace ); |
| 363 | $this->setImportTitleFactory( |
| 364 | new NamespaceImportTitleFactory( |
| 365 | $this->namespaceInfo, |
| 366 | $this->titleFactory, |
| 367 | $namespace |
| 368 | ) |
| 369 | ); |
| 370 | return true; |
| 371 | } else { |
| 372 | return false; |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | /** |
| 377 | * Set a target root page under which all pages are imported |
| 378 | * @param null|string $rootpage |
| 379 | * @return Status |
| 380 | */ |
| 381 | public function setTargetRootPage( $rootpage ) { |
| 382 | $status = Status::newGood(); |
| 383 | $nsInfo = $this->namespaceInfo; |
| 384 | if ( $rootpage === null ) { |
| 385 | // No rootpage |
| 386 | $this->setImportTitleFactory( |
| 387 | new NaiveImportTitleFactory( |
| 388 | $this->contentLanguage, |
| 389 | $nsInfo, |
| 390 | $this->titleFactory |
| 391 | ) |
| 392 | ); |
| 393 | } elseif ( $rootpage !== '' ) { |
| 394 | $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes |
| 395 | $title = Title::newFromText( $rootpage ); |
| 396 | |
| 397 | if ( !$title || $title->isExternal() ) { |
| 398 | $status->fatal( 'import-rootpage-invalid' ); |
| 399 | } elseif ( !$nsInfo->hasSubpages( $title->getNamespace() ) ) { |
| 400 | $displayNSText = $title->getNamespace() === NS_MAIN |
| 401 | ? wfMessage( 'blanknamespace' )->text() |
| 402 | : $this->contentLanguage->getNsText( $title->getNamespace() ); |
| 403 | $status->fatal( 'import-rootpage-nosubpage', $displayNSText ); |
| 404 | } else { |
| 405 | // set namespace to 'all', so the namespace check in processTitle() can pass |
| 406 | $this->setTargetNamespace( null ); |
| 407 | $this->setImportTitleFactory( |
| 408 | new SubpageImportTitleFactory( |
| 409 | $nsInfo, |
| 410 | $this->titleFactory, |
| 411 | $title |
| 412 | ) |
| 413 | ); |
| 414 | } |
| 415 | } |
| 416 | return $status; |
| 417 | } |
| 418 | |
| 419 | /** |
| 420 | * @param string $dir |
| 421 | */ |
| 422 | public function setImageBasePath( $dir ) { |
| 423 | $this->mImageBasePath = $dir; |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * @param bool $import |
| 428 | */ |
| 429 | public function setImportUploads( $import ) { |
| 430 | $this->mImportUploads = $import; |
| 431 | } |
| 432 | |
| 433 | /** |
| 434 | * @since 1.31 |
| 435 | * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames |
| 436 | * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally |
| 437 | */ |
| 438 | public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) { |
| 439 | $this->externalUserNames = new ExternalUserNames( $usernamePrefix, $assignKnownUsers ); |
| 440 | } |
| 441 | |
| 442 | /** |
| 443 | * Statistics update can cause a lot of time |
| 444 | * @since 1.29 |
| 445 | */ |
| 446 | public function disableStatisticsUpdate() { |
| 447 | $this->disableStatisticsUpdate = true; |
| 448 | } |
| 449 | |
| 450 | /** |
| 451 | * Default per-page callback. Sets up some things related to site statistics |
| 452 | * @param array $titleAndForeignTitle Two-element array, with Title object at |
| 453 | * index 0 and ForeignTitle object at index 1 |
| 454 | * @return bool |
| 455 | */ |
| 456 | public function beforeImportPage( $titleAndForeignTitle ) { |
| 457 | $title = $titleAndForeignTitle[0]; |
| 458 | $page = $this->wikiPageFactory->newFromTitle( $title ); |
| 459 | $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable(); |
| 460 | return true; |
| 461 | } |
| 462 | |
| 463 | /** |
| 464 | * Default per-revision callback, performs the import. |
| 465 | * @param WikiRevision $revision |
| 466 | * @return bool |
| 467 | */ |
| 468 | public function importRevision( $revision ) { |
| 469 | if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) { |
| 470 | $this->notice( 'import-error-bad-location', |
| 471 | $revision->getTitle()->getPrefixedText(), |
| 472 | $revision->getID(), |
| 473 | $revision->getModel(), |
| 474 | $revision->getFormat() |
| 475 | ); |
| 476 | |
| 477 | return false; |
| 478 | } |
| 479 | |
| 480 | try { |
| 481 | return $revision->importOldRevision(); |
| 482 | } catch ( ContentSerializationException ) { |
| 483 | $this->notice( 'import-error-unserialize', |
| 484 | $revision->getTitle()->getPrefixedText(), |
| 485 | $revision->getID(), |
| 486 | $revision->getModel(), |
| 487 | $revision->getFormat() |
| 488 | ); |
| 489 | } |
| 490 | |
| 491 | return false; |
| 492 | } |
| 493 | |
| 494 | /** |
| 495 | * Default per-revision callback, performs the import. |
| 496 | * @param WikiRevision $revision |
| 497 | * @return bool |
| 498 | */ |
| 499 | public function importLogItem( $revision ) { |
| 500 | return $revision->importLogItem(); |
| 501 | } |
| 502 | |
| 503 | /** |
| 504 | * Dummy for now... |
| 505 | * @param WikiRevision $revision |
| 506 | * @return bool |
| 507 | */ |
| 508 | public function importUpload( $revision ) { |
| 509 | $status = $this->uploadRevisionImporter->import( $revision ); |
| 510 | return $status->isGood(); |
| 511 | } |
| 512 | |
| 513 | /** |
| 514 | * Mostly for hook use |
| 515 | * @param PageIdentity $pageIdentity |
| 516 | * @param ForeignTitle $foreignTitle |
| 517 | * @param int $revCount |
| 518 | * @param int $sRevCount |
| 519 | * @param array $pageInfo |
| 520 | * @return bool |
| 521 | */ |
| 522 | public function finishImportPage( PageIdentity $pageIdentity, $foreignTitle, $revCount, |
| 523 | $sRevCount, $pageInfo |
| 524 | ) { |
| 525 | // Update article count statistics (T42009) |
| 526 | // The normal counting logic in WikiPage->doEditUpdates() is designed for |
| 527 | // one-revision-at-a-time editing, not bulk imports. In this situation it |
| 528 | // suffers from issues of replica DB lag. We let WikiPage handle the total page |
| 529 | // and revision count, and we implement our own custom logic for the |
| 530 | // article (content page) count. |
| 531 | if ( !$this->disableStatisticsUpdate ) { |
| 532 | $page = $this->wikiPageFactory->newFromTitle( $pageIdentity ); |
| 533 | |
| 534 | $page->loadPageData( IDBAccessObject::READ_LATEST ); |
| 535 | $rev = $page->getRevisionRecord(); |
| 536 | if ( $rev === null ) { |
| 537 | |
| 538 | wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $pageIdentity . |
| 539 | ' because WikiPage::getRevisionRecord() returned null' ); |
| 540 | } else { |
| 541 | $update = $page->newPageUpdater( $this->performer )->prepareUpdate(); |
| 542 | $countKey = 'title_' . CacheKeyHelper::getKeyForPage( $pageIdentity ); |
| 543 | $countable = $update->isCountable(); |
| 544 | if ( array_key_exists( $countKey, $this->countableCache ) && |
| 545 | $countable != $this->countableCache[$countKey] ) { |
| 546 | DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [ |
| 547 | 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] ) |
| 548 | ] ) ); |
| 549 | } |
| 550 | } |
| 551 | } |
| 552 | |
| 553 | $title = Title::newFromPageIdentity( $pageIdentity ); |
| 554 | return $this->hookRunner->onAfterImportPage( $title, $foreignTitle, |
| 555 | $revCount, $sRevCount, $pageInfo ); |
| 556 | } |
| 557 | |
| 558 | /** |
| 559 | * Notify the callback function of site info |
| 560 | * @param array $siteInfo |
| 561 | * @return mixed|false |
| 562 | */ |
| 563 | private function siteInfoCallback( $siteInfo ) { |
| 564 | if ( $this->mSiteInfoCallback ) { |
| 565 | return ( $this->mSiteInfoCallback )( $siteInfo, $this ); |
| 566 | } else { |
| 567 | return false; |
| 568 | } |
| 569 | } |
| 570 | |
| 571 | /** |
| 572 | * Notify the callback function when a new "<page>" is reached. |
| 573 | * @param array $title |
| 574 | */ |
| 575 | public function pageCallback( $title ) { |
| 576 | if ( $this->mPageCallback ) { |
| 577 | ( $this->mPageCallback )( $title ); |
| 578 | } |
| 579 | } |
| 580 | |
| 581 | /** |
| 582 | * Notify the callback function when a "</page>" is closed. |
| 583 | * @param PageIdentity $pageIdentity |
| 584 | * @param ForeignTitle $foreignTitle |
| 585 | * @param int $revCount |
| 586 | * @param int $sucCount Number of revisions for which callback returned true |
| 587 | * @param array $pageInfo Associative array of page information |
| 588 | */ |
| 589 | private function pageOutCallback( PageIdentity $pageIdentity, $foreignTitle, $revCount, |
| 590 | $sucCount, $pageInfo ) { |
| 591 | if ( $this->mPageOutCallback ) { |
| 592 | ( $this->mPageOutCallback )( $pageIdentity, $foreignTitle, $revCount, $sucCount, $pageInfo ); |
| 593 | } |
| 594 | } |
| 595 | |
| 596 | /** |
| 597 | * Notify the callback function of a revision |
| 598 | * @param WikiRevision $revision |
| 599 | * @return bool|mixed |
| 600 | */ |
| 601 | private function revisionCallback( $revision ) { |
| 602 | if ( $this->mRevisionCallback ) { |
| 603 | return ( $this->mRevisionCallback )( $revision, $this ); |
| 604 | } else { |
| 605 | return false; |
| 606 | } |
| 607 | } |
| 608 | |
| 609 | /** |
| 610 | * Notify the callback function of a new log item |
| 611 | * @param WikiRevision $revision |
| 612 | * @return mixed|false |
| 613 | */ |
| 614 | private function logItemCallback( $revision ) { |
| 615 | if ( $this->mLogItemCallback ) { |
| 616 | return ( $this->mLogItemCallback )( $revision, $this ); |
| 617 | } else { |
| 618 | return false; |
| 619 | } |
| 620 | } |
| 621 | |
| 622 | /** |
| 623 | * Retrieves the contents of the named attribute of the current element. |
| 624 | * @param string $attr The name of the attribute |
| 625 | * @return string The value of the attribute or an empty string if it is not set in the current |
| 626 | * element. |
| 627 | */ |
| 628 | public function nodeAttribute( $attr ) { |
| 629 | return $this->reader->getAttribute( $attr ) ?? ''; |
| 630 | } |
| 631 | |
| 632 | /** |
| 633 | * Shouldn't something like this be built-in to XMLReader? |
| 634 | * Fetches text contents of the current element, assuming |
| 635 | * no sub-elements or such scary things. |
| 636 | * @return string |
| 637 | * @internal |
| 638 | */ |
| 639 | public function nodeContents() { |
| 640 | if ( $this->reader->isEmptyElement ) { |
| 641 | return ""; |
| 642 | } |
| 643 | $buffer = ""; |
| 644 | while ( $this->reader->read() ) { |
| 645 | switch ( $this->reader->nodeType ) { |
| 646 | case XMLReader::TEXT: |
| 647 | case XMLReader::CDATA: |
| 648 | case XMLReader::SIGNIFICANT_WHITESPACE: |
| 649 | $buffer .= $this->reader->value; |
| 650 | break; |
| 651 | case XMLReader::END_ELEMENT: |
| 652 | return $buffer; |
| 653 | } |
| 654 | } |
| 655 | |
| 656 | $this->reader->close(); |
| 657 | return ''; |
| 658 | } |
| 659 | |
| 660 | /** |
| 661 | * Primary entry point |
| 662 | * @throws \Exception |
| 663 | * @return bool |
| 664 | */ |
| 665 | public function doImport() { |
| 666 | $this->syntaxCheckXML(); |
| 667 | |
| 668 | // Calls to reader->read need to be wrapped in calls to |
| 669 | // libxml_disable_entity_loader() to avoid local file |
| 670 | // inclusion attacks (T48932). |
| 671 | // phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847 |
| 672 | $oldDisable = @libxml_disable_entity_loader( true ); |
| 673 | try { |
| 674 | $this->reader->read(); |
| 675 | |
| 676 | if ( $this->reader->localName != 'mediawiki' ) { |
| 677 | // phpcs:ignore Generic.PHP.NoSilencedErrors |
| 678 | @libxml_disable_entity_loader( $oldDisable ); |
| 679 | $error = libxml_get_last_error(); |
| 680 | if ( $error ) { |
| 681 | throw new NormalizedException( "XML error at line {line}: {message}", [ |
| 682 | 'line' => $error->line, |
| 683 | 'message' => $error->message, |
| 684 | ] ); |
| 685 | } else { |
| 686 | throw new UnexpectedValueException( |
| 687 | "Expected '<mediawiki>' tag, got '<{$this->reader->localName}>' tag." |
| 688 | ); |
| 689 | } |
| 690 | } |
| 691 | $this->debug( "<mediawiki> tag is correct." ); |
| 692 | |
| 693 | $this->debug( "Starting primary dump processing loop." ); |
| 694 | |
| 695 | $keepReading = $this->reader->read(); |
| 696 | $skip = false; |
| 697 | $pageCount = 0; |
| 698 | while ( $keepReading ) { |
| 699 | $tag = $this->reader->localName; |
| 700 | if ( $this->pageOffset ) { |
| 701 | if ( $tag === 'page' ) { |
| 702 | $pageCount++; |
| 703 | } |
| 704 | if ( $pageCount < $this->pageOffset ) { |
| 705 | $keepReading = $this->reader->next(); |
| 706 | continue; |
| 707 | } |
| 708 | } |
| 709 | $type = $this->reader->nodeType; |
| 710 | |
| 711 | if ( !$this->hookRunner->onImportHandleToplevelXMLTag( $this ) ) { |
| 712 | // Do nothing |
| 713 | } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) { |
| 714 | break; |
| 715 | } elseif ( $tag == 'siteinfo' ) { |
| 716 | $this->handleSiteInfo(); |
| 717 | } elseif ( $tag == 'page' ) { |
| 718 | $this->handlePage(); |
| 719 | } elseif ( $tag == 'logitem' ) { |
| 720 | $this->handleLogItem(); |
| 721 | } elseif ( $tag != '#text' ) { |
| 722 | $this->warn( "Unhandled top-level XML tag $tag" ); |
| 723 | |
| 724 | $skip = true; |
| 725 | } |
| 726 | |
| 727 | if ( $skip ) { |
| 728 | $keepReading = $this->reader->next(); |
| 729 | $skip = false; |
| 730 | $this->debug( "Skip" ); |
| 731 | } else { |
| 732 | $keepReading = $this->reader->read(); |
| 733 | } |
| 734 | } |
| 735 | } finally { |
| 736 | // phpcs:ignore Generic.PHP.NoSilencedErrors |
| 737 | @libxml_disable_entity_loader( $oldDisable ); |
| 738 | $this->reader->close(); |
| 739 | } |
| 740 | |
| 741 | return true; |
| 742 | } |
| 743 | |
| 744 | private function handleSiteInfo() { |
| 745 | $this->debug( "Enter site info handler." ); |
| 746 | $siteInfo = []; |
| 747 | |
| 748 | // Fields that can just be stuffed in the siteInfo object |
| 749 | $normalFields = [ 'sitename', 'base', 'generator', 'case' ]; |
| 750 | |
| 751 | while ( $this->reader->read() ) { |
| 752 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 753 | $this->reader->localName == 'siteinfo' ) { |
| 754 | break; |
| 755 | } |
| 756 | |
| 757 | $tag = $this->reader->localName; |
| 758 | |
| 759 | if ( $tag == 'namespace' ) { |
| 760 | $this->foreignNamespaces[$this->nodeAttribute( 'key' )] = |
| 761 | $this->nodeContents(); |
| 762 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 763 | $siteInfo[$tag] = $this->nodeContents(); |
| 764 | } |
| 765 | } |
| 766 | |
| 767 | $siteInfo['_namespaces'] = $this->foreignNamespaces; |
| 768 | $this->siteInfoCallback( $siteInfo ); |
| 769 | } |
| 770 | |
| 771 | private function handleLogItem() { |
| 772 | $this->debug( "Enter log item handler." ); |
| 773 | $logInfo = []; |
| 774 | |
| 775 | // Fields that can just be stuffed in the pageInfo object |
| 776 | $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp', |
| 777 | 'logtitle', 'params' ]; |
| 778 | |
| 779 | while ( $this->reader->read() ) { |
| 780 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 781 | $this->reader->localName == 'logitem' ) { |
| 782 | break; |
| 783 | } |
| 784 | |
| 785 | $tag = $this->reader->localName; |
| 786 | |
| 787 | if ( !$this->hookRunner->onImportHandleLogItemXMLTag( $this, $logInfo ) ) { |
| 788 | // Do nothing |
| 789 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 790 | $logInfo[$tag] = $this->nodeContents(); |
| 791 | } elseif ( $tag == 'contributor' ) { |
| 792 | $logInfo['contributor'] = $this->handleContributor(); |
| 793 | } elseif ( $tag != '#text' ) { |
| 794 | $this->warn( "Unhandled log-item XML tag $tag" ); |
| 795 | } |
| 796 | } |
| 797 | |
| 798 | $this->processLogItem( $logInfo ); |
| 799 | } |
| 800 | |
| 801 | /** |
| 802 | * @param array $logInfo |
| 803 | * @return mixed|false |
| 804 | */ |
| 805 | private function processLogItem( $logInfo ) { |
| 806 | $revision = new WikiRevision(); |
| 807 | |
| 808 | if ( isset( $logInfo['id'] ) ) { |
| 809 | $revision->setID( $logInfo['id'] ); |
| 810 | } |
| 811 | $revision->setType( $logInfo['type'] ); |
| 812 | $revision->setAction( $logInfo['action'] ); |
| 813 | if ( isset( $logInfo['timestamp'] ) ) { |
| 814 | $revision->setTimestamp( $logInfo['timestamp'] ); |
| 815 | } |
| 816 | if ( isset( $logInfo['params'] ) ) { |
| 817 | $revision->setParams( $logInfo['params'] ); |
| 818 | } |
| 819 | if ( isset( $logInfo['logtitle'] ) ) { |
| 820 | // @todo Using Title for non-local titles is a recipe for disaster. |
| 821 | // We should use ForeignTitle here instead. |
| 822 | $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) ); |
| 823 | } |
| 824 | |
| 825 | $revision->setNoUpdates( $this->mNoUpdates ); |
| 826 | |
| 827 | if ( isset( $logInfo['comment'] ) ) { |
| 828 | $revision->setComment( $logInfo['comment'] ); |
| 829 | } |
| 830 | |
| 831 | if ( isset( $logInfo['contributor']['username'] ) ) { |
| 832 | $revision->setUsername( |
| 833 | $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] ) |
| 834 | ); |
| 835 | } elseif ( isset( $logInfo['contributor']['ip'] ) ) { |
| 836 | $revision->setUserIP( $logInfo['contributor']['ip'] ); |
| 837 | } else { |
| 838 | $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) ); |
| 839 | } |
| 840 | |
| 841 | return $this->logItemCallback( $revision ); |
| 842 | } |
| 843 | |
| 844 | private function handlePage() { |
| 845 | // Handle page data. |
| 846 | $this->debug( "Enter page handler." ); |
| 847 | $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ]; |
| 848 | |
| 849 | // Fields that can just be stuffed in the pageInfo object |
| 850 | $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ]; |
| 851 | |
| 852 | $skip = false; |
| 853 | $badTitle = false; |
| 854 | |
| 855 | while ( $skip ? $this->reader->next() : $this->reader->read() ) { |
| 856 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 857 | $this->reader->localName == 'page' ) { |
| 858 | break; |
| 859 | } |
| 860 | |
| 861 | $skip = false; |
| 862 | |
| 863 | $tag = $this->reader->localName; |
| 864 | |
| 865 | if ( $badTitle ) { |
| 866 | // The title is invalid, bail out of this page |
| 867 | $skip = true; |
| 868 | } elseif ( !$this->hookRunner->onImportHandlePageXMLTag( $this, $pageInfo ) ) { |
| 869 | // Do nothing |
| 870 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 871 | // An XML snippet: |
| 872 | // <page> |
| 873 | // <id>123</id> |
| 874 | // <title>Page</title> |
| 875 | // <redirect title="NewTitle"/> |
| 876 | // ... |
| 877 | // Because the redirect tag is built differently, we need special handling for that case. |
| 878 | if ( $tag == 'redirect' ) { |
| 879 | $pageInfo[$tag] = $this->nodeAttribute( 'title' ); |
| 880 | } else { |
| 881 | $pageInfo[$tag] = $this->nodeContents(); |
| 882 | } |
| 883 | } elseif ( $tag == 'revision' || $tag == 'upload' ) { |
| 884 | if ( !isset( $title ) ) { |
| 885 | $title = $this->processTitle( $pageInfo['title'], |
| 886 | $pageInfo['ns'] ?? null ); |
| 887 | |
| 888 | // $title is either an array of two titles or false. |
| 889 | if ( is_array( $title ) ) { |
| 890 | $this->pageCallback( $title ); |
| 891 | [ $pageInfo['_title'], $foreignTitle ] = $title; |
| 892 | } else { |
| 893 | $badTitle = true; |
| 894 | $skip = true; |
| 895 | } |
| 896 | } |
| 897 | |
| 898 | if ( $title ) { |
| 899 | if ( $tag == 'revision' ) { |
| 900 | $this->handleRevision( $pageInfo ); |
| 901 | } else { |
| 902 | $this->handleUpload( $pageInfo ); |
| 903 | } |
| 904 | } |
| 905 | } elseif ( $tag != '#text' ) { |
| 906 | $this->warn( "Unhandled page XML tag $tag" ); |
| 907 | $skip = true; |
| 908 | } |
| 909 | } |
| 910 | |
| 911 | // @note $pageInfo is only set if a valid $title is processed above with |
| 912 | // no error. If we have a valid $title, then pageCallback is called |
| 913 | // above, $pageInfo['title'] is set and we do pageOutCallback here. |
| 914 | // If $pageInfo['_title'] is not set, then $foreignTitle is also not |
| 915 | // set since they both come from $title above. |
| 916 | if ( array_key_exists( '_title', $pageInfo ) ) { |
| 917 | /** @var Title $title */ |
| 918 | $title = $pageInfo['_title']; |
| 919 | $this->pageOutCallback( |
| 920 | $title, |
| 921 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Set together with _title key |
| 922 | $foreignTitle, |
| 923 | $pageInfo['revisionCount'], |
| 924 | $pageInfo['successfulRevisionCount'], |
| 925 | $pageInfo |
| 926 | ); |
| 927 | } |
| 928 | } |
| 929 | |
| 930 | /** |
| 931 | * @param array &$pageInfo |
| 932 | */ |
| 933 | private function handleRevision( &$pageInfo ) { |
| 934 | $this->debug( "Enter revision handler" ); |
| 935 | $revisionInfo = []; |
| 936 | |
| 937 | $normalFields = [ 'id', 'parentid', 'timestamp', 'comment', 'minor', 'origin', |
| 938 | 'model', 'format', 'text', 'sha1' ]; |
| 939 | |
| 940 | $skip = false; |
| 941 | |
| 942 | while ( $skip ? $this->reader->next() : $this->reader->read() ) { |
| 943 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 944 | $this->reader->localName == 'revision' ) { |
| 945 | break; |
| 946 | } |
| 947 | |
| 948 | $tag = $this->reader->localName; |
| 949 | |
| 950 | if ( !$this->hookRunner->onImportHandleRevisionXMLTag( |
| 951 | $this, $pageInfo, $revisionInfo ) |
| 952 | ) { |
| 953 | // Do nothing |
| 954 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 955 | $revisionInfo[$tag] = $this->nodeContents(); |
| 956 | } elseif ( $tag == 'content' ) { |
| 957 | // We can have multiple content tags, so make this an array. |
| 958 | $revisionInfo[$tag][] = $this->handleContent(); |
| 959 | } elseif ( $tag == 'contributor' ) { |
| 960 | $revisionInfo['contributor'] = $this->handleContributor(); |
| 961 | } elseif ( $tag != '#text' ) { |
| 962 | $this->warn( "Unhandled revision XML tag $tag" ); |
| 963 | $skip = true; |
| 964 | } |
| 965 | } |
| 966 | |
| 967 | $pageInfo['revisionCount']++; |
| 968 | if ( $this->processRevision( $pageInfo, $revisionInfo ) ) { |
| 969 | $pageInfo['successfulRevisionCount']++; |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | private function handleContent(): array { |
| 974 | $this->debug( "Enter content handler" ); |
| 975 | $contentInfo = []; |
| 976 | |
| 977 | $normalFields = [ 'role', 'origin', 'model', 'format', 'text' ]; |
| 978 | |
| 979 | $skip = false; |
| 980 | |
| 981 | while ( $skip ? $this->reader->next() : $this->reader->read() ) { |
| 982 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 983 | $this->reader->localName == 'content' ) { |
| 984 | break; |
| 985 | } |
| 986 | |
| 987 | $tag = $this->reader->localName; |
| 988 | |
| 989 | if ( !$this->hookRunner->onImportHandleContentXMLTag( |
| 990 | $this, $contentInfo ) |
| 991 | ) { |
| 992 | // Do nothing |
| 993 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 994 | $contentInfo[$tag] = $this->nodeContents(); |
| 995 | } elseif ( $tag != '#text' ) { |
| 996 | $this->warn( "Unhandled content XML tag $tag" ); |
| 997 | $skip = true; |
| 998 | } |
| 999 | } |
| 1000 | |
| 1001 | return $contentInfo; |
| 1002 | } |
| 1003 | |
| 1004 | /** |
| 1005 | * @param PageIdentity $page |
| 1006 | * @param int $revisionId |
| 1007 | * @param array $contentInfo |
| 1008 | * |
| 1009 | * @return Content |
| 1010 | */ |
| 1011 | private function makeContent( PageIdentity $page, $revisionId, $contentInfo ) { |
| 1012 | $maxArticleSize = $this->config->get( MainConfigNames::MaxArticleSize ); |
| 1013 | |
| 1014 | if ( !isset( $contentInfo['text'] ) ) { |
| 1015 | throw new InvalidArgumentException( 'Missing text field in import.' ); |
| 1016 | } |
| 1017 | |
| 1018 | // Make sure revisions won't violate $wgMaxArticleSize, which could lead to |
| 1019 | // database errors and instability. Testing for revisions with only listed |
| 1020 | // content models, as other content models might use serialization formats |
| 1021 | // which aren't checked against $wgMaxArticleSize. |
| 1022 | if ( ( !isset( $contentInfo['model'] ) || |
| 1023 | in_array( $contentInfo['model'], [ |
| 1024 | 'wikitext', |
| 1025 | 'css', |
| 1026 | 'json', |
| 1027 | 'javascript', |
| 1028 | 'text', |
| 1029 | '' |
| 1030 | ] ) ) && |
| 1031 | strlen( $contentInfo['text'] ) > $maxArticleSize * 1024 |
| 1032 | ) { |
| 1033 | throw new RuntimeException( 'The text of ' . |
| 1034 | ( $revisionId ? |
| 1035 | "the revision with ID $revisionId" : |
| 1036 | 'a revision' |
| 1037 | ) . " exceeds the maximum allowable size ({$maxArticleSize} KiB)" ); |
| 1038 | } |
| 1039 | |
| 1040 | $role = $contentInfo['role'] ?? SlotRecord::MAIN; |
| 1041 | $model = $contentInfo['model'] ?? $this->slotRoleRegistry |
| 1042 | ->getRoleHandler( $role ) |
| 1043 | ->getDefaultModel( $page ); |
| 1044 | $handler = $this->contentHandlerFactory->getContentHandler( $model ); |
| 1045 | |
| 1046 | $text = $handler->importTransform( $contentInfo['text'] ); |
| 1047 | |
| 1048 | return $handler->unserializeContent( $text ); |
| 1049 | } |
| 1050 | |
| 1051 | /** |
| 1052 | * @param array $pageInfo |
| 1053 | * @param array $revisionInfo |
| 1054 | * @return mixed|false |
| 1055 | */ |
| 1056 | private function processRevision( $pageInfo, $revisionInfo ) { |
| 1057 | $revision = new WikiRevision(); |
| 1058 | |
| 1059 | $revId = $revisionInfo['id'] ?? 0; |
| 1060 | if ( $revId ) { |
| 1061 | $revision->setID( $revisionInfo['id'] ); |
| 1062 | } |
| 1063 | |
| 1064 | $title = $pageInfo['_title']; |
| 1065 | $revision->setTitle( $title ); |
| 1066 | |
| 1067 | $content = $this->makeContent( $title, $revId, $revisionInfo ); |
| 1068 | $revision->setContent( SlotRecord::MAIN, $content ); |
| 1069 | |
| 1070 | foreach ( $revisionInfo['content'] ?? [] as $slotInfo ) { |
| 1071 | if ( !isset( $slotInfo['role'] ) ) { |
| 1072 | throw new RuntimeException( "Missing role for imported slot." ); |
| 1073 | } |
| 1074 | |
| 1075 | $content = $this->makeContent( $title, $revId, $slotInfo ); |
| 1076 | $revision->setContent( $slotInfo['role'], $content ); |
| 1077 | } |
| 1078 | $revision->setTimestamp( $revisionInfo['timestamp'] ?? wfTimestampNow() ); |
| 1079 | |
| 1080 | if ( isset( $revisionInfo['comment'] ) ) { |
| 1081 | $revision->setComment( $revisionInfo['comment'] ); |
| 1082 | } |
| 1083 | |
| 1084 | if ( isset( $revisionInfo['minor'] ) ) { |
| 1085 | $revision->setMinor( true ); |
| 1086 | } |
| 1087 | if ( isset( $revisionInfo['contributor']['username'] ) ) { |
| 1088 | $revision->setUsername( |
| 1089 | $this->externalUserNames->applyPrefix( $revisionInfo['contributor']['username'] ) |
| 1090 | ); |
| 1091 | } elseif ( isset( $revisionInfo['contributor']['ip'] ) ) { |
| 1092 | $revision->setUserIP( $revisionInfo['contributor']['ip'] ); |
| 1093 | } else { |
| 1094 | $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) ); |
| 1095 | } |
| 1096 | if ( isset( $revisionInfo['sha1'] ) ) { |
| 1097 | $revision->setSha1Base36( $revisionInfo['sha1'] ); |
| 1098 | } |
| 1099 | $revision->setNoUpdates( $this->mNoUpdates ); |
| 1100 | |
| 1101 | return $this->revisionCallback( $revision ); |
| 1102 | } |
| 1103 | |
| 1104 | /** |
| 1105 | * @param array &$pageInfo |
| 1106 | * @return mixed |
| 1107 | */ |
| 1108 | private function handleUpload( &$pageInfo ) { |
| 1109 | $this->debug( "Enter upload handler" ); |
| 1110 | $uploadInfo = []; |
| 1111 | |
| 1112 | $normalFields = [ 'timestamp', 'comment', 'filename', 'text', |
| 1113 | 'src', 'size', 'sha1base36', 'archivename', 'rel' ]; |
| 1114 | |
| 1115 | $skip = false; |
| 1116 | |
| 1117 | while ( $skip ? $this->reader->next() : $this->reader->read() ) { |
| 1118 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 1119 | $this->reader->localName == 'upload' ) { |
| 1120 | break; |
| 1121 | } |
| 1122 | |
| 1123 | $tag = $this->reader->localName; |
| 1124 | |
| 1125 | if ( !$this->hookRunner->onImportHandleUploadXMLTag( $this, $pageInfo ) ) { |
| 1126 | // Do nothing |
| 1127 | } elseif ( in_array( $tag, $normalFields ) ) { |
| 1128 | $uploadInfo[$tag] = $this->nodeContents(); |
| 1129 | } elseif ( $tag == 'contributor' ) { |
| 1130 | $uploadInfo['contributor'] = $this->handleContributor(); |
| 1131 | } elseif ( $tag == 'contents' ) { |
| 1132 | $contents = $this->nodeContents(); |
| 1133 | $encoding = $this->reader->getAttribute( 'encoding' ); |
| 1134 | if ( $encoding === 'base64' ) { |
| 1135 | $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) ); |
| 1136 | $uploadInfo['isTempSrc'] = true; |
| 1137 | } |
| 1138 | } elseif ( $tag != '#text' ) { |
| 1139 | $this->warn( "Unhandled upload XML tag $tag" ); |
| 1140 | $skip = true; |
| 1141 | } |
| 1142 | } |
| 1143 | |
| 1144 | if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) { |
| 1145 | $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}"; |
| 1146 | if ( file_exists( $path ) ) { |
| 1147 | $uploadInfo['fileSrc'] = $path; |
| 1148 | $uploadInfo['isTempSrc'] = false; |
| 1149 | } |
| 1150 | } |
| 1151 | |
| 1152 | if ( $this->mImportUploads ) { |
| 1153 | return $this->processUpload( $pageInfo, $uploadInfo ); |
| 1154 | } |
| 1155 | } |
| 1156 | |
| 1157 | /** |
| 1158 | * @param string $contents |
| 1159 | * @return string |
| 1160 | */ |
| 1161 | private function dumpTemp( $contents ) { |
| 1162 | $filename = tempnam( wfTempDir(), 'importupload' ); |
| 1163 | file_put_contents( $filename, $contents ); |
| 1164 | return $filename; |
| 1165 | } |
| 1166 | |
| 1167 | /** |
| 1168 | * @param array $pageInfo |
| 1169 | * @param array $uploadInfo |
| 1170 | * @return mixed |
| 1171 | */ |
| 1172 | private function processUpload( $pageInfo, $uploadInfo ) { |
| 1173 | $revision = new WikiRevision(); |
| 1174 | $revId = $pageInfo['id']; |
| 1175 | $title = $pageInfo['_title']; |
| 1176 | // T292348: text key may be absent, force addition if null |
| 1177 | $uploadInfo['text'] ??= ''; |
| 1178 | $content = $this->makeContent( $title, $revId, $uploadInfo ); |
| 1179 | |
| 1180 | $revision->setTitle( $title ); |
| 1181 | $revision->setID( $revId ); |
| 1182 | $revision->setTimestamp( $uploadInfo['timestamp'] ); |
| 1183 | $revision->setContent( SlotRecord::MAIN, $content ); |
| 1184 | $revision->setFilename( $uploadInfo['filename'] ); |
| 1185 | if ( isset( $uploadInfo['archivename'] ) ) { |
| 1186 | $revision->setArchiveName( $uploadInfo['archivename'] ); |
| 1187 | } |
| 1188 | $revision->setSrc( $uploadInfo['src'] ); |
| 1189 | if ( isset( $uploadInfo['fileSrc'] ) ) { |
| 1190 | $revision->setFileSrc( $uploadInfo['fileSrc'], |
| 1191 | !empty( $uploadInfo['isTempSrc'] ) |
| 1192 | ); |
| 1193 | } |
| 1194 | if ( isset( $uploadInfo['sha1base36'] ) ) { |
| 1195 | $revision->setSha1Base36( $uploadInfo['sha1base36'] ); |
| 1196 | } |
| 1197 | $revision->setSize( intval( $uploadInfo['size'] ) ); |
| 1198 | $revision->setComment( $uploadInfo['comment'] ); |
| 1199 | |
| 1200 | if ( isset( $uploadInfo['contributor']['username'] ) ) { |
| 1201 | $revision->setUsername( |
| 1202 | $this->externalUserNames->applyPrefix( $uploadInfo['contributor']['username'] ) |
| 1203 | ); |
| 1204 | } elseif ( isset( $uploadInfo['contributor']['ip'] ) ) { |
| 1205 | $revision->setUserIP( $uploadInfo['contributor']['ip'] ); |
| 1206 | } |
| 1207 | $revision->setNoUpdates( $this->mNoUpdates ); |
| 1208 | |
| 1209 | return ( $this->mUploadCallback )( $revision ); |
| 1210 | } |
| 1211 | |
| 1212 | /** |
| 1213 | * @return array |
| 1214 | */ |
| 1215 | private function handleContributor() { |
| 1216 | $this->debug( "Enter contributor handler." ); |
| 1217 | |
| 1218 | if ( $this->reader->isEmptyElement ) { |
| 1219 | return []; |
| 1220 | } |
| 1221 | |
| 1222 | $fields = [ 'id', 'ip', 'username' ]; |
| 1223 | $info = []; |
| 1224 | |
| 1225 | while ( $this->reader->read() ) { |
| 1226 | if ( $this->reader->nodeType == XMLReader::END_ELEMENT && |
| 1227 | $this->reader->localName == 'contributor' ) { |
| 1228 | break; |
| 1229 | } |
| 1230 | |
| 1231 | $tag = $this->reader->localName; |
| 1232 | |
| 1233 | if ( in_array( $tag, $fields ) ) { |
| 1234 | $info[$tag] = $this->nodeContents(); |
| 1235 | } |
| 1236 | } |
| 1237 | |
| 1238 | return $info; |
| 1239 | } |
| 1240 | |
| 1241 | /** |
| 1242 | * @param string $text |
| 1243 | * @param string|null $ns |
| 1244 | * @return array|false |
| 1245 | */ |
| 1246 | private function processTitle( $text, $ns = null ) { |
| 1247 | if ( $this->foreignNamespaces === null ) { |
| 1248 | $foreignTitleFactory = new NaiveForeignTitleFactory( |
| 1249 | $this->contentLanguage |
| 1250 | ); |
| 1251 | } else { |
| 1252 | $foreignTitleFactory = new NamespaceAwareForeignTitleFactory( |
| 1253 | $this->foreignNamespaces ); |
| 1254 | } |
| 1255 | |
| 1256 | $foreignTitle = $foreignTitleFactory->createForeignTitle( $text, |
| 1257 | intval( $ns ) ); |
| 1258 | |
| 1259 | $title = $this->importTitleFactory->createTitleFromForeignTitle( |
| 1260 | $foreignTitle ); |
| 1261 | |
| 1262 | if ( $title === null ) { |
| 1263 | # Invalid page title? Ignore the page |
| 1264 | $this->notice( 'import-error-invalid', $foreignTitle->getFullText() ); |
| 1265 | return false; |
| 1266 | } elseif ( $title->isExternal() ) { |
| 1267 | $this->notice( 'import-error-interwiki', $title->getPrefixedText() ); |
| 1268 | return false; |
| 1269 | } elseif ( !$title->canExist() ) { |
| 1270 | $this->notice( 'import-error-special', $title->getPrefixedText() ); |
| 1271 | return false; |
| 1272 | } elseif ( !$this->performer->definitelyCan( 'edit', $title ) ) { |
| 1273 | # Do not import if the importing wiki user cannot edit this page |
| 1274 | $this->notice( 'import-error-edit', $title->getPrefixedText() ); |
| 1275 | return false; |
| 1276 | } |
| 1277 | |
| 1278 | return [ $title, $foreignTitle ]; |
| 1279 | } |
| 1280 | |
| 1281 | /** |
| 1282 | * Open the XMLReader connected to the source adapter id |
| 1283 | */ |
| 1284 | private function openReader() { |
| 1285 | // Enable the entity loader, as it is needed for loading external URLs via |
| 1286 | // XMLReader::open (T86036) |
| 1287 | // phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847 |
| 1288 | $oldDisable = @libxml_disable_entity_loader( false ); |
| 1289 | |
| 1290 | // A static call, to avoid https://github.com/php/php-src/issues/11548 |
| 1291 | $reader = XMLReader::open( |
| 1292 | 'uploadsource://' . $this->sourceAdapterId, null, LIBXML_PARSEHUGE ); |
| 1293 | if ( $reader instanceof XMLReader ) { |
| 1294 | $this->reader = $reader; |
| 1295 | $status = true; |
| 1296 | } else { |
| 1297 | $status = false; |
| 1298 | } |
| 1299 | if ( !$status ) { |
| 1300 | $error = libxml_get_last_error(); |
| 1301 | // phpcs:ignore Generic.PHP.NoSilencedErrors |
| 1302 | @libxml_disable_entity_loader( $oldDisable ); |
| 1303 | throw new RuntimeException( |
| 1304 | 'Encountered an internal error while initializing WikiImporter object: ' . $error->message |
| 1305 | ); |
| 1306 | } |
| 1307 | // phpcs:ignore Generic.PHP.NoSilencedErrors |
| 1308 | @libxml_disable_entity_loader( $oldDisable ); |
| 1309 | } |
| 1310 | |
| 1311 | /** |
| 1312 | * Check the syntax of the given xml |
| 1313 | */ |
| 1314 | private function syntaxCheckXML() { |
| 1315 | if ( !UploadSourceAdapter::isSeekableSource( $this->sourceAdapterId ) ) { |
| 1316 | return; |
| 1317 | } |
| 1318 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 1319 | $oldDisable = @libxml_disable_entity_loader( false ); |
| 1320 | try { |
| 1321 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 1322 | while ( @$this->reader->read() ); |
| 1323 | $error = libxml_get_last_error(); |
| 1324 | if ( $error ) { |
| 1325 | $errorMessage = 'XML error at line ' . $error->line . ': ' . $error->message; |
| 1326 | wfDebug( __METHOD__ . ': Invalid xml found - ' . $errorMessage ); |
| 1327 | throw new RuntimeException( $errorMessage ); |
| 1328 | } |
| 1329 | } finally { |
| 1330 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 1331 | @libxml_disable_entity_loader( $oldDisable ); |
| 1332 | $this->reader->close(); |
| 1333 | } |
| 1334 | |
| 1335 | // Reopen for the real import |
| 1336 | UploadSourceAdapter::seekSource( $this->sourceAdapterId, 0 ); |
| 1337 | $this->openReader(); |
| 1338 | } |
| 1339 | } |
| 1340 | |
| 1341 | /** @deprecated class alias since 1.46 */ |
| 1342 | class_alias( WikiImporter::class, 'WikiImporter' ); |